Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An assessment of the economic impact and modes of evaluation of research and development Schwartz, S. L. 1976

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

Item Metadata

Download

Media
831-UBC_1977_A1 S38.pdf [ 11.99MB ]
Metadata
JSON: 831-1.0094327.json
JSON-LD: 831-1.0094327-ld.json
RDF/XML (Pretty): 831-1.0094327-rdf.xml
RDF/JSON: 831-1.0094327-rdf.json
Turtle: 831-1.0094327-turtle.txt
N-Triples: 831-1.0094327-rdf-ntriples.txt
Original Record: 831-1.0094327-source.json
Full Text
831-1.0094327-fulltext.txt
Citation
831-1.0094327.ris

Full Text

AN ASSESSMENT OF THE ECONOMIC IMPACT AND MODES OF EVALUATION OF RESEARCH AND DEVELOPMENT by Sandra L. Schwartz M.A., Economics, University of Wisconsin, Madison B.A. (Honours) Economics, University of California, Berkeley A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN FACULTY OF COMMERCE i n Interdisciplinary Studies i n Commerce-Economi cs-Ecology We accept this thesis as conforming to the requi red standard The Un ive r s i t y of B r i t i s h Columbia November 1976 © SANDRA L. SCHWARTZ, 1976 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of Brit ish Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of my Department or by his representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. •Oepar-tmerr-tr-of- Faculty of Commerce and Business Administration The University of Brit ish Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 Date November 19, 1976 ABSTRACT R&D can be considered the driving force of the modern economy. The economy is organized to utilize scarce resources. R&D through techno-logical change results in a transformation of scarcities, in the creation of new resources, new products and new prices. "The central stupendous truth about developed economies today is that they can have . . . the kind and scale of resources they decide to have. . . . It is no longer resources that limit decisions, i t is the decision that makes the resources" (Toffler, 1971, p. 15). In technological society resources are created. Yet recent publications indicate a universal decline in R&D investment. This dissertation focuses upon some important aspects of R&D decision making in Canada. The first chapter analyzes available informa-tion about determinants and practices of R&D investment decisions, describes the inventory of normative models developed to improve decision making, and identifies'empirical studies investigating their implementation. A review of the state of the art leads to the identification of the following four areas of information which are deficient: 1 1. i n f o r m a t i o n about t h e n a t u r e o f s e l e c t i v e p e r c e p -t i o n p r o c e s s e s o f R&D d e c i s i o n making, 2. t h e o b j e c t i v e f u n c t i o n s ( e x p l i c i t and l a t e n t ) which g u i d e c h o i c e s among a l t e r n a t i v e R&D i n v e s t m e n t o p p o r t u n i t i e s , i i 3. the impact of R&D upon the positions of prime bargaining units ^organizations, and 4. the impacts of organizational structure and pro-cesses upon implementation of investment decisions. The first two categories of information relate to the question of what different decision units consider relevant in defining their problems and what they value. The last two categories relate to the organizational impact of R&D and the processes by which decisions are reached and imple-mented. These areas provide a focus for required additional research aimed at improvements in R&D decision making. This dissertation attempts to contribute to the first three areas of research outlined above. The focus in Chapter 2 is upon the impact of R&D in shifting resource shares of labour, capital and energy in the total output. Chapter 3 focuses upon processes of information selection in R&D decision making, identifying what environmental conditions are important to whom in making R&D investment decisions. Chapter 4 investigates multi-attribute preferences in R&D project selection. Each chapter draws some normative implications for R&D public policy, and the postscript identifies, promising areas for future research. Some of the major findings of this sequence of studies are: 1. Accepting a neoclassical framework of analysis, in most sectors, R&D has had no impact on input shares, indicating that scale and price effects.dominate the structure of technology. Where R&D has had impact on the structure, i t has sometimes had a labour using and sometimes a capital using bias. 2. Significant differences in patterns of attention to environmental conditions were identified. These differences are related to executive attributes and firm characteristics. i i i 3. High concensus exists with respect to tradeoffs among project attributes across a l l firm-executive group i ngs. 4. Compensatory actuarial models provide a good f i t with observations of R&D investment judgments. Some normative implications of the study for public policy include the following: 1. As R&D impact upon the economic objectives of the major bargaining units in firms is neutral in most cases, perhaps an effort should be made to eliminate technological development as part of the traditional arena of labour-management bargaining. 2. In creating favourable R&D investment climates, government ought to develop a sensitive strategy which recognizes e x p l i c i t l y the selective impact of single dimension interventions on alternative target populations. 3. The role of government as an independent insurance agent for R&D ventures is recommended to replace direct participation in project funding. iv TABLE OF CONTENTS Page ABSTRACT i i LIST OF TABLES v i ACKNOWLEDGMENTS v i i i Chapter 1 R&D PROJECT EVALUATION FROM FIRM BEHAVIOUR TO NORMATIVE MODELS TO IMPLEMENTATION - A STATE OF THE ART REVIEW 1 2 AN ASSESSMENT OF THE ECONOMIC IMPACT OF RESEARCH AND DEVELOPMENT • 40 3 INFORMATION PREFERENCES AND ATTENTION PATTERNS IN R&D INVESTMENT DECISIONS 108 4 MULTI-ATTRIBUTE INVESTMENT DECISIONS: A STUDY OF R&D PROJECT SELECTION .153 POSTSCRIPT 184 REFERENCES. 187 v LIST OF TABLES Table Page 2.1 Textiles, Cost Function Results 69 2.2 Paper and.Allied Products, Cost Function Results 71 2.3 Primary Metals, Cost Function Results , . 73 2.4 Machinery, Cost Function Results 75 2.5 Transportation Equipment, Cost Function Results 77 2.6 Non-metallic Minerals, Cost Function Results 79 2.7 Petroleum and Coal, Cost Function Results 81 2.8 Chemicals, Cost Function Results s 83 2A.1 Capital Data by Sector, 1961-1972 (PQK, R ( l ) , R(2), R ( 3 ) , R (4 ) , PK) . 88 2A.2 Energy Data by Sector, 1961-1972 (PQE, PE) 92 2A.3 Labour Data by Sector, 1961-1972 (Net QL, Net PQL, PL) 96 2A.4 R&D Data by Sector, 1961-1 972 (RD, SRD, PRD) 100 2A.5 Scale of Production by Sector, 1961-1972 (PY, QY, Total Y.) 104 vi Table Page 3.1 Var iab les Relevant for R&D Decis ion Making I l l 3.2 Sample Quest ionnaire 114 3.3 D i s t r i b u t i o n of R&D Laborator ies in Population and in Sample 115 3.4 Executive Cha rac te r i s t i c s . . . 117 3.5 Firm C h a r a c t e r i s t i c s , S t a t i s t i c a l T r a i t s 118 3.6 Firm C h a r a c t e r i s t i c s , Perceived Role 119 3.7 Factor Ana lys i s Results : 127 3.8 Co r re l a t i on Among Factors 128 3.9 F ina l Communality Estimates 130 3.10 Ranking of Items by Average Importance Ratings 133 3.11 Results of Discr iminant Ana lys i s on 44 Economic Items . . . . . . 137 3.12 Discr iminant Dimensions, Locat ion of Group Centro ids , and Important D i scr iminat ing Var iab les 138 4.1 Sample Quest ionnaire 160 4.2 Sample Project P r o f i l e ; 159 4.3 Regression C o e f f i c i e n t s for S i g n i f i c a n t Var iab les . . . . . . . . 162 4.4 Standardized Discr iminant C o e f f i c i e n t s f o r S i g n i f i c a n t Var iab les 167 4.5 Comparison of Ds icr iminant Ana lys i s and Regression Ana lys i s . . 172 4.6 The Winning Set of Projects 180 v i i •ACKNOWLEDGMENTS I would like to express my deep appreciation to Professor I. Vertinsky for his thoughtful and painstaking guidance in the preparation of this dissertation. Numerous stimulating discussions with him and his enthusiastic and generous support made the process of working on this dissertation a very pleasurable experience. I also wish to thank the other members of my committee, Professors W.E. Diewert, C.S., Holling, V.F. Mitchell and J.W.C. Tomlinson for their many helpful suggestions and discussions throughout the preparation of this dissertation. Thanks are also due to Professors E. Appelbaum, R. Barth, J. Claxton and W.T. Stanbury and Messrs. D. Elder and A. Ferrie for a number of helpful comments and activities associated with the pilot study and to Mr. A. Landsberg, Ms. H. Tutek and especially Ms. T.A. Cameron for computational assistance during the various phases of the studies reported here. I would especially like to thank Ms. S. Haller for her a r t i s t i c typing of the dissertation. Special thanks are due to Professors CS. Holling, K.R. MacCrimmon and I. Vertinsky whose thought provoking lectures and discussions on decision making formed an important part of my graduate education and a crucial background for work on this dissertation. I would also like to thank the Department of Industry, Trade and Commerce, Ottawa and the Institute of Industrial Relations at U.B.C. for providing research grants that made possible the extensive data vi i i collection and s t a t i s t i c a l analyses which form a major part of this dissertation. In particular thanks go to Mr. T. Clarke of DITC and Professor M. Thompson of 11R for their part in awarding these grants and for encouragement and guidance in the execution of the work. Finally, thanks also go to my husband, William Ziemba: his comments on previous drafts were helpful and without his impatience this dissertation might have taken longer to complete. ix Chapter 1 R&D PROJECT EVALUATION FROM FIRM BEHAVIOUR TO NORMATIVE MODELS TO IMPLEMENTATION - A STATE OF THE ART REVIEW Introduction R&D activities - innovation, new product development, process improvements to reduce costs and product improvements to extend the l i f e or market of a product - are the tools by which a modern company competes (Levitt, 1966). The economic health of a company depends on how well i t keeps pace with technological change. It is also suggested that R&D is an activity which can provide a solution to the problem of stagnation in the industrial nations. Yet R&D management is fraught with uncertainties aifd risks. The success of research activities depends on the economic environment. Research programmes can quickly be made obsolete by events external to the firm. Technological advances by competitors often require adaptive planning, making current plans unprofitable. The risks involved in such a c t i v i t i e s are assumed to r e s t r i c t commitments by the private sector to levels which are lower than socially desired, especially when a conservative posture is adopted by firms as a response to declining economic environments. It i s , therefore, the policy for many governments to seek direct and indirect 1 2 means to stimulate R&D expenditures. For example, the Canadian federal government has been providing financial assistance for R&D since 1961, this commitment amounted to $100 million in 1973. There are many pressures to increase such commitments, especially in countries whose com-petitive international position has been reduced significantly due to rising costs and the entry of new producers. Yet many doubts prevail as to whether such commitments in fact stimulate, or just replace, private R&D invest-ment. Improved results can only come from better management of the entire R&D effort. This chapter provides a description of the "state of art" in (1) explaining R&D investment behaviour, (2) normative models for R&D decision making, (3) studies of barriers to implementation of proposed R&D decision models and (4) investigation of existing internal R&D investment decision procedures (standard operating procedures). Determinants of R&D Investment Behaviour R&D is an activity aimed at reducing uncertainty about the environ-ment inherent in new product/new process ventures. It is an activity that produces and applies knowledge. R&D management can be analyzed in the general decision making framework. R&D is an activity undertaken in response to perceived gaps in the fulfillment of corporate objectives. Thus, per-ception of this gap and recognition of need are the f i r s t prerequisites. The next essential element is the perception of control and the recognition of opportunities for transforming the environment. This requires both availability of resources and generation of alternative projects. The 3 next requirements for successful R&D management are evaluation and selec-tion procedures. The literature on determinants of R&D investment provides a variety of complementary foci of explanation. These may centre upon different elements inhibiting or stimulating R&D investment by affecting objectives-constraints or benefit calculations. Schumpeter (1971, p. 37) noted an apparent clustering of innovations during economic booms. Prosperity helps induce innovation by increasing the expected returns. Size of the general market is also important in determining industrial research patterns (Quinn, 1966, p. 14). Thus population growth by increasing the market would be expected to have some influence on R&D decisions. Bright (1970, p. 67) labels population trends a prime, but often neglected, signal for the 10-30 year planning horizon. He points out that the Paley report on materials needs for 1975 published in 1952 was based on faulty population forecasts which led to underestimates of demand. Keynes (1964, p. 151) pointed to the role of stock market trends in the formation of long-term expectations and, therefore, to their role in investment and innovation. Keynesian analysis would also point to the influence of interest rates, while Galbraith (1973) would argue that profit and internal savings are more important, as the firms in what he defines as the planning sector are unlikely to borrow funds. Much of R&D is induced by the changing structure of resource costs, thus expected wage settlements, productivity change, inf l a t i o n , and energy requirements are possibly relevant to R&D decisions. (See Smookler, 1966; Rosenberg, 1974; Kami en and Schwartz, 1968; Fellner, 1971 and Hamien and Ruttan, 1974.) 4 The exchange rate plays a dual role depending on the company. It may be an indicator of expected cost changes or an indicator of demand changes. Leonard (1971, p. 234) reported studies that found a positive correlation between export performance and R&D effort in U.S. industries. The role of government as i t influences R&D covers several areas: direct influence on costs, indirect influence on costs via the a v a i l a b i l i t y of funds, supply of information, and influence in the marketplace. The risky nature of research activity results in a free enterprise economy underinvesting in R&D, especially in basic research (Arrow, 1962). Also, public support for R&D is needed when returns accrue to more^than the individual firm. When social benefits outweigh private benefits through diffusion of gains, some government support may be necessary to ensure proper allocation of R&D effort. Thus government subsidies, grants, and loans are made to encourage R&D. Government aid to reduce the cost and risk of R&D and thus increase the benefits may be of much importance. Favourable tax policies imply a cost sharing, also reducing an individual firm's commitment to R&D. In Canada, as in the U.S., the government directly supplies approximately 60% of the funds for R&D (Brooks, 1972). Government contracts influence both the type of research and the atmosphere in which research activities are undertaken (Quinn, 1967). Government support for R&D has tended to be concentrated in defense and areas that bring national prestige (e.g. space). This may result in a misallocation of funds (Leonard, 1971; Brooks, 1972) by limiting the technical resources available for other pursuits (and increasing their costs). On the other hand, de-creases in government funding for R&D have resulted in even greater concern for short-term payoff resulting in the undertaking of few risky and basic research projects (Brooks, 1972; Foster, 1971). 5 Government a l so i n d i r e c t l y inf luences costs by i n f l uenc ing the market i n t e r e s t rate and expectat ions of future p r o f i t a b i l i t y as r e f l e c t e d in the i n t e r e s t r a te . Acce lerated deprec ia t ion increases i n te rna l funds a v a i l a b l e . Government support of f e a s i b i l i t y studies and market deve lop-ment are a l so means of reducing costs — s p e c i f i c a l l y reducing the e x t e r n a l , s oc i a l costs of product innovat ion. The market in f luence of the government includes the ro le of the growth and s i ze of government expenditures. This i s r e f l e c t e d in many ways i nc lud ing formation of expectat ions about the economy, i n f l a t i o n , growth of demand and cost of funds. T a r i f f po l i cy in f luences demand expectat ions . The government ro le in the soc ia l assessment of R&D and the eva luat ion and contro l of technology i s important in d i r e c t l y i n f l uenc ing areas of innovat ion. P o l l u t i o n contro l and environmental p ro tec t i on regu -l a t i o n have in f luenced innovat ive processes f o r handl ing waste mate r i a l s . As noted by Br ight (1970, p. 63) " S o c i a l , p o l i t i c a l and now . . . e co l og i ca l changes may a l t e r the speed and d i r e c t i o n of the innovat ive process . " Technological f o recas t ing must be concerned with soc ia l forces such as spec ia l i n t e r e s t groups and government regulat ions that w i l l i n f luence technology and the acceptance of change (Thurston, 1971). Information gathering and processing c a p a b i l i t i e s are c r u c i a l to R&D dec i s ions . In techno log i ca l l y or iented f irms the time spent on i n f o r -mation gathering i s s tagger ing: "In a t yp i ca l research laboratory s c i e n t i s t s spend 80% of t h e i r time t r y ing to look things up and less than 20% doing what they are paid f o r " (Drucker, 1975). Market s t ruc ture apparently plays a key ro le in i n d u s t r i a l R&D but there i s much disagreement about the prec i se pattern ing of e f f e c t s . 6 Competit ion, concentrat ion and entry ba r r i e r s may be important f o r R&D dec i s ion making. Research spending can be viewed as a means of c rea t ing product d i f f e r e n t i a t i o n . Cooper (1966, p. 176) reports that some companies purposely develop products for small markets be l i ev ing that there w i l l thus be l i t t l e incent i ve fo r competitors to chal lenge.them. R&D would be low in areas where the prospects for d i f f e r e n t i a t i o n are low and a l so where marked d i f f e r e n t i a t i o n already ex i s t s (Comanor, 1967, p. 652). One would expect that competit ion would st imulate firms to innovate in order to acquire competi t ive advantage or to remain compet i t ive. Competit ion would be expected to f a c i l i t a t e f a s t im i ta t ion of technolog ica l innovat ion f o r the same reasons. However, Mansf ie ld (1969, p. 17) found that innovat ions spread less r ap id l y in less concentrated i ndu s t r i e s . Scherer (1971, p. 370) observed greater R&D in indus t r ie s where the minimum p lant s i z e represents 4-7% of the market share and the required investment i s $20-70 m i l l i o n . Beyond these ranges, there appear to be no advantages to entry b a r r i e r s . In con t ra s t , Schumpeter (1971) has argued that monopol i s t ic or o l i g o p o l i s t i c i ndus t r i e s innovated more r ap id l y because the threat o f entry of new f irms causes them to behave as competitors. This view i s supported by others (e .g . , Wi l l iamson, 1964, p. 67). S t a b i l i t y of market, r e f l e c t i n g both the age of indust ry and nearness to the sc ience/technology f r o n t i e r , has been proposed as an impor-tant f ac to r in exp la in ing R&D investment behaviour. A f i rm near the sc ience/ technology f r o n t i e r must be a l e r t to poss ib le innovations by competitors and must p a r t i c i p a t e a c t i v e l y in R&D to maintain i t s market p o s i t i o n . When the industry i s f a r from the boundary (a mature i ndus t r y ) , techno log ica l progress i s evo lut ionary though breakthroughs in other i ndu s t r i e s may make the e n t i r e market obsolete (Ansoff and Stewart, 1967). 7 Sales characteristics (both dynamic and static) play an important role in determining the importance of technology to the firm. Growing sales might be expected to make a firm more responsive to technological change and lead to increased research intensity; however Leonard (1971, p. 254) hypothesized that the causality ran the other way. Lithwick (1969, p. 5) observed that the evidence revealed a negative relationship between R&D intensity and growth. Increasing R&D is an offensive strategy to combat stagnation. Product l i f e cycle considerations have been incorporated in strategic models for R&D (Quinn, 1967; Kotler, 1967; Tilles,» 1966). Ansoff and Stewart (1967, p. 76) stress the importance of l i f e cycles for product innovation. Short cycle products require constant and continual product innovation, quick response and concurrent planning by marketing and engineering divisions. Decisions must be based on approximate and/ in-complete data rather than precise details. Long cycle products can enjoy sequential planning with detailed R&D preceding manufacturing^ and marketing planning. Ready availability of s c i e n t i f i c personnel may be crucial to the decision to undertake a project. In the early sixties in the U.S., the growth of aerospace and defense projects placed a burden on other R&D programmes. The demand for sc i e n t i f i c personnel in these areas increased the cost of other programmes resulting in a slackened pace of c i v i l i a n research. The slow down was especially significant in low technology, mature industries where R&D spending is sensitive to economic cost factors (Brooks, 1972, p. 115). Ansoff and Stewart (1967, p. 79) make the point that for successful innovation i t is not necessary, and often not desirable, to have a high ratio of sc i e n t i f i c to total staff. 8 Availability of funds affects the f e a s i b i l i t y of R&D investment. Williamson (1965, p. 67) contends that the advantages in financing experi-enced by large firms enhance innovative performance. Yet, affluence may lead to complacency. Cyert and March (1963) postulate that innovation is induced by market stress and pressure on profits. An increase in profits relative to the industry rate appears to decrease R&D ac t i v i t y . The 'poor' innovate. However, the empirical evidence is not conclusive. Scherer (1971, p. 364) found that the direction of causality goes the opposite way: profits are an indication of past innovative success. Patents and innovation lead to increased profit with a 3 or 4 year lag. There is much contradictory evidence concerning the relationship between size of firm and R&D effort and effectiveness. R&D activity in-creases with the number of employees up to a level of 5,000. Size (up to $75-200 millions) is also correlated with increased R&D activity (Scherer, 1971, p. 361). In general, the importance of a threshold size has been supported. Scherer puts forth a number of hypotheses to explain the associa-tion of size and R&D activity: 1. a d v a n t a g e s of R&D as a p o r t f o l i o i n v e s t m e n t . :. e n a b l i n g l a r g e f i r m s t o s p r e a d t h e r i s k , 2. economies t o s c a l e i n R&D a c t i v i t y per se3 3. economies t o s c a l e w i t h r e s p e c t , t o o t h e r d e p a r t m e n t s in t h e f i r m ( i n t e r a c t i o n w i t h o t h e r departments may le a d t o g e n e r a t i o n of new i d e a s , m a r k e t i n g c h a n n e l s , . e t c . ) , and 4. a d v a n t a g e s i n p r o c e s s i n n o v a t i o n as c o s t s a v i n g p r o -c e s s e s may have g r e a t e r impact on l a r g e f i r m s w i t h h i g h volume of o u t p u t . 9 Another advantage involves ease of gaining government contracts for research. U.S. Federal grants for R&D, for example, are more concen-trated than internal funds. Firms with 5000 or more employees undertook 88% of the R&D, this included 93% of the federally supported R&D and 83% of privately supported R&D (Scherer, 1971, p. 358). Among the reasons cited that large size inhibits R&D are the following: 1. d e c i s i o n s a r e made by i n d i v i d u a l s n o t f i r m s s o t h a t r i s k s p r e a d i n g may n o t be v a l i d , a n d 2 . o v e r - o r g a n i z a t i o n o f R&D may d r i v e o u t c r e a t i v e , i m a g i n a t i v e p e r s o n n e l . F u g i t i v e s f r o m many f i r m s ( e . g . S p e r r y - R a n d , I . B . M . , W e s t e r n E l e c t r i c ) h a v e f o u n d e d p r i v a t e l a b o r a t o r i e s ( S c h e r e r , 1 9 7 1 ) . The effectiveness of R&D expenditures appears to be negatively related to firm size above a certain threshold (Mansfield, 1964). It has been argued that large firms devote a larger proportion of their R&D funds to basic, more risky and longer term R&D projects than do smaller firms (Mansfield, 1969). But i t has been pointed out (Scherer, 197-1) that large firms appear to have an advantage in the lengthy process of making inventions commercially usable. He views the role of different sized firms as follows: 1. s m a l l f i r m s a n d i n d e p e n d e n t s p l a y a m a j o r r o l e i n g e n e r a t i n g new i d e a s , a n d 2 . l a r g e f i r m s p l a y a ma jo r - r o l e i n d e v e l o p m e n t o f i d e a s t h a t r e q u i r e l a r g e i n v e s t m e n t s ( S c h e r e r , 1 9 7 1 , p . 3 5 7 ) . For example, an independent researcher developed the idea for photocopying but Xerox was ab le to invest the $16 million required for development (Scherer, 1971, p. 355). Dupont is used as an example of the effectiveness of large size for innovation. But many of their inventions 10 resulted from purchasing of rights to new ideas. Only 10 of 25 major inno-vations during 1920-1950 were discovered at Dupont's laboratories (Mueller, 1971, p. 162). Dupont was most successful at making process/product improve-ments rather than inventing new ideas. This seems to be a general pattern. The basic innovativeness of a firm — i t s readiness to perceive and act on innovative opportunities — may be a reflection of i t s past success with R&D. Firms do what they have in the past been good at doing and avoid activities that have in the past led to failures. This positive reinforcement may also be related to a c r i t i c a l mass required for successful innovation. Ansoff and Stewart (1967) find that below some threshold level, R&D expenditure may be totally ineffective. In evaluating R&D opportunities, firms make tradeoffs among three classes of attributes: commitment of resources, expected payoff and risk. Mansfield (1968, p. 55) found that the average project size for electrical equipment manufacturers in 1963-64 was $285,000. He also found that the probability that a firm would fund a project was negatively correlated with the size of the investment required (Mansfield, 1968, p. 310). Require-ments of financial commitment are especially important for small companies or those with constrained access to capital markets, as potentially profit-able projects may have to be abandoned before they have had a real chance to succeed (Cooper, 1966, p. 175). The payback period is a measure of the time commitment to a pro-ject. The payback period for R&D projects is generally required to be shorter than that for investment in plant and equipment. For a l l manufac-turing in the U.S. in 1961, 55% of the projects undertaken had an expected payback of less than 3 years and an additional 34% were in the 3-5 year 11 range (Mansfield, 1968, p. 15). Gerstenfeld (1971, p. 22) found that the payback period varied with size of firm: the average was 4.26 for large firms and 3.5 for small. The high proportion of industrial R&D devoted to development and applied research is indicative of this required short pay-back period (Leonard, 1971, p. 236; Bright, 1968, p. 6). A maximum payback period may also appear as a constraint imposed by management, thus i t may be the deciding factor in project selection (Kotler, 1967, p. 30). A variety of expected payoff indicators are reported to be used by firms. The most frequent criterion is rate of return but other expected performance c r i t e r i a are used such as impact upon market share, increase in sales, etc. No data on specific average levels of expected payoffs are reported in the literature (see Mansfield, 1968; Disman, 1962; Quinn, 1966; Kotler, 1967; Peterson, 1967; and Allen, 1970). Risk is an inherent attribute of R&D a c t i v i t i e s . Risk can be measured by two items: probability of success and patentability. The probability of success incorporates both technical and commercial uncer-tainties. As the risk increases, the expected value of the return and thus the maximum expenditure j u s t i f i e d decreases (Disman, 1962, p. 88). The bulk of R&D is relatively safe (non-risky) and aimed at small improvements in the state of the art. Mansfield (1968, p. 56) found that the ex ante probability of technical success for projects undertaken averaged 80%. It seems that firms generally do not in i t i a t e a project until major technical uncertainties are eliminated. Gerstenfeld (1971, p. 22) found a similarly high average of 71% in his sample. As basic research projects are more risky than applied, a risk avoiding firm will tend to fund more applied than basic projects (Nelson, 1959, p. 304). 12 P a t e n t a b i l i t y reduces the r i s k of a p r o j e c t ' s success by p r o t e c t -ing the innovating company from compet i t ion. Some f irms i n s i s t on patent -a b i l i t y before undertaking developmental r i s k s (Ansoff, 1965, p. 110; Quinn, 1966, p. 124). However, Mansf ield (T968) postulated that patents are becoming less important as the l i f e cyc le for many high technology goods i s qu i te short , and becoming shorter . Normative Models for R&D Project Se lec t ion and Management Many models have been proposed to analyze the R&D management problem. The suggested models of choice span the spectrum from simple rank ing / ra t ing models with minimal data requirements to complex programming models. The s implest models ignore the complexity inherent i n the problem. These are the s ing le c r i t e r i o n p r o f i t a b i l i t y models. Often pro jec t s are ranked by the se lected c r i t e r i o n , the highest ranking p ro jec t s are funded u n t i l the budget i s exhausted. Scor ing models take account o f a number of c r i t e r i a and again r e l y on ranking projects f o r s e l e c t i o n . -Disman's (1962) model i s in th i s category. He proposes to def ine the maximum expenditure j u s t i f i e d (MEJ) for each p ro jec t . This i s fundamentally a present value measure modif ied by the p r o j e c t ' s p r o b a b i l i t y of success. The formulae d i f f e r depending on the type of p ro jec t . The MEJ i s present value m u l t i p l i e d by the p r o b a b i l i t y of techn ica l success in the case of process improvements; while f o r new products, the present value is m u l t i p l i e d by both the prob-a b i l i t i e s of technica l and commercial success. The MEJ d iv ided by cost of the p ro jec t i s then an index of d e s i r a b i l i t y (a b e n e f i t - c o s t r a t i o taking account of uncer ta in ty ) . Projects are ranked by the d e s i r a b i l i t y index and highest ranking projects are to be chosen. The Cranston (1974) model 13 defines another p r o f i t a b i l i t y index that depends on estimation of the probability of technical and commercial success modified by a " c r e d i b i l i t y " estimate. Highest scoring projects are chosen and a project i s replaced wherever another has a higher index. Mottley and Newton (1959) propose an index based on five c r i t e r i a : probability of success, estimation of time to completion, cost of project, strategic need and size of market gain. This model combines consideration of both quantitative and qualitative data. Recognizing that few data are generally available for evaluation of R&D projects, ratings oh each criterion cover broad ranges (e.g. probability of success: 1 = unforeseeable, 2 = f a i r , 3 = high; strategic need: 1 = no apparent, 2 = desirable, 3 = essential). A project is rated on each criterion and the resulting numbers are multiplied to give the project a score. Thus, multiple c r i t e r i a are recognized but tradeoffs among c r i t e r i a are not . evaluated. In none of these models is any reference made to actual applica-tions, nor is practical ju s t i f i c a t i o n given for the scoring components. Other scoring models have been proposed by Williams (1969) and Moore and Baker (1969). Zoppoth (1972) applies system analysis to R&D management at Xerox laboratories. A major part of the paper discusses a system to standardize the definitions of product models for project evaluation (engineering model, pre-prototype model, pre-production model, etc.). This system is integrated with a scheme for programme planning and evaluation of technical qualifications and an identification of risk. Another tool described by Zoppoth is.DRAM (Decision and Risk Analysis for Management) a risk and u t i l i t y analysis model for project evaluation (similar to that developed by Hertz, 1968). Probability distributions are estimated for company-controlled variables (cost, 14 capabilities, etc.). Customer u t i l i t y variables are defined along with a probability distribution of the relative weights for each criterion. Company-controlled variables can be transformed into customer u t i l i t y variables thus giving the value of any possible product development. Simulations are performed. The f i r s t step is to sample from the company variables to derive a "product" which is converted into customer values. Weights for the customer u t i l i t y function are randomly selected and applied to the product values giving a payoff. Payoff is related to sales (placements) and thus related to revenues and costs to yield net present value. The simulation is per-formed a number of times to yield the net present value curve'. No indica-tion is given of how final choice is to be made. A risk analysis type model for budget reallocation is also pre-sented by Bobis et al. (1971). The model has been used as an aid to decision making by the Organic Chemicals Division of American Cyanamid Company. The modelling led to a more farsighted portfolio. The 1967 model budget called for a major redistribution of the research effort and the 197Q budget closely resembled this distribution. The procedure calls for estimates of the prob-a b i l i t y of completion and success for each expenditure level (research curve for the project) and estimates of annual sales for each starting year. These are combined to yield expected sales and probability for each expenditure level. This then expl i c i t l y displays the trade-off between increased expendi-tures and increased funding. Data requirements for the research curve are minimal: the research curve (a logistic curve) can be estimated with only three observations: most lik e l y , optimistic and pessimistic costs. A scoring model was developed to estimate the probability of technical success (five 15 c r i t e r i a with three classes each). It is not specified how the ratings on the c r i t e r i a were aggregated but i t appears that a simple linear model was used. The solution technique is not straightforward. The optimiza-tion procedure was described in Atkinson and Bobis (1969). The objective is to select a portfolio over time to maximize the expected return given annual budget constraints where the annual budgets depend on expenditures in previous years. , An iterative scheme is used to solve the problem. The consideration of the possibility of an i n f i n i t e variation in allocation to each project makes i t very hard to solve the model. This model would probably be unsuitable for laboratories with more than a few projects. The rating models previously described serve as the c r i t e r i a for more structured choice models which consider e x p l i c i t l y resource constraints. One of the most computationally effective optimization methods available for constrained allocation problems is linear programming. LP models enable the explicit maximization of an objective function and permit consideration of a variety of constraints in addition to the budgetary one. The solution is in the form of the entire optimal portfolio. A shortcoming is the lack of consideration of i n d i v i s i b i l i t i e s (in the LP framework, a project can receive funding in any proportion that is optimal, even i f that makes no sense in terms of the project). Integer programming recognizes the inherent project i n d i v i s i b i l i t i e s but loses the relative ease of computation of the LP formulation. An LP model which was tested on U.K. Department of the Environment, Highway Safety data, is described by Moore (1974). The objective is to 16 maximize the expected benefit/cost ratio by selection of an optimal port-fo l i o of highway safety projects. Net benefits are defined as the monetary value of decreases in t r a f f i c deaths, personal injury and property damage by increased highway safety (B) minus the research (R) and implementation (I) costs net of any other losses (L) that may occur (e.g. decrease in deaths may be replaced by increased injury). Costs can be defined either as research costs alone, or as research and implementation costs, depending on the relevant constraints. Uncertainty is included by the estimation of probability of success and probability of implementation; these are applied to cost and benefit calculations to derive the expected net benefit/cost ratios. The values B, R, I, and L are estimated through interviews: subjects are asked to give most li k e l y , optimistic and pessimistic estimates, which are weighted to derive the actual estimates used. The planning period was 1972-2000 and a 10% discount rate was applied. The choice set included 41 on-going projects, 36 variations of these projects, and 21 proposed projects. The constraints included budget and manpower constraints, and the requirement that mandatory projects be undertaken. It was not e x p l i c i t l y stated but integer programming would be required for solution of the problem as formulated since the constraints require that not more than one version of a project be selected and that the version chosen, be funded f u l l y . Nutt (1965) describes a model designed in the U.S. Air Force Flight Dynamics Laboratory. The programme is designed to handle the following attributes of R&D project portfolio selection: needs of the Air Force for various systems development, probability of success, capabilities of research team, degree of support by project for each task, contract-out versus in-17 house development, relationship of support to progress and cost. The objective is to select a portfolio to maximize the total R&D effectiveness (RDE) derived from the budget. The data requirements are substantial though Nutt states that all the information could be collected in any evaluation effort. A mission matrix must be defined giving the system needs of the Air Force and each mission must be given priority values. Each project must be evaluated in terms of its system contribution, probability of success, and capability of the laboratory to attain the required techno-logical advance. A 10-year time horizon is used. Six 10-year plans are generated for each project. The f i r s t is based on the planned resource expenditures, another uses half the resources and the third uses double the resources. A computer model interpolates to define the remaining three project plans. The value (RDE) of each project version is a function of the rate of expenditure, distance from the state of the art and timeliness of completion of the project compared with Air Force goals. The RDE for each project version is defined to be the increased probability of success achieved in the budget period, weighted by the contribution to the goal and the importance of the goal served plus the increase in the confidence level (technical capability) of achieving the goal. Constraints include total budget for in-house and contract research, and contract and in-house engineers. Qualitative aspects such as capability of achieving technological advance and probability of success are recognized. I n d i v i s i b i l i t i e s are handled by defining six project levels, though the interpolation may not be valid ( i f six, why not an infinite range of versions?). The model is a method of collecting and organizing data about the projects. The data bank i t s e l f can serve the function of after the fact evaluation of research progress and research effectiveness. 