Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The impact of airport noise : a case study of Vancouver International Airport Biggs, Andrew John Grainge 1990

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1990_A4_6 B53.pdf [ 10.8MB ]
Metadata
JSON: 831-1.0228306.json
JSON-LD: 831-1.0228306-ld.json
RDF/XML (Pretty): 831-1.0228306-rdf.xml
RDF/JSON: 831-1.0228306-rdf.json
Turtle: 831-1.0228306-turtle.txt
N-Triples: 831-1.0228306-rdf-ntriples.txt
Original Record: 831-1.0228306-source.json
Full Text
831-1.0228306-fulltext.txt
Citation
831-1.0228306.ris

Full Text

THE IMPACT OF AIRPORT NOISE: A CASE STUDY OF VANCOUVER INTERNATIONAL AIRPORT By ANDREW JOHN GRAINGE BIGGS B.Econ.(Hons), U n i v e r s i t y o f Queensland, 1983 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN BUSINESS ADMINISTRATION i n THE FACULTY OF GRADUATE STUDIES (Commerce and Business A d m i n i s t r a t i o n ) We accept t h i s t h e s i s as conforming t o the r e q u i r e d standard THE UNIVERSITY OF BRITISH COLUMBIA February 199 0 © Andrew John Grainge Biggs, 1990 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia Vancouver, Canada Date FelL ZL+ \ <\Q DE-6 (2/88) ABSTRACT The p r i n c i p a l concern of t h i s paper i s the va luat ion of the impact of a i r c r a f t noise associated with Vancouver I n t e r n a t i o n a l A i r p o r t on res idents l i v i n g c lose to the f l i g h t paths. In surveying previous research re levant to the v a l u a t i o n of the impact of a i r p o r t noise , the paper se l ec t s a property value approach i n which the r e s i d e n t i a l property market i s used as a surrogate for a i r p o r t noise nuisance (on the premise that n o i s i e r areas w i l l have lower house p r i c e s than q u i e t e r areas, c e t e r i s p a r i b u s ) . The se lec ted approach involves undertaking hedonic p r i c e modell ing i n which m u l t i p l e regress ion i s used to estimate property va lue , and from which the value of one a t t r i b u t e , exposure to a i r p o r t no i se , may be obtained i m p l i c i t l y . Noise i s measured by the Noise Exposure Forecast (NEF) technique, a widely used procedure for measuring a i r p o r t noise . Several t h e o r e t i c a l concerns with the approach are addressed before reviewing the r e s u l t s of previous a i r p o r t noise s tudies which use hedonic p r i c e models. The r e s u l t s of these models may be reported i n terms of no i se -property value r e l a t i o n s h i p s , measured by a noise d e p r e c i a t i o n s e n s i t i v i t y index (NDSI). I f the func t iona l form of noise i i i s l i n e a r against the natural logarithm of property value , the NDSI w i l l equal the value of the noise c o e f f i c i e n t and w i l l be constant regardless of the absolute p r i c e of the property . The NDSI f igures for previous studies range from about 0.40 to 1.10 percent per dec ibe l for 1967 to 1976 data , with a simple average of about 0.61 percent . Of the two general model types reported, the one i n v o l v i n g i n d i v i d u a l property sa les data i s considered super ior to that using only census information. The model designed for t h i s study uses i n d i v i d u a l p r i c e data for 1987 s i n g l e -detached property sa les i n the Township of Richmond (where the majori ty of no i se -a f fec ted propert ies are s i t u a t e d ) . Data for p h y s i c a l , area , pub l i c sector , a c c e s s i b i l i t y and other c h a r a c t e r i s t i c s are obtained from various sources to enable the model l ing of some 1539 sa les . Two data sets are tes ted with 44 v a r i a b l e s i n i t i a l l y c o l l e c t e d for each: one set i n v o l v i n g only those propert ies in s ide the NEF 25 noise zone (assumed to be a f fec ted by a i r p o r t no i se ) , the other i n v o l v i n g a l l p r o p e r t i e s . For each data set a l t e r n a t i v e func t iona l forms are tes ted , as are severa l approaches for exp la in ing noise (continuous or dummy v a r i a b l e s , and d i f f e r e n t assumed thresho lds ) . Of the eleven models reported , the p r e f e r r e d model includes only those s i n g l e -detached proper t i e s exposed to NEF 25 or higher where the n a t u r a l log of sa les p r i c e i s regressed against s ix teen independent v a r i a b l e s inc lud ing a continuous form of the i i i noise v a r i a b l e (NEF level ) The model has an adjusted R2 value of .634, and a noise c o e f f i c i e n t of -.006484 -imply ing a one un i t increase i n noise r e s u l t s i n more than h a l f of a percent decrease i n property p r i c e (NDSI equals 0.65). The 95 percent confidence i n t e r v a l for t h i s v a r i a b l e i s - .0097 to - .0033. Prel iminary a n a l y s i s of the model type us ing census data i s a lso reported. C e r t a i n matters of in terpre ta t ion are d iscussed before attempting a pre l iminary a p p l i c a t i o n i n the s e t t i n g of Vancouver Internat iona l A i r p o r t i n which the noise impact of a t h i r d runway i s ind icated . What d i s t i n g u i s h e s the modelling for Vancouver Internat iona l A i r p o r t from the previous studies i s the h igh q u a l i t y of i n d i v i d u a l data , the large sample s i z e , the t e s t i n g of s evera l forms for the noise var iab le ( s ) and the explorat ion of a l t e r n a t i v e noise thresholds . Notwithstanding t h i s , the study concludes by o u t l i n i n g severa l areas for further research . i v TABLE OF CONTENTS Page ABSTRACT i i TABLE OF CONTENTS V LIST OF TABLES v i i LIST OF FIGURES ix ACKNOWLEDGEMENTS X CHAPTER 1. INTRODUCTION 1 CHAPTER 2. AIRPORT NOISE 2 2.1 Introduct ion 2 2.2 Impact of noise 2 2.3 Indiv idua l and community reac t ion 5 2.4 A i r p o r t noise sources 9 2.5 Abatement 9 2.6 Measurement 11 2.7 Standards 17 2.8 Noise as an economic problem 2 0 2.9 Vancouver Internat iona l A i r p o r t 21 CHAPTER 3. VALUATION OF AIRPORT NOISE 26 3.1 Introduct ion 2 6 3.2 Property p r i c e approaches: theory 32 3.3 Hedonic p r i c e models: previous s tudies 45 3.4 Hybrid property p r i c e approaches 82 CHAPTER 4. MODEL DESIGN 85 4.1 Introduct ion 85 4.2 Use of the hedonic approach 85 4.3 Study areas 86 4.4 Data sources 87 4.5 Model types 97 4.6 S t r a t i f i c a t i o n 98 4.7 C o l l e c t i o n date 103 v Page 4.8 Sample s i ze 105 4.9 Screening 106 4.10 Noise 107 4.11 Data management 109 CHAPTER 5. ANALYSIS AND RESULTS 111 5.1 Introduction 111 5.2 O r i g i n a l data 113 5.3 Prel iminary analys i s 113 5.4 F i n a l l i s t of poss ib le v a r i a b l e s 117 5.5 The "IN" model 126 5.6 The "ALL" model 140 5.7 Further d iscuss ion 147 5.8 Census model 149 CHAPTER 6. APPLICATION AND CONCLUDING COMMENTS 153 6.1 Introduct ion 153 6.2 Interpretat ion 153 6.3 A p p l i c a t i o n 156 6.4 The impact of a t h i r d runway at Vancouver Internat ional A i r p o r t 157 6.5 Value of the model 160 6.6 Further research 161 BIBLIOGRAPHY 163 APPENDIX A. DESCRIPTION OF NEF TECHNIQUE 183 APPENDIX B. SUMMARIES OF PREVIOUS AIRPORT NOISE STUDIES 184 APPENDIX C. CORRELATION MATRICES 214 v i LIST OF TABLES Table Page 2.1 Community react ion at d i f f e r e n t NEF l e v e l s 8 2.2 Land-use guidance zones 18 3.1 Summary of previous a i r p o r t noise s tudies using hedonic p r i c e models 48 3.2 Phys ica l c h a r a c t e r i s t i c s 66 3.3 Area c h a r a c t e r i s t i c s 67 3.4 Publ ic sector c h a r a c t e r i s t i c s 68 3.5 A c c e s s i b i l i t y c h a r a c t e r i s t i c s 69 4.1 BCAA data 90 4.2 Census enumeration area data 93 4.3 Model types 97 5.1 O r i g i n a l form (untransformed) v a r i a b l e s , a l l Richmond s ingle-detached proper t i e s 114 5.2 F i n a l form p o t e n t i a l v a r i a b l e s , a l l Richmond single-detached propert i e s (ALL) 118 5.3 F i n a l form p o t e n t i a l v a r i a b l e s , Richmond single-detached propert i e s with NEF 25 or higher (IN) 121 5.4 Cases by NEF l e v e l , sample of Richmond single-detached propert i e s 124 5.5 Resul t s , Richmond s ingle-detached proper t i e s with NEF 25 or h igher , pre ferred form model with continuous noise v a r i a b l e (model 1) 128 5.6 Unstandardized c o e f f i c i e n t s , Richmond s ingle-detached proper t i e s with NEF 25 or higher , a l t e r n a t i v e form models with continuous noise v a r i a b l e (models 1, 2, 3, 4) 136 5.7 Resul t s , Richmond s ingle-detached proper t i e s with NEF 25 or h igher , pre ferred form models with a l t e r n a t i v e noise v a r i a b l e s (models 1, 5, 6) 138 v i i Resul t s , a l l Richmond s ingle-detached p r o p e r t i e s , preferred form model with continuous noise var iab l e (model 7) Resu l t s , a l l Richmond s ingle-detached p r o p e r t i e s , preferred form models with a l t e r n a t i v e noise var iab le s (models 7, 9, 10, 11) v i i i LIST OP FIGURES F igure Page 2.1 A t t i t u d e s toward a i r c r a f t noise i n the r e s i d e n t i a l community 7 2.2 Information requirements for NEF procedure 14 2.3 Re la t ionsh ips between noise measures 16 2.4 NEF contours , Vancouver In ternat iona l A i r p o r t , 1987 iw £p**. c*M+Jt 4.1 Median monthly s ingle-detached sa le p r i c e , Greater Vancouver, January 1987 to June 1989 104 i x ACKNOWLEDGEMENTS My i n t e r e s t i n the subject of a i r p o r t noise v a l u a t i o n was prompted by the i n v i t a t i o n of Dr Dean Uyeno and Dr Stan Hamilton to undertake the l i t e r a t u r e review, model design and modell ing for a consultancy project assoc iated with the current proposal to b u i l d a p a r a l l e l runway at Vancouver In ternat iona l A i r p o r t . Consequently, some of the mater ia l wr i t t en for t h i s thes i s i s a lso used for d r a f t i n g the consultancy report presented to Transport Canada i n ear ly 1990. I am g r a t e f u l f or the d i r e c t i o n of my t h e s i s committee: Dr Uyeno, as chairman, for h i s e f for t s i n t r a i n i n g me to balance the academic requirements for the t h e s i s and the demands of the consu l t ing pro jec t ; Dr Hamilton for h i s superv i s ion of the data c o l l e c t i o n and guidance i n the model l ing; and Dr W G Waters II who, being so l i t t l e exposed to the associated consultancy, provided much wise counsel . I acknowledge the cooperation of Transport Canada, the B r i t i s h Columbia Assessment Authori ty and the Real Estate Board of Greater Vancouver, without which c o l l e c t i o n of the data would have been impossible . I express my apprec ia t ion for the e f f o r t s of the many research as s i s tant s involved i n the data c o l l e c t i o n , Teresa Chan's hours of scanning the l i t e r a t u r e , Warren Weir ' s ass i s tance i n us ing SPSS-X on the U n i v e r s i t y ' s mainframe computer, and many usefu l discuss ions with Reagan Prat t on s t a t i s t i c a l concerns. I record my gra t i tude to my employer, the In ter -S ta te Commission, for i t s s trong encouragement and f i n a n c i a l support of my academic pursu i t s outside of A u s t r a l i a . F i n a l l y , I am thankful to those so dear to my hear t : Almighty and sovereign God for the L i f e that I have i n my Lord and Saviour , Jesus C h r i s t , and for being able to undertake the tasks - inc lud ing t h i s t h e s i s - which He gives me to do i n my ear th ly l i f e ; and my wife Tracey and daughter Ruth for t h e i r support and understanding over a very demanding p e r i o d . x CHAPTER 1 - INTRODUCTION The p r i n c i p a l concern of t h i s thes i s i s the v a l u a t i o n of the impact of a i r c r a f t noise associated with Vancouver I n t e r n a t i o n a l A i r p o r t on res idents l i v i n g c lose to the f l i g h t paths. Such v a l u a t i o n i s an important component i n the economic eva luat ion of any changes i n a i r p o r t i n f r a s t r u c t u r e - as has been contemplated i n Vancouver for over two decades. T h i s study i s of considerable contemporary relevance given the s trong community i n t e r e s t i n a i r p o r t no ise . In Chapter 2 the thes i s b r i e f l y surveys re levant noise i ssues and sets the scene for Vancouver In ternat iona l A i r p o r t . The l i t e r a t u r e survey undertaken i n Chapter 3 cons iders var ious approaches to va lu ing noise , focuss ing oh the property p r i c e (or value) approaches and reviewing previous s tudies using hedonic p r i c e models. Chapter 4 o u t l i n e s the model designed for t e s t i n g the property va lue /no i se r e l a t i o n s h i p associated with Vancouver I n t e r n a t i o n a l A i r p o r t . The ana lys i s and r e s u l t s are presented i n Chapter 5. Chapter 6 discusses some matters of i n t e r p r e t a t i o n , undertakes some l i m i t e d a p p l i c a t i o n and makes some concluding remarks. 1 CHAPTER 2 - AIRPORT NOISE 2 .1 INTRODUCTION A i r p o r t s proudly boast of the major p o s i t i v e e x t e r n a l i t i e s they generate for the region ( for example, see Transport Canada 1989); there i s less p r i d e i n the negative e x t e r n a l i t i e s such as extra road t r a f f i c and a i r t r a f f i c each of which generate noise . S t r i c t l y speaking, i n t h i s paper the concern i s with the noise of a i r c r a f t (the p r i n c i p a l source of "airport" n o i s e ) . This chapter cons iders some of the impacts of a i r p o r t noise on i n d i v i d u a l s , soc iety and var ious p a r t i e s , the i n d i v i d u a l and community reac t ion , the sources of a i r p o r t noise , abatement, measurement and standards. The chapter seeks only to summarize these technica l matters, before noting the s tatus of no ise as an economic problem and o u t l i n i n g the s e t t i n g for a i r c r a f t noise at Vancouver I n t e r n a t i o n a l A i r p o r t . 2.2 IMPACT OF NOISE Noise has been defined by Kryter (1985, p . l ) as "audible a c o u s t i c energy that adversely a f f e c t s the p h y s i o l o g i c a l or p s y c h o l o g i c a l wel l being of people". More s imply, noise i s unwanted or excessive sound.' "Sound" may be measured 2 o b j e c t i v e l y , but i t s "unwanted" nature neces sar i l y involves s u b j e c t i v i t y . Much noise i s an unavoidable waste energy by-product of mechanical a c t i v i t y as part of "legit imate" economic / soc ia l a c t i v i t y . Noise cannot be suppressed e n t i r e l y , but i t does not p h y s i c a l l y accumulate - although the damage noise causes may (Nelson 1978a, p . l ) . Noise i s a "bad": i t i s the va luat ion of t h i s neighbourhood disamenity assoc ia ted with an a i r p o r t that t h i s study attempts. The impact of a i r p o r t noise i s d i f f i c u l t to def ine given the s u b j e c t i v i t y invo lved . Numerous impacts are noted i n the l i t e r a t u r e , often with attempts to measure the same. Several broad areas are i d e n t i f i e d : the annoyance experienced by i n d i v i d u a l s , e f fec t s on people 's hea l th , in t er f erence with various human a c t i v i t i e s , v a r i a t i o n i n property va lues , p o l i t i c a l ac t ion , e f f ec t s on the a v i a t i o n indus try and more. Cons ider ing the impact on i n d i v i d u a l s , there are numerous s tudies arguing over the s t a t i s t i c a l v a l i d i t y of r e s u l t s i n d i c a t i n g p a r t i c u l a r a i r p o r t no i s e /hea l th r e l a t i o n s h i p s -f o r example see the dialogue i n Meecham and Shaw (1979), F r e r i c h s et a l . (1980) and Meecham and Shaw (1980) concerning Los Angeles a i r p o r t noise and m o r t a l i t y ra te s . Many authors f i n d the case unproven for l i n k s between a i r c r a f t noise exposure and p s y c h i a t r i c admissions (for example, Jenkins et a l . 1981) and p h y s i c a l growth (for 3 example, S c h e l l and N o r e l l i 1983, S c h e l l and Hodges 1985), but at l e a s t one author considers the ignor ing of heal th impacts "outrageous" (Lane 1986). Whatever the s t a t i s t i c a l merits of some such analyses , there appears to be l i t t l e dispute i n the l i t e r a t u r e that a i r c r a f t noise does i n t e r f e r e to some degree with s l eep ing , speech and comprehension, and the act ions and thought process . Commonly c i t e d are the effects of "jet pause" (as an a i r c r a f t f l i e s over) i n any group i n t e r a c t i o n , with school teaching being a commonly c i t e d example. Th i s i s j u s t one example of a d iverse range of human a c t i v i t y inter ferences (a very comprehensive l i s t i n g of such noise c o n f l i c t s i s to be found i n US 1972, p .57 f f ) . There are a l so economic, s o c i a l and community impl i ca t ions of a i r p o r t no i se . Concern about noise has seen extensive noise regu la t ions which are rendering a large proport ion of the wor ld ' s a i r l i n e f l e e t ( p a r t i c u l a r l y that owned by l e s s developed countr ies) obsolete for serv ing many countr ies or p a r t i c u l a r a i r p o r t s (Taneja 1988, p .58 ) . For the manufacturers, noise c e r t i f i c a t i o n of a i r c r a f t i s an invo lved process ( s t i f l i n g , argues Smith 1983). A i r l i n e equipment purchasing dec i s ions are a f f ec ted , with there being pressure to operate l a r g e r a i r c r a f t l e s s frequent ly - although according to Schonfeld (1984, pp 3-4) , 4 a l l other things being equal , for a given number of passengers, operat ing small planes more frequent ly has the same sound i n t e n s i t y impact. Operating f l e x i b i l i t y i s g r e a t l y inf luenced by noise controls such as a i r p o r t curfews. P r o v i s i o n of a i r p o r t i n f r a s t r u c t u r e i s inf luenced by noise . Muldoon (1984, p.13) suggests that "a i rpor t noise i s the s ing le major impediment to a i r p o r t expansion and to development of new a i r p o r t s i n the U . S . " . Such a statement h i g h l i g h t s the p o l i t i c a l impact of a i r p o r t no ise . The d i v e r s i t y of noise impl icat ions r e s u l t s i n there being numerous p a r t i c i p a n t s i n what may be termed the p o l i t i c s of a i r p o r t noise . Stevenson (1972) i d e n t i f i e s the fo l lowing groups of actors i n the United States context: p r i v a t e and industry ( p i l o t s , a i r l i n e s , manufacturers, r e a l estate developers , b u i l d e r s , a i r p o r t operators) , s tate and l o c a l governments, the court system, the federal l e g i s l a t u r e , f edera l admin i s t ra t ion , and - somewhere - the general p u b l i c . 2.3 INDIVIDUAL AND COMMUNITY REACTION De Vany's (1976, p.206) suggestion that "there i s no noise i f no one complains about i t " leads to cons idera t ion of the s u b j e c t i v i t y of noise r e f l e c t e d i n i n d i v i d u a l s ' react ions to no ise . The l i t e r a t u r e notes that perceived no i s iness v a r i e s 5 by both noise source and by i n d i v i d u a l . Further , a p a r t i c u l a r i n d i v i d u a l may respond d i f f e r e n t l y , i n the case of a i r c r a f t noise , depending on the l e v e l (or i n t e n s i t y ) , time of day, ambient noise l e v e l , d istance to a i r p o r t , the a c t i v i t y engaged i n at the time, and more. Spencer (1980, p.167) and others note the r o l e of an i n d i v i d u a l ' s e thn ic , s o c i a l , c u l t u r a l and economic background. For instance, ethnic d i f ferences are discussed by S c h l i c k (1984), F lorent ine et a l . (1986), Kuwano et a l . (1986). Occupational status i s also important, as i s the perce ived economic importance of the a i r p o r t to the l o c a l economy. P o l i t i c a l ac t iv i sm i n response to a i r p o r t noise i s d iscussed i n Goodman and Clary (1976), and i n Harvey et a l . (1979) who note the i n t e r r e l a t i o n s h i p between household a t t r i b u t e s and cogni t ion of noise and behaviour. Plessas (1973, p.20) concludes that income i s "by f a r the most powerful p r e d i c t o r of noise annoyance". There i s a lso the r o l e of crash fear (the anxiety factor) which causes greater d is turbance i f i n d i v i d u a l s are c loser to the f l i g h t path. Although noise source appears to be important i n determining r e a c t i o n s , the l i t e r a t u r e contains mixed evidence. For ins tance , K r y t e r (1985) and others argue that there i s greater s e n s i t i v i t y to a i r c r a f t noise than to road t r a f f i c no i se , but Brog et a l . (1982) present survey evidence that road t r a f f i c noise i s more i r r i t a t i n g . 6 There i s large variance i n i n d i v i d u a l react ions , but when aggregated across "communities", the reac t ion i s more p r e d i c t a b l e . This has been summarized for varying Noise Exposure Forecast (NEF, a measure of noise l a t e r explained) l e v e l s in US (1977), and reproduced i n Figure 2.1: at NEF 25 about 20 percent of the community rates noise as unacceptable for r e s i d e n t i a l l i v i n g , r i s i n g s t e a d i l y to 80 percent r a t i n g for NEF 45. FIGURE 2.1 - ATTITUDES TOWARD AIRCRAFT NOISE IN THE RESIDENTIAL COMMUNITY 10 20 30 40 NOISE EXPOSURE FORECAST (NEF) Source: US (1977, p.17) In Canada, the Centra l Housing and Mortgage Corporat ion (CHMC 1978, p.6) s tates that "authorit ies must expect complaints where outdoor NEF values exceed 25. Some people, however, are prepared to l i v e where noise exposure i s above t h i s va lue , p a r t i c u l a r l y i f t h e i r dwell ing provides an acceptable indoor noise environment". Table 2.1 l i s t s a pat tern of community r e a c t i o n at d i f f e r e n t NEF l e v e l s which broadly r e f l e c t s Figure 2 .1 . TABLE 2.1 - COMMUNITY REACTION AT DIFFERENT NEF LEVELS Below NEF 30: Sporadic complaints may occur. Noise may i n t e r f e r e occas iona l ly with c e r t a i n a c t i v i t i e s of the res ident . NEF 30 to 35: Sporadic to repeated i n d i v i d u a l complaints . Group a c t i o n i s poss ib l e . NEF 35 to 40: I n d i v i d u a l complaints may be v igorous . Poss ib le group act ion and appeals to a u t h o r i t i e s . Above NEF 40: Repeated and vigorous i n d i v i d u a l complaints are l i k e l y . Concerted group and l e g a l a c t i o n might be expected. Source: CHMC (1978, p.6) S o c i a l surveys of a i r p o r t (and other) noise annoyance are we l l summarized i n Schul tz (1982, chapter 4) . In a d d i t i o n to fac tors a lready mentioned, surveys have revealed the inf luences of seasonal e f f e c t s , a c t i v i t y i n t e r f e r e n c e , background no ise , and indoors and outdoors exposure. 8 2.4 AIRPORT NOISE SOURCES The p r i n c i p a l source of a i r p o r t noise i s a i r c r a f t engine noise which i s p a r t i c u l a r l y loud when a i r c r a f t are approaching or t a k i n g - o f f from a i r p o r t s ; l ow- leve l f l y i n g i s a l so noisy , whereas i n - f l i g h t c r u i s i n g (unless at speeds breaking the sound b a r r i e r ) i s of l i t t l e concern. The propagation of t h i s noise i s a f fected by d i s tance , a i r c r a f t a l t i t u d e , temperature, humidity, wind, temperature and humidity gradients , turbulence, ground impedance and topography ( B u g l i a r e l l o et a l . 1976, p.173) . There i s increas ing awareness of a i r c r a f t ground noise (Lange 1984), although i n t h i s paper the concern i s only with the e f fec ts on nearby res idents of such noise caused by engine run-up and reverse thrus t - and not the apron noise hazard for a i r p o r t employees (for example, as d iscussed by Herzing 1984). 2.5 ABATEMENT Although only of i n d i r e c t concern to t h i s paper, i t i s worthwhile noting the range of abatement p o l i c i e s / s t r a t e g i e s a v a i l a b l e and appl ied - many of them at Vancouver In terna t iona l A i r p o r t - i n an e f f o r t to c u r t a i l or lower noise impacts. The l i t e r a t u r e i d e n t i f i e s s t r a t e g i e s ranging 9 from purely t echn ica l so lut ions through to j u d i c i a l s o l u t i o n s . These may be categorized i n various ways - see, f or example, US (1972, pp 80-1), Spencer (1980a,b) and Bragdon (1983). Methods for reducing noise at source include use of sound absorbent mater ia l on engines, and changes i n engine and a irframe design on new a i r c r a f t and by r e t r o f i t t i n g for o l d . R e s t r i c t i o n s on a i r c r a f t operations include c o n t r o l on r u n -up noise and other ground operat ions , a i r c r a f t type, time of day l i m i t a t i o n s (such as curfews), f lyover noise standards, l i m i t s on t o t a l d a i l y / h o u r l y movements, regulat ions on g l i d e s lope , f l i g h t t rack , a l t i t u d e and reduced t h r u s t , use of p r e f e r e n t i a l runways, cooperation between a i r p o r t s i n a r e g i o n , and more. Such abatement s tra teg ies may be achieved by a v a r i e t y of r e g u l a t i o n , t e c h n i c a l contro l s and s p e c i f i c a t i o n s and f i n a n c i a l incent ives such as no i se -r e l a t e d landing charges. Also i n f l u e n t i a l are a i r p o r t  p lanning and design - i n v o l v i n g such matters as l o c a t i o n , s i z e , runway length and d i r e c t i o n and management. Much e f f o r t i s focussed on imiss ion contro l s (the reduct ion of noise reaching the i n d i v i d u a l ) through the use of opera t iona l procedures already noted, i n s u l a t i o n , cons truc t ion of noise b a r r i e r s , sound masking and land-use c o n t r o l s . Bragdon (1984) i d e n t i f i e s the fo l lowing land-use c o n t r o l s : zoning, comprehensive p lans , land a c q u i s i t i o n , 10 a v i a t i o n a l easement, r e a l estate sales d i s c l o s u r e , b u i l d i n g code, c a p i t a l improvements, development r i g h t s , s i t e design, land banking, tax incent ives , subdiv i s ion regula t ions and environmental impact reviews. A f i n a l category of noise abatement e f f o r t s i s feedback and  c o n t r o l mechanisms i n which l ega l a c t i v i t y i s prominent. The l e g a l bas is for a c t i o n var i e s by country with a i r c r a f t noise i n the a i r not being a basis of l e g a l a c t i o n i n the United Kingdom and many countries of Eng l i sh her i tage (Spencer 1980b). In the United States there i s much more l i a b i l i t y and l i t i g a t i o n . Harper (1988) provides a contemporary survey of the United States scene, and foundational papers inc lude those by Baxter and A l t r e e (1972) and i n Hildebrand (1970), as wel l as Schles inger (1980), B e l l and B e l l (1980) and Dolley and C a r r o l l (1983). 2.6 MEASUREMENT Issues of a i r p o r t noise abatement lead again to questions of noise measurement. As e a r l i e r noted, because noise i s unwanted sound i t s measurement involves s u b j e c t i v i t y which i s hard to capture by any t e c h n i c a l or engineering c r i t e r i a . To t h i s ambiguity one must add the element of nat iona l i sm that has resu l ted i n numerous noise d e s c r i p t o r s , that i s , 11 measurements (Spencer 198 0a, chapter 3). To quote Schultz (1982, pp 2-3), The reasons for the p r o l i f e r a t i o n of noise ra t ings . . . have been mostly h i s t o r i c a l , and have stemmed from the des ire that they should i n some sense mirror the e x t r a o r d i n a r i l y complex working of the human ear and b r a i n , without being any more complicated than necessary for the immediate problem at hand. There i s agreement that descr ip tors are e i t h e r for s i n g l e -event noise (such as EPNdB, the E f f e c t i v e Perceived Noise Leve l i n dec ibels ) or for cumulative noise (such as NNI, the Noise and Number Index). Schultz (1982, chapter 2) descr ibes and evaluates twenty-two d i f f e r e n t single-number r a t i n g sca les and fourteen d i f f e r e n t r a t i n g procedures. Procedures attempt to take account of the context of the noise s t imulus , whereas scales attempt only to descr ibe some aspect of the noise stimulus (Schultz 1982, pp 22-3). Key elements to be considered i n procedures include the l e n g t h / d u r a t i o n of noise , the number of r e p e t i t i o n s , the time of day, season, background noise and the impulsive nature . The use of procedures appears to be an attempt to b e t t e r measure nois iness (that i s , subject ive annoyance), u n l i k e sca les which can more o b j e c t i v e l y measure loudness or no i se . For a i r p o r t noise s tudies the i n t e r e s t i s thus i n measurement procedures. There are various f a m i l i e s of such 12 procedures, but i n t e r - r a t i n g c o r r e l a t i o n i s high. S c h u l t z (1982, p.228) quotes Connor: The b e s t known measures o f a i r c r a f t n o i s e exposure, CNR, NNI and NEF, are more h i g h l y c o r r e l a t e d among themselves than i s any one of them w i t h annoyance, and they are about e q u a l l y e f f e c t i v e i n c o n j u n c t i o n w i t h s o c i a l v a r i a b l e s i n t h e p r e d i c t i o n of annoyance. I t f o l l o w s t h a t t h e e l a b o r a t e procedures which have been d e v i s e d f o r c a l c u l a t i n g the e f f e c t i v e p e r c e i v e d n o i s e l e v e l o f a i r c r a f t f l y o v e r s o f f e r no advantages over r e l a t i v e l y simple measures i n d e a l i n g with community r e a c t i o n , a l t h o u g h they are of g r e a t v a l u e i n comparing a i r c r a f t and e v a l u a t i n g n o i s e r e d u c t i o n . In the North American context, the p r e f e r r e d procedure f o r measuring a i r p o r t n o i s e i n the 1970s was the Noise Exposure F o r e c a s t (NEF). To quote one r e p o r t (NAS 1971b, p.94): The NEF i s now a g e n e r a l l y accepted r a t i o n a l method f o r d e s c r i b i n g the degrees o f community annoyance a s s o c i a t e d w i t h n o i s e exposure i n the v i c i n i t y of e x i s t i n g a i r p o r t s and f o r p r e d i c t i n g t h e l i k e l y impact, i n terms o f community annoyance, of e x t e n s i o n s o f e x i s t i n g a i r p o r t s , i n t e n s i f i c a t i o n of f l i g h t o p e r a t i o n a t e x i s t i n g a i r p o r t s , and of e n t i r e l y new a i r p o r t s . E x p e r i e n c e d u r i n g the p a s t decade has shown a c o r r e l a t i o n between responses p r e d i c t e d by t h i s method and the a c t u a l r e a c t i o n s o f people and e s t a b l i s h e d communities t h a t f a l l w i t h i n such n o i s e exposure zones. The NEF procedure a p p l i e s o n l y t o commercial, not m i l i t a r y , a i r c r a f t . I t i s an e x t e n s i o n o f the Community Noise R a t i n g (CNR) t e c h n i q u e . The procedure t a k e s the E f f e c t i v e P e r c e i v e d Noise L e v e l (EPNL) measure o f the n o i s i n e s s o f a 13 s i n g l e a i r c r a f t sound, re la ted to i n d i v i d u a l subject ive response, and takes account of the frequency of occurrence, time of day and a i r c r a f t track and p r o f i l e . I t thus implies that t o t a l noise exposure i s determined by the occurrence of noise produced by d i f f e r e n t c lasses of commercial a i r c r a f t f l y i n g d i f f e r e n t f l i g h t paths at d i f f e r e n t times of day. Appendix A d e t a i l s the t echn ica l d e s c r i p t i o n of the NEF procedure as explained by Schultz (1982). The information requirements for the NEF procedure are as i n Figure 2.2 FIGURE 2.2 - INFORMATION REQUIREMENTS FOR NEF PROCEDURE Absolute noise leve ls -Noise spectrum Maximum tone Noise duration-A i r c r a f t type Mix of a i r c r a f t Number of operations Runway u t i l i z a t i o n F l i g h t path Operat ing procedures Time of day PNL • EPNL NEF Source: Ashford and Wright (1979, p.424) The NEF procedure has not escaped c r i t i c i s m . Schul tz (1982, p.142) comments on i t s h igh ly d i f f e r e n t i a t e d approach to es t imat ing noise exposure but i t s r e l a t i v e l y u n d i f f e r e n t i a t e d approach to est imating community r e a c t i o n . 14 Abelson (1977, p .358) notes i t s apparent lower r e l i a b i l i t y at low t r a f f i c volumes, i t s not taking account of reverse t h r u s t or ground running, i t s a r b i t r a r y night weighting f a c t o r and i t s poor c o r r e l a t i o n with i n d i v i d u a l noise annoyance l e v e l s . He concludes: "However, although the NEF model does not d iscr iminate f i n e l y between i n d i v i d u a l a t t i t u d e s , i t s i s poss ib le to argue that i t succeeds i n measuring the average community reac t ion to noise . . " In the l a t e 1970s the Federal A v i a t i o n Adminis trat ion i n the Uni ted States adopted another cumulative noise event measure, the day/night average sound l e v e l (L^ or DNL) as the pre ferred method of measuring noise r e s u l t i n g from a i r c r a f t operat ions. The i s an energy average sound l e v e l which treats noise between 2200 and 0700 at lOdB higher than at the res t of the day (see Schultz 1982, pp 161-2 for a t echnica l d e s c r i p t i o n ) . The f i n a l noise measure which should be mentioned i s the Noise and Number Index (NNI), a r e l a t i v e l y s impler measure used i n the United Kingdom. NNI, and NEF are the cumulative event measures most frequent ly found i n the a i r p o r t noise studies reviewed l a t e r i n t h i s paper. They are s i m i l a r i n t h e i r bas ic form, and t h e i r r e l a t i o n s h i p s have been summarized by Ashford et a l . (1984) as reproduced i n F igure 2 .3 . 15 In Canada, and NEF descr iptors are commonly used. Each d e s c r i p t o r has been used in e a r l i e r s tudies at Vancouver In terna t iona l A i r p o r t , although the o f f i c i a l de scr ip tor for Transport Canada noise forecast ing purposes i s NEF. For Vancouver, equals approximately NEF + 32 (Transport Canada 1981a, p .67) . FIGURE 2.3 - RELATIONSHIPS BETWEEN NOISE MEASURES Peak sound energy dE F requencv weighting A-weignted sound level dBIA) Sound exposure level (SE Li Duration Factor Frequency weighting • t Perceived noise level tPNL.) PNdB Pure tone factor Effective perceived noise level IEPNL) Day/night noise level L[)\ Number of / flignts •»„ J Time of \ / / lay S " " V ^ . "cit) • Noise exposure forecast INEF) Noise and number index INNI) Source: Ashford et a l . (1984, p . 56) NEF values can be derived and projected as contours on a map of the area surrounding the a i r p o r t ; r a d i a t i n g from the a i r p o r t , contours are commonly p l o t t e d for NEF 40, 35, 30 and 28 or 25. (The contours for Vancouver In ternat iona l A i r p o r t suppl i ed by Transport Canada are shown l a t e r i n 16 Figure 2.4.) Such values are derived for Vancouver by simulation. Monitoring using mobile and remote monitoring stations also occurs in Vancouver (for example, see Transport Canada 1985). Sophisticated airport noise monitoring is a relatively recent phenomenon in North America according to Bragdon (1985). 2.7 STANDARDS Standards have already been alluded to in the discussion of abatement measures. Although there is multipartite involvement in formulation of noise requirements for aircraft design (for example, the work of ICAO (Zubkov 1988)), there is a lack of unanimity around the world (Feldman 1987, p.37). Similarly, variety exists around the world for land-use guidelines associated with airports. Not only do the units (NEF, NNI, etc.) differ among countries, but so do levels. For example, Spencer (1980b, p^43) notes that in Australia, lower (that is, tougher) NEF standards are applied because that country's residents generally have more open windows and doors, poorer building insulation, and are more often outside. 17 In the United States the Department of Housing and Urban development sets the standards, based on L d n and NEF, for four land use guidance zones as shown i n Table 2.2. TABLE 2.2 - LAND--USE GUIDANCE ZONES Land-use g u i d -Noise exposure c l a s s Noise measurement US Department of Housing and Urban Suggested noise contro l s ance zone ^ NEF Development noise assessment guide l ines A Minimal exposure 0-55 0-20 " c l e a r l y acceptable" Normally requires no spec ia l cons iderat ions B Moderate exposure 55-65 20-30 "normally acceptable" Land use contro l s should be considered C S i g n i f i -cant exposure 65-75 30-40 "normally unacceptable" Noise easements, land use, and other c o m p a t i b i l i t y contro l s recommended D Severe 75 and 40 and " c l e a r l y Containment exposure higher higher unacceptable" wi th in an a i r p o r t boundary or use of p o s i t i v e c o m p a t i b i l i t y contro l s recommended Source: Ashford et a l . (1984, p.77) 18 As an i n d i c a t i o n of Canadian standards, the Central Housing and Mortgage Corporat ion (CHMC 1978, p.6) considers that , Normal cons truct ion i n new r e s i d e n t i a l bu i ld ings should provide an acceptable indoor noise environment up to the 25 NEF l e v e l . The evidence suggests that a broad threshold ex i s t s at the 25 NEF l e v e l above which there i s an ever - increas ing l i k e l i h o o d that normal construct ion w i l l be unable to provide adequate i n s u l a t i o n against a i r c r a f t noise . . . The M i n i s t r y of Transport has advised the Corporat ion , however, that the accuracy of the NEF contours decreases with distance from the runway. While the 3 0 NEF contours i s considered acceptable , the 25 NEF contours, because of deviat ions by a i r c r a f t from s t r a i g h t f l i g h t paths, cannot be accurate ly de l ineated. Consequently, although t h i s contour i s provided by the M i n i s t r y and i s ind ica ted on published contour maps, the Corporat ion i s unable to use i t for the operation of mandatory p o l i c y . The CHMC (1978, pp 6-7) has appl ied the fo l lowing p o l i c y : . where NEF l e v e l s are greater than 35, housing s h a l l be denied f inanc ing under the Nat ional Housing A c t . . where NEF values are between 30 and 35 i n c l u s i v e , housing s h a l l be denied f inancing under the Nat ional Housing Act unless adequate sound i n s u l a t i o n i s provided. . where NEF values are between 25 and 30, the p r o v i s i o n of adequate sound i n s u l a t i o n i s recommended. Housing s h a l l be denied f inanc ing under the Nat ional Housing 19 Act between 28 and 3 0 NEF, when the sound i n s u l a t i o n proposed i s subs tant ia l l y below that considered to be adequate. As l a t e r explained, NEF contour maps suppl ied by Transport Canada are used i n t h i s study. 2.8 NOISE AS AN ECONOMIC PROBLEM Thus f a r , many impacts of noise have been noted, but for t h i s study, e f fects of a i r c r a f t noise such as those on people 's heal th w i l l be of l i t t l e concern except to the extent that such effects are manifest i n market p r i c e s for r e a l property . A i r c r a f t noise i s an economic problem even though i t i s not traded i n an e x p l i c i t market. Several economic c h a r a c t e r i s t i c s of a i r p o r t noise may be noted: i t i s not an ubiqui tous p u b l i c bad, but ra ther i s very l o c a l i z e d ; and i t can be bought i n varying q u a n t i t i e s (Walters 1975, pp 2-3) . This paper assumes median/average noise e f fec ts (as r e f l e c t e d i n average standards) , but acknowledges that v a r i a b l e noise s e n s i t i v i t y does e x i s t (Paul 1971, p.298) . As the OECD (1975, p.242) notes, there w i l l always be some "imperturbable" and some "eterna l ly d i s s a t i s f i e d " res idents 20 i n the a i r p o r t v i c i n i t y . The concern then w i l l be with average noise impact as a step i n the va luat ion of aggregate community impact. Some amount of noise i s good, given the costs and benef i t s of i t s abatement, implying that an optimum noise l e v e l w i l l e x i s t i n any p a r t i c u l a r context . To measure the optimum noise l e v e l for an a i r p o r t , or to assess the impact of any changes a f f ec t ing noise l e v e l s , monetary v a l u a t i o n i s r e q u i r e d . Such va lua t ion i s the object ive of t h i s study. 2.9 VANCOUVER INTERNATIONAL AIRPORT Vancouver Internat iona l A i r p o r t (for which the i n t e r n a t i o n a l designator i s YVR) i s located on Sea Is land, adjacent to the S t r a i t of Georgia, l e s s than t h i r t e e n ki lometres south of downtown Vancouver, j u s t north and east of the Township of Richmond, and less than for ty ki lometres north of the United States . Some 14000 people were d i r e c t l y employed at the a i r p o r t i n 1987, 40 percent of whom l i v e d i n Richmond (25 percent of whose workforce was employed at the a i r p o r t ) and 25 percent i n Vancouver. The a i r p o r t i s the second bus ies t a i r p o r t i n Canada, enplaning and deplaning more than nine m i l l i o n passengers and handl ing over 95 000 tonnes of cargo i n over 325 000 a i r c r a f t movements i n 1988 (Transport Canada, 1989). 21 The a i r p o r t current ly has i n t e r s e c t i n g runways, but uses them v i r t u a l l y as i f they were p a r a l l e l with simultaneous a i r c r a f t movements: both the main east-west runway (08/26) and the crosswind runway (12/30) are heavi ly used. I t i s the crosswind runway that r e s u l t s i n most noise problems for people i n the southern parts of Vancouver (Transport Canada 1981a, chapter 4) . The NEF contour map for Vancouver In ternat iona l A i r p o r t , shown as Figure 2.4, ind icates that only a r e l a t i v e l y small number of Vancouver propert ies are wi th in the NEF 25 contour (although outside t h i s contour, complaints a lso or ig ina te i n other areas of Vancouver), whereas a large part of Richmond i s wi th in the NEF 25 area with many areas s i tuated ins ide the higher noise contours . A i r p o r t construct ion and expansion have always a t t r a c t e d much p o l i t i c a l a t tent ion around the world (Feldman and M i l c h 1982) and i n Canada (Feldman and M i l c h 1983). Behind the o f f i c i a l h i s t o r y of Vancouver Internat iona l A i r p o r t (Transport Canada 1984, pp 669-85), l i e s an i n t r i g u i n g t a l e which need not be d e t a i l e d here. The l e g a l and p u b l i c a c t i v i t y surrounding i t s runway requirements has continued s ince the issues of expropr ia t ion by the Federal Government f i r s t arose i n 1954. In 1967 the Federal Government s ta ted i t s i n t e n t i o n to acquire land on Sea Is land to b u i l d a p a r a l l e l runway. Since then, c a l l s for a p a r a l l e l runway 22 FIGURE 2.4 - NEF CONTOURS, VANCOUVER INTERNATIONAL AIRPORT, 1987 23 24 have a l so been made i n 1972, 1979, 1984, 1988 and most r e c e n t l y on November 6, 1989 when i t s cons truc t ion , subject to r e s u l t s of an environmental impact study, was announced (Smith and Daniels 1989). Numerous s tud ies , reports and the l i k e have been generated, many of them devoting much a t t e n t i o n to a i r p o r t noise - i n p a r t i c u l a r Transport Canada (1975a,b), as we l l as Transport Canada (1976, 1980, 1981a). The current debate includes arguments over r e a l capaci ty requirements, p r i c i n g issues, t r a f f i c c o n t r o l matters, and more (see Lamb 1989c,d,e) , but the key issue i s a i r c r a f t noise - as acknowledged by the A i r p o r t Manager a f t er the 1989 announcement (Daniels 1989). T h i s makes cons iderat ion of a way to value the impact of noise assoc ia ted with Vancouver In ternat iona l A i r p o r t p a r t i c u l a r l y re l evant . 