Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Fluctuations in the turnover of single-family dwellings in metropolitan Vancouver Seek, Ngee-Huat 1975

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


831-UBC_1975_A4_6 S04.pdf [ 5.69MB ]
JSON: 831-1.0099982.json
JSON-LD: 831-1.0099982-ld.json
RDF/XML (Pretty): 831-1.0099982-rdf.xml
RDF/JSON: 831-1.0099982-rdf.json
Turtle: 831-1.0099982-turtle.txt
N-Triples: 831-1.0099982-rdf-ntriples.txt
Original Record: 831-1.0099982-source.json
Full Text

Full Text

FLUCTUATIONS IN THE TURNOVER OF SINGLE-FAMILY DWELLINGS IN METROPOLITAN VANCOUVER by NGEE-HUAT SEEK B.Sc. (Estate Management), University of Singapore, 1973 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN BUSINESS ADMINISTRATION in the Faculty of Commerce and Business Administration We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA May, 1975 In p r e s e n t i n g t h i s t h e s i s in p a r t i a l f u l f i l m e n t o f the r e q u i r e m e n t s f o r an advanced degree at the U n i v e r s i t y o f B r i t i s h C o l u m b i a , I a g r e e that the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and s t u d y . I f u r t h e r a g r e e t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y p u r p o s e s may be g r a n t e d by the Head o f my Department o r by h i s r e p r e s e n t a t i v e s . It i s u n d e r s t o o d that c o p y i n g o r p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n . Department o f Urban Land Economics, Faculty of Commerce and Business Administration. The U n i v e r s i t y o f B r i t i s h Co lumbia 2075 Wesbrook P l a c e Vancouver, Canada V6T 1W5 Date May 20, 1975 ABSTRACT The turnover rate for housing measures the average length of tenure between transactions, and, thus, indicates the level of activity,in the housing market. However, turnover of housing has hitherto "been neglected by urban economists. This thesis •* attempts to study the fluctuations in the turnover in ownership of single-family houses in Metropolitan Vancou-verover the period 19^ 9-1963« It is hoped that such a study will serve to close a gap in housing research by presenting relevant theory and data on this aspect of turnover. Such turnover information will be useful to the policy makers as well as the homebuyers,the mortgage lenders and the real estate profession. The i n i t i a l stage of this study involved the collection of relevant raw data and the computation of turnover rates. The results of this research activity are presented in the f i r s t part of the thesis. The second part is concerned with the formulation of a theoretical base on which a model to explain fluctuations in turnover is developed. In the £Msd section of this thesis, this model is tested quantitatively using the technique of multiple regression on time-series data for both over-a l l (new and existing houses) turnover and turnover of existing houses only. In the final section, the variables used in the regression analy-sis are re-examined qualitatively with the aid of graphical illustrations. This section concludes with a discussion of seasonal variation of turn-over. The study shows that, during the period 19^ 9-196y$ turnover rates fluctuated cyclically, and that there was a definite downward trend after i i 1958. Both quantitative and qualitative analyses indicate' that this phenomenon i s primarily the result of a combination of the following factors: changes i n credit conditions, business cycles, change in the number of family heads, the relative cost of owning versus renting a home, and the rate at which new houses were completed. A further finding of this thesis i s that the volume of sales of single-family houses was subject to significant and consistent seasonal fluctuation. i i i CONTENTS ABSTRACT LIST OF TABLES LIST OF ILLUSTRATIONS ACKNOWLEDGEMENT CHAPTER I INTRODUCTION 1 Purpose and Scope of the Study 1 The Data Basei Technique and Procedure 2 Definition of Turnover Rate ,3 The Significance of Turnover Rate 9 CHAPTER II THEORETICAL CONSIDERATIONS« POTENTIAL DETERMINANTS 14 OF TURNOVER Demographic Influence 20 Income Effects 23 Credit Considerations 28 The Economics of Renting vs Owning 36 Taste and Preference 38 CHAPTER III THE REGRESSION ANALYSIS 42 Turnover of Existing and New Houses 42 Turnover Rates of Existing Houses 52 CHAPTER IV QUALITATIVE ANALYSIS OF THE EMPIRICAL DATA 58 Seasonal Fluctuations in Turnover 58 Turnover, Credit Conditions, Business Cycles and 66 Monetary Policy Turnover and Other Explanatory Variables 73 CHAPTER V SUMMARY AND CONCLUSIONS ?6 BIBLIOGRAPHY 78 APPENDIX 82 iv : . v vi v i i LIST OF TABLES 1 . 1 Turnover Rates, and Their Sampling Errors by Area, 1949 4 and 1963 1 . 2 Municipalities by Sub-areas 5 1 . 3 Turnover Rates for Metropolitan Vancouver and Its Sub-areas, 8 1949-63 2 . 1 Average Age and Age Defferentials 6£ First Marriage for 2 2 Bridegrooms and Brides in Canada 2 . 2 Distribution by Age of Borrowers 2 2 2 . 3 Average Dwelling Costs and Applicants' Income 2 5 2 . 4 Average Ratios of Gross Debt Service to Income and 2 6 Applicants1* Income 2 . 5 Average Down-Payments and Applicants' Income 2 ? 2 . 6 Monthly Payment Per $ 1 0 , 0 0 0 of Mortgage for a 2 0 fear 2 9 Term At Increasing Interest Rate® 2 . 7 Relationship Between Minimum Income, Mortgage Term and 3 0 Monthly Payment At 8% Interest 3 . 1 Correlation Matrix of T • • 4 7 3 . 2 Step-wise Regression of T 5 0 o 3.3 Correlation Matrix of T 55 e J-> 3 A Step-wise Regression of T 5 6 4 . 1 Sample Distribution of Sales by Age 6 3 4 . 2 Sample Distribution of Sales by Value, Overall Area 64 4 . 3 Business Cycle Reference Dates, I 9 4 9 - I 9 6 3 6 9 v LIST OF, ILLUSTRATIONS Figure 2.1 Relationship Between Sales and Price 17 Figure 2.2 Monthly Payment Per $1,000 of Mortgage by Maturity 32 at Various Interest Rate Figure 2.3 Competition for Capatal Funds Chart I A Conceptual Framework Representing the Process of 15 Trading Chart II Distribution of Aggregate Monthly Sales in Number 60 and Value Chart III Distribution of Aggregate Monthly Sales by Age 63 Chart IV Comparison of the Explanatory Variables 67 Chart V Monthly Supply and Interest Rate 68 v i ACKNOWLEDGEMENT I am indebted to my thesis advisors, Professors S.W. Hamilton and David Baxter who have devoted much time and effort in supervising this thesis and for their constructive criticism and encouragement. I am also grateful to Mr. A. Eger for his assistance in computer programming and Mr. and Mrs. C.K. Gan for graciously offering to type this thesis and for an excellent job done. I would also like to take this opportunity to thank Professors S.W. Hamilton, M.A. Goldberg, F. Pennance, and D.Baxter of the Division of Urban Land Economics, Faculty of Commerce and Business Administration for imparting much of their knowledge and for making my two years as a student in the University of British Columbia a fruitful and meaning-ful experience. CHAPTER I INTRODUCTION Purpose and Scope of the Study The purpose of this study i s to measure and explain the fluctuations in the turnover rate of single-family dwellings in Metropolitan Vancouver . over the period 1949-1963. There have been, especially i n the la s t two decades, numerous studies undertaken to explain the fluctuations in housing starts or construction expenditures and mortgage financing, notably byt Grebler, Maisel, Muth, Alberts, Guttentag, Klaman, O'Leary, W. Smith, L. B. Smith, Sparks, Huang, Winger, and Meltzer.* None of these studies, however, have examined the fluctuations in the number of transactions (turnover) in the housing market. This i s probably due to the inavailability of turnover data which are extremely d i f f i c u l t and expensive to gather. It w i l l be shown i n this study that turnover as a dimension of the housing market i s related not only to the cost and av a i l a b i l i t y of mortgage credit but also to the demand for housing. A study of fluctuations in housing expenditure and mortgage financing i s incomplete without including turnover rates. It i s hoped that that this analysis w i l l serve to close this gap in housing studies by presenting relevant theory and data on turnover on a regional basis. I t i s also hoped that such a study w i l l not only be beneficial to the policy makers but also to the home-buyers, the mortgage lenders and the realtors. The time coverage of the study i s solely governed by the avail a b i l i t y of data and financial constraints. The turnover data for I949-I963 i s - 2 -part of the data from a sample housing survey completed in I967 by ^he Faculty of" Commerce and Business Adminstration, University of British Columbia. It will cost $85,000 to carry out a similar survey to up-date the data, and i t is felt that the additional information may not justify the expense. Even though the turnover rates subsequent to may be different, i t is felt that the explanations for the fluctuations in turn-over for the earlier years would s t i l l apply. This is because, generally, there has been no structural changes in the general economic framework. Essentially, this study is conducted in four parts. The f i r s t involves the collection of relevant raw data and the computation of turnover rates. The second part is concerned with the identification of the potential deter-minants of turnover and the formulation of a theoretical base from them. The third part is devoted to quantitative analysis using the technique of multiple regression. In the final part, the seasonal variation of turnover will be studied, and in addition, a l l the variables used in the regression analysis are re-examined qualitatively with the aid of graphical illustra-tions. The Data Base - Technique and Procedure The data cover 7t000 transactions of single-family houses which were situated in the 14 cities and municipalities within Metropolitan Vancouver. These were collected by a random sample of assessment rolls in these munici-palities. The important questions in sampling from this universe were the optimal size of the sample and appropriate stratification of the universe. The sample size was verify on the basis that the sample turnover rate has a 90$ chance of falling within the true turnover rate as shown in Table - 3 -4i$,/As regards stratification of the universe, the objective was to iden-tify distinguishable housing markets with similar turnover rates. It Was found that municipalities with the same rate of population growth tended to have similar rates of new construction and were likely to be similar in such characteristics as age and style of housing and location desirability. The final decision was to stratify the municipalities into 3 sub-areas (see Table .^),mainly on the basis of I95I-6I rates of population growth. The next step was the actual sampling process itself. An estimate of the total number of single-family houses in each municiaplity was made on the basis of data from the various assessment offices. This informa-tion was used to determine the appropriate sample size in each of the three sub-areas. The number of sample units in each subr-area was divided among the municipalities on a proportionate basis. Within each municipality the sampling was random, meaning that every single-family house in the municipality had the same change of being selected. . Because of errors in searching and in the assessment rolls, the number of properties in the final sample was reduced slightly below what was regarded as theoretically desirable. The actual numbers are 1936, 1931 and 1947 for the respective sub-areas. Some houses in the I963 sample were not in existence in earlier years. As the years recede, the number of houses in the sample declines, so that for 1949 the sampling errors are marginally greater, as may also be seen in Table 2. Definition of Turnover Rate For the purpose of this study, the rate of turnover is defined as the ratio between the total number of single-family houses (new and existing) _ k -Table 1.1 TURNOVER RATES, AND THEIR SAMPLING ERRORS BY AREAS, 1949 AND I963 Area Turnover Rate (T h) Err^ t J ? ) 1 90% Confidence Limits* Aggregate 1949 0.135 O.OO65 0.135 ± .0*07 " T h + 0.08T, — h 1963 0.081 0.0038 0.081 + .0063 * T h + 0/08T. — h Subarea I 1949 0.121 0.0086 0.121 + .0142 = Th + 0.12T. — h 1963 0.077 0.0060 0.077 + .0099 = Th ± 0 . 