18 Cochran et al. (1971) describe an integer programming model that has been implemented by the Smith, Kline and French Laboratory to aid management in R&D decision-making. The aim of the model was to recognize the unique problems of the pharmaceutical industry: long lead time (10 years) from conception to commercialization of product coupled with high product at t r i t i o n rates and high R&D costs. The model has two components: project evaluation and portfolio selection. The project evaluation com-ponent reduces the economic data for a l l projects to a single dimension -the expected net present value (ENPV). This calculation requires an estimate of the cash flow from the project, the probability of technical success, and the capital discount factor. A 10-year product l i f e is assumed. The management team for the project estimates the cash flow. The capital cost is the sum of the expected corporate growth rate and the dividend yield (reflecting return to attract new stockholders). The actual calculation of ENPV is unique, accounting for differing probabilities relating to costs and returns: i n i t i a l outlay is certain and weighted one; future outlays are less certain as a project could be terminated, the weight applied is the average of one and the. probability of technical success; returns are un-certain and are weighted by the probability of technical success: (-> . \ K x. N x. ENPV = X L + I V T + P I V r \ 2 J i=2 (1 + R) i-K+1 (1 + R ) 1 _ 1 Where p is the probability of technical success, R is the discount rate, and x- is the net return (negative for costs) of the project in period i . The model allows for sensitivity analysis of ENPV with respect to cost and probability estimates. 19 The portfolio selection component takes the ENPV data and selects a portfolio to maximize total ENPV subject to fi.xed budgetary constraints. An integer programming algorithm is used. Budgets are specified for a number of years (the planning horizon). The model is conversational and user oriented. If a project not included in the portfolio is judged manda-tory despite low ENPV (i.e., i t is selected for non-economic features) the budget variation feature enables a recalculation of the optimal portfolio -netting out mandatory projects from the project l i s t and costs from the budget. This feature makes i t easier to sell the model to managers who recognize that non-economic c r i t e r i a are also valid. Some 1 imitations of the model are: (1) only few projects can be considered, (2) only single versions of the projects are considered and (3) though possible termination of projects is recognized in the calculation of ENPV, there is no feedback loop to i n i t i a t e termination and substitution of other projects within the planning horizon. The discounting of the future costs is incorrect, making the projects seem less costly than they actually are.. It would be better to include the f u l l costs in the ENPV, discounting only the benefits by the probability of technical success. Grossman and Gupta (1974) describe a mixed integer model that is used in the Johnson and Johnson pharmaceutical company. The aim is to develop a portfolio selection procedure to account for different types of research activities (exploratory, developmental, and product support). This model is more general than the Cochran et al. model: multiple, c r i t e r i a are collapsed into a u t i l i t y measure that incorporates more than present value, parallel strategies and interrelation among projects are considered, new and old projects compete for funds over the planning horizon, various 20 funding levels of projects are defined, mandatory projects are also con-sidered but they need not commence in the i n i t i a l period of the plan. The model makes use of decentralized and specialized information of different units in the organization in calculating project value. The model is iterative and requires management participation. The f i r s t phase requires data generation for each project. A novel approach here is to define "families" of projects. The families reflect parallel strategies of development or exploratory research. This feature is a recognition that working on project development from a variety of ways increases the probability of technical success; this is accounted for by Bayesian methods. The project l i s t then includes pseudo projects that are com-binations of projects in the same family. Multiple project levels (normal, accelerated, delayed) are defined for projects and pseudo projects. The assess-ment of project value uti l i z e s a rating mode - various attributes are defined (e.g., growth potential, marketability, competitive products, contribution of new technology to corporate image, s t a b i l i t y , productive a b i l i t y with respect to various resources). Executives and other qualified individuals rate the project on these dimensions on a five point scale giving information only where they feel qualified. The rating task for each individual i s relatively easy. This information is then aggregated into a u t i l i t y value for each project. The objective is to select a portfolio of projects and pseudo projects to maximize u t i l i t y over the planning horizon. Initiating times are selected for each project. The constraints account for' selection of only one version of each project and only one i n i t i a l period. If 21 a project is mandatory equality constraints apply and the programme selects the optimal period for commencing the project. Other constraints relate to budget and manpower availability. It is an interesting attempt to model complex aspects of the R&D management problem. No documentation, however, is supplied for the selection of the c r i t e r i a that make up the u t i l i t y measure nor how tradeoffs among crit e r i a and individual judges are handled given the decentralization. Another integer programming model, somewhat less general but adapted to special problems, has been developed for use at BISRA (British Iron and Steel Research Association). This is an industry research group and as such has special problems: i t s raison d'etre is to perform research that would not otherwise be undertaken by individual firms in the industry and then to ' s e l l ' these projects to the industry. The model develop-ment is reported in Reader et al. (1966), Collcutt and Reader (1967) and Beattie (1970). The approach that has evolved is the use of integer programming with subjective evaluation of probability estimates of technical success. The benefit/cost ratio is the criterion used for evaluating projects. The benefits depend on diffusion of the innovation throughout the industry compared with diffusion i f BISRA had not under-taken the project and later someone else did. It is assumed that diffusion is faster due to BISRA involvement for two reasons: (1) the project is undertaken sooner, and (2) the BISRA selling campaign increases the speed of diffusion. The two diffusion paths are.estimated assuming 22 logistic diffusion curves. The discount rate varies with the project type. Two types of projects are recognized: those resulting in annual savings to firms in the industry and those resulting in once and for al l capital savings. Annual savings affect central funds a v a i l a b i l i t y , the discount rate applied is 12% in real terms. Capital savings affect large capital issues valued at a discount rate of 7%. The benefits are weighted by the possibility of technical success estimated by project leaders. Costs are marginal costs and do not include overhead. Project variations are defined u t i l i z i n g different research team sizes. Benefits, costs and probability of success a l l depend on team size allocated to the project. To compare long and short term projects i t is assumed that a research team assigned to a short project spends the remaining time in the planning horizon working on f i l l - i n projects. Thus the total benefit of the project includes the benefits of the f i l l - i n projects (Reader et al. a 1966). The objective is to select projects to maximize the net benefit of the portfolio subject to manpower and budget constraints. Constraints include selection of only one version of a project, mandatory projects that must be selected in one version and contingent projects where i f one project is selected another must also be selected. Thus some project interdependences and non-economic choices are recognized (Beattie, 1970). Constraints can be expanded: 1) to require that specialists be assigned to their specialities (to maintain group morale, staff satisfaction constraints), 2) to guarantee prestige by requiring that at least one prestigious project be selected, and 3) to ensure diversification by requiring that at least three projects are selected (Reader et al., 1966). 23 Beattie (1970) feels that in use, the system has been worthwhile. Projects that were i n i t i a l l y thought to be worthwhile have been found to be uncompetitive and have been terminated. Some'data are hard to collect but modifications are being made to increase the ease of data collection. Though the costs of implementing the model are high (2-3% of the budget), the returns are substantial (Beattie, 1970, p. 290). The main benefit has been a better organization of the work, with the use of larger work teams to speed up completion of projects. The model could be improved by allowing sequential choice of projects rather than relying on f i l l - i n work to com-plete the portfolio. Bell and Read (1970) developed an LP model for R&D portfolio selection under uncertainty. Their model is based on probabilistic networks and was used at the Central Ele c t r i c i t y Generating Board and at the Gas Council (England). As with the BISRA model, the benefits relate to use by industry of the innovations. The actual form of the benefit function is not clearly specified. Three versions of each project were defined: slow, medium and fast. The constraints account for the requirement that "key" personnel be included on some projects. In practice, 20-40 projects with 40-100 versions are generated. Some projects are designated mandatory. The model is solved using LP though this meant that some projects were accepted at partial levels. As i t does not necessarily hold that partial projects yield the same fraction of benefit, an iterative procedure was used to derive the final portfolio. It would be better to use integer programming. Sensitivity analysis was also run with variations in the discount rate (basic rate 8%) and changes in the time horizon (basic time 10 years). Only a limited number of projects were affected by these changes. 24 Lockett and Freeman (1970) also develop a network model to account for the stochastic nature of resource requirements and project benefits. They apply the model to a case study based on data from an industrial R&D laboratory and compare the solution with that obtained from an expected value model. The example used was small, containing only nine projects and the results obtained were similar to those from the expected value solutions. The authors feel that this evidence validates the procedure though i t seems that this adds credence to the easier to solve expected value models. The assignment model developed by Beged-Dov (1965)- is another example of LP applied to R&D management problems. The problem is how to assign N selected projects among n teams (or laboratories) of researchers. The problem formulation requires three assumptions: 1) projects can be cla s s i f i e d into a small number of groupings by similarity, 2.) most teams can work on a l l the projects with varying efficiency (different costs and time to completion), and 3) i t is possible to estimate the costs of research-ing each project by each team or at least to estimate relative costs. The objective is to allocate projects to teams to minimize the total cost of researching the portfolio subject to constraints on the number of projects each team can handle and the budget that may be allocated to each team. The problem is formulated as a transportation problem to take advantage of the existence of efficient solution codes. Dynamic programming (DP) models take account of time dependent returns and can consider the needs f o r project reappraisal throughout the planning horizon. Decisions made in one period will affect the system and decisions in the future. Thus DP can increase realism by modelling the 25 sequential aspects of decision making, accounting for the fact that i n i t i a l commitments for exploratory research are less than, the final commitment for project commercialization. The solution method is recursive. Data require-ments are more complex than for LP models and the solution techniques are more d i f f i c u l t . Hess (1962) focuses on the sequential nature of the R&D management problem: a decision to init i a t e a project is assumed not to involve any com-mitment of further allocations. He considers R&D activity as a purchase of information. Thus R&D provides better information on which to base future decisions. The objective is to choose a sequence of budget allocations over time to maximize the total expected discounted net profit (maximize the present value of the expected cash flow) from the total R&D budget. R&D projects are considered completed when technical success is achieved. There-fore, projects are terminated before the assessment of economic f e a s i b i l i t y . With this definition i t is not clear how the present value is derived. Net profit is dependent on the period in which technical success is achieved and decreases as time increases. He considers that there is some period beyond which technical success is irrelevant and the project should not be commercialized as i t will result in losses due to competitive considerations. Hess develops models with and without budget constraints and with prob-a b i l i t y of technical success dependent and independent of previous research. He attempts to get an expenditure stream that coincides with that observed in practice, i.e., increasing expenditures over time. .There-fore, he prefers forms of the model that have this characteristic. No example is given so i t is hard to assess the d i f f i c u l t i e s of data collection. Time dependent income streams and probabilities of success are required 26 and these may be hard to estimate. No indication is given of the size of the problem that could be handled by this method. General analytic solu-tions are not available for a l l models; for example the model with budget constraints and time dependent probability of technical success can only be solved numerically and in the case of a large problem this would not be feasible. Dean and Hauser (1967) develop a programming model with the capacity for handling a variety of objectives. They model the hierarchical structure of the problem which leads to sequential decision making. The model is applied to a military R&D problem which has three levels:. 1) the overall system to be developed; 2) materiel concepts or components of the system; and 3) possible technical approaches to developing the components. Step 1 involves choice of the technical approach to maximize the probability of achieving the materiel concept. Step 2 requires funding of the component concepts to maximize the probability of achieving the system. Step 3 in-volves allocation of funds across systems to maximize the total value of the R&D output. Data for steps 1 and 2 are probability of success and costs, and for 3 the payoff or value of each system to the military objectives. Value is measured by military priority of having the system. The model was solved for many levels of budgeting: step 1 was solved at $10,000 incre-ments, steps 2 and 3 at $100,000. The model was solved for nine c r i t e r i a . These reflect interaction with decision makers who, when they do not like the optimal portfolio of systems funded, suggest other constraints to modify the solution. For example when some systems were not funded, they suggest a constraint that a l l systems be funded at some level to maintain continuity of research. The model enables easy analysis of cost effectiveness of the various programmes. 27 Souder (1972) develops a planning and control system that is used at Monsanto Company for aiding R&D decision making. Dynamic programming is used in the planning model, while cost effectiveness is used in the control model. Probability of success in this model depends on the level of funding. A four question procedure is developed for estimating the probability of success. Questions relate to familiarity of the problems to be encountered, avai l a b i l i t y of technology and r e l i a b i l i t y of cost estimates. The relation-ship of the expenditure level (budget/maximum budget) and the probability of success depends on the pattern of yes-no responses. Two objective func-tions are considered: 1) maximize the expected net return on R&D investment and 2) maximize gross return from R&D investment. The model solution gives project selection, project scheduling and project funding. The control model relies on s t a t i s t i c a l quality control. Variance is measured between actual performance and planned performance. If the variance is above a c r i t i c a l level then a decision is called for. Three control variables are suggested: cost variance, progress variance (prob-a b i l i t y of success), and cost/progress variance. Data were available with some modifications. A 10 year planning horizon was required with the same degree of confidence in data for a l l projects. PERT diagrams (decision trees) were used. D i f f i c u l t i e s arose in determining actual probabilities of success to compare with expected as administrators are often unable to assess the actual status of projects immediately. Therefore, Souder defined a new measure that relates to the weighted percentage average of milestones attained. This requires detailed flow charts of projects and knowledge of a l l milestones or information bits required for successful completion. Collection of these data would be 28 d i f f i c u l t and costly. The model serves best for evaluation of projects that are neither exploratory (not enough data beyond day to day activity) nor development (planned relatively easily with other models). The model was in use for one year and then the DP aspect was abandoned though the PERT analysis and control phases are s t i l l used. The DP models generally are hard to understand and data requirements are not easy to interpret (e.g. actual status of project). Kepler and Blackman (1973) use DP to solve a simple example (four basic a c t i v i t i e s , four optimal activities) where the problem is to re-allocate resources as the result of a budget decrement. Three u t i l i t y functions are considered. The DP solution is shown to be better than the conventional solution of equal percentage cost reduction over a l l projects. No comment is made on the relative costs and ease of data collection or the problems that would be expected in solving a more r e a l i s t i c larger problem. Charnes and Stedry (1966) suggest that R&D is characterized by breakthroughs and other emergencies (e.g. changes in the competitive environment). These events of low probability place high resource demands on the system when they.occur. Therefore, R&D management requires an adaptive planning mode; one that can respond to new information between planning periods. They develop a chance constrained model of two stages -planning and control process - which allows for 1) random av a i l a b i l i t y of f a c i l i t i e s in the short and long run; 2) random occurrence of emergency demands at random times during the short run, 3) probabilistic constraints on conformity to availability constraints and emergency demands, and 4) deterministic constraints on desired activity levels. A number of 29 r e a l i s t i c possibilities occur in the solution: 1) i f the binding con-straints reflect research needs, then there is no Change in the plan, 2) i f the binding constraints reflect f a c i l i t y a v a i l a b i l i t y , then there is lessened activity (to protect against an emergency, create slack in the system); and 3) there is increased activity i f the constraints on f a c i l i t i e s are not binding to hedge against emergency. It is not clear what size problem can be handled. Some of the data may be d i f f i c u l t to collect and interpret - such as the probability of an emergency in each research f i e l d . Lockett and Gear (1973) propose a method of R&D portfolio selec-tion that combines decision tree analysis, simulation and linear pro-gramming. For each project a stochastic decision tree is drawn. Resource requirements of each path and in each time period are specified as is the distribution of net benefits associated with each end point. (Gear et al.3 1972 discuss the process of describing a decision tree for individual R&D projects, also see Raiffa, 1968 for the general theory). ^This pro-cedure models the sequential nature of R&D project evaluation. It also enables the analysis of the parallel approaches. The procedure is flexible and allows for consideration of uncertainty in project duration, resource requirements, project outcomes and project value by the appropriate defini-tion of the branches. The problem is how to allocate resources in period 1 in order to be on an optimal path in the future. Various solution techniques are possible. Lockett and Gear mention three: stochastic programming, chance constrained programming and simulation combined with LP. For ease of computation, they select simulation plus LP and describe the process: 30 sample at each probabilistic node, this yields project paths; given a set of paths for a l l projects, solve an LP or integer programming problem; after many such models are solved, search the solutions for stable patterns. The output of the procedure is a number of alternative portfolios with distributions of benefits and probabilities of violating resource con-straints from which management can choose. They present an example using six projects and thirteen versions, with only one resource constraint. For this small example the results are similar to that from a stochastic linear programming model (where only one portfolio would be presented in the final output). However, for large scale problems the stochastic pro-gramming model might not be practical (see Wets, 1976, for algorithmic techniques). The procedure was also applied in an industrial laboratory that had 37 projects, 65 decision nodes and 40 chance nodes, • 4 time periods and 6 resource categories. Chance nodes were sampled 100 times which is not really sufficient for r e l i a b i l i t y of estimates. The procedure for portfolio selection is ad hoc. In this case two attempts were made. The f i r s t approach involved rounding project levels to 0 or 1 and searching for a frequently occurring portfolio. However, there were 72 portfolios each occurring with the same frequency. The second method ranked projects by overall mean of occurrence (average fraction of project selected) then funding projects until the f i r s t resource constraint is reached. There is no indication that the . portfolio selected is optimal or even good. The characteristics of the solution are not known. This procedure is too technical given the ad hoc nature of the solution. A better solution procedure 31 would be to formulate the problem as an integer programme with 65 decision variables, 24 constraints (6 in each period), u t i l i z i n g the expected values defined for the chance nodes. Optimal control models are specified when there are certain variables that guide the evolution of a system continuously or at discrete intervals over time. Solution of the problem requires the selection of these variables to minimize an objective function. Problems are solved by successive approximation and convergence is not guaranteed unless certain convexity assumptions are met (Zangwill, 1969 and Wilde and Beightler, 1967). Lucas (1971) describes a continuous time control theory formula-tion for single project evaluation. The project is assumed to incur costs during a period [0,T] and returns are discounted to time T (not the i n i t i a l period of the investment horizon!). Four cases are generated depending on uncertainty with respect to time and costs: time to comple-tion can be known or unknown, costs can be fixed or variable,, i.e. decreasing in T. The objective function is to maximize the present value of the project i f time to completion is known or to maximize expected present value i f completion time is unknown. When costs are known they are used in the solution. When costs are variable, the optimal level of expenditure is determined at each time period using optimal control theory. Assumptions of the model are: returns on project completion are independent of expenditures and time to completion (i.e. rules out the case of ri v a l r y ) , uncertainty as to completion time is most important (successful completion and payoff on completion are deterministic) and varying expenditures will affect progress toward completion. The solutions have the following characteristics: 32 1) Costs and Time known: solution similar to investment decision, undertake if net present value is positive. 2) Time known, Costs controllable: present value increases as time to completion nears, therefore it is optimal to increase expenditures over time. 3) Costs known, Time unknown: undertake project if mean net present value is positive. It is not optimal to use the expected time to completion as the answer depends on the probability distribution of time to completion. 4) Costs controllable, Time unknown: completion time is dependent on the expenditure plan. Contingency plans are necessary to determine how expenditures will vary with changes in uncontrolled variables, in this case total required effort. -* Though analytic solutions are provided, the model is not applied to any real project. Data collection would probably make these models unattractive. Model 4 seems most r e a l i s t i c with respect to cost and time uncertainties, but would be d i f f i c u l t to apply. Models 2 and 3 do not seem r e a l i s t i c , model 1 is the t r i v i a l case. Optimal control models of R&D management have also^been developed by Kamien and Schwartz (1971, 1974). In the f i r s t model they consider the uncertainty with respect to time or effort to completion similar to Lucas. This is called technical uncertainty. The second model takes account of both technical and market uncertainties (rival behaviour) so that cash flow (benefit) expected from the project is uncertain and depen-dent on time of successful completion of the project. As stated in the f i r s t paper "the objective . . . is not to furnish guidance for R&D managers but rather to provide a theoretical rationale for two basic expenditure patterns which might be empirically observed" (Kamien and Schwartz, 1971, p. 61). The models would be d i f f i c u l t to implement, i.e. data requirements 33 such as the relationship between the level and rate of expenditure on accumulated effort and time to completion are rarely available. The expenditure patterns j u s t i f i e d are: 1) increasing annual expenditures ( i f the completion rate is a non-decreasing function of total effort) and 2) i n i t i a l l y increasing then decreasing annual expenditures ( i f the com-pletion rate is i n i t i a l l y increasing up to a certain amount of total effort and then decreasing). Under rivalry, expenditures are made until some firm successfully completes the project: that firm.collects the benefits and the others lose a l l . Rivals are recognized by a single subjective probability distribution over introduction date of competing products. The objective is to maximize the expected value of the project. The addi-tion of rivalry to the optimal control model does not change the form of the solution. It does, however, lower the expected value of the benefits of the research (the probability exists that a rival will complete the project f i r s t ) and thus may make some projects unprofitable. Again, imple-mentation of this model would be very d i f f i c u l t . Implementation Problems of R&D Decision Models Quantitative R&D project management techniques are not in general use. Many managers have expressed interest in having decision aids and many models have been developed, however, few have been implemented. There is not enough known regarding the evaluation of the usefulness of the various models. An organization does not know in advance the implication of implementing various models. Souder et al. (1972) provide an approach to the assessment of the value of the models as decision aids. They report on two quasi-experiments designed to test the potential usefulness of R&D 34 management techniques. The f i r s t experiment was carried out in an R&D department where the objective was to develop new products and improved products. A dynamic programming model of portfolio selection was intro-duced to the company (see Rosen and Souder, 1965). The objective of the model was to choose a portfolio to maximize total expected net profits given fixed resources. The data requirements included estimates of project l i f e , maximum and minimum annual expenditures and present value of profits. The output was an i n i t i a l allocation of resources for the f i r s t year. This would be periodically updated when new projects were proposed, or when data estimates changed. The model served the purpose of integrating information of the various departments in the company. They report that the model was an important analytic tool, helping to c l a r i f y objectives and constraints of individual departments and focusing attention on needed data. ' The second experiment was undertaken in an R&D laboratory where the major concern was exploratory (and therefore riskier) research. The model implemented was a modification of Hertz's (1968) risk analysis model. The method requires the construction of uncertainty profiles for the key factors. The outcomes considered included anticipated technical or research achievement, market opportunity, market penetration, and profit margin. A group would meet to evaluate individual projects. Then projects would be ranked in order of p r o f i t a b i l i t y given risk. Probability distributions for each possible portfolio were developed. This seems to be an incomplete application of the Hertz method that would normally rely on simulation and estimation of the u t i l i t y function. Without this the model served only as a guide to selection, a means of organizing data. 35 The manager is s t i l l l e f t with the d i f f i c u l t task of selecting among many possible portfolios without a guide. Their main conclusion from these experiments is that R&D manage-ment models can induce the collection and exchange of data, improve use of communication channels and increase integration of departments in an organization. The lack of consideration of group process is another key factor in the limiting use of quantitative models for R&D project/portfolio selec-tion. Experiments by Souder (1974, 1975) and Helin and Souder (1974) take group processes into account and investigate methods to obtain group consensus in R&D project selection. Helin and Souder (1974) report on a Q-sort technique for qualitative group project selection. The procedure involves a cycle of ac t i v i t i e s : 1) each individual sorts items (projects) into five groups by "worth," 2) group discussion period, and 3) phase 1 is repeated again until concensus is reached. The criterion value (worth) determines the priority of each project. For the experiment-13 projects that were being undertaken were selected. Each was given a code number and t i t l e for description. No characteristics (i.e. measures or c r i t e r i a values) were given. This technique is feasible only where the projects are familiar and some underlying concensus exists to begin with. The experiment showed that the Q-sort technique was not very useful in project selection. The method was too imprecise and the procedure would not be useful when many projects are involved. Maclay (1974) in comments suggests that c r i t e r i a should be li s t e d , making the technique more useful for analyzing new projects that are not familiar to the participants. A rating table would be used. Each par-ticipant would enter ratings for a l l the projects for each of the c r i t e r i a , 36 the average rating would be calculated and placed in the center. Then an attempt would be made, without a formal model, to enter overall priority ratings for the projects (a concensus rating). This method too would be of limited usefulness due to the complexities of comparing many projects on many attributes without a formal model. Experience with these qualitative models would indicate that a mixture of modes would be useful. The number of comparisons necessary prohibits use of group processes for project evaluation. Group processes could be designed to reach concensus on organizational objectives and weighting schemes. This information could then be used with quantitative optimization models to select R&D portfolios. Souder (1974b, 1975b) describes some experiments aimed at obtain-ing group concensus for project selection c r i t e r i a . The f i r s t paper reports on an impact method for determining c r i t e r i a p r i o r i t y . The experi-ment used paired comparisons, group discussion and interaction to achieve concensus in R&D project selection in four organizations. Two of the four groups actually achieved concensus. The procedure had the following steps: In step 1 selection c r i t e r i a are solicited from each participant. In Step 2 individuals make paired comparison rankings of their own c r i t e r i a using a tableau with the cr i t e r i a forming both the rows and the columns. Column c r i t e r i a were compared with row c r i t e r i a . If a column criterion is preferred to the row, a "1" is entered; i f dominated a "0" is entered. The priority criterion is that with the highest number of ones. (Consis-tency is met i f the number of 1's equals the number of 0's.) In step 3 group paired comparisons are made. Steps 2 and 3 are done in the same session. Compared with the Delphi technique, the impact method made explicit use of group pressure in arriving at a concensus. 