25 CHAPTER 3 - VALUATION OF AIRPORT NOISE 3 .1 INTRODUCTION A survey of previous research relevant to the valuation of the impact of a i r p o r t noise, reveals studies i n several important areas of the l i t e r a t u r e , as follows: . studies involving the valuation of public goods such as clean a i r , clean water and quiet (or a l t e r n a t i v e l y p u b l i c bads such as a i r , water and noise pollution) and which are generally concerned with s o c i a l cost valuation. . studies concerned with the valuation of r e a l property and the contributing factors. . s p e c i f i c studies r e l a t i n g to the impact of transportation noise on the p r i c e of private goods such as r e a l property. Studies i n a l l three categories are relevant to t h i s review even though the attention - i f any - devoted to a i r p o r t noise and i t s impact on property values varies greatly among the p a r t i c u l a r studies. 2 6 Behind a l l s tudies l i e s debate over s o c i a l cost measurement and the economic concept of e x t e r n a l i t i e s (see, for example, Nash 1978). One e x t e r n a l i t y i s the p u b l i c bad, noise . In t h i s review cons iderat ion i s given to the severa l bas ic approaches used to measure the s o c i a l cost of noise and other e x t e r n a l i t i e s . F ive methods of s o c i a l cost measurement may be noted. Some are d i r e c t ; others are i n d i r e c t . (i) The surrogate market approach (also known as the market p r i c e method) takes advantage of the r e l a t i o n s h i p between p r i v a t e l y marketed goods and publ ic goods by cons ider ing s i t u a t i o n s i n which consumers have a choice between a f i n a n c i a l expenditure on a pr ivate good and s u f f e r i n g the p u b l i c bad. The property market has been used as a surrogate for a i r and noise p o l l u t i o n . When t h i s i s done, the approach i s l a b e l l e d more narrowly as a property value  approach. Where noise i s concerned, property value approaches proceed on the premise that n o i s i e r areas w i l l have lower house p r i c e s than quieter areas , c e t e r i s paribus; that i s , such property owners experience a pecuniary e x t e r n a l i t y (the lower house pr i ce r e f l e c t i n g the t e c h n i c a l e x t e r n a l i t y - the ac tua l noise) . A property value approach i s adopted i n t h i s study. Later , b r i e f reference i s made to the s o - c a l l e d hybr id property p r i c e approaches which d i f f e r i n c e r t a i n ways from the approach undertaken here . 27 ( i i ) The a l t ernat ive cost approach (also re ferred to as the at tenuat ion method or the opportunity cost method) measures the cost of removing the bad. This may include engineering s tudies to determine the cost of amel iorat ing noise by double g laz ing or other soundproofing of b u i l d i n g s . However, complete attenuation may be impossible at any cost , and thus the approach cannot be used to value a l l the s o c i a l cos t . Further , the cost of o f f s e t t i n g the noise may a lso be greater than the cost of the noise annoyance measured by other approaches. ( i i i ) The compensation approach genera l ly leaves measurement to the l ega l system by using court-awarded compensation to indicate market p r i c e changes r e s u l t i n g from the bad. In the United States , courts have found i n favour of property owners near a i r p o r t s and awarded damages for noise (see, for example, Harper 1988). To y i e l d a s a t i s f a c t o r y va lua t ion , such an approach must depend on information obtained from one or more of the a l t e r n a t i v e approaches, and thus cannot stand alone. Compensation may a l so be determined under n o n - j u d i c i a l j u r i s d i c t i o n s such as by a i r p o r t au thor i t i e s when purchasing noise easements or making other no ise -af fec ted property a c q u i s i t i o n s . ( iv) Survey approaches (also known as d i r e c t v a l u a t i o n , d i r e c t quest ioning, or contingent v a l u a t i o n approaches) involve the d i r e c t quest ioning of people regarding t h e i r 28 v a l u a t i o n of environmental impacts. Results are often regarded s u s p i c i o u s l y because of concern that s t r a t e g i c b ias may i n v a l i d a t e the r e s u l t s of any survey. Surveys may seek to d i r e c t l y determine wi l l ingness to pay for lower noise l e v e l s us ing a s ing l e d i r e c t question, converging d i r e c t > quest ions , budget a l l o c a t i o n questions, or t r a d e - o f f games. In some cases, a s imulat ion may be used to determine q u a l i t y choices d i r e c t l y and wi l l ingness to pay i n d i r e c t l y i n actual or optimal markets (Sinden and Worre l l 1979, chapter 14). (v) Votino; approaches include binary (yes/no) referenda on p u b l i c goods supply. Freeman (1979a, p.103) reasons that i n the long run , p o l i t i c a l r a t i o n a l i t y would lead to referendum proposals tending to c l u s t e r around the median voter preference . I f environmental q u a l i t y i s f inanced from wi th in the p o l i t i c a l j u r i s d i c t i o n , t h i s non-market approach i s genera l l y of l i m i t e d use. Furthermore, for noise t h i s approach could never achieve the comprehensive va luat ion r e q u i r e d . The v a l u a t i o n l i t e r a t u r e focusses on two approaches: the surrogate market approach and the d i r e c t survey approach. The surrogate market approach i s used whenever poss ib le so as to take advantage of observed market data , while the survey approach i s used for s i tua t ions i n which market data are harder or impossible to obta in . Several s tudies have compared r e s u l t s us ing these two approaches (for example, 29 Brookshire et a l . 1982, Cummings et a l . 1986, Schechter and Kim 1989). The v e r d i c t i n the l i t e r a t u r e may be summarized as fol lows: although the d i r e c t survey approach may y i e l d i n t e r n a l l y  cons i s tent and v a l i d r e s u l t s , the surrogate market approach  appears to be more r e l i a b l e . When a s p e c i f i c surrogate market approach, the property value approach, i s taken, other advantages are provided: . by using observed market data i t produces ex post va luat ions which may be used at the same or another s i t e , l a t e r i n time for ex ante v a l u a t i o n s . . i t allows an understanding of the dynamics of the property market and the r e l a t i o n s h i p with urban s t r u c t u r e . . i t i s r e l a t i v e l y s tra ight forward (although t h i s depends on the empir i ca l es t imat ion technique used). . the l o g i c i s i n t u i t i v e l y appeal ing . Th i s study thus uses a property value approach to measure the value of a i r p o r t noise nuisance. The l i t e r a t u r e inc ludes var ious techniques to estimate the determinants of the demand for housing c h a r a c t e r i s t i c s , enabl ing the 30 e m p i r i c a l measurement of the value of pub l i c bads such as no ise . F o l l a i n and Jimenez (1985) review f ive empir ica l approaches: simple hedonic, Rosen two-step, b i d - r e n t , Lancas tr ian index and d i s cre t e choice . The appropriate approach depends on the economic issues being considered and the nature of ava i lab le data ( F o l l a i n and Jimenez 1985, p .101) . The major i ty of s tudies , p a r t i c u l a r l y the e a r l i e r ones, have used hedonic p r i c e r e l a t i o n s h i p s to compute d i r e c t l y the marginal e f fec t of a change i n a i r p o r t noise l e v e l s ; others have used the hedonic data i n a second stage of ana lys i s to estimate i n d i v i d u a l inverse demand or marginal w i l l ingness to pay functions for less noise (Freeman 1979a, p.110). There i s much debate concerning the v a l i d i t y of the simple d i r e c t approach, and the assumptions underly ing the use of the two stage (or step) approach. The theory underlying the property p r i c e approaches i s we l l def ined i n the l i t e r a t u r e (see, i n t e r a l i a . F o l l a i n and Jimenez 1985, Freeman 1979a, 1989b, H a r r i s 1981, Nelson 1978a, 1980, Pearce 1978, Rosen 1974, Walters 1975). The e m p i r i c a l app l i ca t ions of hedonic or other models invo lve cons iderable problems which must be overcome as i s l a t e r d i scussed . 31 This chapter summarizes the general ised property p r i c e approach and out l ines a development of the hedonic p r i c e model. This i s followed by a summary of re levant studies which use the hedonic p r i c e model. Further t h e o r e t i c a l debate may be found i n the numerous studies which use the hedonic model. 3.2 PROPERTY PRICE APPROACHES: THEORY The property p r i c e approaches to noise va lua t ion are based on the a b i l i t y of a n o i s e - s e n s i t i v e i n d i v i d u a l to "buy" less noise by purchasing a house i n a quiet area, on the premise that noise i s a neighbourhood disamenity. As Nelson (1981a, p.53 notes, While there are c e r t a i n to be some measurement problems introduced by r e l o c a t i o n costs and avert ing behavior, we can proceed with some assurance that there ex i s t s an i m p l i c i t market for quie t surroundings wherein the p r i c e of qu ie t i s embodied i n the value of r e s i d e n t i a l housing. To quote Pearce (1978, p .32) , Accord ing ly , observation of the behaviour of people who vary i n t h e i r s e n s i t i v i t y to noise should enable us to estimate t h e i r i m p l i c i t (posi t ive) evaluat ions of quiet and hence t h e i r i m p l i c i t (negative) va luat ions of noise . I f , to take an extreme example, an i n d i v i d u a l i s able to choose between two houses, i d e n t i c a l i n every respect except that house A has a peaceful l o c a t i o n and house B a noisy l o c a t i o n , then the existence of n o i s e - s e n s i t i v e people should mean that the p r i c e of A (PA) exceeds the p r i c e of B (PB) . The d i f f e r e n t i a l P A - P B would provide a 32 prima fac ie measure of the extra value of peace and quiet attached to house A. Pearce d e t a i l s four important assumptions required for t h i s genera l ized approach which may be summarized as fo l lows: (i) f u l l mob i l i t y of ind iv idua l s - the e f f ec t of the r e a l i t y of l i m i t e d mobi l i ty i s considered l a t e r . ( i i ) noise l e v e l s vary from one property to another. ( i i i ) noise (or quiet) can be q u a n t i t a t i v e l y measured i n a l i n e a r manner; that i s , there ex i s t s a constant p r i c e per un i t of noise measure. (iv) noise e f fec t s can be i so la t ed from other e f f ec t s on property p r i c e d i f f e r e n t i a l s - as s u c c e s s f u l l y achieved i n many of the studies l a t e r reviewed. These assumptions are not uncontrovers ia l , as evidenced i n fur ther d i scuss ion . Nonetheless they are d e f e n s i b l e , as, i n t e r a l i a . Nelson (1982, pp 118-22) shows. A house and l o t (or for that matter any other type of property) may be considered a composite e n t i t y , represent ing a unique combination of a t t r i b u t e s (or e q u i v a l e n t l y , c h a r a c t e r i s t i c s ) . The sales p r i c e (or rent) w i l l take account of phys i ca l a t t r i b u t e s of the l o t and s t r u c t u r e , 33 neighbourhood a t t r i b u t e s and amenities, a c c e s s i b i l i t y f a c t o r s , p u b l i c sector a t t r i b u t e s , and so on. From these a t t r i b u t e s comes a flow of serv ices y i e l d i n g the fo l lowing funct ion for the t o t a l consumption of housing s e r v i c e s : where C h = consumption of housing s erv i ce s , c 1 , c 2 / . . . , c n = housing a t t r i b u t e s . Each household/consumer has a u t i l i t y funct ion assumed to be of the fo l lowing form: where U = u t i l i t y , X = expenditure on a l l goods other than housing and t r a v e l l i n g to work, D = distance to place of work. Given the consumer has a budget cons tra in t set by h i s income, C h = F ( c 1 # c : (3.1) U = F ^ X ^ D ) , (3.2) 34 Y = X + P h(D) .Ch + CT, (3.3) where Y = income, Ph = price per u n i t of housing, presented here as a function of distance, CT = the cost of t r a v e l l i n g to work. In some models, noise i s treated as being distance dependent (that i s , place of work may coincide with the source of noise). I f one adopts Nelson's (1978a) argument that noise and a l l other attributes other than land services are assumed independent of distance, equations 3.1, 3.2 and a modification of 3.3 r e s u l t i n : Maximize U = F^X,^, ... ,cm,cn,D) , (3.4) subject to: Y = X + l E P i . C i + P n(D).c n + TD, (3.5) i = I where services i=l to i=m = a l l services other than those from land, n t h a t t r i b u t e = the service of land, PA = the price of the i t h service. 35 P i i s the i m p l i c i t or hedonic p r i c e of the a t t r i b u t e s comprising a commodity, here a r e s i d e n t i a l property . Noise p o l l u t i o n may be regarded simply as one such a t t r i b u t e . (For the foundational l i t e r a t u r e on hedonic p r i c e theory, see G r i l i c h e s 1971 and Rosen 1974.) Annual expenditure on housing services alone may be represented as: m H = Y - X - T D = £ p i C i + P n (D) .c n , (3.6) Converting to present value terms, and assuming the time hor izon i s i n f i n i t e , with discount rate , r : m PV(H) = House P r i c e = l / r [ J p i c i + pn ( D ) . c n ] , (3.7) Equation 3.7 may be formulated as a mul t ip le regress ion equation i n which the house p r i c e (the dependent var iab le ) i s a funct ion of var ious a t t r i b u t e s (the independent v a r i a b l e s ) . The regress ion c o e f f i c i e n t s thus represent the i m p l i c i t marginal p r i c e s for the included a t t r i b u t e s , one of which may be noise . More t e c h n i c a l l y , the m u l t i p l e regress ion procedure i s used to estimate the inverse compensated demand funct ions which can be used d i r e c t l y , or 36 to obtain the benefi ts of vary ing p a r t i c u l a r amenity l e v e l s (Harr is 1981). T h i s then i s hedonic p r i c e a n a l y s i s : " in which the values of .independent components of a heterogeneous good are determined i m p l i c i t l y through regress ion analys is" ( M i l l e r 1982, p .33) . This i s a most appealing approach. F i r s t , i t i s very s tra ightforward. Second, i t uses s t a t i s t i c a l techniques which are wel l known and i n widespread use. As Nelson (1978a, p.69) notes, "By i t s e l f the hedonic p r i c e r e l a t i o n s h i p i s rather unexceptional and amounts to l i t t l e more than a simple accounting i d e n t i t y between a un i t p r i c e and expenditures on c h a r a c t e r i s t i c s . There are , however, s evera l t h e o r e t i c a l and empir i ca l issues associated with the s p e c i f i c a t i o n and est imation of hedonic pr i ce s" . Discuss ion of e m p i r i c a l issues and matters of a p p l i c a t i o n i s genera l ly reserved for the next s ec t ion; here, poss ib le l i m i t a t i o n s of the theory need to be noted. The d i scuss ion r e i n f o r c e s the usefulness of the hedonic model, as wel l as h i g h l i g h t i n g the need for care fu l a t t ent ion i n the design of the study. H a r r i s (1981, p.37) i d e n t i f i e s two aspects of the theory at which c r i t i c i s m s are l e v e l l e d : f i r s t , the condit ions under which the observed i m p l i c i t pr i ce s can be assumed to r e f l e c t a household's marginal w i l l ingnes s to pay for a p a r t i c u l a r l e v e l of environmental amenity; and second, the assumptions necessary for the aggregation of estimated i n d i v i d u a l households' 37 marginal wi l l ingness to pay into a market demand funct ion . Drawing further on H a r r i s ' (1981, pp 37-44) assessment of the hedonic p r i c e approach, a t tent ion may be drawn to several t h e o r e t i c a l issues which must be considered i n any empir ica l study. Housing market e q u i l i b r i u m (Issue i) I t i s necessary to estimate marginal i m p l i c i t pr i ce s i n order to measure households' w i l l ingness to pay for a t t r i b u t e s of r e s i d e n t i a l proper t i e s . The economic theory requires a housing market i n e q u i l i b r i u m . Only i n equ i l ibr ium w i l l supply and demand for a t t r i b u t e s be such that marginal i m p l i c i t pr i ce s equal marginal s a l e / b i d p r i c e s . Perfect m o b i l i t y wi th in and between housing markets i s t r a d i t i o n a l l y assumed to ensure an e q u i l i b r i u m . However, with given wealth, given information and i n s t i t u t i o n a l cons tra in t s on i n d i v i d u a l s i n the housing market and given non-instantaneous adjustment of supply and demand, the r e a l i t y i s that i n d i v i d u a l s are not p e r f e c t l y mobile. Thus, the housing market i s not necessar i ly a good example of p e r f e c t l y competit ive e q u i l i b r i u m . Freeman (1979b, pp 160-1) s p e c i f i c a l l y addresses t h i s matter and reasons that such d i s p a r i t y w i l l only introduce random errors to the estimates of marginal w i l l ingness to pay. Whether imperfect ions introduce random or systematic b ias 38 then becomes the issue. Harr i s (1981, pp 38-9) argues that a reasonable expectation i s that given search, t ransact ion , and moving costs , marginal w i l l ingness to pay for housing a t t r i b u t e s i s biased downward, p a r t i c u l a r l y i f house pr i ce s are r i s i n g genera l ly . But he s tates that none of the e m p i r i c a l work has sought to quanti fy the sources of b ias or t h e i r nature . This would c l a r i f y the m o b i l i t y assumption e a r l i e r noted. Housing v a r i e t y (Issue i i ) The hedonic p r i c e model requires a s u f f i c i e n t v a r i e t y of house types to be ava i lab le at each l o c a t i o n ( t echn ica l ly , t h i s i s to s a t i s f y f i r s t - o r d e r condi t ions of u t i l i t y maximization as e q u a l i t i e s ) . Freeman (1979b, pp 161-2) argues that marginal i m p l i c i t p r i c e s w i l l be a reasonable approximation to i n d i v i d u a l marginal w i l l ingness to pay for a t t r i b u t e s i f there i s a large number of house types i n a v a r i e t y of locat ions (such as noise zones) such that the d i f f erences between houses i s continuous i n character . This s a t i s f i e s the noise v a r i a t i o n assumption. Information gap (Issue i i i ) I f the ac tua l l e v e l of an a t t r i b u t e such as noise p o l l u t i o n does not correspond to the l e v e l perce ived by the consuming household, the hedonic p r i c e schedule w i l l not r e f l e c t w i l l i n g n e s s to pay for noise . T h i s i s a p a r t i c u l a r problem for a i r p o l l u t i o n wherein the information gap between higher 39 actua l and lower perceived l eve l s w i l l cause wi l l ingness to pay to underestimate the f u l l damage. Use of measured p o l l u t i o n l eve l s would mis-spec i fy the r e l a t i o n s h i p , g i v i n g biased and incons is tent i m p l i c i t p r i c e estimates; use of perce ived p o l l u t i o n l e v e l s would avoid t h i s problem. In Chapter 2, the v a r i a t i o n i n i n d i v i d u a l percept ion of noise , or the annoyance experienced, was noted, as was the more cons i s tent aggregate, or community r e a c t i o n . Given the d i f f i c u l t i e s in de f in ing any information gap on what i s a subjec t ive matter (except for the equal ly d i f f i c u l t to assess phys i ca l impacts) , i t i s assumed that t h i s i s not a problem. No empir ica l evidence was found concerning the extent of any information gap for noise , although Pearce (1971, p . I l l ) does note that purchasers are cons i s t en t ly o v e r - o p t i m i s t i c i n t h e i r evaluat ion of the noise disamenity. Pearce (1978, p.31) concludes that for noise , perceived and ac tua l p o l l u t i o n are very c l o s e l y l i n k e d , un l ike the s i t u a t i o n for a i r p o l l u t i o n . S t r a t i f i c a t i o n (Issue iv) The p o s s i b i l i t y of s t r a t i f i c a t i o n i n the housing market needs to be considered (Straszheim 1974, p.404, Straszheim 1975a). I f the housing market i n an urban area cons i s t s of separate submarkets (or market segments) each with d i f f e r e n t hedonic p r i c e funct ions , a hedonic p r i c e funct ion estimated for the e n t i r e market w i l l y i e l d inaccurate i m p l i c i t p r i c e estimates for any purchasers i n d i f f e r e n t submarkets. 40 Freeman (1979a, p.142) i d e n t i f i e s two condit ions for the exis tence of d i f f e r e n t hedonic p r i c e functions i n one urban area: there must be b a r r i e r s preventing purchasers i n one market segment p a r t i c i p a t i n g i n other segments (geography, d i s c r i m i n a t i o n , lack of information or des ire for e t h n i c a l l y homogeneous neighbourhoods); and, the demand s t r u c t u r e , supply s t ru c tu re or both must be d i f f e r e n t across regions . I f e i t h e r of these condit ions i s not met, one hedonic p r i c e s t r u c t u r e w i l l be relevant for the whole market. I f both cond i t ions are met, separate functions for each submarket should be estimated. Issues of segmentation i n housing markets are tested by Schnare and Struyk (1976) who conclude for the housing market they study that , "Apparently there i s s u f f i c i e n t s u b s t i t u t i o n on the part of households and/or supply response by housing producers to prevent widespread and persuas ive market segmentation" (p.164). Supply adjustments (Issue v) The schedule of marginal i m p l i c i t p r i c e s e s tab l i shed by the hedonic technique i s l i k e l y to r e s u l t from an i n t e r a c t i o n of demand and supply forces . T e c h n i c a l l y , the i d e n t i f i c a t i o n problem r e f e r r e d to i n the l i t e r a t u r e i s concerned whether an inverse demand function can be i s o l a t e d from supply f o r c e s . Such i d e n t i f i c a t i o n i s poss ib le us ing e i t h e r of two a l t e r n a t i v e models: supply of housing i s assumed i n e l a s t i c at the e x i s t i n g p r i c e l e v e l so that supply i s exogenous (quant i ty i s fixed) (as i n Freeman 1979b, Harr i son and 41 Rubinfeld 1978b), or , supply i s not assumed exogenous i n which case demand i s estimated by simultaneous equations (with a s p e c i f i e d supply equation) (as i n Rosen 1974, Nelson 1978a). H a r r i s (1981, p.41) states: The choice of models depends on the expected speed of adjustment i n the housing market. Only i f the empir ica l researcher can be sure that the supply of housing with a p a r t i c u l a r l e v e l of amenity i s unresponsive to the i m p l i c i t p r i c e of that amenity i n the short run w i l l a c r o s s - s e c t i o n a l est imation of the wi l l ingness - to -pay funct ion i d e n t i f y an inverse demand funct ion . A p p l i c a t i o n of the hedonic technique i n areas where supply i s responsive i n the short run requires c a r e f u l s p e c i f i c a t i o n of a supply equation. Consumer income (Issue v i ) H a r r i s (1981, p.41) s ta tes , i f we assume equ i l ibr ium i n the housing market, the hedonic p r i c e equation r e f l e c t s households' marginal w i l l ingnes s to pay for the a t t r i b u t e s of housing given the supply of housing. The estimated marginal wi l l ingness to pay can be used to estimate an inverse compensated demand curve for the environmental amenity. Once t h i s i s achieved the c a l c u l a t i o n of the benef i t s of a marginal improvement i n environmental q u a l i t y i s r e l a t i v e l y s tra ightforward given a number of assumptions. Among the assumptions i s the c r u c i a l one that the marginal w i l l ingnes s to pay be independent of the consumer's income. This matter i s f a i r l y t e c h n i c a l , but i t s i m p l i c a t i o n s are as fo l lows: 42 The aggregation of ind iv idua l inverse compensated demand curves into an aggregate funct ion requires that the i n d i v i d u a l households have equal marginal propens i t i e s to spend out of income. In p r a c t i c e only i f the group of households i s r e l a t i v e l y homogeneous w i l l an aggregate demand curve e x i s t . Moreover, the aggregation of i n d i v i d u a l household demand funct ions requires that a l l households have i d e n t i c a l u t i l i t y functions. (Harr is 1981, p.43) The way out of the t h e o r e t i c a l d i f f i c u l t i e s may be to include some tas te var iab le s other than income (Freeman 1979b, p.167), or to ensure that the market i s homogeneous with respect to t h i s fac tor i f concerned with underly ing demand parameters ( F o l l a i n and Jimenez 1985, p .81) . Separable u t i l i t y functions (Issue v i i ) Given that c e r t a i n housing a t tr ibutes are l i k e l y to be complements of , or subst i tutes for , p a r t i c u l a r environmental a t t r i b u t e s (for example, double g laz ing and q u i e t ) , u t i l i t y funct ions w i l l not be separable, and the r e s u l t s of the w i l l i n g n e s s to pay est imation w i l l be b iased i f re levant other v a r i a b l e s are excluded. Careful s e l e c t i o n of relevant v a r i a b l e s for the market or submarkets' func t ion i s r e q u i r e d . Many of the above concerns need to be considered i n the design of the hedonic p r i c e model for t h i s study. Otherwise, the hedonic technique, which H a r r i s (1981, p.32) descr ibes as having no l o g i c a l flaws, w i l l be app l i ed less than s a t i s f a c t o r i l y . These l i m i t a t i o n s can be, and i n t h i s 43 study are , taken into account. Other property value approaches are, i f anything, plagued with more problems than the hedonic model ( F o l l a i n and Jimenez 1985). For the purposes of t h i s study which i s concerned with the measurement of the market va luat ion of noise , the simple hedonic method i s judged s u i t a b l e , provided the above concerns are not ignored. I t i s worthwhile quoting Freeman's (1979b, p.155) conc lus ion: . . . one's [s ic] assessment of the hedonic technique seems to depend upon which end of the telescope one looks through i n examining the theory, the assumptions, and the data . The theory i s l o g i c a l and cons i s tent , but i t involves a subs tant ia l s i m p l i f i c a t i o n and abs trac t ion from a complex r e a l i t y . The assumptions are never completely r e a l i z e d i n p r a c t i c e . But t h i s i s a dubious tes t of the v a l i d i t y of an e m p i r i c a l model. I t i s the nature of models i n economics that t h e i r assumptions are to some extent u n r e a l i s t i c . The data are inadequate; v a r i a b l e s are measured with e r r o r ; and the d e f i n i t i o n s of empir i ca l var iab le s seldom correspond p r e c i s e l y to the t h e o r e t i c a l cons tructs . But a l l of these c r i t i c i s m s can be r a i s e d against v i r t u a l l y any empir i ca l work i n economics. The hedonic technique for est imating benef i t s seems to pass the appropriate t e s t s about as w e l l , or as poor ly , as any empir ica l technique for est imating such things as demand funct ions , product ion funct ions , consumption funct ions , and so f o r t h . 44 3.3 HEDONIC PRICE MODELS: PREVIOUS STUDIES The l i t e r a t u r e contains a range of studies which e x p l i c i t l y use hedonic p r i c e models or which are consistent with such models. As intimated i n the in troduct ion to t h i s review of the l i t e r a t u r e , such studies include those concerned with house p r i c e determination (as surveyed i n M i l l e r 1982 and B a l l 1973) and those dea l ing with s p e c i f i c "public" amenities ranging from a i r q u a l i t y (surveyed i n Freeman 1979a), a i r c r a f t noise (surveyed i n Nelson 1980) and t r a f f i c noise (surveyed i n Nelson 1982), to ethnic composition (Schnare 197 6) and nuclear waste dumps (Hageman 1981). A d e t a i l e d taxonomy of r e s i d e n t i a l property value models (by t o p i c with research references) i s provided i n M i l l e r (1982, p.34) . Although i t i s sens ib le to draw on some of these other s tud ies , t h i s review concentrates on those which inc lude a i r p o r t noise as an independent v a r i a b l e i n some form or other . Given that the hedonic p r i c e i s e f f e c t i v e l y a by-product of a regress ion model, there need be l i t t l e i n t e r e s t i n obta in ing the hedonic p r i c e s for other v a r i a b l e s - nor for that matter i n p r e d i c t i n g the dependent v a r i a b l e of property va lue . Of the t h i r t e e n studies considered, two deal with Canadian communities, one with A u s t r a l i a n and the r e s t with the United States . Eleven of the s tudies have been summarized by Nelson (1980); the two others comprise a 45 l a t e r study by Nelson (1979 and 1981a) invo lv ing s ix c i t i e s modelled separately and together, and one by O'Byrne, Nelson and Seneca (1985) i n v o l v i n g two d i f f e r e n t data sets for one c i t y . Any cons iderat ion draws heav i ly on the o r i g i n a l work and the survey work of Nelson who i s the most widely known author i n the subject area . No studies using the hedonic p r i c e model for a i r p o r t noise have been located outs ide North America or A u s t r a l i a . G a u t r i n ' s (1975) work for Heathrow which uses a modified Mohring model of urban land rent to examine the t r a d e - o f f between a minimization of t r a n s p o r t a t i o n costs and disamenity costs i s not considered here. The work of the T h i r d London A i r p o r t study team i s considered b r i e f l y l a t e r i n t h i s chapter. The reviewed studies use a cross sec t ion of property value data ( for a p a r t i c u l a r s u b u r b / c i t y / a r e a , for a p a r t i c u l a r time p e r i o d ) , together with housing c h a r a c t e r i s t i c s and some measure of a i r c r a f t noise exposure. Time-series data are not used i n these s tudies due to the d i f f i c u l t y i n c o n t r o l l i n g for other inf luences on property values over time ( for a time ser ies study, though not using the hedonic approach, see Crowley 1973) . At l eas t one of the s tud ies , Mieszkowski and Saper (1978), f a l l s in to B l a y l o c k ' s (1977, pp 2-3) category of "matched neighbourhood studies" which "compare housing values between noise -af fec ted neighborhoods which, i n most respects , 'match' each other and a c o n t r o l neighborhood experiencing no a i r p o r t noise ." 46 A key problem of such studies i s that the data are not organized by the conceptual s tructure so as to minimize the ambiguity of r e s u l t s . The fo l lowing d i scuss ion aims to present an overview of the previous model l ing, with de ta i l ed cons iderat ion of p a r t i c u l a r aspects re levant i n the model l a t e r used. Summaries of the reviewed studies are contained i n Appendix B and b r i e f d e t a i l s of each study are presented i n Table 3 .1 . Funct ional form Freeman (1979a, p . l39 f f ) discusses some a l t e r n a t i v e func t iona l forms for hedonic p r i c e models i n c l u d i n g : l i n e a r , quadrat ic , l og , semilog, inverse semilog, exponent ia l , semilog exponent ia l , and Box-Cox transformat ion. He notes that " i f the property value funct ion i s l i n e a r , the i m p l i c i t property value funct ion i s l i n e a r , the i m p l i c i t marginal value of p o l l u t i o n i s constant, and est imation of demand curves for [a ir ] q u a l i t y or p o l l u t i o n abatement i s not p o s s i b l e . However most researchers have found non- l inear func t iona l forms to give be t t er f i t s " (pp 140-1). The log and semilog have been the most commonly used funct iona l forms. M i l l e r (1982, p.35) a l so notes the s u p e r i o r i t y of m u l t i p l i c a t i v e and log func t iona l forms. 47 TABLE 3.1 - SUMMARY OF PREVIOUS AIRPORT NOISE STUDIES USING HEDONIC PRICE MODELS PART A: VARIOUS DETAILS Study Community Year Model Sample Dependent R2 NDSI form s i z e v a r i a b l e Abelson (1979) De Vany (1976) M a r r i c k v i l l e , 1972-1973 semilog Rockdale (Sydney) Dal las 1970 Dygert (1973) San Francisco , 1970 i n Nelson San Jose (1978a,1980) Emerson (1972) McDougall (1976) Minneapolis Los Angeles McMillan Edmonton et a l . (1980) 1967 1970 (and others) l i n e a r semilog log log 1975-1976 log 592, 822 1270 128, 198 222 35 352 i n d i v i d u a l house .66 0.4 0 sa le p r i c e .62 0.50 census block mean property value census t r a c t median/mean land value per sq foot i n d i v i d u a l house sa le p r i c e (MLS) census t r a c t median property value i n d i v i d u a l house sa le p r i c e (MLS) 82 83 0. 58 0.80 60 1.50 to 0.70 70 80 0.58 n/a .71 0.50 Maser et a l (1977) Rochester c i t y (also suburbs) 1971 (also l i n e a r 1950, 1960) 393 i n d i v i d u a l house .60 0.55 (990) sa le p r i c e per acre to of land plus s t ruc ture 0.68 TABLE 3.1 (cont'd) - SUMMARY OP PREVIOUS AIRPORT NOISE STUDIES USING HEDONIC PRICE MODELS PART A (cont'd): VARIOUS DETAILS Study Community Year Model Sample Dependent R2 NDSI form s i ze v a r i a b l e Mieszkowski and Saper (1978) Nelson (1978a) Etobicoke, Mississauga (Toronto) 1969-1973 Washington, DC 1970 semilog 509, (also l inear ) 626 log 52 i n d i v i d u a l house sa le p r i c e census t r a c t median property value 81 0.90 86 1.10 Nelson (1979, 1981a) San Franc isco , 1970 St Louis , Cleveland, New Orleans, San Diego, Buffa lo; semilog 113 census block mean .61 0.50 to property value to to 185 .76 0.55 6 c i t i e s pooled 1970 semilog 845 census block mean property value .84 0.50 (adj) to 0. 55 0'Byrne et a l . (1985) At lanta 1979-1980 semilog 96 (also l inear ) i n d i v i d u a l sa le p r i c e house .71 0.67 (adj) At lanta 1970 semilog 248 (also l inear) census block mean .74 0.64 property value (adj) TABLE 3.1 (cont'd) - SUMMARY OF PREVIOUS AIRPORT NOISE STUDIES USING HEDONIC PRICE MODELS PART A (cont'd): VARIOUS DETAILS Study Community Year Model form Sample Dependent s i z e v a r i a b l e NDSI Paik (1972) i n Nelson (1978a, 1980) Paik (1972) remodelled by Nelson (1978a) Pr ice (1974) i n Nelson (1978a, 1980) New York, Los Angeles, Dal las (and pooled) New York, Los Angeles, Dal las (and pooled) Boston 1960 1960 1960, 1970 log log l i n e a r 106, census block 92, median property 94 value 106, census block 92, median property 94 value 270 percent change i n median t r a c t assessed rent over decade .76 2.09 to .79 .78 2.20 ?.50 0.60 TABLE 3.1 (cont'd) - SUMMARY OP PREVIOUS AIRPORT NOISE STUDIES USING HEDONIC PRICE MODELS PART B: SIGNIFICANT VARIABLES Study Phys ica l Area Publ i c sector Access-i b i l i t y Noise Other Abelson (1979) De Vany (1976) l o t frontage, depth, age, not b r i c k , # rooms, ext condi t ion , improvements, rear access, property type, garage, carport , roof type avg # rooms, avg age, % houses with a i r condi t ioning Dygert (1973) i n Nelson (1978a, 1980) Emerson (1972) ? t e r r a i n house s i ze , l o t s i z e , garage space, baths, ranges, f i rep laces , age t r a f f i c l e v e l , road b l i g h t , road width higher density zoning % houses owner-occupied d i s t to a i r p o r t , CBD NEF-25 (also month tested NEF, i n f l . NEF-30, e(NEF-25)/lO (NEF-3CO/10 ' e » zero value for c o n t r o l areas represented by dis tance dumm (very roughly NEF 50, 45, 20) .NEF 25 to 45 ?pub. schools, ?prop tax ?shops e" ?median # of rate ? ind s i t e s (also CNR); people per u n i t , ? a i r p o r t zero value for ?% nonwhite uni t s ?CBD c o n t r o l areas) proximity to freeway, d i s t to open green space, % nonwhite populat ion log 126-CNR (s ig at 90%); threshold CNR 101 (or NEF 30) TABLE 3.1 (cont'd) - SUMMARY OF PREVIOUS AIRPORT NOISE STUDIES USING HEDONIC PRICE MODELS PART B (cont'd): SIGNIFICANT VARIABLES Study- Phys ica l Area McDougall (1976) house s i ze McMillan et a l . (1980) Maser et a l . (1977) house s i ze , age, # zoning bathrooms, # bedrooms, l o t s i z e , # s t o r i e s , duplexes, f i r e p l a c e s , f in i shed basement, garages, b r i c k ext. maintenance, avg # rooms i n owner-occupied uni t s %negro Publ i c sector Access-i b i l i t y Noise Other school q u a l i t y log of weighted f igure of % of planning area exposed to 90dB or more e f fec . d i s t to property CBD tax rate for qu ie t , inverse of (40-NEF); theshold NEF 20 d r i v i n g time to CBD a dummy for for being wi th in lOOdB contour (or NEF 30) TABLE 3.1 (cont'd) - SUMMARY OP PREVIOUS AIRPORT NOISE STUDIES USING HEDONIC PRICE MODELS PART B (cont'd): SIGNIFICANT VARIABLES Study Physical Area Pub l i c sector Access-i b i l i t y Noise Other Mieszkowski and Saper (1978) Nelson (1978a) Nelson (1979,1981a) - i n d i v . c i t i e s 6 c i t i e s pooled house s i z e , avg room s ize , l o t s i z e , # bathrooms, # s t o r i e s , garage, f u l l basement, # rooms i n basement, storms and screens, carpet ing , appl iances , TV antenna, landscaping, f i r e p l a c e s , fac ing , semi-detached, fancy s t y l e , s p l i t l e v e l , condi t ion CNR 95 to 115 month NEF 25 to 45; i n f l . no threshold adjustment # of rooms, age, centra l a i r c o n d i t i o n i n g mean # of rooms, age (for some c i t i e s ) , % a i r condi t ioning (some), substandard plumbing (some) mean # of rooms, % a i r c o n d i t i o n i n g locat ion near r i v e r % owner-occupied units (some) % owner-occupied un i t s , % black pop d r i v time NEF 25, 30, 35 to 75% of ( s ig at 90%); metro empl . threshold NEF 20 NEF 20 or 25 to 45; threshold NEF 20 or 25 NEF 20 or 25 c i t y to 45; dumm. threshold NEF 20 or 25 TABLE 3.1 (cont'd) - SUMMARY OF PREVIOUS AIRPORT NOISE STUDIES USING HEDONIC PRICE MODELS PART B (cont'd): SIGNIFICANT VARIABLES Study Phys ica l Area Pub l i c sector Access-i b i l i t y Noise Other O•Byrne et a l . (1985) - i n d i v . model l i v i n g space area, # of bathrooms, basement, ex ter ior fac ing L d n 65 to 80; i n f l . no contro l census model Paik (1972) i n Nelson (1978a,1980) Paik (1972) remodelled by Nelson (1978a,1980) P r i c e (1974) i n Nelson (1978a,1980) # of rooms # of rooms # of rooms % owner-occupied, % black pop % s ing l e - fami ly , mean people per room, % d e t e r i o r a t i o n (s ig at 9?%) ?decade chg % nonwhite res , ?% o ld people i n 1960, ?% pub renta l hous. u n i t s , ?% age ?% inc i n ? d i s t CBD prop, tax NEF 25 to 45; no contro l log NEF 20, 30, 40; threshold NEF 20 log NEF 20, 30, 40; threshold NEF 20 NEF-25; no apparent threshold c i t y dumm. ?med. rent i n 1960 Nelson (1978a, p.69ff) a l so discusses funct ional form, not ing that i n a housing market which functions imperfect ly (that i s , i n the presence of i n d i v i s i b i l i t i e s , t i e - i n s and j o i n t products i n the supply of housing services) and i n the presence of e l a s t i c supply and demand, marginal hedonic p r i c e s w i l l not be constant and thus not a l i n e a r funct ion . Non- l inear r e l a t i o n s h i p s , i n p a r t i c u l a r for s t r u c t u r a l r e l a t i o n s h i p s , should not be expected for r e s i d e n t i a l housing (see a l so Rosen 1974). Nelson f inds the log model s t a t i s t i c a l l y super ior to l i n e a r and semilog models. More r e c e n t l y , Halvorsen and Pollakowski (1981, p.47) have s tated that previous s tudies employing hedonic p r i c e equations have genera l ly used highly r e s t r i c t i v e func t iona l forms chosen i n a l a r g e l y a r b i t r a r y manner. Since the choice of funct iona l form can have a major e f f ec t on the conclus ions reached i t i s c l e a r l y preferable to base the choice on re levant s t a t i s t i c a l procedures. They then spec i fy a h igh ly general funct ional form which y i e l d s a l l other func t iona l forms as spec ia l cases. Using ac tua l data they t e s t a l t e r n a t i v e funct iona l forms and conclude that the most commonly used forms, l i n e a r , l o g - l i n e a r and semilog are r e j e c t e d . Several s tudies report r e s u l t s using a v a r i e t y of func t iona l forms, although i n such cases the one d e t a i l e d i n Table 3.1 i s for what appears to be the "bes t - f i t equation". 