1 3 T h Subarea II I924.9 0.189 0.0131 0.189 + .0216 = Th + CUT. — h 1963 0.079 0.0060 0.079 + .0099 = Th + o.l3T h Subarea III 1949 0.136 0.0131 0.136 + .0216 ~ T h + 0.16T. - h 1963 0.091 0.0064 0.091 + .0106 = Th + o.l3T h * 90% confidence limits = Turnover rate + I .65 <r . - h Calculated in accordance with the formula for cr given in footnote 2 h - 5 -TABLE 1.2 Municipalities by Subareas Population Growth Rate 1951-61  Subarea I (Slow growth) 0 1;. Vancouver 0.U5 2. New Westminster 0.166 3. North Vancouver (city) 0.508 4. University Endowment Subarea II (Moderate growth) 5. Burnaby O.7I5 6. West Vancouver 0.819 7. Coquitlam (municipality) 0.819 Subarea III (Rapid growth) 8. Richmond 1.258 9. North Vancouver (municipality) I.695 10. Port Goquitlam 1.509 11. Port Moody 1.132 12. Surrey 1.383 13. White Rock 0.975 14. Delta 1.178 - 6 -sold during a given period of time to the total stock of housing at the end of that period. This definition can be expressed as follows: m v e Ti't T =» 0 where, T Q = Turnover rate of new and existing houses Q = Number of new houses sold during period t ^ t Q = Number of existing houses sold during period t e t = Total stock of housing at the end of period t. t It should be noted that is equalled to the total stock of housing at the beginning of the period (S^ ^) plus the number of completions in. : the same period, and refers to the number of sales and not the number of completions. For example, i f the total number of housing stock at the beginning of the year was 1,000 and 60 of them were re-sold during the year, and i f during the same year, 50 new houses were completed, out of which 40 were sold, the turnover rate is calculated as follows^ • 60 + HQ T » - • - - • m > Q 9 5 0 1,000 + 50 The rate of turnover measures the average number of years a house is held before i t changes hand and also indicates the level of market activity at a point in time. To illustrate the former, a turnover rate of .095 a t the end of the year i n d i c a t e s t h a t on the average a house changes hand every 10.5 years. I f the r a t e r i s e s from 0.095 t o 0.15, the average ownership p e r i o d f a l l s from 10.5 years t o 6.67 years. The a c t u a l t u r n o v e r r a t e s f o r the subareas and the whole o f M e t r o p o l i t a n Vancouver are shown i n Table 3. The o v e r a l l turnover r a t e i s a weighted average of the sub-areas (see Table 3). According t o the above d e f i n i t i o n , turnover i s made o f two componentsj hew and e x i s t i n g houses. The important p o i n t t o note here i s t h a t s i n c e new houses are i n c l u d e d i n the computation o f turnover r a t e s , the numner of completions and unsold houses become important market i n d i c a t o r s . As there i s always a demand f o r housing, we can expect most new houses t o be s o l d w i t h i n a s h o r t p e r i o d o f time. A r e l a t i v e l y l a r g e p o r t i o n of the turnovers i s t h e r e f o r e accounted f o r by the s a l e o f new houses. This l e a d s us t o co n s i d e r an a l t e r n a t e d e f i n i t i o n o f turnover r a t e / The r a t e o f turnover can a l t e r n a t e l y be d e f i n e d as the r a t i o o f the number of e x i s t i n g s i n g l e - f a m i l y houses s o l d d u r i n g a given p e r i o d o f time t o the t o t a l housing s t o c k a t the beginning of the same p e r i o d , and t h i s d e f i n i t i o n i s expressed as f o l l o w s j + e s t - i where, S T g - 'Turnover r a t e o f e x i s t i n g houses Q = Number of e x i s t i n g houses s o l d d u r i n g p e r i o d t - v ' t_ % - T o t a l housing stock a t the beginning o f p e r i o d t TABLE 1.3 TURNOVER RATES FOR METROPOLITAN VANCOUVER AND ITS SUBAREAS, 1949-1963 4f-T "for Subareas Overall I' YEAR : ^ — — — _ 1 2 3 T T , , o e 1949 .0.121 •; 0.189 0.136"' J 0.135 0.101 1950 0.135 0.186 0.118 0.141 0.108 1951 0.134 O.I63 0.132 1.139 0.122 1952 0.133 0.159 0.135 0.123 0.105 1953 0.116 O.I54 0.145 0.128 0.102 1954 0.097 O.I83 0.137 0.121 0.087 1955 0.094 0.154 0.158 0.119 0.082 1956 0.111 O.I58 0.189 0.137 0.104 1957 0.097 0.124 O.I56 0.115 0.090 1958 0.097 0.125 0.204 0.129 0.086 1959 0.081 0.130 0.176 0.116 0.077 i960 0.066 0.100 0.108 0.Q84 0.067 1961 0.070 0.086 0.083 0.077 0.061 1962 0.086 0.092 0.087 0.088 0.070 1963 0.077 0.079 0.091 0.081. 0.081 O . U 5 * The formulae for the computation of aggregate turnover rates are given in footnote 3 - 9 -It should he noted that S^ ^ does not include the number of new houses sold during the period. For example, i f the total housing stock at the beginning of the year was 1,000 and 100 of them were re-sold during the year, the turnover rate therefore equalled 0 .1 . In this example, the turnover rate of 0.1 indicates that existing houses were on the average, changing t i t l e every 10 years. The results of the computation of turn-over rate according to this definition are also given in Table 1 . 3 . Under this definition the number of completions in the period will not affect the rate of turnover and will probably reduce certain amount of bias in the regression analysis to be performed in Chapter 3," As shown in Table 1.3» the fluctuations in the turnover rate of existing houses have been closely related to those in the overall turnover rates ( T q ) . For. instance, in 1951. when T q reached to a peak of 0 .139, T also reached a record value of 0.122. At this rate of turnover, exist-e ing houses were changing t i t l e once every 8 . 2 years compared to 7.19 years when both new and existing houses were included in the computation. Simi-larly, when T Q was at an a l l time low (0.077)t T g also reached to its lowest point (0.061), indicating that the average ownership period had increased to 12.98 years and 16.39 years respectively. It can be seen that besides having similar pattern of fluctuation, T was consistently lower than T q. This phenomenon is to be expected as the proportion of sales of existing houses exceeded the new ones by a significant margin. /y* The Significance of Turnover Rate As mentioned earlier turnover rate not only indicates the level of - 10 -market activity but also measures the average number of years a house is held before i t changes hand. A knowledge of the level of market activity is essential to the successful implementation of housing policy. For instance during a real estate boom, the same houses often change hands several times at progressively high prices. The reverse happens during a slump when the number of transactions falls drastically. An abnormally high turnover rate may indicate an a r t i f i c i a l market based on entirely speculative prices. The Florida land boom of the 1920*s is a classic example. " This began as many came to believe that the pleasant climate of Florida would encourage great increases in \ population, as living standard rose. As the boom took off, the basic reasons for property values were lost sight of by the market, which saw only that prices were climbing ever more rapidly. Finally, the bubble burst in 1926 (appropriately enough) when a hurrican devas-tated the State and pricked the euphoria of market h, confidence. " Such a highly speculative market is more likely to happen when the trad-ing involves land. However certain amount of speculation in houses cannot be entirely ruled out, although i t may not be as serious as in the case of land. Nevertheless, an unusually high turnover rate does indicate brisk trading activity in the housing sector and, especially during an inflationary period, may justify investigation into the sources of such activity to determine whether i t reflects normal market adjust-ments or is due to unrealistic expectations or oligopolistis conditions. A knowledge of the average ownership tenure would help mortgage - 11 -lenders to set up their investment portfolios and lending c r i t e r i a . For example, i f the lender i s committed to a 20 year mortgage but the turnover rate indicates an average ownership tenure of only 10 years, there i s a high probability that the mortgage w i l l be prepaid. However, since the introduction of the five year term mortgage, this advantage i s reduced. Anyway, i f the loan i s actually paid off, the l i q u i d i t y position of the lender w i l l be enhanced and this may influence him to relax the borrowing terms and/or re-adjust his investment portfolio to commit more funds in longer term investments or loans. Also, a rise in the rate of turnover may indicate a need for more, funds to finance the increase i n the number of sales. In addition to the mortgage lenders, the home buyer i s also concerned with liquidity, that i s whether he can s e l l his house readily, A rise or f a l l in the turnover rate indicates relative marketability of an average house and this may influence the decision of .the potential purchaser. The realtors should also be interested in the turnover rages as the suc-cess of their business depend on the volume of sales. Changes in the turnover rate signal changes in the level of market act i v i t y . By gaining information on this and the various factors responsible for such changes, a realtor may adapt his marketing techniques to a different set of market conditions. Footnotes; 1. Grebler (7) (8), Maisel (8) (15), Muth (20), Alberts (1), Guttentag (10), Klaman (13), W.Smith (32), L.B.Smith (29$ (30) (31) (32), Sparks (34), Huang (14), Winger (37) and Meltzer (21). - 12 -2. The formula for the subarea i s N h - n : T. (1 - T, ) h J N h n h " 1 and for the overall area i s /N. where N = total number of units i n universe N, =• total number of units i n sub-area h h n h = total number of units i n sample for sub-area h (r. = standard error of T, h h T, = turnover of subarea h h 3. The formula for overall turnoverall turnover rate ( T q ) of existing and new houses i s ^ V i and for turnover rate (T g) of existing house i s d w i V t - i where w^  = the ratio of population size to sample size in subarea i in 1963 = the number of sales in the sample in subarea i in a given period = the number of houses in the sample in subarea i in a given period - 13 -t = end of the period t - 1 = beginning of the period For an explanation of this s t a t i s t i c a l bias, see p.5^7- in Chapter III. Thorneroft (46), p. 246. - 14 CHAPTER II THEORETICAL CONSIDERATIONS - POTENTIAL DETERMINANTS OF TURNOVER This chapter is concerned with the formulations of a theoretical base on which a model to explain fluctuations in turnover rate is developed. For this, i t is assumed that a priori functional relationships between turnover and various explanatory variables as indicated by theoretical considerations exist. In standard economic analysis, the term, 'turnover* refers to the quantity traded. In the case of housing markets, however, i t refers to the transactions within the available stock, and hence changes in owner-ship and market adjustments rather than the total interaction of supply and demand. The ultimate outcomes of such adjustment are recorded as the number of transactions, the aggregate total of which in a given period of time is the 'turnover*. Chart I provides a conceptual framework to i l l -ustrate the working clothe various market forces and the parties involved in the process of trading. The parties involved in the purchasing side of the market are classi-fied into two main groupst former homeowners and former non-homeowners. The latter includes those who were previously renting and new households formed from in-migration and undoubling (those who previously lived with friends or relatives). We assume the following forces have, in one way or another, influenced the decision to purchase a homej demographic factors, changes in income and credit conditions, taste and preferences, govern-mental incentives and the economics of renting versus owning. Of course, not a l l these' factor;/have equal effects on both the homeowners and the PURCHASERS' SIDE I SELLERS' SIDE . D E M O G R A P H I C A I N C O M E A C R E D I T A TASTE. 4 PREFERENCE T O R f l E R HOHE-OV/MERS TRANSACTIONS ! ( T U R N O V E R ) M E W H O U S E S (COMPLETIONS) • E X P E C T E D D E M A N D C R E D I T A • PRICES o r n o u s e s FORCE O ( f i n a n c i a l ctrff\culiies, DEMOGRAPHIC A INCOME. £ T A S T E ' 4 PRiFERLNCt RE,NT v i OWNING I CHART I A CONCEPTUAL FRAMEWORK REPRESENTING THE PROCESS OF TRADING - 16 -non-homeowners; some may affect the homeowners more than the renters and vice versa. This will be discussed when each factor is considered sepa-rately in greater detail in the ensuing sections of this Chapter. On the sellers* side both existing and new (completions) houses de-termine the level of turnover. The decision to sell existing-houses and the decision to build new houses involve different market participants who are motivated by two different sets of reasons. In the f i r s t case the own-ers may be forced to sell because of financial difficulties or a change in the work-place. The other reasons such as demographic influence, the change of taste and preference and the economics of renting versus owning are similar to the demand forces identified above, except that the effects are reversed. As for new houses we should look at the behaviour and the objectives of the suppliers of housing. They are'primarily; profit-maxi-mising enterprises and the decision to build or otherwise depends on their projections of expected demand (which are determined by the forces on the demand side of Ghartll) and the price level of existing houses. The supply of houses may also be impeded or accelerated by changes in the cost and availability of mortgage funds. In addition, i t has been observed that new houses, "may experience some independent inventory moments, i t may adjust only after considerable lag. The existence of such independent reactions will allow construction activity in any period to differ from the related alterations in household~demand." Supply forces are introduced only to give a more comprehensive pic-ture of the working of market mechanism. For the purpose of simplicity in model building, according to Lionel Needleman, "under certain circumstances,.... i t is safe to assume that sales - 1 ? -(number of transactions) are determined primarily by demand fac-2 tors and the influence of supply factors can be disregarded." The diagram (Figure 2.1) represents three different shifts at dif-ferent periods of the relationship between sales and prices. Each shift is brought about by changes in both the supply and demand curves indica-ting changes in the determining variables such as income or credit con-ditions. In such a case, marginal supply is relatively volatile compared to demand; and the sales curve tend to take the shape of the demand curve. "Where there is a large response in supply to small movements in each of the factors affecting both demancTand supply, then the supply relationships can be ignored and the observed sales re-lationships can be treated as i f i t were the demand relationship without great loss of accuracy." Figure 2.1 Relationship between Sales and Prices - 18 -In support of his contention that marginal supply i s highly volatile, Needleman argues in another part of his book thatt ''Because the supply of houses changes so l i t t l e and so slowly, relatively small increases i n the total housing demand can be met only by large fluctuations in marginal supply, that i s , new constructions and conversion." We have, however, chosen to adopt a less extreme hypothesis. While supply factors may be less important i n this context, they cannot be ignored completely, and we feel that the fluctuations in the number of completions per period could be significant in explaining^ the variations in turnover rates. Also some of the supply determinants are accounted for already in the demand variables. As the model i s oriented toward explaining a short-run (quaterly) phenomenon, we have to allow for disequilibrium in the housing market. This short-run disequilibrium i s commonly explained by!the stock-adjustment m o d e l . ( T h i s i s used here s t r i c t l y as an explanation and not a basis for our model). The model in i t s simplest form may be expressed as followss <k - V i = M - *t-i> I t states that the adjustment between desired stocks (Q^ ) and the level of stocks held at the beginning of each period (0^. ^) i s proportional to the gap between the two levels (Qt - Qt_j^ )» where