37 The second paper compared three techniques for achieving con-census: 1) the impact method (combined Delphi and interacting), 2) Delphi; and 3) interacting. All procedures begin with each individual developing a l i s t of c r i t e r i a . The three techniques are variations on a two phase procedure. Step 1: each individual ranks c r i t e r i a and states j u s t i f i c a -tions, this information is exchanged. Step 2: the l i s t s are compared in a group, the group interacts and prepares a group l i s t of c r i t e r i a and ranking. The impact method is a cycling of steps 1 and 2 carried on three times. The Delphi method is step 1 carried on three times. The inter-acting method in step 2 carried on three times. Nine R&D and marketing groups chosen at random participated in the experiment. Leaders were created for each group to control for leadership style (another variable). It was found that the impact model was best for achieving integration and/or concensus between R&D and marketing divisions. The test of the experi-ment would be better carried on with actual representatives of marketing and R&D from the same company participating as a group, given the existing roles and interactions concensus might not be as easily reached as with the random groups. Also given the number of variables (two leadership, two group structures and three modes of interaction) the sample was too small for any conclusions to be drawn. Analysis of values of alternative models for particular organizations and executive groupings is only a f i r s t step toward the study of implementation (or non-implementation) of proposed normative models. Clearly a more universal theory of diffusion and implementation is necessary. In the conclusion, an attempt is made to identify some of these key problem areas important for inducement policy design in the R&D f i e l d . 38 Conclusion This chapter has presented an inventory of the existing knowledge of patterns of R&D investment decisions, government impacts upon them and the normative proposals for improved decision making. If better f i t between target populations and either government inducement policies or management improvement strategies is desired, this mosaic of information is incomplete. There are four broad areas in which information is scarce but necessary for strategic design. They are: 1) i n f o r m a t i o n about t h e n a t u r e of s e l e c t i v e p e r c e p -t i o n p r o c e s s e s o f R&D d e c i s i o n making, 2) t h e o b j e c t i v e f u n c t i o n s ( e x p l i c i t and l a t e n t ) w hich g u i d e c h o i c e s among a I t e r n a t i v e .R&D i n v e s t -ment o p p o r t u n i t i e s , 3) t h e impact o f R&D upon t h e p o s i t i o n s of prime b a r g a i n i n g u n i t s i n o r g a n i z a t i o n s , and J 4) t h e impacts o f o r g a n i z a t i o n a l s t r u c t u r e and p r o -c e s s e s upon i m p l e m e n t a t i o n o f i n v e s t m e n t d e c i s i o n s . The f i r s t two categories of information relate to the question of what different decision units consider relevant in defining their problems and what they value. The last two categories relate to the organizational impact of R&D and the processes by which decisions are reached and implemented. These areas provide a focus for required addi-tional research aimed at improvements in R&D decision making. This dissertation attempts to contribute to the f i r s t three areas of research outlined above. The focus in Chapter 2 is upon the impact of R&D in shifting resource shares of labour, capital and energy in the total output. Chapter 3 focuses upon processes of information selection in R&D decision making, identifying what environmental conditions 39 are important to whom in making R&D investment decisions. Chapter 4 investigates multi-attribute preferences in R&D project selection. Each v.., . chapter draws some normative implications for R&D public policy. The postscript identifies promising areas of future research. Chapter 2 AN ASSESSMENT OF THE ECONOMIC IMPACT OF RESEARCH AND DEVELOPMENT Introduction Challenging arguments of limits to growth, some students of the future emphasize the role technological progress may play in the next decade. "The central stupendous truth about developed economies today is that they can have the kind and scale of resources they decide to have . . . i t is no longer resources that limit decisions, i t is the decision that makes resources" (Toffler, 1971, p. 15). This creation of resources affects the structure of production. The hypothesis that technological R&D ac t i v i t i e s contribute significantly to the position of prime bargaining units in the Canadian economy is the major focus of this chapter. This chapter presents a review of methods of incorporating R&D into the description of the technological relationship and explains some of the structural changes which can be attributed to these activities. In particular, shifts in e l a s t i c i t i e s of substitution among the major factors of production which can be attributed to technological changes will be investigated, and their implications for policies and modes of income allocation examined. The study of R&D impact must take account of the contribution of major primary factors of production as well as changes in their nature over time. Increases in efficiencies of production may not necessarily reflect pure technological changes but may be attributed to improvements in the qualities of production factors. Such improvements may result 40 41 from general processes of quality upgrading (e.g. increases in educational levels in the work force) or from movements of organizations on their learning curves. There are three interrelated methodological developments which are relevant to this study. ( 1 ) D e v e l o p m e n t o f m o d e l s w h i c h c a p t u r e t h e i n t e r -r e l a t i o n s h i p b e t w e e n p r o d u c t i o n f a c t o r i n p u t s a n d o u t p u t s . (2) T h e f r a m e w o r k f o r m e a s u r i n g i n p u t s a n d o u t p u t s . (3) E x t e n s i o n o f p r o d u c t i o n m o d e l s t o c a p t u r e t h e e f f e c t s o f p r o c e s s e s , t h e i m p a c t s o f w h i c h a c c r u e o v e r t i m e ( l e a r n i n g , t e c h n o l o g i c a l c h a n g e s , e t c . ) . Developments in these areas are reviewed in the following section. Sub-sequent sections on methodology and data construction provide an explana-tion of the procedures followed in this study. / Models which Capture Input-Output Relationships in Production: Production Functions Some of the major questions which stimulated the development of production functions were of the following types: . . . how w i l l o u t p u t c h a n g e w i t h c h a n g e s i n f a c t o r . a v a i l a b i I i t y ? . . . how w i l l i n p u t c o m b i n a t i o n s c h a n g e i n r e s p o n s e t o c h a n g e s i n i n p u t p r i c e s a n d f a c t o r a v a i l a b i l i t y ? While theoretically the production function should provide a relationship between quantities of inputs and outputs, practical considera-tion of data availability and theoretical requirements stemming from the need for aggregations dictate the use of values rather than physical quantities. 42 Four major classes of models are reported in the literature: 1. i n p u t - o u t p u t 2. C o b b - D o u g l a s 3. C . E . S . 4. T r a n s l o g ( p r o d u c t i o n o r c o s t ) The f i r s t three are generally well-known and therefore merit only a brief comment. The classical input-output or Leontief production model specifies s t r i c t proportions of inputs without providing accommodation for possible substitution among factors. Estimation of this form requires detailed data of interindustry flows. The data are not generally available in time series form. The Cobb-Douglas (1928) form enables substitution among inputs resulting from changes in input prices. However the el a s t i c i t y of sub-stitution is equal to one and constant for a l l input combinations. The form was developed to explain the observation that the total wage b i l l / output was constant over time. The form is given in 1. Y = AK aL B (1) where a, 3 > 0, a + 3 = 1, Y is the output, A is the efficiency parameter, K and L are capital and labour inputs respectively, and a and 3 are their respective shares in production. A generalized form does not require the constraint a + 3 = 1, but then the parameters cannot be interpreted as share proportions. Estimation of this form is relatively simple. Data are frequently available in time series form and input can either be 43 specified in value terms or in quantity terms. Least squares regression estimation is used on the log of (1). The relative simplicity of the estimation procedure and avail a b i l i t y of data explain to a large extent the high frequency with which this form is used. The C.E.S. (Arrow et al., 1961) function specifies a constant e l a s t i c i t y of substitution among factors. It has the form given in (2). Y = y 6k"r + (1-6)1/ -1/r (2) where y is the efficiency parameter, 6 is the distribution parameter. (0 < 6 < 1) and the elas t i c i t y of substitution a = 1/1+r (r > -1). It reduces to the Cobb-Douglas form when r ->- 0 and to the Leontief when r ->•«>.. This more general form was developed to satisfy the observation that sectoral data can best be explained by assuming different e l a s t i c i t i e s of substi-tution among factors depending on the industry and type of product (Arrow et al., 1961; Schwartz, 1964). The translog functional form is more general. It permits the elas t i c i t y of substitution to vary with changes in input prices and scale of production. The translog production and cost functions are asso-ciated with the Shephard duality theorem which states that technology may be equivalently represented by a production function or a cost function each satisfying certain regularity conditions (Shephard, 1953, 1970; Samuel son, 1953; Diewert, 1971 , 1974). The conditions on the production function are very general and were specified by Diewert (1971, p. 484) as follows: Given an N factor production function f such that y = f(x) where x = (x!,--*^) i s a vector 44 of variable inputs and y is the maximum output that can be produced using x. Assume f satisfies the following conditions: ( a ) f i s a r e a l v a l u e d f u n c t i o n o f x d e f i n e d f o r e v e r y x > 0^  a n d f ( x ) i s f i n i t e i f x i s f i n i t e . , ( E v e r y f i n i t e . b u n d I e o f . i n p u t s g i v e s r i s e t o a f i n i t e o u t p u t . ) ( b ) f ^ O j ^ ~ 0 a n d f i s a n o n d e c r e a s i n g f u n c t i o n o f x. ( Z e r o l e v e l s o f a l l i n p u t s r e s u l t s i n z e r o o u t p u t s a n d t h e r e i s f r e e d i s p o s a l . ) ( c ) f ( x n ) t e n d s t o p l u s i n f i n i t y f o r a t l e a s t o n e n o n n e g a t i v e s e q u e n c e o f v e c t o r s (x11).. ( E v e r y p o s i t i v e o u t p u t l e v e l i s p r o d u c i b l e by s ome i n p u t comb i n a t i o n . ) ( d ) f i s c o n t i n u o u s f r o m a b o v e . ( I f f i s c o n t i n u o u s i t i s c o n t i n u o u s f r o m a b o v e b u t f n e e d n o t b e c o n t i n u o u s a n d t h e r e f o r e t h e p r o d u c t i o n p r o c e s s may e x h i b i t d i s c o n t i n u i t i e s . ) ( e ) f i s a q u a s i - c o n c a v e f u n c t i o n . ( M a r g i n a l r a t e s o f s u b s t i t u t i o n i n c r e a s e w i t h r e s p e c t t o a n y s i n g I e f a c t o r . ) These conditions on the production function define corresponding conditions on production possibility sets and therefore i t is possible to derive the production function from the production possibility sets. (This is the usual relationship between a family of isoquants and a pro-duction function.) The production possibility sets or the production function in turn can be used to define a cost function C(y;p) = min{pTx:f(x) > y:x > 0} which satisfies the following conditions: x 45 C ( y ; p ) i s a p o s i t i v e r e a l v a l u e d f u n c t i o n d e f i n e d a n d f i n i t e f o r a l l y > 0, p » 0^ . ( E v e r y o u t p u t l e v e l h a s p o s i t i v e v f - i n i t e c o s t . ) C ( y ; p ) i s a n o n d e c r e a s i n g f u n c t i o n o f y a n d t e n d s t o p l u s i n f i n i t y a s y t e n d s t o p l u s i n f i n i t y f o r e v e r y p » 0. ( C o s t n e v e r d e c r e a s e s w i t h i n c r e a s i n g o u t p u t . ) C ( y ; p ) i s a n o n d e c r e a s i n g f u n c t i o n o f p . ( G e n e r a l l y c o s t i n c r e a s e s w i t h i n c r e a s i n g p r i c e s . ) C ( y ; p ) i s l i n e a r h o m o g e n e o u s i n p f o r e v e r y -y > 0. ( P r o p o r t i o n a l i n c r e a s e s i n a l l p r i c e s i n c r e a s e c o s t s by t h e same p r o p o r t i o n . ) C ( y ; p ) i s a c o n c a v e f u n c t i o n i n p f o r e v e r y y > 0. ( T h i s i s a n i m p l . i c a t i o n o f c o s t m i n i -m i z i n g b e h a v i o u r , e n a b l i n g t h e d e r i v a t i o n o f d o w n w a r d s l o p i n g demand c u r v e s . ) Similar.ly given C(y;p) satisfying the above conditions we can generate a family of production possibility sets. If a cost function C(y;p) satisfies conditions ( a 1 1 ) - ( e 1 1 ) and i s , in addition, d i f f e r e n t i a t e with respect to factor prices, then Shepard's Lemma indicates that where x.(y;p) is the cost minimizing quantity of input i needed to produce output y given factor prices p. The significance of the Shephard duality theorem i s that given C(y;p) satisfying conditions (a 1) - (e'), C(y;p) may be interpreted as ( a 1 ) Cb' ) (c' ) ( d 1 ) ( e ' ) 46 the total cost function of some underlying production function even i f the production function is not expressed ex p l i c i t l y . The duality provides two methods for deriving the input demand functions x^(y;p): 1. Conventional methods using production functions. This method requires specification of a functional form for f satisfying the regularity conditions (a) - (e), and solving the cost minimization problem f T } min -<-p x:f(x) > y:x > 0--2. Methods based on Shephard's Lemma. This method requires specification of a form for a cost function satisfying conditions (a') -(e 1) and derivation of demand functions by differentiation. Method 2 allows for solution without specification of afunctional form for the production function. A functional form is said to be 'flexible' i f i t provides a second order (Taylor Series) approximation to an arbitrary twice d i f f e r e n t i a t e function. Method 2 can be used to provide a f i r s t order approximation to the input demand functions. On the other hand, i f a flexible form is used with method 1 i t is usually impossible to solve for the derived demand functions as exp l i c i t functions of the unknown parameters of the production function. The translog form has been most generally estimated with the added assumption of constant returns to scale. Usually maximum likelihood or two stage least squares is used to estimate the share equations. Generally the factor demand equations are estimated instead of the cost function i t s e l f , or the inverse demand functions (based on the marginal productivity conditions) are estimated instead of the production function 47 i t s e l f in order to increase the.efficiency of the sta t i s t i c a l estimation property. The data required include quantity and price indices for inputs and outputs. The price indices are rarely available and must be estimated. Framework for Measuring Inputs and Outputs The estimation of production relationships presents problems of aggregation and measurement. Changes in the total value of an input or output incorporate price, quantity and quality effects. Several methods have been proposed for separating these effects. A) Aggregation Aggregate data are used in the estimation of sectoral production relationships. Some aspects of the aggregation problem will be discussed as features of the specific measurement problems (e.g. capital stock is an aggregate of plant and equipment). The general problem, however, relates to the construction of appropriate indices to measure quantities and prices that account for changes in the distribution of items in the composites. Most data are constructed as simple aggregates, e.g. the total wage b i l l is the sum of wages paid to production and non-production workers. Changes in the average wage rate will reflect changes in the composition of production and non-production workers. Thus a price index equal to the average wage will not be a reflection of changes in costs 48 alone. A simple arithmetic index with constant weights, such as the * Laspeyres and Paasche, would not capture the economic effect of substi-tution that may result from shifts in relative prices. These indices are exact for functional forms that do not allow substitution (e.g. Leontief). This problem can be solved by defining the ideal price index to be the geometric mean of the Laspeyres and Paasche indices as suggested by Fisher (Diewert, 1975a). (Generally the Laspeyres and Paasche indices provide bounds on the exact index, see Diewert, 1973a.) The exact price to use varies with the functional form being estimated. The Tbrnquist Index provides the best index for the translog * Laspeyres Price Index: base period quantities used DT = -J Paasche Price Index: current period quantities used v t t ? p i ^ i DT = J i where i = index over input, p^ = price of input i in time t, and qt" = quantity of input i in time t. ~ t A t - f s. + s. i i 2 where s.. = cost share of input i in total cost in time t. PIT = t-1 49 functional form (Diewert, 1975a; Theil, 1967; Christensen and Jorgenson, 1970; Berndt and Wood, 1975). The Divisia Index, is a continuous time Tornquist Index. B) Measurement of Output and Inputs (i) Output For two factor production functions, output is generally mea-sured as value added deflated by the appropriate sectoral price indices. When estimating translog functions, output is constructed as a Divisia Index-of factors, e.g., Berndt and Wood (1975) defined output as a Divisia Index of labour, capital, energy and materials. ( i i ) Labour The quantity of labour is an aggregate of heterogeneous types of labour. Changes in the average wage rate reflect changes^.in the com-position of labour as well as pure price changes. Definition of either a measure of quantity or a price for labour from wage b i l l information, requires adjustments for effective labour units. Improvements in labour reflect both learning curve effects and changes in education and s k i l l levels. The learning curve effect is generally l e f t as part of the residual (Kennedy and ThirlwalT, 1972, p. 38). One example of this is the 'Horndal effect 1 where production increased at an annual rate of 2% for 15 years even though there had been no new investment {ibid., p. 39). However, i t is hard to separate the learning effect from changes in management. 50 The return to education is also not easy to quantify. A number of studies including Schultz (1960), Bowman (1964), Denison (1962) and Becker (1964) have all found a positive rate of production increase associated with increased levels of education of the labour force. Denison estimated that education contributed 23% of the growth of output in the U.S. in the period 1929-56. Dension's results were obtained by assuming that differentials in wages reflect differentials in marginal productivity. These differentials were used to adjust the quantity of labour to derive a measure of effective labour units. Thus i f a college educated employee receives 60% more in wages, then that worker is 1.6 of a standard worker. Becker included on-the-job training and found a doubling of the impact of total education or productivity change. For a measure of labour input Griliches (1963) adjusted labour force data by an education index proportional to the years of schooling. This resulted in 14.4% increase in the quality of the agricultural labour force for the period 1940-60. Similar adjustments were made,,for changes in the distributions of age, sex and race. Berndt and Wood (1975) constructed a measure of labour services as a Divisia Index of production and non-production workers adjusted for quality changes by educational attainment. The price index of labour is the index of total compensation divided by the constructed index of labour. ( i i i ) Capital The measurement of capital provides conceptual problems regard-ing valuation, quantity, depreciation (gross versus net), and embodiment versus disembodiment of technical change (see Kennedy and Thirlwall, 1972). 51 a) F o r t h e v a l u a t i o n o f c a p i t a l i t i s assumed t h a t p r i c e r e f l e c t s t h e v a l u e i n p r o d u c t i o n d u r i n g t h e l i f e span d i s c o u n t e d f o r t i m e . . D e t e r m i n a t i o n of t h e a p p r o p r i a t e d i s c o u n t r a t e i s,:-a p r o b l e m . b) I t i s a l s o d i f f i c u l t t o d e r i v e a measure o f t h e q u a n t i t y o f c a p i t a l as t h e r e i s no s i n g l e p r i c e and i t i s not p o s s i b l e t o j u s t c o u n t machines t h e way we c o u n t w o r k e r s . C a p i t a l i n c l u d e s many h e t e r o -geneous c o m m o d i t i e s w i t h d i f f e r e n t l i f e spans and of d i f f e r e n t ages. c) G e n e r a l l y t h e r e i s o v e r - d e p r e c i a t i o n o f t h e c a p i t a l s t o c k i n c a p i t a l a c c o u n t i n g as p r i c e d e p r e c i a t i o n r e f l e c t s e x h a u s t i o n , d e t e r i o r a t i o n and o b s o l e s c e n c e ( S c h w a r t z , 1973). E x h a u s t i o n r e f e r s t o t h e d u r -a b i l i t y o f equipment, as equipment ages i t has a s h o r t e r r e m a i n i n g p r o d u c t i v e l i f e . D e t e r i o r a t i o n r e s u l t s i n a r e d u c t i o n o f the annual s e r v i c e f l o w w i t h age. O b s o l e s c e n c e i s t h e r e s u l t o f new more p r o d u c t i v e equipment. I t does not r e f l e c t a change i n t h e s e r v i c e f l o w but an i n c r e a s e i n r e p l a c e m e n t c o s t . C a p i t a l becomes o b s o l e t e b e f o r e i t d e t e r i o -r a t e s . T h e r e f o r e t h e f l o w o f c a p i t a l s e r v i c e s w i t h age does n o t d e c r e a s e as much as s u g g e s t e d by d e p r e c i a t i o n r a t e s . d) Solow (1962) and o t h e r s have argued t h a t g r o s s measures o f c a p i t a l a r e n o t adequate i f . t e c h n i c a l change i s embodied. With embodied t e c h n i c a l change o n l y new c a p i t a l b e a r s t h e improvement, w h i l e w i t h d i s e m b o d i e d t e c h n i c a l change, a l l c a p i t a l b e n e f i t s . The concept of embodied technical change presents a circular problem -the capital stock can be corrected only i f the technical change factor is known. Solow assumed a constant rate of technical change and defined effective capital stock as: v where is the effective capital stock in period t, K ^ is the amount of capital from period v s t i l l productive in period t (non-deteriorated), and A, is the rate of embodied technical change. 52 Denison (1964) suggested that the embodied/disembodied question is unim-portant providing that the age distribution of capital is constant. It appears that the age distribution has not changed significantly in the U.S. though in less developed countries this could be an important factor. The data requirements for estimating the role of capital in production vary with the functional form used. Cobb-Douglas and CES production functions can use capital stock data as inputs. Minasian (1969) used gross stock data. Griliches (1963) used a gross capital concept but assumed a 15 year productive l i f e . For each time period he defined capital as the sum of gross investment for the past '15 years. Translog functions require a service flow concept as the parameters are estimated from output or cost share relationships. Berndt and Wood (1975) constructed a rental price of capital taking account of variations in effective tax rate, depreciation, and capital gains:' Then they constructed a Divisia Index of the quantity of services. This follows the work of Christensen and Jorgenson (1969). (iv) Energy Energy may be considered an additional factor of produc-tion. Energy presents an aggregation problem as there are substitution poss i b i l i t i e s among energy sources. Berndt and Wood (1975) constructed a quantity index of energy as a Divisia Index of coal, crude petroleum, refined petroleum products, natural gas and e l e c t r i c i t y . The value of energy is the sum of purchases and the price index is total purchases divided by the quantity index. 53 (v) Material Inputs When the aim is to explain total production (sales plus inven-tory change) of a sector, a l l material inputs (produced inputs immediately consumed in production) are included. The main problem with incorporating material inputs is in deriving price or quantity indices to capture changes in relative proportions. Berndt and Wood (1975) constructed a Divisia quantity index of materials from inter-industry flow tables for ten categories of materials. Griliches (1963) derived a weighted plant nutrient measure for effective f e r t i l i z e r use. This measure indicated an increase in f e r t i l i z e r quality of 8% per year in the period 1947/49-1960. Modifications of Production Models to Capture Effects of Time  Dependent Processes There were four stages in the development of methodologies to study time dependent processes in production: a ) I n t e r p r e t i n g u n e x p l a i n e d r e s i d u a l s a s a c c r u i n g f r o m c h a n g e s i n t e c h n o l o g y b ) T r a n s f o r m a t i o n o f i n p u t m e a s u r e m e n t s t o a c c o u n t f o r i n p u t q u a l i t y u p g r a d i n g a n d a t t r i b u t i n g r e s i d u a l s t o i m p a c t s o f l e a r n i n g a n d t e c h n o l o g i c a l c h a n g e s c ) I n t r o d u c t i o n o f t i m e a s an e x p l i c i t v a r i a b l e i n t o t h e p r o d u c t i o n f u n c t i o n d ) I n t r o d u c t i o n o f R&D e x p e n d i t u r e s a n d o t h e r e x p l a n a -t o r y v a r i a b l e s i n t o t h e p r o d u c t i o n f u n c t i o n i n a n a t t e m p t t o i s o l a t e s p e c i f i c i m p a c t s o f t i m e d e p e n -d e n t p r o c e s s e s i n p r o d u c t i o n . Solow (1957) posited an aggregate production function but l e f t the form unspecified in order to test for neutrality of technical change. He 54 correlated changes in output with per cent changes in the capital labour ratio. As there was no significant correlation, the assumption was that technical change was neutral. He assumed also that there were constant returns to scale and that capital and labour were paid the value of their marginal products. The estimated ratio of technical change for U.S. manu-* facturing in the period 1909-1949 was found to be 1.5% per annum. He estimated that f u l l y 90% of the growth of output per man hour could be attributed to technical change (Solow, 1957 ; Mansfield, 1972, p. 3 2 ; Kennedy and Thirl wall, 1972 , p. 1 8 ) . This discovery of "free" productivity increase was trouble-some to economists d r i l l e d in the maxim that everything has a cost. To quote Abramovitz (Kennedy and Thirlwall, 1972 , pp. 1 7 - 1 8 ) : This result is surprising in the lopsided, importance which it appears to give productivity increase and it should be in a sense sobering if. not discouraging to students of economic growth. Since we know l i t t l e about the causes of productivity increase, the indicated importance of this element may be taken as some measure of our ignorance about the causes of economic growth in the United States and some sort of indication of where we need to con-centrate our attention. * The form obtained by Solow to explain productivity change was the following: £ = £ _ W £ p q \ k where p is a measure of technology, q is output per man, k is capital per man hour, is capital's share of income, and • represents time derivative. This is a Divisia Index for measuring total factor productivity. For proper estimation, the data must be available in continuous form. As economic data comes in discrete form, the Tornquist Index would be the appropriate measure of productivity (Diewert, 1975b) i f the underlying production function were translog. 55 Subsequent research was directed toward eliminating the residual by directly accounting for changes in factor productivity. Education improved the effectiveness of the labour force, and an accounting for this quality change was used to explain differential growth among the OECD countries (Denison, 1966). Accounting for changes in labour quality and other quality changes reduced the residual. Denison (1962) estimated that neutral productivity increases (advance in knowledge) accounted for 40% of the increase in output per man hour in the United States in the period 1929-57. Of this he assumed that research and development was responsible for perhaps 20% (Denison, 1966; Mansfield, 1971,. p. 33). This study greatly reduced the unexplained growth. Accounting for quality improvements in the capital stock also helped explain productivity growth (Jorgenson, 1966). Griliches (1963) argued that estimating simple regressions and then attributing a l l the residual to "technical progress" did l i t t l e to explain that dynamic element of the production process. He derived separate estimates of the labour and capital series that would account for improvements in these factors and then used this information in estimating the production relationship for U.S. manufacturing. This method l e f t very l i t t l e unexplained residual. In estimating these inputs, Griliches found a 1% per annum technical advance factor in both capital and in labour. Accounting for quality improvements in a l l factors, eliminated a l l the unexplained residual and all " f r e e " technical advances (Jorgenson and Griliches, 1967). But this final step lef t l i t t l e room for a contribution of R&D 56 to growth. By attributing a l l changes to factors, R&D would not be directly productive. Actually some changes in factor effectiveness result directly from R&D expenditures: i.e., new processes, new combinations of factors. Even improvements in the capital stock do not come "free." Upgrading the capital stock to account for productivity change does not answer the question of how technical progress is achieved. Technical progress depends on advances in knowledge (and R&D is one major factor in explaining such advances). The next focus of investigation therefore was the relationship between R&D and the rate of growth of^productivity in various sectors. Terleckyi (1960) foundthat the rate of growth of the factor productivity was significantly related to the ratio of R&D to sales. Griliches (1964) investigated the contribution of R&D in U.S. agriculture by incorporating R&D expenditures in the agricultural pro-duction function. He estimated the aggregate agricultural production function using cross sectoral data for three time periods. The data were per farm averages for 39 regions in the U.S. The functional form estimated i s a six factor Cobb-Douglas without pre-specification of returns to scale. Public expenditure for research and distribution of information were included to capture technological change. A lagged variable was defined consisting of the average of expenditure level in t-1 and t-6. The estimated coefficient was .059. This results in a marginal product of $ 1 3 for each additional dollar spent on R&D (a rate of return of 1300%. 57 Mansfield (1965) investigated the relationship between produc-t i v i t y change and the rate of growth of cumulated R&D expenditures: by firm and by industry (sample includes 10 petroleum and chemical firms and 10 manufacturing firms). He estimated the impact of R&D on the rate of productivity change under assumptions of embodied (applicable to new capital only) and disembodied (applicable to the entire capital stock) technical change. One per cent increase in accumulated R&D leads to a .1% increase i f embodied, and only .1% i f disembodied. Brown and Conrad (1967) also found a significant relationship between R&D and productivity increases in U.S. manufacturing industries. They found, as expected theoretically, that a given increase in R&D pro-duces a larger increase in productivity when spent in the durable goods sectors rather than consumer goods. Minasian (1969) estimated a Cobb-Douglas production function for the chemical industry during 1948-59 using data for 17 U.S. firms. The inputs are labour (deflated wage b i l l ) , capital (stock, deflated), R&D (stock, deflated) and output is determined by capital, labour and technology [V^ . = f(Lt» K^ , T^)]. Where technology is a function of the state of knowledge [T, = g(N )] and knowledge depends on expenditures t t t in R&D [N. = H( \ R +)]. Thus output is a function of labour, capital t=0 and the stock of R&D expenditures [v t = k(L t, k t, IRt)]- It was found that a firm's value added was significantly related to the firm's cumu-lated R&D expenditures. The marginal product of capital was .09 and of R&D, .54. That i s , i f the stock of capital increases by $1.00 the value added increases by $9 per year, while the same increase in R&D gives rise to an increase of $54. 