55 Notwithstanding Halvorsen and Pol lakowski 1 s reservat ions , the d e r i v a t i o n of the funct ional form used by Nelson i s reported i n t h i s study because i t appears to be the most commonly used: i t , or s i m i l a r vers ions , are app l i ed i n several of the s tud ies . Nelson (1978a, 1981a) represents the hedonic p r i c e equation for housing with the fo l lowing mult ip le regress ion equation: where b 0 , blf b 2 = constants, X = the set of housing c h a r a c t e r i s t i c s other than noise , N = "subject ive annoyance" due to a i r c r a f t noise , u x = a s tochas t i c error term. This equation i s m u l t i p l i c a t i v e . I t becomes a d d i t i v e i f logs are taken on both s ides . Subject ive annoyance due to a i r c r a f t noise i s measured i n semilogarithmic form as: PV(H) = b 0 X B 1 N D > (3.8) N = c0e' (3.9) where c 0 , c x = constants , e = na tura l logarithm, u 2 = a s tochas t i c error term. 56 NEF i s the noise exposure forecast measured i n d e c i b e l s , the noise d e s c r i p t o r discussed i n Chapter 2.6. The measure i s c l e a r l y explained by Nelson (1980, pp 40-1), as fo l lows: The t o t a l noise exposure produced at a given po int may be viewed as the sum of noise l eve l s produced by d i f f e r e n t a i r c r a f t f l y i n g d i f f e r e n t f l i g h t paths . For the i t h a i r c r a f t on the j t h runway, the NEF algorithm i s NEF i ; j = EPNLi j + 101og(N d i j + 16.7N n i j) - k [3.10] where EPNL i s the average e f f ec t i ve perceived noise l e v e l i n dec ibe l s at a given point i n space, N d i s the number of day-time f l i g h t s 0700 to 2200 hours) , N n i s the number of night-t ime f l i g h t s , and k i s a s ca l e -adjus t ing constant with an assigned value of 88. When summed on an energy bas i s over a l l a i r c r a f t types and f l i g h t paths, noise exposure i s a funct ion of the average perce ived noise l e v e l , time of day, and number of operat ions . (This i s a s l i g h t l y d i f f e r e n t vers ion of the NEF algori thm than the one documented i n Appendix A, i n d i c a t i n g again the v a r i a t i o n i n noise d e s c r i p t o r s . ) S u b s t i t u t i n g N i n the mul t ip l e regress ion equation g ives : PV(H) b2 (3.11) o r , PV(H) 0x e •dl_d2NEF, (3.12) where d 0 b2 57 b2 u 3 = U XU 2 Taking logs, In H = d 0 + d : l n X + d2NEF + u 3 , (3.13) The c o e f f i c i e n t d 2.100 represents the percentage change i n a given property value associated with a dec ibe l change i n noise exposure. "^)H/^)NEF = d2H gives the i m p l i c i t marginal p r i c e of noise - implying that marginal noise damage r i s e s proport ionate ly with property value: that i s , the higher the property value , the greater i s the d o l l a r e f f ec t of a dec ibe l increase i n a i r c r a f t noise , a l l other things remaining constant. Nelson (1982, p.125) notes that , "This i s not incons i s tent with the view that environmental q u a l i t y i s a superior good. The p r i c e of a f a m i l y ' s house i s a good i n d i c a t o r of i t s l e v e l of permanent income." d 2NEF gives the noise e l a s t i c i t y , such that a given percentage increase i n NEF w i l l have a greater impact the higher the NEF l e v e l (Nelson 1981a, p .59) . This model "cardina l izes" the noise scale i n economic terms such that 58 the p r i c e of a u n i t of n o i s e i s the same r e g a r d l e s s of n o i s e l e v e l (Pearce 1978, p .37) . Dependent v a r i a b l e s The dependent v a r i a b l e i n the hedonic p r i c e model i s some measure of property value . Two c lasses of values are reported i n Table 3 .1 . The more common data type i s mean or median property values from census data . The US Census of Populat ion and Housing asks each owner for an estimate of property value (as wel l as gathering other data useful for explanatory v a r i a b l e s ) , with data aggregated by census blocks and t r a c t s and reported as means or medians. Although the census i s a convenient data source, i t s use requires c a r e f u l a t tent ion to s e l e c t i o n of the sample t r a c t s . Several object ions to the use of census data are commonly ra i sed (Freeman 1979a, p.131) . The f i r s t object ion concerns the accuracy of owner-estimated property va lues . Nelson (1978a, p.80) notes c o n f l i c t i n g conclusions i n empir ica l s tudies , but i n h i s study for Washington, DC, compares owner-estimates with profes s iona l assessments of i n d i v i d u a l property p a r c e l s aggregated to the t r a c t l e v e l and f inds a high c o r r e l a t i o n (although i t i s i n t e r e s t i n g to note that owner-estimates were sys temat ica l ly higher by about 3 to 6 percent) . Nelson concludes that , with a large number of households and only random errors i n owner-estimates, the two methods should y i e l d roughly 59 comparable empir i ca l r e s u l t s . F o l l a i n and Malpezzi (1981a, p.54) report , for t h e i r sample i n t h i r t y - n i n e large Selected Metropol i tan Areas in the United States , that most owner/occupants undervalue t h e i r homes by 2 percent , but the r e s u l t v a r i e s , but the r e s u l t var i e s by market and length of tenure. The second objec t ion concerns loss of d e t a i l and reduced a b i l i t y to c o n t r o l for relevant housing and l o c a t i o n c h a r a c t e r i s t i c s because of the l i m i t e d number of v a r i a b l e s reported and the aggregation by census t r a c t . This objec t ion i s of even more concern for explanatory v a r i a b l e s , but may be i n i t i a l l y considered here. Although census t r a c t boundaries are se lected with the aim of having u n i t s with r e l a t i v e l y homogeneous housing and socioeconomic c h a r a c t e r i s t i c s , there i s l i k e l y to be considerable v a r i a t i o n , p a r t i c u l a r l y i n s t r u c t u r a l c h a r a c t e r i s t i c s . Such v a r i a t i o n and i t s inf luence on property values w i l l be hidden by the aggregation. Together with the omission of important v a r i a b l e s due to the noncomprehensive nature of i n d i v i d u a l housing data, the use of aggregate data may b ias parameter estimates considerably . For example, as Freeman (1979a, p.133) notes, census t r a c t data are u n l i k e l y to capture any changes i n the phys i ca l c h a r a c t e r i s t i c s of a house caused by no i se -avert ing behaviour (for example a i r -cond i t i on ing to permit window s e a l i n g ) , yet the hedonic p r i c e funct ion should include terms for these a t t r i b u t e s 60 (although appropriate var iables w i l l d i f f e r depending on the study l o c a l i t y ) . The a l t e r n a t i v e data type used i n the reviewed studies i s the i n d i v i d u a l market transact ion , be i t a sa le p r i c e for owner-occupied housing stock or a monthly r e n t a l for the r e n t a l market. This has the obvious advantages of being ac tua l and disaggregated data; the disadvantages are the costs and d i f f i c u l t y to c o l l e c t a large enough sample s ince market t ransac t ions involve only a small percentage of the owner-occupied housing stock i n any one year . The advantage of us ing sa les data for the dependent v a r i a b l e i s discussed i n more d e t a i l l a t e r . Another source of data for the dependent v a r i a b l e would be p r o f e s s i o n a l appra i sa l s of i n d i v i d u a l p r o p e r t i e s . Freeman (1979a, p.131) notes that as j u r i s d i c t i o n s develop computer-based systems conta in ing appraised values and a v a r i e t y of s i t e and s t r u c t u r a l c h a r a c t e r i s t i c s , these sets of information become valuable - but should be used with caut ion given the p o s s i b i l i t y of systematic b iases for p o l i t i c a l or other reasons in at l eas t some j u r i s d i c t i o n s . The s tudies reviewed general ly examine only s i n g l e - f a m i l y owner-occupied proper t i e s whether census or sa les data i s obtained. Except for Pr i ce (1974) who uses market rents (and t h e i r change over a decade), McDougall (1976) who uses 61 both owner-occupied housing values and apartment rents , and Maser et a l . (1977) who examine d i f f e r e n t property types ( i n c l u d i n g s ing le - fami ly and mul t ip le family r e s i d e n t i a l , i n d u s t r i a l and commercial), the studies ignore other segments of the r e s i d e n t i a l property market and other property types. No study considers vacant land. This neglect poss ib ly r e s u l t s from d i f f i c u l t y obtaining informat ion , but nonetheless may r e s u l t i n an incorrec t p o r t r a y a l of t o t a l community va lua t ion of noise impact i f other market segments d i f f e r . Screening Notwithstanding the above c r i t i c i s m regarding market segmentation, i t may be noted that the majority of the s tudies e x p l i c i t l y screen out p a r t i c u l a r observations for the owner-occupied housing sec tor . The object ive i s to avoid proper t i e s ( i n d i v i d u a l or i n census tracts ) for which there are "unusual" fac tors u n l i k e l y to be captured i n the explanatory v a r i a b l e s . For instance, Abelson (1979) excludes propert ies undergoing s i g n i f i c a n t land-use changes or extensive modi f icat ions; 0'Byrne et a l . (1985) for i n d i v i d u a l sales data exclude cases where there were unusual contrac t terms or l oca t ions near main roads; Maser et a l . (1977) exclude i n d i v i d u a l sa les i n v o l v i n g government agencies , intracorporate or family t rans fers as wel l as t ransac t ions not i n v o l v i n g any d o l l a r exchange; Nelson (1979, 1981a) and 0'Byrne et a l . (1985) for census t r a c t 62 data exclude t r a c t s near l o c a l environmental features , major t ransporta t ion f a c i l i t i e s , commercial development or other features , and (for reasons which are unclear , but r e f e r r e d to la ter ) t r a c t s located i n c e r t a i n pos i t ions r e l a t i v e to NEF contours. Other s tudies exclude census t r a c t s unless at l e a s t 50% of dwell ings are owner-occupied and s i n g l e - f a m i l y (Paik 1972), or unless at l eas t 25% are r e n t a l un i t s (Price 1974). I t may be noted that the greater the number of fac tors def ined as "unusual" and thus warranting exc lus ion , the less use fu l i s the model for exp la in ing noise e f fec t s on property v a l u a t i o n . By v i r t u e of i t s aggregated nature, t h i s makes models using only census data p a r t i c u l a r l y u n a t t r a c t i v e . Market s t r a t i f i c a t i o n The need for s t r a t i f i c a t i o n was considered e a r l i e r . Several of the studies segment the housing market: Abelson (1979) and Emerson (1972) by p r i c e range; Abelson (1979) and Mieszkowski and Saper (1978) by community; Maser et a l . (1977) by property type; Nelson (1979, 1981a) by dis tance from a i r p o r t ; 0'Byrne et a l . (1985) by percentage of owner-occupied housing for census data; and Nelson (1978a) and Maser et a l . (1977) on the bas i s of urban or suburban l o c a t i o n . Not a l l the s tudies t e s t for d i f ferences between the r e s u l t i n g market segments; Nelson (1978a, p.86) does and f inds no evidence of s i g n i f i c a n t d i f ferences i n the i m p l i c i t 63 p r i c e functions for suburban and urban areas i n Washington, DC. However, as Nelson (1978a, p.86) notes, the l i t e r a t u r e argues that the d u r a b i l i t y of housing stocks, the j o i n t nature of many housing services , and general i n e l a s t i c i t y of housing demand and supply tend to produce a segmented housing market, with p o t e n t i a l groupings according to housing c h a r a c t e r i s t i c s , neighborhood c h a r a c t e r i s t i c s , race, income, and so f o r t h . In a segmented market there i s a p o t e n t i a l for large d i f ferences i n the p r i c e funct ion of i n d i v i d u a l housing a t t r i b u t e s , and hence i n the o v e r a l l p r i c e per composite uni t of housing s e r v i c e s . Use of a marketwide or u n s t r a t i f i e d model might, therefore , lead to i n c o r r e c t inferences about the hedonic p r i c e equation. T h i s matter i s a lso discussed i n Freeman (1979a, pp 142-143) who notes the apparently s i g n i f i c a n t e f f ec t of s t r a t i f i c a t i o n by income, employment a c c e s s i b i l i t y and household s o c i a l status on i m p l i c i t p r i c e funct ions for an a i r p o l l u t i o n property value study undertaken by Harr i son and Rubenfeld (1978b). As H a r r i s (1981, p.46) concludes, "The existence of submarkets does not i t s e l f inva l ida te the hedonic technique. Rather i t makes the task of est imating i m p l i c i t p r i c e that much more d i f f i c u l t , e s p e c i a l l y i f no c l e a r c r i t e r i o n for s t r a t i f i c a t i o n can be i so la ted" . Various c r i t e r i a are s e n s i b l y used i n the s tud ie s . 64 Independent v a r i a b l e s A l i s t of the independent (explanatory) v a r i a b l e s considered i n the reviewed studies would be very long s ince often a minor v a r i a b l e proves to be s t a t i s t i c a l l y s i g n i f i c a n t only i n a s i n g l e study. In most cases there may be good reasons for the i n c l u s i o n / e x c l u s i o n of p a r t i c u l a r v a r i a b l e s , but i n d i v i d u a l j u s t i f i c a t i o n of every v a r i a b l e i s not attempted here , nor i s the precise form of each v a r i a b l e described i n most cases . In an e f for t to report the more "important" v a r i a b l e s , s t a t i s t i c a l l y s i g n i f i c a n t v a r i a b l e s from the "best f i t equations" are given i n Table 3.1 under the fo l l owing broad groupings of c h a r a c t e r i s t i c s (part ly a f t er Nelson 1981a, p .58): phys ica l (house and l o t ) , area (or neighbourhood), p u b l i c sector , a c c e s s i b i l i t y , and other c h a r a c t e r i s t i c s . Noise, an area c h a r a c t e r i s t i c , i s shown separate ly because of i t s p a r t i c u l a r relevance to t h i s study. Although i t i s l e f t to Appendix B to comprehensively report the s t a t i s t i c a l s i gn i f i cance or other aspects of v a r i a b l e s used i n the best f i t (or var ious other) equations for each study, general impressions are given here. Tables 3.2 to 3 .5 , for each category, l i s t "general ly s i g n i f i c a n t " v a r i a b l e s ; any commentary regarding v a r i a b l e s of p a r t i c u l a r i n t e r e s t because of t h e i r i n c l u s i o n i n the f i r s t p lace and/or t h e i r s i g n i f i c a n c e , fo l lows . Genera l ly s i g n i f i c a n t v a r i a b l e s are those which were s i g n i f i c a n t at the 95 percent l e v e l . 65 P h y s i c a l c h a r a c t e r i s t i c s TABLE 3.2 - PHYSICAL CHARACTERISTICS G e n e r a l l y s i g n i f i c a n t G e n e r a l l y not s i g n i f i c a n t . number (and type) o f rooms . house s i z e . house age . a i r - c o n d i t i o n i n g / h e a t i n g . number o f f i r e p l a c e s . e x t e r n a l c o n d i t i o n / g e n e r a l house maintenance . t y p e o f s t r u c t u r e . g a r a g e f a c i l i t i e s . l o t d i m e n s i o n , s i z e o r shape s p e c i a l f e a t u r e s (swimming p o o l s and p a t i o s ) r o o f t y p e u t i l i t i e s i n t e r i o r q u a l i t y ( c a r p e t s and drapes) t e r r a i n l a n d s c a p i n g s t r e e t p o s i t i o n The l i s t o f p h y s i c a l house and l o t c h a r a c t e r i s t i c s i s by f a r t h e l o n g e s t o f a l l the c a t e g o r i e s . I t may be r e c a l l e d t h a t t h o s e s t u d i e s r e l y i n g s o l e l y on census d a t a have f a r l e s s scope f o r the i n c l u s i o n o f e x p l a n a t o r y v a r i a b l e s , p a r t i c u l a r l y t h o s e i n v o l v i n g p h y s i c a l c h a r a c t e r i s t i c s . T h i s i s e v i d e n c e d by t h e f a c t t h a t (mean) number o f rooms i s t h e o n l y s i g n i f i c a n t p h y s i c a l c h a r a c t e r i s t i c f o r s e v e r a l s t u d i e s u s i n g census d a t a . In t h i s r e s p e c t , t h e census d a t a a r e much i n f e r i o r t o i n d i v i d u a l h o u s i n g d a t a o b t a i n e d i n o t h e r ways, because t h e y do not c a p t u r e v a r i a t i o n . N o n e t h e l e s s , c ensus d a t a a r e used a lmost u n i v e r s a l l y f o r some o f t h e a r e a c h a r a c t e r i s t i c s . 66 Area c h a r a c t e r i s t i c s TABLE 3.3 - AREA CHARACTERISTICS Genera l ly s i g n i f i c a n t General ly not s i g n i f i c a n t Population . percentage non-white . dens i ty . age L i v i n g condit ions . housing densi ty . length of occupancy . percentage of populat ion i n m u l t i p l e occupancy bui ld ings . percentage of owner-occupation Amenity . prox imity to recreat ion . freeway proximity f a c i l i t i e s (water, park) . amount of road t r a f f i c . school proximity . amount of p u b l i c housing . adjacent land use Area c h a r a c t e r i s t i c s include s o c i a l or p h y s i c a l features of the neighbourhood or community i n which a property i s l o c a t e d . These include populat ion c h a r a c t e r i s t i c s (obtained from census data) , ind ica tors of l i v i n g condi t ions and v a r i a b l e s measuring amenity. One study (McDougall 1976) t e s t s a very e laborate "neighbourhood q u a l i t y and land use c h a r a c t e r i s t i c s sca lar" r e f l e c t i n g landscaping, extent of l i t t e r , number of vacant l o t s , s t r e e t condi t ions , convenience f a c t o r s , land use, l o t 67 s i z e , dwel l ing s i z e and house s tructures - but found i t i n s i g n i f i c a n t , probably because i t captures so much. The amenity of quiet i s a lso tested i n each study, and genera l ly found to be s i g n i f i c a n t as fur ther discussed below. P u b l i c sector c h a r a c t e r i s t i c s TABLE 3.4 - PUBLIC SECTOR CHARACTERISTICS General ly s i g n i f i c a n t General ly not s i g n i f i c a n t . school r e s u l t s . pub l i c education expenditure . pub l i c t r a n s i t s erv ice There i s some overlap between area and p u b l i c sector c h a r a c t e r i s t i c s , but the l a t t e r i s defined to inc lude the q u a l i t y and quant i ty of major publ i c s erv ices to which a property owner i s e n t i t l e d . The models revealed mixed r e s u l t s for the t e s t i n g of property tax rate v a r i a b l e s . A c c e s s i b i l i t y c h a r a c t e r i s t i c s Again , t h i s category overlaps with others , and some proximity c h a r a c t e r i s t i c s have already been noted. A c c e s s i b i l i t y , here, i s concerned with access to economic 68 agents with whom transact ions can be made by house owners. Resul ts were mixed for a i r p o r t access. TABLE 3.5 - ACCESSIBILITY CHARACTERISTICS Genera l ly s i g n i f i c a n t General ly not s i g n i f i c a n t . access to CBD . time required to reach 75 . access to commercial corner percent of metropol i tan employment opportuni t ies T h i s leads to the concern expressed i n the l i t e r a t u r e about the importance of major a i r p o r t s as centres of economic a c t i v i t y ( p a r t i c u l a r l y as employment generators) , such that r e s i d e n t i a l property values ought to d e c l i n e , c e t e r i s  p a r i b u s , as a i r p o r t a c c e s s i b i l i t y f a l l s and commuting costs increase . Account needs to be taken of the two c o u n t e r v a i l i n g e f fects exerted on adjacent property values by a i r p o r t s : deprec iat ion due to noise and apprec ia t ion due to employment a c c e s s i b i l i t y . As Nelson (1981a, pp 65-6) s ta te s , "Empir ica l studies which ignore a i r p o r t a c c e s s i b i l i t y may r e s u l t i n noise exposure c o e f f i c i e n t s that are b iased toward zero, because of p o s i t i v e c o r r e l a t i o n s between noise exposure and a i r p o r t a c c e s s i b i l i t y . " Blaylock (1977) a l so stresses the need to consider the poss ib le interdependence of a i r p o r t - r e l a t e d economic a c t i v i t y and a i r p o r t noise . 69 Nelson (1979, 1981a) seeks to address t h i s issue a f ter not ing the problems i n De Vany's (1976) approach. De Vany p a r t i t i o n s census blocks in to distances from the a i r p o r t so as to use dummy var iab les for c i r c u l a r distance bands. The c o e f f i c i e n t s then r e f l e c t the net impact of the a i r p o r t on property values and should t race out the net renta l g r a d i e n t . Because a i r c r a f t noise contours are elongated, Nelson fee l s that use of c i r c u l a r bands w i l l mi sc la s s i fy many b locks , causing many measurement errors - i f the o b j e c t i v e i s noise discount measurement alone. Nelson's approach i s to use a small study area with a radius of two miles which contains a range of NEF values but as far as p o s s i b l e maintains constant a c c e s s i b i l i t y to a i r p o r t s , CBD, e t c . , as wel l as constant p u b l i c sector c h a r a c t e r i s t i c v a r i a b l e s . Concerned with the p o s s i b i l i t y of s p e c i f i c a t i o n b i a s , Nelson tes t s the robustness of the c o e f f i c i e n t s by r e s t r i c t i n g the sample s izes fur ther and f inds l i t t l e change. He a lso pools the data for the s i x separate a i r p o r t s i n i t i a l l y s tudied, and runs severa l s t a t i s t i c a l t e s t s which a l l suggest that h i s NEF r e s u l t s are r e l a t i v e l y unaffected by a c c e s s i b i l i t y inf luences given the steps he took to avoid t h i s d i f f erence . Noise As s ta ted i n Chapter 2, the Noise Exposure Forecast (NEF) i s the measure of a i r p o r t noise t r a d i t i o n a l l y used i n North 70 America, and in one form or another i s used i n nine of the s tudies reviewed. The r e l a t i v e merits of numerous measures and approaches are wel l debated i n the l i t e r a t u r e and comprehensively surveyed i n Schultz (1982). Two studies use both NEF and Composite Noise Rating (CNR) data (Dygert 1973 and Mieszkowski and Saper 1978), while one employs only CNR data (Emerson 1972). The majori ty of s tudies use a l i n e a r or log form of NEF as a continuous v a r i a b l e graduated i n f ive or ten dec ibe l increments, or i n the case of 0'Byrne et a l . (1978) i n d i v i d u a l data model, (the current ly pre ferred noise d e s c r i p t o r i n the United States , as noted i n Chapter 2) . Only P r i c e (1974) and perhaps Abelson (1979) use a continuous v a r i a b l e graduated i n one dec ibe l increments. Several e a r l i e r s tudies use a logari thmic transformation to convert a noise index in to i t s sound i n t e n s i t y or energy-equivalence l e v e l , the economic meaning of which i s not c l e a r (Nelson 1980, p .41) . In other approaches, De Vany (1976) represents noise by distance dummies, McDougall (1976) merely uses a weighted f igure of the percentage of a planning area exposed to 90dB or more and Maser et a l . (1977) use a dummy v a r i a b l e for being wi th in the lOOdB contour. General ly the s tudies use noise data contemporary to other data, but i n cases where t h i s i s not so they exclude observations which are adjacent to boundaries and therefore may have crossed contours over time (for example, 71 Nelson 1979, 0'Byrne et a l . 1985, Paik 1972). I t i s not apparent what i s gained by such screening when non-adjacent observations w i l l a lso have s h i f t e d . I n t u i t i v e l y , more cons i s tent noise c o e f f i c i e n t s would be obtained i f a l l observations are included - p a r t i c u l a r l y i f NEF contours have sh i f t ed s t e a d i l y . NEF values are observed to be i n the range 15 to 55 with evidence that there i s " l i t t l e " annoyance i n the range 15 to 25, "some" to "much" annoyance i n the 25 to 40 range and "considerable" annoyance above 40 (Nelson 1980, p .41) . Many s tudies assume a threshold noise l e v e l below which the ambient noise l e v e l does not f a l l - for example NEF 20 for McMil lan et a l . (1980), Nelson (1978a), Paik (1972); CNR 101 (approximately equal to NEF 30) for Emerson (1972). Nelson (1979, 1981a) usua l ly uses a threshold of NEF 25, although he uses NEF 2 0 where there i s l i m i t e d r e s i d e n t i a l noise impact. The ass igning of a zero value to c o n t r o l areas, as i s done i n severa l s tudies (for example, Abelson 1979, Dygert 1973), creates measurement e r r o r s given the e f f e c t i v e o r i g i n of the noise index at about NEF 20 (Nelson 1980, p .41 ) . The empir i ca l question should not be the absence/presence of a i r c r a f t noise , but the d i f f e r e n t i a l noise l e v e l . Turning to the r e s u l t s for the noise v a r i a b l e s , they are a l l s i g n i f i c a n t i n one form or another i n a l l the s tudies , 72 although only at the 90 percent l e v e l in the case of Emerson (1972) and Nelson (1978a). Other explanatory v a r i a b l e s The other var iab le s considered i n some studies include those r e l a t i n g to the sale arrangements, f inancing arrangements and the buyer's c h a r a c t e r i s t i c s . For sa le arrangements, mortgage c h a r a c t e r i s t i c s such as low in teres t were found to be s i g n i f i c a n t i n one study and not s i g n i f i c a n t i n one o ther . For f inancing arrangements, a trustee sa le , assessment value and rent were each found s i g n i f i c a n t i n one study. Buyer c h a r a c t e r i s t i c s were only contemplated i n one study and even then were not t e s t ed . Income, for instance, should not be included because the hedonic p r i c e model expla ins property value i n terms of i t s own c h a r a c t e r i s t i c s , not the c h a r a c t e r i s t i c s of the purchaser / renter . The use of a census median income f igure as a proxy for socioeconomic c h a r a c t e r i s t i c s of the area or neighbourhood may, however, be j u s t i f i e d according to Freeman (1979a, p.139). As e a r l i e r noted, income may a l so be used as an a l t e r n a t i v e to house p r i c e for s t r a t i f y i n g the sample. Other v a r i a b l e s include i n f l a t i o n , represented by a time dummy v a r i a b l e , which was s i g n i f i c a n t i n several s tud ie s . C i t y dummy var iab le s were used for s tudies invo lv ing pooled 73 data (Nelson 1979, 1981a, and Paik 1972 when re-modelled by Nelson 1978a). Study r e s u l t s Reference has already been made to the s i g n i f i c a n c e of the v a r i a b l e s included i n the various models i n the s tudies reviewed. Almost without exception the c o e f f i c i e n t s for these v a r i a b l e s have the expected s igns . The publ ished s tudies genera l ly devote l i t t l e a t t ent ion to explorat ion of the model b u i l d i n g process and the ways i n which problems such as m u l t i c o l l i n e a r i t y , o u t l i e r s and he teroscedas t i c i ty are t e s t ed for (although a notable exception i s Nelson 1978a; see a l so Abelson 1979, and Mieszkowski and Saper 1978) . Although the study with the larges t sample s i ze (De Vany 1976) uses census property value data, t h i s i s the exception to the trend of r e l a t i v e l y l arger sample s i z e s for s tudies us ing i n d i v i d u a l house sales data . This i s but one of many immediately obvious d i f ferences between s tudies using the census and those using sales data . As e a r l i e r noted, there i s a major d i s t i n c t i o n between s tudies us ing the census data which tend to favour use of more area c h a r a c t e r i s t i c s , and i n d i v i d u a l sa les studies which favour p h y s i c a l c h a r a c t e r i s t i c s . O v e r a l l , the census models inc lude l e ss v a r i a b l e s . 74 Another d i f ference between the two data c lasses i s the s i ze of the c o e f f i c i e n t s of mul t ip l e determination, R 2 , which measures the proport ionate reduct ion of t o t a l v a r i a t i o n i n house p r i c e associated with the explanatory v a r i a b l e s . Although a large R2 does not necessar i ly imply that the f i t t e d model i s u s e f u l , i t i s a useful summary s t a t i s t i c (although adjusted R2 i s t h e o r e t i c a l l y bet ter ; i t appears to be reported for only one of the s tud ies ) . On t h i s b a s i s , the census value models appear to general ly perform be t ter than i n d i v i d u a l sales p r i c e models - although the range of R2 s t a t i s t i c s reported i s .60 to about .90 for both model types . (It may be noted that such a range i s cons i s tent with r e s u l t s of s tudies other than those concerned with a i r p o r t noise , for example Weicher and H a r t z e l l ' s (1982) study of 59 Metropol i tan Areas i n the United Sta tes . ) The higher R2 cannot be explained by the smaller sample s i z e s , or smal ler number of independent v a r i a b l e s , because, c e t e r i s  par ibus . e i t h e r of these would be expected genera l ly to lead to a lower R 2 . A p o s s i b l e explanation i s that the v a r i a t i o n contained between census t r a c t or block data i s l e s s than that for i n d i v i d u a l data , even though t h i s hides v a r i a t i o n w i t h i n any t r a c t or b lock . The standard dev ia t ion w i t h i n t r a c t s or blocks i s l i k e l y to be greater than that between them. Another poss ib le explanat ion for the apparent d i f f e r e n c e s i n R2 l i e s i n the wide v a r i e t y of s p e c i f i c a t i o n s , data bases 75 and modell ing s o p h i s t i c a t i o n they d i s p l a y . Harr i s (1981, p .45), f o r example, b l unt ly states that "Even the broad consistency of r e s u l t s may of course be spurious i f a l l s tudies have f a i l e d to take into account important c h a r a c t e r i s t i c s of the proper t i e s , households, or neighborhoods." Model s p e c i f i c a t i o n must not be on the bas i s of mere convenience and data a v a i l a b i l i t y , for the exact s p e c i f i c a t i o n may grea t ly inf luence the importance of p a r t i c u l a r hedonic p r i c e s . Abelson (1979) provides one example where the a i r c r a f t noise c o e f f i c i e n t i n h i s model increased by some 60 percent with the dropping of t r a f f i c noise and e x t e r i o r condi t ion v a r i a b l e s . But ler (1982, pp 106-7) f inds that the p r a c t i c a l impact of s p e c i f i c a t i o n biases i s small for s tructure (physical) c o e f f i c i e n t s , but suggests t h i s i s not neces sar i ly the case for non-s tructure c h a r a c t e r i s t i c s which are t y p i c a l l y h igh ly i n t e r c o r r e l a t e d . Despite concern over d i f f e r i n g R2 va lues , there i s evidence that the weaker and the stronger models a l l produce s i m i l a r a i r p o r t no i se /property value r e l a t i o n s h i p s . This i s provided by the standard means of summarizing t h i s r e l a t i o n s h i p . Noise-property value r e l a t i o n s h i p s Using the regress ion models for the var ious s tud ies , the o v e r a l l f ind ings of each can be summarized using the Noise Deprec ia t ion S e n s i t i v i t y Index (see Walters 1975, pp 102-5, 76 Nelson 1978a, pp 99-101). Fol lowing Nelson (1978a), the absolute amount of depreciat ion per dec ibe l (or the p r i c e of quiet) for any two propert ies which are i d e n t i c a l except for noise exposure l eve l s i s defined as: D = d i f f erence i n t o t a l noise discount d i f f erence i n noise exposure (3.14) The percentage rate of deprec ia t ion , or the noise d e p r e c i a t i o n s e n s i t i v i t y index (NDSI) i s defined as: NDSI = (D/property value).100 = d i f f erence i n t o t a l percentage deprec ia t ion di f ference i n noise exposure (3.15) where i t i s assumed that d e f l a t i n g by the average property value f u l l y accounts for d i f ferences i n absolute p r i c e l e v e l s . In other words, NDSI "is a measure of the noise s e n s i t i v i t y of the housing market expressed i n terms of the marginal ra te of deprec ia t ion per dec ibe l over some i n t e r v a l of noise exposure" (Nelson 1978a, p.100). I f the func t iona l form of noise i s l i n e a r against the na tura l log of property value , the NDSI w i l l be constant regardless of the absolute p r i c e of the proper ty . Nelson (1978, p.100) suggests that t h i s i s 77 not incons is tent with a un i tary permanent income e l a s t i c i t y of demand for quiet - which may be v a l i d given the o v e r a l l u n i t y income e l a s t i c i t y of demand for owner-occupied housing (de Leeuw 1971, p .10) . A f t e r making adjustments as appropriate i n each study (such as for funct iona l form, noise l e v e l s , noise d e s c r i p t o r s ) , Nelson (1980) uses the NDSI to summarize the estimated noise -property value r e l a t i o n s h i p s . For nine of the twelve s tudies he reviewed, the range i s about 0.40 to 1.10 percent per dec ibe l for 1967 to 1976 data , or a simple average of about 0.61 percent. Table 3.1 reports these values as we l l as NDSI f igures for the more recent s tudies . Excluded from Nelson's l i s t i s McDougall's (1978) study for which r e s u l t s do not permit c a l c u l a t i o n of an index. Also excluded i s P a i k ' s (1972) study which used 1960 census data to obta in an NDSI of about 2.00 percent per dec ibe l - explained as being so high due to short -run d i s e q u i l i b r i u m i n r e s i d e n t i a l housing markets i n 1960, at the beginning of commercial j e t t r a v e l i n the United States (Nelson 1980, p.44, and 1978a, pp 118-20). Consistency The s tudies reviewed range g r e a t l y i n the s o p h i s t i c a t i o n of t h e i r model l ing: the model func t iona l forms d i f f e r , sample s i z e s range from very small to r e l a t i v e l y large , standard e r r o r s are often h igh , the account taken of s p e c i f i c a t i o n 78 issues i s commonly minimal, and so on. P a r t i c u l a r l a t e r s tudies which appear to be somewhat more thorough (Abelson 1979, Nelson 1978a, Nelson 1979 and 1981a, O'Byrne et a l . 1985) b u i l d on concerns with e a r l i e r s tud ies . Nonetheless, r e s u l t s are i n the same range even for d i f f e r e n t communities and a i r p o r t s , and over t ime. Evidence i s provided by Nelson (1981a) when he addresses the i ssue of the s t a b i l i t y - or homogeneity - of the NEF c o e f f i c i e n t s for h i s s i x study areas to e s t a b l i s h i f each sample i s in some sense unique, rather than being par t of a general market for q u i e t . He f inds that the r e s u l t s are cons i s tent with a s tab le slope for the noise-property value r e l a t i o n s h i p . Cons iderat ion of t h i s issue i s only pos s ib l e when more than one study area i s invest igated - and, although other noise s tudies deal with mul t ip le areas, none of them addresses t h i s i s sue . Studies of a i r p o l l u t i o n have found a consistent r e l a t i o n s h i p across s tudies (Harris 1981, p .45) . But ler (1980, p.451) reports subs tant ia l homogeneity across markets i n d i f f e r e n t metropol i tan areas i n the United States , although not ing that the r e n t a l and ownership markets are subject to d i f f e r e n t inf luences with the l a t t e r having small but s i g n i f i c a n t d i f ferences across markets. (See a l so Weicher and H a r t z e l l 1982.) Further evidence i s provided by O'Byrne et a l . (1985) i n the only study to compare noise discounts obtained us ing the 79 hedonic p r i c e model over time and for i n d i v i d u a l house sa les verses owner-appraised census block aggregate data for the same a i r p o r t . The re su l t s are reported i n Table 3.1 and revea l s i m i l a r estimates across these d i f ferences and when compared to the e a r l i e r studies reviewed. 0 1 Byrne et a l . (1985, p.170) conclude that "biases i n the noise-discount estimate due to aggregation or d i s e q u i l i b r i u m are not obvious, contrary to the general c r i t i c i s m s of housing p r i c e s t u d i e s . . . " Further evidence supporting the r e s u l t s obtained i n the s tudies reviewed i s provided by Frankel (1988a, 1988b) who uses a survey of Realtors and appraisers to determine a i r p o r t noise impact on property va lues . Although a c t u a l l y i n v o l v i n g a survey, such an approach i s s t i l l an i n d i r e c t way of va lu ing noise impact i n that i t i s concerned with observed behaviour r e f l e c t e d i n property values (as reported by agents) . I t i s not a d i r e c t survey approach of the k ind r e f e r r e d to i n the in troduct ion of the l i t e r a t u r e review. A p r i n c i p a l concern of Franke l ' s study i s to assess the e f f ec t of a i r p o r t noise on r e s i d e n t i a l property values i n some f o r t y suburban communities surrounding Chicago's O'Hare A i r p o r t . F indings re levant to t h i s paper inc lude: . the secondary importance of a i r c r a f t noise compared to area amenity, p u b l i c sector and a c c e s s i b i l i t y 80 c h a r a c t e r i s t i c s in communities exposed to moderate or greater a i r p o r t noise (Frankel 1988a, pp 4-8) . . the nature of the market for no i se -a f fec ted r e s i d e n t i a l propert ies where supply i s strengthened, demand weakened and movement slowed i s such that p r i c e weaknesses and p r i c e discounts e x i s t . I n s u f f i c i e n t information among buyers i s seen to contr ibute most d i r e c t l y to i n e f f i c i e n t market outcomes (Frankel 1988a, pp 9-14). . three sets of r e s u l t s based on respondents' general experience (Survey 1), based on neighbourhood-specif ic knowledge (Survey 2) and based on F r a n k e l ' s r e c l a s s i f i c a t i o n of neighbourhoods (Survey 3) , are a l l i n general agreement with the re su l t s of the reviewed hedonic p r i c e models (Frankel 1988a, pp 15-34) . NDSI values c a l c u l a t e d from Real tors ' estimates of no i se -r e l a t e d reduct ions i n the values of s i n g l e - f a m i l y dwel l ings , on the bas i s of a 60 Lj , (approximately NEF 25) noise thresho ld , are (ranging from the lowest to the highest estimates from the three surveys) : .52 to 1.03 for moderate noise l eve l s (65-70 L ^ ) , .77 to 1.04 for s u b s t a n t i a l noise l eve l s (70-75 L ^ ) , and .64 to 1.23 for severe noise l eve l s (75-80 L^) . 81 . estimates of a i r p o r t noise impact are c o n s i s t e n t l y lower for m u l t i - f a m i l y dwell ings than those for s i n g l e -family dwellings and appraisers estimates are cons i s t en t ly lower (on average by 30%) than those by R e a l t o r s . . Real tors and appraisers personal ly exposed to a i r c r a f t noise i n t h e i r own homes are found to g ive sys temat ica l ly higher estimates of property value reduct ion . Conclusions The hedonic method i s i n t u i t i v e l y appeal ing. I t i s a l so r e l a t i v e l y s tra ightforward since i t uses a commonly understood s t a t i s t i c a l technique, regress ion a n a l y s i s . Although the hedonic method has some l i m i t a t i o n s , these l i m i t a t i o n s can be taken into account. Despite concerns r a i s e d i n the l i t e r a t u r e about the homogeneity and s t a b i l i t y of the r e s u l t s of previous studies (Harris 1981, MacLennan 1977, Pearce 1978), the above review of t h i r t e e n s tudies shows that they provide broadly consistent r e s u l t s . 3.4 HYBRID PROPERTY PRICE APPROACHES Pearce (1978, pp 42-7) c l a s s i f i e s the approach of the Research Team of (and of other p a r t i e s assoc iated with) the 82 Commission on the T h i r d London A i r p o r t (CTLA) as one of the h y b r i d property p r i c e approaches. I t i s ' h y b r i d ' i n the sense that i t does not assume a process of continuous adjustment as i n the hedonic a t t r i b u t e models and because, i n p r a c t i c e , mul t ip le regress ion techniques could not be used and resort was had to a d i f f e r e n t method of est imating p r i c e deprec ia t ion . Further , whereas the hedonic models nowhere speak of consumers' surplus - the excess of the i n d i v i d u a l ' s subject ive va luat ion of a house over and above i t s market p r i c e - the d i s c r e t e adjustment procedure assumed i n the CTLA study requ ires an estimate of t h i s surp lus . (Pearce 1978, p.42) The CTLA approach i s described i n CTLA (1970, Appendices 22 and 23) and i n Flowerdew (1972), and i s discussed i n Dasgupta and Pearce (1972, Chapter 9), OECD (1975, Annex A ) , Paul (1971), Pearce (1971, 1976), S tark ie and Johnson (1975, Chapter 4) and Walters (1975). The theme of the approach i s (as i n the approach reviewed thus far) that noise disamenity i s r e f l e c t e d i n property p r i c e s . The methodology involves c a t e g o r i z i n g res idents affected by a i r p o r t noise . The o r i g i n a l set of CTLA categories of non-movers, movers for noise reasons and movers for other reasons has s ince been expanded by Pearce (1978) and others . Abelson (1977), for example, i d e n t i f i e s costs of a i r c r a f t noise a f f e c t i n g f i v e r e s i d e n t i a l groups: natura l out-movers, noise out-movers, s tayer , informed in-movers, uninformed in-movers. Using Abelson's notat ion , the hybr id approach requires es t imat ion of several var iab le s i n c l u d i n g : N, the 83 c a p i t a l i z e d a i r c r a f t noise cost; D, the c a p i t a l i z e d house p r i c e deprec ia t ion due to a i r c r a f t noise; M, household movement c o s t s ; A s , change i n householder surp lus ; and A R, the increased household turnover ( requ ir ing information on the number of households affected and the rate of natural turnover ) . C a l c u l a t i n g values for M may be r e l a t i v e l y s tra ight forward ( i f one ignores psychic costs of movement), but S requires survey methods. Hedonic or other approaches may be used to obtain a value for D. Est imat ion of N, however, i s by f a r the most d i f f i c u l t matter. Abelson (1977, pp 359-60) notes three a l t e r n a t i v e est imation techniques: p r i o r argument (as i n the CTLA), survey (as attempted by Abelson) , or f i t t i n g a noise model to turnover data (apparently never attempted). Abelson completes h i s noise cost measurement by i d e n t i f y i n g taxpayer costs , v i s i t o r costs and business costs . Although i n v o l v i n g impressive theory - a l b e i t c r i t i c i z e d ( for example, Pearce 1971, pp 109-14) - t h i s h y b r i d approach i s beyond the scope of t h i s study: t h i s paper i s in teres ted only i n obta in ing one g e n e r a l l y - a p p l i c a b l e r e l a t i o n s h i p for the impact of a i r p o r t noise on property values - as best done us ing the hedonic method. The same may be sa id about another development of the CTLA approach, the exc lus ion f a c i l i t i e s approach proposed by Stark ie and Johnson (1975, Chapter 5) . T h i s involves modi f icat ion to al low for i n s u l a t i o n of houses, although there are d i f f i c u l t i e s with t h i s method's a p p l i c a b i l i t y (Pearce 1978, pp 47-9) . 84 CHAPTER 4 - MODEL DESIGN 4 .1 INTRODUCTION The o b j e c t i v e of t h i s study was e a r l i e r i d e n t i f i e d as b e i n g t o determine the impact of n o i s e a s s o c i a t e d w i t h Vancouver I n t e r n a t i o n a l A i r p o r t on the v a l u e of nearby p r o p e r t i e s . In d e s i g n i n g the hedonic p r i c e model used f o r t h i s purpose, account i s taken of the many i s s u e s a r i s i n g from the d i s c u s s i o n o f theory and from the l i t e r a t u r e review. The f o l l o w i n g d e s c r i p t i o n of the approach c o n s i d e r s key a s p e c t s o f t h e model. 4.2 USE OF THE HEDONIC APPROACH The i n d i r e c t p r o p e r t y v a l u e approach i s adopted. A l t h o u g h t h e d i r e c t survey approach has m e r i t (as d i s c u s s e d e a r l i e r ) , t h e use of an ex post approach r e l y i n g on a c t u a l p r o p e r t y d a t a t o r e v e a l v a l u e s i s c o n s i d e r e d t o produce more r e l i a b l e e s t i m a t e s o f the a c t u a l impact. Although the adopted approach i s not i n e x p e n s i v e , t h e use o f a survey approach would almost c e r t a i n l y be more c o s t l y . The survey approach c o u l d a l s o be open t o more charges of b i a s , p a r t i c u l a r l y as t h e i s s u e o f a i r p o r t n o i s e r e l a t e s t o t h e p r o p o s a l by T r a n s p o r t Canada t o c o n s t r u c t a t h i r d runway a t Vancouver 85 In ternat iona l A i r p o r t , an issue which, as noted i n Chapter 2, has been subject to much p u b l i c debate i n the past . For t h i s paper a simple hedonic p r i c e model of the form o u t l i n e d i n the l i t e r a t u r e review, a f t e r Nelson (1981a), i s used. The f i n a l prec i se funct iona l form i s discussed l a t e r , but for now note the fol lowing model based on that der ived i n d e t a i l i n Chapter 3: where In H = the log of property va lue , d 0 = a constant term, d x = noise c o e f f i c i e n t , di = non-noise c o e f f i c i e n t s , Xi = property c h a r a c t e r i s t i c s other than noise p o l l u t i o n , NEF = the measure of noise p o l l u t i o n , u 3 = the e r r o r term. 4.3 STUDY AREAS The p r i n c i p a l geographical area of concern i s the Township of Richmond, located to the south and east of Vancouver I n t e r n a t i o n a l A i r p o r t . Figure 2.4 showed Richmond which at the study date was inhabited by some 108 500 people i n about n (4.1) 86 38 000 household un i t s ( S t a t i s t i c s Canada 1988a). Richmond, formerly predominantly a g r i c u l t u r a l land, has experienced steady r e s i d e n t i a l development, with the p o p u l a r i t y of condominiums being a p a r t i c u l a r c h a r a c t e r i s t i c of the area. Figure 2.4 i l l u s t r a t e d the area exposed to a i r p o r t noise . Ins ide the noise zone, defined as propert i e s on or ins ide the NEF 25 contour, over 16 000 r e s i d e n t i a l proper t i e s were a f f ec t ed . As noted l a t e r , the study omits the Steveston area i n the south-west corner of Richmond. Part of the C i t y of Vancouver i s a lso w i th in the NEF 25 contour. Vancouver contains an older and more e s tab l i shed housing stock with v i r t u a l l y no vacant l and , u n l i k e Richmond. In the area defined as being exposed to a i r p o r t noise there are about almost 8000 r e s i d e n t i a l proper t i e s a f f ec ted by noise l e v e l s of NEF 25 or h igher , with two-t h i r d s of these being owner-occupied condominiums and r e n t a l b u i l d i n g s . For reasons noted l a t e r , t h i s study focusses on Richmond. 