the reaction coefficient (&) i s a measure of the f r i c t i o n which prevents the consumer from making a complete adjustment in each period. Reasons for - 19 -such f r i c t i o n may be related to either the supply or demand side of the market. On the supply side, the plausible reasons are lag of response in demand changes because of poor information flow, and shortages of finance. As in the case of the demand, they are ignorance of changes in the market, temporary i l l i q u i d i t y and the inertia to form separate households or to move. It i s clear that imbalances in the housing market may persist for a considerable period of time because the gap between desired and actual levels of stock i s adjusted gradually. Since we are dealing with a disequilibrium adjustment process, short-run variables should therefore be included in the model. Further, we should note the fact that interaction of a l l those forces of supply and demand takes place within the general economic framework. The level of general economic activity w i l l certainly have considerable effect on each of the explanatory variables and the level of turnover. For example, during economic booms, disposal income rises and av a i l a b i l i t y of mortgage funds f a l l s while cost of capital goes up, (see ChartIV 5), i t i s f e l t that there i s no necessity to treat general economic activity as a separate variable because i t i s either taken into account by appropriate;; proxy variables or i s c o l -linear with other explanatory variables that have been included. Instead the alternating business cycles w i l l be used to relate the fluctuations of the variables in the qualitative analysis to be dealt with i n a later chapter. Having identified the potential determinants of turnover, we shall examine the functional relationships of each determinant i n greater detail. - 20 -Demographic Influence, Population growth, family formation and the size and age structure of the population are potential determinants of the level of turnover. I t i s natural that a growing population w i l l augment the need for more housing. In so far as the demand for single-family dwellings i s concerned i t i s more relevant to look at changes in family formation and family size than population growth. Also, families with children are more l i k e l y to own (or purchase) houses than those without. F i r s t , l e t us consider the influence of family size. The need for single-family houses most l i k e l y varied directly with the size of the family. In the case of home-owners, both a rise and f a l l in the size of their .families may cause them to move out of their current houses and either rent or buy more suitable homes. For instance, as the number of children in a family increase and they grow older, more space w i l l be required to accommodate them, and the family may be forced to move into a bigger house, given the income capacity. Conversely, when child-ren have grown up and l e f t home, the house may be too big for the parents, and they may s e l l i t for a smaller house, or even rent an apartment. However, the average size in British Columbia remained f a i r l y constant (between 3.3 to 3.6) during the studied period, 19^9-63.s Hence, we fe e l that family size would not contribute significantly to explain the fluctuations in turnover rates. - 21 -Next, we can further narrow down the scope of demographic influence to the age structure of the population, or more precisely, to the stages of family l i f e cycle. While most married couples set up a household immediately after marriage, the decision to purchase a home i s typically delayed a number of years. This delay i s self-evidenced when we compare the age of marriage (see Table 2.1) to the age of the borrowers of NHA mortgages (see Table 2 . 2 ) . While on the average only about 6-8$ of the borrowers were under 25 years, 22-26$ and 19-21$ were in the 25-29 and 30-34 years age groups, respectively. Another 11-13$ went to the 40-44 age bracket, beyond which the proportion of borrowers declined shrply. Two reasons may be attributed to this decline. F i r s t , most of the household heads over the age of 44 years"were already homeowners. Secondly, income-earning capacity declined, especially for those who had retired or were near to retirement. It therefore appears that the middle stages(25-44 years) of a family cycle are where the need to purchase a house i s the greatest. There are 7 a number of empirical studies which confirmed this contention. Although there i s no unanimous agreement among these studies as to thelupper and lowers-limits of the middle stage of the family cycle, the range does not go beyond the 25 to 44 bracket. A rise in the number of families headed by the 25 to 44 age group augments the need for more housing but i t s translation into effective demand i s limited by the income level and the ease of obtaining mortgage finance for their purchases. The last two aspects w i l l be discussed in the ensuing sections. - 22 -TABLE 2 .1 AVERAGE AGE AND AGE DIFFERENTIALS OF FIRST MARRIAGE FOR BRIDEGROOMS AND BRIDES IN CANADA YEAR BRIDEGROOMS BRIDES AVERAGE DIFFERENCE IN AGE 1940 27.7 24 .4 3.3 1945 27.3 24.3 3.0 1950 26.7 23.8 2.9 1955 26.2 23.5 2.7 196a 25.8 23.0 2.8 1965 25.3 22.6 2.7 1966 25.2 22.6 2.6 1967 25.0 22.6 2.4 I968 25.0 22.6 2.4 Sourcet F.A. Wakil, "Marriage and Family in Fanadaj A Demographic-Cultural Profile," in K. Ishwara, Id., The Canadian TABLE 2 . 2 DISTRIBUTION BY, AGE OF BORROWERS AGE OF BORROWER 1950 1956 1957 1958 1959 i960 1961 1062 ®&XfmM or less 6 . 5 6 . 2 8 . 0 7.1 7 .2 7 .8 7 . 2 6.8 25-29 21.7 23.2 23. k. 25.1 24.9 35.9 35.0 23.8 30-34 24.3 25.7 25.7 25.6 25.0 25.3 25.8 25.3 35-39 20.7 20.0 2 0 . 3 19.7 19.8 19.2 19.2 19.5 40-44 13.0 13.2c 12.2 11.8 11.8 11.2 U . 5 12.5 45-49 7.6 7 .3 6.7 6 . 8 6 . 8 6 . 3 6.7 7 . 3 50 years or more 6 . 2 4 . 4 3.7 3.9 k.5 4.3- 4 . 6 4 . 8 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Average Age of 34.8 33.8 34.4 34.1 Borrower (years) 34.2 33.9 34.2 34.5 Source: C.M.H.C. Statistics. - 23 -Income Effects Demographic changes reflect only the potential demand; sales ^number of transactions) w i l l only increase i f households have the a b i l i t y to purchase. The a b i l i t y to pay i s , however, a function of income. The relationship between income and demand for housing has been the subject of much controversy among economists for more than a century. Engels had observed i n 1857 that housing expenditure tends to be a constant percent-age of income at a l l levels of income. This contention was soon challenged by Schwabewho, with empirical evidence, argued that the percentage of third hypothesis i n which he asserted that demand for housing i s rather income-elastic.^ In more recent years, the same topic has attracted a host of empirical studies. The center of disagreement l i e s in whetherKheusing i s related to "permanent or normal" income or '!current or measured" income. The earlier studies (undertaken by Morton and Winnick)^used only current income data. Both Morton and Winnick, using cross-section data, arrived at an income el a s t i c i t y of demand for housing of about 0 . 6 and 0 . 5 res-pectively. Their findings were consistent with Schwabe/s hypothesis of rent, suggesting therefore that the higher the measured income, the lower the housing expenditure. 10 11 Muth and Reid were f i r s t in using permanent income instead of current income. Both arrived at an income el a s t i c i t y of demand for hous-ing of more than unity with respect to normal income and less than one for measured income. There were many subsequent studies, using both time series and cross-section data; notably by Lee (1964, I 9 6 8 ) * 2 , Winger. ( 1 9 6 8 ) ^ , housing f a l l s as income rises. Marshall presented a - 24 -L.B.Smith (1971)^, and Frank de Leeuw (I97l)1'^'. Although there i s wide variations in the demand e l a s t i c i t i e s derived toy them, their studies are consistent in one result, that i s , housing demand appears to be more responsive to changes in permanent income than to current income. This . finding i s theoretically sound. Since the decision to purchase a house involves a capital commitment of considerable magnitude over a long period of time, i t i s more l i k e l y to be related to expected future streams of income. There i s yet another hypothesis to explain why demand for single-dwelling increases with higher income by G. C a r l i n e r ^ , "Because the imputed rent from owner-occupied homes i s exempted from income taxation, owners receive an add-iti o n a l return on i h i s form of property equal to their marginal tax rate times the imputed rent. As incomes rise marginal tax rates rise, and the advantage of investment in owner-occpied housing over other forms of investment also rises." However, this tax advantage applies only to the non-homeowners. We shall now look at some empirical evidence in the light of these hypotheses. Table 2.3 shows that from 1950 through 1963 the average dwelling costs increased progressively with purchasers' (applicants') income, indicating that households with higher income were willing to commit more money to housing, either for more space or quality. On the other hand Table 2.4 seems to support Schwafee's contention that as income rises, housing expenditure tends to f a l l . The average ratio of gross debt service to incomes tend to decline as income increased. However, Schwabe's TABLE 2.3 AVERAGE DWELLING COSTS AND APPLICANTS* INCOME ~ — : — : Applicant's Income 1950 .1955 : 5$. 5;, 1958; '1959 ,1960 1961 I962 Average Dwelling Costs (Dollars) Under - 3,000 8,266 10,094 10,673 12,471 10,936 11,086 11,321 10,908 11,854 3,000 - 3,999 9,518 10,743 11,276 11,926 12,070 12,241 12,097 11,948 12,085 4,000 - 4,999 10,974 12,124 12,627 13,107 13,329 13,555 13,345 13,217 13,35^ 5,000 - 5,999 12,015 13,364 13,745 14,443 14,288 1^,513 14,498 14,309 14,465 6,000 - 6,999 12,600 14,227 14,667 15,417 14,995 15,227 15,346 15,221 15,475 7,000 - 7,999 13,241 14,841 15,346 16,095 15,806 15,919 16,020 15,9^5 16,187 8,000 - 8,999 13,377 15,305 15,837 16,830 16,296 16,725 16,792 16,792 16,751 9,000 - 9,999 13,561 16,028 16,39^ 17,380 16,776 16,943 16,888 17,162 17,275 10,000 and over 13,903 17,310 18,366 18,366 17,841 17,582 17,809 18,302 18,074 Average Dwelling Cost , 9,725 12,598 13,366 14,512 14,237 14,516 14,380 14,474 14,815 Source! CMHC TABLE 2.4 AVERAGE RATIOS OF GROSS DEBT SERVICE TO INCOME AND APPLICANTS' INCOME Applicant's Income 1950 1955 1956 1957 1958 1959 I960 I96I I962 Average Ratios of Gross Debt Service to Incomes (Per Cent) Under - 3,000 22.1 26.3 26.3 31.6 32.7 31.2 30.9 30.8 33.4 3,000. - 3,999 20.3 22.5 22.8 24.2 26.7 26.8 27.4 . 27.8 27.9 4,000 - 4,999 18.? 20.7 21.0 22.1 24.1 24.5 24.8 25.7 25.7 5,000 - 5,999 16.9 18.7 19.1 20.2 21.2 21.7 22.5 23.6 23.7 6,000 - 6,999 15.1 16.6 16.8 18.2 18.7 19.1 20.0 21.1 21.4 7,000 - 7,999 12.8 14.6 14.9 16.2 16.6 16.9 17.8 18.8 19,1 8,000 - 8,999 11.4 13.2 13.5 14.6 14.9 15,3 16.2 17.2 17.1 9,000 - 9,999 10.1 12.0 12.2 13.3 13.5 13.8 14.7 15.6 16.0 10,000 and over 8.1 9.3 9.3 10.4 10.2 10.7 11.5 12.3 12.7 Average Ratio 19.1 18.6 18.3 18.5 19.9 20.1 21.2 21.7 21.4 Source1 CMHC - 27 -contention may not be confirmed i f Table 2 . 4 i s compared to Table 2 .5 . In Table 2 .5 except for the income level under $3,000 per year, average down-payments rose as income increased. The higher down-payment partly accounted for the corresponding low annual debt service for the higher income household. In fact, the difference in debt service was capitalised in the form of higher down-payment. Hence the housing expenditure of the higher income households may not be proportionally lower than the lower income households. Another relationship worth looking into i s that in Canada, although housing costs have increased, rising incomes have generally matched price increases. The per capita personal disposable income rose by 79.k% between TABLE 2 .5 Applicant's Income 1950 1955 1956 1957 1958 1959 i960 I96I I962 Average Down-payments (Dollars) Under 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 3,000 - 3,999 - 4,999 - 5,999 - 6,999 - 7,999 - 8,999 - 9,999 & over 2,029 2,333 2,957 3,558 3,933 4,631 4,726 5,091 5,345 2,505 2,122 2,465 2,940 3,400 3,813 4,082 4,619 5,417 3,182 2,471 2,798 3,255 3,737 4,147 4,478 4,895 5 ,762 4,304 3 ,043 3 ,126 3 ,603 4,243 4,553 5,117 5,521 6,473 2,629 2,294 2,500 2,924 3,350 3,910 4,252 4,761 5,758 2,529 2,331 2,514 2,938 3,353 3,865 4,360 4,660 5,390 3,027 2,306 2,435 2,938 3,451 3 ,915 4,442 ^,578 5,489 2,407 1,950 1,872 2,233 2,732 3,110 3,670 3,907 4,877 2,935 1,996 1,843 2,119 2,591 2,991 3,277 3,644 4,324 Average Down-Payment 2,544 2,773 3,217 3,826 3,057 3,094 3,033 2,475 2,421 Sourcej CMHC - 28 -19^9-1960compared to an increase of 69.5% for cost of home ownership, indicating that the a b i l i t y to pay for housing had not deteriorated over that period. I t i i s important to note that the income effects on housing should not "be viewed in isolation, but in conjunction with other variables such as demographic influence, credit conditions, taste and preferences of buyers, etc., a l l of which exert considerable;influence on the decision to purchase a house. This partly accounts for the difference in results obtained from the various empirical studies. They vary from a highly simplistic model which assumes a direct relationship between income and housing to a complex model which includes most of the variables mentioned. Ga?edit> Sonsiderations To most average families, purchasing a house involves the largest single capital commitment in their lifetime. Ordinarily, few people can afford to purchase a house with 100% equity. This means that most of them have to get some kind of financing for their purchases. Therefore i t i s logical to expect the demand for houses to be highly sensitive to changes in credit conditions, and as such, sales of houses inevitable vary with ease (or d i f f i c u l t y ) of obtaining mortgage loans. The effects of changes in credit conditions on housing demand are f e l t through both the level of interest rate and the mortgate terms. As shown in Table 2.6 (next page) a 1% point increase in interest rate from 7% to 8% on a twenty year mortgage increases the monthly payments for principal and interest by 7.9% or about $6.00 for a $10,000 loan. However a few dollars increment through a rise in interest rate w i l l - 29 -TABLE 2 .6 Monthly Payment Per $10,000 of Mortgage For A Twenty (20) Year Term At Increasing Interest Rates Interest Rate Monthly Payment Net Increase $ $ I $ 5 66.00 ... X 6 71.64 5.64 8.54 7 77-53 • 5.89 8.22 8 83.65 6.12 7 .89 not alter the housing demand materially unless the rate of increase i s more than 2% or 3% within a short period of time, and which i s unlikely. It i s f e l t that changes in terms are l i k e l y to affect the pur-chasing power of the resources (income and savings) available to buyers of houses more effectively than interest rate alone. Tightening of terms i s usually taken to mean a larger down-payment of the price (or lower loan-to-value ratio), larger monthly payments needed to pay off an amortised mortgage of a particular size or a shortening of the amortisa-tion period. The reverse happens i f mortgage terras are relaxed. Mortgage institutions normally function on the objective of maximum security of capital coupled with minimum risk on inflow of funds. A - 30 -mortgagee i s not only concerned with the value of the property in the ease of default but also the capacity of the mortgager to repay. As a security against the borrower, a minimum income level i s set and this condition must be met to qualify for a mortgage loan. Usually as a guide-line, a certain percentage of the borrower's income i s set as the minimum requirement. For example, assume the monthly mortgage payment 17 to income ratio i s 25$ • A relaxation of mortgage term from 15 years to 30 years at 8$ interest reduces the minimum monthly income require-ment from $382 to $294, a net reduction of 23$. (see Table 2 . 