58 Leonard (1971) found that R&D spending by firms is related to growth of sales with a two-year lag. This sample was based on sixteen U.S. manufacturing industries for the period 1960-67 (R&D to 1957). He also investigated whether federal contracts for R&D had equal impact on output. The results showed that growth was lower for firms in aerospace and electronics when federal funds were included. One explanation put forth is that federal R&D has noncommercial objectives while company R&D is aimed at increasing profit. Another related explanation might be that the concentration of R&D in defense-space areas is a misallocation of resources as the marginal productivity is higher in other industries. Binswanger (1974) using the cost function framework investigated the impact of technological change on agricultural output. He developed two models: (1) factor augmenting technical change where factor prices are corrected for quality changes and (2) neutral technical change where time serves as a proxy for technical change. Model 1 reduces to the normal translog function with adjustments for quality change in inputs; the rate of technical change can vary over time. Model 2 has constant biased technical change; the factor shares vary as a function of time (technical progress) as well as a function of prices. Both models assumed five input categories (land, labour, machinery, f e r t i l i z e r , other). The results indicated strong f e r t i l i z e r using bias induced by rapid decline in the price of f e r t i l i z e r relative to agricultural produce. Labour saving bias was observed after 1944, induced by more rapidly increasing labour costs after 1940. F e r t i l i z e r and labour biases are consistent with neutral innovation possibilities and the hypothesis that relative factor price changes account for biases in production methods. However technical 59 change was machinery using though price increases would have induced machinery saving techniques had the innovation possibilities been neutral. In summary, technical change and the impact of R&D can be con-sidered in each of the production frameworks. In the input-output mode technical change is reflected in discontinuous changes in the relative input proportions and is not implicit in the model. These changes are hard to explain and predict as there are normally few completed input-output tables for each country (Schwartz, 1964). Franchet et al. (1973) describe a method for updating input-output coefficients to account for technological change. The method is based on uniform row correction factors assuming an annual geometric rate of change. See also Bacharach (1970) for a method using row and column correction. In the Cobb-Douglas framework technical change is commonly incorporated in the efficiency parameter A. Alternative formulations (Griliches, 1964; Mi nasi an, 1969) incorporated R&D as a factor of produc-tion. But this is not s t r i c t l y valid as the implication of this treat-ment is that i t is necessary for positive output to have R&D expenditures. In the C.E.S. framework as with the Cobb-Douglas, technical progress is generally reflected in changes in the efficiency parameter. For the translog, technical change has been incorporated using time as a proxy shift parameter to change cost or output shares (Binswanger, 1974; Berndt and Wood, 1975). Methodology In this study the translog functional form was chosen to repre-sent technology. The translog form was f i r s t estimated by Christenson et al. 60 (1971). The translog form is a flexible form, that is i t does not r e s t r i c t the substitution possibilities between inputs. Being a Taylor series expansion in logarithms i t provides a second order approximation to an arbitrary technology. It is assumed that there exist sectoral production functions relating inputs and the stock of R&D to output. The technology can be estimated via the cost function: In C(y,p) = a 0 + £a. In. p. + ^ Y,-1" 9* In p. i ij + l&. In y In p i + e In y + i3(1n y ) 2 i + l\. In RD In p. + TT In RD + £p ln(RD): + co In y In RD t where i = L, K, E, RD = I RD (i.e., the stock of R and D in 'real' terms 0 1 at period t ) , and y is a Tornquist quantity index of inputs. We have I ]" C = a- + Ty. . In p. + S. I n y + A - InRD 3 In p. i y i j K j l J i 61 since }n C = ^ — . ^ - and by-Shephard1 s Lemma d ln p i 3p. C |~- = x- for all i . x. p. Therefore ^ - n = "L 1 , that is the logarithmic derivatives of C are the d in u shares of the input in the total costs. We calculate C as the sum of the inputs times their marginal productivities: c i x i p i i x. p. Defining the cost shares y. - 1 we get the input demand x. p. equations u- = n r 1 = a. + j^ y In p. + 6. In y + X- In RD. Imposing linear homogeneity in prices and symmetry implies the following parameter restrictions y a. = i yv.. = yY.. = o Y • - = Y • y 6. = o I I X. - 0 . (3) i (Linear homogeneity in prices is a condition for the duality of cost and production functions. Symmetry is required in order to interpret the share equations as demand equations.) Interpretation of the input demand equations: demand for an input (the input share) depends on prices, scale of production, and the 62 stock of R&D. The y.- are parameters reflecting the change in input share that results from a one per cent change in the prices of inputs. They are not s t r i c t l y measures of substitution versus complementarity as factor share can decrease though factor use increases. The 6.. are parameters reflecting the change in share that results from a one per cent change in scale df output. The are parameters reflecting the change in share that results from a one per cent change in the stock of R&D. Since the shares sum to one only two of the three equations are independent. In the estimation, one equation is dropped and i t s parameters are identified using the restrictions. For empirical implementation the model has to be embedded within a stochastic framework. It is assumed that the equations of the cost shares are stochastic due to errors in optimization. The additive dis-turbance in the ith equation is defined as e^(t), i=l,•••,n -1. The column vector of disturbances at time t is defined as e^, t=l,***,T. It is assumed that the vector of disturbances is independently and^identically joint normally distributed with mean vector zero and non-singular variance-covariance matrix ft, i.e. E He.(t)}- = 0, E i e(s), e(t) ft,t = s a l l s,t 0,t f s Maximum likelihood estimation is used. This ensures that the results are invariant to the choice of the equation dropped. The translog function with the restrictions in (3) imposes symmetry and linear homogeneity in prices, i t does not, however, ensure that the monotonicity and concavity conditions are satisfied. It is 63 necessary therefore to check whether these condit ions hold. Monotoni-c i t y i s s a t i s f i e d i f the ( f i t t e d ) unit demands are p o s i t i v e . Concavity of the cost funct ion i s s a t i s f i e d i f the Hessian matrix of the cost funct ion i s negative semi -de f i n i t e . The ca l cu l a ted A l l en p a r t i a l e l a s t i c i t i e s (AES) of s u b s t i t u -t i on are a measure of the subs t i tu t i on p o s s i b i l i t i e s in the model. These e l a s t i c i t i e s have been shown by Uzawa (1962) to be given by: 9 2C c / 9C 3C 1 f o r a l l i , j Where a.. = a . , by d e f i n i t i o n . For the t rans log funct ion the AES a re : 0 =hilh^i i j y i y d i f j f o r a l l i , j Y i i + y i " u i a = - J J J 1 i _ -j f o r a l l i Given the AES we can c a l c u l a t e the p r i ce e l a s t i c i t i e s which are given by ( A l l e n , 1938): E. . = a.. u • • Sources of Raw Data and Construct ion of the Var iab les Data were c o l l e c t e d from fourteen sectors o f Canadian manu-f a c t u r i n g as fo l lows: 64 1 - Food and Beverage Industries 2 - Rubber Industries 3 - Textiles Industries 4 - Wood Industries 5 - Furniture and Fixtures Industries 6 - Paper and Allied Industries 7 - Primary Metals Industries 8 - Metal Fabricating Industries 9 - Machinery Industries 10 - Transportation Equipment Industries 11 - Electrical Products Industries 12 - Non-metallic Mineral Products Industries 13 - Petroleum and Coal Industries 14 - Chemical and Chemical Products Industries The time series data cover the period 1961-1972. The data for each sector include the following: CAPITAL - the capital stock series for each industry has been calculated following Woodland (1972). Four types of capital are defined for each industry: building construction, engineering construction, machinery and equipment and operating capital. Woodland's stock data for 1961 are used to i n i t i a l i z e each series. The stock of each type of capital in each year is given as K J i f • : i : t where K... is the stock of capital type i in sector j in year t, I... is investment in 1961 dollars and is the depreciation rate. The invest-ment series is from Statistics Canada General Review of The Manufacturing Industries of Canada on tape on CANSIM. The depreciation rates are from Woodland (1972). To calculate the capital services used in production, a user cost or rental price is estimated. The procedure follows Woodland (1972), 65 Cummings and Diewert (1973) and Jorgenson and Griliches (1967). The method assumes that firms lease their capital. Leasing firms, under the assumption of perfect competition, earn the corporate after tax rate of return, r. Under static expectations (i.e., no expectation of capital gains), the following equality is satisfied: (1+r) Q... + u. [R... - v.. 6.. Q...] - R... = (1-6..) Q..+ \ ] i t t j i t j t j i w j i t J j i t v j i ' ^ j i t where Q.. is the price index of capital type i in industry j in year t, u t is the rate of taxation on capital earnings, R^-t is the "rental price of capital and v ^ is the proportion of depreciation allowable for tax purposes. This relationship states that the purchase price of capital plus one year's opportunity cost (interest) plus tax l i a b i l i t y adjusted for depreciation allowance minus the rental received is equal to the depreciated value of the capital good in the next period. (Only corporate profit tax and personal tax on dividends are considered.) Solving for the rental prices: R . i t - { r t Q j i t + ( 1 - U t V s j i t < i . i t } / 1 - u t The rate of return, r, was assumed to be the average of corporate midyear bond prices in the Canadian economy (McLeod et a l ) . The tax rate, u^ ., is taken to be the weighted average of the corporate tax rate,- u ^ , and the personal tax rate, where ~~* ~ In the case of operating capital, depreciation (turnover) is used only for calculating the stock and not for capital consumption purposes for taxes, so this reduces to: R j i t - v / 1 - V 66 u t = e t u u + (i - et) u 2 t and u^ t is corporate tax dividends divided by corporate profits in manu-facturing, u^ t is personal tax divided by personal income, 8^ is retained earnings (corporate profits) divided by the rental income of capital (value added minus the wage b i l l ) net of the capital consumption allowance, and V j t is capital depreciation divided by the capital consumption allowance. The necessary data are available in Statistics Canada publications of National Income. The cost of capital services is calculated as £ R..^  K j ^ . Given the rental prices and capital stocks, a Tornquist price index is calculated. ENERGY - A Tornquist price index of energy costs was calculated. The General Review referred to above and accessed through CANSIM gives sectoral "cost of fuel and elec t r i c i t y used." The indexes necessary to deflate these values were constructed as follows: For each sector, there is a Statistics Canada publication reviewing the total cost and quantities consumed of 15 energy types. Generally data was available for the industry as a whole. For some industries however, the aggregation level was too low, so the "miscellaneous" category within the sector was used. The fifteen energy types were reduced to seven: Coal (6 types), gasoline, fuel o i l , wood, gases, li q u i f i e d petroleum gases, and e l e c t r i c i t y . The categories steam and other fuel were eliminated as quantity data were not available. A Tornquist price index for energy was derived. For each sector, this was used to deflate the energy cost data. 67 LABOUR - the General Review gives sectoral data on total number of employees (N) and salaries and wages paid to total employees (W). To avoid double counting, the R&D employment and the R&D wage b i l l were netted out of these figures. Then the average wage (net W/net N) was com-puted and normalized to 1961. This value was used as the index to deflate the total wage b i l l . (Given the short time series, no quality adjustments have been made to the labour data.) RESEARCH AND DEVELOPMENT - R&D data were found in Statistics Canada's I n d u s t r i a l Research and Development Expenditures Catalogues #13-203, #13-509, #13-528, #13-520. R&D expenditures are available annually. A wage index for s c i e n t i f i c and technical personnel (Diewert, 1975b) was used to deflate R&D expenditures to constant dollars, as technically trained personnel represent a major portion of R&D costs. R&D employment and wages were also required for modifying the labour data. R&D employment is available biannually prior to 1969 and then annually. Total R&D wages were available for a l l years except 1962 and 1964. The R&D wage b i l l series was completed by using the average ratio of the R&D wage b i l l to total R&D expenditures and multiplying this ratio by the total R&D expenditures. For the R&D employment series, f i r s t the R&D wage rate series was completed by applying year to year changes in the wage rate for sc i e n t i f i c and technical personnel to the available data, and smoothing the trends. R&D employment for the missing years was then calculated as the R&D wage bill divided by the wage rate. OUTPUT - A Tornquist quantity index of inputs (capital, energy and labour) was calculated as the measure of scale of production. 68 All the data are reported in tables in the Appendix: Table Al includes the capital data, Table A2 the energy data. Table A3 the labour data, Table A4 the R&D data, and Table A5 the output data. The Results The input cost share equations were estimated using a TSP routine that imposes symmetry constraints. The programme did not converge for five sectors (food, rubber, wood, furniture, and metal fabricating) and yielded invalid results for the electrical products sector, indicating inappropriateness of the model for these six sectors. FOr the remaining eight sectors, the results are presented in Tables 2.1-2.8. There are two tables for each sector: (A) the regression results including statis-t i c a l tests, and (B) estimated values of input cost shares and price e l a s t i c i t i e s . The Durbin-Watson stati s t i c s indicate autocorrelated variables in four sectors: textiles, machinery, primary metals and transportation equipment. The R 2 ,s are a l l above .82. The own price e l a s t i c i t i e s were negative as required by economic theory except in the cases of machinery (capital positive 1961-65) and non-metallic minerals (capital positive 1961-64). In most sectors a l l resources are substitutes in production (positive cross e l a s t i c i t i e s ) except for textiles where energy and capital are complements and machinery where capital and energy are complements and labour and capital were complements in the period 1961-65. These results for the own price e l a s t i c i t i e s and cross price e l a s t i c i t i e s indicate that the necessary conditions for concavity are satisfied except for textiles where the sufficient deter-minantal conditions are met. Tar-ie 2.1 A: T e x t i l e s - Regression Results and S t a t i s t i c a l Tests TRACE OF MATRIX = 1.99962 LOG OF LIKELIHOOD FUNCTION RIGHT-HAND VARIABLE AL GLL GLK OLRO CLY AK GKK UKRD CKY 77.5212 LSTIMATcD COliFFi C IliNT 0.7210U5 0. U 2 f 0 7 -0.149U2U 0.51U/29i5-U2 -0.165728 0.228u80 0.163olO -0.129&22E-02 0.144974 STANDARD ERROR 0.205653E-U1 0.359081E-01 0.368925E-01 0.24U33E-01 0.312397 0.212733E-01 0.381301E-U1 0.249449E-01 0.323233 T-STATISTIC 35.0592 3. 68739 - 4 . 0 o l 2 0 0 . 2 U d 0 4 -0.530504 10.7471 4. 29083 -0.519632E-01 0.448512 . 2-TAUfc0 T-PRudAOILIT Y 0.0000 • O.00t>9 O.O042 0.0379 0.6111 O.uuOO 0.0031 0.9599 0.o665 EQUATION E05 * * * * * * * * * * * * * * * * * * DEPENDENT VARIABLE: YL3 EQUATION EQ6 ***i|.**v***4.t ****** DEPENDENT VARIABLE: YK3 R-SQUAREC = 0.8762 DUR8IN-WATS0N STATISTIC (ADJ. FOR 0. IMPS) = 1.1337 NUMBER OF OOSERVATICNS * 12. SUM OF SCUARED RESIDUALS • U.165816E-02 STANDARD ERROR OF THE REGRESSION » 0.117550E-01 SUM OF RESIDUALS » 0.596046c-0o MEAN OF DEPENDENT VARIABLE = 0.686107 R-SQUARED « 0.6055 DURBIN-HATSON STATISTIC (ADJ..FOR J . GAPS) => 1.1503 NUMBER OF OBSERVATIONS = 1 2 . SUM OF SQUARED ReSICUALS » 0.177od7E-02 STANOARD ERROR OF THE REGRESSION » 0.121635E-01 SUM OF RESIDUALS » 0.29U023c-0o MEAN OF DEPENDENT VARIABLE - 0.271039 Table 2 . 1 B: T e x t i l e s - Estimated Input Cost Shares and Pri c e E l a s t i c i t i e s ML MK ME 1961 0 . 7 2 1 0 0 5 0 .228680 U .503152E -01 1962 0 . 7 2 C 2 7 4 0 .229695 O . 5 0 0 J 1 2 E - 0 1 1963 0 . 7 1 9 3 9 8 0 .229845 o .5J7^68fc -01 1964 0 . 7 1 7 1 6 3 0 .232222 0 . 5 0 3 9 5 1 E - 0 1 1965 0 . 7 1 6 3 3 3 0 .233690 0 .499 7 76E-01 1966 0 . 7 0 0 7 1 6 0 .250627 U .4do569E -01 1967 0.651:779 0 .252152 u .490o88C -01 1963 0 . 6 9 0 4 1 9 0 .259997 0 .49a643 fc -Q l 1969 0 . 6 3 5 5 0 6 0.319528 U . 4 * 9 o 6 5 E - 0 i 1970 0 . 6 5 6 5 7 9 0.293341 U.400798E-01 1971 0 . 6 8 5 8 8 3 0 .263217 0 .5U6994E -01 1972 0 . 6 8 7 8 1 9 0 .261053 0.511<:a5c-Ql n e e j c i a s c i c i t x e f PSLL PSKK P SEE 1961 . - 0 . 9 5 3 5 2 7 E -01 - 0 . 5 5 U 6 6 7 E -01 - 1 . U 2 2 U 0 19o2 . - 0 . 9 5 8 9 7 4 E -01 - 0 . 5 8 0 1 3 1 c -01 - 1 . 0 2 2 6 9 1963 . - 0 . 9 6 5 4 9 3 E -01 - 0 . 5 U 3 2 8 2 E -01 - 1 . 0 2 0 9 3 1964 . - 0 . 9 8 1 9 5 6 E -01 - 0 . 6 3 2 3 6 1 E -Ox - 1 . 0 2 x 3 2 1965 . - 0 . 9 8 8 2 6 9 E -01 - 0 . 6 6 1 9 3 9 E -01 - 1 . 0 2 2 8 3 1966 . - 0 . 1 1 0 3 2 4 - 0 . 9 6 5 7 0 1 E -01 - 1 . J 2 O I 2 1967 . . - 0 . 1 1 1 7 3 8 - 0 . 9 8 9 9 4 3 c -01 - 1 . 0 2 5 0 8 1968 . - 0 . 1 1 7 8 0 3 - 0 . 110726 - l . O ^ j J O 1969 . - 0 . 1 5 6 1 4 5 - 0 . 1 6 8 4 3 5 - 1 . 0 J 5 9 5 1970 . - 0 . 1 4 U 3 7 1 - 0 . 1 4 8 9 1 3 - 1 . 0 2 / 0 0 1971 . - 0 . 1 2 1 0 7 1 - 0 . 1 1 5 2 0 6 - 1 . 0 i 0 5 9 1972 . - 0 . 1 1 9 6 7 8 - 0 . 1 1 2 2 1 6 - 1 . 0 2 0 0 4 Cross P r i c e E l a s t i c i t i e s P i L K P S L E P S K L P S K £ P S E L P S i K 1961 0 . 2 0 8 7 5 8 E -01 0 . 744768E -Ox 0 . 6 5 8 1 9 1 S - 01 - 0 . 9 9 5 2 7 6 E -02 1.06723 - 0 . 4 5 2 3 4 8 c -01 1962 0 . 2 1 6 7 9 8 c - O i 0.74217-+E -01 u . 6796 521;- 01 - 0 . 9 9 7 0 4 0 c -02 1.06847 - 0 . 4 5 7 7 t 4 E -01 1963 0 . 2 1 5 7 6 8 c -01 0 . 749725E -01 0 . 6 7 3 J 3 7c- 01 - 0 . 9 2 0 5 o 3 E -02 1.06262 - 0 . 4 1 6 6 & 5 E -01 1964 0 . 2 3 3 1 0 0 E i o i 0.748855E -01 0.7 I 9 8 9 6 E - 01 - 0 . 8 7 5 3 7 2 E -02 1.06150 - 0 . 4 0 1 7 7 9 c -01 1965 0 .2453006 -01 0.74296UE -Ox 0 . 751921E- 01 - 0 . b 9 9 d 4 2 c -02 l . 0 o 4 9 0 - 0 . 4 2 0 7 5 o E -01 1966 0 . 3 6 8 0 5 8 E -01 0 .735182E -01 0.1 0 * 9 0 4 - 0 . 6 3 3 3 5 2 E -02 1.05875 - 0 . 3 2 6 2 3 3 E -01 1967 0 .377385C -01 0 .739990c -01 0 . 104a83 - 0 . 5 5 8 8 9 4 C - 0 2 1 . 0 5 J O O - 0 . 2 8 7 2 0 2 E -01 1968 0 . 4 2 9 8 6 5 E -01 0.7 '481 v 63£ -01 0 .114150 - 0 . 3 4 2 4 4 3 E -02 1.04175 - 0 . 1 7 9 3 C 1 E -01 1969 0.337659..- -01 0. 723788E - O i 0 . l o o o O l 0 .183384c -02 1.02292 0 . 1 3 0 J U E -01 1970 0 . 6 5 8 3 9 4 E -01 0 .745317E - O i . 0.1096&7E -02 1.02091 0.OO9091E -02 1971 0 . 4 4 7 7 2 2 E -01 0 .762983E -01 U . l l u u 6 6 - 0 . 1 4 6 0 6 4 E -02 1.02314 - 0 . 7 5 3 3 4 2 E - 0 2 1972 0 . 4 3 2 2 2 2 E - 0 1 0 .764560E -01 0 . 1 x 3 6 8 1 - 0 . 1 6 6 5 7 2 c -02 1.02854 - 0 . 8 5 0 4 8 8 E -02 o Table 2.2 A: Paper and A l l i e d Industries'- Regression Results and S t a t i s t i c a l Tests TRACE OF MATRIX 1.99996 LOG OF LIKELIHOOD FUNCTION RIGHT-HAND VARIABLE AL GLL GLK DLRO CLY AK GKK OKRO CKY 79.9742 ESTIMATED COEFFICIENT 0.523200 0.556S05c-Oi -0. 723t>12i;-01 0.139J65E-01 -0.275510 0.347906 0.101359 -0.179915E-01 0.361J05 STANUARJ ERROR 0.102725E-01 0.19S850E-01 0.139149E-01 0.364397E-02 0.3O6117E-01 0.108615E-01 0.151790E-01 0.379947E-02 0 . 3 3 5 0 0 4 E - 0 1 STATISTIC 50.9323 2.841U9 -5.20025 3.81630 -7.52542 32.0310 6.69073 -4.73526 10.7851 2-TAILcD T-PRO&ABILITY' 0.0000 0.0232 0.0010 0.0058 0.0001 0.0000 0.0002 0.0018 C.OOQO EQUATION EU12 DEPENDENT VARIABLE: YK6 EQUATION E Q U DEPENDENT VARIABLE: YL6 R-SQUAREO » 0.9914 DURQIU-WATSON S T A T I S T I C ( A D J . FOR 0. GAPS) » 2.60f2 NUMBER OF OBSERVATIONS » 12. v-SUM OF SQUARED RESIDUALS = U.143336E-03 STANDARD ERROR OF THE REGRESSION « 0.345610E-02 SUM OF RESIDUALS •> -0.119209c-06 MEAN OF OEPENDENT VARIABLE = 0.515227 R-SOUARED = 0.5945 DURBIN-WATSON STATISTIC (AOJ. FOR 0. vJAPSJ » 2.1314 NUMBER OF OBSERVATIONS * 12. SUM OF SQUARED RESIDUALS » 0.1700O4E-U3 STANDARD ERROR OF THE REGRESSION '» 0.376457E-1 SUM UF RESIDUALS « 0. 715256c-0o MEAN OF DEPENDENT VARIABLE « . 0.3i>1352 Table 2.2 B: Paper and A l l i e d Industries - Estimated Input Cost Shares and Price E l a s t i c i t i e s Cost Shar 3S ML KK Mc 1961 0 . 5 2 3 2 0 0 0 .347906 O.12do94 1962 0 . 5 2 7 4 2 5 0 .341964 i>.130612 1963 0 . 5 2 7 3 2 3 0 .342194 O.130<*84 1964 0 . 5 2 5 3 1 5 0 .345156 0 . I ^ v 5 i 9 1965 0 . 5 2 1 9 2 4 0 .350008 0.1<:o067 1966 0. 516736 0 .357654 0 . 1 2 5 u l 0 1967 0 . 5 1 5 4 6 4 0 .359776 u. 124/61 1968 0 .513U14 0 .363855 0.12JJ.31 1969 0 . 4 8 9 0 0 3 0 .397962 0. i i .30o5 1970 0 .497014 0 .307246 J . 1 1 5 / 4 0 1971 0 . 5 0 7 2 0 9 0 .372629 . U.120162 1972 0 .510282 0 .369218 0. l«£Ut99 Own P r i c e E l a s t i c i t i e s . PSLL PSKK PSfcc 1961 . -0.370420 -0.360180 -O .7 7tio7 1962 . -0.367047 -0.361050 -0.7/J724 1963 . -0.367128 -0.361019 - 0 . 7/J759 1964 . -0.368733 -0.360604 -v.7/4008 1965 . -0.371435 -0.359831 - J . 7 / f J 6 9 1966 . -0.375553 -0.358388 -0.7/4917 196 7 . -0.376559 -0.357941 -0.7/5089 1968 . -0.378493 -0.357027 -0.775->93 1969 . -0.397177 -0.346841 -0.77o426 1970 . -0.351000 -0.350495 -0.7/6304 1971 . -0.383056 -0.354625 -0.7/5855 1972 . -0.38C644 -0.355718 - 0 . 7/5&09 Cross P r i c e E l a s t i c i t i e s P S L K P S L E ps*E PSKL PSEK 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 O.2C5601 0.160318 U.3i5209 0.449710c- 01 0.652784 0.12138t 0.204766 0.162280 0.3x5820 ., 0.452300c-01 0.655305 0.1x8420 0.204970 V 0.162158 0.315660 0.45i552c-01 C.655528 0.116430 0.207408 0.161324 U. 315oo7 0.449370c- 01 0.654264 0.119744 0.211365 0.160069 0.315183 0.446478E- 01 0.652545 0.1<:2023 0.217619 0.157933 0.31't4l4 0.439740c- 01 0.O49708 0.125209 0.219395 0.157164 0.314335 0.436J59L- 01 0.649^42 0.125747 0.222604 0.155688 0.3x4x40 0.428860E- 01 0.643604 0.126729 0.249985 0.147192 0.307173 0.396679c- 01 0.636767 0. 139658 0.241654 0.149346 0.3x0153 0.403419c- 01 0.641327 0.134977 0.229964 0.153092 0.3x3018 0.418064E- 01 0.646210 0.129644 0.227412 0.153231 U.314*98 0.414201E- 01 0.048894 0.X26914 Table 2.3 A: Primary Metals - Regression Results and S t a t i s t i c a l Tests TRACE OF MATRIX 1.99973 LOG OF LIKELIHOOD FUNCTION » RIGHT-HAND VARIABLE AL GLL GLK OLRD CLY AK GKK DKRD CKY 7 2 . 3 4 2 8 E S T I M A T E D C u E F F i C ItUT 0 . 5 4 4 0 2 7 0 . 9 4 2 - » 6 9 c - 0 1 - 0 . 1 1 O J 0 6 0 . 1 5 2 7 1 6 E - 0 1 - 0 . 1 7 1 0 1 6 0 . 3 4 6 3 5 3 0 . 1 4 4 1 9 5 - 0 . 1 4 5 J . 4 9 E - 0 1 0 . 1 7 7 2 4 3 STANOARO ERROR 0 . 3 0 2 4 8 4 E - 0 1 0 . 1 7 4 U 0 2 E - 0 1 0 . 2 0 5 3 0 4 E - 0 1 0 . 1 1 0 7 9 7 E - 0 1 0 . 8 9 0 0 < » 2 E - 0 1 0 . 4 0 9 0 9 0 E - 0 1 0 . 2 7 5 6 5 7 E - 0 1 0 . 1 4 9 8 5 0 E - 0 1 0 . 1 2 0 1 3 8 STATISTIC 1 7 . 9 8 5 3 5 . 3 9 1 0 5 - 5 . 6 6 5 0 5 1 . 3 7 3 3 4 - 1 . 9 2 1 4 3 8 . 4 7 1 3 1 5 . 2 3 0 9 6 - 0 . 9 6 8 6 2 7 1 . 4 7 5 3 3 2 - T A I L E D T -P R 0 8 A B I L I T Y 0 . 0 0 0 0 0.UU08 . 0 . 0 0 0 6 0 . 2 0 7 8 0 . 0 9 3 3 0 . 0 0 0 0 0 . 0 0 1 0 0 . 3 6 2 9 0 . 1 8 0 8 EQUATION EQ13 #**»«****#•***«*** OEPENObNT VARIABLE: YL7 EOUATION EQ14 * * * * * * * * + * * t n c 4 : * * * DEPENDENT VARIABLE: YK7 R-SQUARED « 0.5500 OURBIN-WATSON STATISTIC (AOJ. FOR J . GAPS) - 1.5470 NUMBER OF OBSERVATIONS « 12. SUM OF SCUARED RESIOUALS - U.5^9373E-03 STANDARD ERROR OF THE REGRESSION « 0.664187E-02 SUM OF RESIDUALS • 0.238419c-0o MEAN OF OEPENOENT VARIABLE » 0.549037 R-SOUAktO * 0.5401 DURBIN-WATSON STATISTIC (ADJ. FOR J . GAPS) » 1.4129 NUMBER OF OBSERVATIONS » 12. SUM OF SOUARcO RESICUALS * 0.964632E-03 STANDARD ERROR OF THE REGRESSION » 0.896582E-Q2 SUM OF RESIDUALS » 0.119209c-06 MEAN OF DEPENDENT VARIABLE « 0.351864 CO Table 2.3 B: Primary Metals - Estimated Input Cost Shares and Price E l a s t i c i t i e s }ost Shares 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 ML MK Mc 0.544027 0.346553 '0. 109420 0.546230 0.343865 0.109905 0.544832 0.345622 0.109346 0.539983 ' 0.351652 0.106365 0.534866 0.357977 0. 107i56 0.519280 0.377354 0.10336O 0.517811 0.379191 o.102997 0.512694 0.385537 0.101719 0.469851 0.438758 0.9x3907c 0.49C823 0.412792 0.9ood49E 0.509090 0.390163 0.1007*6 0.511898 0.386736 J.101306 Own Price E l a s t i c i t i e s PSLL PSKK P SCc 1961 . -0.282752 -0.237363 -U.837.j65 1962 . -0.231248 -0.236799 -0.837*35 1963 . -0.282203 -0.237174 -U.0-17320 1964 . -0.285499 -0.238298 -0.83/921 1965 . -0.288946 -0.239218 -0.000324 1966 . -0.299244 -0.240525 -0.040322 1967 . -0.300198 -0.240539 -J.040t90 1968 . -0.303499 -0.240451 -U.d4i.U5B 1969 • -0.329581 -0.232598 -0.844920 1970 . -0.317179 -0.23 7892 -0.0^3^25 1971 . -0.305801 -0.240260 —j.841478 1972 . -0.304C09 -0.240413 -0.841212 Cross Price E l a s t i c i t i e s PSLK PSLE PSKL PSKE PSEL PSe'K 1961 1962 1963 1964 1965 1966 1967 1960 1969 1970 1971 1972 0.132766 0.130941 0.132152 0.136264 0.140529 0.153379 0.154531 0.150735 0.191221 0.175831 0.161705 0.159531 0.149986 0.150307 0.150051 0.149235 0.148417 0.145865 0. 14561.7 0.144764 0.133360 0.141348 0.144096 0.144477 J . 2 0 8 4 2 0 0 . 2 0 d 0 0 0 0 . 2 0 8 3 2 1 0 . 2 0 9 ^ 4 2 0 . 2 0 9 9 6 9 0 . 2 1 1 0 6 6 0 . 2 1 1 U 9 1 l i . 2 i . 1 0 O i . u . 2 0 4 / 7 2 0 . 2 0 9 0 6 9 0 . 2 1 0 9 9 5 0 . 2 1 1 1 6 2 0.209434E-01 0.287991E-01 0.288525E-01 0.29055oE-01 0.292400E-01 0.294586E-01 0.294475E-01 0.295896C-01 0.278262E-01 O.288220E-01 0.29265OE-01 0.2925U6E-01 0.745715 0 . 7 4 / 0 2 9 0.746233 0.743634 0 .740815 0 .732780 0 . 732077 0 .729o50 0.711328 0 .719780 0 .728143 0 . 7 2 9 o l 3 0.91o6SSE-01 0.901052L-01 0.910311E-01 0. 942672E-01 0.977066E-01 0.107543 0.10d413 0.1i.l<i07 0.133591 0.123437 0.113335 0.111599 Table 2.4 <V: Machinery - Regression Results and S t a t i s t i c a l Tests T R A C E OF M A T R I X = 1 . 9 9 9 6 7 L I K E L I H O O D F U N C T I O N - 0 3 . 8 6 5 0 R I G H T - H A N D c S T M A T E U S T A N D A R D T - 2 - T A I L E U T -V A R I A B L E C O E F F i C I c N T E R R O R S T A T 1 S T I C P R O f l A B I L I T Y A L 0 . 9 2 7 ^ 1 2 0 . 1 6 3 2 1 9 E - O 1 5 6 . 8 2 0 0 0 . 0 0 0 0 G L L 0 . 5 4 9 o 5 6 e - U l 0 . 2 2 7 3 3 7 E - 0 1 2 . 4 1 8 1 5 0 . 0 4 3 9 G L K - 0 . 6 6 5 d 5 7 f c - U l 0 . 1 6 7 7 9 6 6 - 0 1 - 3 . 9 6 8 2 5 0 . 0 0 4 7 D L R D - 0 . 1 7 3 : > 6 7 c - 0 1 0 . 7 6 6 4 8 0 E - 0 2 - 2 . 3 2 9 7 0 0 . 0 5 0 3 C L Y 0 . 1 2 3 6 4 2 0 . 3 9 2 6 2 9 E - 0 1 3 . 2 7 6 4 3 0 . 0 1 2 3 A K 0 . 4 6 5 4 6 1 E - 0 1 0 . 1 4 0 9 7 0 E - 0 1 3 . 3 0 1 8 5 0 . 0 1 1 9 G K K 0 . 7 U 2 6 9 E - O 1 0 . 1 4 4 8 3 6 E - 0 1 4 . 9 1 0 8 5 0 . 0 0 1 4 D K R D 0 . 1 8 7 5 8 4 E - 0 1 0 . 6 5 6 1 6 8 E - O 2 2 . 8 5 5 7 4 0 . 0 2 2 8 C K Y - 0 . 1 2 1 6 5 9 0 . 3 4 1 3 3 1 E - 0 1 - 3 . 5 7 0 1 2 0 . 0 0 8 1 E Q U A T I O N E Q 1 7 * * * * * * * * * * * * * * * * * * D E P E N D E N T V A R I A B L E : YL9 E Q U A T I O N E Q 1 8 * * * * * * * * * * * * * * * * * * D E P E N D E N T V A R I A B L E : Y K 9 R - S Q U A R E D = 0 . 9 4 1 8 0 U R 8 I N - W A T S O N S T A T I S T I C I A D J . F O R 0 . G A P S ) = 1 . 3 3 7 7 N U M B E R O F O B S E R V A T I O N S • 1 2 . S U M OF S Q U A R E D R E S I D U A L S « O . 3 0 0 1 0 4 E - 0 3 S T A N D A R D E R R O R OF T H E R E G R E S S I O N » O . 5 0 O 0 8 7 E - O 2 S U M OF R E S I D U A L S = 0.0 M E A N . O F O E P E N D E N T V A R I A B L E • 0.868667 R - S Q U A R E D = 0 . 5 5 9 1 D U R 8 1 N - W A T S C N S T A T I S T I C I A D J . F O R 0 . G A P S ) = 1 . 4 3 8 7 N U M B E R OF O B S E R V A T I O N S « 1 2 . S U M OF S Q U A R E D R E S I D U A L S = 0 . 2 3 6 7 6 6 E - 0 3 S T A N D A R D E R R O R OF T H E R E G R E S S I O N * 0 . 4 4 t l 9 0 E - 0 2 S U M O F R E S I D U A L S * Q . 2 3 8 4 1 9 e - 0 o M E A N O F D E P E N D E N T V A R I A B L E « ' 0 . 1 1 0 3 6 3 Table 2.4 B: Machinery - Estimated Input Cost Shares and Price E l a s t i c i t i e s Cost Shares ML MK HE 1 9 6 1 0 . 9 2 7 4 1 2 0 . 4 6 5 4 6 1 E - 0 1 0 . 2 o 0 4 2 2 E - 01 1 9 6 2 0 . 9 2 1 6 8 2 0 . 5 2 0 0 3 7 E - 0 1 u . 2 o 3 1 4 2 E - 0 1 1 9 6 3 0 . 9 1 9 2 7 6 0 . 5 3 9 7 7 2 E - 0 1 0 . 2 6 7 W 1 E - 01 1 9 6 4 0 . 9 1 2 9 2 7 0 . 6 0 7 1 3 3 c - 0 1 0 . 2 o 3 a 4 2 E - 01 1 9 6 5 0 . 9 0 2 0 8 7 0 . 7 2 1 0 3 7 E - 0 1 0.23OO94fc- 0 1 1 9 6 6 0 . 8 9 2 0 0 5 0 . 8 2 2 0 0 7 E - 0 1 0 . 2 5 7 9 4 0 c - 01 1 9 6 7 0 . 8 8 9 9 2 7 0 . 8 4 0 1 4 3 E - 0 1 0 . 2 6 0 5 9 0 c - 01 1 9 6 8 0 . 8 8 3 2 4 1 0 . 9 0 3 1 0 0 E - 0 1 0 . 2 o 4 * 9 1 E - 01 1 9 6 9 0 . 8 5 8 2 9 7 0 . 1 1 6 4 5 0 0 . 2 5 2 3 3 1 c - 01 1 9 7 U 0 . 8 6 1 2 1 3 0 . 1 1 2 6 4 5 0 . 2 6 1 4 1 2 E - 01 1 9 7 1 . 0 . 8 6 8 1 6 8 0 . 1 0 5 0 7 7 0 . 2 6 7 5 5 0 E - 01 19 72 0 . 8 7 1 5 5 1 • 0 . 1 0 1 1 2 7 U . 2 7 3 2 2 0 E - 01 Own P r i c e E l a s t i c i t i e s 1 9 0 1 . - 0 . 1 3 2 9 9 0 E - 0 1 0 . 5 7 4 6 4 2 - 1 . 2 4 5 0 1 1 9 6 2 . - 0 . 1 8 6 6 0 1 E - 0 1 0 . 4 1 9 7 3 1 - 1 . 2 4 1 9 4 1 9 6 3 . - 0 . 2 0 9 1 0 3 E - 0 1 C . 3 7 1 6 9 9 - 1 . 2 3 7 1 6 1 9 6 4 . - 0 . 2 6 8 4 2 6 E - 0 1 0 . 2 3 2 1 4 2 - 1 . 2 4 1 4 9 1 9 6 5 . - 0 . 3 6 9 5 9 4 c - 0 1 0 . 5 8 5 5 5 8 E - 0 1 - 1 . 2 4 7 6 9 1 9 6 6 . - 0 . 4 6 3 5 2 1 E - 0 1 - 0 . 5 2 5 1 6 2 E - 0 1 - 1 . 2 4 7 8 7 1 9 6 7 . . - 0 . 4 8 2 8 6 7 E - 0 1 - 0 . 6 9 3 8 0 3 E - 0 1 - 1 . 2 * 4 8 2 1 9 6 8 . - 0 . 5 4 5 0 4 8 E - 0 1 - 0 . 1 2 2 1 0 3 - l . i 4 0 4 3 • 1 9 6 9 . - 0 . 7 7 6 3 9 8 E - O 1 - 0 . 2 7 2 7 5 8 - 1 . 2 5 4 2 7 1 9 7 0 . - 0 . 7 4 9 3 9 9 E - 0 1 - 0 . 2 5 5 9 3 1 - 1 . 2 4 J 8 8 1 9 7 1 . - 0 . 6 8 4 9 7 1 E - 0 1 - 0 . 2 1 8 0 2 1 - 1 . 2 3 7 0 8 1 9 7 2 . - 0 . 6 5 3 5 9 8 E - O 1 - 0 . 1 9 5 5 3 1 - 1 . 2 3 1 0 3 ?'< ' E l a s t i c i t i e s P S L K P S L E P S K L P S K b P S E L P S E K . 1 9 6 1 . - 0 . 2 5 2 5 1 2 E - 0 1 C . 3 8 5 5 0 2 E - 0 1 - 0 . 5 0 3 1 2 0 - 0 . 7 1 5 2 3 0 E - 0 1 1 . 3 7 2 8 5 - 0 . 1 2 7 8 3 6 1 9 6 2 . - 0 . 2 0 2 3 9 9 E - 0 1 0 . 3 8 9 0 0 0 E - 0 1 - 0 . 3 5 3 7 2 0 - 0 . 6 1 0 1 1 8 E - 0 1 1 . 5 o 2 5 1 - 0 . 1 2 0 5 7 5 1 9 6 3 . - 0 . 1 8 4 5 5 6 E , - 0 1 0 . 3 9 3 6 5 8 = - 0 1 - 0 . 3 x 4 3 1 3 '-. - 0 . 5 7 3 3 6 2 E - 0 1 1 . 3 5 2 9 7 - 0 . 1 1 5 6 0 8 1 9 6 4 . - 0 . 1 2 2 1 0 1 E - 0 1 0 . 3 9 0 6 0 7 E - 0 1 - 0 . 1 8 5 7 0 5 - 0 . 4 8 4 3 8 3 E - 0 1 1 . 3 5 3 0 9 - 0 . x i l 5 9 8 1 9 6 5 . - 0 . 1 / C 9 2 1 E - 0 2 0 . 3 3 6 6 8 6 E - 0 1 - 0 . 2 1 3 6 J 3 E - 01 - 0 . 3 7 1 7 3 1 E - 0 1 1 . 3 5 1 5 4 - 0 . 1 0 3 8 5 1 I960 0 . 7 5 5 3 5 1 E - 0 2 0 . 3 8 7 9 8 5 E - 0 1 0 . 8 1 9 6 7 3 E - 0 1 - 0 . 2 9 4 5 2 2 E - 0 1 1 . 3 4 1 7 2 - 0 . 9 3 8 5 3 5 E - 0 1 1 9 6 7 0 . 9 1 9 2 7 3 E - 0 2 0 . 3 9 0 . 9 3 B E - 0 1 0 . 9 7 j / f 6 E - 0 1 - 0 . 2 7 9 9 - W E - O 1 1 . 3 5 5 0 7 - 0 . 9 0 2 5 4 9 E - 0 1 1 9 6 8 0 . 1 4 9 2 2 1 E - 0 1 0 . 3 9 5 8 2 7 E - 0 1 0 . 1 4 5 9 3 9 - 0 . 2 3 8 3 6 3 E - 0 1 1 . 3 2 1 6 2 . - 0 . 8 1 3 8 8 5 c - 0 1 1 9 6 9 0 . 3 8 8 7 1 5 E - 0 1 0 . 3 8 7 6 8 3 c - 0 l 0 . 2 d o 5 0 2 - 0 . 1 3 7 4 4 5 E - 0 1 1 . 3 1 7 o 5 - O . o 3 3 6 0 4 c - 0 1 1 9 7 0 0 . 3 5 3 2 9 3 E - 0 1 0 . 3 9 6 1 0 6 E - 0 1 0 . 2 7 0 1 0 5 - 0 . 1 4 1 7 3 6 E - 0 1 1 . 3 0 4 9 6 - 0 . O 1 0 7 5 9 E - 0 1 19 71 0 . 2 8 3 8 O 5 c - 0 i 0 . 4 0 U 6 5 E - 0 1 0 . 2 3 4 4 8 5 - 0 . 1 6 4 6 3 5 E - 0 1 1 . 3 0 x 7 3 - 0 . 6 4 o 5 8 5 E - 0 1 1 9 7 2 0 . 2 4 7 2 8 1 E - 0 1 0 . 4 0 6 3 1 7 E - 0 1 . 0 . 2 x 3 x 1 6 - 0 . 1 7 5 8 4 6 E - 0 1 1 . 2 9 6 1 2 - 0 . o 5 0 3 o l S - 0 1 Table 2.5 A: Transportation Equipment - Regression Results and S t a t i s t i c a l Tests T R A C E OF M A T R I X = 1 . 9 9 9 8 4 : L I K E L I H O O D F U N C T I O N * 8 2 . 7 5 0 1 R I G H T - H A N O V A R I A B L E E S T I M A T c D C J c F F I C I c N T S T A N D A R D E R R U R T -S T A T I S T I C 2 - T A I L E D T -P R 0 8 A 8 I L I T Y A L 0 . 6 6 4 4 9 8 0 . 6 0 8 4 3 7 E - 0 1 8 . 2 1 9 5 3 0 . 0 0 0 1 G L L O . U B l 2 7 i i : - 0 1 0 . 3 8 1 7 6 1 E - 0 1 2 . 3 0 8 4 3 0 . 0 5 1 9 G L K - 0 . 9 3 3 7 5 6 E - O 1 0 . 3 7 6 8 3 3 E - 0 1 - 2 . 4 7 7 9 0 0 . 0 4 0 1 D L R D 0 . 4 3 1 U 3 2 E - 0 1 0 . 2 4 6 6 0 0 E - 0 1 1 . 7 4 7 9 0 0 . 1 2 1 1 C L Y - 0 . 2 7 0 d 8 7 0 . 1 3 8 8 3 6 - 1 . 9 5 0 4 2 0 . 0 8 9 4 A K 0 . 2 9 7 8 1 2 0 . 7 9 9 4 1 0 E - 0 1 3 . 7 2 5 4 0 0 . 0 0 6 5 G K K 0 . 9 8 3 0 3 8 E - 0 1 0 . 3 7 2 6 2 4 E - 0 1 2 . 6 3 8 1 5 0 . 0 3 1 5 D K R O - 0 . 4 0 3 3 9 9 E - 0 1 0 . 2 4 3 8 4 5 E - 0 1 - 1 . 6 5 4 3 3 0 . 1 3 9 2 C K Y 0 . 2 6 1 7 9 0 0 . 1 3 7 3 4 7 1 . 9 0 6 0 4 0 . 0 9 5 5 E Q U A T I O N E Q 1 9 * * * * * * * • * * * • » * * * * * » D E P E N D E N T V A R I A B L E : Y L 1 0 R - S U U A R E D = 0 . 8 2 7 4 D U R B I N - W A T S O N S T A T I S T I C ( A D J . F O R 0 . G A P S ) " 1 . 1 0 5 0 N U M B E R O F O B S E R V A T I O N S » 1 2 . S U M OF S Q U A R E D R E S I D U A L S * ^ 0 . 1 9 8 4 3 5 E - 0 2 S T A N D A R D E R R O R O F T H E R E G R E S S I O N » 0 . 1 2 8 5 9 3 E - 0 1 SUM O F R E S I D U A L S » 0 . 6 5 5 6 5 1 e - 0 6 M E A N O F D E P E N D E N T V A R I A B L E « 0 . 7 7 6 0 2 3 E Q U A T I O N E Q 2 0 D E P E N D E N T V A R I A B L E : Y K 1 0 R - S Q U A R E D » 0 . 8 4 1 9 O U R B I N - W A T S O N S T A T I S T I C ( A D J . F O R 0 . G A P S ) • 1 . 1 1 9 4 N U M B E R O F J 3 S E R V A T I O N S « 1 2 . SUM OF- S Q U A R E D R E S I D U A L S * 0 . 1 9 3 9 6 0 E - 0 2 S T A N D A R D ERROR OF THE R E G R E S S I O N «. 0 . 1 2 7 1 3 5 E - I S U M OF R E S I D U A L S = 0 . 1 0 7 2 8 3 E - 0 5 M E A N O F D E P E N D E N T V A R I A B L E » 0 . 1 9 5 4 2 2 Table 2.5 B: Transportation Equipment - Estimated Input Cost Shares and P r i c e E l a s t i c i t i e s Cost Shares ML KK ME 1 9 6 1 0 . 6 6 4 4 9 3 0 . 2 9 7 8 1 2 O . 3 7 6 9 0 2 E - 0 1 1 9 6 2 0 . 6 6 8 7 7 9 0 . 2 9 3 1 8 4 0.3600 7 8 E - 0 1 1 9 6 3 0 . 6 7 1 0 9 0 C . 2 9 0 7 5 8 O . i d i 5 2 1 E - 0 1 1 9 6 4 0 . C 6 7 5 4 0 0 . 2 9 4 4 3 2 O . 3 6 0 2 0 5 E - C 1 1 9 6 5 0 . 6 6 6 4 8 7 0 . 2 9 5 5 2 5 u . 3 7 9 8 8 0 i - 0 1 1 9 6 6 0 . 6 5 1 7 5 4 0 . 3 1 1 0 1 3 O . 3 7 2 J 3 0 E - 0 1 1 9 6 7 0 . 6 4 9 0 1 3 0 . 3 1 3 8 7 7 0 . 3 7 1 U 9 9 E - C 1 1 9 6 8 0 . 6 5 1 2 3 5 0 . 3 1 1 4 3 3 O . 3 7 3 o l 9 t - 0 1 1 9 6 9 0 . 6 1 0 C 9 9 0 . 3 5 4 7 1 1 U . 3 3 1 9 0 3 E - 0 1 1 9 7 0 0 . 6 2 6 3 0 5 0 . 3 3 7 6 0 4 0 . 3 6 0 9 0 7 E - 0 1 1 9 7 1 0 . 6 4 4 1 6 5 0 . 3 1 8 7 7 2 0 . 3 7 U o 2 7 E - 0 1 1 9 7 2 0 . 6 5 2 9 8 9 0 . 3 0 9 4 3 5 O . 3 7 5 7 6 3 E - 0 1 Own P r i c e E l a s t i c i t i e s P S l t P S K K P S E c 1 9 6 1 . - 0 . 2 0 2 8 8 0 - 0 . 3 7 2 1 0 1 - 0 . 9 7 0 3 0 8 1 9 6 2 . - 0 . 1 9 9 4 4 8 - 0 . 3 7 1 5 1 9 - 0 . 9 7 U o 0 3 1 9 6 3 . - 0 . 1 9 7 5 9 0 - 0 . 3 7 1 1 4 7 - 0 . 9 7 U 2 4 4 1 9 6 4 . - 0 . 2 0 C 4 4 2 - 0 . 3 7 1 6 9 2 - J . 9 7 0 3 9 4 1 9 6 5 . - 0 . 2 0 1 2 8 6 - 0 . 3 7 1 8 3 4 -0.970444 1 9 6 6 . - 0 . 2 1 3 0 3 0 - 0 . 3 7 2 9 1 1 - 0 . 9 7 1 3 7 0 1 9 6 7 . - 0 . 2 1 5 2 0 1 - 0 . 3 7 2 9 3 1 - 0 . 9 7 1 5 2 1 1 9 6 8 . - 0 . 2 1 3 4 4 2 - 0 . 3 7 2 9 1 7 - 0 . 9 7 1 2 4 8 1 9 6 9 . - 0 . 2 4 5 4 5 4 - 0 . 3 6 8 1 5 1 - 0 . 9 7 3 9 1 2 1 9 7 0 . - 0 . 2 3 2 9 8 6 - 0 . 3 7 1 2 1 5 - 0 . 9 72 7 34 1 9 7 1 . - 0 . 2 1 9 0 2 6 - 0 . 3 7 2 8 4 5 - 0 . 9 7 1 5 7 9 1 9 7 2 . - 0 . 2 1 2 0 5 2 - 0 . 3 7 2 8 7 7 - 0 . 9 7 0 9 4 8 • Cross P r i c e E l a s t i c i t i e s P S L K P S L E P S K L P S K E P S E L P S c K 1 9 6 1 0 . 1 5 7 2 9 1 0 . 4 5 5 8 8 6 E - 0 A u. 3 5 0 9 5 9 0 . 2 1 1 4 2 2 E - 0 1 0 . 8 0 3 7 5 1 0 . 1 6 7 C 5 7 1 9 6 2 0 . 1 5 3 5 6 2 0 . 4 5 3 8 5 7 ; - - 0 1 U . 3 5 U 2 9 0 0 . 2 1 2 2 8 7 c - 0 1 0 . 8 0 6 7 5 9 0 . 1 O J 6 2 4 1 9 6 3 0 . 1 5 1 6 1 7 s 0 . 4 5 9 7 2 9 c " - 0 1 u. 3 * 9 9 4 4 H 0 . 2 1 2 0 2 7 c - 0 1 0 . 0 0 3 o 5 7 0 . 1 o l 5 6 o 1 9 6 4 . . 0 . 1 5 4 5 5 1 0 . 4 5 8 9 0 9 i - 0 1 0 . 3 5 0 4 0 1 0 . 2 1 2 9 0 6 E - 0 1 0 . 8 0 5 5 5 4 0 . 1 6 4 8 4 0 1 9 6 5 0 . 1 5 5 4 2 3 C . 4 5 8 6 2 8 1 3 - 0 1 0 . 3 3 U J 2 2 • 0 . 2 1 3 1 2 0 c - 0 1 0 . 6 0 4 6 4 9 0 . 1 6 5 7 9 5 1 9 6 6 0 . 1 6 7 7 4 5 0 . 4 5 2 6 5 9 c - 0 1 0 . 3 3 1 5 2 3 • 0 . 2 1 3 8 7 4 E - 0 1 0 . 7 9 2 7 1 8 0 . 1 7 d o 5 2 1 9 6 7 0 . 1 7 0 0 0 4 0 . 4 5 1 9 6 8 c - 0 1 0 . 3 5 1 5 2 2 0 . 2 1 4 O 8 9 E - 0 1 0 . 7 9 0 4 . 4 4 0 . 1 8 1 0 7 6 1 9 6 8 0 . 1 6 8 0 5 1 0 . 4 5 3 9 1 2 E -or 0 . 3 5 1 4 0 9 0 . 2 1 5 0 7 8 c - 0 1 0 . 7 9 1 3 2 4 0 . 1 7 9 4 2 4 1 9 6 9 0 . 2 0 1 6 6 1 0 . 4 3 7 9 2 9 ^ - 0 1 0 . 3 4 0 6 5 5 0 . 2 1 2 9 6 8 E - 0 1 0 . 7 5 9 2 4 5 0 . 2 1 4 6 o 7 1 9 7 0 . 0 . 1 8 8 5 1 5 C . 4 4 4 7 0 8 E - 0 1 0 . 3 4 9 7 2 2 0 . 2 1 4 9 3 3 E - 0 1 0 . 7 7 1 7 3 0 0 . 2 0 1 0 5 5 1 9 71 . 0 . 1 7 3 3 1 6 0 . 4 5 2 1 0 4 c - 0 1 0 . 3 5 1 2 4 2 0 . 2 1 6 0 2 6 E - 0 1 0 . 7 8 5 7 7 6 0 . 1 8 5 6 0 3 1 9 7 2 , 0 . 1 6 6 4 3 3 0.456139E - 0 1 0.351227 0 . 2 1 6 4 9 9 E-01 0 , 7 9 2 6 6 4 0 . 1 7 8 2 8 4 Table 2.6 A: Non-metallic Minerals - Regression Results and S t a t i s t i c a l Tests T R A C E . O F M A T R I X « 1 . 9 9 9 9 4 L O G OF L I K E L I H O O O F U N C T I O N = 7 9 . 5 1 5 3 h1GHT-HAND V A R I A B L E E S T I M A T E D C O E F F I C I E N T S T A N D A R D E R R O R T -S T A T I S T I C 2 - T A I L E U T -P R U B A B I L I T Y A L O. 5 7 5 a 5 0 0 . 1 2 3 9 3 5 E - 0 1 4 6 . 4 3 9 8 0 . 0 0 0 0 G L L 0 . 1 0 9 2 4 6 0 . 1 4 1 1 3 2 E - 0 1 7 . 7 3 7 9 4 0 . 0 0 3 1 G L K - 0 . 1 5 0 o 7 B 0 . 2 0 8 7 3 4 E - 0 1 - 7 . 2 1 6 9 3 0 . 0 0 0 1 D L R D 0 . 2 3 9 J 6 2 C - O 1 • 0 . 1 3 3 2 2 1 E - 0 1 1 . 7 9 6 7 3 0 . 1 1 2 6 C L Y - 0 . 1 5 1 0 4 5 0 . 9 7 2 3 3 2 E - 0 1 - 1 . 5 5 3 4 3 0 . 1 6 1 4 A K 0 . 2 8 1 8 5 3 0 . 1 8 7 1 8 8 E - 0 1 1 5 . 0 5 7 2 0 . 0 0 0 0 G K K 0 . 2 0 5 2 1 4 0 . 3 1 5 2 2 0 E - 0 1 6 . 5 1 0 1 8 0 . 0 0 0 2 D K R D - 0 . 2 7 7 6 9 8 E - 0 1 0 . 2 0 1 1 3 1 E - 0 1 - 1 . 3 8 0 6 8 0 . 2 0 7 1 C KY 0 . 2 2 2 3 8 5 0 . 1 4 6 5 7 9 1. 5 1 6 5 3 0 . 1 6 9 6 E Q U A T I O N E Q 2 3 * * * * * * * * * * * * * * * * * * E Q U A T I O N E Q 2 4 D E P E N D E N T V A R I A B L E : Y L 1 2 -D E P E N D E N T V A R I A B L E : Y K 1 2 R - S Q U A R E O = 0 . 9 4 7 0 D U R B I N - W A T S O N S T A T I S T I C < A D J . F O R 0 . G A P S ) = 1 . 7 5 6 2 N U M B E R O F O B S E R V A T I O N S » 1 2 . S U M ' O F S Q U A R E O R E S I D U A L S » U . 5 b ^ o 3 6 E - 0 3 S T A N D A R D E R R O R O F T H E R E G R E S S I O N - 0 . 6 8 4 1 2 7 E - 0 2 S U M O F R E S I O U A L S » 0 . 5 9 6 0 4 o c - 0 7 M E A N O F D E P E N D E N T V A R I A B L E * 0 . 5 6 6 9 1 1 R - S Q U A R E D = 0 . 9 4 5 9 O U R B I N - W A T S O N S T A T I S T I C J A D J . F O R 0 . G A P S ) =» 1 . 7 2 7 1 N U M B E R OF O B S E R V A T I O N S = 1 2 . SUM OF S Q U A R E D R E S I C U A L S » 0 . 