4.4 DATA SOURCES Dependent v a r i a b l e The l i t e r a t u r e review discusses two broad data categor ies for the dependent v a r i a b l e - census data and ac tua l 87 i n d i v i d u a l property sales data . For the reasons mentioned e a r l i e r , the i n d i v i d u a l data are general ly considered s u p e r i o r . For t h i s study i n d i v i d u a l data were a v a i l a b l e for the e n t i r e number of sales for the per iod January 1987 to June 1989. Census data from the 1986 Census of Population are a v a i l a b l e for comparative purposes. Data for the average value of owner-occupied pr iva te dwel l ing (defined as the amount expected by the owner i f the dwel l ing were to be sold) are der ived from a 20% sample of households. In the study area , there are few census t r a c t s : the Richmond study area , for ins tance , comprises j u s t twenty census t r a c t s . Census data were obtained for much smal ler geographical areas, enumeration areas (which comprise a maximum of 375 households) , boosting the sample s i ze to 97 i n the case of the Richmond study area (not inc lud ing enumeration areas e x c l u s i v e l y for h o s p i t a l s , hote l s and the l i k e ) . T h i s lower l e v e l of aggregation i s expected to mask less of the v a r i a t i o n (Goodman 1977, p.487) . Independent v a r i a b l e s The l i t e r a t u r e review categorized independent or explanatory v a r i a b l e s in to several broad groupings. Given the s i g n i f i c a n c e of v a r i a b l e s i n each, i n i t i a l l y a wide range of v a r i a b l e s were se lected which were genera l ly found to be s i g n i f i c a n t i n e a r l i e r s tudies and/or which a p r i o r i appear 88 relevant for t h i s study. I n i t i a l l y , no sens ib le , a v a i l a b l e v a r i a b l e was excluded from cons iderat ion . Thus, tes ted v a r i a b l e s inc lude some which proved i n s i g n i f i c a n t i n e a r l i e r s tudies (such as c e r t a i n a c c e s s i b i l i t y c h a r a c t e r i s t i c s ) , but exclude some v a r i a b l e s which are s i g n i f i c a n t i n other s tudies (such as % black populat ion) . Phys i ca l c h a r a c t e r i s t i c s To obtain the p h y s i c a l c h a r a c t e r i s t i c s of i n d i v i d u a l proper t i e s previous s tudies have e i ther used in tens ive f i e l d research or have taken information from sales documentation such as mul t ip l e l i s t i n g serv ice (MLS) sheets. Where s tudies are concerned with observations for census t r a c t s , very l i m i t e d census data on phys ica l c h a r a c t e r i s t i c s have been used. For t h i s study, the primary data source was the B r i t i s h Columbia Assessment A u t h o r i t y ' s (BCAA) inventory of a l l r e s i d e n t i a l property records (under the s t i p u l a t i o n that i n d i v i d u a l property data not be re leased) . These records provide numerous i n d i v i d u a l r e s i d e n t i a l property c h a r a c t e r i s t i c s . Table 4.1 l i s t s items of information a v a i l a b l e from BCAA records inc lud ing a l l those p h y s i c a l c h a r a c t e r i s t i c s se lected for cons iderat ion i n any of the models reviewed (these are i d e n t i f i e d by an a s t e r i s k ) . The Assessment Author i ty data are contemporary to the data c o l l e c t i o n for t h i s study (August to September 1989). In p r a c t i c e the assessments may not have been a l t e r e d for up to 89 TABLE 4.1 - BCAA DATA . land use, ac tua l use, *tenure . l o t s ize* - width, depth, area . zoning* . s erv ices - water, hydro, w e l l , roads, sewers, curbing and g u t t e r i n g , underground conduit , a l l e y * , sidewalk . other l o t c h a r a c t e r i s t i c s - common beach logged, double frontage, cu l -de-sac , corner loca t ion* , waterfront , bordering waterfront, wooded, r e s t r i c t e d use, panhandle, p ie shape, *extreme shape, f lood ing , subdiv idable , standard l o t , view*, water l o t * , s t r a t i f i e d land . topography* - undulat ing, bank, she l f , s lope, above road, below road, rock extrus ion , d i f f i c u l t access . improvement - nontypical improvement, under, over, l e g a l nonconforming, poor u t i l i z a t i o n , no o f f - s t r e e t park ing , exce l l en t landscaping* . s t r a t a t i t l e (condominium) - workshop, games room, sauna, poo l , playground . swimming pool* - types . outbui ld ings - various c h a r a c t e r i s t i c s . roof type* . dormers . e x t e r i o r wal l s* . veneers . chimneys . e l e c t r i c a l . i n t e r i o r f i n i s h . f l o o r s . i n t e r i o r features . f i r e p l a c e s - type*, number*, q u a l i t y . windows . plumbing - number*, type* of bathrooms . cabinets . basement - presence*, f i n i s h , f l o o r covering . heat ing type* . porches and decks* . yard improvements . rooms - type, number* . a t t i c . s t ruc ture age* . s t ruc ture remodell ing* . number of s t o r i e s * . carport* - type, q u a l i t y . garage* - type, number, e l e c t r i c door, q u a l i t y . i n s u l a t i o n . f u n c t i o n a l p lan . i n t e r i o r condi t ion* . e x t e r i o r maintenance* 90 severa l years (the Au th or i t y ' s object ive i s to make a p h y s i c a l inspect ion every f ive years ) , although i f there has been any sales t ransac t ion , phys ica l inspect ion w i l l g e n e r a l l y be made and changes noted wi th in twelve months. There i s a s l i g h t r i s k that the assessment d e t a i l s r e f e r to the property i n i t s s tate before changes re levant for the 1987 sa l e occurred, but there i s more r i s k that the contemporary d e t a i l s re fer to changes subsequent to the 1987 sa le (as l a t e r noted, only 1987 sales are tested i n t h i s s tudy) . The p r i n c i p a l d i f f i c u l t y with these data i s that they are i r r e l e v a n t for propert ies demolished fo l lowing t h e i r p a r t i c u l a r sale i n 1987. Such sa les were dropped from the sample, although a secondary data source, MLS s t a t i s t i c s ( a v a i l a b l e for a major proport ion of the r e s i d e n t i a l marketplace) were sometimes used for the phys i ca l c h a r a c t e r i s t i c s for such sa les . The Assessment Author i ty data are usefu l for subsequent sales (during 1987 and l a t e r ) i n v o l v i n g propert ies on the same l o t s . The 1986 Census of Population y i e l d s aggregate information at the enumeration area l e v e l on the average dwel l ing va lue , s t r u c t u r a l type of dwel l ing , tenure, average number of rooms, per iod of cons truct ion , presence of c e n t r a l heating and p r i n c i p a l heating f u e l . These c h a r a c t e r i s t i c s , a l l of which have been se lected for cons iderat ion i n previous models, are much more l i m i t e d than those a v a i l a b l e from the Assessment A u t h o r i t y , and were a l l the more so at the time 91 of t h i s study because of the u n a v a i l a b i l i t y of data for average dwell ing value and average number of rooms. Area c h a r a c t e r i s t i c s For area , p u b l i c sector and a c c e s s i b i l i t y c h a r a c t e r i s t i c s , f i e l d research could be undertaken to e s t a b l i s h many v a r i a b l e s by observat ion, c a l c u l a t i o n or quest ioning of occupants. Th i s would be a massive task given the sample s i z e . Instead, a v a r i e t y of census data, c a l c u l a t i o n s and observations are used for these var iab l e ca tegor ies . Enumeration areas for census data allow use of f a i r l y homogeneous data for area c h a r a c t e r i s t i c s (being more homogeneous even than census t r a c t s which are purposely def ined to meet the fo l lowing c r i t e r i o n (among others ) : "the area must be as homogeneous as poss ib le i n terms of economic status and s o c i a l l i v i n g condi t ions") . Table 4.2 l i s t s the enumeration area data (based on a 100% or 2 0% sample) a v a i l a b l e from S t a t i s t i c s Canada (1988a,b). P u b l i c sector c h a r a c t e r i s t i c s For the p u b l i c sector c h a r a c t e r i s t i c s , jus t two are considered: p u b l i c t r a n s i t p r o v i s i o n q u a l i t y and p u b l i c school ing p r o v i s i o n . Given the stated uni formity of t r a n s i t s e r v i c e i n the study area as measured by d is tance to a bus route and bus stop spacing (BC T r a n s i t 1983), i t was decided that l i t t l e would be gained by using such a v a r i a b l e . 92 TABLE 4.2 - CENSUS ENUMERATION AREA DATA . populat ion density* . average age* . average age at immigration . family l i v i n g arrangements . family status . family composition . household type . household s ize* . mar i ta l s tatus . mother tongue . home language . o f f i c i a l language . c i t i z e n s h i p . ethnic o r i g i n . place of b i r t h . mobi l i ty status* . per iod of immigration . average employment income . average census family income . average household income* . average i n d i v i d u a l income . t o t a l household income . median census family income . median household income . median i n d i v i d u a l income . c lass of worker . industry . labour force a c t i v i t y . occupation . weeks worked annually . owner's major payments (to secure she l ter ) . owner's major payments or gross rent as % of household income* . sources of income . labour force p a r t i c i p a t i o n rate . highest l e v e l of school ing . major f i e l d of study at school ing Distance to the nearest Jun ior Secondary School was c a l c u l a t e d "as the crow f l i e s " from the c e n t r o i d of the map i d e n t i f i e r (one-sixteenth of a f u l l s ec t ion or square mi le 93 for Richmond; one-half of a sector or one-ninth of a square mi le for Vancouver) i n which an observation i s l oca ted . A c c e s s i b i l i t y c h a r a c t e r i s t i c s Two sets of v a r i a b l e s are ca lcu la ted for a c c e s s i b i l i t y c h a r a c t e r i s t i c s . The f i r s t set uses distance from the i d e n t i f i e r centro id "as the crow f l i e s " . Distance to the southern end of the Oak Street Bridge i s used as a proxy for access from Richmond to economic a c t i v i t y i n the CBD and elsewhere i n the C i t y of Vancouver. For access to the p r i n c i p a l commercial corner i n Richmond a foca l po int on the commercial s t r i p along No. 3 Road i s chosen, with d is tance being taken to the Lansdowne Park Shopping Centre. As a measure of a c c e s s i b i l i t y to the a i r p o r t , d istance to the southern end of the Dinsmore Bridge i s used because i t provides the only access to Sea Is land for most Richmond re s ident s . Concern about serious d i s t o r t i o n s a r i s i n g from use of these dis tance measures which do not take account of t r a v e l time v a r i a t i o n during rush hours and at other times l e d to use of a second set of a c c e s s i b i l i t y v a r i a b l e s , based on t r a v e l time. Data produced i n the Greater Vancouver Regional D i s t r i c t (1989) survey i s used for the morning peak t r a v e l times from 34 zones i n Richmond to the CBD, to the A i r p o r t and to the Lansdowne Park Shopping Centre area . P a r t i c u l a r care was exerc ised i n the matching of cases to time zones as the geographical areas for zones are not 94 cons i s tent with other geographic i d e n t i f i c a t i o n used i n the study. Noise Data for one area c h a r a c t e r i s t i c , noise , i s taken as a g iven i n t h i s study. Transport Canada has suppl ied a map of NEF contours associated with operations at Vancouver I n t e r n a t i o n a l A i r p o r t for 1987. The contour map, reproduced i n Chapter 3 as Figure 3.4, was produced by a s imulat ion which uses as i t s main input the number of f l i g h t s by type of plane by runway during the year . There has been no publ i shed v a l i d a t i o n of these NEF values by actua l measurement at the Remote Monitoring Stat ions es tab l i shed severa l years ago by Transport Canada at var ious l o c a l s i t e s . The merits of the NEF d e s c r i p t o r were out l ined i n the d i s cus s ion of noise measurement i n Chapter 2. M u l t i c o l l i n e a r i t y There are many p o t e n t i a l independent v a r i a b l e s ; for models i n v o l v i n g e i t h e r i n d i v i d u a l sale or census value data, the r i s k of ser ious m u l t i c o l l i n e a r i t y (high c o r r e l a t i o n s between independent var iables ) i s great . This encourages the very c a r e f u l s e l e c t i o n of f i n a l v a r i a b l e s tak ing in to account c o r r e l a t i o n and other s t a t i s t i c a l r e l a t i o n s h i p s , and a p r i o r i reasoning regarding sens ib le r e l a t i o n s h i p s . I t may be noted, however, that i f the only use of the model i s to estimate the noise parameter; ser ious m u l t i c o l l i n e a r i t y i s 95 not of concern so long as such m u l t i c o l l i n e a r i t y as ex i s t s does not involve the noise parameter. I f the model i s used i n an e f for t to determine the r e l a t i v e importance of i n d i v i d u a l independent v a r i a b l e s , h igh m u l t i c o l l i n e a r i t y i s of greater concern. I f the model i s to be used for p r e d i c t i v e purposes, high m u l t i c o l l i n e a r i t y i s not an issue i f the future i n t e r c o r r e l a t i o n pat tern i s assumed not to change (thus rendering i t useless i f any one of the independent v a r i a b l e s changes) (Stewart 1976, pp 102-3) . The i m p l i c a t i o n of a l l t h i s i s that being s t r i c t l y concerned only with noise , there i s no need to develop a model i n which a l l independent v a r i a b l e s are unaffected by each other . I f the model i s to be used for p r e d i c t i n g house p r i c e , m u l t i c o l l i n e a r i t y i s l i k e l y to be of major concern, and would require cons iderat ion of a l t e r n a t i v e s t a t i s t i c a l techniques such as p r i n c i p a l components or fac tor a n a l y s i s , two- or three-stage l eas t squares modell ing techniques, simultaneous equation techniques, and nonl inear models or non l inear funct iona l forms ( M i l l e r 1982, p .36) . For example: Mark (1980) uses both p r i n c i p a l components a n a l y s i s and regress ion analys i s to t e s t the r e l a t i o n s h i p between environmental e x t e r n a l i t i e s and housing p r i c e s ; and Krumm (1980) undertakes an empir i ca l i n v e s t i g a t i o n of housing and neighbourhood amenities us ing fac tor a n a l y s i s . 96 4.5 MODEL TYPES The two model types considered are summarized i n Table 4.3. The pre ferred model i s type A i n that i t involves the l e a s t aggregation of v a r i a b l e s and the most comprehensive l i s t of explanatory v a r i a b l e s . Compared to type A, the data requ ired for type B are a l l averages based on enumeration areas ( d i r e c t l y or i n d i r e c t l y from i d e n t i f i e r s ) and involve TABLE 4.3 - MODEL TYPES Type Dependent Independent v a r i a b l e v a r i a b l e Phys. Area Pub. sect . Access. Noise i n d i v i d . BCAA census ca lcu la ted c a l c u l a t e d NEF house EA for for for sa l e i d e n t i f i e r i d e n t i f i e r i d e n t i f i e r p r i c e s and time zone B census census census ca lcu la ted c a l c u l a t e d NEF EA EA b EA (for iden) (for iden) (for iden) average averaged averaged averaged dwel l ing for EA for EA for EA va lue a Notes: a census data for average dwel l ing value at the enumeration area l e v e l were not a v a i l a b l e ; for each EA a v a r i a b l e using census weighted (by proport ion of s ingle-detached houses i n an EA) dwel l ing value was ca l cu la ted from BCAA i n d i v i d u a l data . b census data for average number of rooms at the enumeration area l e v e l were not a v a i l a b l e ; for each EA a v a r i a b l e using census-weighted (by proport ion of s ingle-detached houses i n an EA) number of rooms was c a l c u l a t e d from BCAA i n d i v i d u a l data . 97 l e s s phys i ca l c h a r a c t e r i s t i c s than type A. I t i s to be expected that far more re l i ance can be placed on model type A r e s u l t s , although for p r e d i c t i v e purposes i f model type B performed reasonably sens ib ly , i t i s more a t t r a c t i v e by v i r t u e of i t s less r igorous - and thus less expensive - data requirements. The focus i n t h i s study, though, on the i n d i v i d u a l data and model type A; reference i s made i n the fo l lowing pre l iminary d i scuss ion to the a l t e r n a t i v e model type B which i s tested a f t e r model type A. 4.6 STRATIFICATION Two necessary condi t ions for market s t r a t i f i c a t i o n were i d e n t i f i e d i n the review of the theory (issue i v ) : (i) market b a r r i e r s , and ( i i ) d i f f e r e n t demand and/or supply s t r u c t u r e s . I f e i t h e r condi t ion e x i s t s , separate models should be designed for the market segments. Condi t ion (i) i s more d i f f i c u l t to i d e n t i f y ; condi t ion ( i i ) may be evidenced by demand and/or supply s tructures d i f f e r e n t i a t e d by property types, tenure, p r i c e , l oca t ion (such as a c c e s s i b i l i t y to the a i rpor t ) and m u n i c i p a l i t y . Property type For the i n d i v i d u a l property data , a p r i o r i , i t i s s ens ib le to separate d i f f e r e n t property types: s ingle-detached 98 dwe l l ing , m u l t i - u n i t condominiums, commercial and i n d u s t r i a l property , p u b l i c use property and vacant land. As noted i n the l i t e r a t u r e review, most s tudies deal e x c l u s i v e l y with one property type: s ingle-detached r e s i d e n t i a l u n i t s . In recent years the r e l a t i v e market share of m u l t i - u n i t condominiums has increased markedly. Th i s submarket should not be ignored, p a r t i c u l a r l y i n Richmond where t h i s form of tenure i s very popular. A l s o , i t i s p o s s i b l e that a i r p o r t noise a f f ec t s s ingle-detached and m u l t i - u n i t (and other) condominiums d i f f e r e n t l y because: . condominiums are l i k e l y to be be t t er soundproofed because of t h e i r higher dens i ty l i v i n g arrangements. . condominiums allow less outdoor a c t i v i t i e s , thus p o s s i b l y l i m i t i n g d i r e c t noise exposure. . condominium occupants are l i k e l y to be more mobile and thus discount for the noise e f f e c t l e s s . Unfor tunate ly , assembly of the data set f or t h i s property type was not able to be completed to al low t e s t i n g of condominiums as part of t h i s t h e s i s . The va lue of commercial and i n d u s t r i a l property i s a f fec ted by d i f f e r e n t var iab le s than r e s i d e n t i a l property; i t i s 99 excluded from th i s study. I t may be noted that the impact of a i r p o r t noise on these propert ies i s dependent upon the q u a l i t y of soundproofing, i n t e r i o r / e x t e r i o r a c t i v i t y , ambient noise l eve l s and so on. A c c e s s i b i l i t y to the a i r p o r t i s l i k e l y to be more important than for r e s i d e n t i a l p r o p e r t i e s so that the d i f f i c u l t y of i n s e p a r a b i l i t y of t h i s e f f e c t from the a i r p o r t noise e f fec t may be c r u c i a l . Pub l i c -use property i s even more involved i n i t s c h a r a c t e r i s t i c s . Furthermore, i t i s not amenable to a property value approach given the few, i f any, ac tua l sa l e s . I t i s thus excluded from t h i s study. Vacant land i s of p a r t i c u l a r i n t e r e s t i n Richmond, where cons iderable t r a c t s are s t i l l a v a i l a b l e for development. No reference to models for vacant land has been found i n the l i t e r a t u r e . The ef fects of a i r p o r t noise p o l l u t i o n are expected to be c a p i t a l i z e d into land va lues , except that p o r t i o n which i s r e f l e c t ed i n super ior cons truc t ion for n o i s e - m i t i g a t i o n purposes. Thus, the pos s ib l e e r r o r s assoc iated with est imating improvement values are e l iminated . Therefore, for the vacant land market segment, the estimated re su l t s could be expected to be more p r e c i s e . Pre l iminary t e s t ing suggested, however, the sample s i ze was too sma l l , and the v a r i a t i o n too great . Therefore model l ing of vacant land was not attempted as part of t h i s t h e s i s . 100 For models using census data, enumeration areas could be i d e n t i f i e d as predominantly single-detached houses, apartments (defined as 5 or more s t o r i e s ) , or a l l other types ( inc luding semi-detached, row-houses, other s i n g l e -detached houses, apartments i n a detached duplex and apartments i n a b u i l d i n g that has less than f i v e s tor ie s ) and the sample s t r a t i f i e d on t h i s bas i s . Such an approach i s not sens ible i n Richmond because of the mix i n most EAs, the c l a s s i f i c a t i o n of almost a l l condominiums as " a l l other types" (because they are less than 5 s t o r i e s ) , and the small number of EAs. Thus, i n t h i s study, pooled census data are used for dwel l ing value and phys ica l c h a r a c t e r i s t i c s with average EA values for these var iab les being c a l c u l a t e d by weighting by the proport ion of s ingle-detached houses i n the EA. Tenure Rental u n i t s , both condominium and t r a d i t i o n a l income p r o p e r t i e s , represent yet another degree of occupant m o b i l i t y and thus a d i f f e r e n t market. The noise discount fac tor may again be lower, g iv ing further reason to segment owner-occupiers and renters . Census data for enumeration areas may ind ica te predominantly owner- or renter -occupied p r i v a t e dwel l ings and the sample s t r a t i f i e d on t h i s b a s i s . However, the l i m i t e d sample s i ze would cause problems. As a proxy, the proport ion of owner-occupied p r i v a t e dwel l ings may be used as an independent var iab l e i n an u n s t r a t i f i e d 101 model. For i n d i v i d u a l sa les , previous or subsequent tenure s tatus i s unava i lab le , and the same proxy may be used as an area c h a r a c t e r i s t i c . P r i c e Some of the reviewed studies s t r a t i f y by p r i c e . For t h i s study, cons iderat ion of p r i c e data (by property type) d i d not suggest i t i s worthwhile t e s t i n g to see i f there are any s i g n i f i c a n t d i f ferences i n models estimated for the t o t a l market and by p r i c e range. A i r p o r t a c c e s s i b i l i t y The r o l e of a i r p o r t a c c e s s i b i l i t y has been considered i n the l i t e r a t u r e review and must not be ignored i n t h i s study. Nelson's (1979, 1981a) approach of using a small study area such that a c c e s s i b i l i t y to the a i r p o r t (and CBD a c c e s s i b i l i t y and p u b l i c sector c h a r a c t e r i s t i c s ) i s constant might be fol lowed by segmenting the Richmond area into markets of s i m i l a r distance from the a i r p o r t . In the case of Richmond, the distance to the Dinsmore Bridge or t r a v e l time to the a i r p o r t could be used to segment the market, ra ther than j u s t being used as the a c c e s s i b i l i t y v a r i a b l e s i n one u n s t r a t i f i e d model as i s done here. M u n i c i p a l i t y Several previous studies have segmented by m u n i c i p a l i t y or by c i t y and suburbs. The markets i n t h i s study are qui te 102 d i s t i n c t , r e q u i r i n g Richmond, where the majori ty of the no i se -af fec ted propert i e s are located, to be considered separately from Vancouver. The regressions for Richmond could be modelled for both c i t i e s together, by adding a dummy v a r i a b l e for Vancouver, although t h i s i s not attempted here. 4.7 COLLECTION DATE For the i n d i v i d u a l sa les data, observations are a v a i l a b l e for January 1987 through June 1989. The review of the theory (issue i ) suggested that i n order to ensure that the housing market i s "close" to equ i l ibr ium, per iods i n which p r i c e s are r e l a t i v e l y s table (after taking in to account the general i n f l a t i o n rate) should be used. Figure 4.1 graphs the median monthly s ingle-detached sale p r i c e s for Greater Vancouver for the per iod for which data are a v a i l a b l e . The data shown here revea l the much less s table market during 1988. The more s table per iod of 1987 i s pre f erred because the housing market i s more l i k e l y to be i n e q u i l i b r i u m . During t h i s per iod i t may be assumed that supply i s not responsive i n the short run (that i s , supply i s i n e l a s t i c at the e x i s t i n g p r i c e l eve l ) which takes account of theory i ssue v . A monthly dummy i s tested for any i n f l a t i o n a r y e f f e c t s , although the per iod i s not marked by an abnormal ra te of general i n f l a t i o n . Because there are obvious 103 o FIGURE 4.1 - MEDIAN MONTHLY SINGLE-DETACHED SALE PRICE, GREATER VANCOUVER. JANUARY 1987 TO JUNE 1989 Thousand dol lars 300 250 200 150 100 50 0 1987 L _ L 1988 = e' I. 1989 j J.. J F M A M J J A S O N D J F M A M J J A S O N D J F M A M J Month f luc tua t ions through the year quarter ly dummies are also tes ted i n the model. Convenient ly , the se lected period matches reasonably wel l the BCAA data . The census enumeration data for June 3, 1986 s l i g h t l y predates the data c o l l e c t i o n p e r i o d , but t h i s should not be of too much concern, given the aggregate nature of the data - as long as housing market equ i l ibr ium i s assumed. Publ i c sector and a c c e s s i b i l i t y d is tance measures are constant across time. T r a v e l times are those a p p l i c a b l e for 1989 and the NEF values are those applying to 1987. 4.8 SAMPLE SIZE I t was e a r l i e r noted that some of the previous models have small sample s i z e s , p a r t i c u l a r l y those us ing only census data . The s t a t i s t i c a l value of such models may be questioned i f they attempt to model numerous explanatory v a r i a b l e s . Th i s i s a major reason for p r e f e r r i n g models us ing the i n d i v i d u a l sa les data to those using census average values . Of the t o t a l 4000 single-detached property sa les i n Richmond during 1987, a systematic sample i s taken to obta in a sample of 1539 cases. Using t h i s number of observat ions (or l e ss for fur ther s t r a t i f i e d models) s a t i s f i e s the t h e o r e t i c a l l i m i t a t i o n of housing v a r i e t y (issue i i ) . 105 In a further e f f o r t to meet the t h e o r e t i c a l l i m i t a t i o n r e q u i r i n g s u f f i c i e n t housing v a r i e t y , the sample composition must be v a r i e d such that a l l NEF l eve l s are represented, i n c l u d i n g those areas "unaffected" by a i r p o r t noise . The sample of Richmond sales includes 894 p r o p e r t i e s outside the NEF 25 noise zone (excluding Steveston which has completely d i f f e r e n t c h a r a c t e r i s t i c s ) . 4.9 SCREENING Grounds for screening out p a r t i c u l a r observat ions have been reported for previous s tudies . The c o l l e c t i o n of i n d i v i d u a l sa les data i n t h i s study excludes a l l sa les which for any reason appear not to involve arm's length t ransac t ions , as we l l as those where data implied demol i t ion . Without the examination of each property or the use of very d e t a i l e d maps, i t i s d i f f i c u l t to screen for those proper t i e s adjacent or appropr ia te ly near l o c a l environmental features , major transport f a c i l i t i e s , commercial developments, cemeteries , schools and the l i k e . The r e l a t i v e importance of l o c a t i o n - r e l a t e d fac tors , as rated by Rea l tors and appra i ser s , has been reported i n Frankel (1988a, pp 4-8). To the extent that these factors are not captured i n the model, the i n c l u s i o n of observations whose value i s af fected 106 by such may d i s t o r t the r e s u l t s . This may or may not a f f ec t the noise c o e f f i c i e n t , but i t almost c e r t a i n l y a f f ec t s the o v e r a l l model i f L i and Brown's (1980) r e s u l t s are a p p l i c a b l e to t h i s study. They present evidence that "micro-neighbourhood" factors such as proximity to c e r t a i n n o n - r e s i d e n t i a l land uses and v i s u a l q u a l i t y do have s i g n i f i c a n t impact on other v a r i a b l e s . The same appl i e s for census data, although for almost every enumeration area (being an aggregate of up to 375 households) , at l eas t one reason for exc lus ion may always be found because some propert ies wi th in the EA are a f fec ted -t h i s i s why many of the previous models screened out a large number of census areas. By not doing t h i s , the i m p l i c i t assumption i s that v a r i a t i o n between EAs for these l o c a t i o n a l fac tors i s not important. By i n c l u d i n g census data for proport ion of owner-occupation there i s no need to screen on t h i s bas is as i s done i n many of the reviewed models. 4.10 NOISE In the d i s cus s ion of t h e o r e t i c a l i s sues , i t was asserted that the information gap for noise i s u n l i k e l y to be l a r g e ; that i s , the ac tua l noise l e v e l corresponds c l o s e l y with perce ived noise l e v e l . Assuming then that NEF i s an 107 appropriate measure, and that the values suppl ied by Transport Canada are accurate , there need be l i t t l e concern about theory issue i i i . The d i scuss ion of standards i n Chapter 2 h inted at a range of noise l eve l s appropriate as thresholds for a i r p o r t noise impact. For t h i s study, NEF i s accepted as an appropriate measure of c o l l e c t i v e community response. I f one accepts arguments i n the l i t e r a t u r e for a threshold noise l e v e l of NEF 25, a l l observations i n the contro l areas should be assigned NEF 24. As l a t e r discussed, i t i s p o s s i b l e to experiment with the noise l e v e l for the c o n t r o l area . For those proper t i e s with noise l eve l s of NEF 25 or h igher , NEF values from 25 upwards are a l loca ted to the c e n t r o i d of each map i d e n t i f i e r on the bas i s of an NEF contour map for 1987, suppl ied by Transport Canada. Given the averaging required i n t h i s process and the arguments for a downward-bias i n v a l u a t i o n (issue i ) , these centro id NEF l e v e l s are , when there i s any uncer ta inty , set to a higher NEF l e v e l rather than a lower one (eg. 37, instead of 36). I t may be speculated that the NEF l eve l s are lower bounds given the vary ing to lerance of people to a i r p o r t no ise . As argued i n the l i t e r a t u r e review, the l eas t averse persons are l i k e l y to s e l ec t noisy areas. The impact of t h i s could be v e r i f i e d by using a hybr id noise approach - which i s not attempted here. 108 4.11 DATA MANAGEMENT A l l computerized analys i s for th i s study was c a r r i e d out us ing the software, S t a t i s t i c a l Package for S o c i a l Sciences Extended Vers ion (SPSS-X), Release 3.0 . The c o l l e c t i o n and manipulation of data involved severa l phases: . the recording of sales information and c e r t a i n geographic i d e n t i f i e r s . . the recording of l e g a l , l o c a t i o n and p h y s i c a l c h a r a c t e r i s t i c for each property. . the a l l o c a t i o n of propert ies to enumeration areas. . the recording of census data by EAs, and thus by i n d i v i d u a l p r o p e r t i e s . . the recording of NEF l eve l s and d is tances by i d e n t i f i e r s , and thus by i n d i v i d u a l p r o p e r t i e s . . the a l l o c a t i o n of i d e n t i f i e r s and EAs to time zones. . the recording of t r a v e l times by time zones, and thus by i n d i v i d u a l proper t i e s . 109 I t was i n i t i a l l y envisaged that the phys ica l data would be a v a i l a b l e from the BCAA on computer tape. Accurate ly a n t i c i p a t e d delays i n obtaining the data i n t h i s form required the recording of phys ica l data on hardcopy from the r e g i o n a l o f f i ce s of the Assessment Author i ty . These data were subsequently t rans ferred to the computer data f i l e . T h i s par t of the data c o l l e c t i o n process involved the highest r i s k and cost , and i t may be noted that e r r o r s made here took considerable e f f o r t to clean up. Otherwise, i t was a normal laborious e f f o r t reading maps, t r a n s f e r r i n g other computer output and double-checking. V e r i f i c a t i o n of o r i g i n a l data entry was undertaken using the CROSSTAB, CONDESCRIPTIVE and other procedures i n SPSS-X. The f i n a l data set comprises 92 separate pieces of information (many of which are "missing" for p a r t i c u l a r property types - because they are i rre l evant ) s tored i n 8 records for each sa le ( i d e n t i f i e d by r o l l number for a property so ld , with mul t ip le sales being recorded where a p p r o p r i a t e ) . Pre l iminary checks, using SPSS-X FREQUENCIES, CORRELATIONS, CROSSTABS and PLOT procedures, revealed miscoding, programming e r r o r s , mi s labe l l ed sa l e s , formatt ing and other pred ic tab le problems - a l l of which were s a t i s f a c t o r i l y correc ted . 110 CHAPTER 5 - ANALYSIS AND RESULTS 5.1 INTRODUCTION The p r i n c i p a l hypothesis being tested i s that a i r p o r t noise i s important i n an hedonic p r i c e model where property value i s a funct ion of various c h a r a c t e r i s t i c s . Ex ante, many explanatory v a r i a b l e s i n several categories are considered appropr iate for t e s t i n g , but t h e i r i n c l u s i o n depends on which p a r t i c u l a r data are being used as wel l as t h e i r r e l i a b i l i t y and/or value (for instance, during data c o l l e c t i o n i t became evident that the Assessment A u t h o r i t y d i d not comprehensively record a l l the items l i s t e d on t h e i r record ing forms). Although i n t e s t i n g the hypothesis the concern i s with the a p p l i c a t i o n of a u n i v a r i a t e s t a t i s t i c a l technique, m u l t i p l e regres s ion , there need only be i n d i r e c t concern with i t s t r a d i t i o n a l goal of p r e d i c t i n g the value of the dependent v a r i a b l e from a set of independent v a r i a b l e s . As a lready noted, the values of the independent v a r i a b l e s - other than any dea l ing with noise - are s t r i c t l y i r r e l e v a n t i f the v a r i a t i o n i n noise i s s a t i s f a c t o r i l y i s o l a t e d . Given the p o s s i b i l i t y of i n t e r c o r r e l a t i o n with noise , there i s a need to be aware of m u l t i c o l l i n e a r i t y , and for the purposes of p o r t a b i l i t y and reproduct ion i t would be valuable i f other 111 c o e f f i c i e n t s have sens ible values (and appropriate s i gns ) . The prime concern, then, i s with the s ign and value of the unstandardized B c o e f f i c i e n t for the noise v a r i a b l e . In a l i n e a r form against the natural logarithm of property value as the dependent v a r i a b l e , the value of t h i s c o e f f i c i e n t can be d i r e c t l y l inked to percent change i n property value caused by l o c a t i o n i n areas exposed to d i f f e r e n t noise l e v e l s . For a l l modell ing (regardless of model type) two a l t e r n a t i v e data sets are tes ted . F i r s t , only those cases for proper t i e s ins ide the NEF 25 noise zone are tested (the IN model). Second, a l l cases, that i s propert ies outside as we l l as in s ide the NEF 25 noise zone, are tested (the ALL model). Given the t o t a l lack of v a r i a t i o n i n the independent v a r i a b l e of concern (noise) for those cases outs ide NEF 25 ( a l l such propert ies being assigned the same NEF l e v e l for lack of any other est imates) , and uncerta inty as to what the appropriate threshold NEF l e v e l may be, l e s s confidence i s placed i n the r e s u l t s of the ALL model. Th i s chapter out l ines the pre l iminary ana lys i s for the model l ing of Richmond s ingle-detached propert i e s us ing i n d i v i d u a l sa les p r i c e data (model type A ) , reports and discusses the r e s u l t s for the IN and ALL models, and comments b r i e f l y on modell ing of Richmond proper t i e s us ing census data (model type B) . 112 5.2 ORIGINAL DATA O r i g i n a l data p e r t a i n i n g to some 44 var iab les were c o l l e c t e d for each s ingle-detached sale sampled i n Richmond. The names, descr ipt ions and summary s t a t i s t i c s of the o r i g i n a l , untransformed v a r i a b l e s (for the f i n a l sample of 1539 cases) are presented i n Table 5 .1 . 5.3 PRELIMINARY ANALYSIS The o r i g i n a l v a r i a b l e s inc lude s i x dummy ( indicator) v a r i a b l e s with the balance being (discrete) continuous v a r i a b l e s (measuring proport ions or numbers). Pre l iminary a n a l y s i s of these data tes ted for out-of-range va lues , p l a u s i b l e measures of c e n t r a l tendency and d i s p e r s i o n s , and the degree and sense of l i n e a r assoc ia t ion . V a r i a b l e s and cases with missing data were examined, as were v a r i a b l e s with o u t l i e r s . Tests were conducted for the normal i ty of the data (although skewness i s l e s s of a concern with such a large sample), l i n e a r i t y and homoscedasticity. Although t e c h n i c a l l y not n e c e s s a r i l y an issue except for noise (as e a r l i e r noted), i n d i c a t o r s of m u l t i c o l l i n e a r i t y and s i n g u l a r i t y were cons idered. Treatment of problems emerging i n these checks inc luded the c o r r e c t i o n or d e l e t i o n of cases and the d e l e t i o n , recoding 113 TABLE 5.1 - ORIGINAL FORM (UNTRANSFORMED) VARIABLES, ALL RICHMOND SINGLE-DETACHED PROPERTIES V a r i a b l e Mean Standard Minimum Maximum Number V a r i a b l e d e s c r i p t i o n name dev ia t ion of cases PRICE 139099.641 39769.220 46000 390000 1539 Sales p r i c e ($) WIDTH 74.549 245.484 20.0 4356.0 1533 Lot width (feet) DEPTH 132.275 70.084 1.0 1089.0 1533 Lot depth (feet) STORIES 1.396 .492 1 3 1539 No. s t o r i e s i n b u i l d i n g FIREPLAC 1.414 .714 0 4 1539 No. f i r e p l a c e s i n dwel l ing POOL n/a n/a 0 1 1539 Pool f a c i l i t i e s (no/yes) SQFEET 1716.474 637.544 440 4404 1529 Sq. f t . i n dwel l ing NOROOMS 9.653 2.879 5 22 1534 No. rooms BEDROMS 3.299 .841 1 7 1529 No. bedrooms FBATHS 1.315 .511 0 4 1506 No. f u l l bathrooms PBATHS .843 .691 0 3 1506 No. par t bathrooms NOROOMB 1.466 2.056 0 9 1523 No. rooms i n basement BALCONY .628 .485 0 2 1539 No. balconies YRBUILT 68.959 13.408 28 87 1539 Year b u i l t EFFECYR 70.208 13.095 39 88 125 E f f e c t i v e year b u i l t YRREMOD 74.793 8.908 56 89 58 Year remodelled EXWALL n/a n/a 1 56 1533 Type of e x t e r i o r wal l (var) EXCOND n/a n/a 0 2 1539 E x t e r i o r condi t ion (var) ROOFT n/a n/a 0 3 1535 Roof type (various) BSUITE n/a n/a 0 1 1539 Basement su i t e (no/yes) CORNER n/a n/a 0 1 1539 Corner l o t (no/yes) NONMOVRS 52.138 14.713 0 79 1539 Prop, non-movers MIGRANTS 22.710 7.824 6 59 1539 Prop, migrants A65 8.155 4.500 2 34 1539 Prop, aged 65 and over A15 21.181 4.075 2 31 1539 Prop, aged 15 and l e s s TABLE 5.1 (cont'd) - ORIGINAL FORM (UNTRANSFORMED) VARIABLES, ALL RICHMOND SINGLE-DETACHED PROPERTIES V a r i a b l e Mean Standard Minimum Maximum Number V a r i a b l e d e s c r i p t i o n name dev ia t ion of cases LABOUR 72. 237 6. 170 46 91 1539 Prop. labour force employed OWNED 77. 206 16. 534 0 96 1539 Prop. owned dwel l ings SINGLE 75. 754 20. 385 0 100 1539 Prop. s i n g l e dwel l ings YRCONST 16. 334 21. 621 0 100 1539 Prop. constructed s ince 1981 ONEPERS 12. 089 7. 419 0 46 1539 Prop. one-person households TWOPRS 29. 905 6. 175 17 61 1539 Prop. two-person households AVGPRS 3. 062 • 373 1 .72 3 .78 1539 Avge. persons i n household AVGCHILD 1. 250 • 199 .23 1 .71 1539 Avge. c h i l d r e n i n household AVGINC 43323. 164 7037. 958 20332 66426 1539 Avge. income i n household MOTHER 72. 246 9. 069 45 .00 92 .00 1539 Prop. E n g l i s h mother tongue DISTSCH 1. 167 • 983 .20 11 .90 1539 D i s t . to Jun. Sec. Sch. (km) DISTDINS 3. 401 1. 385 .00 13 .70 1539 D i s t . to Dinsmore Bridge (km) DISTLANS 3. 377 1. 210 .20 12 .50 1539 D i s t . to Lansdowne Park (km) DISTOAK 4. 823 1. 722 .30 11 .70 1539 D i s t . to Oak S t . Bridge (km) AIRT 16. 149 2. 490 7 .56 23 .72 1539 Time to a i r p o r t (minutes) SHOPT 7. 263 2. 629 4 . 12 20 .22 1375 Time to shops (minutes) CBDT 27. 496 3. 082 19 .90 32 .95 1539 Time to CBD (minutes) NEFLEVEL 19. 110 10. 629 10 41 1539 NEF l e v e l MONTH 6. 311 3. 086 1 12 1538 Month of sa le or otherwise transforming of cer ta in v a r i a b l e s . Each of the r e s u l t i n g set of v a r i a b l e s was then tested i n i t s na tura l logarithm and l i n e a r funct ional forms. The logar i thmic form (ac tua l ly the n a t u r a l log of the l inear v a r i a b l e p lus one) i s preferable on t h e o r e t i c a l grounds, and for the continuous v a r i a b l e s data i n t h i s form generally were more normally d i s t r i b u t e d . However, for the dummy v a r i a b l e s with large number of zero observat ions , e i ther approach introduces some s p e c i f i c a t i o n e r r o r (Nelson 1981a, p.73) . Thus, the funct iona l form for each var iab le i s l i n e a r for dummy v a r i a b l e s and log or l i n e a r for continuous v a r i a b l e s -depending on the data f i t . These checks were repeated when those cases with noise l e v e l s of NEF 25 or higher were modelled separate ly (IN models). L i t t l e new of s t a t i s t i c a l concern emerged, although for some continuous var iab le s , the se lec ted func t iona l form does d i f f e r . I t may be noted that as modelling took p lace , graphic a n a l y s i s of r e s i d u a l s was undertaken to a s c e r t a i n that the fo l lowing departures from the mult ip le l i n e a r regress ion model are not made: n o n - l i n e a r i t y , non-consistency of e r r o r var iance , non-independence of error terms, presence of o u t l i e r s , non-normality of error terms, and omission of important independent v a r i a b l e s . 116 5.4 FINAL LIST OF POSSIBLE VARIABLES L i s t s of a l l poss ib le v a r i a b l e s ( inc luding o r i g i n a l ones and those transformed af ter pre l iminary analys is ) and t h e i r summary s t a t i s t i c s , are presented for the ALL models i n Table 5.2, and for the IN models i n Table 5.3. These tab le s a l so report c o r r e l a t i o n s of the independent v a r i a b l e s with the dependent v a r i a b l e , LPRICE (natural log of PRICE + 1) . I t may be noted that the l i s t of v a r i a b l e s i s not qui te i d e n t i c a l for the ALL and the IN models: for some v a r i a b l e s one model uses a natural logar i thmic form and the other uses a l i n e a r form, and d i f f e r e n t transformations of NEFLEVEL are requ ired depending on the sample (as discussed l a t e r ) . C e r t a i n of the var iab les were not tested because of the large number of missing values (eg. IREMOD, remodelled), the almost homogeneous nature of the sample (eg. IBSUITE, presence of a basement s u i t e ) , or spurious accuracy (eg. IHIPROOF, IGABLROO, IFLATROO, IOTHROOF, roof types ) . Of those that were tested, a s table subset c l e a r l y emerged as a s e n s i b l e group for further t e s t i n g . As already noted, m u l t i c o l l i n e a r i t y may be less of a concern given the importance of one independent v a r i a b l e , but a watch was kept on to lerance and unstable c o r r e l a t i o n c o e f f i c i e n t s to ensure i t was under c o n t r o l . 117 TABLE 5.2 - FINAL FORM POTENTIAL VARIABLES, ALL RICHMOND SINGLE-DETACHED PROPERTIES (ALL) V a r i a b l e Mean Standard Minimum Maximum Number C o r r e l a t i o n V a r i a b l e d e s c r i p t i o n name dev ia t ion cases with LPRICE LPRICE 11 .805 .273 10.74 12. 87 1539 1.0000 In sa les p r i c e LNWIDTH 4 .088 .251 3.04 5. 30 1533 -.1215 In act . /nom. width LLOTSIZE 8 .893 .472 7.51 11. 21 1533 -.1149 In s i z e of l o t LARGELOT n/a n/a .00 1. 00 1533 .0497 Lot s i ze > 25000 HUGELOT n/a n/a .00 1. 00 1533 .0221 L o t s i z e > 50000 SMALLLOT n/a n/a .00 1. 00 1533 -.0502 L o t s i z e < 15000 ISTORIES n/a n/a .00 1. 00 1539 .4677 S ing le /mul t , s t o r i e s IFIREPL n/a n/a .00 1. 00 1539 .3736 E x i s t , of f i r e p l a c e s I POOL n/a n/a .00 1. 00 1539 .0847 Existence of pool LSQFEET 7 .380 .376 6.09 8. 