7 ) . TABLE 2.7 Relationship Between Minimum Income Mortgage Term And Monthly Payment At 8$ Interest Term Monthly Payment Minimum Monthly Income Net Change In  Income Require- ment From 15 Year Term Years 2 15 20 25 30 95-57 83.65 77.18 73.38 3.82 3.35 3.09 2.94 47 73 88 12.3 19.1 23.0 - 31 -However, in practice, i t i s l i k e l y that an increase in the length of maturity w i l l be followed by a rise i n interest rate as the loan i s now more risky. The amount of down-payment w i l l also affect a person's desire to purchase a house. The opportunity cost of committing his savings by way of a down-payment to buy a house i s the return foregone from an alternate investment. Assuming that he i s economically rational, i f the rate of return on the alternate investment i s lower than the imputed return from the house, he w i l l be prepared to make a larger down-payment. If, however, imputed return i s lower, he w i l l be reluctant to make a down-payment of any appreciable sum. In the case of current homeowners, the amount of down-payment i s less l i k e l y to be a problem. They can pay off a l l or part of the down-payment from the equity build-up of the houses they have owned, Hence, the diff e r e n t i a l between the down-payment and the equity would more l i k e l y to their concern. We should not however overlook the inter-relationships among monthly payment, down-payment, amortisation period and interest rate. For example, the monthly payment per $10,000 of mortgage i s $69.10 at % interest and 18-| year maturity, as well as at k% and l6j year and 6% and 2i-| year (see Figure 2.2). Supply of housing i s also sensitive to changes in credit conditions. During periods of credit restraint, builders are forced to reduce hous-ing construction not only because of the decline in effective demand, but also because of the high cost of borrowing which means lower profit margin and d i f f i c u l t i e s of obtaining committments from mortgage lenders - 32 -Figure 2.2 HeatUy Hjmtal p*r |10OO of Hortgag* by Maturity mt Virions Intorut RiUs $10.00 9.00 8.00 7.00 6.00 5.00 ti.00 3.00 1.00 t, i , I I I I I I I I I I L 0 6 0 12 lb 16 IS 20 22 a 26 28 30 32 3b Katorltjr, In jroirs J L 33 UO Sourcet R. Wood, "Credit Terms and Demand for Residential ; Construction", Study of Mortgage Credit, Op. Cit . on permanent financing. The latter reason is expecially relevant in an industry characterised by numerous small firms. We shall now examine the overall market effects of changes in mortgage credit terms. It appears that relaxation ,of credit terms, ceteris paribus, wil l result in a rapid increase in turnover, greater use of credit and a rise in prices of houses. However turnover wil l slow-down and prices wil l stabilise after a few months. As mortgage terms relax and more households can qualify to buy a house, the number of bidders entering the market increases. The net result is a rapid rise in turnover and prices of existing houses. - 33 -When prices rise to a high enough level to exceed costs of construction, developers will!.be induced to build more. However the upsurge i n the number of transactions and prices eventually wear-off as ava i l a b i l i t y of funds f a l l s and credit terms tighten, or as prices rise to a level as to destroy the original credit advantages enjoyed by the bidders. A detailed theoretical study, more or less along the same line of argument 18 was done by Ramsay Wood . The next question i s why are there alternate periods of l i b e r a l i -sation and tightening of credit terms. Changes in credit terms are the result of fluctuations in the supply of mortgage funds, which can be explained by two principal reasonsi ii'X) Corporate demand for capital funds i s relatively less sensitive than that of the housing industry to changes in credit conditions, and (2) Bond yields are relatively more flexible than mortgage yields to changes in credit conditions. The mortgage market i s only a sub-market of the overall capital market. Mortgages have to compete with other capital market instru-ments such as public (federal, provincial and municipal) and corporate bonds. The strongest competition traditionally comes from the business 19 sector, especially during general economic expansion. During economic upswings, aggregate demand increases, and as a result businesses follow an expansionary policy. When their retained earnings rise less than the planned increase in their investment in inventories and fixed capital, they have to look for external finance. - 34 -However, as demand for funds increase when expansion in general business occurs, interest rates rise and borrowing terms lighten. Avai l a b i l i t y of funds i s further limited when monetary policy shifts from ease to restraint as capacity ceilings are approached. The expanded demand for funds by the business sector tends to reduce the fund available for housing. Figure 2.3 i l l s t r a t e s this contention. Figure 2.3 - Competition for Capital Funds Assume a fixed supply of funds i s distributed equally between corporate bonds and mortgages represented by the shaded areas, Q 0P O L c and Q J J O P ^ . Note that the demand for mortgage funds i s comparatively more elastic than that of corporate funds, being more sensitive to changes i n credit conditions. As business a c t i v i t i e s expand, G N P and income increase, and the demand for mortgage funds and corporate funds - 35 -shift upward from D m to Dm,and D c to DQ, respectively. The demand schedule of forporate funds "being less elastic, attracts more funds as represented hy Q c0P l M (i.e. Q ^ O P ^ Q ^ Z ) . The whole process i s reversed when general economic activity contracts. Demand for capital funds f a l l and this i s followed by easing of credit terms and f a l l i n g interest rates (and generally an easing of monetary policy). Though income f a l l s when general activity contracts, the ava i l a b i l i t y of funds for the housing sector tends to increase (Q^OPgX> Q^OPgK). This i s because the demand schedule for mortgage funds i s more elastic. So far we have been looking at the demand side of the mortgage K market. The reasons for the relative i n f l e x i b i l i t y of mortgage yield 20 and the consequential reduction i n ava i l a b i l i t y of mortgage funds have to be examined from the supply side of the mortgage market. Part of the problem arises from the asymmetry of the assets and l i a b i l i t i e s of the financial institutions which supply most of the mortgage funds. Apart from the insurance companies (whose residential recent ?1 mortgage portfolio has decreased considerably in^years) , credit unions, trust companies, mortgage loan companies and banks provide highly l i q u i d assets for individuals to hold; such assets are close substitutes for short-term market instruments on which yields are highly elastic. While their l i a b i l i t i e s are relatively short-term, their assets being mainly i n mortgages (except banks), are long-term. Since the average yields on their earning assets (mortgages) do not respond readily to changes in market interest rates, their inflow of - 36 -funds derived from consumers' savings are competed away during i n f l a t i o n -ary periods. With a f a l l in the inflow of funds and a rise in market interest rates, these lending institutions have to re-adjust their investment portfolio. Mortgage lenders are principally profit maximizing i n s t i t u t -ions; therefore, when credit conditions change from ease to restraint and "bond yields rise faster than mortgage yields, they tend to switch out of mortgeges into other more profitable forms of investment leading to a further reduction of supply of mortgage funds. Conversely,Muring economic downswings, bond yields decline faster than mortgage yields and by the same reasoning supply of mortgage funds increase. The effects of monetary policy, mentioned only b r i e f l y above deserves more detailed comments. One main objective of monetary control i s to ensure the s t a b i l i t y of the economy, i.e. to adopt a policy of restraint during inflationary periods and ease during downswings. Such intervention generally distorts the distribution of funds among the various sectors i n the economy. Because of the characteristics of the mortgage market and the nature of i t s relationship with the overall capital market, a v a i l a b i l i t y of mortgage funds tends to be a 'victim* of monetary policy. I t accentuates the shortage of mortgage funds during .economic upswings and the reverse occurs during economic down-swings . The Economics of Rinting Versus Owning There are essentially two alternatives open to a person looking for accommodation; he has a choice of either renting or-buying. - 37 -Because the "pre-ownership" sentiment i s built deeply into the North American culture, the desire to purchase a house i s often governed by y ndn-pecuniary factors. On the other hand, i t i s not unreasonable to assume that most people are somewhat economically rational, and as such, the relative cost of renting to owning a house i s considered in making a decision to purchase a house. If the cost of renting exceeds that of owning, we can expect more people to buy and vice versa. We consider this notion of renting versus owning significant enough to warrant sepa-rate treatment. Economic reasoning suggests that under equilibrium conditions i n the long-run, the costs of renting should match the cost of home-owning.. If renting i s cheaper than owning a home, more people w i l l s h i f t to the rental market, thus pushing up rental values. In the long-run, rents w i l l rise to a point where the costs of renting equal that of owning. In reality, however, owing to market imperfections, principally poor flow of information and market inertia, this equilibrium point may not be reached. Moreover we are only interested i n the short-run during which equilibrium i s unlikely to be reached. 22 John-Shelton's analysis i s one of the most definitive and com-prehensive among the many studies done on this issue of renting versus owning. He concludes that i t i s cheaper to own a house than to rent i n the long-run? the break-even point i s 3i years when tax advantages are taken into account and 5 years, when they are not. In other words, i f a person expects to l i v e in a house for more than 3r years, he w i l l buy, given the financial capacity. He assumes that the land lord transfers a l l his costs which include property tax, maintenance, obsolescence, interest on mortgage, vacancy - 38 -allowance, management and return on investment in the form of rents to the tenants. The home-owner has to pay for most of the cost items too hut saves on tax differences, management costs, vanancy allowance and maintanance. There is one other non-recurring cost item for home-ownership, that i s , transfer costs which consists of realtor commiss sions plus an allowance for certain fixed costs such as t i t l e search, deed recording, etc. The 3T years-break-even point i s arrived at by dividing a 2% per annual savings into the transger costs of about 7% of the value of the house. In the case of Canada mortgage repayments are not deducti-ble for income tax purposes and therefore the home-owner here does not enjoy the tax advantages that his counterpart i n the United State has. The break-even point in Canada should therefore be closer to 5 years. Again, this tax advantage does not apply to a prospective buyer who has already owned a house. Taste and Preference Besides demographic and financial considerations, the decisions to purchase a house i s also affected by a host of factors at a micro-level, commonly summed up as "taste and preference factor". A number 23 of surveys, notably the two conducted by the University of Michigan 24 and University of Connecticut, were undertaken specially to determine consumers' behaviour and preferences in the home buying process in local housing markets. The act of buying a home i s the f i n a l outcome of a process of interaction between the buyer and his social-psychological environment. /His evaluation of the relative s u i t a b i l i t y or desirability of the housing alternatives i s influenced by his personal values, motivation and attitudes. Before making the f i n a l choice, he i s l i k e l y to engage in an information-seeking process and his original preferences may be modified by the information gathered. Also, his original preference may be limited by what the market can offer. "In this context, the home that i s purchased may not necessarily entirely reflect what the buyer wanted, but rather, reflects an adaptive response to what he finds. His f i n a l choice i s li k e l y to be influenced by the following con-siderationsrlin iaddition to what has been discussed already i i . location i i . neighbourhood i i i i . size or space of house iv. features of house such as design, type of construction and general condition. Unless a survey i s specifically designed to identify the beha-vioural variables attributable to the home buying process, they are not normally discernible, l e t alone quantifiable. It i s assumed in this study that the effects, of behavioural factors tend to cancel out in the process of aggregation. They are discussed here only for the purpose of giving a more comprehensive account of the market forces which may have influenced the level of turnover- . * - 40 -Footnotes t 1. Maisel ( 2 3 ) , pp.76-77. 