1 2 8 2 2 5 E - 0 2 S T A N D A R D E R R O R OF T H E R E G R E S S I O N - 0 . 1 0 3 3 7 O E - C S U M OF R E S I O U A L S * 0 . 5 3 6 4 4 2 c - 0 o M E A N OF O c P E N D E N T V A R I A B L E • 0 . 3 0 6 7 6 4 Table 2.6 B: Non-metallic Minerals - Estimated Input Cost Shares and Price E l a s t i c i t i e s Cost Shares Own Price E l a s t i c i t i ML MK ME 1 9 6 1 0 . 5 7 5 5 5 0 C . 2 8 1 8 5 3 . 0 . 1 4 2 5 9 7 1 9 6 2 0 . 5 7 9 1 2 6 0 . 2 7 6 7 6 7 0 . 1 4 4 1 0 8 1 9 6 3 0 . 5 7 5 7 3 2 0 . 2 8 1 2 3 9 0 . I t 3 0 2 d 1 9 6 4 0 . 5 7 2 7 5 1 0 . 2 b 5 J 0 3 0 . 1 4 1 9 4 6 1 9 6 5 0 . 5 6 5 0 2 2 C . 2 9 5 6 8 0 0 . 1 3 9 ^ 9 3 1 9 6 b ' . 0 . 5 4 8 1 1 6 0 . 3 1 U 5 8 3 0 . i 3 3 - > 0 1 1 9 6 7 0 . 5 4 5 4 1 5 0 . 3 2 2 1 8 7 0 . 1 ^ 2 3 9 8 1 9 6 8 0 . 5 3 8 6 3 4 0 . 3 3 1 2 8 4 0 . 1-JU082 1 9 6 9 0 . 4 8 8 3 5 4 0 . 3 9 9 6 4 6 0 . 1 1 2 0 0 0 1 9 7 0 0 . 4 9 7 1 1 8 0 . 3 8 7 0 5 5 0 . 1 1 5 8 2 7 1 9 7 1 0 . 5 1 8 1 9 8 C . 3 5 8 1 6 0 0 . 1 2 J O 4 2 1 9 7 2 0 . 5 3 6 5 8 9 0 . 3 3 2 9 4 8 0 . 1 3 0 4 6 3 S PSLL PSKK PSEt 1 9 6 1 . - 0 . 2 3 4 6 3 9 0 . 9 9 4 1 1 9 E - 0 2 - 0 . 7 6 5 5 0 4 1 9 6 2 . - 0 . 2 3 2 2 3 5 0 . 1 8 2 3 5 2 E - O i -0. 7 6 * 9 5 6 1 9 6 3 . - 0 . 2 3 4 5 1 6 i 0 . 1 0 9 1 6 0 c - 0 1 - 0 . 7 O 5 J 5 0 1 9 6 4 . - 0 . 2 3 6 5 1 0 0 . 4 5 8 7 4 0 E - 0 2 - 0 . 7 o 5 7 3 3 1 9 6 5 . - 0 . 2 4 1 6 J 0 - C . 1 0 2 7 9 2 c - 0 1 -0. 7 o 6 o 2 6 1 9 6 6 . - 0 . 2 5 2 5 7 3 - 0 . 3 7 2 7 1 5 E - 0 1 -0. 7 6 0 3 9 1 1 9 6 7 . - 0 . 2 5 4 2 8 6 - C . 4 0 8 7 3 0 E - 0 1 - 0 . 7 6 8 o 2 4 1 9 6 8 . - 0 . 2 5 8 5 4 6 - 0 . 4 9 2 6 6 2 E - 0 1 -0. 7 6 9 1 7 7 1 9 6 9 . - 0 . 2 8 7 9 4 4 - 0 . 8 o 8 6 5 l E - 0 1 - o . 7 7 0 9 9 5 1 9 7 0 . - 0 . 2 8 3 1 2 3 - 0 . 8 2 7 5 2 0 c - 0 1 -0. 7 7 1 0 3 4 1 9 7 1 . - 0 . 2 7 0 9 8 3 - 0 . 6 8 8 7 2 9 E - 0 1 - 0 . 7 7 0 3 7 0 1 9 7 2 . - 0 . 2 5 S 8 1 8 - 0 . 5 0 6 9 U 3 E - 0 1 -0. 7 6 9 0 9 1 Cross Price E l a s t i c i t i e s P S L K P S L E P S K . L P S K E P S c L P S E K 1 9 6 1 0 . 2 0 0 5 5 4 E - 0 1 0 . 2 1 4 5 8 3 0 . 4 0 9 3 3 7 c - C l - 0 . 5 0 8 9 5 0 E - 01 0 . Q 6 6 1 C 2 - G . 1 0 0 5 9 6 1 9 6 2 0 . 1 6 5 8 5 5 E - O l 0 . 2 1 5 6 4 9 . 0 . 3 4 7 0 * 7 c - 01 - 0 . 5 2 9 4 0 1 E - 0 1 0 * 6 6 6 6 J 1 - 0 . i O i 6 7 4 1 9 6 3 0 . 1 9 5 2 4 7 E - 0 1 V 0 . 2 1 4 9 9 2 ' 0. 3 9 9 0 9 6 c - G l - 0 . 6 0 8 6 5 7 c - 0 1 0 . 8 6 5 * 0 8 . - 0 . 1 0 0 0 5 8 1 9 6 4 0 . 2 2 2 2 5 6 = - 0 1 0 . 2 1 4 2 8 4 0 . 4 4 o l 8 4 c - 01 - 0 . 4 9 2 0 o l E - 0 1 0 . 8 6 4 6 3 4 - 0 . 9 6 9 0 1 0 E - 0 1 1 9 6 5 0 . 2 9 0 0 3 9 E - 0 1 0 . 2 1 2 6 2 6 0 . 5 5 4 2 4 2 E - 01 - 0 . 4 5 1 * 5 2 £ - 0 1 0 . 8 6 2 4 5 3 - 0 . 9 5 8 i o 9 c - 0 1 1 9 6 6 0 . 4 3 6 8 1 9 E - 0 1 0 . 2 0 8 8 9 1 0. 7 3 0 3 7 E - 0 1 - 0 . 3 7 8 8 2 3 E - 0 1 0 . 8 5 6 9 2 8 - 0 . 9 0 5 3 6 S E - 0 1 1 9 6 7 0 . 4 5 9 2 4 9 E - 0 1 0 . 2 0 8 3 6 1 v 0. 7 / 7 4 4 1 E - 0 1 - 0 . 3 6 8 7 1 3 E - 0 1 0 . 6 5 3 3 4 9 . - O . 8 9 7 2 5 o c - 0 1 1 9 6 8 0 . 5 1 5 4 3 5 E - 0 1 0 . 2 0 7 0 0 2 J. 630044E- 01 - 0 . 3 4 5 5 6 4 E - 0 1 0 . 8 5 7 1 3 7 - 0 . 8 7 ' < 5 9 d £ - 0 1 1 9 6 9 0 . 9 1 1 0 4 4 E - 0 1 0 . 1 9 6 8 3 9 0 . 1 1 1 3 2 7 - 0 . 2 4 4 6 1 7 c - 01 0 . 8 5 3 2 6 1 - - 0 . 8 7 2 6 6 1 E - 0 1 1 9 7 0 0 . 8 3 9 5 2 5 E - 01 0 . 1 9 9 1 7 1 0. 1 0 7 6 2 5 - 0 . 2 5 U / 3 6 E - 0 1 0 . 6 5 * 6 2 2 - 0 . 8 3 / 6 / 5 E - 0 1 1 9 7 1 0 . 6 7 3 8 7 5 E - 01 0 . 2 0 3 5 9 5 Urn 9 7 4 9 8 6 E - 01 - 0 . 2 8 6 2 5 8 S - 0 1 0 . 8 5 3 2 9 2 - 0 . 8 2 9 2 1 6 E - 0 1 1 9 7 2 . 0 . 5 2 1 4 1 9 E -01 0 . 2 0 7 6 7 6 0. 8 4 0 3 3 4 E - 01 - 0 . 3 3 3 3 5 3 E - 01 0 . 8 5 4 1 6 4 • - 0 . 8 5 0 7 3 6 E - 0 1 Table 2.7 A: "Petroleum and Coal - Regression Results and S t a t i s t i c a l Tests T R A C E O F M A T R I X = 1 . 9 9 9 9 8 L O G OF L I K E L I H O O D F U N C T I O N R I G H T - H A N D V A R I A B L E A L G L L G L K D L R D C L Y A K GKK OKRD C K Y 7 3 . 3 1 1 2 E S T I M A T E D C U E F F 1 C I c N T 0 . 4 4 2 * 0 5 O . 8 4 8 7 0 3 c - O l - 0 . 9 3 6 j . l 5 c - U . l - 0 . 1 2 4 7 9 7 E - 0 1 - 0 . 2 5 8 1 3 2 0 . 5 1 1 7 0 3 0 . 1 0 4 2 3 2 0 . 1 2 0 0 1 6 E - 0 1 0 . 2 7 8 6 1 4 S T A N D A R D E R R O R 0 . 1 5 4 5 2 5 E - 0 1 0 . 3 4 6 4 6 8 E - 0 1 0 . 2 6 3 2 4 3 E - 0 1 O . 4 1 9 7 1 5 E - 0 2 O.193774E-01 0 . 1 3 7 5 9 4 E - 0 1 0 . 2 3 4 4 5 8 E - 0 1 O . 3 5 2 8 3 0 E - 0 2 0 . 4 2 4 6 0 9 E - 0 1 S T A T I S T I C 2 8 . 6 3 0 0 2 . 4 4 9 5 8 - 3 . 5 5 6 0 9 - 2 . 9 7 3 3 8 - 5 . 2 2 7 7 3 3 7 . 1 8 9 4 4 . 4 4 5 6 6 3 . 4 0 1 5 3 6 . 5 6 1 0 7 2 - T A I L E U T -P R Q B A B I L I T Y 0 . 0 0 0 0 0 . 0 4 1 9 0 . 0 0 6 3 0 . 0 1 9 1 0 . 0 0 1 0 0 . 0 0 0 0 0 . 0 0 2 5 0 . 0 1 0 3 0 . 0 0 0 2 E Q U A T I O N E 0 2 5 * * * * * * * * * * * * * * * * * * D E P E N D E N T V A R I A B L E : • Y L 1 3 R - S Q U A R E D = 0 . 9 5 8 0 O U R B I N - w A T S O N S T A T I S T I C ( A D J . F O R 0 . G A P S ) = 2 . 0 6 8 5 N U M B E R OF O B S E R V A T I O N S » 1 2 . SUM OF S Q U A R E D R E S I D U A L S » 0 . 8 2 5 9 3 9 E - 0 3 S T A N D A R D E R R O R OF T H E R E G R E S S I O N - 0 . 8 2 9 6 2 8 E - 0 2 SUM O F R E S I D U A L S » 0 . 6 5 5 6 5 1 C - 0 6 M E A N OF D E P E N D E N T V A R I A B L E = 0 . 3 8 2 8 2 2 E Q U A T I O N E Q 2 6 * * * * * * * * * * * * * * * * * * D E P E N D E N T V A R I A B L E : Y K 1 3 R - S Q U A R E D = 0 . 5 6 9 2 D U R B I N - W A T S O N S T A T I S T I C ( A D J . F O R o . G A P S 1 => 2 . 2 5 1 4 NUMBER OF 0 3 S E R V A T I C N S » 1 2 . SUM OF S Q U A R E D R E S I D U A L S » 0 . 6 3 5 d 0 1 c - 0 3 •1 S T A N D A R D E R R O R O F T h E R E G R E S S I O N » 0 . 7 5 5 9 7 7 E - 0 2 SUM O F R E S I D U A L S » 0 . 8 3 4 4 6 5 c - 0 6 M E A N OF D E P E N D E N T V A R I A B L E » 0 . 5 7 0 6 5 0 Table 2.7 B: Petroleum and Coal - Estimated Input Cost Shares and Price E l a s t i c i t i e s Cost Shares ML MK . ME 1 9 6 1 0 . 4 4 2 4 0 5 0 . 5 1 1 7 0 3 U . 4 5 8 9 2 3 C - 0 1 1 9 6 2 0 . 4 5 C 4 9 1 0 . 5 0 2 7 0 3 d . * o 6 U 6 2 c - 0 1 1 9 6 3 0 . 4 5 0 6 6 3 0 . 5 0 2 4 8 0 J . 4 o 3 j 7 6 6 - 0 1 1 9 6 4 0 . 4 5 C 7 9 3 0 . 5 0 2 4 3 9 u . 4 o 7 6 3 4 E - 0 1 1 9 6 5 0 . * 5 1 1 o 0 0 . 5 0 1 9 6 0 0 . 4 O 6 8 0 7 E - 0 1 1 9 6 6 0 . 4 4 1 3 2 4 0 . 5 1 3 1 6 6 0 . 4 J J i 0 5 E - 0 1 1 9 6 7 0 . 4 3 6 2 3 3 0 . 5 1 8 8 5 9 O . 4 * v O 8 5 c - 0 1 1 9 6 3 0 . 4 3 3 3 8 8 0 . 5 2 2 0 6 7 0 . 4 4 5 4 4 1 E - 0 1 1 9 6 9 0 . 3 9 7 1 2 1 0 . 5 6 2 5 4 8 0 . 4 J J 3 0 9 C - 0 1 1 9 7 0 0 . 4 1 3 3 1 6 0 . 5 4 4 5 4 8 O . * 2 1 J O 5 E - 0 1 1 9 7 1 0 . 4 2 5 6 3 5 0 . 5 3 0 8 8 1 0 . 4 j * 6 4 i e - 0 1 1 9 7 2 0 . 4 2 6 2 1 3 0 . 5 2 8 0 5 3 U . 4 3 / J 3 3 E - 0 1 a s t i c i t i e s P S L L P S K K P i E c 1 9 6 1 . - 0 . 3 6 5 7 5 7 - 0 . 2 8 4 6 0 0 - 0 . 9 i 5 1 5 0 1 9 6 2 . - 0 . 3 6 1 1 1 4 ' - 0 . 2 6 9 9 5 3 - o . 9 i 3 0 3 6 1 9 6 3 . - 0 . 3 6 1 0 C 3 - 0 . 2 9 0 0 8 4 - 0 . 9 1 3 0 3 1 1 9 6 4 . - 0 . 3 6 0 9 5 5 - C . 2 9 0 1 0 9 - u . 9 1 J 0 4 2 1 9 6 5 . - 0 . 3 6 0 7 2 5 - 0 . 2 9 U 3 9 0 - 0 . 9 i J J 2 5 1 9 6 6 . - 0 . 3 6 6 3 6 8 - 0 . 2 6 3 7 1 8 - 0 . 9 1 J 1 B 9 1 9 6 7 . - 0 . 3 6 9 2 1 5 - 0 . 2 8 0 2 5 4 - 0 . 9 i j 2 ^ 6 1 9 6 8 . - 0 . 3 7 C 7 8 2 - 0 . 2 7 8 2 8 0 - 0 . 9 1 3 2 5 9 1 9 6 9 . - 0 . 3 8 9 1 6 5 - 0 . 2 5 2 1 6 6 - O . 9 1 J 0 6 * 1 9 7 0 . - 0 . 3 8 1 3 4 4 - 0 . 2 6 4 0 4 1 - Y . 9 1 3 2 5 5 . 1 9 7 1 . - 0 . 3 7 4 9 6 8 - 0 . 2 7 2 7 8 1 - 0 . 9 1 3 2 9 0 ' . 1 9 7 2 . - 0 . 3 7 3 5 9 1 - 0 . 2 7 4 5 5 7 - U . 9 1 3 2 0 7 E l a s t i c i t i e s P S L K P S L E P S K L P S K b P S E L P S E K 1 9 6 1 0 . 3 0 0 1 0 6 0 . 6 5 6 5 0 6 E - 0 1 0 . 2 5 9 4 6 4 . 0 . 2 5 1 3 6 4 E 0 . 2 5 6 7 8 7 E - 0 1 0 . 6 3 2 8 7 7 0 . 2 8 0 2 7 3 1 9 6 2 0 . 2 9 4 9 0 4 V 0 . 6 6 2 0 9 9 E - 0 1 0 . 2 6 4 2 75 - 0 1 0 . 0 3 7 2 * 4 0 . 2 7 5 / 9 2 1 9 6 3 . 0 . 2 9 4 7 6 9 0 . 6 6 2 3 3 0 c - 0 1 0 . 2 6 4 3 8 3 0 . 2 5 7 0 0 8 E - 0 1 0 . 6 3 7 3 1 0 0 . 2 7 5 7 2 1 1 9 6 4 0 . 2 9 4 7 8 1 0 . 6 6 1 5 3 9 E - 0 1 0 . 2 6 4 4 8 3 0 . 2 5 6 2 4 9 E -0.1 0 . 6 3 / 7 2 1 0 . 2 7 5 3 2 0 1 9 6 5 0 . 2 9 4 4 6 9 0 . 6 6 2 5 5 6 E - 0 1 0 . 2 o * o 6 8 ' 0 . 2 5 7 2 1 9 c - 0 1 0 . 6 3 7 6 1 6 0 . 2 7 5 * 0 9 1 9 6 6 0 . 3 0 1 0 5 1 0 . 6 5 3 1 7 2 E - 0 1 . 0 . 2 4 8 1 3 8 E - 0 1 0 . 6 3 3 3 9 3 0 . 2 7 9 7 9 5 1 9 6 7 0 . 3 0 4 2 6 3 0 . 6 4 9 4 6 4 c - 0 1 0 . 2 5 5 6 1 5 0 . 2 4 4 5 8 9 E - 0 1 0 . 6 3 0 8 7 7 0 . 2 8 2 J 5 9 1 9 6 8 0 . 3 0 6 0 6 8 0 . 6 4 7 1 J S E - 0 1 ^ . 2 3 4 0 7 9 0 . 2 4 2 0 0 3 E - 0 1 0 . 6 2 9 o 2 5 0 . 2 6 3 6 3 3 1 9 6 9 0 . 3 2 6 8 2 3 . 0 . 6 2 3 4 2 3 E - 0 1 0 . 2 3 0 / 1 5 0 . 2 1 4 5 1 0 c - 0 1 0 . 6 1 5 3 5 8 0 . 2 9 9 2 0 6 1 9 7 0 0 . 3 1 8 0 5 9 0 . 6 3 2 8 5 5 E - O i d . 2 4 1 * 0 9 0 . 2 2 6 3 2 6 E - 0 1 0 . 6 2 0 7 6 5 0 . 2 9 2 4 9 0 1 9 7 1 0 . 3 1 0 9 4 7 0 . 6 4 O 2 0 9 E - 0 1 0 . 2 4 9 3 0 2 0 . 2 3 4 7 8 0 c - 0 1 0 . 0 2 6 6 5 5 0 . 2 8 6 6 3 5 1 9 7 2 0 . 3 0 9 4 4 3 0 . 6 4 1 4 6 5 c - 0 1 0 . 2 5 0 9 3 6 0 . 2 3 6 2 0 6 E -oi 0 . 6 2 8 0 8 6 0 . 2 6 5 2 0 1 00 ro Table 2.8 A: Chemicals - Regression Results and S t a t i s t i c a l Tests T R A C E OF M A T R I X 1 . 9 9 9 9 6 L O G O F L I K E L I H O O D F U N C T I O N R I G H T - H A N D V A R I A B L E A L G L L G L K D L R D C L Y A K G K K OKRD C KY 8 1 . 2 6 3 2 E S T I M A T E D C O E F F I C I E N T 0 . 5 3 d 5 5 1 0 . 6 7 4 J 5 1 C - 0 1 - 0 . 7 2 8 7 2 6 c - U l 0 . 1 5 8 3 6 4 E - 0 1 - 0 . 3 1 4 2 3 0 0 . 3 6 2 8 8 0 0 . 9 8 1 1 6 9 E - 0 1 - 0 . 1 7 6 O 3 8 E - 0 1 0 . 3 3 0 4 8 1 S T A N D A R D E R R O R . 0 . 2 1 8 2 6 1 E - 0 1 0 . 1 3 8 1 6 3 E - 0 1 0 . 1 4 3 8 7 3 E - 0 1 0 . 5 7 1 9 0 9 E - 0 2 0 . 3 9 2 7 0 3 E - 0 1 0 . 2 3 9 6 7 8 E - 0 1 0 . 1 6 3 1 9 5 E - 0 1 0 . 6 2 9 2 0 2 E - 0 2 0 . 4 4 5 6 1 2 E - 0 1 T -S T A T I S T I C 2 4 . 6 7 4 7 4 . 8 7 8 6 6 - 5 . 0 6 5 0 7 2 . 7 6 9 0 4 - 8 . 0 0 1 7 3 1 5 . 1 4 0 3 6 . 0 1 2 2 4 - 2 . 7 9 7 7 9 7 . 4 1 6 3 4 2 - T A I L E D T -P R U B A o I L I T Y 0 . 0 0 0 0 0 . 0 0 1 5 0 . 0 0 1 2 0 . 0 2 5 9 0 . 0 0 0 1 0 . 0 0 0 0 0 . 0 0 0 4 0 . 0 2 4 8 0 . 0 0 0 1 E Q U A T I O N E Q 2 7 * * * * * * * * * * * * * * * * * * D E P E N D E N T V A R I A B L E : Y L 1 4 E Q U A T I O N E Q 2 B D E P E N D E N T V A R I A B L E : Y K 1 4 R - S Q U A R E O = 0 . 9 9 1 5 D U R B I N - H A T S O N S T A T I S T I C I A D J . F O R 0 . G A P S ! • 2 . 5 5 8 2 N U M B E R O F O B S E R V A T I O N S » 1 2 . V S U M OF S Q U A R E D R E S I D U A L S ' U . 2 U 5 d l l E - 0 3 S T A N D A R D E R R O R OF T H E R E G R E S S U N * 0 . 4 1 4 1 3 7 E - 0 2 S U M O F R E S I D U A L S » 0 . 6 5 5 6 5 1 E - 0 & M E A N O F D E P E N O E N T V A R I A B L E «• 0 . 5 3 9 3 1 2 R - S Q U A R E D = 0 . 5 9 2 8 U U R B I N - W A T S O N S T A T I S T I C ( A D J . F O R 0 . G A P S ) « 2 . 3 7 6 5 N U M B E R ' ' O F O B S E R V A T I O N S • 1 2 . S U M OF S Q U A R E D R E S I D U A L S » J . 2 4 5 1 1 6 E - 0 3 S T A N D A R D E R R O R OF T H E R E G R E S S I O N = 0 . 4 5 1 9 5 4 E - 0 2 S U M O F R E S I D U A L S » 0 . 5 9 6 0 4 6 c - 0 6 M E A N OF D E P E N O E N T V A R I A B L E « 0 . 3 6 5 4 0 2 Table 2.3 B: Chemicals - Estimated Input Cost Shares and Price Ela s t i c i t i e s ML MK ME 1 9 6 1 0 . 5 3 8 5 5 1 0 . 3 6 2 8 8 0 ' U . 9 8 5 6 8 4 E - 0 1 1 9 6 2 0 . 5 4 1 2 9 5 0 . 3 6 0 3 5 0 U . 9 d 3 5 5 0 E - 0 1 1 9 6 3 0 . 5 4 2 6 4 1 0 . 3 5 9 2 6 0 0 . 9 d 0 9 8 8 E - 0 1 1 9 6 4 0 . 5 3 9 7 7 3 0 . 3 6 3 6 3 7 0 . 9 o S o 9 6 c - 0 1 1 9 6 5 0 . 5 3 6 4 1 5 0 . 3 6 7 3 4 9 0 . 9 o 2 3 o 4 E - 0 1 1 9 6 6 0 . 5 2 8 5 3 4 0 . 3 7 8 7 8 1 0 . 9 * o b 5 0 E - 0 1 1 9 6 7 0 . 5 2 5 5 4 3 0 . 3 8 4 6 1 6 O . d 9 b 4 i i E - 0 1 1 9 6 b 0 . 5 2 4 0 9 8 0 . 3 8 8 1 2 6 0 . 8 7 7 7 6 6 E - 0 1 1 9 6 9 0 . 4 9 8 6 0 0 0 . 4 2 3 8 0 1 0 . 7 7 5 9 9 5 E - 0 1 1 9 7 0 0 . 5 0 9 6 0 1 0 . 4 1 2 2 2 3 0 . 7 6 1 7 6 1 5 - 0 1 1 9 7 1 0 . 5 2 3 5 5 5 0 . 3 9 3 6 9 3 0 . 8 2 7 5 1 8 E - 0 1 1 9 7 2 0 . 5 3 1 8 9 1 0 . 3 8 4 3 7 2 0 . 8 3 7 3 6 8 E - 0 1 Own Price. E l a s t i c i t i e s ' PSLL PSKK PSEC 1 9 6 1 . - 0 . 3 3 6 2 8 9 - 0 . 3 6 6 7 3 6 - 0 . 7 0 0 7 9 0 1 9 6 2 . - 0 . 3 3 4 1 0 0 - 0 . 3 6 7 3 6 8 - 0 . 7 0 0 5 6 8 1 9 6 3 . - 0 . 3 3 3 1 4 3 - 0 . 3 6 7 6 3 1 - 0 . 7 0 0 * 9 9 1 9 6 4 . - 0 . 3 3 5 3 5 0 - 0 . 3 6 6 5 4 1 - J . 6 9 0 o 5 9 1 9 6 5 . - 0 . 3 3 7 9 2 7 - 0 . 3 6 5 5 5 7 - 0 . 6 9 6 2 6 0 1 9 6 6 - 0 . 3 4 3 9 3 4 - 0 . 3 6 2 1 8 6 - 0 . 6 9 ^ 9 3 8 1 9 6 7 . - 0 . 3 4 6 1 9 9 - 0 . 3 6 0 2 8 0 -0.6900*7 1 9 6 8 . - 0 . 3 4 7 2 9 1 - 0 . 3 5 9 0 7 8 - 0 . 6 0 6 9 1 4 1 9 6 9 . - 0 . 3 6 6 2 1 2 - 0 . 3 4 4 6 8 3 - 0 . 6 6 7 5 4 2 1 9 7 0 . - 0 . 3 5 8 1 2 8 - 0 . 3 4 9 7 5 8 - O . 6 o 6 d 4 6 1 9 7 1 . - 0 . 3 4 7 7 0 0 - 0 . 3 5 7 0 8 5 - 0 . 6 7 8 2 5 8 1 9 7 2 . - 0 . 3 4 1 3 8 2 - 0 . 3 6 0 3 6 2 - 0 . 6 8 0 0 8 4 • Cross Price E l a s t i c i t i e s , • PSLK. PSLE PSKL PSKE PSEL PSEK. 1 9 6 1 0 . 2 2 7 5 6 3 V 0 . 1 0 8 7 2 0 0 . 3 3 7 7 3 4 0 . 2 9 0 0 1 7 E - 0 1 0 . 5 9 4 0 2 0 0 . 1 0 6 7 7 0 1 9 6 2 0 . 2 2 5 7 2 4 0 . 1 0 8 4 5 6 0 . 3 3 9 0 & 3 O . 2 8 2 9 9 9 E - 0 1 0 . 3 9 6 8 8 4 0 . 1 0 3 6 8 4 1 9 6 3 0 . 2 2 4 9 6 8 C . 1 0 3 1 7 4 O . 3 3 9 6 0 0 0 . 2 7 8 3 1 2 E - 0 1 0 . 5 9 8 3 7 5 0 . 1 0 1 9 2 4 1 9 6 4 0 . 2 2 8 6 3 1 0 . 1 0 6 7 1 9 0 . 3 J 9 3 7 4 0 . 2 7 1 6 7 d c - Q i 0.590373 0 . 1 0 2 2 8 1 1 9 6 5 0 . 2 3 1 4 9 8 0 . 1 0 6 4 2 9 0 . 3 3 0 0 4 0 0 . 2 7 5 1 6 0 E - 0 1 0 . 5 9 3 2 2 8 0 . 1 0 3 0 3 3 1 9 6 6 0 . 2 4 0 9 0 4 , 0 . 1 0 3 0 2 9 0 . 3 3 6 x 4 7 0 . 2 6 0 3 8 7 E - 0 1 0 . 5 8 7 5 2 3 0. 1 0 0 4 1 4 1 9 6 7 0 . 2 4 5 9 5 4 0 . 1 0 0 2 4 5 u . 3 3 6 0 7 4 0 . 2 4 2 0 5 9 c - 0 1 0 . 5 8 6 4 0 0 0 . 1 0 3 6 2 7 1 9 6 0 . 0 . 2 4 9 0 0 2 0 . 9 8 2 0 8 7 E - 0 1 0 . 3 3 6 3 4 3 0 . 2 2 7 3 4 9 E - O 1 0 . 5 S 0 3 B O 0 . 1 0 0 5 2 8 1 9 6 9 0 . 2 7 7 6 4 6 0 . 0 8 5 6 5 1 6 - 0 * 0 . 3 2 6 6 5 0 0 . 1 S 0 3 2 9 E - 0 1 0 . 5 6 9 ^ 5 7 0 . 9 0 4 6 4 5 E - 0 1 1 9 7 0 . 0 . 2 6 9 2 2 3 0 . 0 8 9 0 5 0 c - 0 1 0,3J2822 0 . 1 & 9 3 6 5 E - 0 1 0 . 5 7 9 5 3 9 0 . S 9 3 U 0 1 E - 0 1 1 9 7 1 0 . 2 5 4 5 0 5 0 . 9 3 1 9 4 7 6 -Oi 0 . 3 3 8 4 5 5 0 . 1 8 6 2 9 8 E - 0 1 0 . 5 8 9 6 2 6 0 . 0 0 6 5 1 6 c - 0 1 1 9 7 2 . 0 . 2 4 7 3 6 5 0 . 9 4 0 1 6 0 E -Oi 0 . 3 4 2 3 0 2 0 . 1 8 0 5 9 9 E - 0 1 0 . 5 9 7 1 0 4 0 . 0 2 0 9 9 3 E - 0 1 85 During the period 1961-72 there was a sharp decline in the rela-tive price of energy. Nominal energy prices declined in the f i r s t part of the period and over a l l generally rose less than 20%. The prices of capital and labour increased in about the same proportion (almost doubling), but with different time trends. Capital prices increased sharply in 1969 and then stabilized; while labour prices increased steadily throughout the period. The cost shares in most sectors indicate slightly decreased shares of labour, slightly increased shares of capital with either constant or slightly decreased shares of energy. The own price e l a s t i c i t i e s of energy are the largest of a l l the inputs for a l l sectors. Hence energy use is the most sensitive :of a l l factor inputs to relative price changes. In the period 1961-72 with declining relative energy prices in a l l sectors, this has led to increased use of energy. For energy planners, this has important significance since energy-use is not fixed but is highly dependent upon pricing policies. R&D has contributed significantly (at the .05 level) to changing; input shares in four sectors: paper, petroleum and coal, machinery, and chemicals. In paper and chemicals, R&D contributed to an increase in the shares of labour and energy and a decrease in the share of capital. In petroleum and coal and machinery, R&D contributed to a decrease in the shares of labour and energy and an increase in the capital share. In these four sectors scale of output was also important. Increase in scale led to declines in labour's share in paper, petroleum and coal and chemicals and an increase in labour's share in machinery. The impact is reversed for capital's share. In a l l cases energy's share: increased. In these sectors*the impact of scale is 10-20 times as large, as that of R&D and i t is also greater than the impact of prices. 86 Some conclusions to be drawn from this study are: ( 1 ) i n m o s t s e c t o r s R&D had no i m p a c t ton i n p u t s h a r e s , m o s t p r o b a b l y i n d i c a t i n g t h a t s c a l e a n d p r i c e e f f e c t s a r e d o m i n a n t i n d e t e r m i n i n g t h e s t r u c t u r e o f t e c h n o l o g y , a n d (2) t h e i m p a c t o f R&D i s n o t a l w a y s c a p i t a l u s i n g , b u t a f f e c t s t h e s h a r e o f i n p u t s d i f f e r e n t l y , d e p e n d i n g o n t h e s e c t o r . Therefore as the impact of R&D on the shares of the major factors of production is essentially neutral, perhaps an effort should be made to eliminate technological development from the traditional arena of labour-management bargaining. However, the above results must be interpreted cautiously since they are obtained using data which have been aggregated over both firms and goods. APPENDIX: DATA 87 Table 2A.1 Capital Data by sector 1961 - 1972 (PQK, R ( l ) , R<2), R(3), R(4), PK) key: PQK = ca p i t a l services used i n production R(l) = rental price of building construction R(2) = rental price of engineering construction R(3) = rental price of machinery and equipment R(4) = rental price of operating c a p i t a l PK = Tornquist price index of c a p i t a l services. PQK, ( U l ) , R12] i t R13J , R«4J, PK: SECTOR 1 217326.50000 0.12426 0.12072 0.15273 0. 03498 A . 0JO00 226572.1Z500 0.11870 0.11574 0.15464 J . 06267 0. 9918* 241528.12500 0.11965 0.11464 0.15870 0. 06419 1. 01141 268933.37500 0.12364 0.11690 0.17467 0. 09176 1. 06y97 299400.68750 0.13202 0.129-+4 0.18852 0. 099o7 1. W287 356278.00000 0.15723 0.15136 0-21197 0. 1198o 1. j**85 394 735.87500 0.17620 0.17093 0.21657 0. 12833 1. t!958 45623*. 56250 0.19444 0.19017 0.23631 0. 14663 1. P5341 619504.37500 0.26158 0.26079 0.30301 0. 20625 2. 625223.56250 0.25556 0.25106 0.29356 0.1934o 1. 97t79 633920.25000 0.25091 0.24746 0.28186 0. 13016 1. 3*1076 665163.62500 0.25738 0.25974 0.28429 0. 162x7 1. 93987 PGKi R ( l ) , R12J t R13), R(4), PK: SECTOR 2 24800.58203 0.12124 0.11797 0.20579 0. 06496 1. uOOOO 2579o.61719 0.11732 0.11422 0.21325 0. 0d305 ±m 0x442 27366.48828 0.11691 0.11552 0.22530 0. 06536 1. 05427 29287.57813 0.12090 0.11911 0.24240 0. 09160 1. 1^228 33444. 59375 0.13076 0.12993 0.26489 0. 0*6 33 1. 2<iJ00 39111.29683 0.15437 0.15068 0.28956 0. 1<-103 1. 36 721 47293.92188 0.17622 0.17251 0.29740 0. 13U36 i . *ta26 52150.51563 0.19215 0.19232 0.312b8 0. 1*6 37 1. 5*176 69558.12500 0.25824 0.25959 0.38354 0. 20746 1. 95x07 75334. 12500 0.2536* 0.25031 0.37442 0. 192o6 1-*u523 7992o.25000 0.24826 0.2*327 0.36647 0. lo072 X.6o317 99582.06250 0.25709 0.26375 0.38378 0. 16217 1. 9*328 PyK, R ( l ) , R(2) t R(3J, R(4K PK: SECTOR 3 77063.81250 0.12665 0.12252 0.15714 0. 06498 1. LtOOOO 79328. 62500 0.12282 0. 11384 0.16816 0. 06262 1. 0*424 82050.68750 0.12288 0.11959 0.17552 0. 064 7 7 1. 07937 83752.81250 0.12672 0.12326 0. 18724 0. 09213 1. 1*31* 102588.06250 0. 13551 0.13411 C.19879 0. 10002 1. 21603 129635.06250 0.16177 0.15631 0.22773 0.11924 1. *^745 1*5397. 3 7500 0.13167 0.17755 0.23^42 0. 12398 i . •+7293 160984.50000 0.19925 0.19806 0.24983 0. 14652 1. 59344 207073.31250 0. .26459 0.26525 0. .31565 0. . 2 0 6 7 3 2. , U t 5 0 4 205144.06250 0. 26056 0.25650 0. 30430 0. 19271 1. 9 6 2 2 2 2 0 3 6 9 9 . 5 6 2 5 0 0. 25533 0.25597 0. 29501 0. 17939 1. 92655 214715.75000 0. . 263,90 0.26930 0- 30864 0. .18155 2. 00764 POK, R ( l ) , R(2] 1 t R 13 ) , R ( 4 ) t PK: SECTQR 4 ' 67625.31250 0.14842 0.14842 0. 15812 0. 06493 1. 00000 69709.18750 0 .14283 0.14418 0. 1613 6 0. 03196 0.99999 72668.00000 0. 14427 0.14297 0. 16698 0. 06471 1. J<:572 79600.68750 0. 14853 0.15573 0. 17660 0- 09193 1. 0o733 88173.13750 0. 15847 0.16392 0. 13850 0. 10002 1. 15x32 105271.25000 0. 18634 0.18569 0. 20711 0. 1201o 1. 29190 121455.06250 0. 20321 0.21731 0. 22674 0. 1294J 1. 42631 131928.93750 0. 22533 0.23036 0. 23914 0. 14710 1. 31758 171198.37500 0. 23955 0.29882 0. 29481 0- 2u58 5 1. 9U249 189730.18750 0. 28718 0.29049 0. 2 82 91 0. 19S-*5 1. 64240 211498.75000 0. 28777 0.29439 0. 28879 0. 17999 1. 86625 233863.43750 0. 29685 0.31447 0. 29193 0. 16172 1. 9 u l 4 5 POK. i R I D , R(2) 1 i . R<3) , R14), PK: SECTOR 5 13193.621C9 0. 15414 0.08498 0. 16471 0.06493 1. OOOOO 12313.30859 0. 14801 0.07965 0. 16570 0. 07944 0. 93520 13425.26125 0. 15058 0.07939 0. 17319 0. 067o6 i . 0x363 14702.57422 0. 15381 0.06214 0. 13419 0. Od7_>3 1. 06514 16516. 10933 0. 16263 0.03953 0. 19451 0. 09766 1. x2627 20107. 64 063 0. 19112 0.10857 0. 21257 0. 1234 J . 1. 27010 2460o.26906 0. 21545 0.i2719 0. 23491 0. 1J>43JL 1. 4x531 28352.77344 0. 23157 0.14740 0. 24945 0. 15i 32 1. Pi.22 7 39082.05766 0.30639 0.20991 0. 31439 0. 20867 1. 9ab99 37624.39344 0. 30i26 0.19932 0. 2946i 0. 1692 8 1. 6 o 9 i l 38600.71875 0. 29477 0.19360 0. 30100 0. 17939 1. 8 7 0 i t 39569.71434 0. 3 0842 0.20232 0. 30007 0.17706 1. 90493 PCK, R ( l ) , R12) » R.13), R(4), PK: s ECTOR 6 276665.06250 0. 12282 0.11941 c. 17096 0. 06493 1. UJOOO 273847.93750 0. 11696 0.11538 0. 16730 0. 0. 9 7x34 287734.5o250 0. 11786 0.11607 0. 17303 0. 06445 0.99674 317091.50000 0. 12i37 0.11992 0. 18603 0. 0 9 i o 4 1. 05873 373412.62500 0. 13020 0.12836 0. 20270 0. 09930 1. i 4 3 7 1 465952.25000 0. 15519 0.15137 0. 22422 0 . U 9 6 1 1. 29799 560133.56250 0. 17463 0.17262 0. 23174 0. 1<:906 1. ->7590 653495-93750 0. 19233 0.19201 0. 24667 0. 1465 6 1. 4d253 849407.50000 0.25819 0.25936 0' 303 93 0. 205 U 1. 69 935 862130.06250 0. 25279 0.24964 0. 30330 0. 19343 1. 8o203 919139.3i250 0. 24969 0.24922 ' 0.30401 0. io0-t3 1. o;>41 5 1003646.87500 0. 25713 0.26629 0. 31492 0. x3204 1. 9x913 POK, RJ1J. R(2J > R(3 ) , R(4J, PK: SECTOR 7 260428.00000 0. 131 78 0.12653 0. 17005 0. 06493 1. 00000 2563-38. 93750 0. 12601 0.12142 J . 17165 0. 06235 0. •9 9038 283867. 12500 0. 12803 0.12417 c. 18055 0. 06434 1. 02972 311016.50000 0. 13242 0.12844 0. 19602 0. 0914 7 1. X 0 1 2 0 360862. 68750 0 - 14232 0. 13301 0.21415 0 . 09971 1 .19717 427701. 93 750 0 .16817 0 . 16265 0.23833 0 . li.952 1 .35608 490635.93750 0. 18799 0 .18254 0.24446 0 . 12911 1 .43136 545213. 87500 0 .20554 0 . 20371 C.26030 0 . 14669 1 .53991 706199.75000 0 . 27198 0 . 27112 0.32714 0 . 20611 1 .97343 702423.93750 0. 26532 0 . 26201 0.31485 0 . 19340 i .90695 735947. 37500 U. 26302 0 . 25933 0.30847 0 . 18030 1 .37197 817932.13750 0 . 27174 0 . 27619 0.33196 0 . 18134 1 .9o729 PQK, FU1), R(2J » RJ3) , R(4 ) , PK: SECTOR 3 83920.93750 0 . 12410 0.12023 0 . 1 6 8 8 6 0 . 06498 1 .UOUOO 84627.81250 0 . 11970 0 .11691 0 .17604 0 . 081H3 1 .01296 91127. 8J.250 0 . 12304 0 . 11921 0.18893 0 . 06-+J.2 1 .07044 99729. 68750 0 . 12724 0 . 12407 0.20822 0 . 09193 1 .i 3 5 4 5 113212.56250 0 .13644 0 . 13530 0.22625 0 . 1U017 1 .25309 140203. 56250 0 . 16256 0 . 15707 0.25649 0 . ii.973 1 .44037 161469. 13750 0 . 18187 0 . 17337 0.26017 0 . 1^8oi 1 .5.1.101 186381. 0 0 0 0 0 0 . 20090 0 . 19936 0.28101 0 . 14663 1 .64t>22 256743. 12500 0 . 26895 0 . 26876 0.36423 0 . 2o528 2 . 10OU0 261751. 25C00 0.26283 0 . 25851 0.34321 0 . 19393 2 .07884 266371. 31250 0 . 25817 0 . 25801 0.33563 0 . 17979 2 . 01*484 276791. 43750 0 . 26477 0 . 27067 0.34146 0 . ldxoO 2 •0a544 PQK, R ( l ) , R12) t R(3) , R(4J, PK: SECTOR 9 23226.94922 0 . 09220 0 . 09149 0.10046 0 . 00493 1 .UUOOO 26325. 39453 0 . 09470 0 .09344 0 .11783 0 . 03233 1 . i u 9 3 5 28587.96094 0 . 09631 0 . 09545 0.12750 0 . 0o524 1 .1/493 34173. 75C00 0 . 10302 0. 10161 0 . 1 4 7 0 3 0 . 09 0 7 3 1 . 5 73 43874. 75000 0 . 11491 0 . 11516 0.17296 0 . 09925 1 .3X86^ 53346. 83203 0. 13655 0 . 13387 0.19318 0 . i i 9 2 1 1 . 7o^>99 60815. 10156 0 . 15369 0 . 15227 0 .19527 0 . 12 972 1 .8243 3 71190. 12500 0 . 16960 0 . 17151 0.21203 0 . 14522 1 .99550 91378. 63750 0 . 22072 0 . 22417 0.25414 0 . 20411 2 .^7746 98803. 75000 0 . 22434 0 . 22371 0 .26353 0 . 19246 2 . 54301 102753. 06250 0 . 22126 0. 22376 0.25696 0 . 17826 2.4tfo2i 106004. 31250 0.22692 0 . 23413 0 . 2 6 0 2 9 0 . 10033 2 .5^334 PCK, R(I ), R{2 J • RI 3 ), R(4J, PK: SECTOR 10 115103. 1 8750 0 . 13523 0 . 129o0 0 .15201 0 . 06493 1 .00000 114052. 63750 0 . 12882 0 . 12433 0.15410 0 . 05-500 0 .99091 117775. 13750 0. 13043 0 .12631 0.16082 0 . 0842i 1 . 0 ^ 3 1 9 130094.31250 0. 13397 0. 12963 0 .17494 0 . 09140 1 .U9140 153838. 62 500 0. 14314 0 . 13878 0.19106 0 . 09 9 95 1 .1835 3 205454. 75000 0. i 7006 0. 16435 0.21791 0 . 11939 i .3o877 245638. 06250 0. 1 903 1 0. 18403 0.22344 0 . 1*:893 1 .44677 289364.81250 0. 20919 0. 2J390 0.24437 0 . IH -7 03 X .56676 394398. 25000 0. 27354 0. 27316 0.32273 0 . 205u9 2 .10427 395782. 0 0 0 0 0 0. 27173 0. 26423 0-30600 0 . 19368 2 .0x3 56 416732. 25000 0. 26952 0. 26331 0 .29443 . 0. 16067 1 .95o49 428088. •062 50 0. 27605 0.27906 0 .29906 0. 131 7 7 1 .9*432 PQK, R ( i ) , R(2J, R ( 3 ) , R(4), PK: SECTOR 11 91 59190. 58970. 62651. 68925. 77189. 92829. 110405. 130244. 172423. 176802. 182271. 196216. 82422 99219 91406 68750 ooooo 56250 75000 75000 87500 81250 25000 00000 0.13421 0.12765 0.12915 0.13337 0. 14244 0.16742 0.18764 0.20395 0.27332 0.26917 0.26563 0.27208 0.12870 0.12367 0.124-r8 0.12858 0. 13957 0. 16037 0.18263 0.20370 0.27162 0.26289 0.26315 0.27672 0.17447 0.17538 0. 18255 0. 19746 0.21388 0.23750 0.24389 0.26183 0.32994 0.31985 0.31279 0.31905 0.08498 0. 06239 0.06417 0.09160 0.0^976 6.12009 0.12842 0.14722 0.20495 0.19443 0.17939 0.16231 1.00000 0.98975 1.01895 1.06747 1.17369 1.32565 1.39869 1.5x409 1.94289 1.39143 1.65333 1-39345 PQK, R I1), RC2J, R(3), R(4), 90113. 89810. 94195. 101963. 117135. 143406. 170843. 196 C37. 257542. 261096. 269345. 270761. 62500 56250 12500 75000 2 5000, 12500 50000 12500 68750 12500 12500 68750 0. 13992 0. 13403 0. 13533 0. 14036 0.15057 0. 17696 0.19779 0.21634 0.2 8431 0.27939 0.27694 0.28308 R(4), PK: SECTOR 12 0.133C9 0.15893 0. 0849d 1. OOOOO 0.12816 0.16019 0. 08213 0. 93634 0.12730 0.16671 0. 06435 1. Ol 664 0.12956 0.X805U 0. 09160 1. U6402 0.15513 0.19694 0. 09932 1. 13074 0.16451 0.22092 0. li.949 1. 34184 0.18803 0.22809 0. 12898 1. 4^453 0.21309 0.24533 0. lte> 59 1. 5<+569 0.28192 0.31263 0. 20627 1. 99434 0.27010 0.30137 0. 19291 1. 9_>231 0.28467 0.29437 0. 16026 1. o9o39 0.28038 0.29671 0. 18193 1. 92077 PQK t R H ) , R{2), R(3), R(4), PK 128070. 120629. 125472. 129116. 132990. 151897. 168857. 192996. 276547. 278378. 312086. 360511. 93750 43750 coooo 12500 87500 56250 37500 13750 25000 75COO 2 50 00 12500 0. 1482 4 0.14002 0. 13962 0.14117 0.15365 0.18332 0. 20256 0. 18382 0.30211 0.28x41 0.28371 0.28737 PUK, R ( l j , R(2), R(3), 158737, 166414. 173x05. 191 19 7. 216375. 282548. 335523. 390026. 541310. 540520. 540632. 555672. 6 3 7 5 0 4 3 7 5 0 1 2 5 0 0 2 5 0 0 0 0 6 2 5 0 5 6 2 5 0 6 2 5 0 0 2 5 0 0 0 1 2 5 0 0 43 750 OOOOO O o 2 5 0 0.12349 0. 11033 0. 11902 0.12320 0.13112 0. 15700 0.i7fa57 0.19470 0.26233 0.25633 0.25128 0.25699 4), PK: SECTOR 13 0.14037 0.17017 0 .03498 1.UJOOO 0. 13414 0.16763 0 .06555 0.9o174 0. 13642 0.17528 0 .06^93 0. V8204 0. 13949 0.18970 0 . 06621 x.ux557 0.14779 0.20163 0, .09o33 1.07633 0.17190 0.22573 0.1^024 1. ^ 4636 0.19237 0.23135 0, .12939 1.3/063 0.21375 0.24378 0. .14497 1.49932 0.26661 0.32078 0, .20495 2. J1303 0. 27522 0.30545 0. .19x21 1.9^938 0.27702 0.30054 0. .183*6 1 . 9 ^ . 4 9 5 0.29534 0.30738 0. .16109 2.U478B R(4), PK: SECTOR 14 0.12003 0.17249 0. 03496 1.OOOOO 0.11547 0. 17466 0. 06247 0.99196 0.11824 0.1798 7 0. 06427 1 . 0 x 4 3 3 0.12111 0.19625 0. 0-jis>0 1-03513 0 . 12939 0.21392 0. 09951 1.x 7356 0.x5340 0.23960 0. U 9 40 1-3*5 79 0. 17329 0.25035 0. l i 3 73 l.*t240 0 . 19451 0.26502 0. i4694 1.55299 0 . 26305 0.33360 0. 2u 5 b 3 2.00843 0. 25424 0.321o9 0. 1931 7 1.9*309 0.25135 0.30603 0. I60i6 1.67300 0.26551 0.30204 0. 16160 1.66715 Table 2A.2 Energy Data by sector, 1961 - 1972 (PQE, PE) key: PQE = energy costs PE = Tornquist price index of energy PQE, PE: SECTOR 71226.0 74591.0 77455.0 82125.0 87681.0 91893.0 95189.0 99410.0 102871.0 101532.0 10933t>.0 117479.0 1.00000 1.01027 0.93665 1.01644 1.02972 1.00703 1.02469 1.03080 C. 96209 1.03061 1.13765 1.20886 PQE, PE: SECTOR 5276.0 5550.0 5921.0 6315.0 6707.0 7523.0 7938.0 7948.0 8607.0 8882.0 9623.0 11896.0 1 .00000 0.97021 0.94245 0. 94556 0.94040 0.95778 0. 93705 1.0026 8 1.02529 1.05624 1.16665 1.22279 PQEi Pc: SECTOR 3 15114.0 1.00000 15758.0 1.07594 16298.0 0.96928 1734-+- 0 1. 01120 19348.0 1.16933 20306.0 1.23699 21673.0 1.19004 22801.0 1.03817 23900.0 1.05014 24 3 02.0 0.97426 27853.0 1.04833 31091.0 1.13428 PQEi PE: SECTOR 4 20460.0 23626.0 25176.0 27749. 0 29564.0 30737-0 31307.0 33842.0 37414.0 362 74.0 43500.0 52958. 0 PQE, PE: 3892.0 4035.0 4156.0 4561.0 5246.0 5626.0 6067.0 6525.0 7395.0 6790.0 7287.0 8354.0 PQEi Pt: 1222 71.0 127933.0 132166.0 144993 .0 158338.0 171482.0 180679.0 139853.0 208196.0 ;2i4544. 0 244242.0 256760.0 PQE, PE: 38754-0 83719.0 93346.0 106304.0 1198 95.0 123 2 29 .0 1319 5 5.0 142 944.0 142501-0 170690.0 1.00000 0.97740 0.96176 C.96570 1.01322 0.99255 1.01749 1.01720 0. 96 049 0.97929 0.99105 1.00723 SECTOR 5 1.00000 1.00337 0.98109 0.96251 0.97056 0.90036 0.90073 0.88366 0.96652 0.99413 0.99955 1.04142 SECTOR 6 1.00000 1.03139 1.04375 1.05876 1.08930 1.11826 1.13365 1.12071 1.15134 1.15416 1.32835 1.31519 SECTOR 7 1.00000 0.94224 0.93342 0.93595 1.01551 0.96216 0.99648 0.93350 0.96391 1.02442 183414.0 1.H779 186299.0 1.17419 PQEt Pc : SECTOR 19718.0 1.00000 21463.0 1.00834 22505.0 1.01568 24479.0 1.00372 27457.0 0.94990 29974.0 0.93381 30712.0 1.01215 32256.0 0.97090 34671.0 0.92304 34189.0 1.03691 36637.0 1.13470 39334.0 1.16142 PQE, Pt: SECTOR 9 6931.0 1.00000 7119.0 0.95358 7416.0 0.92028 8441.0 0.96151 923o.0 0.98109 10527.0 0.97432 11232.0 0.99122 11470.0 0.97241 12768.0 1.02660 13004.0 1.06347 14311.0 1.098<+2 17084.0 1.15899 PQE, Pt: 18087. G 19773.0 21027.0 22336.0 26081.0 29791.0 32 709.0 36408.0 39467.0 39338.0 46012.0 51699.0 SECTOR 10 i .00000 ' 0.34460 0.91384 0.84325 C.90024 0.39220 0.90665 0.904 7 7 0. 93537 0.94675 1.0^455 1.09187 PQE, PE: 11338.0 12145.0 13171.0 i 3'9i3.0 1493 9.0 1583o.0 SECTOR 11 1-00000 C. 87868 0.91427 0.92186 0.90C30 0.88371 17389.0 1.02622 18189.0 0.85363 19397.0 0.98607 19494.0 1.09967 21019.0 1.08256 22954.0 1.20065 PQE, PE 46566. 0 50815. 0 51706. 0 55685. 0 62425. 0 67391. 0 64868. 0 67968. 0 71837. 0 74112. 0 83340.0 94533. 0 SECTOR 12 1.00000 0.92910 0.87781 0.92655 0.91458 0.91663 0.95331 0.97613 1.04668 0.79162 0.82646 0.90801 PQE, PE: 113 5.1.0 10850.0 . 11337.0 12579.0 12576.0 13399.0 13695.0 16763.0 18522.0 19178.0 21421.0 23875.0 SECTOR 13 1.00000 1.04434 1.07662 1.02235 1.16630 0.97010 1.01510 1.07076 1.09667 1.15359 1.23957 i.33631 PQE, PE: 54660.0 56047.0 59901.0 63677. 0 69941. 0 80559.0 83693.0 88946.0 96929.0 104 708.0 112038.0 117690.0 SECTOR 14 1.00000 0.96941 0.97238 0.97767 1.06192 1.03350 1.045 56 1.05543 1.09149 1.00979 1.11761 1.14794 Table 2A.3 Labour Data by sector, 1961 - 1972 (Net QL, Net PQL, PL) key: Net QL = quantity of labour net of R&D employment Net PQL = wage b i l l net of R&D wages PL = price index of wages NET QL, NET PQL, PL: SECTOR 1 210430.0 782156.0 1 .00000 209868.0 815c24.0 1 .C4548 209649.0 845921.0 1 .C8555 214513.0 902778.0 1 .13225 220223.0 968312.0 1 .13295 226731.0 1054331.0 1 .25107 228224.0 1135657.0 1 .33875 225947.0 1206192.0 1 .43623 223589.0 1288128.0 1 .54997 221235.0 1377935.0 1 .67567 217744.0 1464839.0 1 .80991 219867.0 1590J.76.0 1 .94581 NET QLi NE T PQL, PL: SECTOR 21665.0 94995.0 1 .COOOO 22622.0 103331.0 1 .04224 23946.0 110126.0 1 .04685 24750.0 121447.0 1 . 11910 25976.0 132808.0 1 . 16603 27565.0 147102.0 1 .21703 26636.0 154038.0 1 •3189i 24552.0 152812.0 1 .41948 24976.0 16 385 8.0 A .54190 23784.0 172614.0 1 . 657H 23503.0 162450.0 1 .77043 25383.0 214859.0 1 .93049 NET QL, NE T PQL, PL: SECTOR 64 828.0 223974.0 1 .00000 67773.0 242241.0 1 .03456 70028.0 261710.0 1 .03172 74196.0 290036.0 i - 13165 76307.0 312903.0 1 .13689 76919.0 33 9266.0 1 .27665 77016.0 357159.0 1 .342 2 9 72881.0 361453.0 1 .43550 75022.0 396617.0 1 .53020 72363. 0 410856.0 1 .64338 72315-0 443909.0 73939.0. 478505.0 NET QL, NET PQL, PL: 82075.0 83446.0 86856.0 89375.0 91562.0 91908.0 89828.0 90256.0 92483.0 87756.0 91775.0 102619.0 292664.0 311905.0 340636.0 366886.0 398773.0 428920.0 450837.0 490269-0 540927.0 551115.0 637465-0 770117.0 1.77677 1.87191 SECTOR 4 1.00000 1.04323 1.09985 1.15122 1.22138 1.30377 1.40750 1.52335 1.64023 1.76119 1.94793 2.104o0 NET QL, NET PQL, PL: 33454-0 34341 .0 35895.0 37965.0 40353.0 43532.0 43882.0 43158.0 44224.0 42188.0 42969.0 46854.0 97520.0 99 300.0 100775-0 105435. 0 108953.0 115593.0 117279.0 116646.0 120653.0 120722. 0 119422 .0 119858.0 117029.0 125077.0 134344.0 148096.0 164019.0 189681.0 201675.0 210971.0 232644. 0 234732.0 253031.0 293749.0 489517.0 513022.0 534541.0 530475.0 625463.0 717080.0 770962.0 824483-0 914196.0 972035. 0 10338-+2.0 1124051.0 SECTOR 5 1.COOOO 1.04116 1.06989 1.11510 1.16177 1.24415 1.31377 1.39738 1.50379 1.59052 1.68334 1.82270 SECTOR 6 1.COOOO 1.02923 1.05671 1.09679 1.14364 1.23579 1.30960 1.40811 1.50942 l-o0406 1.72463 1.86329 NET QL, NET PQL, PL:-NET QL, NET PQL, PL: 89 038.0 90 740.0 92992 .0 993 34.0 106558.0 112578.0 471793.0 492570.0 52 2 413.0 5 7 707 0. 0 644720.0 703339.0 SECTOR 7 1.COOOO 1.02503 1.06081 1.09 69 3 1.14249 1.18819 ' 111640.0 111890.0 109417.0 114975.0 112578.0 112210.0 741991.0 792667.0 324426.0 943195.0 1000316.0 1091492.0 1.25501 1.33773 1.42277 1.54905 1.67784 1.33678 NET QL, NET PQL, PL: SECTOR 8 100846. 0 456791 . 0 1 . OOOOO 109313. 0 503147. 0 1. 02626 112923. 0 541984. 0 1. C5956 120712. 0 60033 9. 0 1. C9688 133716. 0 689580 . 0 1. 13852 143031 . 0 792815. 0 1. 22372 138 929.0 815645. 0 1. 29613 137268. 0 862140.0 1. 33659 141125. 0 955297. 0 1. 49443 139227. 0 1008871. 0 1. 59975 135584. 0 1048042. 0 1. 70652 138011 . 0 1147806. 0 1. 33610 NET OL, NET PQL, PL : SECTOR 9 50137. 0 241428. 0 1. coooo 53908. 0 269454. 0 1. U3801 58096. 0 302346. 0 1. 03076 63151. 0 342746. 0 1. 12710 70060. 0 395212. 0 1. 17147 74833. 0 450550. 0 1. 25032 78277. 0 497184. 0 1. 31903 75450. 0 511560.0 1. 40802 80658. 0 592912. 0 1 . 52656 78519 . 0 624323. 0 1. 65122 69279. 0 566446. 0- i X . 70395 76244 . u 663557. 0 1. 80735 NET QL, NET PQL, PL: 96933.0 103031.0 110291 . 0 122039.0 133140.0 ' 144810 . 0 147724.0 147.030. 0 155319.0 144821.0 143263.0 155837.0 483090 .0 533216.0 606793.0 693431.0 811071.0 903498.0 958971.0 1032013.0 120430 7.0 1185736.0 1315 074.0 1438851. 0 SECTOR 10 1.OOOOO 1.04820 1.10451 1.14070 1.22298 1.25255 1.30323 1.47688 1.55661 1.64371 1.78068 1.91300 SECTOR 11 1.COOOO 1.02409 NET QL, NET PQL, PL: 86960.0 395844 .0 94195 .0 439107.0 97974.0 101595.0 108939.0 119817.0 121629.0 11852 7. C 120752.0 114041.0 116469.0 115196.0 NET QL, 467432.0 501871.0 552133.0 633 099.0 663226.0 697728.0 758105.0 782935.0 851407.0 903550.0 PQL, PL: 1.04810 i .08521 1.11341 1.16078 1.20693 1.29320 1.57921 1.50830 1.60591 1.72310 SECTOR 12 43160.0 45306.0 45837.0 48311 .0 51052.0 53009.0 51098.0 51491.0 51709.0 49784.0 51601.0 52911 .0 NET QL, N 16047.0 15392.0 15004.0 14546. 0 13731. 0 14744.0 14988.0 14903.0 14905.0 14952.0 14864.0 14 733.0 6 1 3 1 8 . 0 6 1 7 4 9 . 0 6 3 1 5 4 . 0 6 4 9 8 o . 0 6 8 3 5 3 . 0 7 0 6 5 7 . 0 7 2 2 4 9 . 0 7 4 0 6 0 . 0 7 5 7 5 6 . 0 764 5 4 . 0 74 791 . 0 72161 . 0 190935.0 209157.0 217168.0 233969.0 267700.0 295592.0 300080.0 3243 86.0 355346.0 3o2728.0 406320.0 456024.0 9 3 3 7 5 . 0 1 0 2 1 5 2 . 0 9 3 4 0 4 . 0 9 9 3 1 8 . 0 9 3 6 8 7 . 0 1 1 4 0 7 5 . 0 1 2 2 6 2 0 . 0 1 3 1 3 0 7 . 0 1 4 3 9 9 1 . 0 1 5 2 7 5 8 . 0 1 6 4 8 3 8 . 0 1 7 7 0 3 0 . 0 2 0 6 9 6 5 . 0 3 1 9 7 3 4 . 0 3 4 0 9 3 0 . 0 5 6 1 7 o l . 0 393 8 6 1 . 0 4 3 0 9 3 7 . 0 4 6 3 5 3 9 . 0 5 1 1 2 3 8 . 0 5 6 5 1 2 8 . 0 6 0 3 2 8 6 . 0 6 4 0 9 1 0 . 0 66 7 6 9 2 . 0 1.COOOO 1.04355 1.07096 1.11813 1.18531 1.25196 1.32748 1.42405 1.55558 1.64698 1.779 94 1.S4822 1.CCOOO 1.04852 1.06983 1.11376 1.17238 1.26207 1.33452 1.43722 1.57584 1.666 53 1.80897 i.95396 SECTOR 14 1 . 0 0 0 0 0 1 . 0 3 4 4 9 1 . C 7 3 3 6 1 . 1 1 1 9 9 1 . 1 5 0 9 4 1 . 2 1 8 4 5 1 . 2 3 1 6 0 1 . 3 7 3 9 2 1 . 4 9 0 1 5 1 . 5 8 9 3 0 1 . 7 1 1 7 3 1 . 8 4 8 3 0 NET QL, NET PQL, PL: ET PQL, PL: SECTOR 13 Table 2A.4 R&D Data by sector, 1961 - 1972 (RD, SRD, PRD) key: RD = constant dollar R&D expenditures SRD = constant d o l l a r stock of R&D PRD = price index of R&D expenses RD, SRD, PRD: SECTOR 1 2 . 99400 2. 99400 1 .COOOO 3. 44200 6. 31238 1 .03725 4. 95600 10. 85920 1 .03999 6 . C0100 16. 04243 1 .15777 6 . 56100 21- 35780 1 .23434 9 . 13200 27. 97581 1 .37987 9 . COOOO 33. 79512 1 .54657 10. 00000 39. 75901 1 .68240 10. 20000 45. 274i7 1 .84276 10. 80000 50. 61960 2 .02042 12. 00000 56.06821 2 .20239 12. 10000 61 . 14075 2 .38539 RD, SRD, PRD: SECTOR 2 1. 53400 1. 53400 1 •COOOO 1. 66500 3. 13920 1 .03725 2. 03600 5. 0 0710 1 .08999 2. 40100 7. 08092 1 .15777 2. 83700 9. 37930 1 .23434 3. 30700 11. 77590 1 .37937 3. 70000 14. 16829 1 .54657 4. 10000 16. 60529 1 .68240 4. 50000 19. 04726 1 .34276 4. 30000 21. 17552 2 .02 042 4. 80000 23. 35497 t. .20259 5. 20000 25. 