39 1529 .3138 In s q . f t . dwel l ing LLOTSQFT 1 .735 .458 .88 4. 53 1523 -.3054 In dwel l ing s q . f t . per l o t s q . f t . LSQFTROM 5 .162 .479 3.77 6. 38 1528 -.1197 In dwel l ing s q . f t . per room LNOROOMS 2 .331 .263 1.79 3. 14 1534 .6064 In no. rooms BEDROMS 3 .299 .841 1 7 1529 .3522 No. bedrooms LFBATHS .817 .205 .00 1. 61 1506 .5113 In no. f u l l bathroom LPBATHS .535 .401 .00 1. 39 1506 .4323 In no part bathroom LTBATHS 1 .108 .297 .69 1. 95 1506 .6033 In t o t a l bathrooms LTROOMS 1 .838 .240 1.10 2. 56 1502 .5686 In no. baths+bedroom LNOROOMB .604 .742 .00 2. 30 1523 -.0412 In no. rms basement IBALCON n/a n/a .00 1. 00 1539 .1991 Existence of balcony LAGE 2 .518 1.126 .00 4. 09 1539 -.6391 In property age INEW n/a n/a .00 1. 00 1539 .3573 Const, i n 1987 INEWCONS n/a n/a .00 1. 00 1539 .4897 Const, i n 1986-1987 TABLE 5.2 (cont'd) - FINAL FORM POTENTIAL VARIABLES, ALL RICHMOND SINGLE-DETACHED PROPERTIES (ALL) V a r i a b l e name Mean Standard Minimum deviat ion Maximum Number cases C o r r e l a t i o n with LPRICE V a r i a b l e d e s c r i p t i o n IMOD n/a n/a .00 1.00 1539 .5639 Const, s ince 1977 IREMOD n/a n/a .00 1.00 59 . 3624 Recent remodell ing ISSWALL n/a n/a .00 1.00 1533 . 0701 S i d i n g and/or stucco IPOORCON n/a n/a .00 1.00 1539 -.0223 Below avg. ext. cond IGOODCON n/a n/a .00 1.00 1539 -.0093 Above avg. ext. cond IHIPROOF n/a n/a .00 1.00 1539 -.0214 Hip roof IGABLRF n/a n/a .00 1.00 1539 -.0261 Gable roof IFLATRF n/a n/a .00 1.00 1539 .0401 F l a t roof IOTHROOF n/a n/a .00 1.00 1539 .0569 Other roof IBSUITE n/a n/a .00 1.00 1539 .0226 Presence of b. su i t e ICORNER n/a n/a .00 1.00 1539 -.0373 Corner l o t NONMOVRS 52 .138 14.713 0 79 1539 -.1838 Prop, non-movers LMIGRANT 3 .113 .334 1.95 4.09 1539 .1013 In prop, migrants LA65 2 .119 .428 1.10 3.56 1539 -.0323 In prop, age 65+ A15 21 .181 4.075 2 31 1539 .0355 Prop. age 15-LABOUR 72 .237 6.170 46 91 1539 -.0379 Prop. l a b . force emp OWNED 77 .206 16.534 0 96 1539 -.0332 Prop. own. dwell ings SINGLE 75 .754 20.385 0 100 1539 -.1367 Prop. s i n g , dwell LYRCONST 2 .095 1.316 .00 4.62 1539 . 3248 In prop, const.>1981 LONEPERS 2 .421 .563 .00 3.85 1539 -.0995 In prop. 1-per. h/h LTWOPERS 3 .412 .196 2.89 4.13 1539 -.0106 In prop. 2-per. h/h AVGPERS 3 .062 .373 1.72 3.78 1539 .0782 Avge. persons h/h AVGCHILD 1 .250 . 199 .23 1.71 1539 .0887 Avge. c h i l d r e n h/h AVGINC 43323 .164 7037.958 20332 66426 1539 .1482 Avge. income h/h MOTHER 72 .246 9.069 45.00 92.00 1539 -.1904 Prop. Eng. tongue TABLE 5.2 (cont'd) - FINAL FORM POTENTIAL VARIABLES, ALL RICHMOND SINGLE-DETACHED PROPERTIES (ALL) V a r i a b l e Mean Standard Minimum Maximum Number C o r r e l a t i o n V a r i a b l e d e s c r i p t i o n name deviat ion cases with LPRICE LDISTSCH .715 .314 . 18 2. 56 1539 -.1400 In d i s t . JS school LDISTDIN 1 .427 .354 .00 2. 69 1539 .1491 In d i s t . D in . Bridge LDISTLAN 1 .436 .296 .18 2. 60 1539 .1166 In d i s t . Lansdowne DISTOAK 4 .823 1 .722 . 30 11. 70 1539 .2002 D i s t . Oak S t . Bridge AIRT 16 .149 2 .490 7 .56 23. 72 1539 . 1782 Time to a i r p o r t LSHOPT 2 .072 .267 1 .63 3. 05 1375 -.2483 In time to shops CBDT 27 .496 3 .082 19 .90 32. 95 1539 .2449 Time to CBD NEFLEV24 26 .879 4 .447 24 .00 41. 00 1539 -.1906 NEF l e v e l (t/h=24) NEFLEV2 0 24 .660 6 .043 20 .00 41. 00 1539 -.1677 NEF l e v e l (t/h=2 0) INZONE24 n/a n/a .00 1. 00 1539 . 0873 NEF 24 INZ0NE25 n/a n/a .00 1. 00 1539 .0360 NEF 25 TO 29 INZONE30 n/a n/a .00 1. 00 1539 -.0206 NEF 30 TO 34 INZONE35 n/a n/a .00 1. 00 1539 -.0192 NEF 35 TO 39 INZONE40 n/a n/a .00 1. 00 1539 -.2280 NEF 40 TO 44 NZONE24 27 .921 5 .543 24 .00 44. 00 1539 -.1756 NEF zone (t/h=24) NZONE20 25 .598 7 .271 20 .00 44. 00 1539 -.1576 NEF zone (t/h=20) MONTH 6 .311 3 .086 1 12 1538 .1034 Month of sa le QI n/a n/a .00 1. 00 1539 -.1091 Sold i n quarter 1 Q2 n/a n/a .00 1. 00 1539 -.0134 Sold i n quarter 2 Q3 n/a n/a .00 1. 00 1539 .0806 Sold i n quarter 3 TABLE 5.3 - FINAL FORM POTENTIAL VARIABLES, RICHMOND SINGLE-DETACHED PROPERTIES WITH NEF 25 OR HIGHER (IN) V a r i a b l e Mean Standard Minimum Maximum Number C o r r e l a t i o n V a r i a b l e d e s c r i p t i o n name dev ia t ion cases with LPRICE LPRICE 11 .777 .289 10 .74 12.87 645 1.0000 In sales p r i c e LNWIDTH 4 .127 .254 3 .43 5.30 642 .0404 In act . /nom. width LLOTSIZE 9 .028 .532 7 .84 11.21 642 .0612 In s i z e of l o t LARGELOT n/a n/a .00 1.00 642 .0876 L o t s i z e > 25000 HUGELOT n/a n/a .00 1.00 642 .0487 L o t s i z e > 50000 SMALLLOT n/a n/a .00 1.00 642 -.1622 Lot s i ze < 15000 ISTORIES n/a n/a .00 1.00 645 .4240 S ing le /mul t , s t o r i e s IFIREPL n/a n/a .00 1.00 645 .4015 E x i s t , of f i r e p l a c e s I POOL n/a n/a .00 1.00 645 .1347 Existence of pool LSQFEET 7 .384 .388 6 .09 8.39 642 .3849 In s q . f t . dwel l ing LLOTSQFT 1 .846 .532 .89 4.53 639 -.1784 In dwel l ing s q . f t . per l o t s q . f t . LSQFTROM 5 .214 .470 3 .77 6.23 641 -.0192 In dwel l ing s q . f t . per room LNOROOMS 2 .288 .264 1 .79 3.14 644 .5438 In no. rooms BEDROMS 3 .272 .891 1 7 643 .3850 No. bedrooms LFBATHS .800 . 194 .00 1.39 643 .4123 In no. f u l l bathroom LPBATHS .502 .407 .00 1.10 643 .4312 In no. part bathroom LTBATHS 1 .075 .300 .69 1.79 643 .5449 In t o t a l bathrooms LTROOMS 1 .815 .256 1 .10 2.40 641 .5398 In no. baths+bedroom LNOROOMB .616 .759 .00 2.30 644 .0108 In no. rms basement IBALCON n/a n/a .00 1.00 645 . 1476 Existence of balcony LAGE 2 .745 1.029 .00 4.09 645 -.5876 Property age INEW n/a n/a .00 1.00 645 .2719 Const, i n 1987 INEWCONS n/a n/a .00 1.00 645 .4005 Const, i n 1986-1987 TABLE 5.3 (cont 'd) - FINAL FORM POTENTIAL VARIABLES, RICHMOND SINGLE-DETACHED PROPERTIES WITH NEF 25 OR HIGHER (IN) V a r i a b l e Mean Standard Minimum Maximum Number C o r r e l a t i o n V a r i a b l e d e s c r i p t i o n name dev iat ion cases with LPRICE IMOD n/a n/a .00 1. 00 645 .4878 Const, s ince 1977 IREMOD n/a n/a .00 1. 00 22 .2795 Recent remodell ing ISSWALL n/a n/a .00 1. 00 644 .0337 S id ing and/or stucco IPOORCON n/a n/a .00 1. 00 645 -.0559 Below avg. ext . cond IGOODCON n/a n/a .00 1. 00 645 .0123 Above avg. ext . cond IHIPROOF n/a n/a .00 1. 00 645 -.0361 Hip roof IGABLRF n/a n/a .00 1. 00 645 -.0848 Gable roof IFLATRF n/a n/a .00 1. 00 645 . 0849 F l a t roof IOTHROOF n/a n/a . 00 1. 00 645 . 1667 Other roof IBSUITE n/a n/a .00 1. 00 645 .0093 Presence of b. su i te ICORNER n/a n/a .00 1. 00 645 -.0462 Corner l o t NONMOVRS 48 .653 15.998 0 79 645 -.1542 Prop, non-movers LMIGRANT 3 .167 .326 2.48 4. 09 645 .1034 In prop, migrants LA65 2 .212 .462 1.10 3. 47 645 -.1158 In prop, age 65+ A15 20 .581 4.720 2 30 645 .0946 Prop, age 15-LABOUR 71 .183 7.187 55 91 645 -.0904 Prop. l a b . force emp OWNED 69 .656 19.247 0 94 645 -.0230 Prop. own. dwell ings SINGLE 70 .622 22.155 0 99 645 -.1661 Prop. s i n g , dwell LYRCONST 2 .320 1.274 .00 4. 62 645 .3171 In prop, const.>1981 LONEPERS 2 .635 .570 1.10 3. 85 645 -.1387 In prop. 1-per h/h LTWOPERS 3 .426 . 197 3.04 4. 09 645 -.0880 In prop. 2-per h/h AVGPERS 2 .964 .415 1.72 3. 69 645 .1308 Avge. persons h/h AVGCHILD 1 .209 .222 .23 1. 54 645 . 1610 Avge. c h i l d r e n h/h LAVGINC 10 .605 . 183 10.09 11. 09 645 .2164 In avge. income h/h MOTHER 72 .036 10.819 45.00 92. 00 645 -.1307 Prop. Eng. tongue TABLE 5.3 (cont'd) - FINAL FORM POTENTIAL VARIABLES, RICHMOND SINGLE-DETACHED PROPERTIES WITH NEF 25 OR HIGHER (IN) V a r i a b l e Mean Standard Minimum Maximum Number C o r r e l a t i o n V a r i a b l e descr ip t ion name dev iat ion cases with LPRICE LDISTSCH .760 .321 .26 2.10 645 -.1215 In d i s t . JS school DISTDINS 2.477 1.072 .00 8.50 645 . 1229 D i s t . D in . Bridge DISTLANS 2.433 .844 .20 7.30 645 .1862 D i s t . Lansdowne DISTOAK 3.349 1.479 . 30 6.80 645 .3364 D i s t . Oak St . Bridge AIRT 14.468 2.143 7.56 23.72 645 .3020 Time to a i r p o r t LSHOPT 2.165 .273 1.63 3.05 494 -.0937 In time to shops CBDT 25.341 2.666 19.90 31.82 645 .3239 time to CBD NEFLEVEL 30.870 4.446 25 41 645 -.2804 NEF l e v e l INZONE25 n/a n/a .00 1.00 645 .1460 NEF 25 TO 29 INZONE30 n/a n/a .00 1.00 645 .0418 NEF 30 TO 34 INZONE35 n/a n/a .00 1.00 645 .0026 NEF 35 TO 39 INZONE40 n/a n/a .00 1.00 645 -.3117 NEF 40 TO 44 NZONE 33.357 4.738 29.00 44.00 645 -.2722 NEF zone l e v e l MONTH 6.166 3.072 1 12 644 .1041 Month of sa le QI n/a n/a .00 1.00 645 -.1139 Sold i n quarter 1 Q2 n/a n/a .00 1.00 645 -.0235 Sold i n quarter 2 Q3 n/a n/a .00 1.00 645 .0862 Sold i n quarter 3 The t o t a l number of cases for ALL and IN models gives a high c a s e / v a r i a b l e r a t i o , these being about 100:1 and 40:1 for the respect ive f i n a l models presented l a t e r . This makes i t l e s s important that res idua l s be normally d i s t r i b u t e d (Tabachnick and F i d e l l 1988, p.92) and allows the mul t ip le regress ion model to bet ter cope with skewed v a r i a b l e s , small e f f e c t s and substant ia l measurement e r r o r s . TABLE 5.4 - CASES BY NEF LEVEL, SAMPLE OF RICHMOND SINGLE-DETACHED PROPERTIES Noise zones NEF l e v e l Number of cases T o t a l cases 24 or l e s s 24 or less 894 894 25 - 29 25 73 26 44 27 32 28 62 29 61 272 30 - 34 30 103 31 2 32 79 33 62 34 2 248 35 - 39 35 30 36 12 37 14 38 1 39 4 61 40 or more 40 39 41 25 64 TOTAL a l l 1539 1539 124 Of the t o t a l 1539 cases i n the sample, 645 cases have an NEF l e v e l of 25 or higher. The number of cases at each NEF l e v e l i s l i s t e d in Table 5.4. Also shown i n Table 5.4 i s the grouping into noise zones for ins ide and outside the def ined noise area. For the ALL models, a dec i s ion needs to be made as to the threshold NEF l e v e l , that i s , the noise l e v e l below which propert ies are assumed to have n o n - c r i t i c a l NEFs because of ambient noise a lready e x i s t i n g i n a l o c a t i o n . Although a threshold of NEF 25 recommended by the Centra l Housing and Mortgage Corporat ion (outl ined i n Chapter 2) i s accepted, the a r b i t r a r y a l l o c a t i o n of proper t i e s located below t h i s threshold to NEF 24 and to NEF 20 i s tested, i n each case with a l l other propert ies having NEF 25 or greater . Because of the uncertainty of the noise l e v e l for those proper t i e s (the a i r p o r t noise s imulat ion model does not capture non-a i r p o r t no i se ) , the IN models are considered super ior . For the IN and ALL models, three types of noise v a r i a b l e s are tes ted: . one continuous v a r i a b l e for a l l i n d i v i d u a l NEF l e v e l s , which assumes a l i n e a r r e l a t i o n s h i p between p r i c e and noise . . dummy var iab le s f or the noise zones, which al low d i f f e r e n t p r i c e - n o i s e r e l a t i o n s h i p s i n d i f f e r e n t zones. 125 . a scaled continuous v a r i a b l e which s t i l l assumes a l i n e a r pr i ce -no i se r e l a t i o n s h i p , but which only uses a few NEF l eve l s as values (representing ranges of i n d i v i d u a l NEF l e v e l s ) ; t h i s does not require the a l l o c a t i o n of NEF l eve l s to propert ies to be as accurate . 5.5 THE "IN" MODEL For the 645 cases with noise l e v e l s greater than or equal to NEF 25, the zero-order c o r r e l a t i o n with the dependent v a r i a b l e LPRICE i s -.280 suggesting that , ignoring a l l other r e l a t i o n s h i p s , the higher the noise l e v e l , the lower the p r i c e . A (simple) c o r r e l a t i o n matrix of a l l v a r i a b l e s i n the f i n a l IN models i s presented as Table C . l i n Appendix C. The tab le shows that c o r r e l a t i o n s of NEFLEVEL with other independent var iab le s i n the f i n a l continuous IN model range ( in absolute terms) from -.011 (NONMOVRS) to -.203 (LTBATHS). The c o r r e l a t i o n s of other independent v a r i a b l e s with LPRICE range ( in absolute terms) from -.023 (Q2) to -.588 (LAGE). The r e s u l t s of the f i n a l continuous IN model (Model 1), us ing m u l t i p l e regress ion with pairwise d e l e t i o n of missing va lues , are presented i n Table 5.5. Of a l l the IN models 126 discussed, t h i s model i s the most important; thus Table 5.5 includes a comprehensive set of s t a t i s t i c s , whereas d i s cus s ion of other models i s l ess d e t a i l e d . The adjusted R2 value of .634 (used i n preference to the R2 value of .643) gives an i n d i c a t i o n of the model's goodness of f i t , implying that the model expla ins jus t over 63 percent of the t o t a l v a r i a b i l i t y i n LPRICE. The o v e r a l l F value (70.170, associated with the Ana lys i s of Variance table) used to tes t the hypothesis that a l l c o e f f i c i e n t s are zero, i s h igh ly s i g n i f i c a n t . Of immediate i n t e r e s t i s the noise v a r i a b l e , NEFLEVEL, which has a B (unstandardized) c o e f f i c i e n t of -.006484, implying a one u n i t increase i n noise re su l t s i n more than h a l f of a one percent decrease i n property p r i c e . The 95 percent confidence i n t e r v a l for t h i s v a r i a b l e i s -.0097 to - .0033. Further d i scuss ion of the impl i ca t ions of these noise r e s u l t s i s deferred u n t i l l a t e r . Other v a r i a b l e s have B c o e f f i c i e n t s with sens ib le signs and magnitudes, as far as such can be determined a p r i o r i . Although these v a r i a b l e s are of l e ss concern, i t i s reassur ing when these aspects of the model make sense. Consider ing each c h a r a c t e r i s t i c type, commentary may be made on those i n the model and those on the l i s t of f i n a l form p o t e n t i a l v a r i a b l e s which were not inc luded / t e s t ed . 127 TABLE 5.5 - RESULTS, RICHMOND SINGLE-DETACHED PROPERTIES WITH NEF 25 OR HIGHER, PREFERRED FORM MODEL WITH CONTINUOUS NOISE VARIABLE (MODEL 1) Var iab le Unstand- Stand. 95% conf. i n t . (B) Stand- Stand. Zero- Part P a r t i a l ardized error ardized e r r o r order c o r r e l . c o r r e l . c o e f f i c . for B lower upper c o e f f i c . for c o r r e l . (B) (Beta) Beta LPRICE dependent LNWIDTH .267428 .0363 .196212 . 338644 .235387 .0319 .0404 .1765 .2835 SMALLLOT - .153187 .0239 - .200165 - .106209 -.175421 .0274 -.1622 -.1533 -.2487 ISTORIES .127870 .0191 .090348 .165393 .211084 . 0315 .4240 . 1602 .2592 IFIREPL .066920 .0229 .021940 .111899 .086950 .0298 .4015 .0699 . 1164 LSQFEET .126849 .0232 .081277 .172422 .170378 .0312 .3849 . 1309 .2141 I POOL .117394 .0375 .043686 .191101 .075479 .0241 . 1347 .0749 . 1244 LNOROOMS .131965 .0465 .040689 .223240 .120649 .0425 .5438 . 0680 . 1131 LTBATHS .134675 .0376 .060924 .208426 .140092 . 0391 .5449 .0859 . 1423 LAGE - .092947 .0114 - .115378 - .070516 -.331285 .0407 -.5876 -.1948 -.3102 NONMOVRS - .001489 .0005 - .002408 - .000570 -.082516 .0259 -.1542 -.0762 -.1266 A15 - .004654 .0016 -.007758 - .001550 -.076102 .0258 .0946 -.0705 -.1172 LDISTSCH - .101220 .0224 - .145281 - .057159 -.112565 .0250 -.1215 -.1080 -.1780 QI - .128336 .0220 -.171623 - .085049 -.182479 .0313 -.1139 -.1394 -.2273 Q2 - .043765 .0200 - .083018 - .004513 -.072247 .0330 -.0235 -.0524 -.0875 Q3 - .007066 .0212 - .048716 .034584 -.010646 .0320 -.0862 -.0080 -.0134 NEFLEVEL - .006484 .0016 - .009693 - .003276 -.099870 .0252 -.2804 -.0950 -.1571 Constant 10 .064584 .2450 9 .583404 10 .545765 Condit ion number bounds: 3.151, 443.347 TABLE 5.5 (cont'd) - RESULTS, RICHMOND SINGLE-DETACHED PROPERTIES WITH NEF 25 OR HIGHER, PREFERRED FORM MODEL WITH CONTINUOUS NOISE VARIABLE (MODEL 1) Var iab le Tolerance F S i g n i f . Unique Step of R* a f t er of F Rz change entry entry LPRICE dependent LNWIDTH .562531 54.380 .0000 .03117 2 .4452 SMALLLOT .763747 41.005 . 0000 .02350 7 . 5849 ISTORIES .576115 44.787 .0000 .02567 4 .5239 IFIREPL .647146 8.536 .0036 .00489 12 . 6310 LSQFEET .589931 29.878 . 0000 .01712 6 .5699 I POOL .984189 9.783 .0018 .00561 10 . 6210 LNOROOMS .317409 8.061 .0047 .00462 14 . 6398 LTBATHS .375554 12.860 .0004 .00737 3 . 4899 LAGE .345801 66.215 . 0000 .03795 1 .3453 NONMOVRS .852328 10.125 .0015 .00580 11 .6261 A15 .858135 8.671 .0034 .00497 13 .6351 LDISTSCH .920604 20.352 .0000 .01166 8 .6035 QI .583465 33.898 .0000 .01943 5 .5487 Q2 .526438 4.794 .0289 .00275 15 .6438 Q3 .561253 .111 .7391 .00006 16 .6435 NEFLEVEL .905037 15.749 .0001 .00903 9 .6147 Constant 1687.186 .0000 M u l t i p l e R: .80218 R square: .64350 A d j . R square: .63433 Standard e r r o r : .17454 Analys i s of Variance DF Sum of squares Regression 16 34.20428 Residual 622 18.94947 Mean square 2.13777 0.03047 F = 70.17037 S i g n i f . F = .0000 The p h y s i c a l c h a r a c t e r i s t i c s are by far the most important v a r i a b l e s , comprising nine of the s ixteen independent v a r i a b l e s i n the model. Four phys i ca l v a r i a b l e s - LAGE, LNWIDTH, LTBATHS, ISTORIES - among them account for over h a l f of the t o t a l v a r i a b i l i t y i n LPRICE. They are the f i r s t four v a r i a b l e s to enter (on the bas is of s i g n i f i c a n c e ) . Each contr ibutes great ly to the explanat ion of the unique v a r i a n c e . The values (for l i n e a r var iab le s ) and signs appear sens ib le for each v a r i a b l e , with a younger age house (LAGE) , a wider property (LNWIDTH) , more than one storey (ISTORIES), more square feet (LSQFEET), a l o t s i ze greater than 15000 square feet (SMALLLOT), more rooms (LN0R00MS), a pool (IPOOL), more bathrooms (LTBATHS) and one or more f i r e p l a c e s (IFIREPL), each p o s i t i v e l y a f f e c t i n g p r i c e . A l t e r n a t i v e cu to f f points for each of the SMALLLOT and IMOD dummies as we l l as mult ip le cutof f po ints for mult ip le dummies for l o t s i ze and house age were tes ted , as was a quadrat ic form for age, before s e l e c t i n g SMALLLOT and LAGE i n t h e i r f i n a l form. Also tested were p h y s i c a l i n t e r a c t i o n v a r i a b l e s , but these proved h igh ly i n s i g n i f i c a n t . Few of the (census) area c h a r a c t e r i s t i c s are important. Only two are included i n the f i n a l model, A15 and NONMOVRS, the s igns of which appear sens ib le . The higher the p r o p o r t i o n of the population aged 15 or l e s s , the lower i s p r i c e , r e f l e c t i n g perhaps the i n a b i l i t y to a f ford more expensive housing at t h i s stage of t h e i r l i f e c y c l e . The 130 higher the proport ion of the populat ion which has not moved i n the l a s t f ive years, the lower i s p r i c e , i n d i c a t i n g that more expensive houses (which are a lso general ly the newer houses) are more l i k e l y to be i n areas of higher turnover (to f i l l the new houses). The only p u b l i c sector v a r i a b l e se lected for t e s t i n g , LDISTSCH, i s important, suggesting that the further a house i s from a school (and i t s associated le isure/community/sports f a c i l i t i e s ) , the lower w i l l be p r i c e . A c c e s s i b i l i t y c h a r a c t e r i s t i c s , e i t h e r i n terms of d is tance or time, do not appear i n the f i n a l models. These rece ived p a r t i c u l a r a t tent ion because of the extremely low to lerance of some a c c e s s i b i l i t y v a r i a b l e s with the noise v a r i a b l e . T h e i r exc lus ion perhaps makes sense i n a s o c i a l c l imate where for any fami ly , mul t ip l e in teres t s i n employment, not to mention other commercial a c t i v i t y , may tend to make any l o c a t i o n a complicated compromise of a c c e s s i b i l i t y . Bender and Hwang (1985) expla in such i n s i g n i f i c a n t r e l a t i o n s h i p s between house p r i c e and d i s t a n c e / t r a v e l time to the CBD by i l l u s t r a t i n g the s i g n i f i c a n c e of secondary employment centres when introduced to t h e i r model. That ne i ther of the a i r p o r t a c c e s s i b i l i t y v a r i a b l e s are anywhere near s i g n i f i c a n t i n t h i s study, suggests that the v a r i a t i o n i n d is tances from Richmond for those employed at the a i r p o r t i s 131 l e s s than i n e a r l i e r studies where such has been important -even though 25 percent of Richmond workforce works on Sea Is land (Transport Canada 1989). The q u a r t e r l y dummies capture the e f f ec t of seasonal f l u c t u a t i o n and perhaps i n f l a t i o n , exp la in ing more v a r i a b i l i t y than d id the month v a r i a b l e . Only January to March (QI) and A p r i l to June (Q2) appear to be s i g n i f i c a n t l y d i f f e r e n t from the res t of the year. Note that the i n t e r p r e t a t i o n of the c o e f f i c i e n t s for dummy v a r i a b l e s i n t h i s model (using a semi- logar i thmic equation) i s not the same as for continuous l i n e a r v a r i a b l e s . For example a quar ter ly dummy c o e f f i c i e n t of -.128 a c t u a l l y impl ies a r e l a t i v e p r i c e e f fect of about - .120 (Halvorsen and Palmquist 1980, p.475). For "smaller" c o e f f i c i e n t s , the r e l a t i v e e f f ec t i s l i t t l e d i f f e r e n t . Otherwise, p o s i t i v e c o e f f i c i e n t s have larger r e l a t i v e e f f e c t s , and negative c o e f f i c i e n t s have a l g e b r a i c a l l y l a r g e r and abso lute ly smal ler r e l a t i v e e f f ec t s . Other s t a t i s t i c s reported include the s tandardized regress ion c o e f f i c i e n t s (Beta) and t h e i r standard errors for each independent v a r i a b l e . These g ive some i n d i c a t i o n of the r e l a t i v e (not absolute) importance of the c o e f f i c i e n t s , although l i k e the B c o e f f i c i e n t s , these are contingent on 132 other independent var iab le s in the equation and are affected by c o r r e l a t i o n s . Tests of the unique contr ibut ion of a v a r i a b l e against the n u l l hypothesis for the B, Beta, p a r t i a l or semipart ia l c o r r e l a t i o n s s t a t i s t i c s , a l l use the i n d i v i d u a l F value and i t s s i g n i f i c a n c e l e v e l . This s i g n i f i c a n c e t e s t i s only s e n s i t i v e to unique variance that a p a r t i c u l a r independent v a r i a b l e adds to R 2 . Thi s i s i l l u s t r a t e d i n the unique c o n t r i b u t i o n of each independent v a r i a b l e reported as the (unique) R2 change for the exclusion of the B value for the complete model. In t o t a l , the independent v a r i a b l e s account for .2116 of the v a r i a b i l i t y uniquely . Shared or common v a r i a b i l i t y (the balance of R2) explained by the var iab le s i s .4319. Note again that these R2 changes are not a measure of the proport ion of unexplained v a r i a t i o n that an independent v a r i a b l e contr ibutes . T e c h n i c a l l y , these R2 changes are the square of the s e m i p a r t i a l (part) c o r r e l a t i o n s which measure the c o r r e l a t i o n between LPRICE and an independent v a r i a b l e when l i n e a r e f f ec t s of the other independent v a r i a b l e s are removed from p a r t i c u l a r independent v a r i a b l e s . A l so l i s t e d i n Table 5.5 are the p a r t i a l c o r r e l a t i o n s which measure the c o r r e l a t i o n between LPRICE and an independent v a r i a b l e when the l i n e a r e f f ec t s of other independent v a r i a b l e s are removed from the p a r t i c u l a r independent v a r i a b l e and LPRICE. 133 An i n d i c a t i o n of the importance of var iab les i s reported i n the order of entry (on the basis of smallest p r o b a b i l i t y of F) and the accompanying cumulative R z s . These show how LAGE alone accounts for some 34 percent of t o t a l var iance , or more than h a l f of the t o t a l v a r i a b i l i t y explained by the model. Other s t a t i s t i c s reported i n Table 5.5 suggest that there are no c o l l i n e a r i t y problems with the v a r i a b l e of concern, NEFLEVEL, or with other independent v a r i a b l e s . The tolerances measure the proport ion of v a r i a b i l i t y i n a p a r t i c u l a r independent v a r i a b l e not explained by other independent v a r i a b l e s . I f very low, the to lerance ind ica tes excessive interdependence between t h i s and other independent v a r i a b l e s . The inverse of these tolerance f igures are variance i n f l a t i o n fac tors (VIF). VIFs measure how much the variances of the estimated regress ion c o e f f i c i e n t s are i n f l a t e d as compared to when the independent v a r i a b l e s are not l i n e a r l y r e l a t e d - although these factors cannot d i s t i n g u i s h between severa l simultaneous m u l t i c o l l i n e a r i t i e s (Neter, et a l . 1985, p.393) . The upper and lower bound of the condi t ion numbers are less than 1000, r e i n f o r c i n g the view that m u l t i c o l l i n e a r i t y i s not l i k e l y to be a problem i n t h i s model. 134 Model 1 appears to provide an i n t u i t i v e l y and s t a t i s t i c a l l y reasonable explanation of the determination of property values and i n p a r t i c u l a r the ro l e of a i r p o r t noise l e v e l . Notwithstanding t h i s conclus ion, a l t ernat ive func t iona l forms were tes ted . These include the fo l lowing funct iona l forms commonly used i n previous models: . regress ing LPRICE against natural logarithms of a l l independent v a r i a b l e s except NEFLEVEL and the i n d i c a t o r v a r i a b l e s (c loser to the t h e o r e t i c a l form der ived i n the model design) (model 2) . . using a semilogarithmic form regress ing LPRICE against a l l l i n e a r independent var iab le s (model 3) . . using a l i n e a r form regress ing PRICE against a l l l i n e a r independent v a r i a b l e s (model 4) . Table 5.6 summarizes models 1, 2, 3 and 4. I t may be noted that compared with model 1, R2 f luctuates somewhat for some of the a l t e r n a t i v e models. Compared with the NEFLEVEL value of -.006498 i n model 1, model 2 has a value of -.006057, model 3 has a value of -.006776 and, convert ing to percent change for the mean house p r i c e , model 4 has a value of -.006438. 135 TABLE 5.6 - UNSTANDARDIZED COEFFICIENTS, RICHMOND SINGLE-DETACHED PROPERTIES WITH NEF 2 5 OR HIGHER, ALTERNATIVE FORM MODELS WITH CONTINUOUS NOISE VARIABLE" (MODELS 1, 2, 3, 4) V a r i a b l e s Model ( funct ional form) 1 2 3 4 (preferred) (mixed) (semilog) ( l inear) Dependent LPRICE LPRICE LPRICE PRICE Phys i ca l LNWIDTH NWIDTH SMALLLOT ISTORIES IFIREPL LSQFEET SQFEET I POOL LNOROOMS NOROOMS LTBATHS TBATHS LAGE AGE Area NONMOVRS LNONMOVR A15 LA15 267428 153187 127870 066920* 126849 117394* 131965* 134675 092947 001489* 004654* P u b l i c sector LDISTSCH -.101220 DISTSCH Other QI Q2 Q3 Noise NEFLEVEL -.128336 -.043765* -.007066ns 264949 155995 128872 065000* 127255 118887* 131303* 133056 094790 -.006484 Constant 10.064584 -.040563 -.077200* -.115179 -.126561 -.041072* -.005402ns -.006057 10.300316 .003528 .165595 .120342 .060925* .000070 .105477* .019407 .034798* .005947 .001817 .005833 -.043362 -.128841 -.044972* -.005106ns -.006776 11.880837 489.81 •24919.67 18238.63 1624.69ns 9.87 20583.85 3470.10 4408.41* -805.85 -202.65* -1028.81 -7231.92 -18752.74 -6946.48* -2205.38ns -875.12 c 152382.52 136 TABLE 5.6 (cont'd) - UNSTANDARDIZED COEFFICIENTS, RICHMOND SINGLE-DETACHED PROPERTIES WITH NEF 25 OR HIGHER, ALTERNATIVE FORM MODELS WITH CONTINUOUS NOISE VARIABLE8 (MODELS 1, 2, 3, 4) Adjusted R2 .63433 .63673 .62504 .57154 Notes: a a l l independent var iab le s are s i g n i f i c a n t at p < .001, unless i d e n t i f i e d by * i f s i g n i f i c a n t at p = .05, or by ns i f not s i g n i f i c a n t at p = .05. b converting to percent change for the mean house p r i c e , NEFLEVEL would have the value -.006438. There appears to be more s t a t i s t i c a l j u s t i f i c a t i o n i n us ing the pre ferred model form reported i n Table 5.5 (model 1), even though i t does not s t r i c t l y follow the t h e o r e t i c a l i d e a l . This model i s at l eas t as f a i t h f u l to the i d e a l as a l t e r n a t i v e s reported here i n Table 5.6 (and used i n previous s tud ie s ) . As e a r l i e r noted, a l t e r n a t i v e ways of inc lud ing noise i n the model are tes ted . The r e s u l t s for several a l t e r n a t i v e models (with the same other independent v a r i a b l e s as i n model 1) are summarized i n Table 5.7 which presents for each model the adjusted R2 and key s t a t i s t i c s for the noise v a r i a b l e . Although the amount of v a r i a b i l i t y explained by a l t e r n a t i v e continuous model 6 i s only marginal ly d i f f e r e n t from the p r e f e r r e d model 1, model 5 using zone dummies does capture s l i g h t l y more. This increase i n adjusted R2 i s to be 137 TABLE 5.7 - RESULTS, RICHMOND SINGLE-DETACHED PROPERTIES WITH NEF 25 OR HIGHER, PREFERRED FORM MODELS WITH ALTERNATIVE NOISE VARIABLES8 (MODELS 1, 5, 6) Model Noise v a r i a b l e and form Adjusted R2 B(noise) 95% confidence i n t e r v a l (B) S ig F 1" continuous NEFLEVEL (NEF 2 5 , . . . , 4 1 ) .63433 -.006484 -.009693 -.003276 .0001 5b three zone dummies INZONE25 (NEF 25 to 29) INZONE30 (NEF 30 to 34) INZONE35 (NEF 35 to 39) .64075 .130976 .095366 .147174 .079957 .040876 .079934 .181996 .149856 .214415 .0000 .0006 .0000 6b continuous NZONE (NEF 29,34,39,44) .63419 -.006012 -.009009 -.003015 .0001 Notes: a a l l noise var iab le s are s i g n i f i c a n t at p < .001. b model 1 (the preferred model, reported i n Table 5.5) regresses LPRICE against LNWIDTH, SMALLLOT, ISTORIES, IFIREPL, LSQFEET, IPOOL, LNOROOMS, LTBATHS, LAGE, NONMOVRS, A15, LDISTSCH, QI, Q2, Q3, NEFLEVEL; other models d i f f e r only with respect to the subs t i tu t ion of a l t e r n a t i v e noise v a r i a b l e ( s ) . expected given the use of three v a r i a b l e s for noise rather than one. Use of the a l t e r n a t i v e continuous noise v a r i a b l e i n model 6 (where propert ies are grouped and a l loca ted the highest NEF value for the group) does r e s u l t i n a lower noise c o e f f i c i e n t . This i s perhaps because of reduced v a r i a t i o n i n the noise v a r i a b l e when us ing four instead of seventeen values . The zero-order c o r r e l a t i o n c o e f f i c i e n t for the continuous NZONE v a r i a b l e used i n model 6 i s - .272, only marginal ly lower than the -.280 c o r r e l a t i o n for the continuous NEFLEVEL v a r i a b l e used i n model 1. Note that t h i s cruder continuous (scaled) v a r i a b l e form was most commonly used i n previous s tudies . The c o e f f i c i e n t s for the noise zone dummy v a r i a b l e s used i n Model 5 (where the base zone i s NEF 40 plus) suggests the fo l lowing pattern of n o i s e - p r i c e r e l a t i o n s h i p s which d i f f e r between zones: c e t e r i s par ibus , p r i c e s w i l l be about the same i n the NEF 25 to 29 and NEF 35 to 39 zones, higher i n the NEF 3 0 to 34 zone and even higher i n the NEF 40 plus zone. This impl ies that the n o i s e - p r i c e r e l a t i o n s h i p i s not l i n e a r , a f ind ing of some of the previous s tud ies . This may be the case, although l i t t l e confidence i s p laced i n these p a r t i c u l a r r e s u l t s because of: . the small number of cases which are i n the NEF 35 to 39 zone (and which are biased to the lower end of the range) . 139 . the a r b i t r a r y a l l o c a t i o n of the a l loca ted NEF l e v e l s i n the f i r s t place (several NEF contours may cross a s i n g l e i d e n t i f i e r , r e q u i r i n g the a r b i t r a r y a l l o c a t i o n of one NEF l eve l which may "misallocate" many proper t i e s to an adjacent noise zone). . the a r b i t r a r y zone borders and widths (d i f ferent r e s u l t s were obtained when experimenting with the b o r d e r s ) . I f more confidence could be p laced i n the a l l oca ted noise l e v e l s than i n the s ing le NEF l e v e l , more sophi s t i ca ted border s e l e c t i o n might be v a l i d , but i n the circumstances t h i s i s not so; th i s encourages use of a continuous model which although g iv ing jus t one ( l inear) r e l a t i o n s h i p , i s most representat ive of the s i t u a t i o n . 5.6 THE "ALL" MODEL Adding the 894 cases with noise l e v e l s below the NEF 25 thresho ld to the 645 cases of the IN model gives the t o t a l sample of 1539 s ingle-detached proper t i e s which are used i n the ALL model. As e a r l i e r noted, two vers ions are tes ted for continuous v a r i a b l e s . The f i r s t , reported as Model 7, i s more s t r i c t l y correct i f one accepts the noise threshold 140 l e v e l of NEF 25, below which a l l propert ies should be a l l o c a t e d to NEF 24. The second, reported as model 8, experiments by more a r b i t r a r i l y a l l o c a t i n g propert i e s below the threshold to NEF 20. This second approach e f f e c t i v e l y impl ies that propert i e s between NEF 20 and 24 are above the ambient noise l e v e l (and that the threshold NEF l e v e l should be at NEF 20) but introduces some d i s t o r t i o n by a l l o c a t i n g them a l l to NEF 20. A l t e r n a t i v e l eve l s to NEF 24 other than NEF 20 were a l so tested but are not reported . The r e s u l t s for model 7 which assumes an NEF 25 threshold and a l loca tes a l l other propert i e s to NEF 24 are examined i n d e t a i l before being compared with the a l t e r n a t i v e continuous model (model 8) and a l t e r n a t i v e noise v a r i a b l e s . The zero-order c o r r e l a t i o n of NEFLEV24 with the dependent v a r i a b l e LPRICE i s -.1906 (compared with -.1677 for NEFLEV20) suggesting - as i n the IN model - tha t , ignoring a l l other r e l a t i o n s h i p s , the higher the noise l e v e l , the lower the p r i c e . A simple c o r r e l a t i o n matrix of a l l v a r i a b l e s i n the f i n a l ALL models i s presented as Table C.2 i n Appendix C. The tab le shows that c o r r e l a t i o n s of NEFLEV24 with other independent v a r i a b l e s i n the f i n a l continuous ALL model 8 range ( in absolute terms) from -.007 (IPOOL) to .0196 (LAGE). The c o r r e l a t i o n s of other independent v a r i a b l e s with LPRICE range ( in absolute terms) from -.013 (Q2) to -.6309 (AGE). 141 The r e s u l t s of the f i n a l continuous ALL model, using mul t ip l e regression with pairwise de l e t ion of missing va lues , are presented i n Table 5.8. The adjusted R2 value of .638 suggests that the continuous ALL model explains marginal ly more of the t o t a l v a r i a b i l i t y i n LPRICE than does the continuous IN model (with an adjusted R2 value of .634). The noise v a r i a b l e , NEFLEV24 has a B (unstandardized) c o e f f i c i e n t of -.003921, implying a one un i t increase i n noise r e s u l t s i n under h a l f of a one percent decrease i n property p r i c e which i s l ess than that impl ied by the IN model. The 95 percent confidence i n t e r v a l for t h i s v a r i a b l e i s -.0059 to - .0019, compared with -.0097 to -.003 3 for the IN model. Both models thus suggest a lower bound of around one quarter of one percent, with the IN model suggesting a higher upper bound than the ALL model. Model 7 has the same independent v a r i a b l e s as model 1, a l l of which have B c o e f f i c i e n t s with the same signs and i n most cases roughly the same magnitudes. S i m i l a r experimentation to that reported i n the IN r e s u l t s was undertaken for the ALL models. In t o t a l , the independent v a r i a b l e s i n model 7 account for .17533 of the v a r i a b i l i t y uniquely . Shared or common v a r i a b i l i t y (the balance of R2) explained by the v a r i a b l e s i s .46615. LAGE alone accounts for some 40 percent of t o t a l var iance , or over 60 percent of the t o t a l v a r i a b i l i t y explained by the model. Other s t a t i s t i c s 142 TABLE 5.8 - RESULTS, ALL RICHMOND SINGLE-DETACHED PROPERTIES, PREFERRED FORM MODEL WITH CONTINUOUS NOISE VARIABLE (MODEL 7) V a r i a b l e Unstand- Stand. 95% conf. i n t . (B) Stand- Stand. Zero- Part P a r t i a l ardized error ardized e r r o r order c o r r e l . c o r r e l . c o e f f i c . for B lower upper c o e f f i c . for c o r r e l . (B) (Beta) Beta LPRICE dependent LNWIDTH .251813 .0243 .204071 .299555 .231350 .0224 -.1215 . 1608 .2594 SMALLLOT - .115771 .0191 - .153286 - .078256 -.107515 . 0178 -.0502 -.0941 -.1552 ISTORIES .087135 .0122 .063217 .111054 .155930 .0218 .4677 .1111 . 1824 IFIREPL .069269 .0159 .038163 .100376 .081488 .0187 .3736 .0679 .1127 LSQFEET .113356 .0145 .084994 .141717 .155832 . 0199 .3138 . 1219 . 1994 I POOL .070969 .0237 .024389 .117548 .046939 .0157 .0847 .0465 .0773 LNOROOMS .185599 .0286 .129594 .241604 .178502 . 0275 .6064 .1010 . 1664 LTBATHS .164382 .0223 .120560 .208204 .178629 .0243 .6033 . 1144 . 1876 LAGE - .090990 .0068 - .104324 - .077656 -.375020 . 0280 -.6391 -.2080 -.3282 NONMOVRS - .001644 .0003 - .002269 - .001020 -.088521 .0171 -.1838 -.0803 -.1329 A15 - .006440 .0011 - .008588 - .004291 -.096034 .0163 .0355 -.0914 -.1509 LDISTSCH - .074240 .0143 - .102342 - .046138 -.085432 .0165 -.1400 -.0805 -.1333 QI - .098702 .0135 - .125193 - .072212 -.146795 .0201 -.1091 -.1136 -.1864 Q2 - .036300 .0123 - .060365 - .012236 -.062579 .0211 -.0134 -.0460 -.0766 Q3 - .012255 .0128 - .037325 .012815 -.019908 .0208 .0806 -.0149 -.0249 NEFLEV24 - .003921 .0010 - .005925 - .001917 -.063808 .0166 -.1906 -.0597 -.0991 Constant 9 .980048 .1510 9 .683834 10 .276261 Condit ion number bounds: 3.249, 450.367 TABLE 5.8 (cont'd) - RESULTS, ALL RICHMOND SINGLE-DETACHED PROPERTIES, PREFERRED FORM MODEL WITH CONTINUOUS NOISE VARIABLE (MODEL 7) V a r i a b l e Tolerance F S i g n i f . Unique Step of R2 a f t er of F R2 change entry entry LPRICE dependent LNWIDTH .483167 107.043 .0000 .02586 2 .4827 SMALLLOT .765822 36.643 .0000 .00885 9 .6184 ISTORIES .507398 51.066 .0000 .01234 7 .6057 IFIREPL .694162 19.080 .0000 .00461 11 .6285 LSQFEET .611504 61.466 .0000 .01485 5 .5783 I POOL .979405 8.932 .0028 .00216 15 .6413 LNOROOMS .320400 42.257 .0000 .01021 4 .5607 LTBATHS .409921 54.141 .0000 .01308 3 .5421 LAGE .307769 179.166 .0000 .04328 1 .4085 NONMOVRS .822033 26.663 .0000 .00644 12 .6332 A15 .905683 34.574 .0000 .00835 10 .6209 LDISTSCH .888889 26.854 .0000 .00649 8 .6120 QI .598880 53.418 . 0000 .01291 6 .5927 Q2 .540147 8.756 .0031 .00212 14 .6392 Q3 .560453 .919 .3378 .00022 16 .6415 NEFLEV24 .874222 14.733 .0001 .00356 13 .6369 Constant 4367.764 .0000 M u l t i p l e R: .80093 Analys i s of Variance R square: .64168 DF Sum of squares Mean square A d j . R square: .63762 Regression 16 71.85629 4.49102 Standard e r r o r : .16450 Residual 1484 40.15984 0.02706 F = 165.95361 S i g n i f . F = .0000 reported i n Table 5.8 suggest that there are no problems of ser ious m u l t i c o l l i n e a r i t y invo lv ing the noise or other v a r i a b l e s . The same comments regarding the j u s t i f i c a t i o n for using the f u n c t i o n a l form i n model 1 apply to model 7. Various ways of inc lud ing noise i n the model are tested and the r e s u l t s f or several a l t e r n a t i v e models are summarized i n Table 5.9 which presents for each, the adjusted R2 and key s t a t i s t i c s f o r the noise v a r i a b l e . Table 5.9 shows that the continuous model 8 (using NEFLEV20) has a s l i g h t l y lower adjusted R2 va lue , and a much lower (absolutely) noise c o e f f i c i e n t - which makes sense given the reduced NEF l e v e l value a l l o c a t e d to more than h a l f the cases . Model 7 i s pre ferred because i t more c o n s i s t e n t l y (though s t i l l somewhat a r b i t r a r i l y ) a l loca tes a l l proper t i e s located outside the noise threshold of NEF 25, noise values of NEF 24. The same absolute reduct ion i n the noise c o e f f i c i e n t i s evident when comparing the a l t e r n a t i v e continuous models 10 and 11 (where propert ies are grouped and a l l o c a t e d the highest NEF value for the group). As i n the IN models, these a l t e r n a t i v e continuous models exp la in v i r t u a l l y the same amount of v a r i a b i l i t y , but with lower noise c o e f f i c i e n t s ; again, the zero-order c o r r e l a t i o n c o e f f i c i e n t s for the continuous NZONE24 AND NZONE20 v a r i a b l e s i n models 10 and 11 are lower than the continuous 145 TABLE 5.9 - RESULTS, ALL RICHMOND SINGLE-DETACHED PROPERTIES, PREFERRED FORM MODELS WITH ALTERNATIVE NOISE VARIABLES" (MODELS 7, 8, 9, 10, 11) Model Noise var iab l e Adjusted and form R2 B(noise) 95% confidence i n t e r v a l (B) S ig F 7b continuous .63762 NEFLEV24 (NEF 24 ,25 , . . . , 41 ) -.003921 -.005925 - .001917 . 0001 8C continuous .63650 NEFLEV20 (NEF 20 ,25 , . . . , 41 ) -.002431 -.003930 - .000931 .0015* 9 e four zone dummies .64108 INZONE24 (NEF 24 or less) INZONE25 (NEF 25 to 29) INZONE30 (NEF 30 to 34) INZONE35 (NEF 35 to 39) .099166 .116613 .066214 .118922 .054476 .069786 .017884 .057908 .143855 .163440 .114544 .179937 .0000 . 0000 .0073* .0001 10 f continuous .63705 NZONE24 (NEF 24,29,34,39,44) -.002893 -.004506 - .001281 . 0004 l l 8 continuous .63631 NZONE20 (NEF 20,29,34,39,44) -.001932 -.003171 - .000692 .0023* Notes: a a l l noise var iab le s are s i g n i f i c a n t at p < .001, unless the s i g of F has an * . b model 7 (the preferred model, reported i n Table 5.8) regresses LPRICE against LNWIDTH, SMALLLOT, ISTORIES, IFIREPL, LSQFEET, IPOOL, LNOROOMS, LTBATHS, LAGE, NONMOVRS, A15, LDISTSCH, QI, Q2, Q3, NEFLEV24; other models d i f f e r only with respect to the subs t i tu t ion of a l t e r n a t i v e noise v a r i a b l e ( s ) . NEFLEV24 and NEFLEV2 0 var iab le s used in models 7 and 8, though more so than i s the case with the IN models. This makes sense given the large number of extra cases i n the ALL model having no v a r i a t i o n i n NEF l e v e l . Model 9 using zone dummies, captures s l i g h t l y more of the v a r i a b i l i t y compared to the continuous models. The c o e f f i c i e n t s f or the noise zone dummy v a r i a b l e s suggest the fo l lowing pat tern of no i s e -pr i ce r e l a t i o n s h i p s which d i f f e r between zones: c e t e r i s par ibus , pr ices w i l l be about the same i n the area outside the NEF 25 threshold , the NEF 25 to 29 and NEF 35 to 39 zones. As with the IN models, t h i s suggests that the n o i s e - p r i c e re la t ionsh ip i s not l i n e a r . Of p a r t i c u l a r i n t e r e s t i s the suggestion that i n two of the four noise zones the r e l a t i o n s h i p i s not s i g n i f i c a n t l y d i f f e r e n t from the zone outside the no i se -af fec ted area . Th i s only confirms the concerns out l ined i n the d i s cus s ion of a l t e r n a t i v e noise v a r i a b l e s for the IN models; for the same reasons, a f t e r experimentation with the borders , the continuous model i s again considered to be the most representat ive . 5.7 FURTHER DISCUSSION Model 1 remains the pre ferred model because i t only considers proper t i e s in s ide the NEF 25 l e v e l - thus avoiding 147 the problem of def in ing the exact threshold, and uses a continuous noise v a r i a b l e - thus avoiding the need to accurate ly ident i fy zone borders (although acknowledging that the implied l i n e a r r e l a t i o n s h i p i s probably not accurate ) . Model 1 suggests that the pr i ce -no i se r e l a t i o n s h i p , or the Noise Depreciat ion S e n s i t i v i t y Index (NDSI) i s - .0065; more important ly , the NDSI l i e s somewhere i n the 95 percent confidence i n t e r v a l of -.0033 to - .0097. This IN model 1 confidence i n t e r v a l includes the upper l i m i t s of the i n t e r v a l s for the ALL continuous model 7 and for the a l t e r n a t i v e IN continuous model 6. The range may thus be considered to be very l i k e l y to include the ac tua l va lue . I f anything, the value i s l i k e l y to be at the lower end of the range. The model l ing es tabl i shes a r e l a t i o n s h i p between proper t i e s exposed to d i f f e r e n t NEF l e v e l s and p r i c e i n 1987. I t i s reasonable to expect that t h i s can be appl ied i n l a t e r per iods assuming a l i n e a r r e l a t i o n s h i p . In the absence of a continuous r e l a t i o n s h i p , however, l e s s confidence could be placed i n such an a p p l i c a t i o n i f noise l eve l s had been c o n s i s t e n t l y changing ( s t ead i ly or otherwise) across Richmond (due to t echno log ica l and other changes a f f e c t i n g a i r p o r t noise and/or changes i n other community noise sources ) . Such cons i s tent a l t e r a t i o n i n noise l e v e l s might a l t e r the impact of any p a r t i c u l a r NEF l e v e l on p r i c e i n 148 absolute terms, but not r e l a t i v e to d i f f e r e n t NEF l e v e l s (which i s what the continuous model measures). The IN model does not i d e n t i f y or tes t for the noise thresho ld . The model assumes that i t i s no higher than NEF 25, by using a continuous noise v a r i a b l e with values ranging from NEF 25 upwards i n the IN model. 