2. Needleman(2?)t p.53. 3 . Ibid. p.149. 4 . Ibid., p.150. 5. Huang(l4), Lee(18), Muth(26) & Silver ( 3 1 ) 6. D.B.S. 7. Atkinson ( 3 ) , Hempel(13), Lansing ( l 7 ) , Maisel ( 2 3 ) , Rossi ( 2 9 ) . 8. Refer to the opening paragraphs of Reid(28), Muth(26), Lee(18). 9. Morton(25), Winnick(44). 10. Muth(26). 11. Reid(28). 12. Lee(18) 13. Winger(42). 14. L.B. Smith(32) & (35). 15. de Leeuw(20). 16. Carliner(6), p.114. 17. In the case of N.H.A. mortgages, the maximum debt-service ration was 30$. 18. Wood(46). 19. This i s based on the assumption that corporations account for a large proportion of the nation's capital formation. 20. It i s generally believed that conventional mortgage rates tend to lag 3 to 6 months behind changes i n bond yeild. (16) (54) 21. Life insurance companies accounted for less than 10$ of the total - 41 -volume of new lending by private financial institutions in 1973. The importance of insurance companies as a source of supply of residential mortgage funds has declined. (15) 22. Shelton(30). 23. Lansing (17). 24. Hempel(ll). 25. Hempel(13), p.103. - 42 -CHAPTER III THE REGRESSION ANALYSIS This section attempts to quantify the functional relationships "between turnover rates and the various factors discussed in the pre-vious chapter by means of multiple regression analysis. Both the over-a l l (existing and new houses) turnover rate and turnover rate of existig houses w i l l be tested separately. In the analysis, time-series data from the last quarter of 1949 through to the last quarter of I963 are used. The reason for using quarterly instead of annual data i s to pro-vide sufficient degrees of freedom for the regression and also to study the seasonal fluctuations of turnover. The sources and the procedure and methods of obtaining data for a l l the variables are given i n the appendix. Turnover of Existing and New Houses The choice of appropriate independent (or explanatory) variables to f i t into the regression equation i s based on the theoretical considerations discussed previously and i s limited by the ava i l a b i l i t y of suitable data. We assume a series of a p r i o r i functional relationships which i s expressed as follows 1 T - f ( A H , AY, LV, rm, A, rm-rb, R/P, C, e) (1) where T Q m turnover rates of existing and new houses AH = change in. the number of families headed by 25-44 year married males - 43 -£Y = change in the real disposable per capita personal income LV = loan-to-value ratio of conventional mortgage rm = interest rate of conventional mortgage A .= length of maturity of conventional mortgage rm - rb = interest d i f f e r e n t i a l between yields of conventional mortgage and long-term corporate bond VR/? = rent index to price index ratio G = completion ratio e » error disturbance AH and AY measure the need for housing and the income and economic position of the purchasers respectively; both variables are expected to vary directly with turnover rates. Credit changes are represented by LV, A and rm. The f i r s t two variables measure changes i n the terms of borrowing and are positively related to turnover. The cost of finance i s measured by rm and i s expected to be negatively related. The interest di f f e r e n t i a l (rm - rb) measures the av a i l a b i l i t y of mortgage funds. A rise in the interest differential w i l l encourage more funds to be made available for mortgage financing. We expect a rise in R/P to be f o l l -owed by an increase in lurnover as i t i s now cheaper to rent than to own a house. Completion ratio i s calculated by dividing the completed dwelling units during the quarter by the standing housing stock at the end of the quarter. It measures the rate at. which new houses are : available i n the market. As there i s always a demand for housing, we expect most new houses to be sold within a short period of time. A large portion of the turnovers (or number of transactions) i s therefore - U n -accounted for by the sale of new houses; in fact, completion ratio i s an integral component of turnover rate. Ordinary least square method i s used in the regression analysis, and the s t a t i s t i c a l problems normally associated with regression involving time series data are also taken into account. A linear relationship between turnover rate and the other explana-tory variables as given in equation (i) i s assumed. There i s , however, no a p r i o r i reason to assume that this relationship i s necessarily linear, but i t is useful as a f i r s t step towards deciding whether non-linear regression i s appropriate. Model I i s entirely static with no lagged variables. It i s assumed that the adjustment of turnover rates to changes in the independent variables was instantaneous, or at least was completed within one quarter. The estimated regression equation i s as follows j T = .0216 + .0053 H - .0006 Y - .0092LV - .I369rm + .OOOlA 0 (4.616) ((.8201) (1.926) (1.034) (2.529) + .0329(rm - rb) + .0113R/P + .1534c (2) (0.154) (2.146) (3.429) R2 =? 0.8182 " KW.' - 1.4463 SER - 0.0031 Significance tests are performed on each of the independent variables at a 95% significance level to determine i f the estimated coefficients are significantly different from zero. I f they are, we can conclude that the independent variables are significantly related to the dependent varia-ble, T. The t - s t a t i s t i c at 9% level of significance and 48 degree - 45 -of freedom is I.679. The figures in parenthesis are the t-statistics of the respective coefficients. In equation (2) only the coefficients of AH, A, R/P and G are statistically significant. Though rm and rm - rb are insignificant, they have the correct signs and should not be rejected. 2 The coefficient of determination, R is 0.8182, indicating that 81.82% of the variation in turnover rates is accounted for by the regres-sion equation. The Durbih-Watson statistic (D.W.) is commonly used in time-series regression to check for serial correlation. It is felt that Model I is unrealistic as the adjustment process could not be instantaneous. Therefore, certain lagged variables are introduced into the equation. In Model II, H, G, rm and rm - rb are lagged one period and the regression results are as followst Tot * .0372 + .0024AHt_1 - .00UYt - .0094LVT - . 3 3 6 2 x 1 1 1 ^ (5.888) (1.876) (1.245) (1.708) (4.596) + ,00002At + ,3681(rm - rb ) t _ 1 + .0064R/P + .100lC t ± (1.983) (1.811) (1.458) (2.567) .....(3) R 2 = 0.7597 D.W. - 1.5393 S E R - 0.0035 In model II, besides A H , A and G, rm and rm - rb are also signifi-cantly different from zero; but R / P , though has the correct sign, is insignificant. There are, however, strong theoretical grounds to believe R/P to be positively related to turnover. L V and AY remain insignificant and they also have the wrong signs. The statistical results should not - 46 -be taken as entirely conclusive as the presence of multicollinearity and s e r i a l correlation could have caused gross distortions. One of the basic assumptions of regression model i s that the explanatory variables are independent of each other. However, because of the non-experimental data used, i t i s possible that the explanatory variables are in fact correlated with each other. The correlation matrix of the variables are shown in Table 3.1. While regression analysis explains how the variables are related, correlation analysis shows us the degree to which these variables are related. The problem of multicollinearity i s less serious than expected. A l l the partial correlation coefficients between the independent varia-bles are less than 0.4, with the exception of the following in order of seriouness j = - . 6 6 2 6 tyP . rm rLV.A " '^28 r ( r m * - r b ) . A = r(t»-A).LV- • M f 6 ° It i s not surprising that these variables are highly correlated, A rise i n the cost of mortgage finance causes developers' cost to increase and in the short-run a reduction i n profit margin, and then in starts, the net result of which i s a rise i n price. This indicates that rm and R / P are negatively correlated. As expected, easier credit terms i n the form of higher loan-to-value ratio and longer length of maturity are the results of an increase in the a v a i l a b i l i t y of mortgage funds TABLE 3.1 - CORRELATION MATRIX OF T~ T 1.0000 o H 0.3858 Y 0.0393 LV -0.0489 R/P 0.6072 G 0.4827 rm -0.6900 A 0.2347 vrmsrb 0.2059 AH AY 0.3858 0.0393 1.0000 -0.1700 -0.1700 1.0000 0.0230 0.0212 0.1082 0.2796 0.3501 0.0813 -0.0921 -0.0530 0.0185 0.1281 0.3002 0.2059 LV R/P -0.0489 0.6072 0.0230 0.1082 0.0212 0.2796 1.0000 -0.0885 - 0 . 0 8 8 5 1.0000 0.0120 0.3012 0.1808 -0.6626 0.5728 -0.1314 0.4460 0.0828 G rm 0.4827 -0.6900 0.3501 -0.0921 0.0813 -0.0530 0.0120 0.1808 0.3012 -0.6626 1.0000 -0.1455 -0.1455 1.0000 0.1832 0.0283 0.2229 0 . 2 5 U A rm-rb 0.2347 0.2059 0.0I85 0.3002 0.1281 0.2058 0.5728 0.4460 SO.1314 0.0828 0.1832 0.2229 0.0283 0 . 2 5 U 1.0000 0.5347 0.5347 1.0000 - 48 -as indicated by a rise in rm - rb. This relationship. explains why LV, A and rm - rb are positively correlated, and perhaps also accounts why LV i s st a t i s t i c a l l y insignificant. Both R/P and LV are superfluous. The income variable (AY) has been a poor performer in both equations in that i t has a wrong sign and i s s t a t i s t i c a l l y not distinguishable from zero. It i s possible that the dependent variable (T) i t s e l f has already contained a great deal of information on the income and economic position of the purchasers. On the other hand, from the point of view of need, changes in the number of household heads appear to have some role to play in explaining the fluctuations i n turnover rates. Nevertheless, i t should be noted that for predictive purposes, the presence of multicollinearity does not reduce the c r e d i b i l i t y of the equ-ation, although structural questions cannot be answered accurately. Another s t a t i s t i c a l problem often encoutered in regression analysis involving time-series date i s s e r i a l correlation of errors. One basis v assumption of the regression model i s that the residual, e^ i s f u l l y ran-dom and that successive values of e^ are s t a t i s t i c a l l y independent. How-ever, i t i s often observed empirically that the residual, e^ at any time t i s correlated with one or more of the previous values (e^ ^, e^ 2» etc.). This i s termed s e r i a l correlation. The Durbin-Watson Statist i c (D.W.) i s used as a check for s e r i a l correlation. As the mathematical derivation of D.W. can be found in any standard econometric text, i t shall not be dealt with here. The upper and lower limits for the values of D.W. which indicate s e r i a l correlation at various significanca levels, by number of observations have been tabulated. The values of the D.W. of both equations (2) and (3) are within the inconclusive range which i s between the upper - 49 -and lower "bounds. In other words,it cannot be proven whether s e r i a l correlation i s present or not. There are mathematical techniques to deal with this problem, but their practical application in econometric work i s opened to doubt. For example, the Cochrane-Orcutt iterative technique, which i s incorporated into the TSP (Time-Series Processor) program available at U.B.C, has been found to make things worse than 1 using ordinary least square technique on the original model. To demonstrate the problems of multicollinearity and, to a lesser extent, s e r i a l correlation, the stepwise procedure i s used. Successive independent variables are regressed against .the dependent variable as shown i n Table 3 . 2 . After equation ( 7 ) , each successive addition of independent variable 2 does not improve R significantly, and more important, the values of D.W. actually f a l l . Such s t a t i s t i c a l behaviour indicates the presence of multicollinearity as well as s e r i a l correlation. Equations (4) to (7) do not shown any unusually high partial corre-lation between the independent variables. The only partial correlation coefficients exceeding 0 . 3 are between H and rm— rb ( .3002) in equation (6) and between AH and G (.3501) in equation (7) . However as from equation (8) onwards, the values of the pa r t i a l coefficients of certain variables increase sharply as shown below» Equation (8) r ( r m _ r b ) # A = 0 . 5 3 4 7 Equation (9) r^.R/p = O.6626 Equation (10) r A < L Y =0.5728 TABLE 3 .2 - STEPWISE REGRESSION OF T o EQU. WO. CONST AH t_ 1 sm t - 1 ( r m - r b ^ C t_ 1 R/Pt LV t 4Y t •" R D.W. SER 4 . 0 2 7 0 ( 3 ° ° | i ) . 1 ^ 9 1 .7091 . 0 0 6 3 5 ' ° 5 1 5 ($73) ( 7 ^ 0 ) . 5 8 1 0 1.0321 . 0 0 4 5 6 . .0^ 65 (J?^o) (IfllS)- • - 6 7 1 3 1 > 2 3 3 9 - 0 0 4 0 _ i • ' 7 0 4 2 1 - 0 0 2 1 - ; 4 2 6 8 . 5 0 6 0 : .1253 ™ 1 • N W ^ . 7 * W 2 1 ( I . 7 8 6 ) ( 9 . 2 4 3 ) (3 . 5 1 3 ) (3.396) - 7 3 1 0 1 < 6 0 6 8 - 0 0 3 6 p 0 Z , 1 7 .0023 - . 4 2 0 6 . 4 1 9 4 . 1 2 1 4 . 8 9 6 1 , ^ _ 8 . . 0 4 1 7 ( 1 > Q i f Q ) ( 8 > 9 9 ^ ) ( 2 f 4 l 6 ) ( 3 . 2 6 4 } , ( . 8 9 6 1 ) - 7 3 5 1 1-5639 .0036 Q .0026 - . 3 6 8 3 .3660 . 1 0 9 8 . 0 0 0 0 1 . 0 0 4 0 : „ « « „ ' ' ' " « „ , o r / ' ^ / 9 * - 0 3 6 7 (2.115) ( 5 . 0 6 2 ) ( 1 . 5 4 1 ) (2.796). ( 1 . 2 4 2 ) ( . 9 3 9 3 ) 5 , 7 3 9 7 1 -5789, .0036 1 n .0027 - .3^72 . 3 3 7 8 .1001 .00002 .0049 - . 0 0 8 5 „ M n « e h n l , n n ^ , 1 0 * 0 3 a 5 ( 2 . 2 1 2 7 ) ( 4 . 7 5 3 ) ( 1 . 6 6 4 ) (2. :55^) ( 1 . 8 3 4 ) ( 1 .158) (1.555) , 7 5 2 G U 5 k 2 k ' 0 0 3 5 . 11 0372 - ° 0 2 4 "-3362 . 3 6 8 1 . 1 0 0 1 . 0 0 0 0 2 . . 0 0 6 4 - . 0 0 9 4 - . 0 0 1 . 7 £ - q 7 . ™ n n ^ 11 .0372 ( 1 < 8 ? 