53490 2 .38539 RD, SRD, PRD: SECTOR 3 1.57900 1.66100 2.76500 3.42 00 0 4.16800 4.8ol00 4.20000 4.70000 5.10000 4.30000 1.57900 3.18034 5.71706 3.67101 12.04770 15.57050 18.28618 21.07960 23.84738 25.97565 1.00000 1.03725 1.08999 1.15777 1.23434 1.37987 1.54657 1.68240 1.84276 2.02042 4.00000 27.79184 2.20239 4.60000 29.72023 2.38539 ), SRD, PRD • SECTOR 4 0.16500 0 .16500 1 .00000 0.23800 0 .39445 •1 .03725 0.19800 0.57610 1 .08999 0.24400 0 .78685 1 .15777 0.41700 1 .12469 1 .23434 0.44900 1 .45008 1 .37987 1.30000 2 .29065 1 .54657 0.80000 2 .76616 1 .68240 0.70000 3 .14602 1 .84276 1.30000 3 . 73 945 2 .02042 1.30000 4 .37972 ~> *— .2 023 9 1.30000 4.92470 2 .38539 », SRD, PRD • SECTOR 5 0.11400 0 .11400 1 .00000 0.12400 0 .23355 1 .03725 0.11900 0 .34272 1 .08999 0 .12900 0 .45414 1 -15777 0.10400 0 .53840 1 .23434 0.11100 0 .61834 1 .37987 0.20000 0 .74616 1 .54657 0.20000 0 .86704 1 -o8240 0.30000 1 .02983 1 .84276 0.60000 1 .32680 2 .02042 0.70000 1 .64464 2 .20239 0.80000 1 .98001 2.38539 SRD, PRD SECTOR 6 6. 95000 6.95000 1 •COOOO 8. 49100 15.13604 1 .03725 15. 01300 26.90952 1 .08999 20. 13 600 46.30156 1 .15777 26. 56500 67.82312 1 .23434 27. 69200 87.89166 1 .3798 7 25. 80000 104-57370 1 .54657 23. 10001 113.30409 1 .68240 22. 60001 130.36828 1 .84276 21. 60001 141.25914 2 .02042 18. 80 00 0 149.79530 2 .20239 18. 60001 157.59276 2 .33539 RD, SRD, PRD: SECTOR 7 8.46700 9.20600 14.58600 17.39200 21-45500 25.32800 8.4o700 17.34236 30.72^09 4 5. 74 6 06 63.12776 81.48311 1.COOOO 1.03725 i.03999 1.15 777 1.23434 1-37987 26-39999 22.89999 28.20000 31.80000 34.00000 36.30000 98.55310 112.16461 127.46771 143.20703 158.64478 173.86240 1.54657 1.68240 1.64276 2.02042 2.20239 2.38539 RD, SRD, PRD: SECTOR 8 2.55900 3.42300 4.01300 3.56600 2.45600 3.11500 3.20000 4.40000 5.10000 4.20000 5.60000 5.40000 2.55900 5.85906 9.54073 12.62079 14.61051 16.86797 18.93704 21.55235 24.31993 26.3 93 70 28.94138 31.20514 1.00000 1.03725 1.08999 1.15777 1.23434 1.37907 1.54657 1.68240 1.84276 2.02042 2.202J9 2.38539 RD, SRD, PRD: SECTOR 9 5.49100 6.12000 6.81800 8.11800 8.61400 11.16600 14.20000 16.80000 18.50000 20.OOOOO 2i.60001 21.60001 5.49100 11.39120 17.64o27 24.65002 31.65663 39.72860 48.91026 58.89601 68.93527 78.83421 63.64i72 97. 696.34 1.00000 1.03725 1 .08999 1. 15777 1.23434 1.37937 1.54657 1.63240 1.84276 2.02042 2.20239 2.38539 RD, SRD, PRD: SECTOR 10 17.38699 17.38699 1 .OOOOO 18. 34599 35.07403 1 .03725 32. 03701 64.51187 1 .03999 42. 70100 101.59397 1 .15777 59. 74001 i49.7 9216 1 . 23434 52. 97301 188.18199 1 .37987 45.39999 217.53722 1 .54657 48. 8000 0 246.54343 I A. .63240 58. 50000 278.28906 1 .84276 48. 10001 302.09593 2 .02042 47. 50000 323 .66333 2 .20239 61. 39999 349.40332 2 .38539 RD, SRD, PRD: SECTOR 11 31.93100 31.93100 1.00000 32.06000 62.83954 1.03725 43. .74699 102. .97462 1.08999 54.42200 149. .98048 1.15777 72. .57300 203. .77934 1-23454 87. .12601 271, . 91992 1.37967 95, .20000 333.47334 1.54657 89. .70000 336. .79199 1-68240 103. , 89999 443. , 17456 1.84276 105.50000 495. .39136 2.02042 117.20000 548. .60596 2.20239 107.20000 593.54614 2.38539 RD, SRD, PRD: SECTOR 12 1. 52700 1. 52700 1.COOOO 1. 57100 3 . 04158 1.03725 2 . 07100 4. .94159 1.03999 1. 99600 6. 66559 1-15777 1. 34400 . 3. 15950 1.23434 2 . 7950 0 10. 18505 1.37967 3 . 00000 12. 12483 1.54657 3 . 20000 14. 02637 1.68240 3 . 60000 15. 98046 1.84276 3. 70000 17.31175 2.02042 3. 60000 19. 44633 2.20239 4 . 20000 21. 20705 2.38539 RD, SRD, P SECTOR 13 6. 06900 6. 06900 1.00000 7. 46000 13. 26107 1.03725 11. 19400 23. 53035 1 .03999 18. 23801 39. 28354 1.15777 22. 54601 5 7. 54912 1.23434 19. 23900 71. 52795 1.37987 20. 7000 0 84. 91238 1.54o5 7 23. 30000 93. 76166 1.68240 23. 500J0 111. 51424 1.84276 17. 70000 120. 27460 2.02042 17. 50000 123. 22069 2.20239 19. 20000 136. 26967 2.38539 RD, SRD, PRD: SECTOR 14 23. 25101 23.25101 1 .COOOO 24. 0&7C0 46. 4 5363 1 .05725 27. 10300 71. 31891 I .03999 35 . 86400 102. 29508 1 .15777 39.87601 134. o0112 1 .23434 45. 23100 167. 38028 1 .37987 4o. 20000 i97. 25273 1 .54657 45. lOOOi 224. 05975 1 .63240 56. 60001 254. 7 744 9 1 -84276 54. 20000 281 . 60059 2 .02042 40. 39999 303. 57666 2 .20239 50. 10001 324. 5 7935 2 .38539 Table 2A.5 Scale of Production by sector, 1 9 6 1 - 1 9 7 2 (PY, QY, Total Y) key: PY = Tornquist price index of output QY = Tornquist quantity index of output Total Y = Total output as measured by inputs PY, QY, TOTAL Y: SECTOR 1 1. .00000 1. .00000 1070708, .00 1 .03201 1. .01069 1116737. .00 1. .06324 1.02326 1164904. .00 1. .11575 1. .04960 1253886. .00 1. .17087 1.08115 1355393. .00 1. .25523 1. .11795 1502502-00 1. .33478 1. .13745 1625581. . 00 1. .43495 1. .146 74 17ol836. .00 1. .62154 1, .15312 2010503.00 1. .69626 1. .15897 2104690. .00 1. . 77610 1. .16125 2203C95. . 00 1. . 87759 1. .18043 2372323. ,00 u QY, TOTAL Y: SECTOR 2 1. ,00000 1. 00000 125071. 56 1. 03367 1. 04212 134727. 56 1. 04513 1. 09729 143433. 44 1. 11185 I 12936 . 157049.56 1. 16624 1. 1 3576 172959. 56 1. 23 2 74 1. 25655 193736. ,25 1. 32683 1. 26103 209269. 38 1. 42125 1. 19730 212910. 50 1. 60366 1. 23170 247023. 13 1. 67421 1. 22 759 257030. 13 1- 74488 1. 24646 271999. 25 1. 87431 1. 39184 326337. 06 QY, TOTAL Y: S ECTOR 3 1. 00000 1. 00000 316151. 31 1. 03879 1. 02713 337327. 63 1. 07536 1. 05357 360058. 69 1. 12351 1. 11184 596682. 81 1. 19265 1. 15 324 434839. 06 1 .30568 1. 18527 439257 . 06 1. 36 604 1. 21388 524229. 33 1. 45336 1. 1 3663 545233. 50 1. 63836 i . 21172 627595. 3 i 1. 6923 8 1. 19694 640332. 06 1. 76983 1. 20726 675461. 56 1.86055 1.23144 724311.75 PY, QY, TOTAL Y: SECTOR 4 1 .00000 1 .00000 380749. 31 1 .03557 1 .02777 405240. 19 1 .07840 1 .06790 .438480. 00 1 .12843 1 .10377 474235. 69 1 .19603 1 .13423 516515- 19 1 .28575 1 . i 5399 564928. 25 1 .38626 1 .14360 603599.06 1 .48909 1 .15712 656039. 94 1 .64562 1 .19632 749539. 58 1 .72005 1 .16667 777119. 19 1 .85474 1 .26364 892463. 75 1 .97H8 1 .40833 1056933. 00 QY, T O T A L ,Y: SE CTOR 5 1 .00000 1 .00000 134114. 56 1 .03464 1 .02316 141975. 25 1 -06231 1 .06635 151925. 25 1 .10567 1 .12861 167359. 56 1 .15241 1.20204 185733. 06 1 .23512 1 .30043 215414. 63 1 . 309-fO 1 .32311 232343. 25 1 .39047 1 .31637 245843. 75 1 .53266 1 .35803 279121 . 06 1 .59 736 1 .3031* 279146. 33 1 .67600 1 .32997 298918. 69 1 .79964 1 .43645 346672. 69 QY, Ti J T A L Y: S E C T O R 6 1 .00000 1 .00000 883453 .06 1 .01149 1 .01798 914322 .94 1 .03624 1 .03670 954441 .56 1 .07967 1 .08687 1042564 .50 1 .13773 1 .14473 1157213 .00 1 .23951 1 .22999 1354514 . 00 1 .30652 1 .30256 15119 74 .00 1 .39221 1 .34640 1667331 .00 1 .59345 1 .33348 1971799 .00 1 .63130 1 .413 59 2048709 .00 1 .71083 ]. .44559 2i97273 .00 1 .80048 1 .49080 2384657 .00 , QY, T O T A L Y: S E C T O R 7 1 .OOOuO 1 .00000 820975 .00 1 .00499 1 .01770 639677 . 94 1 .03662 1 .05689 89962b .13 1 . 03 00 3 .1 .12148 994390 .50 1 . 14522 1 .19707 1125475 .00 1 .21380 i .26877 1264319 .00 1 .27860 1 .29998 1364581 .00 1.35735 1.54251 1.59684 1.66647 1.79592 1.32883 1.32128 1.38554 1.40320 1.42146 1480824.00 1673126.00 1816308.00 1919677.00 2095723.00 PY, QY, TOTAL Y : SECTOR 8 1. OOOOO 1. OOOOO 560429. 94 1. 02371 1. 07063 614237. 81 1. 05942 1. 10422 655616. 81 1- 10301 1. 17290 725047. 69 1. 14713 1. 29143 830249.56 1. 24050 1. 38517 962992. 56 1. 31283 1. 36983 1007846. 19 1. 40410 1.37343 103C777. 00 1. 56503 1. 42155 1246711. 00 1. 64160 1. 41340 1304811. 00 1. 71926 1. 40233 1351050. 00 1. 82842 1. 42877 1463933. 00 PY, QY, TOTAL Y : SECTOR 9 1. . OOOOO 1. .00000 271585, .94 1. .04181 1. .07054 302398. .38 1. .08443 1. .14883 338349. .94 1. .13797 1. .24689 385360. .75 1-19391 1. 38281 449372.75 1-28107 1. .47358 514423. .81 1. .35014 1. 35241 569231. .06 1. 44279 1. 51652 594220. , 13 1. . 59034 1. 61508 697558. .69 1. 70695 1. 58796 736130. 75 1. 74 8 22 1. 44389 685510. .06 1. 34333 1. 57140 786645. 31 U QY, TOTAL . Y : SECTOR 10 1. OOOOO 1. OOOOO 616280. 19 •1. 03119 1. 05749 672041. 69 1. 03381 1. 11626 745595. 19 1. 12151 1. 22379 845861. 31 1. 20439 1. 33 51i 990990.63 1. 25784 1. 46904 1133743. 00 1. 31134 1 . 53053 1237313. 00 1. 4719 3 1. 55257 1408305.00 1. 63225 1. 62870 1638172. 00 1. 63118 1. 56456 1620856. 00 1. 77435 1. 62552 1777813. oo' 1. 33716 1. 69286 1968633. 00 QY, TOTAL Y: SE CTOR 11 1. OOOOO 1. OOuOO 466372. 01 1. 01611 1. 07668 510230. 94 1. 04106 1. 11890 543254. 08 1. 0 8 1 0 4 1 . 1 5 9 7 5 5 8 4 7 1 4 . 69 1. 1 1 4 7 7 1. 2 3 9 1 9 6 4 4 2 6 1 . 00 1. 1 7 1 8 7 1 . 3 5 7 3 1 7 4 1 8 1 4 . 56 1. 2 2 3 7 1 . 1 . 3 9 4 8 0 7 9 6 0 2 0 . 75. 1. 3 0 5 9 6 1. 3 8 9 3 0 8 4 6 1 6 1 . 75 1. 4 3 9 1 1 1 . 4 1 5 4 5 9 4 9 9 2 5 . 88 1. 5 4 1 6 3 1 . 3 6 2 1 6 9 7 9 2 3 1 . 81 1. 6 1 4 9 6 1. 4 0 0 4 4 1 0 5 4 6 9 7 . 00 1. 7 1 9 2 6 1 . 4 0 0 3 3 1 1 2 2 7 2 0 . 00 P Y , QY, TOTAL Y : SECTOR 12 1. 0 0 0 0 0 1 . 0 0 0 0 0 3 2 7 6 1 4 . 63 1. 0 1 1 5 5 1. 0 5 5 4 7 3 4 9 7 3 2 . 56 1 . 0 2 6 4 5 1 . 0 7 9 6 6 3 6 3 0 6 9 . 13 1. 0 7 9 1 7 1 . 1 2 1 8 0 3 9 6 6 1 7 . 75 1. 1 4 0 o 4 1. 1 9 7 0 0 4 4 7 3 1 0 . 25 1. 2 2 0 3 3 1. 2 6 1 6 1 5 0 4 3 8 9 . 13 1. 2 9 1 2 6 1. 2 6 6 5 8 5 3 5 7 9 1 . 50 1. 3 3 2 7 6 1. 2 9 8 3 8 5 8 3 3 9 1 . 13 1. 5 9 8 9 7 1. 3 0 8 1 5 6 8 5 2 2 5 . 69 1. 5 8 0 4 4 1. 3 * 3 0 4 6 9 7 9 3 c 13 1. 6 4 3 5 4 1. 4 0 9 7 0 7 5 9 0 0 5 . 13 1. 7 5 1 9 1 1. 4 3 1 1 0 8 2 1 3 3 3 . 69 P Y , QY, TOTAL Y : SECTOR 13 1 . 0 0 0 0 0 1 . 0 0 0 0 0 2 3 7 7 9 6 . 94 1 . 0 0 1 6 3 0 . 9 3 0 8 9 2 3 3 6 3 1 . 44 1 . 0 2 2 8 9 0 . 9 6 7 0 0 2 3 5 2 1 3 . 00 1 . 0 5 6 2 2 0 . 9 5 9 5 3 . 2 4 1 0 1 3 . 13 1 . 1 2 1 7 2 0 . 9 1 5 7 0 2 4 4 2 5 3 . 88 1 . 2 3 9 1 4 0 . 9 4 3 1 0 2 7 9 3 7 1 . 56 1 . 3 3 8 1 2 0 . 9 5 9 0 5 3 0 5 1 7 2 . 38 1 i -45232 0 . 9 8 7 5 9 3 4 1 0 7 1 . 19 1 . 7 9 3 0 2 1 . 0 2 9 6 8 4 3 9 0 6 0 . 25 1 . 7 8 1 6 3 1. 06 282 4 5 0 3 1 4 . 7 5 1 . 8 4 0 2 0 1 . 1 3 8 7 5 4 9 8 3 4 5 . 25 1 . 9 6 2 0 1 1. 2 0 3 3 3 5 6 1 4 6 6 . 13 P Y , QY, TOTAL Y : S E C T O R ' 1 4 1 . 0 0 0 0 0 1. 0 0 0 0 0 5 2 0 3 6 2 . 69 1 . 01 43 9 1. 0 2 7 2 7 5 4 2 2 4 5 . 44 1 . 0 4 7 1 1 1 . 0 5 3 5 3 57 3 9 5 6 . 13 1 . 0 8 9 0 6 1 . 0 8 3 1 1 6 1 6 6 3 5 . 25 1 . 1 4 3 5 5 1. 1 3 8 0 6 6 3 0 1 7 7 . 06 1 . 24 44 2 1 . 2 2 6 3 4 7 9 4 0 9 4 . 56 1- 3 0 6 9 0 1. 2 9 3 1 0 3 8 2 7 6 0 . 63 1. 3932 5 1. 3 6 0 9 7 9 9 0 2 1 0 . 25 1 . 6 2 3 8 9 1. 4 2 4 1 8 1 2 0 3 3 6 7 . 00 1 . 6 3 9 3 7 1. 4 6 9 0 7 1 2 5 3 5 1 4 . 00 1 . 68 38 9 1. 4 72 03 1 2 9 3 5 3 0 . 00 1 . 7640 3 1 . 4 6 1 0 0 1 3 4 1 0 5 4 . 00 Chapter 3 INFORMATION PREFERENCES AND ATTENTION PATTERNS IN R&D INVESTMENT DECISIONS Introduction Simon and others have argued that firms make decisions on the basis of partial knowledge (e.g. see March and Simon, 1958; Cyert and March, 1963). High costs associated with obtaining and processing information, constrained computational a b i l i t i e s and limited spans of attention induce the development of search heuristics. These heuristics are often.institu-tionalized in the form of standard operating procedures for the selection and evaluation of alternatives. The problem of R&D project selection constitutes one of the more complex areas for firm decision making. It deals with risky alternatives, the outcomes of which are realized over a long time horizon and are sub-jected to risks from both internal and environmental sources. The importance of information preference and attention patterns for R&D policies are often ignored. Generally, the focus of R&D inducement policies has been upon creating economic incentives or direct investment in R&D. Attempts to employ information strategies to induce R&D seem to ignore the fact that what is good news to some is no news to others. By 108 109 i d e n t i f y i n g d i f f e rences in information attended t o , s t ra teg ie s and i n t e r -ventions to create favourable c l imates f o r R&D can be designed, in p a r t i c u l a r , fo r target populat ions , thereby increas ing t h e i r po tent ia l impact. In th i s study, we focus upon empir ica l i n ve s t i g a t i on of in forma-t i o n preferences and a t tent ion patterns in R&D dec i s i on making. In par -t i c u l a r , the a s soc ia t i on of preference and a t ten t ion patterns to dec i s i on makers' a t t r i bu te s are inves t i ga ted. Methodology On the basis of extensive l i t e r a t u r e search, an " in format ion basket" was constructed, combining dimensions judged re levant to R&D d e c i s i o n making, according to a v a r i e t y of normative perspect ives and behavioural theor ie s . The " in format ion basket" was used as a bas is f o r con s t ruc t i ng a quest ionnaire to s o l i c i t preferences fo r information items. A sample of execut ives was asked to ind ica te preferences f o r in format ion. Ana l y s i s of these responses to i d e n t i f y r e l a t i on sh ip s between preferences and i n d i v i d u a l and organ izat iona l a t t r i b u t e s was conducted. In the fo l lowing sect ions descr ib ing the methodology are provided d e t a i l s concerning: (1) t h e c o n s t r u c t i o n a n d a - p r i o r i j u s t i f i c a t i o n o f t h e " i n f o r m a t i o n b a s k e t , " ( 2 ) t h e e x p e r i m e n t , ( 3 ) t h e sample, (4) t h e s t a t i s t i c a l a n a l y s i s , a n d ( 5 ) t h e c o n f i r m a t i o n o f t h e a p r i o r i i n f o r m a t i o n d i m e n s i o n s a n d t e s t i n g o f t h e r e l e v a n c e o f t h e " i n f o r m a t i o n b a s k e t . " 110 (1) Construction of the Information Basket Information items were selected after a review of the R&D, general economic and decision making literatures. Table 3.1 l i s t s the items and their sources in the literautre (also see Chapter 1 for general discussion). The items were classified into six general dimensions: general economic trends and expectations, government role in the economy, information, market, firm and project attributes. (2) The Experiment Forty-seven items were selected to represent the various sub-dimensions in the information basket. Subjects were asked to rate their importance on a seven point Likert scale. The questionnaire is presented in Table 3.2. , (3) The Sample A sample of 330 executives was randomly selected from.the directory of R&D establishments in Canada (Ministry of State for Science and Technology, 1974) plus the remainder of 'Top 100' firms (Morgan, 1975) not included in the R&D directory. A second mailing was sent to replace those firms which had declined to participate. Reminders were sent to those firms which had not responded i n i t i a l l y . The questionnaire was sent to the R&D director or to the president of the firm. The response "rate was 40%. Distributions of the R&D population and of the respondents by industry of primary involvement are presented in Table 3.3. The response, distribution corresponds well with the population distribution indicating Table 3.1 Var iables Relevant for R & D Decision Making Concep t Item Source ECONOMY. GENERAL P a s t & C u r r e n t Trends average p r o f i t rate in the economy short term bank i n te re s t ra tes * . stock market trends* general trends i n inventor ies general trends of growth unemployment Keynes (1964), Galbralth (1973), Preston (1975), Trendicator (Raguley and Booth, 1975), U.S. Department of Commerce (1975) Demand Changes expected growth of real GNP growth of population Schumpeter (1971) Quinn (1966), Br ight (1970) Inducements t o Change expected wage settlements expected general p roduct i v i t y changes expected rate of i n f l a t i o n expected energy requirements Schmookler (1966), Rosenberg (1974), Kamien & Schwartz (1968), Fe l lner (1971), Hamlen & Ruttan (1970) Demand and C o s t Changes expectations with respect to the foreign exchange rate Leonard (1971) ECONOMY. GOVERiljjENT ROLE" " " ' . " C o s t s , D i r e c t government subsidies fo r R&D p ro jec t s * low i n te re s t government loans for R&D pro jec t s * p o s s i b i l i t y of gaining a new government contract f o r part of the p ro jec t * favourable tax p o l i c i e s fo r R&D projects T i l ton (1971), Brooks (1972), Quinn (1966), Hamberg (1966) Leonard (1971), Foster (1971), Arrow (1962) . Costs, I nd i rec t accelerated depreciat ion of R&D expenditures for tax purposes accelerated deprec iat ion o f new cap i ta l equipment expenditures f o r tax purposes* low i n t e re s t rates on government bonds high i n te re s t rates on government bonds i n t e re s t rates on government bonds increas ing i n te re s t rate on government bonds dec l in ing Hamberg (1966) Keynes (1964), G a l b r a l t h (1973). CONTINUED Table 3.1 (Continued) Concept Item Source Information government funding of feasibility studies for R&D projects availability of sound government information on tech-nological change* government support and promotion for market development availability of government surveys of market potential Thurston (1971), Brooks (1972), Bright (1968), Foster (1971), Drucker (1975) Market Influence growth of government expenditures favourable tariff policy pollution control measures (environmental concern) Bright (1968), Foster (1971) Thurston (1971), Smith (1973) INFORMATION Private Government availability of private surveys of market potential availability of government surveys of market potential* availabil ity of sound government information on technological change* government funding of feasibi l ity studies for RJD projects* f Thurston (1971), Brooks (1972), Bright (1968), Foster (1971), Drucker (1975) MARKET average profit rate of industry group stabi1ity of market barriers to entry in the market Ansoff & Stewart (1967) Mansfield (1959), Schumpeter (1971), Scherer (1971), Comanor (1967), Williamson (1955), Baldwin & Childs (1969), Cooper (1966), Galbraith (1973) FIRM Demand recent growth of sales of firm expected growth of sales of firm stage in l i fe cycle of existing products-Lithwick (1959), Leonard (1971), Mansfield (1968). Hamberq (I960) Kotler (1967), Fjuinn (1965), Tilles (1965), Ansoff and Stewart (1967) Supply, Factors availability of scientif ical ly trained personnel Brooks (1972), Cooper (1966) Liquidity average profit rate of firm accelerated depreciation of new capital equipment expenditures for tax purposes* short term bank interest rates* stock market trends* government subsidies for R&D projects* low interest government loans for RSD projects* Scherer (1971), Williamson (1965), Gal bra 1th (1973) Cyert & March.(1963), Mansfield (1968), Hamberg (1966), Tllton (1971), Brooks (1972) CONTINUED Table 3.1 (Continued) Concept Item Source Ir.novativeness h i s to ry of success with R&D ( f i r m ' s ) Ansoff & Stewart (1967) PROJECT-Ccmnitment, Money cost of the R8D project r e l a t i v e to to ta l sales of f i rm Mansfield (1964), Scherer (1971), Gerstenfeld (1971), Cooper (1966), T i l l e s (1965), Ansoff & Stewart (1967), Mottley & Newton (1959), A l len (1970) and Resources p o s s i b i l i t y of gaining a new government contract fo r part of the p ro jec t * Brooks (1972), T i l ton (1971), Hamberg (1966) CoTinitment. Tine expected payback period for the R&D project Leonard (1971), Mansfield (1958), Gerstenfeld (1971), Br iaht (1968), Kot ler (1967), Ansoff & Stewart (1957) Brooks (1972), T i l l e s (1966), Cooper (1966), A l l e n (1970), Hamberg (1963) P r o f i t a b i l i t y rate of return f o r the R&D project Mansfield (1969), Disman (1962), Quinn (1966), Kot le r (1967) Peterson (1967), A l l en (1970) expected impact of the R&D project on market share Mansfield (1968), Peterson (1967), Mottley & Newton (1959) expected change in sales a t t r i bu ted to R&D project Peterson (1967) Risk p robab i l i t y of technica l success estimated for the R&D project \ Scherer (1971), Mansfield (1968), Nelson (1959), Gerstenfeld (1971), Disman (1962), Quinn (1956), Ansoff & Stewart (1967), McRlauchlin (1968), Thurston (1971), A l l en (1970), Cooper (1966), T i l l e s (1965), Cranston (1974), Mottley and Newton (1959) - p a t e n t i b i l i t y of innovation P h i l l i p s (1966), Scherer (1971), Ansoff (1965), Mansfield (1968), Quinn (1966), Foster (1971) * ind icates items that appear i n more than one dimension. v Table 3.2 Sample Questionnaire (1-4)  1 General Economic Environment Factors (5) 1  In th is sect ion you are asked to rate the importance of various factors in R&D project evaluation. Please use the rating values from a seven point sca le , where 1 2 3 4 5 6 7 minimum maximum (unimportant) (critical) Enter the ra t ing on the line to the left of each factor. If totally irrelevant enter 0. expected rate of inflation . ' accelerated depreciation of new capital equipment expenditures for tax purposes (29) government funding of feasibility studies for R&D projects (30) average profit rate of firm p o s s i b i l i t y of gaining a new government contract for part of the project (31) patentibility of innovation h i s tory of success with R&D (firm's) (32) expected growth of sales of firm lev: i n te res t rates on government bonds increasing (33) _ expectations with respect to the foreign exchange rate low in teres t government loans for R&D projects (34) _ . pollution control measures (environmental concern) i n te re s t rates on government bonds increasing (35) stage in l i fe cycle of existing products general trends of growth (36) accelerated depreciation of R&D expenditures for tax purposes a v a i l a b i l i t y of sound government information on technological (37) probability of technical success estimated for the R&D project change (38) favourable tax policies for R&D projects a v a i l a b i l i t y of private surveys of market potential (39) _ average profit rate of industry group expected payback period for the R&D project (40) favourable tar i f f policy expected energy requirements H D _ growth of government expenditures expected general productivity changes (42) recent growth of sales of firm a v a i l a b i l i t y of scientif ical ly trained personnel (43) _ stock market trends bar r i e r s to entry in the market (44) unemployment availability of government surveys of market potential (45) _ general trends in inventories interest rate on government bonds declining (46) _ high interest rates on government bonds expected impact of the R&D project on market share (47) _ growth of population government support and promotion for market development (43) _ government subsidies for R&D projects expected growth of real GNP -v . (49) stability of market average profit rate in the economy (50) short term bank interest rates expected wage settlements (51) _ cost of the R&D project relative to total sales of firm expected change 1n sales attributed to R&D project (52) rate of return for the R&D project 115 Table 3.3 Distribution of R&D Laboratories in Population and in Sample Industry Per Cent in Population Per Cent in Sample Agriculture Forestry Mines and Wei 1s Food and Beverages Tobacco Rubber Leather Textiles Wood Furniture Paper Printing Primary Metals Metal Fabricating Machinery Transport Equipment Electrical Products Non-metallic Minerals Petroleum and Coal Chemicals Mi seellaneous Manufacturing Construction Transportation and Communication Services .9 .7 5.8 5.8 .4 5.5 .4 1 .6 .2 3 .2 2.8 5.6 11.5 5 16 2.4 1.9 18 4.5 .4 2.2 4.5 1.5 .8 5.3 6.8 4.5 .8 .8 > 4.5 3.8 5.3 8.3 3.0 13.6 2.3 3 22 3.8 1.5 4.5 3.8 116 that the sectoral breakdown in the sample is representative of R&D estab-lishment frequencies in the population. The questionnaire asked for supplemental information relating to the attributes of the executive and his firm. The distributions of these attributes for the respondents are summarized in Table 3.4. Included in the sample are 15 corporate presidents. Sixty-one per cent of the respondents are currently employed in the R&D department of their firms. The general level of formal education is quite high: two-thirds of the executives have had at least some post-graduate training. The characteristics of the firms are divided into statistical traits and role perception. Table 3.5 displays the statistical characteristics. Half of the sample have sales over $50 million; also half of the sample is either privately owned or controlled by a few interests. The sample is equally split in terms of Canadian or foreign control and represents well: the state of ownership in Canadian manufacturing. For an indication of the perceived role of the firm, respondents were asked to rate on a seven point Likert scale market type (stable/volatile), innovation (follower/innovator) and importance of R&D (no involvement/ extremely important). Table 3.6 presents the distributions of the results. Approximately 50% considered their market somewhat stable while 12% con-sidered their market extremely volatile. Few considered their firm followers. R&D involvement was extremely important for 22%. (4) The Statistical Analysis A number of multivariate techniques were used to analyze the data. Factor analysis was used to determine latent environmental dimensions important E x e c u t i v e C h a r a c t e r i s t i c s POSITION AS GIVEN BY- RESPONDENT RELATIVE . _ jTSyryniTE FKEGGTN'CY-CAVEGORY LABEL CODE FREQUENCY (PERCENT) ""TTuT'TTvT N 0 S &. 8 Pses; of„N_r _ l is i! . A V.P. .CLKTRCLLP.R Z IB 13.& ~CM I H F E N i i r ^ r f i P T s T ~ s T . T _3_I RECTOR 4 23 1 7 . A A S S T . O I R E C T O R 5 2 1.5 f-iuSYuPT ti 51 3T76 S U P cn V I SU_R H c AD 7 3 2.3 G P E R . FLANNEK, ACVIS 8 3 2.3 ' R E S E A K C H A s s n i . 7 H i ; . 9 3 2.T TOTAL 132 100 .0 AGE RELATIVE - ~AB SCLUTE~"'F REQU ENCY -CAIEOJKY LAOEL CCCE FR ECUENCY (PERCENT J NO V"&IV EN 0" 6 ATS . 20-2S___. 1 2 U 5 25-30 2 7 5.3 ""JO-AO i ! 3 33 ^ 2 5.0' _ A C - SO ' i _ A A_2 31.8 50«- 5 A2 31.8 —TCTAI T32 ~TCCTC 0EPT_ PRESENT DEPARTMENT IN YCUR CC«PANJr_ RELATIVE "» B"S"C LU T E~" ~F RE wU F N C Y ~ CATEGORY LABEL CODE FRECLENCY (PERCENT) NUT GIVEN 0 e 6 .1 PRODUCT ICN • _ _ 1_ 3 2.3 SALES, MKTG, ADVERT 2 6 A.5 FINANCE AND ACCTING 3 A 3.0 RESEARCH, DEVELCPMNT 6 e i 61.A GENERAL A0MIN1STRN 7 2 2 16.7 OTHER 8 8 6.1 TOTAL 132 100.0 EOUC FCRfAL ECLCATION, HIGHEST LEVEL CCWPLETEp CATEGORY LABEL CCOE A6SC LLT E" FRECLENCY RELATIVE ~FRE COENCY (PERCENT) NOT GIVEN 0 6 A.S HIGHSCHCCL 1_ A 3.0 CERTIFICATE ' DEG 2 2 1.5 • eACTTE'LCR* s OEGREE 3 3 5 26.5 POST-GKAD TRAINING' A 25 18.9 MASTER'S DEGREE 5 35 26.5 OOCTOR'S CECKEE 6 25 18."3 _ 1 TOTAL 132 ICC.O Table 3.5 F i r m C h a r a c t e r i s t i c s , S t a t i s t i c a l T r a i t s S A L E S A P t - R C X I E A T E S I Z E C F C C E P A N V ' S S A L E S ' C O R E L A T I V E "' AT5"5TCUTE FPFCOHMTr C A T E G O R Y L A B E L CODE F R E Q U E N C Y ( P E R C E N T ) NOT GIVEN 0 ' ~h 3.H J>~5??J? l 13 9._R 1, CL.0-9,999 2 3C 22 .7 TcTu OO-'. 9,9 99 3 25 T1T79 5 0 , U 0 C - 7 4 , 9 9 9 4 8 1 ; 6 . 1 7 5 , 0 0 0 - 9 9 , 9 9 9 5 4 3.0 "Tub, 0 0 0 ^ 2 5 0 , 0 0 0 '• 6~~ 13 9.8 2ii_C;, 0CL+ 7 3 4 . 2 5 . 8 TOTAL 132 100.0 E M P L E f P L C Y E E S I N T O T A L COMPANY RELAT IVE A B T C l U T FT5TC l~E N X T " C A T E G Q R Y L A B E L COOE F R E L L E N C Y ( P E R C E N T ) NQT~GTVEN a 12 sTT __L.l°L.i5l _ l 14 i c . 6 5C TO 99 2 U P.3 iooTlo 2 4 9 — ~ ~ ~ ~ 3 l T ~ 1 C.6 _ 2 5 O . J 0_4 9 9 4 16 12 . 1 500 TO 749 5 e 6.1 75 0~~fcT 999 6 ~ 4 ""ITc 1 0 0 0 TO 1 1 4 9 9 7 8 6.1 1500 TO 2999 8 15 11.4 " 3 0 0 0 TO 4 9 9 9 9 4" 3~."0 _ _ 5 0 0 0 f 10. 26 19 .7 TOTAL 132 100 .0 CWNcD P U B L IC JL*-_POLGH PR_1VATJ C V N F P S H I P RELATIVE . S'B"s"C CUT E FPfEOU EN C Y ~ C A T E G U R Y L A B E L COCE-' F R E Q U E N C Y I P E R C E N T ) "NTJT G l " VEN ' 0 3 << 6~7i P U B L I C — h l D E L Y h E L C 1 3 3 _ . _ 2 5 . 0 P U B L I C — C N T R L BY FEW 2 4 8 3 6 . 4 s ~Pfl'I v T f E L Y ~ OV.NEO 3 4 3 7~12~.b T O T A L 1 3 2 1 C 0 . 0 CONTROL IS C A N A D I A N SHARE ENCUCr FOR C C N T P C L R E L A 1 ! V F ' A f c S C L u T E " F R E L L E N C Y " " " C A T E G O R Y L A B E L C C C E F P E C L E N C Y ( P E R C E N T ) NOT G I V E N 0 S 6 . 8 Y E S I 6 0 4 5 . 5 NO 2 6 3 4 7 . 7 T C T A L ' 1 3 2 1 0 0 . 0 Table 3.6 F i r a Cha rac te r i s t i c s , Perceived Role X A W t ; T Y P E S T A tf L F = 1 VCLAT1LE = 7 FIRM FOLLCV»SR= l IN'NCVATCR = 7 RELATIVE RELATIVE AT3TCITJTF. "FRETCfOENTY A5SCLLTE FREQUENCY CAT EGCRY LABEL CODE FREQUENCY (PERCENT) CATEGORY LABEL COOE FREQUEN'CY (PERCENT) 0 6 4.5 0 4 3 .0 : _ i IL _ 8.3 - - ... _ . J... 1 ..Cl.?. " 2 2 t 19.7 2 4 3.0 3 26 19.7 3 14 10.6 4 27 20 .5 4 22 16 .7 5 20 15.2 5 41 31.1 b 10 7.6 6 25 18 .9 7 6 4 .5 7 21 15.9 TOTAL 132 1C0.0 TOTAL 132 100.C RDIMPORT RSD NG INVCL~VE = 1 EXTRfKE=7 CATEGORY LABEL CODE AB"S"CL"UTF~" FREOLit NCY RELAT IVF ' F K E C U F N C Y -(PERCENT) 0 2 5 4 3.8 3.0 V 3 16 i.12.1 4 5 21 28 15.9 21.2 6 29 22.0 7 TOTAL 29 132 22.0 100.0 120 for R&D decision making and confirm (or reject! the hypothesized dimensions. Discriminant analysis was used to determine differences in "attention" patterns among executives grouped by organizational and individual attributes. Factor Analysis Factor analysis is a technique to uncover independent sources of variation in the data. Meaningful patterns depend on the covariation of the items. Items that vary uniformly cluster together in the data space and provide the basis for the dimensions or factors. Classical factor . analysis techniques seek to explain the common variation of the variables. The basic model is Z. = T a. .F. + d.U. J ; JI 1 J J 1 1 where i is the number of factors, j is the number of variables, F- i s the common factor i , U. is the unique factor j , a., is the loading of variable j on Factor i , and Z. is the standardized value of variable j . Each variable (in standard form) is explained by the common factors and i t s unique factor. A number of factoring methods are available. The method: most commonly used is principal factoring or principal axis factoring. This was the method used here. The principal axes (factors) are the minimum number of orthogonal dimensions required to explain the data. 121 Two major problems with factor analysis stem from some a r b i t r a t i -ness introduced into the communality estimation procedure and the determina-tion of the number of significant factors extracted. The communality of a variable is that portion of the variance that is related to the other variables. The correlation matrix that is being factored has 'ones' on the diagonal, but these refer to the total variance of the variable. If a variable were completely explained by the factor space, the communality would indeed be one. Thus one provides an upper bound on the communality. The lower bound is the squared multiple correlation coefficient of the variable on all the variables (R2) (Rummel, 1970, pp. 317-318). The most, common approach is to replace the ones on the diagonal of the matrix with. R 2s. The factoring of the matrix is conducted iteratively in the following way: after an i n i t i a l factoring, communalities are estimated, these are then inserted in the diagonal, the procedure is repeated until there is l i t t l e difference between the i n i t i a l and estimated communality. This was the procedure followed in this study. The other problem is the determination of the number of factors to be extracted. The aim is to reduce the variable space to a more compact, factor space. The rank of the correlation matrix factored is generally equal to (or one less than) the number of variables. Extracting a l l the-'factors' would not be in harmony with the aim of data reduction. Most factoring techniques extract the factors in descending order of variation explained. Therefore for the unrotated factor matrix, a stopping rule is not necessary. Inspection will determine which factors to include. How-ever, rotated factor loadings are sensitive to the number of factors. While overfactoring leads to less distortion than underfactoring (Rummel,. 122 1970, p. 365), i t is s t i l l best to be parsimonious. A rule is needed to ensure extraction of only significant factors. The most common procedure is to extract those factors with eigenvalues greater than or equal to one. This often results in too many factors, depending on the number of items. The eigenvalue is related to the per cent variation explained by the factor: the per cent variation explained equals the eigenvalue divided by the number of variables (Mulaik, 1972, p. 176). Two alternatives are possible: (1) observe the relationship between the eigenvalue and the number of factors, or (2) postulate the number of factors on the basis of theory. When the relationship between factors and eigenvalue is graphed, discon-tinuities or a scree line emerge. (The scree line is the area where the eigenvalue approaches an asymptote.) Either is a good choice of cutoff for the number of factors (Rummel, 1970). The unrotated factors generally do not give a meaningful pattern-ing of the variables. Factoring techniques extract factors in order of importance. The f i r s t factor tends to be a general factor, loading sig-nificantly on every variable. The remaining factors tend to be bipolar with positive loadings on half the variables and negative on the rest. Rotation serves to simplify the factors making them more interpretable. Two types of rotations are possible: orthogonal and oblique. Orthogonal rotation assumes that the factors are independent. The most common orthogonal rota-tion is varimax which attempts to simplify columns of the factor matrix. A varimax simplified factor has the maximum number of high .and low loadings. The rule is to maximize the variance of the squared loadings in each column (Kim, 1975, p. 485). Inspection of graphs of the orthogonal factors provides a test of the independence of the factors. 123 Oblique rotation assumes that the factors are dependent and often provides a more distinct clustering of the variables. Oblique rotation also provides an account of the actual correlation among the factors. Most oblique rotation methods are computed indirectly on the reference axes (axes perpendicular to the primary axes) (_see Rummel, 1970; Halm, 1973); others calculate directly on the principal axes (direct oblimin, see Kim, 1975, p. 486). In this study direct oblimin is used. The oblique rotation results in two matrices: factor pattern and factor structure. The pattern matrix gives the direct contribution of each factor toward explaining the variance of each variable. The pattern loadings (a-.) are equivalent to the regression coefficients of the variables on each factor. The structure loadings ( r-jj) give the direct and indirect contribution of each factor toward explaining the variance, i.e., the correlation of the variable with the factor and of the factor with' other factors (Rummel, 1970, pp. 397-399). Discriminant Analysis Discriminant analysis is a technique that is useful for deter-mining which variables help explain groupings of individuals. For dis-criminant analysis the groups must be specified a p r i o r i . Linear discrimi-nant functions are derived as predictors of group memberships on the basis of item ratings. The maximum number of functions is one less than the number of groups. Discriminant functions are of the form: 124 where the are standard scores on the p discriminating^variables. Each discriminant function represents a dimension in the reduced discriminant space. Each group is represented by its centroid in this space, thus each group i s located relative to the others. The variables that are most important in distinguishing the groups are determined by a stepwise pro-cedure. Variables are selected to maximize the difference in group centroids. The Wilks Lambda criterion will be used to determine at each step which new variable should enter the function. This accounts for the difference among a l l group centroids and also the cohesiveness within groups. This criterion is equivalent to maximizing the multivariate F ratio that tests the difference among group means. The standardized dis-criminant function coefficients (in absolute value) measure the relative importance of each variable. The variables with high coefficients are then used to define the dimensions as in factor analysis. Canonical cor-relations are calculated for each discriminant function. They are similar to R2 for multivariate regression analysis. The square root,of the canonical correlation is a measure of the per cent of the variance among the groups that is explained by the discriminant function. The discriminant functions are then used to reclassify the respondents. Classification functions are derived for each group using the group centroids and the within group covariance matrix. Individuals are assigned to the group for which they have the highest score. Adjust-ments are made to account for differences in prior group membership prob-a b i l i t i e s . The accuracy of the discriminant functions can be compared using two measures: the probability of correct allocation by chance and the probability of maximum allocation (Morrison, 1969). The probability 125 of maximum a l l o c a t i o n ind icates the per cent of cases that would be c o r -r e c t l y c l a s s i f i e d i f a l l cases were assigned to -the group with the highest a p r i o r i p r o b a b i l i t y of membership. When one group has a r e l a t i v e l y l a rge representat ion th i s becomes a strong te s t . The p r o b a b i l i t y of co r rec t a l l o c a t i o n by chance, which is a more r e a l i s t i c comparison, i s the per cent of cases that would be c o r r e c t l y a l l oca ted to a l l groups on the bas i s o f a p r i o r i group membership p r o b a b i l i t i e s . This i s ca l cu l a ted as c L. " i i where p^ i s the p r o b a b i l i t y of membership in group i (Morrison, 1969; Mos te l le r and Bush, 1954). Indiv idual scores on the d i scr iminant funct ions measure the number of standard dev iat ions from the means. Group centro ids measure the d i s tance among groups on each dimension. The d i scr iminant va lues , when graphed, present a p i c ture of the c l u s t e r i n g of i nd i v idua l s Ju each group. This in conjunct ion with group averages w i l l provide an in s i gh t in to d i f f e rences in a t tent ion s e l e c t i v i t i e s of executives making R&D eva lua t ions . In p a r t i c u l a r the impact upon a t ten t ion patterns of i nd i v idua l a t t r i b u t e s such as education background and pos i t ion,and f i rm a t t r i b u t e s such as s i z e and ownership w i l l be studied. (5) Confirming the A - P r i o r i Dimensions of the " Information Basket" and  Test ing the Relevance of the " Information Basket" Factor ana lys i s was used to confirm whether the a . p r i o r i dimen-sions correspond to empir ica l patterning of v a r i a b l e s . Both general theory 126 and observation of the scree line indicated that six to eight factors could be reasonably extracted. Factor analysis was run therefore with varimax rotation being constrained to respectively six, seven and eight factors. Analysis of this factor range indicated that seven factors yielded the optimal granularity level. Results of the varimax rotation indicated that, some of the items had high loadings on a number of factors defining cor-related factors. Therefore, direct oblimin was performed for a somewhat correlated factor space (delta = -.05). The resulting factors were: (1) general economic conditions, (2) p r o f i t a b i l i t y and sales, (3) government support for R&D, (4) project evaluation, (5) information, (6) taxes, and (7) inducements for R&D. Table 3.7 presents the significant factor loadings with items ordered according to their a priori theoretical clusters (see Glossary of Item Abbreviations at end of the chapter). The relative weight's for each factor were calculated as Y 1?./YY I 2 , where 1.. is the factor pattern I I J JJ TJ i j loading of item i on variable j for the variables loading significantly on each factor. The weights are presented in the bottom line of the table. Analysis of inter-factor correlations revealed the following associations: The tax dimension correlates positively with general economic trends and more weakly with government support for R&D. P r o f i t a b i l i t y and sales correlates positively with project evaluation (both dimensions include R&D project evaluation items). Information correlates negatively with both the p r o f i t a b i l i t y and inducement factors. Government support and information are also somewhat positively correlated, while government support and taxes are somewhat negatively correlated. (The correlations among factors are reported in Table 3.8). 