5.8 CENSUS MODEL As discussed e a r l i e r , model type B using census data i s the approach taken i n many of the reviewed previous s tud ies . The r e s u l t s of t e s t i n g of t h i s model type for the Richmond area are now reported . The independent v a r i a b l e s tested include the same area , p u b l i c sector , a c c e s s i b i l i t y , noise and other c h a r a c t e r i s t i c s as i n the i n d i v i d u a l model (type A ) , although these are averaged for each enumeration area (EA) by taking the mean values of each of these c h a r a c t e r i s t i c s for a l l sales i n an EA ( e f f e c t i v e l y weighting) . Model type B d i f f e r s from model type A i n two ways: . by using census enumeration area averages for the dependent and a l l independent v a r i a b l e s . 149 . by using census data for the only independent v a r i a b l e for phys ica l c h a r a c t e r i s t i c s and the dependent v a r i a b l e . For model type B Chapter 4 proposed using census data for average number of rooms per dwel l ing and average dwel l ing va lue , each at the enumeration area l e v e l . However, these p a r t i c u l a r enumeration l e v e l data were s t i l l not a v a i l a b l e at the time of the study, and for each EA, these two v a r i a b l e s were ca lcu la ted from Assessment Author i ty i n d i v i d u a l data using census-weighted (by proport ion of s ingle-detached houses i n an EA) number of rooms and dwel l ing value , r e spec t ive ly . Pre l iminary ana lys i s of the data, s i m i l a r to that reported for the i n d i v i d u a l data, was undertaken for the data averaged for enumeration area . Transformations were l e ss complicated because of the absence of phys i ca l c h a r a c t e r i s t i c s . Assumptions regarding the funct iona l form are more important given the small number of EAs (which determine the sample s i z e ) . Again , the choice was made between the l i n e a r and natura l logar i thmic form for each v a r i a b l e . As with the i n d i v i d u a l data , model type B was tes ted for two areas: 150 . the IN model using 48 EAs (cases) which contained any p r o p e r t i e s exposed to NEF 25 or higher (even i f not a l l of an EA was ins ide the noise zone). . the ALL model, using a l l 97 EAs i n the Richmond study area . The NEF l e v e l for any p a r t i c u l a r EA was obtained by taking the mean of the NEF values for a l l proper t i e s so ld i n the EA. In so doing each EA's NEF i s i m p l i c i t l y weighted by the d i s t r i b u t i o n of sa les . As in the modell ing for type A, continuous and dummy noise v a r i a b l e s , determined s i m i l a r l y , are t e s t ed . Model l ing revealed high m u l t i c o l l i n e a r i t y between area v a r i a b l e s based on census data and between noise and some a c c e s s i b i l i t y v a r i a b l e s (as i n the i n d i v i d u a l model l ing) . The r e s u l t s for a l l model v a r i e t i e s tes ted (regardless of noise v a r i a b l e form, funct ional form, IN or ALL) c o n s i s t e n t l y ind ica te that most of the v a r i a b i l i t y i n LCWPRICE (the natura l logarithm of census-weighted sales p r i c e + 1) i s explained by LCWROOMS (the n a t u r a l logarithm of census-weighted rooms + 1). In fact the only other s i g n i f i c a n t independent var iab l e i s the noise v a r i a b l e . Such a s i m p l i s t i c model i n which l ess than h a l f the v a r i a b i l i t y i n average dwell ing value i s explained (adjusted R2 i s l e s s than .5 ) , and predominantly by one v a r i a b l e 151 (average number of rooms) i s not considered comprehensive enough for the purpose of i d e n t i f y i n g the p r i c e - n o i s e r e l a t i o n s h i p . The unstandardized c o e f f i c i e n t for noise v a r i a b l e s i s less than h a l f that recorded for the same v a r i a b l e s i n model type A, but t h i s value i s use less . The overwhelming importance of LCWROOMS makes sense given the dominance of phys ica l v a r i a b l e s i n the i n d i v i d u a l model l ing using type A. As the only phys i ca l v a r i a b l e "avai lable" (and even then a proxy has had to be used) for t e s t i n g i n the type B model, i t s entry i s to be expected. The unimportance of other v a r i a b l e s , inc lud ing those which are s i g n i f i c a n t i n the i n d i v i d u a l s ingle-detached model l ing, i s perhaps explained by the use of averaging across EAs. A l l the model l ing emphasizes the importance of p h y s i c a l v a r i a b l e s which are not a v a i l a b l e from census data for type B models. This lack of p h y s i c a l data i s unfortunate given the f i n a n c i a l a t tract iveness of us ing census data rather than i n d i v i d u a l data to determine the n o i s e - p r i c e r e l a t i o n s h i p . 152 CHAPTER 6 - APPLICATION AND CONCLUDING COMMENTS 6.1 INTRODUCTION T h i s concluding chapter f i r s t comments on the i n t e r p r e t a t i o n of the hedonic model l ing. Discuss ion of a p p l i c a t i o n leads to an i l l u s t r a t i o n of the usefulness of the Noise Deprec iat ion S e n s i t i v i t y Index by attempting a pre l iminary a p p l i c a t i o n i n the s e t t ing of Vancouver Internat iona l A i r p o r t . F i n a l l y , general comments are made regarding the value of the modell ing along with suggestions for further research i n the context of Vancouver. 6.2 INTERPRETATION The model developed i n t h i s paper should not be used for p r e d i c t i o n of house p r i c e s . Rather, the c o e f f i c i e n t for noise y i e l d s the Noise Depreciat ion S e n s i t i v i t y Index, a marginal va lue . For marginal changes, t h i s value should perhaps be considered a minimum est imate. Pearce (1976, p.37) argues that most of the d e f i c i e n c i e s suggest undervaluat ion rather than overvaluat ion of noise nuisance cos t s . Abelson and Markandya (1985) consider the bias of such estimates for environmental v a r i a b l e s when the future i s not s t a t i c . They conclude that biases are l i k e l y us ing 153 e i t h e r l i n e a r or l o g - l i n e a r models, with the regress ion form and the nature of changes i n the s p a t i a l pat tern of the environmental c h a r a c t e r i s t i c determining the d i r e c t i o n of the b i a s . I f the future noise patern i s presumed to be l i t t l e d i f f e r e n t from the present, then regardless of f u n c t i o n a l form there i s no b ia s . B lay lock (1977, p.23) makes an important po int regarding the i n t e r p r e t a t i o n of the hedonic r e s u l t s : an unfortunate tendency, developed i n the hedonic l i t e r a t u r e , of r e f e r r i n g to the estimated marginal d o l l a r spent on a hedonic c h a r a c t e r i s t i c as the p r i c e of that c h a r a c t e r i s t i c s has been taken up i n the hedonic studies of housing. Such a misunderstanding of the i n t e r p r e t a t i o n of the hedonic marginal valuations can under c e r t a i n circumstances r e s u l t i n t h e i r misuse i n est imating d e p r e c i a t i o n due to noise. The problem i s that the hedonic re la t ionsh ips can be considered a p r i c e i n the c l a s s i c a l sense only under very r e s t r i c t i v e circumstances. B l a y l o c k ' s (1977, p.24) suggestion that the term "discount rate due to noise" may be the best d e s c r i p t i o n , i s p r e f e r r e d ; hence the use i n t h i s paper of the Noise Deprec ia t ion S e n s i t i v i t y Index. The noise c o e f f i c i e n t i s best understood as the response at the margin to supply and demand fac tors i n a p a r t i c u l a r housing market. The c o e f f i c i e n t r e f l e c t s a market e q u i l i b r a t i n g p r i c e or opportunity locus i n a p a r t i c u l a r supply-demand s i t u a t i o n , making i t inappropr ia te for use i n 154 va lu ing non-marginal changes (McMillan et a l . 1980, p.315). The r e s u l t contains only l i m i t e d wi l l ingness - to -pay informat ion: no wi l l ingness - to -pay or marginal va luat ion schedule i s i d e n t i f i e d (Nelson 1981a, p .70) . Abelson (1979, pp 27-28) s tates : hedonic pr ices may not represent wi l l ingness to pay amenity values because markets are imperfect or i n d i s e q u i l i b r i u m or because preferences vary and average values do not equal marginal va lues . One r e s u l t of t h i s i s that hedonic p r i c e s may vary between areas. Most of these d i f f i c u l t i e s can be reso lved and relevant wi l l ingness to pay pr i ce s estimated with p l a u s i b l e assumptions about market imperfections and d i s e q u i l i b r i u m e f f ec t s , and with survey data about consumer preferences . A l t e r n a t i v e l y , i f the hedonic pr i ce s represent e q u i l i b r i u m s i t u a t i o n s , w i l l ingness to pay p r i c e s may be ca lcu la ted with the a i d of a Freeman-type two-step econometric model to estimate the demand curve for amenities, the f i r s t step of which i s obta in ing amenity p r i c e s . . . Out of a l l t h i s , i t may be concluded that although the hedonic p r i c e s obtained i n the modell ing r e f l e c t equ i l ibr ium w i l l i n g n e s s to pay, t h e o r e t i c a l l y they are able to be used d i r e c t l y only under very r e s t r i c t e d circumstances. Nonetheless, the regress ion c o e f f i c i e n t s may be used to approximately ca lcu la te aggregate marginal e f f ec t s of a change i n noise l e v e l s . True s o c i a l benef i ts w i l l not be much d i f f e r e n t , i f such changes are not subs tant ia l and i f households do not re locate (Nelson 1978a, p.78; see a lso McMil lan et a l . 1980, p.325) . 155 6.3 APPLICATION The r e s u l t s from modelling such as that undertaken i n t h i s paper may be appl ied to r e a l s i t u a t i o n s invo lv ing changes i n a i r p o r t noise . This i s t y p i c a l l y done e i t h e r by undertaking a b e n e f i t - c o s t analys is (BCA) of noise abatement projec t s i n t h e i r own r i g h t (eg. see Nelson 1978a, NRC 1977), or as part of a bene f i t - cos t evaluation of a p a r t i c u l a r a i r p o r t i n f r a s t r u c t u r e development (such as c u r r e n t l y under way for Vancouver Internat ional A i r p o r t ) . Rigorous cons iderat ion of the use of benef i t -cos t analyses to determine the economical ly most e f f i c i e n t use of resources with respect to a i r p o r t noise i s beyond the scope of t h i s paper, but some general comments may be made. Bearing i n mind the above comments concerning d i r e c t a p p l i c a t i o n to only marginal changes (and then c a u t i o u s l y ) , the NDSI f igure obtained from the model may be appl ied to a l l a f fec ted propert ies and the t o t a l e f fec ts aggregated across the affected area to g ive a c a p i t a l i z e d value of annual noise costs . Here, the hedonic p r i c e model i s being used to obtain the " c a p i t a l i z e d present value of some unknown future stream of quietude" - not an annual p r i c e for quie t (Nelson 1978a, p.132) . In BCA terms, t h i s value may be represented as: 156 N = n 0 + iVtl+r) + rVfl+r) 2 + . . . = ^nt/fl+r)' (6.1) where n t = annual noise cost , r = s o c i a l discount r a t e . For Net Present Value (NPV) c a l c u l a t i o n s , the NPV of a p r o j e c t equals the present value of noise costs p lus the net present value of non-noise costs and benef i t s , or : 6.4 THE IMPACT OF A THIRD RUNWAY AT VANCOUVER INTERNATIONAL With a major benef i t - cos t study of a proposed t h i r d runway at Vancouver Internat iona l A i r p o r t being conducted at the same time as the research for t h i s thes i s was undertaken, t h i s paper does not attempt to make a d e f i n i t i v e v a l u a t i o n of the t o t a l noise impact; such would require d e t a i l e d assumptions about the extent and t iming of ac tua l noise i m p l i c a t i o n s of the proposed a d d i t i o n a l runway, as we l l as dec i s ions on threshold noise l e v e l s and other i ssues l e f t unresolved i n t h i s study. Nonetheless the broad impact of small changes i n noise l e v e l s r e s u l t i n g from the (6.2) AIRPORT 157 cons truc t ion and operation of a t h i r d runway, may be postu lated as an ind ica t ion of the a p p l i c a t i o n of the r e s u l t s obtained i n t h i s study. Without d i scuss ing the ent ire nature of the proposed p r o j e c t , i t may be r e c a l l e d from Chapter 2 that the r e s u l t i n g a i r p o r t operations would be expected to reduce noise i n some areas and increase i t i n others . The net e f f e c t i s , of course, the i n t e r e s t i n g issue from "the p u b l i c ' s " perspect ive . The NDSI of 0.65 percent obtained from the model l ing could be used (assuming, among other th ings , that the noise impact i s l i n e a r and that the r e l a t i o n s h i p appl ies not only to s ingle-detached propert ies i n Richmond but to a l l property types i n Richmond and Vancouver), along with p r o j e c t i o n s of changes i n NEF l eve l s for each i d e n t i f i e r i n Richmond (and Vancouver) to estimate the s o c i a l impact. For each i d e n t i f i e r , the t o t a l number of, and a current average (median) sa les p r i c e for each r e s i d e n t i a l property type, could be used. As a hypothet ica l example, assume the fo l l owing: . vacant land may be ignored. . the median sales pr i ce s for s ingle-detached and condominiums are the same i n a l l i d e n t i f i e r s . 158 . the runway i s b u i l t immediately. . there i s no change i n market condi t ions , r e l a t i v e l y or abso lute ly , for s ingle-detached and condominiums i n Richmond or adjacent property markets. . there i s no r e l o c a t i o n (to outside Richmond) because of the change i n noise l e v e l s . . the hedonic p r i c e s may be used. . the 1989 t h i r d quarter median pr i ce s of $220 000 for Richmond s ingle-detached propert ies and $132 000 for Richmond condominiums apply. . there are 6800 s ingle-detached propert ies and 8500 condominiums i n the no i se -af fec ted area. An extreme scenario for Richmond may be used to i d e n t i f y a pos s ib l e maximum cost impact. Assume that a l l proper t i e s i n s i d e the noise zone and 30 percent of t h i s number immediately outside the noise zone experience an increase i n noise r a i s i n g t h e i r NEF l e v e l by three . The t o t a l d o l l a r cost impact, or the c a p i t a l i z e d present value of annual noise costs (N) would be $66 m i l l i o n . I t must be r e i t e r a t e d that t h i s i s not the annual noise cos t . Despite t h i s being 159 an extreme example, with many assumptions and much averaging for i l l u s t r a t i v e purposes, the f i n a l t o t a l d o l l a r impact may be considered at l eas t i n d i c a t i v e . Af ter a l l , extremes, assumptions and averaging are the s t u f f of which BCAs are made. 6.5 VALUE OP THE MODEL The simple hedonic model, one of the property value approaches, adopted for use i n t h i s paper was se lec ted i n preference to a l t e r n a t i v e va lua t ion approaches. The model i s designed i n a manner s i m i l a r to previous a i r p o r t noise s tud ies , taking account of var ious issues ra i sed i n the t h e o r e t i c a l l i t e r a t u r e as we l l as other s tudies . What p a r t i c u l a r l y d i s t ingu i shes the modell ing for Vancouver Internat iona l A i r p o r t from the previous studies i s the data (the high q u a l i t y of i n d i v i d u a l data for sales t ransac t ions and property c h a r a c t e r i s t i c s ; the less aggregated census enumeration area data - compared with census t r a c t s ; and the large sample s i ze - r e l a t i v e to models using i n d i v i d u a l data ) , the t e s t i n g of severa l forms for the noise v a r i a b l e ( s ) , and explorat ion of a l t e r n a t i v e noise thresho lds . The NDSI r e s u l t of 0.65 percent for the preferred model i s c lose to the simple average of 0.61 percent for previous s tudies (ranging from 0.4 to 1.1 percent ) . 160 6.6 FURTHER RESEARCH The fo l lowing areas for further research are apparent a f t er undertaking the modell ing for t h i s paper: . t e s t i n g the model for a l t e r n a t i v e property types (condominiums and vacant land) . . model l ing for Vancouver (separately and/or combined with Richmond by the use of a dummy v a r i a b l e ) . . examining the e f fects of r a p i d l y r i s i n g property p r i c e s (post-1987). . fur ther experimentation with the a i r p o r t noise thresho ld . . use of a l l sales (not ju s t a sample) to t e s t the r e l a t i o n s h i p only for propert ies roughly equid i s tant from the a i r p o r t (after Nelson 1981a), p a r t i c u l a r l y to i d e n t i f y any n o n - l i n e a r i t y . . screening for l o c a l environmental f a c t o r s . . r e f i n i n g of independent v a r i a b l e s f or c e r t a i n c h a r a c t e r i s t i c s - p a r t i c u l a r l y a c c e s s i b i l i t y and the most important p h y s i c a l v a r i a b l e , dwel l ing age. 161 . use of more sophis t icated a l t e r n a t i v e funct iona l forms. . t e s t i n g of a l t e r n a t i v e noise d e s c r i p t o r s , such as Notwithstanding t h i s l i s t of poss ib le future d i r e c t i o n s to re f ine the model, the model presented i s considered va luable : the a i r p o r t noise /property value r e l a t i o n s h i p i t y i e l d s i s an important input i n any evaluat ion of the t o t a l noise impact of any changes at Vancouver In ternat iona l A i r p o r t . 162 BIBLIOGRAPHY Abelson, P W (1977), "The Pol icy Problems and Economics of A i r c r a f t Noise", Transportat ion Research, 11, pp 357-64 . Abelson, P W (1979), "Property Pr ices and the Value of Amenities", Journal of Environmental Economics and  Management. 6, pp 11-28. Abelson, P W, and A Markandya (1985) , "The In terpre ta t ion of C a p i t a l i z e d Hedonic Prices i n a Dynamic Environment", Journa l of Environmental Economics and Management. 12, pp 195-206. A l l e n , W (1973), "Noise; the S o c i a l Impact", i n A i r p o r t s  f or the 80s. Proceedings of the 4th World A i r p o r t s Conference, 3-5 A p r i l 1973, The I n s t i t u t i o n of C i v i l Engineers , London. Anderson, R J , J r , and T D Crocker (1971), " A i r P o l l u t i o n and Res ident ia l Property Values", Urban Studies , 8, pp 171-80. Ashford , N, H P M Stanton, and C A Moore (1984), A i r p o r t  Operat ions . John Wiley and Sons, New York. Ashford , N, and P H Wright (1979), A i r p o r t Engineer ing , John Wiley and Sons, New York. Ashford , N, and P H Wright (1984), A i r p o r t Engineer ing , 2nd e d . , John Wiley and Sons, New York. Baldwin, B C (1980), The Management Structure of A i r p o r t Operations i n Canada and the United States , Centre for Transportat ion Studies , U n i v e r s i t y of B r i t i s h Columbia, Vancouver. B a l l , M J (1973), "Recent Empir ica l Work on the Determinants of Re la t ive House Pr ices" , Urban Studies . 10, pp 213-33. Baxter, W F , and L R A l t r e e (1972), "Legal Aspects of A i r p o r t Noise", Journal of Law and Economics. 15, pp 1-113. BC T r a n s i t (1983), "Vancouver Regional T r a n s i t System Serv ice Design Guidel ines", Vancouver, BC. 163 Bednarz, R S (1975), The E f f e c t of A i r P o l l u t i o n on Property  Value in Chicago. U n i v e r s i t y of Chicago, Department of Geography, Chicago, I l l i n o i s . B e l l , R B, and L M B e l l (1980), "Airport Noise: Legal Developments and Economic Al ternat ives" , Ecology Law  Q u a r t e r l y . 8 (4), pp 607-53. Bender, B, and H Hwang (1985), "Hedonic Housing P r i c e Indices and Secondary Employment Centers", Journal of  Urban Economics,. 17, pp 90-107. Bishop, D E (1966), "Judgments of the Relat ive and Absolute A c c e p t a b i l i t y of A i r c r a f t Noise", Journal of the  A c o u s t i c a l Society of America. 40 (1), pp 108-22. B lay lock , J E , III (1977), "Airport Noise and Housing Values: An Inves t igat ion into the Hedonic Theory of Housing and the Value of Quiet", Unpublished Ph.D. d i s s e r t a t i o n , Texas A & M U n i v e r s i t y . Bor ins , S F (1981), "Mieszkowski and Saper's Estimate of the E f f e c t s of A i r p o r t Noise on Property Values: A Comment", Journal of Urban Economics. 9, pp 125-8. Bragdon, C R (1971), Noise P o l l u t i o n : The Unquiet C r i s i s . U n i v e r s i t y of Pennsylvania Press, P h i l a d e l p h i a , Pennsylvania. Bragdon, C R (1983), "The Contro l of A i r p o r t Noise i n the United States: An Overview", Inter-noise 83 -Proceedings, 1983 Internat iona l Conference on Noise Contro l Engineering, Edinburgh, pp 57-60. Bragdon, C R (1984), "Land Planning Noise Contro l Techniques Around A i r p o r t s " , Inter -no i se 84 - Proceedings. 1984 In ternat iona l Conference on Noise Control Engineer ing , Honolulu, pp 635-638. Bragdon, C R (1985) , "Airpor t Noise Monitoring Systems i n North America", Issues i n Transpor ta t ion -re la t ed  Environmental Q u a l i t y . Transportat ion Research Record 1033, Transportat ion Research Board, Nat ional Research C o u n c i l , Washington, DC, pp 43-6. B r i t t l e , C (1982), "Role of A i r p o r t Noise A l l o c a t i o n s i n a Regional A i r p o r t System", A i r p o r t and Highway Noise  C o n t r o l . Planning, and A n a l y s i s . Transportat ion Research Record 865, Transportat ion Research Board, Nat iona l Research C o u n c i l , Washington, DC, pp 25-32. Brog, W, G Haberle , and B Mettler-Meibom (1982), "Comparison of I r r i t a t i o n Caused by Noise Generated by Road T r a f f i c and A v i a t i o n T r a f f i c " , A i r p o r t and Highway Noise  C o n t r o l . Planning, and A n a l y s i s . Transportat ion 164 Research Record 865, Transportat ion Research Board, Nat ional Research Counc i l , Washington, DC, pp 32-38. Brookshire , D S, M A Thayer, W D Schulze, and R C D'Arge (1982) , "Valuing Publ ic Goods: A Comparison of Survey and Hedonic Approaches", American Economic Review. 72 (1), pp 165-77. Brown, J N, and H S Rosen (1982), "On the Estimation of S t u c t u r a l Hedonic Pr i ce Models", Econometrica, 50 (3), pp 765-8. B u g l i a r e l l o , G, A Alexandre, J Barnes, C Wakstein (1976), The Impact of Noise P o l l u t i o n : A Soc io-Technologica l  Introduct ion . Pergamon Press , New York. B u l l e n , R B, and A J Hede (1986), "Comparison of the Ef fect iveness of Measures of A i r c r a f t Noise Exposure by Using S o c i a l Survey Data", Journal of Sound and  V i b r a t i o n . 108 (2), pp 227-45. B u l l e n , R B, A J Hede, and E Kyriacos (1986), "Reaction to A i r c r a f t Noise i n R e s i d e n t i a l Areas around A u s t r a l i a n A i r p o r t s " , Journal of Sound and V i b r a t i o n . 108 (2), pp 199-225. B u t l e r , R V (1980), "Cross - sec t ional V a r i a t i o n i n the Hedonic Relat ionship for Urban Housing Markets", Journa l of Regional Science. 20 (4), pp 439-53. B u t l e r , R V (1982), "The S p e c i f i c a t i o n of Hedonic Indexes for Urban Housing", Land Economics. 58, pp 96-108. C e n t r a l Housing and Mortgage Corporat ion (CHMC) (1978), New  Housing and A i r p o r t Noise. Ottawa. C e n t r a l Housing and Mortgage Corporat ion (CHMC) (1984), Road  and R a i l Noise: Ef f ec t s on Housing, Ottawa. Chalupnik, J D (ed.) (1970), Transportat ion Noises: A  Symposium on A c c e p t a b i l i t y C r i t e r i a . U n i v e r s i t y of Washington Press, S e a t t l e . Cobb, S A (1977), "site Rent, A i r Q u a l i t y , and the Demand f o r Amenities", Journal of Environmental Economics and  Managementf 4, pp 214-218. Cobb, S (1984), "The Impact of S i t e C h a r a c t e r i s t i c s on Housing Cost Estimates", Journa l of Urban Economics. 15, pp 26-45. Commission on the T h i r d London A i r p o r t (CTLA) (1970), Papers  and Proceedings f Volume 7, (Chairman: The Hon Mr J u s t i c e R o s k i l l ) , HMSO, London. 165 Commission on the Third London A i r p o r t (CTLA) (1971), Report (Chairman: The Hon Mr J u s t i c e R o s k i l l ) , HMSO, London. Crowley, R W (1973), "A Case Study of the Ef fec t s of an A i r p o r t on Land Values", Journal of Transport Economics  and P o l i c y , pp 144-52. Cummings, R G, D S Brookshire, and W D Schulze (1986), Va lu ing Environmental Goods: An Assessment of the  Contingent Valuat ion Method. Rowman and A l l e n h e l d , Totowa, NJ. Dan ie l s , A (1989), "Airport manager backs f u l l s crut iny for new runway", Vancouver Sun. November 8. D a r u i , Z , C X i u l a n , and C Tong (1986), "Chinese D e f i n i t i o n of Loudness, Noisiness and Annoyance and the A t t i t u d e towards Noise", Inter-noise 86 - Proceedings. 1986 Internat iona l Conference on Noise Control Engineering, Cambridge, USA, pp 855-60. Dasgupta, A K, and D W Pearce (1972), Cost -Benef i t A n a l y s i s :  Theory and Prac t i ce . Macmil lan, London. de Leeuw, F (1971), "The Demand for Housing: A Review of Cross -Sect ion Evidence", Review of Economics and  S t a t i s t i c s . 53 (1), pp 1-10. DeLoach, R (1984) , "Model for Assessing A i r p o r t Community Noise Impact", Transportat ion Research C i r c u l a r . No. 274, Transportat ion Research Board, Nat ional Research C o u n c i l , Washington, DC, pp 35-8. De Vany, A S (1976), "An Economic Model of A i r p o r t Noise P o l l u t i o n i n an Urban Environment", i n S A Y L i n (ed . ) , Theory and Measurement of Economic E x t e r n a l i t i e s . Academic Press, New York, pp 205-14. Diamond, D B, J r , and G S T o l l e y (eds) (1982), The Economics  of Urban Amenities. Academic Press , New York. D o l l e y , L C, and D G C a r r o l l (1983), "Airport Noise P o l l u t i o n Damages: The Case for Local L i a b i l i t y " , Urban Lawyer, 15 (3), pp 621-36. Doyle, R H (1988), "Can a Busy A i r C a r r i e r A i r p o r t Rea l ly be Compatible with i t s Neighbours?", i n A i r p o r t s for  People. Proceedings of the 8th World A i r p o r t s Conference, organized by the I n s t i t u t i o n of C i v i l Engineers , 2-5 June 1987, Thomas T e l f o r d , London. Draper, N, and H Smith (1981), Appl i ed Regression A n a l y s i s . 2nd e d . , John Wiley and Sons, New York. 166 Dygert, P K (1973), "Estimation of the Cost of A i r c r a f t Noise to Residual A c t i v i t i e s " , Unpublished Ph.D. d i s s e r t a t i o n , U n i v e r s i t y of Michigan, summarized i n Nelson (1978a, 1980). Edmonds, L D, P M Layde, and J D Erickson (1979), "Airpor t Noise and Teratogenesis", Archives of Environmental  Heal th . 34 (4), pp 243-7. E l l i c k s o n , B (1981), "An A l t e r n a t i v e Test of the Hedonic Theory of Housing Markets", Journal of Urban Economics. 9, pp 56-79. Emerson, F C (1972), "Valuation of Res ident ia l Amenities: An Econometric Approach", Appra i sa l J o u r n a l . 40 (2), pp 268-78. Fayerman, P (1989), "Ratepayers' group worried about noise from new runway", Vancouver Sun. November 7, p . C2. Feldman, E J , and J M i l c h (1982), Technocracy Versus Democracy: The Comparative P o l i t i c s of In ternat iona l  A i r p o r t s , Auburn House, Boston, Massachusetts. Feldman, E J , and J Mi l ch (1983), The P o l i t i c s of Canadian  A i r p o r t Development: Lessons for Federalism, Duke U n i v e r s i t y Press, Durham, NC. Feldman, J M (1987), "Noise growing as major U .S . orphan", A i r Transport World. 22, pp 32-7. Ferguson, B (1984) , "The E f f e c t of the Vancouver Advanced R a i l T r a n s i t System on S ing le Family Property Values", Unpublished MSc (Business Administration) t h e s i s , U n i v e r s i t y of B r i t i s h Columbia, Vancouver, BC. F i s h e r i e s and Environment Canada (1978), "Guidelines for Preparat ion of an Environmental Impact Statement on the Proposed P a r a l l e l Runway at Vancouver In ternat iona l A i r p o r t " , June 29. F l o r e n t i n e , M, S Namba, and S Kuwano (1986), "Concepts of Loudness, Nois iness , Noise, and Annoyance i n the USA, Japan, and England", Inter -no i se 86 - Proceedings, 1986 Internat iona l Conference on Noise Control Engineer ing , Cambridge, USA, pp 831-4. Flowerdew, A D J (1972), "Choosing a S i t e for the T h i r d London A i r p o r t : the R o s k i l l Commission's Approach's i n P R G Layard (ed. ) , Cos t -Benef i t Ana lys i s : Se lected  Readings, Penguin, Harmondsworth, pp 431-51. F o l l a i n , J R, and E Jimenez {1985), "Estimating the Demand for Housing C h a r a c t e r i s t i c s : A Survey and C r i t i q u e " , Regional Science and Urban Economics. 15, pp 77-107. 167 F o l l a i n , J R, J r , and S Malpezzi (1981a), "Are Occupants Accurate Appraisers?", Review of Pub l i c Data Use, 9, pp 47-55. F o l l a i n , J R, J r , and S Malpezzi (1981b), "The F l i g h t to the Suburbs: Insights Gained from an Analys i s of C e n t r a l -C i t y vs Suburban Housing Costs", Journal of Urban  Economics, 9, pp 381-398. F r a n k e l , M (1985), "Amenity Changes, Property Values, and Hedonic Pr ices i n a Closed C i t y " , Journal of  Environmental Economics and Management. 12, pp 117-31. F r a n k e l , M (1988a), "The Ef fec t s of A i r c r a f t Noise and A i r p o r t A c t i v i t y on Res ident ia l Property Values: A Survey Study", Working Paper Number 60, Of f i ce of Real Estate Research, Un ivers i ty of I l l i n o i s , Urbana-Champaign. F r a n k e l , M (1988b), "The Impact of A i r c r a f t Noise on R e s i d e n t i a l Property Markets", I l l i n o i s Business  Review, 45 (5), pp 8-13. Franken, P A, and D Standley (1970), A i r c r a f t Noise and  A i r p o r t Neighbors: A Study of Logan Internat ional  A i r p o r t , prepared for US Department of Transportat ion and US Department of Housing and Urban Development. Freeman, A M, I I I (1974), "On Est imating A i r P o l l u t i o n Contro l Benefits from Land Value Studies", Journal of  Environmental Economics and Management. 1, pp 74-83. Freeman, A M, I I I (1979a), The Benef i ts of Environmental  Improvement: Theory and P r a c t i c e . Resources for the Future , Johns Hopkins U n i v e r s i t y Press , Balt imore. Freeman, A M, I I I (1979b), "Hedonic P r i c e s , Property Values and Measuring Environmental Benef i t s : a Survey of the Issues", Scandinavian Journal of Economics. 81, pp 155-73. Freeman, A M , III (1980), "Land Pr ices S u b s t a n t i a l l y Underestimate the Value of Environmental Q u a l i t y : A Comment", Review of Economics and S t a t i s t i c s . 62, pp 154-6. F r e r i c h s , R R, B L Beeman, and A H Coulson (1980), "Los Angeles A i r p o r t Noise and M o r t a l i t y - Faul ty Ana lys i s and P u b l i c Po l i cy" , American Journa l of Pub l i c Hea l th . 70 (4), pp 357-62. G a r b e l l , M A (1979), "The Environmental Stress between an A i r p o r t and i t s Neighbours", i n A i r p o r t s for the  Community. Proceedings of the Conference of the 6th 168 World A i r p o r t s Conference, 5-7 June 1979, The I n s t i t u t i o n of C i v i l Engineers, London. G a u t r i n , J - F (1975), "An Evaluation of the Impact of A i r c r a f t Noise on Property Values with a Simple Model of Urban Land Rent", Land Economics. 51 (1), pp 80-6. George, K D, and J Shorey (1978), The A l l o c a t i o n of Resources; Theory and P o l i c y . George A l l e n and Unwin, London. G i f f o r d , K H (1980), "Noise Control and Compensation for Noise P o l l u t i o n " , Austra l ian Law J o u r n a l . 54, pp 408-18. Goldberg, M A (1989), "Addit ional Evidence about the Impacts of Vancouver's Light R a i l T r a n s i t System on Housing and Land Pr i ce s" , Paper presented to the Weimer School of Advanced Studies i n Real Estate and Land Economics of Homer Hoytt Advanced Studies I n s t i t u t e , Singer Is land, F l o r i d a , January. Goodman, A C (1977), "A Comparison of Block Group and Census T r a c t Data i n a Hedonic Housing P r i c e Model", Land  Economics. 53 (4), pp 483-7. Goodman, A C (1978), "Hedonic P r i c e s , P r i c e Indices and Housing Markets", Journal of Urban Economics. 5, pp 471-84 . Goodman, R F , and B B Clary (1976) , "Community At t i tudes and A c t i o n i n Response to A i r p o r t Noise", Environment and  Behavior. 8 (3), pp 441-70. Greater Vancouver Regional D i s t r i c t (GVRD) (1989), "GVRD Transporta t ion Study", unpublished, Vancouver. G r i l i c h e s , Z (ed.) (1971), Pr ice Indexes and Qua l i ty Change:  Studies i n New Methods of Measurement. Harvard U n i v e r s i t y Press, Cambridge, Massachusetts. Hageman, R K (1981), "Nuclear Waste D i sposa l : Potent ia l Property Value Impacts", Natural Resources J o u r n a l . 23 (4), pp 789-810. Halvorsen , R, and R Palmqauist (1980), "The In terpre ta t ion of Dummy Var iab le s i n Semi-logarithmic Equations", American Economic Review. 70 (3), pp 474-5. Halvorsen , R, and H O Pollakowski (1981), "Choice of Funct iona l Form for Hedonic P r i c e Equations" f Journal  o f Urban Economics. 10, pp 37-49. Harper, D V (1988), "The Continuing Dilemma of A i r c r a f t Noise", Transportat ion Journa l , 28 (2), pp 33-42. 169 H a r r i s , A H (1981), "Hedonic Technique and Valuat ion of Environmental Qual i ty" , i n V K Smith (ed.) , Advances i n  Appl i ed Microeconomics,, Volume 1, JAI Press, Greenwich, Conn. , pp 31-49. H a r r i s o n , D, J r , and D L Rubinfeld (1978a), "The D i s t r i b u t i o n of Benefits from Improvements i n Urban A i r Qua l i ty" , Journal of Environmental Economics and  Management, 5, pp 313-32. H a r r i s o n , D, J r , and D L Rubinfeld (1978b), "Hedonic Housing Pr i ce s and the Demand for Clean A i r " , Journal of  Environmental Economics and Management. 5, pp 81-102. Harvey, M E , J W F r a z i e r , and M Matu l ion i s (1979), "Cognition of a Hazardous Environment: Reactions to Buffa lo A i r p o r t Noise", Economic Geography. 55 (4), pp 263-86. Haynes, K E , F Y P h i l l i p s , and G M White (1976), "A Market-consumer Model of Res ident ia l Property Values: the A i r p o r t - n o i s e - Land-use Impact Problem", Environment  and Planning A. 8, pp 271-87. Heggie, I G (1972), Transport Engineering Economics. McGraw-H i l l , London. Herzer , W (1988), "Evolving Theories i n the Cost Impact of A i r c r a f t Noise", Unpublished term paper, Commerce 545, Facu l ty of Commerce and Business Adminis trat ion , U n i v e r s i t y of B r i t i s h Columbia. Herz ing , W (1984), "Noise Control on the Apron of Munich A i r p o r t " , Inter -no i se 84 - Proceedings. 1984 Internat iona l Conference on Noise Control Engineering, Honolulu, pp 663-668. Hi ldebrand, J L (ed.) (1970), Noise P o l l u t i o n and the Law. W i l l i a m S Hein, Buffa lo , NY. Hogan, C M, and J Ravnkilde (1985), "Design of A c o u s t i c a l Insu la t ion for E x i s t i n g Residences i n the V i c i n i t y of San Jose Munic ipa l A i r p o r t " , Issues i n Transpor ta t ion- r e l a t e d Environmental Q u a l i t y . Transportat ion Research Record 1033, Transportat ion Research Board, Nat ional Research C o u n c i l , Washington, DC, pp 54-9. Holsman, A J , and V Aleksandric (1977), " A i r c r a f t Noise and the R e s i d e n t i a l Land Market i n Sydney", A u s t r a l i a n  Geographer. 13, pp 401-8. Horonjef f , R, and F X McKelvey (1983), Planning and Design  of A i r p o r t s . 3rd e d . , McGraw-Hi l l , New York. 170 Horowitz, J L (1987), "Ident i f i cat ion and Stochast ic S p e c i f i c a t i o n i n Rosen's Hedonic P r i c e Model", Journal  of Urban Economics. 22, pp 165-73. Jenkins , L , A Tarnopolsky, and D Hand (1981), "Psychiatr ic admissions and a i r c r a f t noise from London A i r p o r t : f our -year , three -hosp i ta l s ' study", Psychological  Medic ine , 11, pp 765-82. Johansson, P-0 (1987), The Economic Theory and Measurement  of Environmental Benefits . Cambridge U n i v e r s i t y Press , Cambridge. Johnston, J (1972), Econometric Methods. 2nd e d . , McGraw-H i l l , New York. Jud , G D, and J M Watts (1981), "Schools and Housing Values", Land Economics, 57 (3), pp 459-70. K a i n , J F , and J M Quigley (1970), "Measuring the Value of Housing Q u a l i t y " , Journal of the American S t a t i s t i c a l  A s s o c i a t i o n . 65, pp 532-48. Kamerud, D B, and C R von Buseck (1985) , "The Ef fec t s of T r a f f i c Sound and Its reduction on House Pr ices" , Issues i n Transportat ion-re la ted Environmental Q u a l i t y . Transpor ta t ion Research Record 1033, Transportat ion Research Board, National Research C o u n c i l , Washington, DC, pp 16-22. Kennedy, P (1979), A Guide to Econometrics. MIT Press, Cambridge, Massachusetts. K i s h , L , and J B Lansing (1954), "Response E r r o r s i n Es t imat ing the Value of Homes", American S t a t i s t i c a l  A s s o c i a t i o n J o u r n a l , 49, pp 520-38. Kneese, A V , and B T Bower (eds) (1972), Environmental  Q u a l i t y A n a l y s i s : Theory and Method i n the S o c i a l  Sc iences . Resources for the Future , I n c . , Johns Hopkins Press , Bal t imore . Krumm, R J (1980), "Neighborhood Amenities: An Economic A n a l y s i s " , Journal of Urban Economics. 7, pp 208-24. K r y t e r , K D (1985), The Effects of Noise on Man. 2nd e d . , Academic Press , Orlando, F l o r i d a . Kuwano, S, S Namba, and H J F a s t i (1986), "Loudness E v a l u a t i o n of Various Sounds by Japanese and German Subjects", Inter-noise 86 - Proceedings. 1986 I n t e r n a t i o n a l Conference on Noise Contro l Engineering, Cambridge, USA, pp 835-40. 171 Lamb, J (1989a), "Healthy a i r p o r t v i t a l to province", Vancouver Sun, June 21, p .A7. Lamb, J (1989b), "Runway plan w i l l r a i s e summer hackles", Vancouver Sun, Ju ly 10. Lamb, J (1989c), "The case against b u i l d i n g another runway", Vancouver Sun. J u l y 22, p .B5. Lamb, J (1989d), u n t i t l e d column, Vancouver Sun. J u l y 26. Lamb, J (1989e), "Ottawa set to throw $100 m i l l i o n at a i r p o r t " , Vancouver Sun. J u l y 29, p .B5 . Lamb, J (1989f), "Airport growth: i t ' s a long saga", Vancouver Sun. September 20, p.A13. Lancaster , K J (1966), "A New Approach to Consumer Theory", Journa l of P o l i t i c a l Economy. 74, pp 132-57. Lane, S R (1986), "Airport Noise P o l l u t i o n and Adverse Health Ef fec t s" , Inter-noise 86 - Proceedings. 1986 In ternat iona l Conference on Noise Contro l Engineering, Cambridge, USA, pp 799-804. Large, J B (1984), "Development of Techniques for Est imating A i r c r a f t Ground Noise", Inter -no i se 84 - Proceedings, 1984 Internat iona l Conference on Noise Control Engineer ing , Honolulu, pp 673-6. L i , M M , and H J Brown (1980), "Micro-Neighbourhood E x t e r n a l i t i e s and Hedonic Housing Pr i ce s" , Land  Economics f 56 (2), pp 125-41. Lucas, R E B (1975), "Hedonic P r i c e Funct ions", Economic  Ingu iry . 13, pp 157-78. McDougall , G S (1976), "An Inguiry in to the Demand for A i r c r a f t Noise Abatement", Review of Regional Studies . 6 (3), pp 59-69. McDougall , G S, and C Wright (1980), "A Proposal for Improving the Measurement of Benef i t s from P o l l u t i o n Abatement", Journal of Environmental Economics and  Management. 7, pp 20-29. Maclennan, D (1977), "Some Thoughts on the Nature and Purpose of House Pr i ce Studies", Urban Studies , 14, pp 59-71. M c M i l l a n , M L (1979), "Estimates of Households' Preferences for Environmental Qual i ty and Other Housing C h a r a c t e r i s t i c s from a System of Demand Equations", Scandinavian Journal of Economics. 81, pp 174-87. 172 McMil lan , M L , B G Reid , and D W G i l l e n (1980), "An Extension of the Hedonic Approach for Est imating the Value of Quiet", Land Economics. 56 (3), pp 315-28. Maddala, G S (1977), Econometrics. McGraw-Hi l l , New York. Maler , K-G (1974), Environmental Economics: A T h e o r e t i c a l  Inquiry . Resources for the Future, I n c . , Johns Hopkins U n i v e r s i t y Press , Baltimore. Mark, J (1977), "Determinants of Urban House P r i c e s : A Methodological Comment", Urban Studies . 14, pp 359-63. Mark, J H (1980), "A Preference Approach to Measuring the Impact of Environmental E x t e r n a l i t i e s " Land Economics. 56 (1), pp 103-16. M a r t i n , W H (1972), "Aviat ion - Economic Impact on Real Estate", A p p r a i s a l J o u r n a l f 40, pp 231-4. Maser, S M, W H R i k e r , and R N Rosett (1977), "The Ef fec t s of Zoning and E x t e r n a l i t i e s on the Pr ice of Land: An E m p i r i c a l Ana lys i s of Monroe County, New York", Journal  of Law and Economics. 20, pp 111-32. Meecham, W C, and N Shaw (1979), "Effects of J e t Noise on M o r t a l i t y Rates", B r i t i s h Journal of Audiology, 13, pp 77-80. Meecham, W C, and N Shaw (1980), "Comments on 'Los Angeles A i r p o r t Noise and M o r t a l i t y - Faul ty Ana lys i s and Publ i c P o l i c y ' " , l e t t e r , American Journal of Pub l i c  Heal th . 70 (5), p. 543. Menge, C W (1985), "The One-Minute L e q Measurement Method", Issues i n Transpor ta t ion-re la ted Environmental Q u a l i t y . Transportat ion Research Record 1033, Transportat ion Research Board, Nat ional Research C o u n c i l , Washington, DC, pp 22-4. Mieszkowski, P, and A M Saper (1978), "An Estimate of the E f f e c t s of A i r p o r t Noise on Property Values", Journal  of Urban Economics. 5, pp 425-40. M i l l e r , N G (1982), "Resident ia l Property Hedonic P r i c i n g Models: A Review", i n C F Sirmans (ed . ) , Research i n  Real E s t a t e . Volume 2, JAI Press, Greenwich, Conn. , pp 31-56. Mishan, E J (1970), "What i s Wrong with R o s k i l l ? " , Journal  of Transport Economics and P o l i c y . 4 (3), pp 221-34. Mishan, E J (1971), "The Postwar L i t e r a t u r e on E x t e r n a l i t i e s : An Interpretat ive Essay", Journa l of  Economic L i t e r a t u r e . 9, pp 1-28. 173 M o r r a l l , J F , I I I (1979), "Reducing A i r p o r t Noise", i n M i l l e r , J C, I I I , and B Yandle (eds), Benef i t -cost  Analyses of S o c i a l Regulation: Case Studies from the  Counci l on Wage and Pr ice S t a b i l i t y . American Enterpr i se I n s t i t u t e for Publ ic P o l i c y Research, Washington, DC, pp 147-60. M o r r i s , C L , (1978), "Noise P o l l u t i o n : Attempted Federal Contro l of Airp lane Noise", Natural Resources J o u r n a l . 18 (3), pp 621-38. Muldoon, J P (1984), " A i r c r a f t Noise - i t s Effects/Consequences", Transportat ion Research  C i r c u l a r . No. 286, Transportat ion Research Board, Nat ional Research Counc i l , Washington, DC, pp 13-5. Muskin, J B, and J A Sorrent ino, J r (1977), " E x t e r n a l i t i e s i n a Regulated Industry: The A i r c r a f t Noise Problem", American Economic Review. 67, pp 770-4. Muskin, J B, and J A Sorrent ino, J r (1978), "An ' E f f l u e n t ' Charge Approach to A i r c r a f t Noise Abatement", Journa l  of Environmental Economics and Management. 5, pp 333-50. Nash, C A (1978), "The Theory of S o c i a l Cost Measurement", i n D W Pearce (ed . ) , The Valuat ion of S o c i a l Cost . George A l l e n and Unwin, London. Nat ional Academy of Sciences, Nat ional Academy of Engineering (NRS) (1971a), Jamaica Bay and Kennedy  A i r p o r t : A M u l t i d i s c i p l i n a r y Environmental Study.  