6 ) ( 4 > 5 9 6 ) ( M n ) ( ; 2 t 5 6 ? ) ( l ^ 8 3 | ( 1 . 4 5 8 ) (1.7P8) ( 1 . 2 4 5 ) ' 7 5 9 7 1 , 5 3 9 3 - 0 0 3 5 - 51 -(rm - rbJ'LV The above s t a t i s t i c s indicate that certain variables are superfluous, that i s , the addition of them into the equation increases multicollinearity. We therefore suggest the omission of A, LV, and R/P i n order that more reliable results of the other varibles may be obtained. However, w i t cannot be said that the omitted variable has no effect but merely that i t s effect i s confounded with that of the other explanatory variables and 2 cannot be isolated." Also,:only equation (?) has a Durbin-Watson Statis t i c (1.6068) which, at 99$ level of significance, indicates no evidence of positive s e r i a l correlation - in fact i t i s just outside the upper l i m i t of 1.55. However, at this level of significance, AH i s s t a t i s t i c a l l y insignificant. Never-theless, we believe that there are strong theoretical grounds as discussed i n Chapter II to include H i n the equation. Moreover, i t i s significant at the 95$ level. For a l l these reasons we f e e l that equation (7) i s s t a t i s t i c a l l y the best model both for predictive purposes and for explaining the structural relationships between the dependent and the independent variables. T o t = , o Z |' 2 1 + , 0 0 2 l A H t - l " ^ 2 6 8 : n V l + • 5 0 6 o(™-*b) t _ 1 (1.786) (9.243) (3.513) + .l253Gt,1 (7) (3.396) R2 - O.73IO D.W. « 1.6068 SER - 0.0036 - 52 -The model should however be used only as a guideline bearing in mind i t s s t a t i s t i c a l limitations and, especially for predictive purposes, to make allowance for changes in exogenous events such as the introduction of condominium housing into the market. Turnover Rates of Existing Houses The regression equation for turnover rate of existing houses (T ) i s based on the assumptions discussed previously and i s expressed as follows» T =c?(kH, AY, rm, LV, A, rm - rb, R/P, e) .(12) Note that the only difference between "equations (1) and (12) i s the change of the dependent variable from T Q to T g and the exclusion of the indepen-dent variable of completion ratio (C). It i s hoped that the exclusion of explanatory variables w i l l reduce a certain amount of s t a t i s t i c a l bias in the regression. This s t a t i s t i c a l bias in equation (1) can be illustrated as followst T Q = f ( x i f X g , Xy C) + Q u But, G =» ^ , where Q n and are the number of new houses sold and unsold, respectively, and S i s the total stock of 'housing, Q + Q n e and,- T Q = g , where Q g i s the number of existing houses sold. Q Q Q Q Therefore, ^ + constant + ^ + + f| x^ + ((g + ^ ) Since both sides of the equation retain the same component, - , i t i s - 53 -expected t h a t T Q w i l l be h i g h l y c o r r e l a t e d w i t h G and the f i t n e s s o f the r e g r e s s i o n equation w i l l be enhanced. Hence, i t i s not s u r p r i s i n g t h a t the r e g r e s s i o n r e s u l t s of the o v e r a l l turnover r a t e (T ) are s u p e r i o r t o t h a t of the turnover r a t e of e x i s t i n g houses ( T q ) . By removing Q Q from both s i d e s of the equation, i t i s hoped t h a t the s t a t i s t i c a l b i a s w i l l be reduced. However, under t h i s new d e f i n i t i o n , there may be a b i a s i n the sampling. The m a j o r i t y of the s a l e s of e x i s t i n g houses were expected t o be i n the slow growth area ( i . e . Sub-area: i ) , and t h e r e f o r e , T may © be l e s s r e p r e s e n t a t i v e of the r a p i d growth area. From the purpose of con s i s t e n c y , the same lagged v a r i a b l e s as i n equation (3) a r e used i n t h i s model. The estimated r e g r e s s i o n equation i s as f o l l o w s j V - .0311 + .003? A H t _ 1 - .0003 A Y t - . 0 0 7 l L V t (3.350) (0 . 4 3 5 ) (1.424) + ,0064R/P t - .282lrm t_ 1 + ,00002A t (1.677) (4.227) (2.28?) + . 2 6 6 2(rm-rb) t_ 1 (1.419) " ( 1 3 ) R 2 - 0.717^ D.W. » 1.3875 SER - 0 .0033 - 54 -At 95% level of s i g n i f i ^ h c e , -only the coefficients of AH, rm and A are significantly different from zero, and R / P and rm-rb have the correct sighs, though s t a t i s t i c a l l y insignificant. AY and LV remain insignificant. These s t a t i s t i c a l results are very much i n line with those of equation (3) i n terms of significance tests. However, i n terms of fitness of the equation and presence of s e r i a l correlation, i t i s i n f e r i o r to equation ( 3 ) . Only 7 l . ? 4 % of the variation i n turnover rates i s accounted for by this regression equatin as against 75.97% by the other. The Durbin-Watson s t a t i s t i c (D.W.) i s just within the inconclusive region at 95% level of significance. There appears to be no improvement i n terms of multi-collinearity. The same partial correlations between the indepen-dent variables are present, (see Table 3.3) , Nevertheless the step-wise regression i s performed and the results are given i n Table 3 . 4 . Equation (17) appears to be the best model. A l l the independent variables (AH, rm, R / P , A) are s t a t i s t i c a l l y significant. 2 After equation (17)» R does not increase significantly and i t has the highest D.W.(1.3508). Unfortunately, at 95% level of significance, the D.W;. s t a t i s t i c indicates that positive s e r i a l correlation exists. Morever rm i s highly collinear with R / P ( .26626). We can therefore conclude" that the s t a t i s t i c a l results obtained from the regression of turnover rate of existing houses (T ) are certainly much inferior to those obtained from the regression of overall turnover rate (T ). For this reason, i t i s suggested that the latte r be used for pre-e dictive purposes, given i t s s t a t i s t i c a l limitations. Besides the s t a t i s t i c a l limitations, certain implicit assumptions were made i n constructing the models,sandSthese assumptions further l i m i t TABLE 3 . 3 CORRELATION MATRIX OF T T AH e T Q 1.0000 0.4225 H 0.4225 1.0000 Y 0.0685 -0.1700 LV -0.0200 0.0230 R/P 0.5849 0.1082 rm -0.6828 -0.0921 A 0.2363 0.0185 rm-rb 0.2132 0.3002 AY LV R/P 0.0685 -0.0200 0.5849 -0.1700 0.0230 0.1082 1.0000 0.0212 0.2796 0.0212 1.0000 -0.0885 0.2796 -0.0885 1.0000 -0.0530 0.1807 -O.6626 0.1281 0.5728 -0.1314 0.2059 0.4460 0.0828 rm A rm-rb -0.6828 0.2363 0.2132 -0.0921 0.0185 O .3OO2 -0.0530 0.12.81 0.2059 0.1808 0.5728 0.4460 -O.6626 -0.1314 0.0828 1.0000 0.0283 0.2511 0.0283 1.0000 0.5347 0.2511 0.5347 1.0000 TABLE 3 . 4 - STEP-WISE REGRESSION OF T EQU. NO. CONST ^H t_ 1 r m ^ R/Pt A t ( r m - r b ) ^ AY . LV R 2 D.W. SER 14 .0214 ( 3 ° ° J 7 ) 0.1787 0.7389 0.0053 15 .0^23 ( 4 ° ? ^ ) l^k%z) 0,5968 u m z °'0038 16. .03365 ^ ( ^ ° 3 ) 0.6206 1.1341 0.0037 n o o k .0044 - .2520 . 0080 . 00002 n A q M O ^ Q a n f m 17. .0274 ( 4 > 4 ? 3 ) ( i + # ? 3 l ) ( 2 t ^ 3 ) ( 3 < 5 9 4 ) 0.6961 1.3508 0 .0033 18 0300 ' ° 0 3 8 -•• 3 °° 5 , 0 ° 5 5 - 00002 . 2350 • 0 70S4 1 426? 0 0<m 18. .0300 ( 3 > 5 8 1 ) (4.594) (1.498) (1.828) (I .265) 0 , 7 0 5 4 1 , 4 2 6 2 0 , 0 0 3 3 . n ? Q 7 7 .0037 -.2988 .0055 .00002 .2402 -.0002 0 0 0 1 1108 0 19. .02977 ( 3 < 3 8 0 ) ( 4 < 5 0 3 ) ( 1 > 4 9 7 ) ( 1 < 8 2 6 ) ( l t 2 ? 3 ) ( t 2 5 2 4 ) 0.7057 1.3308 0 .0033 on - 0 ° 3 7 -V2821 .0064 .00002 .2662 - .0003 -.0071 n 7 1 7 4 ^o 7. n n f m 20. .03113 ( 3 # 3 5 0 ) ( 4 > 2 2 ? ) ( 1 > 6 7 ? ) ( 2 , 2 8 7 ) ( 1 # 2 , 1 8 ) ( , 4 3 5 ) ( 1 > 4 2 4 ) 0.7174 1.3875 0.0033 1 1 - 57 -the practicality of them. There are two major areas which need to be , " elaborated upon. The f i r s t stems from the heterogeneity of the housing commodity. Within the whole Metropolitan area, or even the sub-areas, the housing market i s made of a diverse range of dwellings i n terms of designs, age, location, etc. By aggregating them into a single measure, turnover rate (number of sales divided by number of stock in the same period), we have in fact assigned equal weight to the dwellings, i.e. we have ignored the qualitative differences between Ihem. We hope to take some of the qua-l i t a t i v e differences into account i n the next chapter. The other major problem i s the adjustment process of supply and demand i n the housing market. Because' of imperfect information flow, high cost of production and the great length of time required to increase or reduce supply, the whole system i s plagued by unstable forces result-ing lags, inertias or acceleration. I t i s virtually impossible to iden-t i f y the magnitude and timing of these forces l e t alone to quantify them. In this model we only assume a one-period lag. Footnotesi 1. Dennis J. Aigner, (43), p.133. 2. G.E.V. Lesser, (42), p.28. " - 58 -CHAPTER IV QUALITATIVE ANALYSIS OF THE EMPIRICAL DATA As noted previously, the quantitative tests are not entirely con-clusive; so i n this chapter a qualitative analysis of certain more important factors and of certain interested aspects not dealt with in the quantitative analysis i s made. More specifically, the l a t t e r refers to the effects of business cycles, certain policy implications and the seasonal fluctuations of turnover. Seasonal Fluctuations , i n Turnover Graph 1 shows very clearly that the movement of turnover rates was highly seasonal over the entire period studied. (Data on turnover rates used for the quantitative tests were seasonally adjusted) The f i r s t and the last quarters of every year were consistently lower than the second and third. The seasonal pattern i s further amplified when we examine Chart II which gives the distribution of the aggregate monthly sales i n terms of number and value not broken down by year, over the entirelperiod (i.e. for example, January i s equaled to the sum of a l l January months from 1949 to 1963). The aggregate monthly sales are also subdivided into the different subareas as defined i n Chapter I. To be exact, the months with the lowest number of sales were the Winter months of December, January and February, then followed by November, October and March. This i s true of the overall area as well as the sub-areas. The highest months were June, July and August with the exception - 59 -60 -CHART II - Distribution of Aggregate Monthly Sales in number and value number value Graph 2 -Overall Area Graph 3 -Subarea III Graph 4 -Subarea II Graph 5 -Subarea I J F M A M J J A S O N D - 61 -of Subarea I which had a unusually high number of sales in September. This bias could probably be attributable only to a couple of abnormal months of September in the entire period. As shown in Chart II, this seasonal pattern is true in terms of number as well as value, for a l l the four areas. There were, however, more months in which the percentage of sales in value terms were lower than in number, indicating that in these months the houses sold were in the lower value range. In fact, there were only four months in which sales in terms of value exceeded that In number, with the exception of of Sub-area I which had five months. Chart III takes into account the age differences between Ihe houses. They were grouped into three age brackets« below 16 years, 16-30 years and over 30 years. Although the graphs exhibit more variations than those in Chart II, the general seasonal pattern is similar. Among the three age groups, the 0-15 year group was the most consistent in terms of seasonal fluctuations. This same group accounted for more sales than the sum of the other, with the exception of Subarea I, and therefore the more impor-tant component of the total volume of sales (see Table 4.1). While Sub-area I was an exception, i t had the most consistent pattern among the three subareas and thus should not be responsible for any distortion. Anyway, i t is reasonable to expect Subarea I which was classified as the mature (or slow growth) area to trade more heavily in older houses. The same seasonal pattern was repeated when differences in value were also taken into consideration (see Table 4.2) and will therefore not be elaborated upon here. - 62 -- 63 -TABLE 4.1 - SAMPLE DISTRIBUTION OF SALES BY AGE SUBAREA I SUBAREA II SUBAREA III OVERALL YEAR , 0 - I 5 I 6 - 3 0 3 1 + 0 - 1 5 16 -30 3 1 + 0 - 1 5 16 -30 3 1 + 0 - 1 5 16 -30 31+ JAN 5 6 3 9 6 5 9 4 37 2 2 § 8 9 41 19 2 3 9 1 1 ? 1 0 6 1EB 5 4 41 7 0 1 2 7 39 2 7 1 1 0 2 5 1 2 2 9 1 1 0 5 1 0 9 MAR 61 6 9 104 1 3 0 5 7 2 5 164 3 8 1 3 3 5 5 164 142 APR 7 2 5 9 8 7 1 7 3 6 6 3 6 167 4 9 1 7 412 1 ? 4 140 MAY 81 6 9 9 0 128 6 8 3 1 149. 3 ? 24 3 5 8 1 ? 4 145 JUNE 82 6 5 104 181 6 6 k5 176 5 5 1 7 4 3 9 186 166 JULY 7 6 6 2 1 1 5 140 4 3 3 9 188 4 3 2 2 404 148 I 7 6 AUG 7 8 48 100 1 7 7 5 0 4 4 194 5 0 18 449 148 I 6 5 SEPT 7 7 7 7 1 1 3 1 3 5 5 6 41 1 5 5 4 3 24 3 6 7 1 ? 6 I 7 8 OCT 7 1 49 84 134 7 5 40 166 3 9 18 3 7 1 163 V*2 NOV 5 8 48 97 137 5 9 42 142 5 3 16 337 1 6 0 I 5 5 DEC ¥> 3 6 5 6 1 2 5 3 5 37 117 3 2 14 288 1 0 3 1 0 7 Source: U.B.C. Sample Housing Survey, I 9 6 7 . - 64 -TABLE 4.2 SAMPLE DISTRIBUTION OF SALES BY VALUE, OVERALL AREA PERIOD LOW-RANGE MID-RANGE HIGH-RANGE JAN 3 5 8 8 6 18 FEB 411 7 4 2 0 MAR 516 1 2 5 2 0 APR 571 1 2 5 3 0 MAY 540 116 2 1 JUN 616 144 3 1 JUL 5 8 3 114 3 1 AUG 5 6 3 1 5 7 42 SEPT 5 6 2 1 2 6 3 3 OCT 5 1 9 1 2 2 3 5 NOV 4 9 6 124 3 2 DEC 3 7 4 1 0 0 24 Notet The c r i t i c a l value i s the upper li m i t of the price range for the relevant year and i s calculated by the following formulaj Average of a l l Sales under $X The following are the c r i t i c a l values of the respective yearsj 1949 1 9 5 0 1951 1 9 5 2 1 9 5 3 1 9 5 4 1 9 5 5 1956 Low Range(xl00$) 60 80 86 77 90 67 8~5 I4"2 Mid Range(xl00$) 100 1 1 3 124 I30 145 I56 I75 200 High Range(xl00$) - open ended -1 9 5 7 1958 1 9 5 9 I960 1961 1962 1963 Low Range(xl00$) 1 S 9 I 7 6 " I 8 2 I 9 5 173" IBO IBT Mid Range(xlOO$) 2 3 2 228 230 2 5 0 245 2 4 5 2 5 5 High Range(xl00$) - open ended -- 65 -There are obvious reasons to account for the seasonal fluctuations i n turnover. The unfavourable weather conditions in the Winter months certainly detered potential purchasers from moving and thus brought the number of sales down. The high volume of sales during the Summer months could be attributed f i r s t l y to the favourable weather conditions and secondly, to the fact that June to August were the vacation months for school children. During these months school children need not be inter-rupted in the course of their study and in addition, parents had more time to look for houses and to arrange to move house. (We have assumed most purchasers of single-family houses to have children.) An important digression i s called for at this juncture. The st a t i s t i c s show that more than half the to t a l number of sales were in the 0-15 year age group (refer Table 4 . 