127 Table 3.7 Factor Analys is Results : General Economic Condit ions P r o f i t a b i l i t y and Sales Governnent Support Project Evaluation Information Taxes Induce-ments 1. ECONOMY, GENERAL A. Trends AV PR ECON 511 T BANK I STK MKT TR INVENT TR GEN TRO GR UNEMPLMI .44 .51 .61 .32 .42 ' .57 .47 • --B. Future Expectations EXP GNP GR POP GR .48 .46 C. Cost Chanqes ^Inducements) EX W SETTL FOR EXCH R EXP INFL EXP PRD CH EXP E REQ .51 .44 .46 .44 .48 .56 11. ECONOMY, GOVERN-MENT ROLE y A. B. Costs Direct GOVT SUBS L I GOVT L P GOVT CON FAV TX POL Indirect AC DEPR RD ACCL DEPR •1 I GOVT B .54 - .62 - . 3 8 - .72 - . 3 9 .60 .79 .72 C. Market Influence GOV EXP GR FAV TAR IF POLL CONT .40 - .45 .33 .53 0 . Informat ip_n_J& Support GOV SLIP KK FEAS STL'O GOVT SURV GOV INF Cil - .66 - . 8 0 - . 7 3 - . 6 0 I I I . OTHER INFORMATION PRIV SURV - . 7 0 • IV. MARKET AV PR GRP STABL MKT BAR TO ENT .35 .40 V. HRM A. B. Demand REC SLS GR EXP SLS GR LIFE CYCL Supply, FactofS SCI TR PRS .47 .73 .4? • C. L iqu id i t y AV PR FIRM .59 D. Innovaliveness RO HISTRY .30 VI. PROJECT A. B. CoM-^itnient COST SLS EXP PAY B P r o f i t a b i l i t y ftO ROR RD MKT SUR RD SALE CH .59 .59 \ .56 .87 . .67 C. Riisk P TECH SUC PATENTS .36 .32 .33 RELATIVE WEIGHTS .22 • .20 .18 .13 ' .10 .11 .07 Table 3.8 Cor re la t i on Among Factors General Economic P r o f i t a b i l i t y Government Project T T , ^ , - T^J.^^™* Condit ions & Sales Support Project Information Taxes Inducement General Economic 1.00 .0.14 , -0.16 0.04 -0.24 0.30 0.23 Condit ions & r s l l e s b i 1 i t y 1 - 0 0 ' ° - 2 0 0 , 3 1 " ° - 3 3 ° ' 1 2 Government Support Project , 0 0 Evaluat ion 0.22 1.00 -0.20 0.29 -0.27 -0.09 •0.23 0.14 0.16 Information 1.00 -0.26 -0.32 Taxes 1.00 0.20 V Inducement 1.00 129 The final communality estimates indicate the proportion of the variance of each item that is explained by the factor space. The ordered communality estimates are presented in Table 3.9. The factor space explains about 60% of the variation of the items. A few items load significantly on two factors but analysis reveals that such duality is theoretically warranted. For example the item general trends in growth, loaded on both the factor representing the dimension of general economic conditions and the factor representing the p r o f i t a b i l i t y and sales dimension. Expected inflation (loading on factors 1 and 6) affects the formation of general economic expectations and appears as a tax or reduction of real profits. Favourable tax policies (loading on factors 3 and 6) appears as a tool of government support for R&D and as a reduction in tax. Expected growth of government expenditures (loading on factors 1 and 3) affects formation of expectations and appears as government support for R&D apparently reflecting the assumption that growth of govern-ment expenditures means more contracts and grants for R&D and market support. Two items are not significant in the factor space: average profit of the industry group and avai l a b i l i t y of s c i e n t i f i c a l l y trained personnel. The factor analysis confirmed to a large extent the hypothesized decision dimensions (see Table 3.1). Government items separate into five groups clustering with other items depending on function. Interest on government bonds and growth of government expenditures cluster with the general economic items to form factor 1, general economic conditions. Government surveys and government information cluster with private surveys: to form factor 5, information. Table 3.9 Final Communality Estimates FEAS STUD 0.79 AC DEPR RD 0.74 EXP PAY B 0.73 GOVT SURV 0.72 FAV TX POL 0.70 EXP SLS GR 0.67 GOV SUP MK 0.65 GOVT SUBS 0.61 STK MKT TR 0.61 EXP INFL 0.60 AV PR FIRM 0.48 COST SLS . 0.-48 P GOVT CON 0.48 GOV EXP GR 0.46 REC SLS GR 0.45 FAV TARIF 0.44 RD MKT SHR 0.43 EXP E REQ 0.43 LIFE CYCL 0.43 POP GR 0.42 GEN TRD GR 0.41 GOV INF CH 0.58 SH T BANK I 0.57 BAR TO ENT 0.57 PRIV SURV 0.57 EXP GMP GR 0.55 ACCL DEPR 0.54 STABL MAKT 0.54 I I GOVT B 0.54 UNEMPLMT 0.53 AV PR ECON 0.53 EX W SETTL 0.51 RD SALE CH 0.50 FOR EXCH R 0.50 RD ROR 0.50 EXP PRD CH 0.37 L I GOVT L 0.37 PATENTS 0.36 AV PR GRP 0.35 INVENT TR 0.34 SCI TR PRS 0.33 P TECH SUC 0.33 POLL CONT 0.31 RD HISTRY 0.16 131 Tax related items cluster with inflation to form factor 6, taxes., Pollution control and favourable t a r i f f s cluster with expected produc-t i v i t y change and energy requirements to form factor 7, inducements to R&D. The remaining government items form factor 3, government support of R&D. The market characteristics and project characteristics also s p l i t up. Factor 2, sales and p r o f i t a b i l i t y , includes firm characteristics and market s t a b i l i t y and project related items that reflect change in sales and risk. Barriers to entry (as a risk item) and patentability cluster with project commitment items (relative size and payback) and the rate of return. It is of interest to report the factors that join together as fewer factors are extracted and those that s p l i t as more factors are extracted. With six factors, information and inducements cluster'together, reflecting the correlation of factors 3 and 6. With five factors, the tax factor is eliminated, the items loading on different factors: acceler-ated depreciation of capital joins factor 1, and accelerated depreciation of R&D expenditures and favourable tax policy join factor 3 as government support items. This reflects the correlation of factor 6 with both 1 and 2. With four factors, factors 2 and 4 combine, forming one factor that encompasses a l l project evaluation items, along with firm and market items. Moving toward extracting more than seven factors, interesting patterns can also be observed. With eight factors the general economic factor divides into two groups, with stock market trends, inventory trends,, unemployment, and foreign exchange forming one factor and expected growth of GNP, general trends in growth, average profit of the economy, expected wage settlements and population growth forming another. With nine factors, 132 the project characteristics s p l i t from the p r o f i t a b i l i t y factor leaving two project factors and one firm factor. One project factor relates to sales, the other to p r o f i t a b i l i t y . (This would be another reasonable breaking point; adding the two factors adds 5% of the variation of the items explained by the factor space.) With 10 factors, low interest government loans and increasing interest on government bonds combine to form a separate factor. With eleven factors, R&D history forms a separate factor. General Patterns of Information Preferences Table 3.10 presents the average importance ratings of the items. The most important items are those pertaining to the following attribute dimensions: the project, the firm and the market. Most of the high ranking environmental items were project specific (with expected growth of sales for the firm completing the set of important decision cues). The items receiving the lowest ratings included indicators erf general economic trends and monetary policies. The particular details of item rankings are provided below for each attribute dimension. 1. General Economic The most important of the general economic items is general trends of growth followed by expected general productivity .changes.. Each was given an importance rating of 5 or more by approximately 50%. Expected rate of inflation, expected real GNP growth and average profit rate in the economy are bimodally distributed and are important for some of the respondents. Table 3.10 Ranking of Items by Average Importance Ratings High RD ROR P TECH SUC EXP PAY B RD SALE CH RD MKT SHR EXP SLS GR COST SLS Above Average AV PR FIRM SCI TR PRS BAR TO ENT GEN TRD GR PATENTS FAV TX POL RD HISTRY STABL MKT GOVT SUBS EXP PRD CH LIFE CYCL Average REC SLS GR AC DEPR RD POLL CONT PRIV SURV FEAS STUD AV PR GRP ACCL DEPR EXP E REQ GOV SUP MK EXP GNP GR FAV TARIF EXP INFL Low 5.71 AV PR ECON 2.97 5.64 P GOVT CON 2.94 5.42 EX W SETTL 2.89 5.35 GOV INF CH 2.67 5.05 GOVT SURV 2.56 4.99 L I GOVT L 2.49 4.84 POP GR 2.39 SH T BANK I 2.32 GOV EXP GR 2.08 FOR EXCH R 2.01 4.72 X 4.53 4.42 V^y Low 4!32 INVENT TR 1.91 4.32 UNEMPLMT 1.77 4.32 STK MKT TR 1.72 4.23 I I GOVT B 1.39 4.22 H I GOVT B 1.36 4.17 D I GOVT B 1.27 4.04 L I GOVT B 1.23 4.02 3.87 3.83 3.54 3.43 3.42 3.39 3.39 3.33 3.27 3.13 3.08 3.05 134 2. Government Government subsidies and favourable tax policies for R&D re-ceived high ratings from at least 60% of the subjects. The po s s i b i l i t y of receiving government contracts, which also diminishes the firm's financial commitment to R&D, was bimodally distributed; i t received high importance ratings from 35% and low from 50% of the subjects. Accelerated depreciation of R&D expenditures was also important for 45% of the respondents, while accelerated depreciation of capital equipment has no special importance for R&D decision making. Interest rates on government bonds have l i t t l e importance to R&D decisions, supporting Galbraith's thesis that R&D projects are internally funded or that costs are passed on. Originally i t was hypo-thesized that the effect of government bond interest rates on R&D decision making would vary depending on level and direction of change. For this reason four interest items were included. However given the low importance ratings of a l l and the similarity of their distributions, i t * was apparent that they a l l measure the same thing. Therefore, only one interest item (increasing interest on government bonds) was retained for further analysis. Government funding of f e a s i b i l i t y studies is important to 40% of the sample. Other forms of government information a l l rank low. Pollution control measures are important to 40% of the sample perhaps reflecting low concern with the environmental impact of changing technologies. 135 3. Information Though no information item receives a high importance rating, private information sources are generally more important than governmental. 4. Market At least 50% of the respondents rate st a b i l i t y of market and barriers to entry as important. This probably indicates the requirement of low risk levels,or the comfort of concentration for R&D investment decisions. 5. Firm The expected future is much more important than the past in making R&D decisions except with respect to the R&D history of the firm. Expected growth of sales of the firm is quite important, rated high by 70% of the subjects, while recent growth of sales was rated iiigh by 50%. The firm's history of past success with R&D is given a high rating by 50% of the subjects. The profit rate of the firm is important for 60% of the respon-dents, while outside sources of funds (except government subsidies) are a l l low (again indicating preference for internal funding). Life cycle considerations have high importance ratings or greater than average importance for 5 5 % of the subjects. 136 6. Project The most important project attributes are the probability of success (rated high by 90%) and the rate of return (80%). These are closely followed by expected payback period and change in sales attributed to the project (80% high). Other important project attributes were the expected impact on market share (70% high) and the cost of the project relative to sales (60% high) and patentability (50%). To examine differences in information preferences and relate them to executive attributes and firm characteristics, discriminant function analysis was employed. Using the levels of each attribute and some combinations of attributes, alternative classifications of subjects were defined. The ratings of the information items were used as potential independent variables in the various discriminant functions. The per cent of the respondents correctly classified is a measure of the adequacy of the analysis. Half the groupings yielded accurate predictions for at least 80% of the cases (see Table 3.11). The discriminant analysis performed better than the probability of maximum allocation in all cases. On average the discriminant analysis performed 30 percentage points better than the probability of allocation by chance, indicating a high degree of accuracy. The canonical correlations for each function are also presented in Table 3.11. The important discriminating variables for each classification of subjects are presented in Table 3.12 along with the group centroids. When there are only two groups, the group centroids are separated by at least one standard deviation. Important variables are defined to be those with standardized coefficients of .4 and above. The important Table 3.11 Results of Discriminant Analys is on 44 Economic Items Groupings Var iables Number of Groups De f i n i t i on of Groups Percent Correct Chance Probabil i t y Maximum P robab i l i t y Canonical Co r r e l a t i on * Pos i t ion 3 1. Presidents 2. Other Management 3. S ta f f 94 66 80 .71, .56 Department 2 1 . R A D 2. Others 75 54 65 .55 Age 2 1. <40 2. >40 73 56 67 .52 education 2 1. Bachelor ' s Degree 2. Postgraduate 82 59 71 .56 Location 2 1. Ontario 2. Quebec 84 58 70 .65 Industry 3 1. High RD (> 3% value added) 2. Medium RD~(l-3% value added) 3. Low RD (<1S value added) 87 47 52 .74, .62 Market 2 1. Stable (1-3) 2. V o l a t i l e (4-7) 79 50 50 .57 Fi rm 2 1. Follower (1-5) 2. Loader (6,7) 77 54 64 .54 RD Importance 2 1. L i t t l e involvement (1-4) 2. Important (5-7) 83 56 68 .65 Sales 3 1. < $1 mi 11 ion 2. $1-50 m i l l i o n 3. > $50 m i l l i o n 71 41 46 .70, .40 Employment 3 1. < 100 employees 2. 100-1000 employees 3. > 1000 employees 75 36 44 .72, .53 Owned 3 1. Pub l i c , widely held 2. Rub l i c , control by few 3. Pr i vate 64 35 39 .55, .43 Canadian 3 1. 100% ownership 2. 50-99% ownership 3. < 50% ownership 81 40 51 .68, .60 Control 2 1. Yes 2. No 82 50 51 .62 * In o r d e r o f Impo r t ance when t h e r e 1s more t h a n one d i s c r i m i n a n t f u n c t i o n . T a b l e 3 . 1 2 D i s c r i m i n a n t D i m e n s i o n s , L o c a t i o n o f G r o u p C e n t r o i d s , a n d I m p o r t a n t D i s c r i m i n a t i n g V a r i a b l e s G r o u p i n g V a r i a b l e s Narae(s) o f D i m e n s i o n ( s ) L o c a t i o n o f G r o u p C e n t r o i d O r d e r o f I m p o r t a n t V a r i a b l e s and G r o u p A t t e n t i o n F u n c t i o n 1 F u n c t i o n 2 F u n c t i o n 1 F u n c t i o n 2 P o s i t i o n 1 . G e n e r a l E c o n o m i c C o n d i t i o n s and I n f o r m a t i o n 2 . G o v e r n m e n t S u p p o r t and P r o j e c t E v a l u a t i o n 1 . 2 . 3 . 1 . 5 9 - . 1 0 - 1 . 5 7 - . 8 1 . 2 6 - 1 . 5 5 SH T BANK I GOV INF CH INVENT TR FOR EXCH R I I GOVT B EX W SETTL 3 , 1 , 2 3 , 1 , 2 3 , 2 , 1 3 , 2 , 1 3 , 2 , 1 1 , 3 , 2 GOVT SURV BAR TO ENT RD ROR EXP PAY 3 GOV SUP MK 3 , 2 , 1 3 , 2 . 1 2 , 1 , 3 3 , 2 , 1 3 , 1 , 2 DepartiT-'-nt 1 . G e n e r a l E c o n o m i c C o n d i t i o n s , G o v e r n -ment S u p p o r t a n d S a l e s 1 . 2 . . 4 0 - . 7 5 GOVT SURV EX W SETTL RD MKT SHR FAV TX POL •EXP SLS GR GOV INF CH SH T BANK I 2 , 1 2 , 1 . 1 , 2 2 , 1 2 , 1 1 , 2 2 , 1 A c e 1 . G e n e r a l E c o n o m i c C o n d i t i o n s , T a x , and S a l e s 1 . 2 . . 7 3 - . 3 7 AV PR ECON AV PR FIRM FAV TX POL ' ACCL DPR UNEMPLMT I I GOVT B 2 , 1 1 , 2 2 , 1 1 , 2 2 , 1 1 , 2 E d u c a t i o n 1 . G e n e r a l E c o n o m i c C o n d i t i o n s , S a l e s , P r o j e c t E v a l u a t i o n and I n f o r m a t i o n 1 . 2 . . 8 6 - . 3 6 FEAS STUD SHT BAIIK I COST SLS EXP CNP GR EXP SLS GR 1 , 2 2 , 1 2 , 1 2 , 1 2 , 1 L o c a t i o n 1 . G e n e r a l E c o n o m i c C o n d i t i o n s , I n f o r -m a t i o n and G o v e r n -ment S u p p o r t 1 . 2 . - . 4 3 . 9 9 SH T BNK I GOV INF CH INVENT TR GOVT SURV P GOVT CON EX W SETTL 1 , 2 1 , 2 2 . 1 2 , 1 1 , 2 2 , 1 I n d u s t r y 1 . I n f o r m a t i o n , T a x e s , I n d u c e m e n t a n d A v a i l -a b i l i t y o f R e s o u r c e s 2 . G e n e r a l E c o n o m i c C o n - v d i t i o n s , R i s k , G o v e r n -ment S u p p o r t , I n d u c e -m e n t s a n d A v a i l a b i l i t y o f R e s o u r c e s 1 . 2 . 3 . . 7 4 - . 4 6 - 2 . 2 9 V . 3 2 - . 4 5 2 . 4 4 AC DEPR RD GOV INF CH SCI TR PRS ,. POLL CONT 3 , 2 , 1 3 , 2 , 1 1 , 2 , 3 2 , 1 , 3 SCI TR P R S * PATENTS* INVENT T R * P GOVT CON* EXP E REQ 1 , 2 , 3 3 , 1 , 2 1 , 3 , 2 2 , 3 , 1 2 , 1 , 3 * . ' . CONTINUED i m p o r t a n c e r a t i n g s < . 4 . Table 3.12 (Continued) Grouping Variables Name(s) of Dimension(s) Location of Group Centroid Order of Important Variables and Group Attention Market Function 1 Function 2 Function 1 Function 2 1. General Economic Conditions, Govern-ment Support and Project Evaluation 1. .57 2. -.57 I I GOVT B 1,2 FOR EXCH R 2,1 FAV TX POL 2,1 FEAS STUD 1,2 RD ROR 1,2 INVENT TR 1,2 EXP PAY B 1 ,2 r i rm 1. General Economic Con-ditions, Sales and Government Support 1. -.41 2. .72 P TECH SUC 2,1 EXP SLS GR 2,1 GOV SUP MK 2,1 L I GOVT L 1,2 EXP GNP GR ?.l RD Importance 1. Government Support and Sales • 1. .94 2. -.45 GOV SUP MK 2,1 RD HISTRY 2,1 Sales 1. General Economic Con-dition and Sales 2. General Economic Con-ditions, Sales and Government Support 1. -1.51 2. -.35 3. .66 -.81 .41 -.20 SH T BANK I 2,1,3 EXP SLS GR 1,2,3 ' I I GOVT B 3,2,1 LIFE CYCL 1,2,3 SH T BANK I 2,1,3 FOR EXCH R 1,3,2 P GOVT CON 1,2,3 LIFE CYCL 1,3,2 Empl o y T i e n t 1. General Economic Con-ditions, Inducement, and Seles 2. Tax, Government Support, Sales and Information 1. -1.15 2. -.22 3. .72 .58 -.70 • .28 SH T BANK I 1,2,3 INVENT TR 3,2,1 POLL C0NT 3,2,1 .POP GR 3,2,1 REC SLS GR 1,2,3 PRIV SURV 2,3,1 AC DEPR RD 3,1,2 FAV TX POL 2,1 ,3 L I GOVT L 1=2,3 REC SLS GR. 1,2,3 P GOVT CON 1,3,2 Owned 1. General Economic Condi tions 2. General Economic Con-ditions, Sales and Inducements 1. -.08 2. -.58 3. .71 -.70 ,30 .21 SH T BANK I 3,1,2 INVENT TR 2,1,3 STK MK TR 1,3,2 FOR EXCH R 3,1,2 UNEMPLMT 1,3,2 UNEMPLMT 1,3,2 FAV TARIF 2,3,1 I I GOVT B 3,2,1 LIFE CYCL 1,3,2 Canadian 1. General Economic Con-ditions, Governments Support and Sales 2. General Economic Con-V ditions, Sales, Project. Evaluation, Government Support and Tax 1. -.97 . 2. .32 3. .52 V. -.09 1.32 -.38 FAV TX POL 1,3,2 I I GOVT B 1,3,2 COST SLS 1,3,2 " EXP GNP GR 1,3,2 GOV SUP MK 1,3,2 BAR TO ENT 1,3,2 FAV TX POL 1,3,2 . AC DEPR RD 3,1 ,2 EX GNP GR 1,3,2 RD HISTRY 3,1,2 Control :' 1. General Economic Con-ditions, Government Support and Tax 1. -.63 2. .60 FAV TX POL 1,2 EX GNP GR 2,1 SH T BANK I 1,2 I I GOVT B 2,1 AC DEPR RD 1,2 •• W3 140 variables are helpful in naming the discriminant dimensions. Here the dimensions are defined by the factors associated :with the important variables. In the case of two functions, some variables are important for both, varimax rotation would be a useful tool to give better separa-tion of the dimensions but this option was not available-Determination of the group that gives highest rating to the items helps in interpreting the group distinctions. The discussion of results that follows highlights the relationship of differences in infor-mation preferences to: (1) decision makers' attributes, and (2) organiza-tional characteristics and individual attribute interactions. Executive Attributes and Differences in Information Preferences Generally, a preference for a broader basket of information for R&D project evaluation, was indicated by the following overlapping groups: (1) t h e young ( 2 ) t h e e x e c u t i v e s w i t h l o n g e r t i m e o f s c h o o l i n g , and ( 3 ) e x e c u t i v e s p e r c e i v i n g t h e i r f i r m s as " l e a d e r " r a t h e r than " f o l l o w e r " Presidents focused relatively more than other executives upon evaluation of general growth trends and wage settlements, i.e. general expectations of investment climates. Senior managers seemed to focus upon information related to expected returns on specific proposals and av a i l a b i l i t y of government support. Other executives (staff) were less discriminating indicating a broader set of variables as important to 141 R&D decision making. This result confirms the expectation that top level executives concern themselves with strategic problems, while the next echelon of senior management focuses upon tactical problems. Analysis of the departmental membership of subjects yields support to the observation that information selection reflects parochial tendencies. R&D managers in comparison to others, focus more upon infor-mation about technical change while executives in other departments dis-play higher interests in general market information. The exception to this rule is a higher interest of R&D managers in assessment of the impact of R&D on market share as opposed to general investment climates and contribution to overall company growth. Executives based in Quebec tend more than Ontario executives to value information concerning inventory trends, government support for market development and expected wage settlements, while the latter pay more attention to conditions of short-term financing and direct government demand for R&D services. These distinctions were judged to .be a function of different industrial structures in the two provinces. Differences in Information Preferences and Organizational Attributes The analysis of discriminating information preferences among subject classifications by organizational attributes, f i r s t focused upon the level of commitment in the firm to R&D. Firms with high R&D commit-ment displayed a relatively low interest in information about inducements, and incentives for R&D. This is perhaps a reflection of the institu-tionalization of R&D investments as a standard operating procedure in 142 the perceived general market role of these organizations. Low commitment to R&D was associated with higher concern with tax incentives and patent-a b i l i t y of developments. Medium R&D involvement was associated with preference for information about derived demands for R&D. For example, the probability of government contracts and expected energy requirements received higher ratings by executives in firms with medium R&D involvement. Executives in firms with highly stable markets showed higher preferences than other executives, for information on general economic trends, reflecting a longer planning horizon. Those in volatile markets tended to have higher preferences for short-term information, such as, foreign exchange rates and existing tax incentives. Executives in firms which they perceived to be leaders, tend to pay higher attention to general economic conditions than those per-ceiving their firms to be followers. The latter showed more interest in positive inducements for R&D involvement. Sales size effects upon information preferences of executives seemed to confirm Galbraith's claim that large firms are concerned with government bond interest as a measure of general economic conditions. Executives from medium and small firms, in contrast, paid higher attention to bank interest rates which affect their costs. Discriminant functions derived to predict classification of executives by the employment size of their firms indicated the following preferences: Executives from smaller firms were more attentive to infor-mation items concerned with liquidity and direct support for R&D. The distance in group centroids between executives from medium.and large firms was less than the distance between either of them and executives 143 from small firms. This may reflect a size threshold phonomenon in organi-zational decision procedures. Ownership and firm control have several interesting associations with information behaviour of executives. Executives from firms which are publically owned and the shares of which are widely held tended to pay more attention than other executives to stock market trends, unemploy-ment levels and product l i f e cycle characteristics. Those from privately owned firms showed higher concern than others with interest rates and foreign exchange rates. These differences may reflect the alternative potential financial sources perceived by these groups for R&D investment funding as well as differences in performance evaluation c r i t e r i a . The three groups of executives classified by the degree of Canadian control in their firms were equally distant in the discriminant space. Both dimensions included general economic conditions, government support and sales. The second dimension included also project evaluation and tax. The f i r s t dimension items were a l l rated as highly important by group 1 (100% Canadian ownership) followed by group 3 (less than 50%. ownership). This dimension separates group 1 from the other groups. The second dimension separates group 2 from groups 1 and 3. Group 3 was most concerned with accelerated depreciation of R&D and R&D history; while group 1 rated higher the remaining items in the second dimension. Where Canadian ownership was sufficient for control, the execu-tives were concerned with favourable tax policies, short-term bank interest and accelerated depreciation of R&D. In firms where control is not held by Canadian interests, executives were concerned with general economic conditions such as expected growth of GNP and the interest on government bonds. 144 Firm - Executive Attribute Interactions To identify possible impacts of interactions of individual and organizational attributes upon information behaviour, individual a t t r i b u t e s — position and department, were cross-classified with the firm attributes — industry, market, firm R&D importance, sales, employment, ownership and Canadian control to identify new groupings of subjects. To keep the number of groups manageable, only two position groups were defined: presidents and other management. Table 3.13 summarizes the results of the analysis. The discriminant function always performed better than both the probability of chance allocation and the probability of maximum allocation. The highlights of the analysis will be described for the two major discriminating dimensions of the groupings. 1. Position and Location Dimension 1 included government bonds, government information, favourable tax policy, inventory trends, and short-term bank interest. Dimension 2 included expected wage settlements, sales change attributed to the R&D project, pollution control measures and unemployment. Presidents were similar on dimension one which explained 61% of the variation. Presidents of Ontario firms were similar to managers from Quebec firms on dimension two explaining 22% of the variation. 2. Pos i t i on and Industry Dimension 1 included government contracts, government support of markets, government f e a s i b i l i t y studies, average profit of the firm, 145 Table 3.13 Results of Discr iminant Ana lys i s fo r Cross C l a s s i f i e d Groups on 44 Economic Items Number Per Cent P r o b a b i l i t y P r o b a b i l i t y Grouping Var iab les Per Cent of Maximum K A l l o c a t i o n A l l o c a t i o n Pos i t i on and Locat ion 4 88 44 60 Industry 6 84 37 44 Market 4 79 39 47 Firm 4 79 41 57 R&D Importance 4 81 43 56 Sales 6 78 33 •»•. 43 Employment 6 77 31 40 Ownership 6 67 25 34 Control 4 73 39 50 Department and Locat ion 4 79 32 47 Industry 6 77 26 '33 Market 4 61 28 38 Firm 4 66 30 45 R&D Importance . 4 69 31 42 Sales 6 67 . 25 37 Employment 6 71 22 34 Ownership 6 58 19 28 Control 4 72 28 38 Locat ion and Industry 6 72 27 37 Market 4 82 29 38 Firm 4 67 32 47 R&D Importance 4 79 33 47 Sales 6 63 24 33 Ownership 6 67 20 27 Control 4 79 29 37 146 foreign exchange rates, pollution control, inventory trends, population growth and short-term bank interest. Dimension.2 included accelerated depreciation and government contracts. Presidents in firms with medium and low R&D commitment and managers in firms of high and medium R&D commitment were similar on dimen-sion one (explaining 35% of the variation). Presidents and managers of medium commitment firms were similar on dimension two (30% of the variation). 3. Position and Market Dimension 1 included government bonds, government information, pollution control and probability of technical success of the project. Dimension 2 included government bonds, sales change attributed to the R&D project, average profit of the firm, foreign exchange rate, inventory trends, and short-term bank interest. Dimension one separated presidents of volatile markets from other presidents and a l l the managers (explaining 51% of the variation). On dimension two presidents and managers in volatile markets were similar. 4. Position and Sales Dimension 1 included government bonds, inventory trends and short-term bank interest. Dimension 2 included general trends in growth, expected wage settlements, expected sales growth, average profit of. the industry group, short-term bank interest and the relative size of the project. 147 For the largest firms, position was unimportant. For the smallest firms the presidents and other managers were far apart in the discriminant space. On dimension one management in small and medium sized firms were close. For the medium size firms, the distance between presidents and management was moderate. 5. Position and Canadian Control Dimension 1 included government bonds, private surveys, expected wage settlements, and short-term bank interest. Dimension 2 included government information, government support of markets, average profit of the firm, inventory trends, population growth, short-term bank interest, and the rate of return of the project. Managers were similar whether the firm was Canadian controlled or not. However, presidents of each group differed. For Canadian con-trolled firms, presidents and management were similar on dimension two (explaining 29% of the variation). While in the non-Canadian group, presidents were similar to managers on dimension one (explaining 59% of the variation). 6. Department and Market Dimension 1 included expected productivity change, change in market share attributed to the R&D project, average profit-in the economy, expected wage settlements, patents, favourable tax policies, and average profit of the group. Dimension 2 included accelerated depreciation, change in market share attributed to the R&D project, expected wage 148 settlements, expected and recent sales growth, and rate of return of the R&D project. .; On dimension one R&D managers were similar whether the markets were volatile or stable (explaining 46% of the variation). However on dimension two, R&D managers in stable markets were different from the cluster of a l l the other groups (explaining 36% of the variation). 7. Department and Employment Dimension 1 included expected sales growth, pollution control, favourable tax policies, inventory trends, and short-term bank interest. Dimension 2 included government loans, general trends in growth, change in market share attributed to the R&D project, average profit of the group, and growth of government expenditures. Employment size of the firm was more important than department in determining similarity of items attended. R&D and other executives of large firms were similar on both dimensions. On dimension one a cluster formed of both executive groups, in medium size firms and another cluster of executives in small firms (explaining 41% of the variation). 8. Location and Sales Dimension 1 included accelerated depreciation, government con-tracts, expected sales growth, and population growth. Dimension 2 .. includes government information, ava i l a b i l i t y of s c i e n t i f i c a l l y trained personnel, change in market share attributed to the R&D project, expected wage settlements and expected sales growth. 149 Sales size was the basis of clusters on the f i r s t dimension rather than location (explaining 38% of the variation).. On the second dimension, Ontario firms of a l l sizes and large Quebec firms formed another similar group (26% of the variation). Policy Implications The study has identified significant differences in informa-tion selection patterns among executives. These were associated to differences in executive and firm attributes. The study suggests that strategies aimed at improvements in the specific attributes of investment opportunities will be universally attended to. In contrast measures aimed at improvements of specific climate attributes or measures which provide specific inducements for R&D stimulation will have a highly selective impact. "Social marketing" strategies to stimulate R&D must provide a f i t in content to"'prevailing information selection patterns of executives and organizations. Similarity groupings such as those identified by this study, will constitute the appropriate target populations for specific strategic designs. Clearly i t is also necessary to ensure that other characteristics of the information diffusion process provide a f i t with search and evaluation procedures in firms (e.g. f i t in media type, form of messages, etc.) and that barriers to actions are removed. Further studies to provide this information are necessary for improvement in the impact of intervention upon R&D investment. 