Volume 1: Conclusion. Recommendations. Summary. Washington, DC. Nat iona l Academy of Sciences, Nat ional Academy of Engineering (NRS) (1971b), Jamaica Bay and Kennedy  A i r p o r t : A M u l t i d i s c i p l i n a r y Environmental Study.  Volume 2. Washington, DC. Nat iona l Research C o u n c i l , Committee on Appra i sa l of S o c i e t a l Consequences of Transportat ion Noise Abatement (NRC) (1977), Noise Abatement: P o l i c y A l t e r n a t i v e s for  Transporta t ion: A Report to the U . S . Environmental  Protec t ion Agency. Nat ional Academy of Sciences , Washington, DC. Nelson, J P (1977), " A c c e s s i b i l i t y and the Value of Time i n Commuting", Southern Economic J o u r n a l . 43 (3), pp 1321-9. Nelson, J P (1978a), Economic Ana lys i s of Transportat ion Noise Abatement. B a i l i n g e r , Cambridge, Massachusetts. 174 Nelson, J P (1978b), "Residential Choice, Hedonic P r i c e s , and the Demand for Urban A i r Qua l i ty" , Journal of Urban  Economics. 5, pp 357-69. Nelson, J P (1979), "Airport Noise, Locat ion Rent, and the Market for Res ident ia l Amenities", Journal of  Environmental Economics and Management. 6, pp 320-31. Nelson, J P (1980), "Airports and Property Values: A Survey of Recent Evidence", Journal of Transport Economics and  P o l i c y . 14 (1), pp 37-52. Nelson, J P (1981a), "Measuring Benef i ts of Environmental Improvements: A i r c r a f t Noise and Hedonic Pr ices" , i n V K Smith (ed . ) , Advances i n Appl ied Microeconomics. Volume 1, JAI Press, Greenwich, Conn. , pp 51-75. Nelson, J P (1981b), "Three Mi le Is land and Res ident ia l Property Values: Empir ica l Ana lys i s and P o l i c y Impl icat ions" , Land Economics. 57 (3), pp 363-72. Nelson, J P (1982), "Highway Noise and Property Values: a Survey of Recent Evidence", Journa l of Transport  Economics and P o l i c y , 16, pp 117-38. Neter, J , W Wasserman, and M H Kutner (1985), Appl ied L inear  S t a t i s t i c a l Models: Regression. Ana lys i s of Variance ,  and Experimental Designs. Irwin, Homewood, 111. Niskanen, W A, and S H Hanke (1977), "Land Pr i ce s S u b s t a n t i a l l y Underestimate the Value of Environmental Q u a l i t y " , Review of Economics and S t a t i s t i c s . 59, pp 375-7. Niskanen, W A, and S H Hanke (1980), "Land Pr i ce s S u b s t a n t i a l l y Underestimate the Value of Environmental Q u a l i t y : A Rejoinder", Review of Economics and  S t a t i s t i c s , 62, p.156. Noise Advisory Counc i l (1971), A i r c r a f t Noise: F l i g h t Routeing Near A i r p o r t s ; Report by a working group of  the C o u n c i l . HMSO, London. Norus i s , M J (1988), SPSS-X Introductory S t a t i s t i c s Guide  f o r SPSS-X Release 3. SPSS I n c . , Chicago, I l l i n o i s . Nourse, H O (1967), "The Ef fec t of A i r P o l l u t i o n on House Values", Land Economics. 43, pp 181-9. O'Byrne, P H, J P Nelson, and J J Seneca (1985), "Housing Values , Census Estimates, D i s e q u i l i b r i u m , and the Environmental Cost of A i r p o r t Noise: A Case Study of A t l a n t a " , Journal of Environmental Economics and  Management. 12, pp 169-71. 175 Organisat ion for Economic Co-operation and Development (OECD) (1975), A i r p o r t s and the Environment. P a r i s . Organisat ion for Economic Co-operation and Development (OECD) (1988), Transport and the Environment, P a r i s . Pa ik , I K (1970), Impact of Transportat ion Noise on Urban  R e s i d e n t i a l Property Values with Spec ia l Reference to  A i r c r a f t Noise. Urban Transportat ion Center, Consortium of U n i v e r s i t i e s , Washington, DC. Paik , I K (1972), "Measurement of Environmental E x t e r n a l i t y i n P a r t i c u l a r Reference to Noise", Unpublished Ph.D. d i s s e r t a t i o n , Georgetown U n i v e r s i t y , summarized i n Nelson (1978a, 1980). Palmquist , R B (1982), "Measuring Environmental Ef fec t s on Property Values without Hedonic Regressions", Journal  of Urban Economics. 11, pp 333-47. Palmquist , R B (1984), "Estimating the Demand for the C h a r a c t e r i s t i c s of Housing", Review of Economics and  S t a t i s t i c s . 66, pp 394-404. Pau l , M E (1971), "Can A i r c r a f t Noise Nuisance be Measured i n Money?", Oxford Economic Papers. 23, pp 297-322. Pearce, D (1971), "The Economic Evaluat ion of Noise-Generating and Noise Abatement Projects" , i n Organisat ion for Economic Co-operat ion and Development, Problems of Environmental Economics. P a r i s , pp 103-18. Pearce, D (1976), S o c i a l Cost of Noise. Organisat ion for Economic Co-operation and Development, P a r i s . Pearce, D W (1978), "Noise Valuat ion" , i n D W Pearce (ed . ) , The Valuat ion of S o c i a l Cost . George A l l e n and Unwin, London. Pedhazur, E J (1982), M u l t i p l e Regression i n Behavioural Research: Explanation and P r e d i c t i o n . 2nd e d . , H o l t , Rinehart and Winston, New York. Pendakur, V S, and H McLean (1989), "Automated Light R a i l T r a n s i t (ALRT) i n Vancouver, Canada: Measured and Perceived Noise Impacts", Paper No. 880645, The Transportat ion Research Board, 68th Annual Meeting, January 22-26, 1989, Washington, DC. P lessas , D J (1973), "Airport Noise: Some A n a l y t i c and P o l i c y Perspectives", Land Economics. 49, pp 14-21. P o l i n s k y , A M, and D L Rubinfe ld (1977), "Property Values and the Benefits of Environmental Improvements: Theory and Measurement", i n L Wingo and A Evans (eds), P u b l i c 176 Economics and the Qual i ty of L i f e . Resources for the Future , Johns Hopkins U n i v e r s i t y Press , Balt imore, pp 154-80. Po l insky , A M, and S Shavel l (1975), "The A i r P o l l u t i o n and Property Value Debate", The Review of Economics and  S t a t i s t i c s , 57 (1), pp 100-4. Po l insky , A M, and S Shavel l (1976), "Amenities and Property Values i n a Model of an Urban Area", Journal of P u b l i c  Economics, 5, pp 119-129. P r i c e , I (1974), "The S o c i a l Cost of A i r p o r t Noise as Measured by Rental Changes: The Case of Logan A i r p o r t " , Unpublished Ph.D. d i s s e r t a t i o n , Boston U n i v e r s i t y , summarized i n Nelson (1978a, 1980). Quig ley , J M (1982), "Nonlinear Budget Constra ints and Consumer Demand: An A p p l i c a t i o n to Pub l i c Programs for R e s i d e n t i a l Housing", Journal of Urban Economics. 12, pp 177-201. Radford, K J , and M 0 Giesen (1984), The Ana lys i s of C o n f l i c t s Over the Locat ion of A i r p o r t s Near Major  Populat ion Centres. Research Report No. 87, U n i v e r s i t y of Toronto/York Univers i ty J o i n t Program i n Transporta t ion , Toronto. Real Estate Board of Greater Vancouver (1989), Monthly  S t a t i s t i c s , November. Ridker , R G, and J A Henning (1967), "The Determinants of R e s i d e n t i a l Property Values with Spec ia l Reference to A i r P o l l u t i o n " . Review of Economics and S t a t i s t i c s . 49 (2), pp 246-57. Rosen, S (1974), "Hedonic Pr ices and I m p l i c i t Markets: Product D i f f e r e n t i a t i o n i n Pure Competition", Journa l  of P o l i t i c a l Economy. 82 (1), pp 34-55. Sasak i , R (1989), "On a Poss ib le Bias i n Estimates of Hedonic Pr i ce Functions", Journa l of Urban Economics. 25, pp 138-42. Schechter, M, and M Kim (1989), "Valuation of P o l l u t i o n Abatement Benef i ts: D i r e c t and I n d i r e c t Measurement", Working Paper, U n i v e r s i t y of H a i f a , I s r a e l , and the U n i v e r s i t y of B r i t i s h Columbia. S c h e l l , L M, and D C Hodges (1985), "Longitudinal Study of Growth Status and A i r p o r t Noise Exposure", American  Journa l of Physical Anthropology. 66, pp 383-9. 177 S c h e l l , L M, and R J N o r e l l i (1983), "Airpor t Noise Exposure and the Postnatal Growth of Ch i ldren" , American Journal  of Phys ica l Anthropology. 61, pp 473-82. Schick, A, S Kuwano, and S Namba (1984), "Cross - cu l tura l Examinations of Noise i n the Neighbourhood", In ter - noise 84 - Proceedings. 1984 Internat iona l Conference on Noise Control Engineering, Honolulu, pp 967-72. Schles inger , G (1988), "Airport Noise: The P r o p r i e t o r ' s Dilemma", Transportat ion Law J o u r n a l . 16 (2), pp 333-42. Schnare, A B (1976), "Racial and Ethnic Pr i ce D i f f e r e n t i a l s i n an Urban Housing Market", Urban Studies . 13, pp 107-20. Schnare, A B, and R J Struyk (1976), "Segmentation i n Urban Housing Markets", Journal of Urban Economics. 3, pp 146-66. Schonfeld, P M (1984), "Airport E f f e c t s of A i r l i n e r Size on Transportat ion System E f f i c i e n c y " , Transportat ion  Research C i r c u l a r . No. 274, Transportat ion Research Board, Nat ional Research C o u n c i l , Washington, DC, pp 2-10. Schul tz , T J (1982), Community Noise Rat ing . 2nd e d . , Appl ied Science Publ i shers , London. Sealy , K R (1976), A i r p o r t Strategy and Planning. Oxford U n i v e r s i t y Press, London. Sinden, J A, and A C Worre l l (1979), Unpriced Values: Decisions without Market P r i c e s . John Wiley and Sons, New York. Smith, D (1989), "Runway b i l l c a l l e d no worry", Vancouver  Sun, November 7, pp C I , 2 . Smith, D, and A Daniels (1989), "Third runway to be b u i l t i f environment panel OKs", Vancouver Sun. November 6, pp A l , 1 4 . Smith, M J T (1983), "Isn't Society Now Overplaying the A i r c r a f t Noise Issue?", Inter -no i se 83 - Proceedings. 1983 In ternat iona l Conference on Noise Contro l Engineering, Edinburgh, pp 41-4. Smith, V K (1977), "Resident ia l Locat ion and Environmental Amenities: a Review of the Evidence", Regional  Studies , 11 (1), pp 47-61. Spencer, F A (1980a), Transport J e t A i r c r a f t Noise Abatement  i n Foreign Countr ies : Growth. S t r u c t u r e . Impact. F i n a l 178 Report. Volume 1; Europe, prepared for National Aeronautics and Space Admin i s tra t ion , Transportat ion Center, Northwestern U n i v e r s i t y , Evanston, I l l i n o i s . Spencer, F A (1980b), Transport J e t A i r c r a f t Noise Abatement  i n Foreign Countries: Growth. S tructure . Impact. F i n a l  Report. Volume 2: P a c i f i c Bas in , prepared for Nat ional Aeronautics and Space Admin i s t ra t ion , Transportat ion Center, Northwestern U n i v e r s i t y , Evanston, I l l i n o i s . SPSS Inc. (1988), SPSS-X User's Guide. 3rd e d . , SPSS I n c . , Michigan, I l l i n o i s . S t a r k i e , D N M, and D M Johnson (1975). The Economic Value  of Peace and Quiet . Saxon House, Westmead. S t a t i s t i c s Canada (1988a), Census T r a c t s . Vancouver: Part 1. P r o f i l e s . Cat. No. 95-167. S t a t i s t i c s Canada (1988b), Census T r a c t s . Vancouver: Part 2. P r o f i l e s . Cat. No. 95-168. S t e l l i n g , J (1982), "Don't Confuse Me with Facts , or How to Succeed i n an A i r p o r t Noise Study", Interfaces . 12 (1) , pp 66-8. Stevenson, G M, J r (1972), The P o l i t i c s of A i r p o r t Noise. Duxbury Press, Belmont, CA. Stewart, J (1976), Understanding Econometrics. Hutchinson, London. Stoddard, A T, I I I , and G P F i s h e r (1985), "A Disaggregate Noise Interference Model for Est imat ing A i r p o r t Noise Impact", Issues i n Transpor ta t ion -re la ted Environmental  Q u a l i t y . Transportat ion Research Record 1033, Transportat ion Research Board, Nat ional Research C o u n c i l , pp 47-54. Straszheim, M (1974), "Hedonic Est imat ion of Housing Market P r i c e s : A Further Comment", Review of Economics and  S t a t i s t i c s , 56, pp 404-6. Straszheim, M R (1975a), An Econometric Ana lys i s of the  Urban Housing Market. Nat ional Bureau of Economic Research, New York. Straszheim, M R (1975b), " E f f i c i e n c y and Equity Considerat ions i n the Financing of Noise Abatement A c t i v i t i e s at A i r p o r t s " , In ternat iona l Journal of  Transport Economics. 2 (1), pp 3-15. Sugden, R, and A Will iams (1978), The P r i n c i p l e s of Cost - benef i t A n a l y s i s . Oxford U n i v e r s i t y Press , Oxford. 179 Tabachnick, B G, and L S F i d e l l (1983), Using M u l t i v a r i a t e  S t a t i s t i c s . Harper and Row, New York. Taneja, N K (1987), Introduction to C i v i l A v i a t i o n , Lexington Books, Lexington, Massachusetts. Taneja, N K (1988), The Internat ional A i r l i n e Industry;  Trends. Issues and Challenges, Lexington Books, Lexington, Massachusetts. T a y l o r , R (1970), Noise. Penguin, Harmondsworth. T a y l o r , S M, F L H a l l , and S E B i r n i e (1981), "A Comparison of Community Response to A i r c r a f t Noise at Toronto In terna t iona l and Oshawa Munic ipa l A i r p o r t s " , Journal  of Sound and V i b r a t i o n , 77 (2), pp 233-44. Transport Canada, A i r p o r t Planning Committee (1975a), Aeronaut ica l Noise Study. Volume 1 Main Report. Document PI69-9i , Vancouver, BC. Transport Canada, A i r p o r t Planning Committee (1975b), Aeronaut ica l Noise Study. Volume 2 Appendices A. B. C.  D to Main Report, Document P169-9 i i , Vancouver, BC. Transport Canada, A i r p o r t Planning Committee (1976), Vancouver Internat ional A i r p o r t F i n a l Report. Vancouver, BC. Transport Canada (1980), Vancouver In ternat iona l A i r p o r t :  Master Plan Study Report. Vancouver , .BC. Transport Canada (1981a), Vancouver In terna t iona l A i r p o r t :  Master P l a n r TP 2375, Vancouver, BC. Transport Canada, C i v i l Av ia t ion Branch, P a c i f i c Region (1981b), "Aeronautical Noise Produced i n Communities Surrounding Vancouver Internat iona l A i r p o r t , November 1979 - June 1980", C i v i l A v i a t i o n Planning , TP 3399E, Vancouver, BC. Transport Canada, C i v i l Av ia t ion Branch, P a c i f i c Region (1984), "Vancouver Internat ional A i r p o r t : Engine Run-up - Noise A n a l y s i s , March 1982 - A p r i l 1983", A i r Navigat ion System Requirements, C i v i l A v i a t i o n Planning , TP 5531E, Vancouver, BC. Transport Canada, Canadian A i r Transport Admin i s tra t ion , P a c i f i c Region (1985), "Vancouver In ternat iona l A i r p o r t A i r c r a f t Noise Monitoring System: S e l e c t i o n of Remote Moni tor ing Si tes" , prepared by K C Simpson, A i r Navigat ion System Requirements, TP 6686E, Vancouver, BC. 180 Transport Canada, Canadian A i r Transport Admin i s t ra t ion , P a c i f i c Region (1986), " A i r c r a f t Noise Levels Measured During the T r i a l SID2 Procedure Vancouver Internat iona l A i r p o r t , Canada", prepared by K C Simpson and T C Spurgeon, A i r Navigation System Requirements, TP 7374E, Vancouver, BC. Transport Canada, A i r p o r t s (1989), The Economic Impact of  Vancouver Internat iona l A i r p o r t . TP 9820, Canada. Troy , P N (1973), "Residents and t h e i r Preferences: Property Pr ices and Res ident ia l Q u a l i t y ' , Regional  Studies , 7, pp 183-92. Turvey, R (1957), The Economics of Real Property: An Analys i s of Property Values and Patterns of Use. George A l l e n and Unwin, London. United States , Department of Housing and Urban Development (1972), A i r c r a f t Noise Impact: Planning Guide l ines for  Local Agencies, prepared by Wilsey & Ham, and B o l t , Beraneck and Newman, I n c . , TE/NA-472. United States , Department of Transportat ion , Federal A v i a t i o n Adminis trat ion (1977), Impact of Noise on  People. Of f i ce of Environmental Q u a l i t y . United States , Department of Transportat ion , Federal A v i a t i o n Adminis trat ion (1982), INM. Integrated Noise  Model. Vers ion 3: User 's Guide ( inc lud ing changes 1, 2, 3), Washington, DC. U t l e y , W A (1985), "Descriptors for Ambient Noise", I n t e r - noise 85 - Proceedings. 1985 In ternat iona l Conference on Noise Control Engineering, Munich, pp 1069-72. Vancouver Sun (1989), E d i t o r i a l : "The t h i r d runway humbug element", November 13, p .A8. Wabe, J S (1971), "A Study of House Pr ices as a means of E s t a b l i s h i n g the Value of Journey Time, the Rate of Time Preference and the Valuat ion of some Aspects of Environment i n the London Metropol i tan Region", Appl i ed  Economics. 3, pp 247-55. Walters , A A (1975), Noise and P r i c e s . Clarendon Press , Oxford. Walters , A A (1978), "Airports - An Economic Survey", Journal of Transport Economics and P o l i c y . 12 (2), pp 125-160. Walther, H 0 (1970), "Land Values Near A i r p o r t s " , i n World A i r p o r t s : The Way Ahead. Proceedings of the Conference held at the Queen E l i z a b e t h H a l l , London, 23,24,25 181 September, 1969, The I n s t i t u t i o n of C i v i l Engineers , London. Waters, W G, II (1977), "Investment C r i t e r i a and the Expansion of Major A i r p o r t s i n Canada", Canadian P u b l i c  P o l i c y . 3, pp 23-35. Weicher, J C, and D H a r t z e l l (1982), "Hedonic A n a l y s i s of Home P r i c e s : Results for 59 Metropol i tan Areas", i n C F Sirmans (ed. ) , Research i n Real Estate . Volume 2, JAI Press , Greenwich, Conn. , pp 267-91. Weisberg, S (1985), Appl ied Linear Regression. 2nd e d . , John Wiley and Sons, New York. Wesler, J E (1986), "Aviat ion Noise Descr iptors f or Federal Programs", Inter -no i se 86 - Proceedings. 1986 Internat iona l Conference on Noise Contro l Engineer ing , Cambridge, USA, pp 1379-82. Whitbread, M (1978), "Measuring the Costs of Noise Nuisance from A i r c r a f t : a Review A r t i c l e " , Journal of Transport  Economics and P o l i c y . 12, 202-8. Whitbread, M, and H B i r d (1973), "Rent, Surplus and the Evaluat ion of R e s i d e n t i a l Environments", Regional  Studies , 7, pp 193-223. Wiley , J R (1986), A i r p o r t Administrat ion and Management. Eno Foundation for Transportat ion , I n c . , Westport, Connect icut . Wi lk inson , R K (1973), "House Pr ices and the Measurement of E x t e r n a l i t i e s " , Economic Journa l . 83, pp 72-86. World Health Organizat ion (1980), Noise. Geneva. Zimmer, T M, and W E Burhenne (1972), " A i r f i e l d Noise Abatement i n the Federal Republic of Germany", Natural  Resources J o u r n a l . 12, pp 354-87. Zubkov, V D (1988), "Economic impl icat ions of j e t - n o i s e r e s t r i c t i o n s studied", ICAO B u l l e t i n . 43, pp 19-20. 182 APPENDIX A - DESCRIPTION OP NEF TECHNIQUE The fol lowing d e s c r i p t i o n of the NEF technique reproduces that given by Schultz (1982, pp 140-1): The bas ic measure, E (effective) PNL, i s a refinement of PNL [Perceived Noise Level] that takes account of s i gna l duration and the presence of pure tones: EPNL = PNL + D + L (A . l ) where PNL =maximum ca lcu la ted perceived noise l e v e l at any ins tant of time during the f l y o v e r , ca l cu la ted from th ird-oc tave band noise l e v e l s . D = 10 log t / 15 , where t i s the time i n t e r v a l i n second during which the noise l e v e l i s w i th in 10 dB of the maximum PNL. F = correc t ion for the presence of d i s c r e t e frequency components; the c o r r e t i o n i s tabulated according to the t h i r d -octave band i n which the tone l i e s and the extent to which the tone l e v e l exceeds the mean l e v e l i n the adjacent bands. The t o t a l noise exposure at a given point i s viewed as being composed of noise produced by d i f f e r e n t a i r c r a f t f l y i n g d i f f erent f l i g h t paths. For a s p e c i f i c c l a s s of a i r c r a f t i , on f l i g h t path j , the NEFij can be expressed: NEFij = EPNLij + 10 log - C (A.2) where n D and n„ are the numbers of operat ions for daytime (0700-2200) and night-t ime (2200-0700), r e s p e c t i v e l y , of a i r c r a f t c las s i , on f l i g h t path j . KD = 20; K„ = 1.2; C = 75: The choice of these constants s i g n i f i e s that a s i n g l e night-t ime f l i g h t contr ibutes as much to the NEF as 17 daytime f l i g h t s ; the value of C i s a r b i t r a r i l y chosen so that NEF numbers t y p i c a l l y l i e i n a range where they are not l i k e l y to be confused with other composite noise r a t i n g s . The t o t a l NEF at a given ground p o s i t i o n i s determined by summation of a l l the i n d i v i d u a l NEFij values on an energy bas is : NEF = 10 l og ^> ^ anti-log N£F;J (A.3) 183 APPENDIX B - SUMMARIES OF PREVIOUS STUDIES Thi s appendix provides b r i e f summaries of the t h i r t e e n i n d i v i d u a l a i r p o r t noise studies reviewed i n Chapter 3. They are presented i n a lphabet ica l order, with each summary commencing on a new page. 184 ABELSON, P W (1979), "Property Prices and the Value of Amenities", Journal of Environmental Economics and  Management, 6, pp 11-28. STUDY The study models the v a r i a t i o n i n house pr i ce s i n Sydney, A u s t r a l i a , between January 1972 and September 1973 and estimates hedonic pr i ce s for a i r c r a f t noise as wel l as road t r a f f i c , road widening, views, spacious s t ree t s , good access to shops, and high q u a l i t y neighbourhoods. MODEL To improve the p r e c i s i o n of the r e s u l t s i n the l i g h t of an imperfect property market, varying household preferences and c o s t l y household moves wi th in (as wel l as between) c i t i e s , the study area i s l i m i t e d to two large neighbourhoods, M a r r i c k v i l l e and Rockdale. Each of these areas i s modelled separate ly; i t i s reasoned that the r e s u l t i n g i m p l i c i t amenity pr i ce s cannot be appl ied to other areas of Sydney because the supply and demand condit ions which determine p r i c e s vary from area to area. The dependent v a r i a b l e i n the study i s house sa les p r i c e , used i n four regress ion model types: (i) ordinary l i n e a r regress ion i n which property a t t r i b u t e s have absolute d o l l a r va lues ; ( i i ) regress ion models i n which some of the independent var iab le s are expressed i n log or exponential form, but dummy var iab le s are unchanged ( logari thmic form impl ies d iminishing marginal costs or benef i t s ; exponential impl ies increas ing) ; ( i i i ) models i n which log of house p r i c e i s a function of l i n e a r v a r i a b l e s , so that housing a t t r i b u t e s are valued at a given percentage of house p r i c e ; and (iv) models i n which log of house p r i c e i s a funct ion of l og v a r i a b l e s and l i n e a r ones. Various regress ion models are estimated, inc lud ing separate ones for low and high p r i c e houses to check i f household incomes a f f ec t i m p l i c i t amenity p r i c e s . Proper t i e s which had undergone, or were l i k e l y to undergo, s i g n i f i c a n t land use change fo l lowing t h e i r sa le are excluded, as are propert ies with h igh ly unusual features (very extensive modi f icat ions , subdiv i s ions in to apartments, custom-bui l t apartments), g iv ing two f i n a l samples of 592 and 822 house sa les . Independent var iab le s tes ted i n l i n e a r and log form ( in at l e a s t one model) inc lude: . P h y s i c a l : l o t frontage, l o t depth, age of house, cons truct ion type, number of rooms, e x t e r i o r c o n d i t i o n , 185 improvements, block access, block l e v e l , property type, garage type , roof type. . Area: road t r a f f i c l e v e l , road b l i g h t , road width, higher zoning, view, publ ic transport q u a l i t y , access to shops, s o c i a l s tatus , proximity to sea. . Noise: arguing that the r e l a t i o n s h i p between NEF and household annoyance from a i r c r a f t noise i s not n e c e s s a r i l y l i n e a r , several forms of the NEF v a r i a b l e are t e s t ed , inc lud ing NEF, NEF-25 and e ( N E F " 2 5 ) / 1 ° . The l a t t e r two forms assume that at l ess than 25 NEF the NEF l e v e l can be considered equal to zero. In other words, i t i s assumed that a i r c r a f t noise i s i r r e l e v a n t at these lower l e v e l s . . Other: monthly i n f l a t i o n . Other data c o l l e c t e d , but not used, include buyer's occupation, buyer 's n a t i o n a l i t y , personal business , spec ia l rooms, swimming p o o l , r e g u l a r i t y of shape and s p e c i a l f a c t o r s . Other v a r i a b l e s considered, but for which data were not c o l l e c t e d , include house i n t e r i o r c h a r a c t e r i s t i c s , p r i v a c y , amount of sun i n winter, a i r p o l l u t i o n and distance to CBD. The l a s t v a r i a b l e was not included because a l o c a l area i s sampled for which there i s uniform q u a l i t y of p u b l i c t ransport and a v a i l a b i l i t y of l o c a l jobs . RESULTS M u l t i c o l l i n e a r i t y between qua l i ty a t t r i b u t e s (eg. views and good external house condition) i s lower than might have been expected. Nonetheless, i t i s found that area s o c i a l status c o r r e l a t e s with other p o s i t i v e amenities of the house and i t i s found that road t r a f f i c noise corre la te s with a i r c r a f t noise . The r e s u l t i n g R2 vary from .62 to .68 for M a r r i c k v i l l e and .37 to .62 for Rockdale. R2 are general ly s l i g h t l y h igher for models where ordinary house p r i c e rather than log of house p r i c e i s used, but because l i n e a r v a r i a b l e s are assoc iated with higher r e s i d u a l s , l og of house p r i c e i s p r e f e r r e d . Nearly a l l c o e f f i c i e n t s are s i g n i f i c a n t at the 95% l e v e l and possess expected s igns . The major determinants of house pr i ce d i f f e r e n t i a l s are house q u a l i t y and s i z e , land s i ze , and i n f l a t i o n . Environmental f ac tors and neighbourhood e f f ec t s exp la in a smal ler , but s t i l l s i g n i f i c a n t , proport ion of v a r i a t i o n . The importance of p a r t i c u l a r i n d i v i d u a l v a r i a b l e s d i f f e r s cons iderably f o r the two areas, but for M a r r i c k v i l l e , number of rooms accounts for 28% of v a r i a t i o n i n house p r i c e , l o t frontage for 9%, monthly i n f l a t i o n 8%, cons truc t ion type 4%, property type 7%, road b l i g h t 2%, e ( N E / " 2 5 ) / 1 ° 2%, with everything e l s e contr ibut ing less than 1%. Hedonic p r i c e s for a i r c r a f t noise suggest that the r e l a t i o n s h i p between 186 p r i c e and NEF i s bet ter represented by a nonl inear funct ion such as e ( N E F" 2 5 ) / 1 0 i n both areas. In M a r r i c k v i l l e , w i th in NEF 25, the model suggests a p r i c e d i f f e r e n t i a l of 6% (or 0.4% per NEF) for noisy houses. The model l ing reveals the importance of inc luding a large number of v a r i a b l e s : for example for M a r r i c k v i l l e , the NEF c o e f f i c i e n t increases by 60% i f t r a f f i c noise and e x t e r i o r cond i t ion var iab les are not inc luded. In Rockdale, no s i g n i f i c a n t r e l a t i o n s h i p for noise i s i d e n t i f i e d , except for higher p r i c e d homes where there i s a p r i c e d i f f e r e n t i a l of 10% (or 0.5% per un i t on the NEF scale) for noisy homes (but only s i g n i f i c a n t at 90%). Suggested reasons for t h i s inc lude the existence of some counterva i l ing a t t r i b u t e (eg. access to work) which i s not otherwise taken in to account. A l s o , the NEF model i s not always a good measure of annoyance, e s p e c i a l l y at low numbers of a i r c r a f t movements. I t i s stated to be "absurd to suggest that there i s no a i r c r a f t noise problem i n Rockdale". Household survey r e s u l t s are quoted which show 80% of households underestimated a i r c r a f t noise before they moved into the area . Resul ts are also presented for road t r a f f i c , road widening, views, block l e v e l s , spacious roads and other environmental amenit ies . Of note i s the p r i c e d i f f e r e n t i a l for proximity to a rai lway: such houses are 12% lower i n p r i c e . Th i s v a r i a b l e was expected to r e f l e c t the advantage of p u b l i c t ransport access, but t h i s i s apparently outweighed by noise disadvantage. Nelson (1980) derives an NDSI of 0.40% for M a r r i c k v i l l e and 0.50% for Rockdale. 187 DE VANY, A S (1976), "An Economic Model of Airport Noise Pollution in an Urban Environment", in 8 A Y Lin (ed.)/ Theory and Measurement of Economic Externalities, Academic Press, New York, pp 205-14. STUDY The study presents a t h e o r e t i c a l analys i s based on the argument that the slope of housing pr ices as one moves away from an a i r p o r t must represent a balancing of t r a n s p o r t a t i o n costs and noise exposure that achieves e q u i l i b r i u m across the d i s tance . This n o i s e - a c c e s s i b i l i t y t r a d e - o f f theory for d e r i v i n g the t o t a l premium associated with increas ing a c c e s s i b i l i t y i s used to exp la in some well-known patterns of land-use such as commercial zones forming around a i r p o r t s , and the fr inge of apartments a l so surrounding a i r p o r t s . The study then uses an empir i ca l model to t e s t the r e l a t i o n s h i p between a i r p o r t noise and housing p r i c e s for Dal las i n 1970. MODEL A l i n e a r model i s t e s ted , with 1270 observations from block l e v e l data from the 1970 Census of Housing being used for housing c h a r a c t e r i s t i c s and p r i c e s . The dependent v a r i a b l e i s the census block mean property values (as estimated by owners). Explanatory v a r i a b l e s include: . P h y s i c a l : average number of rooms per house, average age of housing, percentage of homes which are a i r -condi t ioned. . Area: percentage of homes that are owner-occupied, average length of stay of res idents . . A c c e s s i b i l i t y : d is tance from a i r p o r t (using four dummy v a r i a b l e s for four c i r c u l a r distance r ings around the a i r p o r t ) , d is tance from CBD (four dummies, l ikewise for downtown). . Noise: noise e f f ec t s are captured i n the v a r i a b l e for distance from the a i r p o r t . Because the t o t a l a c c e s s i b i l i t y premium i s not represented by a s i n g l e s lope , the p a r t i t i o n i n g by distance i s requ ired to i s o l a t e the inf luence of a i r p o r t noise . RESULTS The publ i shed regress ion has an R2 of .8193, wi th the fo l lowing v a r i a b l e s s i g n i f i c a n t for a two- ta i l ed 95% t e s t : owner-occupied, number of rooms, age, a i r - c o n d i t i o n i n g , and 188 severa l of the distance dummies. The re su l t s suggest a s i z eab le reduction i n value i f located with in 1 mile of the a i r p o r t , an inconsequential e f f ec t on value i f between 1 and 2 mi l e s , an increase i n value i f between 2 to 3 mi les , and no net e f fect on value a f t er 3 mi les . The conclus ion i s that environmental degradation due to a i r p o r t noise overwhelms loca t ion advantages wi th in 1 mi le ; the opposing forces of fset each other from 1 to 2 mi les ; the proximity more than of fsets environmental e f fec ts for 2 to 3 mi les ; beyond 3 miles , values are not affected by the presence of the a i r p o r t . (Distance to the CBD i s s i m i l a r l y discussed. ) I t i s concluded that the o v e r a l l e f fec t of the a i r p o r t on housing values i s p o s i t i v e , with the d o l l a r los s of value by homes with in 1 mile of the a i r p o r t being more than o f f se t by the much bigger value gain by homes l y i n g 2 to 3 miles from the a i r p o r t . Nelson (1978a, p . I l l ) der ives an NDSI of 0.80%. When the sample i s p a r t i t i o n e d to include only propert ies 2 to 3 miles from the a i r p o r t , the NDSI i s lower at 0.58% (NRC 1977, p.139). Nelson (1979, 1981a) discusses the d i f f i c u l t i e s with the model's use of c i r c u l a r r ings to approximate noise d i f f e r e n t i a l s and the r e s u l t i n g noise deprec ia t ion impact when noise contours for the a i r p o r t are d i s t i n c t l y n o n c i r c u l a r i n shape. 189 DYGERT, P K (1973), "Estimation of the Cost of Aircraft Noise to Residual Activities", unpublished Ph.D. dissertation, University of Michigan, SUMMARIZED IN NELSON (1978a, 1980). STUDY The study invest igates the r e l a t i o n s h i p between a i r c r a f t noise and property value for two communities, San Francisco and San Jose, i n 1970. MODEL A two-equation model i s used, with the equation of p a r t i c u l a r i n t e r e s t being i n semilog form and having as the dependent v a r i a b l e the log of median (or mean) assessed land value per square foot from census t r a c t data . Explanatory v a r i a b l e s inc lude: . P h y s i c a l : t e r r a i n . . Area: p u b l i c schools, median number of people per u n i t , percentage of nonwhite u n i t s . . P u b l i c Sector: property tax r a t e . . A c c e s s i b i l i t y : shopping centres , i n d u s t r i a l s i t e s , a i r p o r t terminals , CBD. . Noise: a v a r i a b l e i n NEF u n i t s (NEF 25 to 45) was entered i n exponential form. Nelson (1978a, p.105) s tates that areas outside the NEF 25 contour are assigned a noise l e v e l of zero. T h i s w i l l b ias the noise c o e f f i c i e n t toward zero, l ead ing to a measurement e r r o r . Some estimates of Composite Noise Rating (CNR) are a l so used. The f u l l sample s i ze for San Franc isco i s 128 observations, and for San Jose i s 198. A number of regress ions are tes ted based on p a r t i t i o n i n g the two samples by noise l e v e l s , by the propor t ion of s ing l e - fami ly dwel l ings , and by the census median owner-assessed property value i n each t r a c t . RESULTS For San Franc i s co , the noise c o e f f i c i e n t v a r i e s considerably by noise l e v e l , tending to c l u s t e r around -0.005 although ranging from -0.004 to -0.034. For San.Jose , c o e f f i c i e n t s range from -0.007 to -0.015 but do not revea l much tendency to c l u s t e r around a c e n t r a l va lue . T y p i c a l R2 range from 0.60 to 0.70. Nelson (1978a, p.106) quest ions the study's conclus ions because of the non-allowance for background or 190 r e s i d u a l noise l e v e l s , the uncerta in requirement for a two-stage l eas t squares simultaneous equation procedure, and the p o s s i b l e e f fec t of the use of un i t land values rather than property values . Nelson further suggests that the NDSI for San Franc i sco i s at l e a s t 1.5%, but could be as high as 2.0%, and for San Jose i s apparently 0.7%, but could be as high as 1.5% us ing a r e s t r i c t e d sample. 191 EMERSON, F C (1972), "Valuation of Residential Amenities: An Econometric Approach", Appraisal Journal. 40 (2), pp 268-78. STUDY The study models house pr i ce s using 1967 data for a p r i m a r i l y r e s i d e n t i a l par t of Minneapolis to determine the e f f ec t s of environmental amenities such as proximity to parks , freeways and a i r p o r t operat ions. The study includes an elementary d i s cus s ion of the regress ion technique used. MODEL The mul t ip le regress ion model used i n the study takes a logar i thmic form. Data comes p r i m a r i l y from M u l t i p l e L i s t i n g Service (MLS) sheets and includes only dwel l ings on l o t s zoned for s i n g l e - f a m i l y r e s i d e n t i a l use. Where MLS sheets noted d e t e r i o r a t i o n of the s t ruc ture , the observat ion i s excluded, g i v i n g a complete sample of 222 for two-and-a-h a l f months i n 1967. Both the f u l l sample and three segments of i t grouping residences by p r i c e are analyzed. The dependent v a r i a b l e i s actual sale p r i c e (as recorded by the MLS sheet) . Independent var iab le s which are tes ted inc lude: . Phys i ca l : house s i z e , l o t s i z e , garage space, baths, f l o o r s , cons truc t ion type, ranges, carpets and drapes, f i r e p l a c e s , extras (eg. porches) , house age, corner l o c a t i o n , a r t e r i a l l o c a t i o n . . Area: d i s tance to open green space, nonwhite proport ion of populat ion , neighbourhood d i l a p i d a t i o n , distance to schoo l , freeway access, proximity to freeway. . Publ ic sec tor: number of blocks to bus. . A c c e s s i b i l i t y : distance to commercial corner . . Noise: the l e v e l of exposure to a i r c r a f t noise for each l o c a t i o n i s indexed using the Composite Noise Rating (CNR), with observed CNR ranging from 90 to 125 i n a neighbourhood about a mile and a h a l f o f f the northwest end of the main runway. The study assumes some background community noise l e v e l by s e t t i n g various thresho ld CNRs: CNR 100 (which equals about NEF 30) i s used as a threshold i n the f i n a l f u l l sample model. The "quietness" or "freedom-from-aircraft -nuisance" v a r i a b l e used i s thus equal to the l e s s e r of (126-CNR) and 26. I t i s noted that at the time of the study there was an unusually high concentrat ion of 192 operat ions on the main runway because of two other runways being c losed for r e p a i r s . This suggests that the per iod i s poss ib ly not representat ive of long-run c o n d i t i o n s . . Other: low in teres t mortgage. The large number of independent v a r i a b l e s tes ted makes m u l t i c o l l i n e a r i t y a very l i k e l y problem. There i s no i n d i c a t i o n of the e f fec t of any a l t e r n a t i v e model s p e c i f i c a t i o n s (Nelson, 1978a, p.102) . RESULTS The f u l l sample regress ion model r e s u l t s revea l about h a l f of the v a r i a b l e s tested are s i g n i f i c a n t . S i g n i f i c a n t v a r i a b l e s (using a one- ta i l ed t - t e s t ) inc lude the fo l lowing: w i t h i n two l o t s of freeway, open green space, nonwhite p o p u l a t i o n , square feet i n house, square feet on l o t , garage space, baths, ranges, f i r e p l a c e s , age ( a l l at 95%), distance to s choo l , freedom from a i r c r a f t no ise , stone fac ing ( a l l at 90%), f l o o r s , stucco fac ing ( a l l at 85%). This gives an R2 of .798. For the three segments, grouping by p r i c e , the R2 ranged from .418 to .702 (noted by Nelson 1978a, p.102). The f u l l sample model y i e l d s an e l a s t i c i t y of freedom-from-a i r c r a f t - n u i s a n c e of 0.00316, though the v a r i a b l e i s only s i g n i f i c a n t at the 90% l e v e l . Of more concern i s the marked d i f f e r e n c e s i n noise c o e f f i c i e n t s when p a r t i t i o n e d regress ions are run: c o e f f i c i e n t s range from 0.0014 for medium-priced propert ies to 0.0628 for low-value proper t i e s . (Nelson 1978a, p.102). Implied reduct ions i n (nuisance-free) p r i c e s for residences exposed to a CNR of 125 were 9.8% i n the f u l l sample. The impl ied marginal r e l a t i o n s h i p i s such that Nelson (1980) ca l cu la te s the damage cost to equal about a 0.58% reduction i n average property value for each u n i t increase on the NEF sca le (NDSI). 193 McDOUGALL, G S (1976), "An Inquiry into the Demand for A i r c r a f t Noise Abatement", Review of Regional S tud ies , 6 (3) , pp 59-69. STUDY The study estimates i m p l i c i t demand, as revealed by r e s i d e n t i a l s i t e choice and therefore property va lues , f or a i r c r a f t noise abatement. I t uses t h i s to estimate the amount households are w i l l i n g to pay for an increase i n environmental q u a l i t y associated with a reduct ion i n a i r c r a f t noise . Hedonic regressions are used to model the environmental qual i ty-house value gradient for Los Angeles i n 1970. MODEL A log s p e c i f i c a t i o n of the model i s used for data der ived from information concerning a sample of "representative" census t r a c t s wi th in the t h i r t y - f i v e planning areas of the C i t y of Los Angeles. Only t r a c t s located i n the centre of a p lanning area are considered, to avoid any external or s p i l l - o v e r e f fec ts between contiguous planning areas . The dependent v a r i a b l e i s the census t r a c t median property value for owner-occupied housing u n i t s . Explanatory v a r i a b l e s inc luded i n the model are: . P h y s i c a l : average house s i z e . . Area: neighbourhood q u a l i t y and land use c h a r a c t e r i s t i c s are measured by a s c a l a r r e f l e c t i n g landscaping, amount of l i t t e r , number of vacant l o t s , s t ree t condi t ions , convenience fac tors , land use, l o t s i z e , dwel l ing s i z e and house s tructures , based on subject ive rankings by the C i t y Planning Department. . Pub l i c sector: r e l a t i v e q u a l i t y and convenience of bus serv ice (the sum of the average times spent walking to the nearest bus stop and wait ing for a bus to a r r i v e , weighted by the average number of t r i p s per day); the a v a i l a b i l i t y of l o c a l l y - p r o v i d e d c o l l e c t i v e goods as measured by the median reading score on the "Iowa Tests of Education Development" weighted by school s i ze and the proport ion of each school neighbourhood wi th in a planning area (assuming i n the long-run that p u p i l s attend the c loses t schools to t h e i r homes). . A c c e s s i b i l i t y : no measure of distance or t r a v e l time to the CBD i s u t i l i z e d because economic o p p o r t u n i t i e s and a c t i v i t i e s are widely dispersed throughout the C i t y of Los Angeles. 194 . Noise: r e c i p r o c a l of the r e l a t i v e noise i n t e n s i t y as measured by a weighted f igure of the percent of each planning area that i s subject to noise l e v e l s equal to or greater than 90 dec ibe l s (areas with noise i n t e n s i t y equal to or greater than 100 dB are given twice the weight of other areas ) . RESULTS The Household Opportunity Locus r e s u l t i n g from the model shows house s i ze and q u a l i t y of school to be s i g n i f i c a n t . With a c o e f f i c i e n t of .015766, the noise v a r i a b l e ( s ign i f i cance l e v e l not reported) suggests that house value i s r e l a t i v e l y i n e l a s t i c with respect to noise i n t e n s i t y d i f f e r e n t i a l s . Nelson (1980, p.49) states that a NDSI cannot be ca l cu la ted from t h i s information. The h igh degree of c o l l i n e a r i t y between school and neighbourhood v a r i a b l e s introduces a problem of "near m u l t i c o l l i n e a r i t y " . These r e s u l t s are used to speci fy a demand equation r e l a t i n g to both p r i c e s and income. A two-stage l eas t squares estimate of the demand r e l a t i o n for environmental u t i l i t y , us ing a l o g - l i n e a r s p e c i f i c a t i o n , i s made to y i e l d d i r e c t demand e l a s t i c i t y est imates. By i n t e g r a t i n g over the i n d i v i d u a l 1 s demand curve and then summing over the re levant populat ion, the authors estimated the benef i t associated with non-marginal changes i n a i r c r a f t no ise . The r e s u l t i n g benef i t f igure may be considered the compensating v a r i a t i o n associated with noise p o l l u t i o n , or the amount households would be w i l l i n g to pay for a decrease i n experienced noise l e v e l s . 195 MCMILLAN, M L, B G REID, and D W GILLEN (1980), "An Extension of the Hedonic Approach for Estimating the Value of Quiet", Land Economics, 56 (3), pp 315-28. STUDY The s tudy's o v e r a l l object ive i s to determine households' true w i l l i n g n e s s to pay for quiet from information provided by hedonic es t imat ion , such that estimates of the value of non-marginal changes can be es tabl i shed. An estimate of the p r i c e of qu ie t i s obtained by applying an hedonic model to the r e s i d e n t i a l area i n Edmonton for the per iod September 1975 to September 1976. MODEL The model uses i n d i v i d u a l data for 352 house sa les i n the af fected area . A log model i s used for the f i n a l form of the estimated equation, although other func t iona l forms are t e s t ed . Information for the dependent v a r i a b l e , sa les p r i c e and house c h a r a c t e r i s t i c s are obtained from M u l t i p l e L i s t i n g data . Explanatory v a r i a b l e s include: . P h y s i c a l : f l o o r area, age, number of bathrooms, l o t s i z e , number of bedrooms, f i r e p l a c e , number of s t o r i e s , duplex, f i n i s h e d basement, b r i c k e x t e r i o r , garages. . Area: zoning. . P u b l i c sector: e f fec t ive property tax r a t e . . A c c e s s i b i l i t y : distance to CBD. . Noise: NEF 20 to 39, spec i f i ed ( for quiet) as the inverse of (40-NEF); background noise l e v e l of NEF 20. . Other: monthly dummy. The study reasons that even though (other) p u b l i c sector v a r i a b l e s are omitted because of lack , of data , i t may be assumed that p u b l i c serv ices are uniformly provided because a l l observat ions are wi th in one m u n i c i p a l i t y . RESULTS A l l v a r i a b l e s except b r i c k exter ior and zoning are s i g n i f i c a n t at the 95% l e v e l (one- ta i l test ) with a l l c o e f f i c i e n t s having the expected s igns . The R2 f or the f i n a l form i s .713. In the reported equat ion, the marginal value of qu ie t depends on the l e v e l of q u i e t . Nelson (198 0, p.49) c a l c u l a t e s NDSI to be about 0.5% for e a r l i e r , but only s l i g h t l y d i f f e r e n t , r e s u l t s from t h i s study. 