1 ) . Since new houses also f a l l within this category, there i s a high probability that new houses actual-ly accounted for a large proportion of the total sales. This somewhat confirms our original hypothesis that new houses (or completions) were an important component of turnover rates and reinforces the conclusions of our quantitative tests. Another important point to note i s that more than half of the number of houses sold were in the low value range (see Table 4 . 2 ) . I t i s therefore not unreasonable to assume that a substantiate portion of the total sales within this value range came from relatively young families who did not have the income capacity to purchase houses in the higher value range. The lower income households were thus the most i n -fluential among the purchasers in terms of the proportion of share i n the total volume of sales or turnover rates. This perhaps reinforces - 66 -our ea r l i e r contention in the chapter on quantitative analysis that the dependent variable (i.e. turnover rates) i t s e l f has already contained a great deal of information on the income position of the purchasers and therefore explains why the income variable was s t a t i s t i c a l l y insignificant. Turnover, Credit Conditions, Business Cycles and Monetary Policy Graph 8 (in Chart IV) records the movement of the seasonally adjusted turnover rates. It shows a considerable amount of c y c l i c a l fluctuations and that after 1958 there was a distinct f a l l in the turnover rates. This c y c l i c a l pattern i s believed to be correlated with the cost and av a i l a b i l i t y of mortgage funds, which in turn are very much the result of changes i n the general business conditions and monetary and f i s c a l policy. Although this i s a regional study, credit and business conditions in Vancouver, as a functional component in Canada's urban economy, are very much conditional by national economic factors such as business cycles and federal monetary and f i s c a l policy. For this reason, national s t a t i s t i c s are applicable in this section of the study. To il l u s t r a t e the relationships between these variables, a year-to-year commentary over the whole period of 19^9-I963 i s made. For this purpose, Graphs 8 , 14 (mortgage interest rate) and 15 (interest differential between long-term bond and conventional mortgages) in Chart IV and Chart V are used as illustrations. The movement of mortgage interest rates not only indicates changes in-:the cost of mortgage financing but also changes in the mortgage terms. It has been established in Chapter III that the interest rate variable i s CHART V MONEY SUPPLY AND INTEREST RATES Sourcei Report of the Royal Commission on Banking and Finance, Ottawa, 1964. - 69 -a sufficiently good proxy for loan-to-value ratio (LV) and length of ma-turity (A ) . A rise or f a l l in interest rates w i l l probably signal a corresponding tightening or easing of borrowing terms. The a v a i l a b i l i t y of mortgage funds'is represented by the interest di f f e r e n t i a l (rm - rb) graph. Basing on the theory of supply of funds expanded in Chapter I I , i t i s expected that a v a i l a b i l i t y of mortgage credit w i l l increase as interest differential widens. Chart V shows the results of government's monetary policy over the studied period, and Table 4 . 3 gives detail on the business cycles speci-fying the dates of the turning points and the alternating periods of ex-pansion and contraction. The recession phases of the cycles are shaded i n a l l the graphs. TABLE 4 . 3 BUSINESS CYCLE REFERENCE DATES, 1949-63 YEAR QUARTER MONTM1 : : TURNING POINT TYPE AND NO. OF PHASE BEGUN 1949 I I I SEPTEMBER TROUGH EXPANSION 1 1953 I I MAY PEAK CONTRACTION 1 1954 I I JUNE TROUGH EXPANSION 2 1957 I I APRIL PEAK CONTRACTION 2 1958 I I A P R I L TROUGH EXPANSION 3 I960 I JANUARY PEAK CONTRACTION 3 1961 I MARCH TROUGH EXPANSION 4 Source1 Report of the Royal Commission on Banking and Finance, Ottawa, "1964, p.402. ~ ~ - 70 -19^9 to 1950 - During this period the Canadian Economy was s t i l l reco-vering from the aftermath of the War. There was a backlog of domestic demand in a l l sectors and a high degree of l i q u i d i t y brought about by heavy war-time accumulation of l i q u i d assets. Monetary controls were not used during the early postwar years. Bank rate was set at a low level and rate on government long-term bonds were kept between 2 . 6 $ to 3%. With no monetary restraint, high l i q u i d i t y , mortgage credit was readily a v a i l -able, the net result of which was a strong effective demand for housing as indicated by the relatively high turnover rates. 1951 to mid-1952 - After late 1950 a policy of monetary restraint was adopted. Bank rate was raised and banks were given directives to curb their lending a c t i v i t i e s . This tight-money policy was introduced because the outbreak of the Korean War was threatening inflation. Interest rates were high and there was an unprecedented demand for capital funds by the corporate sector. However, despite stringent credit terms, the supply of mortgage funds was not reduced. The slackened demand for housing was the result of soaring construction costs i n the later part of 1950 and early 1951 rather than the lack of funds. In 1951 consumer price index rose over 10$ and construction costs over 15$. Mid-1952 to 1955 - By 1952 the inflationary pressures had somewhat ease and an increase in the demand for housing followed. The strength of demand was strong enough to restore interest rate to 6% by early 1953. Business activity reached a new peak i n the second quarter of 1953» after, which the economy lapsed into the f i r s t real post-war recession and a policy of ease was pursued well into 1955 when the investment boom had - n -already picked up momentum. Bank rate was actually lowered i n February of 1955. A general f a l l i n corporate demand for capital funds during the recessionary period increased the avai l a b i l i t y of mortgage funds substan-t i a l l y . Hence, despite high conventional mortgage rate, demand for housing remained unfaltered. I956 to Mid-1957 - The economy moved into another period of expansion beginning in late 1954 and reached to a peak in the second quarter of 1957. There was heavy demand for capital funds from a l l sectors of the economy. Credit conditions were subsequently tightened and a policy of monetary restraint was pursued. Bank rate was raised several times u n t i l i n November 1956 i t was set at a rate of •§• of 1% above the price of 91 day treasury b i l l s . Despite governmental efforts, the strain on the capital market caused a sharp decline in the ava i l a b i l i t y of mortgage funds and a consequential f a l l i n demand for housing. Though turnover rates climbed above the 0.04 mark in the f i r s t quarter of 4956, the effects of the credit squeeze were f e l t thereafter u n t i l the next period of busi-ness recovery as witnessed by the sharp decline i n turnover rates. Mid-1957 to Mid-1958 - By April of 1958 the strain on the national re-sources was f i n a l l y f e l t and the economy moved over the peak into another period of contraction. Monetary conditions were subsequently eased in the later part of 1957. As yields on long-term bonds and corporate demand for funds declined, a v a i l a b i l i t y of mortgage funds increased. During this period CMHC was pm an a l l out effort to stimulate demand by engaging i n . direct lending and easing borrowing terms. However, these efforts became - 72 -effective only after the recessionary months. Turnover began to move upwards only i n the f i r s t quarter of 1 9 5 8 • Mid-1958 to i960 - This period i s best described by the -Report of the  Royal Commission on Banking and Finance, 1 9 6 4 as follows j " influenced by continuing price increases and fears of inflation, the monetary authorities i n late 1958 v i r t u a l l y froze the money supply despite the fact that the recovery then underway was only in i t s early stages and had shown few signs of strength..... .this policy of vigorous restraint contibuted to the fact the 1958 -60 expansion was the weakest 1 since the end of the war " It was only after the recession had started i n i960, that money supply was expanded. During this period we witnessed considerable d i f f i c u l t i e s i n the financial market. The tight-money situation was further accentuated by the ill-timed conversion loan of July-September 1958, "which at one-step lengthed the maturity date of k0% of the outstanding marketable debt;" This was res-posible for the sharp rise in interest rate to 7>5% in early i960 because the public's holding of more li q u i d assets was reduced substantially. The combined effect of these factors brought about a steady decline in the turnover rates which began as early as the end of 1 9 5 8 and conti-nued even after the recession. I961 to I963 - During this period the monetary authorities practiced a - 73 -policy of ease, except i n mid-1962 when i t was interrupted temporarily "by the foreign exchange c r i s i s . Money supply was increased and bank rate was lowered to stimulate demand. Turnover rates did eventually rose but remained relatively lower than the pre-1958 periods. There were obviously factors, other than credit influences, which were responsible to this 3 unusual behaviour. Turnover and other Explanatory Variables In general, the business cycles, the alternating periods of monetary ease and restraint and the consequential changes in the cost and availa-b i l i t y of mortgage funds more or less explain the c y c l i c a l fluctuations i n turnover rates. However, these events alone could not substantiate the downward trend, especially after 1958. We therefore have to examine the other explanatory variables ( i . e . changes i n households, rent-to-price ratio and completion ratio) f o r an explanation of this phenomenon. The graph on changes in the number of household heads (Graph 9, Chart IV) measures the need as opposed to the desire for housing. Before 1958, there were periods of ups and downs, fluctuating between a low .25% to a high 1.8% and running almost parrel to the movement of the turnover rates. However, after 1958, there was a definite decline in the rate of change and as a result the need for housing also f e l l . This therefore partially explains the downward trend in turnover rates, especially after 1958. Completion ratio also declined sharply as from the end of I958 and did not recovered thereafter (see Graph 13, Chart IV). Graph 12 shows the movement of the rent-to-price ratios which was - ?4 -on a downward trend throughout the second half of the studied period, indicating a sh i f t against home-ownership. Another point of view i s that the homeowners had a low propensity to move, i.e., they were reluctant to s e l l their houses. Shelton*s study (refer to Chapters?) reveals that i t costs less to own than to rent a house i f a person lives in i t for 5 years or more. I t was therefore possible that as the years went by, more and more people became less inclined to s e l l their houses, other things taken to be constant. Hence, sales of existing houses accounted for less and less of the turnover rates. (Note that turnover rate i s more sensitive to of a change in the number/existing houses than new houses because the la t t e r increase the denominator (stock of housing) and the former do not). The combined effect of homeowners' reluctance to s e l l , and a f a l l in the need for housing and ava i l a b i l i t y of new houses i s a f a l l i n g turnover rate. Changes in real per capita income (Graph 10) was the only variable which moved upwards in the second half of the studied period, though the fluctuations i n the f i r s t half more or less coincided with that of the turnover rates. Perhaps the income change (increase in this case) was not strong enough to offset the influence of the other factors. In conclusion, we note that the va r i a b i l i t y i n the turnover rates was the result of a combination of a number of factors. Any attempt to explain this v a r i a b i l i t y on the basis of only one or two factors may lead to some dangerously biased conclusions, which we have taken care to avoid in the course of this study. Footnotes» 1. Report of the Royal Commission on Banking and Finance, Ottawa, 1964. p.<+12 - 75 -2. Ibid. p.414. 