196 MASER, S M, W H RIKER, and R N ROSETT (1977), "The Effects of Zoning and Externalities on the Price of Land: An Empirical Analysis of Monroe County, New York", Journal of  Law and Economics, 20, pp 111-32. STUDY The objec t ive of t h i s study i s to determine whether zoning s i g n i f i c a n t l y modifies outcomes i n the urban land market or whether market forces negate the forces of regu la t ion . As a s ide i s sue , i t examines the e f fect of a i r p o r t noise in the urbanized area of Rochester, New York i n 1971. I t i s suggested that a p r i o r i e i ther hypothesis can be argued. Samples are drawn of r e a l estate t r a n s a c t i o n s , c h a r a c t e r i s t i c s for the parcels that a f f e c t the p r i c e of land are measured, and regression a n a l y s i s i s used to estimate an hedonic p r i c e index. The zoning categories are inc luded as independent var iab les s ince zoning creates a r t i f i c i a l s c a r c i t i e s and thus rents . MODEL A l i n e a r form of the hedonic model i s used and data c o l l e c t e d for a random sample of about one-s ix th of a l l r e a l estate t ransact ions recorded during each of three years: 1950, 1960 and 1971. Altogether ten samples are analyzed, depending on the year, the property type ( s ing le - fami ly , m u l t i p l e - f a m i l y , commercial and i n d u s t r i a l ) and whether for the c i t y or suburbs. Of in teres t i s the ana lys i s of s i n g l e -family proper t i e s for c i t y (393 observations) and suburbs (990 observations) for 1971. The dependent v a r i a b l e i s the i n d i v i d u a l house sa le p r i c e per acre of land plus s tructure . Observations are excluded from the sample i f the transfers were not arm's- length t ransac t ions ( for example, sales i n v o l v i n g government agencies , in tracorporate or i n t r a f a m i l y t rans fer s or no pecuniary exchange). Data were obtained from the terms of sa le from the deed recorded i n the County C l e r k ' s O f f i c e . The same source i s used for data on p h y s i c a l c h a r a c t e r i s t i c s . Local municipal o f f i c e s , Po lk ' s "City D irec tory" , the 1969 "Rochester Metropo l i tan Transportat ion Study" and the Census of Housing Block S t a t i s t i c s provide other data . For metropolitan one-family houses i n 1971, explanatory v a r i a b l e s include : . P h y s i c a l : maintenance. . Area: v i s i b l e uses (mul t i - fami ly r e s i d e n t i a l , commercial, i n d u s t r i a l , dump or slum, body of water, land r e c r e a t i o n , expressway or busy s t r e e t , pub l i c b u i l d i n g , vacant land) , adjacent uses (two-family r e s i d e n t i a l , walk-up apartment, p r o f e s s i o n a l o f f i c e , 197 neighbourhood, business, CBD or shopping centre , l i g h t indus try , heavy industry , gas s t a t i o n , dump or slum, r e c r e a t i o n , p u b l i c b u i l d i n g , vacant land) , type of s t r e e t , populat ion densi ty , % black populat ion , % 62 and over, % i n t en-un i t b u i l d i n g , average number of rooms owner-occupied, average number of rooms for renter-occupied , zoning variance (area, dens i ty , commercial) . . Pub l i c sector: school q u a l i t y (per p u p i l expenditure) and property tax assessment (highly c o r r e l a t e d with each other and other v a r i a b l e s , and never t e s ted) . . A c c e s s i b i l i t y : average d r i v i n g time to CBD. . Noise: a dummy v a r i a b l e to ind ica te whether the property i s wi th in the 100 dB ground noise contour for the Monroe County A i r p o r t . . Other: equal ized assessed value of s t ruc ture d i v i d e d by acreage, sale by trus tee , mortgage. RESULTS S i g n i f i c a n t v a r i a b l e s i n the 1971 f i n a l form equation for the C i t y of Rochester one-family houses inc lude: equal ized assessment value of s tructure d iv ided by acreage, access to CBD, a i r p o r t noise , sa le by trus tee , mortgage, maintenance, % b lack populat ion , and average number of rooms i n owner-occupied u n i t s . For t h i s sample the R2 i s .60. For 1971 suburban s i n g l e - f a m i l y use, the R2 i s .79. Tes t ing the hypothesis that zoning or any other v a r i a b l e has an e f f e c t on land p r i c e s , the e f fec t of excluding the v a r i a b l e i s noted. Various tes t s reveal no evidence e i t h e r that zoning has a p r i c e - d i s t o r t i n g e f f e c t , or that the uses of land which zoning i s supposed to prevent impose ex terna l costs on neighbours. However, a i r p o r t noise has a s t a t i s t i c a l l y s i g n i f i c a n t e f f ec t with the cost per house i n the c i t y being lower than the cost per house i n the suburbs (where houses per acre are fewer). Nelson (1978a, p.49) derives an NDSI of about 0.55% to 0.68% per NEF. 198 MIESZKOWSKI, P, and A M SAPER (1978), "An Es t i m a t e o f the E f f e c t s o f A i r p o r t Noise on P r o p e r t y V a l u e s " , J o u r n a l o f  Urban Economics, 5, pp 425-40. STUDY The study uses a sample of housing transact ions i n the proximity of Toronto A i r p o r t to see whether houses s i tuated i n high noise contours are so ld at a discount r e l a t i v e to houses located i n areas unaffected by a i r c r a f t noise . The study notes the usefulness of discount measurement i n p u b l i c p o l i c y for such things as compensation determination, cost ings of a i r p o r t a l t e r n a t i v e s and evaluat ion of various land use a l t e r n a t i v e s . I t i s a p a r t i a l or l i m i t e d study i n the sense that i t d e t a i l s only the s tructure of r e s i d e n t i a l land values i n the v i c i n i t y of a p a r t i c u l a r a i r p o r t rather than the global e f fects which w i l l inf luence absolute land p r i c e s as we l l . MODEL The model uses two samples of housing transact ions drawn from the " r e l a t i v e l y stable" per iod January 1969 to June 1973 (after some re levant p o l i t i c a l dec is ions had been announced and when t r a f f i c volume, but not s t r u c t u r e , was changing): 509 i n Etobicoke and 626 i n Miss i ssauga. The authors claim these communities comprise r e l a t i v e l y small areas (although Nelson 1978a, p.108 h ints otherwise) so as to abstract from a c c e s s i b i l i t y factors and matters r e l a t e d to p u b l i c serv ices , taxes and neighbourhood e f f e c t s . Contro l groups were c a r e f u l l y chosen (with assumed zero NEF or CNR values) so that houses i n the noise less areas are roughly of s i m i l a r type and age, e tc . For Miss issauga, there are no houses outs ide the noise contours which are contiguous to noise a f fec ted areas, causing problems. For Etobicoke, one contro l area i s chosen by v i s u a l in spec t ion , and one by sampling an arc of transact ions outs ide the af fected area about the average width of one noise contour area away. This process i s only p a r t i a l l y success fu l as houses i n the contro l groups are on average super ior to the no ise -af fec ted houses. Because of t h i s c o l l i n e a r i t y problem, i t i s argued that one should be wary of accept ing r e s u l t s at face value . The model uses l i n e a r and semilog models with dummy v a r i a b l e s to represent p a r t i c u l a r time i n t e r v a l s (to c o n t r o l for i n f l a t i o n ) . The dependent v a r i a b l e i s i n d i v i d u a l house sa le p r i c e . Explanatory v a r i a b l e s inc lude: . Phys i ca l : house s i z e , house s i ze squared, average room s i ze , l o t s i z e , l o t s i z e squared, number of bathrooms, number of bedrooms, number of u t i l i t y rooms, 199 number of s t o r i e s , number of garages, type of garage, forced a i r heat ing, f u l l basement, number of rooms i n basement, storms and screens, carpet ing , b u i l t - i n appl iances , TV antenna, draper ies , own hot water heater, fenced l o t , sun room, landscaping, p a t i o , number of f i r e p l a c e s , s o l i d b r i c k , semidetached, fancy s t y l e , s p l i t l e v e l , house condi t ion , time s o l d . Most of these are descr ibed by dummy v a r i a b l e s , many of which are m u l t i p l i e d by house s i z e . . Noise: a number of d i f f e r e n t s p e c i f i c a t i o n s . There i s a reasonably c lose correspondence between NEF 25 to 35 (1971 data) and CNR 95 to 115 (1975 to 1976 data) noise contours; both are t r i e d for Etobicoke and CNR alone i s t r i e d for Miss issauga. The study notes that the varying tolerance of noise by d i f f e r e n t segments of the populat ion suggests that any estimate of a noise discount w i l l be a lower bound estimate for aversion to a i r p o r t noise for a t y p i c a l person, given that persons l e a s t adverse to noise w i l l n a t u r a l l y se lec t the s i t e s i n the n o i s i e s t areas. I t i s argued that errors due to vary ing preferences and other complications are l i k e l y to be small r e l a t i v e to the est imation e r r o r s due to sampling v a r i a b i l i t y and the imperfect s p e c i f i c a t i o n of the house value r e l a t i o n s . RESULTS Using a l l observat ions , R2 i s .81, but there are a large number of v a r i a b l e s which perform i n an incons i s tent way or have very small e f f ec t s and/or s t a t i s t i c a l s i g n i f i c a n c e . S i g n i f i c a n t v a r i a b l e s inc lude: house s i z e , average room s i z e , log s i z e , number of bathrooms, number of s t o r i e s , garage, f u l l basement, storms and screens, carpet ing , b u i l t -i n appl iances , TV antenna, landscaping, f i r e p l a c e , s o l i d b r i c k , semidetached, fancy s t y l e , s p l i t l e v e l and noise . The regress ion r e s u l t s imply s i g n i f i c a n t discounts for a i r p o r t noise o v e r a l l , with most forms y i e l d i n g h igh ly s i g n i f i c a n t est imates . The Mississauga discount r e s u l t s are much higher (two or three times) although less reasonable than those for Etobicoke . For the Etobicoke sample, semilog and l i n e a r r e s u l t s are qui te consis tent for both CNR and NEF measures, but greater d i s p a r i t y ex i s t s between semilog and l i n e a r estimates for the Mississauga sample. The r e s u l t s inc lude a very wide range of discount estimates: from 3.5% to 6.1% for CNR 95, 5.1% to 10.0% for CNR 100, 7.6% to 15.0% for CNR 105. For Etobicoke, the discount estimates are 6.4% for NEF 25, 4.6% for NEF 30 and 7.8% for NEF 35. There i s a dec l ine i n some noise discounts with higher noise l e v e l s : t h i s may be due to small samples at the higher noise l e v e l s , or a l t e r n a t i v e l y the enhanced 200 commercial value or superior a c c e s s i b i l i t y of those areas c l o s e s t to the a i rpor t (Nelson 1978a, p.108). Nelson (1978a, p. 108) uses the NEF regressions for Etobicoke to obtain a weighted mean NDSI of 0.9%. The study uses discount est imates, number of houses and p r i c e information to estimate the t o t a l c a p i t a l i z e d s o c i a l l o s s assoc iated with a i r p o r t no ise . Th i s loss i s used to determine an a i r p o r t passenger tax to f u l l y compensate r e s i d e n t s . Then, taking into account increases i n populat ion and a i rpor t t r a f f i c , the tax i s adjusted a c c o r d i n g l y . 201 NELSON, J P (1978a), Economic Analysis of Transportation  Noise Abatement, Ballinger, Cambridge, Mass. STUDY The chapter e n t i t l e d "Empirical A n a l y s i s of Res ident ia l Property Values and Transportat ion Noise for the Washington, D .C . SMSA" has the object ive of determining the c o v a r i a t i o n between r e s i d e n t i a l property values and transportat ion noise i n the Washington, DC Standard Metropol i tan S t a t i s t i c a l Area for 1970. The study undertakes a pre l iminary analys i s before obta in ing empir ica l r e s u l t s . These r e s u l t s , together with those for other studies reviewed i n the subsequent chapter, are used to estimate a range of marginal damage for a i r c r a f t noise . MODEL The t o t a l study area includes a l l 630 r e s i d e n t i a l census t r a c t s i n the metropol i tan area for 1970. Of those a v a i l a b l e , r e s i d e n t i a l t r a c t s are considered i f they include at l e a s t 200 s i n g l e - f a m i l y , owner-occupied housing u n i t s , g i v i n g a bas ic sample of 467 census t r a c t s from which a large set of data i s assembled. I t i s argued that although the area may i n some ways be unique from a property value standpoint , the empir ica l r e l a t i o n s h i p s should be at l eas t representat ive of the s i t u a t i o n i n other urban areas. Because of the aggregation of 414 000 s i n g l e - f a m i l y dwel l ings and 485 000 m u l t i p l e - f a m i l y u n i t s in to 467 bundles, the f u l l range of v a r i a t i o n for some v a r i a b l e s i s u n l i k e l y to be observed. The dependent v a r i a b l e i s the census t r a c t mean property value , as estimated by the owners. The study compares these estimates with p r o f e s s i o n a l l y assessed estimates (based on r e a l estate tax assessment records) and f inds the zero-order c o r r e l a t i o n s are qui te h igh . The mean values for the two a p p r a i s a l methods d i f f e r by only 4 per cent of the census mean va lue , suggesting that large response errors are u n l i k e l y . (It i s a l so noted that , to the extent that response e r r o r s i n the census data are random a f t e r aggregation, the best l i n e a r unbiased propert ies of the least -squares est imators are unaffected so long as measurement e r r o r s occur only i n the dependent v a r i a b l e . ) The study discusses the poss ib le consequences of high m u l t i c o l l i n e a r i t y . I t notes that the s i z e and diverse nature of the sample used i n the study may help overcome the problem of i n t e r c o r r e l a t i o n among the explanatory v a r i a b l e s . Because a ra ther large number of v a r i a b l e s may p o t e n t i a l l y inf luence r e s i d e n t i a l property va lues , i t i s more l i k e l y that the problem of s p e c i f i c a t i o n e r r o r outweighs that of m u l t i c o l l i n e a r i t y . Two measures of m u l t i c o l l i n e a r i t y based 202 on c o r r e l a t i o n coe f f i c i en t s are included i n the ana lys i s . To cope with the problem, pre l iminary a n a l y s i s i s c a r r i e d out with models not inc luding t r a f f i c or a i r c r a f t noise to s e l e c t the funct iona l form, the sample, and the i n i t i a l set of explanatory var iab le s representing housing c h a r a c t e r i s t i c s . Af ter t h i s stage a measure of t r a f f i c noise i s added and the robustness of the noise coe f f i c i en t examined i n the l i g h t of a l t ernat ive model s p e c i f i c a t i o n s . The same i s done with a i r c r a f t noise . The s e l e c t i o n of funct ional form using a r e s t r i c t e d set of explanatory v a r i a b l e s includes cons iderat ion of l i n e a r , l o g , semilog and inverse semilog models. The log model i s s e l ec ted for further use because summary s t a t i s t i c s i n d i c a t e d i t s general super ior i ty (although only marginal ly so) , and the model of hedonic p r i c e s suggests that i t would be unwise to place too many r e s t r i c t i o n s on these p r i c e s . In s e l e c t i n g samples, s t r a t i f i c a t i o n by urban and suburban t r a c t s i s t es ted but found to be unnecessary. Other tes ts are performed on i n d i v i d u a l regress ion c o e f f i c i e n t s , but these a l so suggest an u n s t r a t i f i e d housing market. The regress ion res iduals are screened for o u t l i e r s and for h e t e r o s c e d a s t i c i t y . Plots do not suggest the presence of h e t e r o s c e d a s t i c i t y , but do indicate severa l o u t l i e r s which lead to the exc lus ion of several census t r a c t s on the bas is of unique features of the Washington property market. In s e l e c t i n g housing a t t r ibutes for i n c l u s i o n i n the model, explanatory v a r i a b l e s experimented with inc lude the f o l l o w i n g : . P h y s i c a l : number of rooms, house s i z e , l o t s i ze , age, c e n t r a l a i r - c o n d i t i o n i n g , l ack of f l u s h t o i l e t . . Area: % of t o t a l uni ts occupied by b lacks , property crime r a t e , l oca t ion near r i v e r , two measures of a i r p o l l u t i o n (average monthly p a r t i c u l a t e concentrat ion, average summer oxidant concentrat ion) . . P u b l i c sector: e f f ec t ive property tax r a t e , educat ional l e v e l per p u p i l i n appropriate pub l i c school d i s t r i c t . . A c c e s s i b i l i t y : automobile t r a v e l time to CBD, t r a v e l t ime to 75% of metropolitan employment opportuni t i e s . . Noise: day-night average sound l e v e l ( L ^ , i n dec ibe l s ) for t r a f f i c noise; f or a i r c r a f t noise , NEF 25, 30, 35, or 20 for background noise l e v e l . For the a i r c r a f t noise model, the se l ec ted sample consis ts of 52 census t r a c t s (only ten of which have NEF values grea ter than NEF 20) which are considered to be r e l a t i v e l y 203 homogeneous i n terms of most housing c h a r a c t e r i s t i c s , but which s t i l l exh ib i t a f a i r range of v a r i a t i o n i n the dependent v a r i a b l e and some independent v a r i a b l e s inc lud ing a i r c r a f t noise . RESULTS The model for a i r c r a f t noise and r e s i d e n t i a l property values has an R2 of .863, with the fo l lowing var iab le s s i g n i f i c a n t at the 95% confidence l e v e l (one- ta i l ed t - t e s t ) : number of rooms, age, c en tra l a i r - c o n d i t i o n i n g , l oca t ion near a r i v e r . The NEF l e v e l i s s i g n i f i c a n t at the 90% l e v e l . The r e s u l t s a l low the NDSI to be ca lcu la ted at 1.1% 204 NELSON, J P (1979), " A i r p o r t N o i s e , L o c a t i o n R e n t , and th e Ma r k e t f o r R e s i d e n t i a l A m e n i t i e s " , J o u r n a l o f E n v i r o n m e n t a l  Economics and Management, 6, pp 320-31; and NELSON J P (1981a ) , " M e a s u r i n g B e n e f i t s o f E n v i r o n m e n t a l Improvements: A i r c r a f t N o i s e and Hedonic P r i c e s " , i n S m i t h , V K ( e d . ) / Advances i n A p p l i e d M i c r o e c o n o m i c s , Volume 1, J A I P r e s s , G r e e n w i c h , Conn., pp 51-75. STUDY The study examines the influence of a i r c r a f t noise on housing p r i c e s at San Francisco , St Lou i s , Cleve land, New Orleans , San Diego, and Buffalo for 1970. I t seeks to take account of De Vany's (1976) argument that e a r l i e r studies contain a s p e c i f i c a t i o n error by f a i l i n g to c o n t r o l for a c c e s s i b i l i t y to a i r p o r t s . This i s important because of the dual inf luence of a i r p o r t s on adjacent property values of a deprec ia t ion e f fec t due to a i r c r a f t noise and an apprec ia t ion e f fec t due to employment a c c e s s i b i l i t y . A f t e r not ing the CBD a c c e s s i b i l i t y / s p a c e trade o f f and the r e f l e c t i o n of e x t e r n a l i t i e s such as environmental amenities, l i v i n g condi t ions and s o c i a l features i n land rents , the study s tates that for a i r p o r t s the e m p i r i c a l problem i s one of d i s en tang l ing the ex terna l i ty and a c c e s s i b i l i t y rents ( i f the focus i s on the s ize of the marginal noise d iscount) . The s tudy's sampling procedure attempts to c o n t r o l for the e f f ec t of a c c e s s i b i l i t y on l o c a t i o n rent (and hence housing pr ice ) while a l lowing a i r c r a f t noise v a r i a t i o n . In an e f f o r t to avoid the s p e c i f i c a t i o n b ias or m u l t i c o l l i n e a r i t y , the study uses dummy c o e f f i c i e n t s to capture the net impact of the a i r p o r t on land values , and to t race out the rent gradient . Noting the problems of using c i r c u l a r d is tance r ings (as i n De Vany 1976) such that noise contours are elongated (and thus any r i n g contains NEF l e v e l s ranging from ambient l e v e l s to peak l e v e l s ) , the study uses contiguous geographic areas . MODEL The study uses Nelson's standard hedonic p r i c e equation for housing. Each study area i s about 2 miles i n rad ius and contains NEF l e v e l s ranging from 20 to 45 i n an attempt to ensure constant a c c e s s i b i l i t y to the a i r p o r t t ermina l s , CBD and other major f o c a l po ints ; v a r i a t i o n i n l o c a l p u b l i c sector v a r i a b l e s should also be quite smal l . The study uses census block data and screens (using a e r i a l photographs) 300 to 400 contiguous blocks for each of the s i x c i t i e s , so that f i n a l samples range from 113 to 185 205 blocks and average about 140 b locks . Blocks are excluded i f near l o c a l environmental features (parks, cemeteries, g o l f courses) , major t ranspor ta t ion f a c i l i t i e s (freeways, major s t r e e t s , r a i l r o a d t r a c k s , commercial developments (shopping centres , apartment complexes), or other spec ia l neighbourhood features (sewage treatment p l a n t , cana l s , naval bases) . Blocks are a lso excluded i f adjacent to ac tua l NEF 25 or 30 contour l i n e s , or wi th in NEF 30 but adjacent to NEF 35. Th i s i s done for a number of reasons: to avoid blocks which may have sh i f t ed across contours because the 1972 NEF data were gathered l a t e r than the census data and to avoid bordering blocks causing measurement errors given the use of 5dB increments (though how exc lus ion of p a r t i c u l a r blocks avoids t h i s i s not apparent) and a des i re to draw samples conta ining a r e l a t i v e l y large proport ion of high NEF l e v e l s . As we l l as the above exc lus ions , for i n d i v i d u a l areas , f i n a l samples r e f l e c t the fo l lowing cons tra int s : blocks are excluded i f less than 50% of housing un i t s are s i n g l e -fami ly , owner-occupied, i f they contain fewer than 10 s i n g l e - f a m i l y , owner-occupied u n i t s , i f less than 80% of the r e s i d e n t i a l uni t s reported values i n the 1970 Census. Explanatory v a r i a b l e s inc lude: . P h y s i c a l : mean number of rooms per u n i t , % of owner-occupied houses with substandard plumbing, % housing u n i t s b u i l t before 1939, % housing uni t s with c e n t r a l a i r - c o n d i t i o n i n g . . Area: % of t o t a l housing un i t s that are owner-occupied, % of b lack populat ion . . Noise: NEF 20 or 25 to NEF 45 i n 5dB increments; threshold i s u s u a l l y set at 25, though where there i s l i m i t e d r e s i d e n t i a l noise- impact , at 20. V a r i a b l e s for p h y s i c a l and area c h a r a c t e r i s t i c s are c o l l e c t e d from e i t h e r block or t r a c t census data . The t r a c t v a r i a b l e s tend to be c o r r e l a t e d with the NEF v a r i a b l e given t h e i r broader geographic coverage. Regressions are run for the s i x i n d i v i d u a l samples, p lus the pooled sample (for a t o t a l of 845 observat ions) . RESULTS For the i n d i v i d u a l areas , NEF and the mean number of rooms are always s i g n i f i c a n t . The remaining v a r i a b l e s are s i g n i f i c a n t about h a l f the time. NEF c o e f f i c i e n t s range from -.29% to -.74% (-.51% weighted) for equations with R2 from .611 to .762. 206 For the pooled sample cons i s t ing of 845 observat ions , a Chow t e s t i n d i c a t e d nonhomogenous samples. The study therefore includes f i v e a i r p o r t dummy v a r i a b l e s . The r e s u l t i n g pooled equations have an R2 of about .84. S i g n i f i c a n t var iab les include NEF, mean number of rooms, % owner-occupied, % black , % a i r - c o n d i t i o n i n g , and four of the f i v e c i t y dummies. The robustness of pooled sample r e s u l t s i s examined by t e s t i n g the s t a b i l i t y of regress ion c o e f f i c i e n t s . This i s done f i r s t by us ing dummy var iables appl ied to the NEF l e v e l . The author concludes that the NEF-property r e l a t i o n s h i p may not be simple but the amount of b ias i s smal l r e l a t i v e to the range of values for i n d i v i d u a l study areas. Second, the study p a r t i t i o n s regress ions us ing l i n e a r distance from the a i r p o r t terminal . This g ives NEF c o e f f i c i e n t s of -.3% for 0 to 2 miles from t e r m i n a l , -.61% for 1 to 3 mi les , -.55% for 2 to 4 miles , -.36% for 3 to 5 mi les . These r e s u l t s are cons i s tent with s ing l e and pooled samples, although there i s some suggestion of a higher marginal noise discount at about 2 to 3 miles from the a i r p o r t . A simple measure of a i r p o r t a c c e s s i b i l i t y i s constructed and the model rerun with the pooled sample i n c l u d i n g t h i s v a r i a b l e . S i g n i f i c a n t c o r r e l a t i o n was found with NEF; l inear dis tance i s not s i g n i f i c a n t , but NEF i s s t i l l s i g n i f i c a n t with the same r e l a t i v e magnitude - suggesting r e s u l t s are r e l a t i v e l y unaffected by a c c e s s i b i l i t y in f luences . The NDSI are ca l cu la ted to be i n the range, 0.50 - 0.55% for both i n d i v i d u a l and pooled data. 207 O'BYRNE, P H, J P NELSON, and J J SENECA (1985), "Housing Values , Census Est imates , D i s e q u i l i b r i u m , and the Environmental Cost of A i r p o r t Noise: A Case Study of A t l a n t a " , Journal of Environmental Economics and Management, 12, pp 169-71. STUDY The study appl ies the hedonic p r i c e model to two sets of data for a r e s i d e n t i a l area near A t l a n t a Internat iona l A i r p o r t for 1979-1980 and 1970-72, between which dates populat ion was unchanged. I t i s noted that previous studies have t y p i c a l l y r e l i e d on owner-appraised property values , aggregated and averaged for census t r a c t s or blocks for the census years 1960 or 1970. The study seeks to address the issue of comparabi l i ty of e m p i r i c a l r e s u l t s from aggregate census data versus i n d i v i d u a l sa les values , and the issue of the homogeneity and s t a b i l i t y of r e su l t s over time and across markets. The study states that for the housing market near an a i r p o r t , homeowner expectation about growth and change can be expected to play an important r o l e i n determining asset va lues . E m p i r i c a l r e s u l t s may thus be subject to a r e s u l t i n g bias i f no formal model of homeowner expectations i s inc luded . In the ana lys i s of a s ing l e set of cross sec t ion data, the inves t iga tor assumes that d i f f e r e n t households are homogeneous, except for d i f ferences i n measurable v a r i a b l e s of the problem and a random e r r o r term. Homogeneous expectations about future environmental l e v e l s w i l l be r e f l e c t e d i n the constant term of the regress ion . A dynamic, t ime-ser ies model would be needed i f one i s in teres ted i n adjustment patterns over t ime. For cross s ec t ion data, biased estimates can be expected i f d i s e q u i l i b r i a forces are present and expectations are not homogeneous. MODELS AND RESULTS I n d i v i d u a l property sales (1979-80) Some 90% of t o t a l i n d i v i d u a l property sa les data for 1979-80 are obtained, of which 30 observations are dropped (because of incomplete data, unusual contrac tua l terms, or loca t ions c lose to major s treets and highways) g i v i n g a sample of 96. The dependent v a r i a b l e i s the i n d i v i d u a l house sa le p r i c e (exclusive of c l o s i n g costs , taxes, and other contrac tua l extras) used i n a semilog model form. Independent v a r i a b l e s modelled inc lude: 208 . P h y s i c a l : square feet of l i v i n g space, number of bathrooms, ex ter ior fac ing , presence of basement, number of rooms, number of bedrooms, presence of c e n t r a l a i r - c o n d i t i o n i n g and presence of c e n t r a l heat ing . . Noise: 1980 day-night sound l e v e l for the range 65 to 80, i n 5dB increments. . Other: time of sale ( i n f l a t i o n ) . The regress ion equation for the model has an adjusted R2 of .7077. The fol lowing var iab le s are s i g n i f i c a n t at the 95% l e v e l f or a two- ta i l ed tes t : L ^ , i n f l a t i o n , number of bathrooms, presence of basement and e x t e r i o r f a c i n g . A l i n e a r formulation of the model would y i e l d comparable r e s u l t s . Given the s t a b i l i t y of the c o e f f i c i e n t i t may be used to estimate an NDSI of about 0.67% per d e c i b e l . Census block data (1970) Census block and t r a c t data for 1970 and noise exposure forecas t (NEF) contour data for 1972 are used i n a semilog model. The study area i s considered small enough to render constant any important a c c e s s i b i l i t y and l o c a l p u b l i c sector v a r i a b l e s . The approach of Nelson (1979, 1981a) i s used to s e l e c t v a r i a b l e s and c o l l e c t data . The dependent v a r i a b l e i s the mean property value for report ing owner-occupied housing u n i t s i n each census b lock . Explanatory v a r i a b l e s i n c l u d e : . P h y s i c a l : mean number of rooms, substandard plumbing, age, c e n t r a l a i r - c o n d i t i o n i n g . . Area: % of t o t a l housing un i t s that are owner occupied, % of populat ion that i s b lack . . Noise: NEF 25 to 45 (Ljn = NEF + 35 dB(+/" 3dB) ) ; important r e s i d e n t i a l areas are those for which NEF exceeds 30. Blocks which border d i r e c t l y on a noise contour are e l iminated to reduce poss ib le measurement errors caused by s h i f t s i n the noise contours between the 1970 census date and the 1972 NEF date. In order to contro l for some of the environmental , neighbourhood and a c c e s s i b i l i t y c h a r a c t e r i s t i c s not represented i n the model, the study a lso screens out blocks adjacent to outdoor recrea t ion areas , major t ranspor ta t ion f a c i l i t i e s , and other s p e c i a l neighbourhood features . Two samples are used: the r e s t r i c t e d sample which excludes blocks with l e s s than 50% owner-occupied housing un i t s (248 b l o c k s ) , and the f u l l sample which excludes only blocks with l e ss than 25% owner-occupied un i t s (258 b l o c k s ) . 209 Resul ts are general ly consistent with those reported i n Nelson (1981a). For the r e s t r i c t e d sample, the adjusted R2 i s .7409. The NEF c o e f f i c i e n t for the r e s t r i c t e d sample of the more heav i ly r e s i d e n t i a l blocks impl ies a NDSI of 0.64% per d e c i b e l . The use of a l i n e a r model y i e l d s comparable r e s u l t s for the noise-discount est imates . CONCLUSIONS The study ind icates that noise cost estimates obtained using 1970 census block data are very s i m i l a r to those obtained us ing 1979-80 i n d i v i d u a l sales p r i c e data . Assuming independent random sampling and normal populat ions with equal var iance , a small sample t e s t of s i g n i f i c a n c e can be undertaken. At the 95% confidence l e v e l , the d i f ference between the noise discounts for the i n d i v i d u a l sales data and the r e s t r i c t e d census data sample i s not s t a t i s t i c a l l y s i g n i f i c a n t , but i s for the i n d i v i d u a l sa les data and the f u l l census sample data . The study s tates that the most appropriate comparison i s the one us ing the r e s t r i c t e d census sample, given d i f f i c u l t i e s of c o n t r o l l i n g for a l l p o s s i b l e fac tors that inf luence property values i n non-r e s i d e n t i a l areas (that i s , those areas excluded from the f u l l sample). The r e s u l t s are i n general agreement with e a r l i e r a i r p o r t no i se /proper ty value studies suggesting that NDSI f igures are a p p l i c a b l e across markets and across t ime. T h i s , i n t u r n , suggests that the hedonic p r i c e model w i l l be stable when app l i ed to a i r p o r t noise r e l a t i o n s h i p s i n the 1980s. 210 PAIK, I K (1972), "Measurement of Environmental Externality in Particular Reference to Noise", unpublished Ph.D. dissertation, Georgetown University, SUMMARIZED IN NELSON (1978a, 1980). (See a l so PAIK, I (1970), Impact of  Transportat ion Noise on Urban R e s i d e n t i a l Property Values  with Spec ia l Reference to A i r c r a f t Noise. Urban Transportat ion Center, Consortium of U n i v e r s i t i e s , Washington, DC.) STUDY The study models the e f fec t s of a i r c r a f t noise on r e s i d e n t i a l property values i n New York, Los Angeles and Dal las i n 1960. MODEL The model has a log funct iona l form. I t uses census block data to obtain samples r e s t r i c t e d to blocks where at l eas t 50 percent of residences were s i n g l e - f a m i l y and owner-occupied, g i v i n g samples of 106, 92 and 94 r e s p e c t i v e l y for New York, Los Angeles and D a l l a s . A pooled sample i s a l so modelled. The dependent v a r i a b l e i s the median owner-assessed property value . Explanatory v a r i a b l e s inc lude: . P h y s i c a l : median number of rooms per house. . Area: mean number of people per household, absolute number of s i n g l e - f a m i l y houses, absolute number of nonwhite houses, % of de ter iora ted houses. . Noise: NEF 20, 35, 40, with a threshold noise l e v e l of NEF 20. Because NEF contours are for 1965, f i ve years l a t e r than the census data, blocks located near the contour boundaries are excluded from the sample. RESULTS Only the v a r i a b l e s for the number of rooms and noise are s i g n i f i c a n t . Regressions with only these two explanatory v a r i a b l e s y i e l d noise e l a s t i c i t i e s of -0.796 for the pooled sample and from -0.576 to -0.763 for the three c i t i e s . Nelson (1978a, p.103) c a l c u l a t e s an NDSI of 2.09%. Nelson (1978a, p.103) reports problems with the study, being susp ic ious of the s p e c i f i c a t i o n of severa l v a r i a b l e s , i n p a r t i c u l a r those using the absolute number of houses. He i s a l so concerned about pool ing the three areas without c o n t r o l l i n g for interurban d i f f e r e n c e s . Nelson uses the s tudy's data to obtain another set of log models, regress ing median property value on median number of rooms, a i r c r a f t no i se , % of s i n g l e - f a m i l y houses, mean number of people per 211 room, % of de ter iorated houses, % of nonwhite houses i n each census b lock , noise , and c i t y dummy v a r i a b l e s for the pooled sample. S i g n i f i c a n t var iab le s found by Nelson at the 95% confidence l e v e l (two-tailed) included noise , rooms, % s i n g l e - f a m i l y , mean people per room, Los Angeles dummy, and, at 90% s i g n i f i c a n c e , % deter iorated . Nelson suggests that r e l a t i v e l y few housing var iab les are s i g n i f i c a n t not because of m u l t i c o l l i n e a r i t y but because of the r e l a t i v e homogeneity of the study areas. The noise c o e f f i c i e n t i s qui te s table across the three areas, and suggests a 2.2% NDSI for the pooled sample (Nelson 1980, p .50) . Nelson (1978a, p.105) reasons that t h i s high NDSI may be due to the fact that 1960 was the beginning of commercial j e t t r a v e l i n the United States . 212 PRICE, I (1974), "The S o c i a l Cost of A i r p o r t Noise as Measured by Rental Changes: The Case of Logan A i r p o r t " , unpublished Ph.D. d i s s e r t a t i o n , Boston U n i v e r s i t y , SUMMARIZED IN NELSON (1978a, 1980). STUDY The study examines the percentage change from 1960 to 1970 i n the median contract rent for Boston as af fected by a i r p o r t noise . MODEL A l i n e a r model i s used, with data from 270 census t r a c t s being used for the dependent v a r i a b l e which i s % change i n median contract rent between 1960 and 1960. Explanatory v a r i a b l e s inc lude: . Area: change i n the % of residences occupied by nonwhites over the decade, % of nonwhites i n 1960, % of people over 65 years i n 1960, % of housing un i t s b u i l t p r i o r to 1939, % of housing un i t s b u i l t s ince 1960, % of r e n t a l housing un i t s that are p u b l i c housing. . Pub l i c sector: % increase i n property tax r a t e . . A c c e s s i b i l i t y : distance to CBD. . Noise: NEF-25 for 1970, with no apparent adjustment for background noise . . Other: median contract rent i n 1960. RESULTS Nelson (1978a, p.107) uses the r e s u l t s to obtain an NDSI of 0.60% for Boston. 213 APPENDIX C - CORRELATION MATRICES Table C . l i s a c o r r e l a t i o n matrix of a l l the v a r i a b l e s used i n the f i n a l IN models (models 1, 2, 3, 4, 5, 6 ) . Table C.2 i s a c o r r e l a t i o n matrix of a l l the v a r i a b l e s used i n the f i n a l ALL models (models 7, 8, 9, 10, 11). 214 TABLE C . l - CORRELATION MATRIX OF ALL VARIABLES USED IN FINAL IN MODELS (MODELS 1, 2, 3, 4, 5, 6) LPRICE LNWIDTH SMALLLOT ISTORIES IFIREPL I POOL LSQFEET LNOROOMS LTBATHJ LPRICE 1.0000 .0404 -.1622 .4240 .4015 . 1347 .3849 .5438 .5449 LNWIDTH 1.0000 -.4006 -.3416 -.0486 -.0039 .2183 -.3357 -.2870 SMALLLOT 1.0000 .0684 . 1707 -.0034 -.0193 .0707 .1450 ISTORIES 1.0000 .1117 .0529 -.0597 .6124 .4070 IFIREPL 1.0000 .0644 .4299 .3319 .4644 I POOL 1.0000 .0377 .0811 .0467 LSQFEET 1.0000 .0541 .3957 LNOROOMS 1.0000 .6450 LTBATHS 1.0000 LAGE NONMOVRS A15 LDISTSCH QI Q2 Q3 NEFLEVEL INZONE25 INZONE30 INZONE35 INZONE40 NZONE TABLE C . l (cont 'd) - CORRELATION MATRIX OF ALL VARIABLES USED IN FINAL IN MODELS (MODELS 1/ 2, 3/ 4, 5, 6) LAGE NONMOVRS A15 LDISTSCH QI Q2 Q3 NEFLEVEL INZONE2 LPRICE -.5876 -.1542 .0946 -.1215 -.1139 -.0235 .0862 -.2804 .1460 LNWIDTH .4196 .2682 . 1794 .0743 .0091 -.0363 .0068 -.0983 .2529 SMALLLOT -.0994 -.0711 .0282 -.1829 .0137 .0078 -.0197 .0141 -.0989 ISTORIES -.4974 -.2308 .1552 -.0612 .0482 .0288 -.0570 -.0226 -.1086 IFIREPL -.3884 -.0157 .1451 -.0773 -.0270 .0248 .0242 -.1865 .0667 I POOL -.0561 .0131 .0525 .0191 -.0188 -.0349 .0421 -.0320 .0559 LSQFEET -.1969 .1421 .0249 .0197 .0222 -.0804 .0204 -.1547 .1025 LNOROOMS -.7338 -.2070 .3114 -.0751 .0504 -.0011 -.0010 -.1490 -.0118 LTBATHS -.6632 -.0872 .2685 -.0009 .0346 -.0343 .0162 -.2026 -.0062 LAGE 1.0000 .2395 .2639 .0711 -.0429 .0049 -.0164 . 1448 .0590 NONMOVRS 1.0000 . 1719 .1344 .0291 -.0009 -.0324 -.0106 .0794 A15 1 .0000 .0221 .0070 .0295 -.0376 -.0145 .0738 LDISTSCH 1.0000 -.0633 -.0034 .0278 .0964 -.1771 QI 1.0000 -.3806 -.3034 .0144 -.0092 Q2 1.0000 -.4242 -.0116 .0035 Q3 1.0000 .0090 -.0342 NEFLEVEL 1.0000 -.7482 INZONE25 1.0000 INZONE30 INZONE35 INZONE40 NZONE TABLE C . l (cont 'd) - CORRELATION MATRIX OF ALL VARIABLES USED IN FINAL IN MODELS (MODELS 1, 2, 3, 4, 5, 6) INZONE30 INZONE35 INZONE40 NZONE LPRICE .0418 .0026 - .3117 -.2722 LNWIDTH - .3191 .0827 .0205 -.0931 SMALLLOT .1239 - .0064 - .0319 .0294 ISTORIES .1933 - .0466 - .0896 -.0144 IFIREPL .0673 . 0468 - .2655 -.1880 I POOL - .0489 .0236 - .0359 -.0445 LSQFEET - .0306 .0967 - .2138 -.1587 LNOROOMS .1642 - .0638 - .1851 -.1304 LTBATHS .1951 .0062 - .3127 -.1923 LAGE - .2328 .0470 .2354 . 1323 NONMOVRS - .1314 .0912 - .0067 -.0175 A15 - . 1488 .0658 .0558 .0171 LDISTSCH .1244 .2640 - .1682 .0678 QI - .0005 .0123 .0039 .0110 Q2 - .0075 - .0021 .0084 .0028 Q3 .0537 - .0173 - .0140 .0037 NEFLEVEL .0992 .3709 .7114 .9540 INZONE25 - .6749 - .2760 - .2834 -.7858 INZONE30 1 .0000 - .2554 - .2623 . 1074 INZONE35 1 .0000 - .1073 .3852 INZONE40 1 .0000 .7461 NZONE 1.0000 TABLE C.2 - CORRELATION MATRIX OP ALL VARIABLES USED IN FINAL ALL MODELS (MODELS 7, 8, 9, 10, 11) LPRICE LNWIDTH SMALLLOT ISTORIES IFIREPL I POOL LSQFEET LNOROOMS LTBATHS LPRICE 1.0000 -.1215 -.0502 .4677 .3736 .0847 .3138 .6064 .6033 LNWIDTH 1.0000 -.3347 -.4619 -.0029 .0813 .2845 -.4318 -.3493 SMALLLOT 1.0000 .0840 .2073 -.0199 -.0278 .0872 . 1170 ISTORIES 1.0000 .0832 -.0040 -.1220 .6444 .4569 IFIREPL 1.0000 .0457 .3691 .2834 .3633 I POOL 1.0000 .0794 .0273 .0478 LSQFEET 1.0000 -.0293 .3142 LNOROOMS 1.0000 .6395 LTBATHS 1.0000 LAGE NONMOVRS A15 LDISTSCH QI Q2 Q3 NEFLEV24 NEFLEV20 INZONE24 IZNONE25 INZONE30 INZONE35 INZONE40 NZONE24 NZONE20 TABLE C.2 (cont'd) - CORRELATION MATRIX OF ALL VARIABLES USED IN FINAL ALL MODELS (MODELS 7, 8, 9, 10, 11) LAGE NONMOVRS A15 LDISTSCH LPRICE - .6391 -.1838 .0355 - .1400 LNWIDTH .5468 .2820 -.1857 .0921 SMALLLOT - .1289 -.0099 .0528 - .2672 ISTORIES - .6025 -.2732 .1589 - .0970 IFIREPL - .2948 .0130 .1010 .1509 I POOL .0270 .0325 -.0021 - .0020 LSQFEET - .0713 .1203 -.0044 - .0261 LNOROOMS - .7491 -.2204 .1815 - . 1242 LTBATHS - .6510 -.1497 .1895 - .0376 LAGE 1 .0000 .2619 -.1594 .0915 NONMOVRS 1.0000 -.1924 .0267 A15 1.0000 .0175 LDISTSCH 1 .0000 QI Q2 Q3 NEFLEV24 NEFLEV20 INZONE24 IZNONE25 INZONE30 INZONE35 INZONE40 NZONE24 NZONE20 QI Q2 Q3 NEFLEV24 NEFLEV20 -.1091 -.0134 .0806 -.1906 -.1677 .0312 -.0513 -.0065 .0612 .0787 .0198 -.0153 -.0026 -.1338 -.1544 .0268 .0088 -.0058 -.0725 -.0794 -.0172 -.0010 .0343 -.1959 -.1822 .0014 -.0173 .0403 -.0071 -.0032 .0080 -.0351 .0032 -.0596 -.0470 .0113 -.0122 .0427 -.1685 -.1673 -.0036 .0121 .0234 -.1593 -.1485 .0021 -.0116 -.0230 .1855 .1913 .0204 -.0112 .0000 -.1583 -.2034 .0043 .0009 -.0277 -.1023 -.1076 -.0258 .0172 -.0047 . 1338 .1273 1.0000 -.3619 -.3119 .0147 .0120 1.0000 -.4284 .0161 .0187 1.0000 -.0210 -.0278 1.0000 .9738 1.0000 TABLE C.2 (cont 'd) - CORRELATION MATRIX OF ALL VARIABLES USED IN FINAL ALL MODELS (MODELS 7, 8, 9, 10, 11) INZONE2 LPRICE .0873 LNWIDTH - .1349 SMALLLOT .1856 ISTORIES .0830 IFIREPL .1375 I POOL -, .0088 LSQFEET - , .0096 LNOROOMS .1387 LTBATHS .0951 LAGE -, .1706 NONMOVRS .2013 A15 .1249 LDISTSCH -, .1215 QI -, .0112 Q2 -, .0275 Q3 .0324 NEFLEV24 -, .7625 NEFLEV20 -, .8732 INZONE24 1, .0000 IZNONE25 INZONE30 INZONE35 INZONE40 NZONE24 NZONE20 INZONE25 INZONE3 .0360 - .0206 .2123 - .1092 - .1710 - .0064 - .1026 .0615 - .0328 - .0275 .0359 - .0233 .0627 - .0126 - .0821 .0201 - .0552 .0620 .1223 - .0298 - .0629 .1830 - .0218 - .1600 - .0318 .1331 .0011 .0055 .0170 .0100 - .0359 .0124 .0103 .4485 .1778 .4910 - .5455 - .5160 1 .0000 - .2031 1 .0000 INZONE35 INZONE4 - .0192 - .2280 .0849 .0462 - .0498 - .0717 - .0484 - .0751 .0014 - .2280 .0173 - .0209 .0650 - .1362 - .0734 - .1507 - .0187 - .2235 .0678 .1767 .0142 - .0540 .0180 .0099 .1984 - .0779 .0105 .0052 .0053 .0121 - .0184 - .0166 .4153 .6332 .3803 .5424 - .2392 - .2452 - .0941 - .0965 - .0890 .0913 1 .0000 - .0423 1 .0000 NZONE24 NZONE20 - .1756 -.1576 .0787 .0966 - .1409 -.1578 - .0742 -.0791 - . 1931 -.1845 - .0090 -.0045 - .0507 -.0360 - .1625 -.1616 - .1497 -.1401 .1854 .1877 - .1745 -.1877 - .0970 -.1079 .1260 .1290 .0133 .0132 .0239 .0257 - .0257 -.0284 .9766 .9517 .6787 .9833 - .8331 -.9067 .0902 .2169 .4808 .5067 .4062 .3746 .6044 .5274 1 .0000 .9887 1.0000 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0228306/manifest

Comment

Related Items