3. A l l information contained in the year-to-year commentary was from Report of the Royal Gimmission on Banking and Finance, Chapters 14 and 20 • L.B. Smith's The Postwar Canadian Housing and Residential Mortgage Markets and the Role of the Government, pp.58-65; and E.Sussman's Role of Mortgage Banking in the Canadian Economy, Chapter 4. CHAPTER V SUMMARY AND CONCLUSIONS This study has shown that over the period 1949-1963, the turnover rates of single-family houses fluctuated c y c l i c a l l y and that there was a definite downward trend, especially after 1958. This thesis i s basi* cally an attempt to explain this phenomenon "by quantitative as well as qualitative analyses. Essentially, two regression models were constructed to explain the functional relationship between turnover rate and the various explanatory variables. The f i r s t model uses turnover rate of existing and new houses as the dependent variable and the second uses turnover rate of only exist-ing houses. Unfortunately, due to the presence of serious multicollinear-it y and s e r i a l correlation, the practicality of the second model i s in doubt. However, the f i r s t model, which includes the explanatory variables of change in the number of family heads, mortgage interest rate, interest differential between mortgage and long-term bond and completion ratio, i s comparatively free from multicollinearity and s e r i a l correlation, and. could be used with certain reservations for predictive purposes as well as for explaining the structural relationships. As the regression analysis has not been entirely conclusive and com-prehensive, i t i s supplemented with qualitative considerations with the aid of graphical illustrations. It has been found that the cy c l i c a l fluctuations more or less coincided with the business cycles and the alternating periods of credit ease and restraint. The downward trend - 77 -i s probably explained by the movements of the;other variables, such as change in the number of family heads, rent-to«price ratio, and comple-tion ratio. A l l of them show a definite downward trend in the second half of the period studied. Another interesting finding of this study i s that the fluctuations in turnover rate was highly seasonal. The number of sales of houses were consistently low i n the Winter months of December, January and February and high in the Summer months of June, July and August. This seasonal pattern was true of the whole of Metropolitan Vancouver as well as the sub-areas. Finally, i t i s hoped that further research w i l l be undertaken to study this hitherto neglected dimension of housing. The obvious conti-nuation of this study i s an updating of the data used to study total turnover of housing. This w i l l be particularly interesting because in the post-1963 period, there have been a number of major changes in the housing market in Metropolitan Vancouver. These includet a) the trend to high density residential development and condominium housing (since I968), b) changes in the Income Tax Act (1972) which substantiately altered tax position with respect to real property and owner occupied housing, and c) the implementation of rent control in 1974. - 78 -BILIOGRAPHY 1. Aigner, D.J., Bsisic Econometrics, Prentice-Hall Series in Mathe-matical Economics, 1971. 2. Alberts, W.W., "Business Cycles, Residential Construction Cycles and the Mortgage Market", Journal of P o l i t i c a l Economy, Vol. 70, June 1962, pp. 263-281. 3. Atkinson, L.J., "Factors in the Housing Market", Survey of Current  Business, Ap r i l i 9 6 0 . 4. Abrams, C., "Credit Terms and Effective Demand for New Housing", Study of Mortgage Credit, Committee on Banking and Currency Sub-committee on Housing U.S. Senate, 85th Congress, 2nd Session, Dec, 1958, pp. 81-86. 5. Boardssof Governors of the Federal Reserve System, "Ways to Moderate Fluctuations i n the Construction of Housing", Federal Reserve Bulletin, March 1972, pp. 215-225. 6 . Carliner, G., "Determinants of Home Ownership", Land Economics, Vol. L, No.2, May 1974. 7. Fisher, E.M., Urban Real Estate Markets} Characteristics and Financing, National Bureau of Economic Research, N.Y., 195L 8. Frpmm, g., "Econometric Models of the Residential Construction cSector: A Comparison" i n R.R.Ricks, ed., National Housing Models, Lexington Books, 1973. 9 . Greater Vancouver Regional Di s t r i c t , The Housing Issue, 1973• 10. Grebler, L., Housing Issues in Economic Stabilization Policy, Nation-a l Bureau of Economic Research, I960. 11. Grebler, L. & Maisel, S.J., "Determinants of Residential Construction A Review of Present Knowledge", Impacts of Monetary Policy, Prentice-H i l l , N.Y., 1963. i>2. Guttentag, J.M., "The Short Cycle in Residential Construction, 1946-1959", The American Economic Review, Vol. LI, June 1961. 1*3. Hempel, D.J., A Comparative Study of the Home Buying Process i n Two  Connecticut'Housing Markets, Center f o r Real Estate and Urban Economic Studies, University of Connecticut, 1970. 14. Huang, D.S., "Effect of Different Policies on Housing Demand" in I. Friend, ed., Study of the Saving and Loan Industry, Federal Home Loan Bank Board, Washington, D.C., 1969, pp 1211-1240. - 79 -15. Jarrett, R.F.S., "The Residential Mortgage Market in Canada", Bank of Canada Review, May 1974, pp. 3-19. 16. Klaman, S.B., "The Availability of Residential Mortgage Credit" in Study of Mortgage Credit, Op. Cit., pp.189-209. 17. Lansing, J.B., Residential Location and Urban Mobility1 The Second Wave of Interview, Ann Arborj University of Michigan, Survey Research Centre, Institute for Social Research, I966. 18. Lee, T.H., "The Stock Demand E l a s t i c i t i e s of Non-Farm Housing", Review of Economics and Statistics, Vol. L I U , Feb. 1971. 19. Lesser, C.E.V., Econometric Techniques and Problems, John Wiley & Sons, 1970. 20. Leeuw de F.,."The Demand for Housing! A Review of Cross-Section .Evidence", The--Review-of-Economics and Statistics, Vol. LIII, Feb. 1971. 21. Maisel, S.J., "A Theory of Fluctuations in Residential Construction Starts", American Economic Review, Vol. LIII, June 1963, pp. 359-383. 22. , Financing Real Estate, McGraw-Hill, I965. 23. » "Rates of Ownership, Mobility, and Purchase" in Essays in Urban Land Economics, University of California, Los Angeles, 1966, pp. 76-108. 24. Meltzer, A.H., "The Building Industry and Business Cycle", The  Journal of Finance, Vol. 29, June 1974. 25. Morton, W.A., Housing Taxation, Madison! University of Wisconsin Press, 1955. pp.42-50. 26. Muth, R.F., "The Demand for Housing", i n A.C. Harberger, ed., The Demand for Durable Goods, University of Chicago Press, Chicago, 111. I960. 27. Needleman, L., Economics of Housing, Staples Press, I965, London. 28. Reid, M., Housing and Income, University of Chicago,Press, 1962. 29. Rossi, P.H., Why Family Move! A Study i n the Social Psychology of  Urban Residential Mobility, Free Press, N.Y., 1955 30. Shelton, J.P., "The Cost of Renting Versus Owning a Home," Land  Economics, Vol. 44, Feb.1968, pp. 60-72 9 - 80 -31 . Silver, Irving, "A Model of Housing Demand in Metropolitan Areas" in J.J. Sullivan, ed., Explorations i n Urban Land Economics, Johns C. Lincoln Institute, University of Hartford, 1970. 3 2 . Smith, L.B. Postwar Canadian Housing and Residential Mortgage  Market and the Role of the Government, University of Toronto Press 33. , "Postwar Canadian Housing Policy i n Theory and Practice", '"Land Economics, Vol. 4 4 . , Aug. I 9 6 8 , pp. 339-349. 34. , "On the Economic Implications of the Yeild Ceiling on Government-Insured Mortgages", Canadian Journal of Economics and  P o l i t i c a l Science, Vol. 33 , Aug. 196? pp. 420-431. 35. , "The Canadian Housing and Mortgage Markets", Bank of Canadian"Research Studies, No. 6 , 1970. 36. Smith,W.L., "The Impact of Monetary Policy on Residential Constru-ction" i n Study of Mortgage Credit, Op. Cit. 37. Smith, Wallace, ''Housingt The Social and Economic Elements, u University of California Press, 1970. 38. Sparks, G.R., "A Econometric Analysis of the Role of Financial Intermediaries i n Postwar Residential BuildingffCycles" in.R. Ferber, ed., Determinants of Investment Behaviour, National Bureau of Economic Research, 196?. 39. Sussman, E., The Role of Mortgage Banking i n the Canadian Economy, Ph. D. Thesis, Dept. of Economics and P o l i t i c a l Science, McGill University, I963. 40. Thorncroft, M., Principles of Estate Management, The Estate Gazette Ltd., London, 41. University of Bri t i s h Columbia, Faculty of Commerce and Business Adminstration, Vancouver Housing Market, I 9 6 6 - I 9 6 8 , Joint Report of the Seminar in Governmental Urban Land Policies, May, 1968. 42. Winger, A.R. "Housing and Income", Western Economic Journal, Vol. $fcfJiitoea 1908v8.ppv.2£6?&&* 4 3 . ,. , "Trade-Offs in Housing", Land Economics, Vol. 4 5 , Nov. 1969, pp.413-417. 4 4 . Winnic'k, L., "Housing! Has There been a Downward Shift i n Consumer Preference?", Quarterly Journal of Economics, Vol. LXIX, Feb. 1955, PP. 57-58. - 81 -&5> Wonnacott, R.J. & Wonnacott, T.H., Econometrics, John Wiley & Sons, 1970. 46. Wood, R., "Credit Terms and Demand for Residential Construction" i n Study of Mortgage Credit, Op. Cit. - 82 -APPENDIX I Sources of Data 1. Turnover Rates (T and T ) o e A l l data on turnover rates were obtained from the housing sample survey as explained in Chapter 1 and were computated in accordance with the for-mulae given i n footnote 3 of Chapter i . 2. Families headed by 25-44 year old males ( A H ) Population census figures are available for the years 1951» 1956, 1966 f o r the whole of Metropolitan Vancouver and annual estimates of provincial total are also available. Provincial data on the number of married males with age-group breakdowns are available i n D.B.S.'s annual publication Population Estimates by Marital Status, Age and Sex for Canada and Provinces. We assume that a l l families are headed by married males. The data used in the regression analysis are calculated by the following formulas Number of Married Males (25-44) Population in Metro. Van, in t in B.C. i n Period t K Population i n B.C. i n t In accordance with this formula we assume that the proportions of growth in each period are the same in Metropolitan Vancouver as i n B.C. 3. Real Per Capita Disposable Personal Income (AY) Data on per Capita disposable income in B.C. were obtained from D?B;S.'s publication National Income and Expenditure. The series was deflated by the consumer price index for Vancouver appearing in Prices and Price Indexes, D.B.S. - 83 -4. Interest Differential (rm-rb) Interest differential data were obtained from Macleod, Young and  Weir Series. The bond rate (rb) is a weighted average of a l l long-term bonds. 5. The Rent/Price Ratio (R/P) This ratio is calculated by dividing a rent index by a price index with last quarter of 19^ 9 equal to 100 for both indices. The rent index is the rent component of the consumer price index for Vancouver. This unpublished series was made available by D.B.S. The housing price index is another result of the sample survey which yielded the turnover data. 6. Loan-to-Value Ratio (L/V), Length of Maturity (A), Mortgage Rate (rm) Data for these three variables are also derived from the sample survey. The figures for L/V and A are median values. 7. The Completions Ratio Number of completions! The "single completions" figures were obtained from CMHC for each municipality in metropolitan Vancouver for the years 1951-1963• "Single" here includes semi-detached and duplex units as well as single detached, but the latter make up a l l but a very small percentage of the total. Precise data for earlier years were not available, but, on the basis of incomplete data in the Dominion Bureau of Statistics (DBS) periodical New Residential Construction; we made crude approximations for metropolitan Vancouver though not for the - 84 -subareas. Stock of single-family housesj In constructing the sample for calculating the turnover ratio',,, the number of single-family dwelling units i n each subarea was obtained from the assessment r o l l s for the end of 1963• Th e number of units i n the 1963 sample was calculated in accordance with the formula given ln footnote 2 of Chapter 1. As the years recede, the sample size shrinks, since some houses in existence at the end of I963 were not in existence in earlier years. The hous-ing stock in any earlier year t was calculated using the'formulat Housing stock in year t Sample size in year t Sample size i n I963 x Housing stock i n 1963 This formula assumes that the sample in each year represents the same proportion of the housing stock in that year. APPENDIX I I DATA USED I N REGRESSION ANALYSIS PERIOD (QTRS) T AH rm rm-rb C R/P A LV AY 19^9 4 .0398 1.3591 .0500 .0115 .0721 1.000 119.0 .440 - .289 1950 1 2 3 •V .0307 .0421 .0383 .0358 .4425 ^3911 :'-*fiB20 .0500 .0500 .0500 .0450 .0122 .0122 .0122 .0122 .0716 .0631 .0632 .0545 .965 .932 .905 . 8 ? ! 150.0 239.0 120.0 180.0 .630 .490 .570 .520 1.064 • 957 .758 .941 1951 1 2 3 k ' .0397 .0320 .0368 .0362 1.3905 1.5209 1.4948 1.3129 .0500 .0500 .0550 .0600 .0196 .0163 .0172 .0159 .0529 .0445 .0425 .0356 .874 .878 .,.882 .886 U9.5 114.0 81.0 60.0 .550 .570 .520 .410 - .180 :- .280 - .187 - .094 1952 1 2 3 .0344 .0301 .0320 .0346 • 9852; .8650" ,'7134 .5611 .0600 .0600 .0550 .0575 .0133 .0130 .0152 .0140 .0373 .0373 .0349 .0317 .895 .905 .914 .922 120.0 87.5 120.0 150.0 .545 .455 .520 .490 . 188 .375 .561 .836 1953 1 2 3 .0311 .0342 .0366 .0368 .3976 .3512 .4198 .5515 .0600 .0525 .0600 .0600 .0152 .0151 .0150 .0160 .0367 .0366 .041? .0410 .930 .935 .941 .950 60.0 240.0 71.0 120.0 .490 .705 .485 .480 1.382 1.364 1.525 1.325 1954 i 2 3 .0319 .0319 .0361 .0361 .?820 .6879 .5462 .4033 .0600 .0600 .0550 .0550 .0170 .0315 .0225 .0229 .0440 .0407 .043? .0405 .958 .965 .970 • 975 132.5 180.0 240.0 300.0 • 755 .540 .6?0 .810 0.000 0.000 0.000 00.000 1955 i 2 3 4 .0335 .0326 .0324 .0339 ,Q^|80 .4038 .2682 .2647 .0600 .0550 .0500 .0500 .0230 .0209 .0209 .0213 .0462 .0453 .0510 .0491 .940 .900 .860 .890 240.0 300.0 300.0 300.0 .?80 .795 .765 .6I5 1.307 1.37? i . 6 1 3 1.504 PERIOD (QTRS) T AH rm rm-rb C R/P A LV 1956 1 .0412 1.6574 .05620 .0201 .0513 .758 266.5 .700 1.64? 2 .040? 1.7097 .0600 .0182 .0464 .706 180.0 .550 1.457 3 .0367 1.8068 .0600 .0191 .0485 .655 240.0 .620 1.756 4 .0297 1.3803 .0600 .0162 .0462 .606 238.0 .595 1 .255 1957 1 .0322 .7062 .O65O .0147 .0462 • 585 176.5 .730 - .38? 2 .0292 • 5595 .06?5 .0157 .0419 .565 240.0 .560 - .389 3 .0296 .5549 .0675 .0152 .0440 .553 120.0 .506 - .390 4 .0316 .9722 ,0700 .0164 .0403 .541 120.0 • 530 - .549 1958 1 .0326 1.3730 .0600 .0198 .0092 .580 300.0 .795 - .473 2 .0374 1.201? .0600 .0195 .0454 .541 300.0 .760© - .396 3 .0336 .8893 .0600 .0183 .0518 .543 24© JO .640 - .47? 4 .0380 .9885 .0600 .01?0 .0505 .544 300.0 .700 - .319 1959 1 .0335 . *2625 .0600 .0161 .0539 • 535 240.0 .705 ,» /160 2 .0358 .4641 .0600 .0145 .0496 .520 240.0 • 330 .320 3 .0299 .4616 .0675 .0127 .0143 .502 240.0 .700 .470 4 .0304 .5688 .0700 .010? .0480 .489 180.0 .580 .397 i960 1 .0259J. .,5162 .0737 .0104 .0469 .485 114.0 .620 - .079 2 .0206 .5692 .0725 .0134 .0389 .486 180.0 .600 - .079 3 .0219 .5647 .0724 .0158 .0373 .494 144.0 .570 .000 4 .0234 .3340 .0725 .0151 .0305 .504 177.5 .600 - .079 1961 1 .0208 .2526 .0725 .0135 .0301 .520 144.0 .600 .238 2 .0205 .2238 .0700 .0133 .0248 .540 87.5 .630 .384 3 .0197 .2225 .0700 .0154 .0238 .564 180.0 .650 .315 4 .0187 .3301 .0700 .0159 .0209 .581 180.0 .670 .471 1062 1 ,0222 .1252 .0700 .0162 .022? .570 148.0 .620 .781 2 .0238 .2218 .0?00 .0148 .0212 .556 180.0 .660 1.008 3 .0241 .2760 .0700 .0119 .0229 .543 180.0 .650 .844 4 .0176 .2724 .0700 .0144 .0213 .532 162.0 .650 .989 1963 1 . O f 8? .2482 .0700 .0158 .0227 .52? 165.0 .674 .756 2 .020.3 . .2189 .0700 .0154 .0209 .538 121.0 .670 .673 3 .0251 .2186 .0700 • .014? .0220 .520 180.0 .370 .743 4 .0170 .2156 .0700 .014? .0242 .517 150.0 .370 .737 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items