150 GLOSSARY OF ITEM ABBREVIATIONS ACCL DEPR accelerated depreciation of new capital equipment expendi-tures for tax purposes AC DEPR RD accelerated depreciation of R&D expenditures for tax purposes AV PR ECON average profit rate in the economy AV PR FIRM average profit rate of firm AV PR GRP average profit rate of industry group BAR TO ENT barriers to entry in the market COST SLS cost of the R&D project relative to total sales of firm D I GOVT B interest rate on government bonds declining EXP E REQ expected energy requirements EXP GNP GR expected growth of real GNP EXP INFL expected rate of inflation EXP PAY B expected payback period for the R&D project EXP PRD CH expected general productivity changes EXP SLS GR expected growth of sales of firm EX W SETTL expected wage settlements FAV TARIF favourable t a r i f f policy FAV TX POL favourable tax policies for R&D projects FEAS STUD government funding of f e a s i b i l i t y studies for R&D projects 151 FOR EXCH R expectations with respect to the foreign exchange rate GEN TRD GR general trends of growth GOV EXP GR growth of government expenditures GOV INF CH avail a b i l i t y of sound government information on techno-logical change GOV SUP MK government support and promotion for market development GOVT SUBS government subsidies for R&D projects GOVT SURV availability of government surveys of market potential H I GOVT B high interest rates on government bonds J I I GOVT B interest rates on government bonds increasing INVENT TR general trends in inventories LIFE CYCL stage in l i f e cycle of existing products L I GOVT B low interest rates on government bonds L I GOVT L low interest government loans for R&D projects' PATENTS patentability of innovation P GOVT CON possibility of gaining a new government contract for part of the project POLL CONT pollution control measures (environmental concern) POP GR growth of population PRIV SURV ava i l a b i l i t y of private surveys of market potential -P TECH SUC probability of technical success estimated for the R&D project REC SLS GR recent growth of sales of firm RD HISTRY history of success with R&D (firm's) RD MKT SHR expected impact of the R&D projection market share RD ROR rate of return for the R&D project RD SALE CH expected change in sales attributed to R&D project SCI TR PRS avail a b i l i t y of s c i e n t i f i c a l l y trained personnel SH T BANK I short-term bank interest rates STABL MKT st a b i l i t y of market STK MKT TR stock market trends UNEMPLMT unemployment Chapter 4 MULTI-ATTRIBUTE INVESTMENT D E C I S I O N S : A STUDY OF R&D PROJECT SELECTION Introduction In Chapter 3 the relationships between a variety of environ-mental variables and project attributes and R&D investment decisions were examined. The chapter concluded that managers tend to focus upon project specific information in their evaluation of R&D investment oppor-tunities rather than search for general information about their economic environments. This chapter attempts to provide additional information about the role project attributes play in investment selections. In particular the paper focuses upon imputed trade-offs among pfoject attributes, and investigates possible correspondences between attention and judgment patterns and the characteristics of decision makers and their organizations. The attributes upon which we focus are: C l ) c o s t o f t h e p r o j e c t r e l a t i v e t o t o t a l R&D b u d g e t , ( 2 ) t h e p a y b a c k p e r i o d , ( 3 ) t h e p r o b a b i l i t y o f t e c h n i c a l a n d c o m m e r c i a l s u c c e s s , (4) m a r k e t s h a r e i m p a c t , 153 154 (.5) e x p e c t e d r a t e o f r e t u r n , a n d (.6) a v a i l a b i l i t y o f g o v e r n m e n t p a r t i a l f u n d i n g f o r t h e p r o j e c t . \-Cost of the project relative to the total R&D budget of the firm is a measure of resource commitment. Economic theory would suggest that the cost by i t s e l f would not be important (a measure of p r o f i t a b i l i t y should be considered). However, Mansfield (1968, p. 310) found that the probability that a firm would fund a project was negatively correlated with the size of the investment required. Concern with financial commit-ment is such that, especially for small companies, potentially profitable projects may be abandoned before they have had a real chance to succeed (Cooper, 1966, p. 175). For other references on cost, see Scherer (1970),. Gerstenfeld (1971), T i l l e s (1966), Ansoff and Stewart (1967), Mottley and Newton (1959) and Allen (1970). The payback period is a measure of the time commitment to a project. Payback period norms reflect the subjective time discount and time horizon of the firm. It is also a risk measure in that the longer the time commitment, the less certain the p r o f i t a b i l i t y and other estimates. The payback period for R&D projects is generally required to be shorter than that for investment in plant and equipment. For a l l manufacturing in the U.S. in 1961, 55% of the projects undertaken had an expected payback of less than 3 years and an additional 34% f e l l in the 3-5 year range (Mansfield, 1968, p. 15). Gerstenfeld (1971, p. 22) found that the payback period varied with size of firm; the average was 4.26 for large firms and 3.5 for small. The high proportion of industrial R&D devoted to development and applied research is indicative of this required short payback period (Leonard, 1971, p. 236; Bright, 1970, p. 6). A 155 maximum payback period may also appear as a constraint imposed by manage-ment, thus i t may be the deciding factor in project selection (Kotler, 1967, p. 30). For other references, see Ansoff and Stewart (.1967), Brooks (1972), Tilles (1966), Cooper (19.66), Allen. (1970) and Hamberg (.1963). The probability of technical and commercial success is a measure of risk. It is useful to modify profitability estimates by an estimate of the probability of success. .As the risk increases, the expected value of the re-turn and thus the maximum expenditure justified decreases (Disman, 1962, p. 88). The bulk of R&D is relatively safe (non-risky) and aimed at small improve-ments in the state of the art. Mansfield (1968, p. 56) found that the ex ante probability of technical success for projects undertaken averaged 80%. It seems that firms generally do not initiate a project until major technical uncertainties are eliminated. Gerstenfeld (1971, p. 22) found a similarly high average of 71%. As basic research projects are more risky than applied, a risk avoiding firm will fund more applied projects (Nelson, 1959, p. 304). See also Scherer (1970), Gerstenfeld (1971), Quinn (1956), Ansoff and Stewart (1967), McGlauchlin (1968), Thurston (1971), Cooper (1966), Tilles (1966), Cranston (1974), Mottley and Newton (1959) and Allen (1970). Market share is often a subsidiary goal of firms. From an economic point of view, it is a reflection of the competitive power of the company and of market security. Mottley and Newton (1959) propose market gain as an auxiliary variable in a scoring model for project selection. See also Peterson (1967). Expected rate of return (R0R) is a measure of profitability in certain environments. Projects can be ranked by the R0R and selected 156 in descending order until the R&D budget is exhausted. Alternatively a minimum acceptable ROR can be imposed as a constraint. The ROR has the advantage of incorporating both costs and revenues. For references, see Mansfield (1968), Disman (1962), Quinn (1966), Kotler (1967), Peterson. (1967) and Allen (1970). Availability of government funding reduces a firm's commit-ment to R&D. In Canada as in the U.S., the government directly supplies approximately 60% of the funds for R&D (Brooks, 1972). Government con-tracts influence both the type of research and the atmosphere in which research act i v i t i e s are undertaken (Quinn, 1966). Government support for R&D has tended to be concentrated in defense and areas that bring national prestige (e.g., space). This may result in a misallocation of funds (Leonard, 1971; Brooks, 1972) by limiting the technical resources available for other pursuits (and increasing their costs). On the other hand, decreases in government funding for R&D have resulted in even greater concern for short-term payoff resulting in the undertaking of few risky and basic research projects (Brooks, 1971; Foster, 1971). Methodology To identify latent structures of decision processes s t a t i s t i c a l modelling methods based upon the "actuarial" approach are employed. In this approach, from repeated investment decisions (in experimental or real settings), models are estimated to relate functionally project selections and the underlying project attributes. Studies using this approach have investigated decision behaviour in diverse areas: c l i n i c a l 157 psychology (Goldberg, 1971), graduate student admissions (Dawes, 1971), student performance (Einhorn, 1971), stock selection (Slovic, 1969), judicial decisions (Kort, 1968), tenure evaluation (Green and Carmone, 1974), and physician decision making (Schwartz et al. , 1976). Different rules were postulated to describe the process by which attributes are combined to yield a judgment. A majority of the studies, however, con-cluded that the linear model provides an excellent paramorphic represen-tation of decision makers in many situations (Dawes, 1971). Goldberg (1971), for example, concludes that i f one's purpose is to reproduce the responses of most judges, then a simple linear model will normally permit the reproduction of 90-100% of their reliable judgment variance. As our prime objective is to develop a "black box" model to predict rather than explain the selection process we focus, in our analysis, mainly on simple linear or quasi-linear models. (For a discussion of black box models in contrast to process-explanations, see Green, 1968). There have been some criticisms of this method when used for prediction of judgments and determination of causal relationships. Green and Carmone (1974) pointed out that the process of modelling might be influenced by ut i l i z a t i o n of the rating scale when subjects could supply reliably only ordinal ratings. In addition, i f rating data i s ordinal, some configural models (e.g. Einhorn's conjunctive model) cannot be differentiated from the data f i t of the additive model. This is the case when the configural models constitute in fact order preserving transfor-mations of the linear model (e.g. in Einhorn 1s conjunctive model a log-arithmic transformation of data is used to produce a linear model). Birnbaum (1973, 1974) claimed that using correlations of theoretical 158 predictions and data as indicators of "correctness" might mislead. He claimed that such correlations could be higher for incorrect models than for correct ones. He suggested that functional measurement provided a sounder basis for model evaluation by placing scaling in the context of model f i t t i n g and by testing deviations from predictions rather than concentrating upon overall goodness of f i t . To cope with some of the d i f f i c u l t i e s of employing general linear regression analysis to estimate "actuarial" decision models, two alternative frameworks were utilized to derive linear models. The f i r s t modelling route uti l i z e s orthodox regression models with the addition of a test procedure for screening against models based upon ordinal per-ceptions of attribute scales. This procedure f i r s t suggested by Dawes and Corrigan (1974) and further developed by Einhorn and Hogarth (1975) call s for repeating the regression analysis with equal weights. If R2 values obtained by means of unequal weights are meaningfully higher than R2 based on equal weights, then trade-offs can be imputed from the regression coefficients. Otherwise, no trade-off inferences are possible on this basis. The second modelling route permits more f l e x i b i l i t y with respect to scaling requirements of the dependent variable. This route employs discriminant analysis to identify those attributes (and their corresponding weights) which explain judgments classified into groups according to subject preferences. The employment of these two competing modelling alternatives will permit a further test of validity, the test of robustness. 159 The Experiment Subjects were presented with a hypothetical economic environ-ment in which they were instructed to make R&D investment decisions. The projects were described by six attributes: relative cost, payback, probability of success, per cent increase in market share, rate of return and per cent of government funding. Subjects were requested to indicate the probability that they would recommend funding of each project. Table 4.1 presents the description of the economic environment and the instructions. Table 4.2 exhibits a sample profile. For the f i r s t ten profiles the attributes were presented in random order to avoid attribute ordering effects. For the remaining profiles the attribute order was standardized. The same sample of subjects participating in the information preference and attention patterns study (Chapter 3) have participated in this experiment. The response rate for this experiment w"as 30%. The subjects were then presented with sixty project profiles. Table 4.2 Sample Project Profile % gov't $ payback period % mkt prob. success ROR cost R&D 52% 2 yrs 4 mo 26% 57% 34% 11% probability of funding = % Table 1.1 Sample Questionnaire 160 II. Ufjl) D e c i s i o n M:ik i IIK : I ' re ject l iva lua t ion In t h i s s e c t i o n you w i l l lie a s k e d to e v a l u a t e K5D p r o j e c t s on the b a s i s o f s e v e r a l key a t t r i b u t e s w i t h i n the c o n t e x t o f a hypothe t Lea l economic env i r onment . lxonomic. S v n o p s i s : Tako a h y p o t h e t i c a l s i t u a t i o n i n which the economic environment i s d e s c r i b e d as f o l l o w s : hast year the Canad ian economy had been o p e r a t i n g in an i n t e r n a t i o n a l env i ronment o f s t r o n g demand, s t r a i n e d p r o d u c t i v e c a p a c i t y , h i gh i n f l a t i o n and r i s i n g i n t e r e s t ' r a t e s . In the c u r r e n t y e a r , a d e c e l e r a t i o n in growth appears t o bo under i ." way in a number o f c o u n t r i e s , w i t h a r e t u r n to lower r a t e s o f expans ion and i n some cases r e a l c o n t r a c t i o n in growth o f o u t p u t : The r a t e o f growth i n the Canad ian e c o -nomy peaked a yea r or two ago at a rea l , r a t e o f 65 and i s c u r r e n t l y s l o w i n g down a n d , i n the o p i n i o n o f exper t f o r e c a s t e r s , may even approach ze ro . Unemployment r a t e s which h i t a low o f S . l » are now b e g i n n i n g t o . i n c r e a s e . I n t e r e s t r a t e s on government bonds reached an a i l time h i gh o f 10 1/2% but are now at 6 1/25. Banks and f i n a n c i a l i n s t i t u t i o n s r e p o r t very u n s t a b l e g e n e r a l f i n a n c i a l p o s i t i o n s where they have to borrow s h o r t term funds at h i gh r a t e s o f i n t e r e s t and t h e r e f o r e are r e l u c t a n t to commit themselves long term. The Consumer P r i c e Index f o r the l a s t twelve months shows i n f l a t i o n runn ing at an annua l r a t e o f 13%. S tock markets u n i v e r s a l l y show marked weakness. The va lue o f the C a n a d i a n d o l l a r i s e x p e c t e d to s t a b i l i s e a t p a r wi th the U.S. $. Most deve loped c o u n t r i e s are p u r s u i n g m i l d l y expans ionary p o l i c i e s . O p e r a t i n g i n t h i s env i ronment you are p r e s e n t e d w i t h a p o r t f o l i o o f RSD p r o j -e c t s p r o v i d e d by a c o n s u l t i n g group o f t e c h n i c a l e x p e r t s . Eacii p r o j e c t i s summar-i z e d by s i x key a t t r i b u t e s as f o l l o w s . [ T i t l e s i n b r a c k e t s arc those used i n the^. p r o j e c t p o r t f o l i o which f o l l o w s ] : 1. e x p e c t e d RyC cos t o f the p r o j e c t measured as a pe rcen tage o f t o t a l R5U budget of the f i r m (where c o s t i s t o t a l o u t l a y t i l l commerc ia l phase) 2. e x p e c t e d payback p e r i o d f o r the p r o j e c t [payback p e r i o d ] 3. p r o b a b i l i t y o f succe s s , i n c l u d e s t e c h n i c a l and commerc ia l sue- , , , r- c A - *i r . • i [prob. s u c c e s s ] cess ( i f there i s no s u c c e s s , the c o s t o f the p r o j e c t i s l o s t ) ' 1 4. e x p e c t e d i n c r e a s e in market s h a r e measured as the p e r c e n t change i n c u r r e n t market sha re I. e x p e c t e d r a t e o f r e t u r n , i . e . the r a t e tha t d i s c o u n t s the net cash f low to equa l the i n i t i a l o u t l a y on an a f t e r tax ba s i s 6. p r o p o r t i o n o f funds a v a i l a b l e from th< government f o r t h i s p r o j e c t ( the o the r p r o j e c t a t t r i b u t e s are independent o f [t g o v ' t $] these funds) Assume a l l o the r a t t r i b u t e s ( e g . a v a i l a b i l i t y o f l abour ) arc- equal f o r a l l project;-, For e a c h - p r o j e c t you are a sked to i n d i c a t e the chance that you would recommend approva l o f the p r o j e c t f i o M a f i r m w i t h cl:a rac t o r j st i cs Cnarket , s i z e , e t c . ) s im -i l a r to y o u i s . * * l o r each p r o j e c t , e n t e r a number between 0 and 1OO, where )()() menus the project, would de Ti II i le 1 y be app roved , aa;! ') means that tho project, wou ld dv f i :\i to I y not be approved. Note: In the p r o f i l e s that f o l l o w the p r o j e c t a t t r i b u t e s appear in random o r d e r . v " " : i : 1 < d t o ;;ive your judgment o f each p r o j e c t , there .-ire n o correct -co s t L R5D J [$< n k t . ] [ROR] 161 The questionnaire asked for supplemental information relating to the attributes of the executive and this firm. Included in the sample are 9 corporate presidents. Fifty-six per cent of the respondents are currently employed in the R&D department of their firms. The general level of formal education is quite high: more than 50% of the executives have had at least some post-graduate training. The characteristics of the firms are classified into two major information dimensions: statis t i c a l traits (e.g. size) and role percep-tion (e.g. leadership). Half of the organizations represented in the sample have sales over $50 million; seventy per cent of the firms represented in the sample are either privately owned or controlled by a few interests. The sample is almost equally s p l i t in terms of Canadian control and represents well the state of ownership in Canadian manufacturing. Approxi-mately 50% of the subjects considered their markets somewhat stable and 12% considered their market extremely volatile. Ten per cent considered their firm followers, while 35% were leaders. R&D involvement was extremely important for 40% of the firms. The Results Regression Models of Individual Decision Makers Stepwise linear regression analysis was performed to estimate the decision model for each executive. Table 4.3 presents the coefficients of the significant attributes. Eighty-three per cent of the regressions had R2 greater than .80. The Einhorn test was performed and the original regression equations provided a better f i t in a l l cases (see Table 4.3), 162. T a b l e 4 . 3 R e g r e s s i o n C o e f f i c i e n t s f o r S i g n i f i c a n t V a r i a b l e s E x e c u t i v e I . D . ft COST PAYB PSUC MKT ROR ^TTT. : GOVT R 2 1 - . 2 4 - . 1 0 1.27 . 8 4 . 3 7 3 - . 5 0 - . 2 7 .76 . 3 5 . 8 8 . 3 9 5 - . 4 4 .45 .44 .61 . . 3 1 6 - . 4 8 - . 6 1 1 .05 . 5 9 . 8 2 . 5 0 12 - . 1 8 . 6 5 .27 . 2 2 . 9 7 .51 13 - . 6 2 .47 1 .18 . 8 3 . 3 3 14 - . 5 6 - . 4 2 . 7 0 . 6 3 .81 • . 4 0 19 - . 2 3 . 4 8 . 4 9 . 7 7 . 3 8 20 - . 1 8 . 2 8 .51 . . 4 2 . 8 3 . 1 9 26 - . 7 4 .65 \ .79 . . 3 4 43 - . 5 4 . 5 0 1.18 . 8 3 . 2 4 45 1 . 2 0 . 6 6 . 3 7 47 1 .49 . 9 0 .31 49 - 1 . 5 5 . 2 6 .72 .31 .57 . 8 8 • ' ' . 14 68 - . 2 4 . 4 6 1 .32 . 8 9 . 5 7 74A .15 1 . 0 0 . 8 4 . 2 9 85 - . 2 2 . 9 8 .19 . 9 6 . 3 5 90 - . 5 3 . 7 3 . 4 2 . 8 9 . 4 7 90B - . 4 9 - . 5 1 .86 .61 . 2 4 . 9 2 . 7 6 90G. - . 5 0 - . 3 6 .75 . 4 5 . 8 3 . 5 2 93 - . 5 7 . 6 0 . 9 6 . 8 8 . 3 9 93A - . 5 8 .62 . 9 6 . 8 9 . 4 3 94 - . 3 1 - . 2 3 . 2 9 1 .27 . 9 3 . 4 9 94B - . 4 6 - . 4 7 .52 . 5 9 1 .27 . 8 9 . 5 3 94C - . 4 6 - . 2 6 .37 1.14 .34 . 18 . 9 3 . 4 5 94D - . 4 5 - . 1 6 .54 .31 .41 . 8 6 . 3 6 97A - . 4 3 1 .13 . 8 9 " . 2 8 101 .21 . 3 8 " ^ 8 4 . 2 0 102 - . 2 6 . 2 9 . 7 6 . 8 2 . 1 7 105 - . 1 9 . 6 9 1 . 2 6 . 9 3 . 2 9 115 - . 7 0 .31 1 .22 .74 .41 120 - . 2 6 - . 2 5 .78 . 9 0 . 9 6 . 5 5 121 - 1 . 0 3 .92 . 5 8 . 14 . 9 5 . 3 6 127 - . 7 7 .72 .74 .74 . 2 5 129 - . 4 8 .44 1 .29 . 8 8 .41 136 - . 4 2 . 4 8 . 3 5 . 2 9 . 9 0 .51 .150 - . 4 3 . 5 8 .94 . 9 4 . 5 7 152 - . 2 3 .86 .29 . 9 8 . 3 4 154 - . 4 0 .42 .72 .79 . 2 8 179 - . 0 6 .49 .15 . 8 3 . 9 9 • . 6 0 182 - . 4 5 .51 1 .22 . 8 9 . 4 3 184 .16 .52 .71 . 2 9 185 -.44 . 5 3 1.14 . 6 0 .85 . 5 0 CONTINUED Table 4.3 (Continued) Executive I.D. # COST PAYB PSUC MKT ROR GOVT R2 RE 191 -.24 .56 1.32 .88 .42 204 .43 .67 .93 .32 227C -.83 .89 .60 .41 .88 .52 2270 -.17 .52 .69 .27 .92 .49 231 -.85 -.20 .55 .49 .64 -.20 246 -.61 1.12 .87 .29 250 -.93 .15 .58 .27 .86 .13 253 1.50 .81 .33 253A -1.01 -.18 .67 .67 .84 .50 253C -.36 -.20 .39 1.29 .87 .53 253D -.47 -.26 .49 1.04 .87 .41 255 -.21 .63 .32 1.06 .92 .45 258 -.30 -.52 .92 .51 .92 .50 264 -.76 .87 .49 .43 .87 .41 269 .39 -.30 .34 1.04 .86 .25 271 -.29 -.21 .41 .49 .69 * .18 281 -.39 .69 .86 .20 295 -.40 -.17 .53 -.31 .93 .86 .36 299 -.61 .49 .55 .71 .29 325 -.61 -.13 .50 .78 .97 .51 335 -.40 -.18 .26 .48 .47 .36 337 1.07 .63 .40 358 -.43 .40 1.70 .86 .37 367 -.58 .71 .23 .41 • -92 .43 378 -.62 .76 .33 .93 .44 382 -.49 • .89 .54 .86 .32 389 -.48 .29 1.67 .86 .31 391A -.31 -.37 .58 .43 1.34 .93 .56 391B -.15 .61 .32 1.01 .95 -•-.49' 399 -.41 -.26 .40 1.86 .92 .51 400 -.28 .79 .68 .94 .46 412 -.62 .66 1.45 .69 .29 416 -.20 .65 .41 .93 .43 432 .28 1.54 .81 .14 436 -.79 .31 1.09 .93 .35 .86 .37 442 -.34 .76 .68 .20 456 -.17 .44 2.24 .96 .41 458 -.31 -.13 .56 .27 .24 .90 .29 463 -.23 .68 .75 .24 .93 .37 465 -.46 -.59 .97 .34 .79 .90 .53 473 -.41 .45 .74 .32 .88 .34 483 -.52 .92 .35 .93 .46 490 -.80 .60 .69 .69 - .32 495 -.49 -.15 .42 .64 .48 .86 .41 CONTINUED 164 Executive I.D. S COST 709 710 -.43 711 -.63 712 715 -.37 720 -.22 Average Model - - 3 0 Table 4.3 (Continued) PAYR PSUC MKT ROR ^'•'GOVT R2 -.32 .32 1.20 .87 -.23 .22 1.36 .85 -.37 .52 -.38 1.77 .83 -.30 .82 .45 .88 -.42 .41 1.63 .90 -.40 .78 .59 .47 .12 .98 -.29 .54 .22 .63 .14 .98 1 (-.37) (.79) (.14) (.42) (.16) R E .55 • .46 .40 .36 .46 .67 165 thus, suggesting that meaningful inferences about trade-off patterns can be derived from the regression weights. On average 3.5 attributes were found to be significant indicating a span of attention of 3 to 4 cues. The attributes in order of frequency of significance are: probability of success (significant for 90% of the executives), payback (77%), rate of return (.67%), cost (46%), impact on market share (42%), and government funding (24%). Thus the dimension consisting of success probability, time commitments and p r o f i t a b i l i t y assumed a prominent role in the selec-tion process. This result is consistent with ratings of importance reported in Chapter 3. ' j For the most part, the sign of the cues corresponded to those indicated by theory: cost and payback negative and the rest positive. J The Group Regression Model A regression model was run using as the dependent variables the average probability ratings of a l l subjects for funding each of the projects. The results are presented at the bottom of Table 4.3. To compare the relative importance of the attributes, normalized regression coefficients (scale free) are also given. All the attributes appear as significant in the group model. The order of significance of the a t t r i -butes i s : probability of success (.79), rate of return (.42), payback (-.37), cost (-.19), government funding (.16) and impact on market share (.14). 166 Discriminant Analysis Models Discriminant analysis was performed for each executive by grouping profiles into high and low rated categories (probability of funding for group 1 profiles was less than 50%, and for group 2, greater than or equal to 50%). Other groupings were also tried and for some individuals a better classification model was obtained, but on the average the 50-50 breakdown obtained highest accuracy of re c l a s s i f i c a -tion. The results, in the form of the standardized coefficients for the significant attributes are presented in Table 4.4. The classification obtained by discriminant analysis performed better than the probability of allocation by chance for a l l subjects. It yielded more accurate classification than the allocation to the maximal group with two exceptions: for subject #1 the discriminant analysis was only 3 percentage points better, and for subject #184 they were equal. In each of these cases an alternative grouping performed better. In case #1, the grouping less than 40% and greater than 60% has^96% correct allocation versus 88% probability of maximum allocation. This function included an additional attribute, government funding. In case #184 grouping project profiles to those with less than 20% chance of being funded and those with funding probabilities of 40-59% yielded a c l a s s i f i -cation function with 87% correct allocations versus 63% yielded by alloca-tion to the maximal group. This executive had not given high ratings to any of the profiles. The significant attributes are the same as. those obtained by the 50:50 classification. Probability of success was a significant discriminating variable for 95% of the executives, payback for 72%, rate of return for 70%, market 167 Table 4.4 Standardized Discriminant Coe f f i c i en t s for S i gn i f i can t Variables Executive COST PAYB I.D. t PSUC MKT ROR GOVT PCORR PKAX PCH 1 .36 .87 78 75 63 3 -.18 -.41 .84 .18 90 76 63 5 -.17 -.72 .57 .42 .22 92 73 59 6 -.45 .72 .42 .15 .26 88 58 51 12 .90 .23 .25 85 60 52 13 -.32 .18 .90 -.37 85 67 56 14 -.40 -.62 .59 .43 .17 92 77 .61 19 -.20 .73 .45 .16 82 67 56 20 .46 .57 -.37 .44 73 63 53 26 -.74 .56 .19 97 83 • 72 43 -.13 .17 .56 .75 .24 88 50 50 45 -.42 -.33 .49 .56 82 70 58 47 .28 .25 .88 90 58 51 49 -.94 .11 .10 .19 97 53 50 68 -.25 .51 .19 .65 .16 90 63 53. 74A .46 .81 78 62 53 85 -.17 .91 .42 90 70 58 '90 --19 -.85 .45 .42 82 57 51 90B -.24 -.35 .70 .35 .21 88 58 51 90G -.35 .82 .30 88 68 56 93 .13 -.55 .29 .64 88 57 51 93A -.55 .33 .64 90 55 51 94 -.36 -.32 .21 .77 83 58 51 94B -.28 -.73 .14 .33 .44 .17 82 55 51 50 94C -.40 -.28 .22 .88 .16 83 52 94D -.64 -.62 .54 83 77 65 97A .20 -.15 .93 93 52 50 101 .19 .54 '. .40 .21 .56 , 93 80 68 102 -.50 -.69 .20 .52 -.42 / 97 8 7 ^ 77 105 -.18 .50 .86 88 75 63 115 .75 .30 .68 88 63 53 120 -.27 .68 .44 .26 92 62 53 121 -.63 .49 .48 .16 .17 83 60 52 127 -.77 .39 .35 93 65 55 129 -.45 .37 .69 -.15 87 57 51 135 -.22 -.87 .50 .26 90 80 68 150 -.18 -.64 .42 .48 90 50 50 152 -.23 .92 .29 87 68 56 154 -.75 .33 .43 88 78 .66 179 -.15 -.20 .25 .86 93 73 61 182 .67 .11 .57 .17 90 57 51 184 .74 .55 77 77 65 186 -.18 -.65 .24 .44 .55 83 55 51 -191 .45 .81 87 67 56 CONTINUED 168 Table 4,4 (Continued) Executive ' I.D. 0 COST PAYB PSUC MKT ROR GOVT PCORR PMAX PCH 204 .21 .28 .98 90 57 51 227C -.84 .48 .32 .23 ... 80 52 50 227D -.38 .39 .28 .50 .47 80 65 55 231 -.86 -.54 .17 .28 92 87 77 246 -.15 -.75 .67 85 53 50 250 -.96 .22 93 83 72 253 .45 .21 .79 90 58 51 253A -.73 -.34 .40 .32 .13 92 72 60 253C .54 .40 .61 .19 92 53 50 253D -.30 -:21 .43 -.21 .68 87 63 53 255 -.20 .42 .17 .70 .25 90 77 65 258 -.15 -.61 .64 .30 93 50 50 264 -.71 .51 .39 '.17 82 57 51 269 .16 -.41 .16 .79 85 53 50 271 -.40 -.32 .61 .63 90 80 68 281 .16 .94 .20 85 58 51 295 .52 .64 .29 82 68 56 299 -.81 .19 .21 .72 85 72 60 325 -.27 .50 .77 93 50 50 335 -.43 -.55 .35 .53 92 90 82 337 .43 .41 .68 .20 87 77 70 358 -.19 -.50 .16 .74 83 55 51 367 -.38 -.14 .66 .39 .21 82 60 52 378 -.29 .71 .17 .43 90 55 51 382 -.66 .74 .22 78 50 50 389 .16 -.38 .15 . .84 87 60 52 391A -.23 .42 .32 .73 90 60 52 391B . -.14 .32 .24 .84 92 73 61 399 -.13 ' .42 .76 .16 95 60 52 • 400 -.40 .56 .20 .55 90 6 8 ^ 56 412 -.16 -.80 .41 .23 . 2 4 / 92 75 63 416 .81 .52 77 57 51 432 .29 .90 78 62 53 436 -.73 .14 .47 .37 87 52 50 442 -.23 .98 75 65 55 456 -.31 .22 .88 93 73 61 458 -.26 -.51 .72 .26 .40 60 52 463 -.31 .72 .62 .26 ' 87 72 - 60 465 -.47 .72 .17 .33 87 57 51 CONTINUED 169 T a b l e 4.4 ( C o n t i n u e d ) E x e c u t i v e I.D. t COST PAYB PSUC MKT ROR GOVT PCORR PMAX PCH 473 -.74 .33 .60 .39 80 57 51 488 -.79 .61 .11 90 52 50 490 -.21 .67 .57 82 60 52 495 -.62 -.34 .43 .52 .19 85 72 60 709 -.29 -.33 .28 .21 .68 .23 90 53 50 710 -.15 -.24 .29 .82 90 65 55 711 -.24 -.32 .35 .15 .72 92 62 . 53 712 -.23 .89 .24 83 52 50 715 -.36 .39 .73 93 60 52 720 -.13 -.44 .47 .29 .56 -.20 97 75 63 Average Models A -.43 .62 .49 • 78 35 24 B -.38 .64 .48 92 55. 53 170 share for 52%, cost for 48%, and government funding for 41%. This order is similar to the one obtained from the regression analysis (market and cost have changed ranks but are of similar importance). Government fund-ing is significant for twice as many executives as indicated by the regression models. As with the regression models, the signs are generally in the theoretically expected direction. The Group Discriminant Model Two group discriminant models were estimated, assigning profiles to groups on the basis of the average scores. Two groupings were tried: model A had five groupings (<20%, 20-39%, 40-59%, 60-79%, and 80-100%) and model B had three groupings (<20%, 40-59%, 80-100%). Both models performed better than allocation by chance and maximum allocation. The significant discriminating variables and their coefficients were similar in both models. Probability of success was most important, followed by the rate of return and payback. The results are reported at'the bottom of Table 4.4. Comparison of Regression and Discriminant Models The comparison of discriminant models and the regression models obtained for the same subjects provide additional evidence on the va l i d i t y and robustness of the relationships. Table 4.5 presents the ordered significant attributes for both the regression and the discriminant models (for ease of comparison, normalized regression coefficients are used). In twenty-three of the cases the order of attribute significances. 171 is exact [exact here is defined such that the order of the attributes is the same though one model may have more attributes than the other). In 68 of the cases there are only minor differences among the models. In only two of the cases the models are quite different. In case #20, time commitment (payback) in the regression model is replaced by resource commitment (cost) in the discriminant model, and risk (probability of success) is replaced by pr o f i t a b i l i t y (rate of return). Though the discriminant model performed better than the maximum allocation, i t was quite inaccurate in allocating profiles to group 2. In this case the regression analysis would be preferred. In case #49 payback- and market share appear in the regression model. They are replaced by government funding in the discriminant model. There are two cases of counter consensus direction of impact of the attributes that appear in both the regression and the discriminant models. In the other cases the sign is correct in the alternative model and/or the attribute is of low significance or does not appear at a l l in the alternative model. For executive #269, cost had positive impact perhaps reflecting an empire builder. For executive #97A market share has a negative impact perhaps reflecting a concern with antitrust legislation. The Group Models The order of the three most significant variables is the same in both models indicating that the probability of success, rate of return and payback are the most important attributes for R&D project evaluation and selection. Table 4.5 Comparison of Discriminant Analysis and Regression Analysis Executive I.D. # Significant Variables in Descending Order Regression ti,Discriminant 1 R C PB R PS 3 PS PB C R PS PB C=R 5 PS PB M PB PS M G C 6 PS PB M C PS PB M G R 12 PS G PB M PS R M 13 R PS C R G C PS 14 PS PB M C PB PS M C G 19 PS R PB PS R PB G 20 - G PS M PB H C G R 26 PS C C PS G 43 M PS C M PS R PB C 45 R R PS C PB 47 R R PS M 49 PS C R PB M C G PS R 68 R PS PB R PS PB M G 74A R PS R PS 85 PS PB M PS M PB 90 PS PB G PB PS G C 90B PS PB R C G PS . PB=M C G 90G PS PB C R PS PB R 93 PS PB R R PB PS C 93A PS PB R R PB PS 94 R PS PB C /R C PB PS 94B R PS PB M C PB R M C G" P 94C M PS PB C R G M C PB PS G 94D PS C R PB=M C PB PS 97A G M G PS M 101 G PS G PB PS R C 102 M PS PB PB M C R PS 105 PS M PB M PS PB 115 R PS C R PS M 120 PS R PB C PS R PB G 121 PS C M G C PS M G R 127 PS PB R PB PS R 129 R PS PB R PB PS G 136 PS PB G M PB PS G C " 150 PS R PB PB R PS C 152 PS M C PS M C 154 PS PB R PB R PS 179 PS R M PB R PS PB C CONTINUED Table 4.5 (Continued) Executive I.D. § Significant Variables in Descending Order Regression Discriminant 182 R PS C PS R G M 184 R PS PS R 186 R G PB M PB G R MC 191 R PS PB R PS 204 G M G M C 227C PS PB G M PB PS M G 227D PS R G PB R G PS PB M 231 PS C M PB C PB M PS 246 PS PB PB PS C 250 - PS C PB=R C PS 253 R R PS M 253A PS C R PB C PS PB R G 253C R PS PB C R PS M G 253D PS R PB C R PS C PB=M 255 PS R PB M R PS G PB M 258 PS PB R C PS PB R C 264 PS PB M R PB PS M R 269 R PS PB C R PB PS=C 271 PS M PB C M PS C PB 281 PS C PS G PB 295 PS R C PB M R PS G 299 PB PS G PB G M PS 325 G PS C PB G PS C 335 PS R C PB PB R C PS ' 337 R R PS M G 358 R PS PB R PB C PS 367 PS C R M PS R C G PB 378 PS C G PS G C R 382 PS PB R PS PB G 389 R PB PS R PB C PS 391A R PS PB H C R P f H PB 391B PS R M PB R PS M PB 399 R PS PB C R PS G PB 400 PS R PB PS R PB M 412 PS PB M PB PS R M C 416 PS G PB PS G 432 R PS R PS 436 PB H R PS G PB M R PS 442 PS PB PS PB 456 R M PB R PB M 458 PS G C M PB PS PB G PS=C CONTINUED Table 4.5 (Continued) 174 Executive I.D. # Significant Variables in Descending Order Regression Discriminant 463 PS M PB G PS M PB G 465 PS PB R C M PS PB R M 473 PS PB M G PB M G PS 488 PS PB R PB PS M 490 PS C R PS R C 495 PS M R C PB C M PS PB R 709 R PS PB . R PB c PS G 710 R PS PB C R PS PB C 711 " R PS PB C M R PS PB C M 712 PS PB R PS R PB 715 R PS PB C R PS PB 720 PS PB M R C=G R PS PB M G Average Model s PS R PB C G M PS R PB / 175 External Validity Some measure of external validity is provided by the respon-dents' comments. Consider a few examples: a. Executive #93A: Description: Subject is a manager of the sales department in a chemical firm with sales of $1-10 million, privately owned, not Canadian controlled. Market is somewhat volatile, firm is a leader and R&D is extremely important. Subject's Comments: "In the case of my own company R&D is largely financed by sales of existing products so that the influence of government subsidies and changes in the money market do not generally enter into our product (or project) evaluations. We generally look for a rate of return of 30% or more, and a pay-back period of around 3 years. Our impact on the market and the probability of marketing or of technical success is also c r i t i c a l to • our evaluation process. This requires a considerable amount of market research and c l i n i c a l research before a product is launched." Models: The models are largely in congruence with these comments. In the discriminant model the attribute order of significance i s : rate of return, payback, and prob-a b i l i t y of success. The attribute order is reversed in the regression model. Market share, however, does not appear in either model. But, impact upon market is captured perhaps by the profit indicator. b. Executive #97A Description: Subject is a manager of the R&D department of an electrical products firm with sales of $50-75 million, publicly owned with control by a few interests, not Canadian controlled. Market is stable, firm is a follower, and R&D is extremely important. 176 Subject's Comments: "In reviewing the answers we have pro-vided you might be interested in noting that we are a very conservative Company that is involved in develop-ing and marketing state-of-the-art products. This leads to an interesting dilemma which has been solved through the use of Government funding (mainly Canada and the U.S.A.). Every attempt is made to get any R&D effort f u l l y paid from external sources. Very often we sell our R&D effort outright. This has the advantage of achieving 100% funding but the disadvantage that we do not have exclu-sive rights to the resulting product. We rely on our i n i t i a l experience with the resulting product as a means of remaining competitive in any potential production requirement." Models: Government funding is most significant in both the regression and discriminant models. Market share,is also important (negative) in both, while the discriminant model includes probability of success. c. Executive #121 Description: Subject is a manager of an R&D department in a-firm in the transportation and communications sector, with sales greater than $250 million, publicly owned, Canadian controlled. Market is extremely stable, firm a follower, and R&D is of moderate importance. Subject's Comments: "We evaluated the sixty itemized conditions to indicate probability of funding by allocating points as follows to arrive at a ranking order. Probability of success 50% 50 to 60% 60 to 70% 70 to 80% 80 to 100% Points 0 1 2 3 4 R&D cost as related to total R&D expendi ture >30% <30% 0 3 177 Points Per cent Market <10% 0 >10% 2 Per cent Government <10% 0 support >10% 1 We consider that the above four c r i t e r i a are the most important ones in our decision making. Probability of. success is the highest c r i t e r i a commanding the maximum points (4) for probability of success between 80 to 100%. After establishing the ranking order we ar b i t r a r i l y divided the sixty situations into ten parts giving 100% probability of funding to the top six in the ranking order and working our way down the ladder." Summary of the model provided: probability of success and cost are most important, followed by impact on market share and government funding. Models: The models are confirmed by the comments. In the regression model the order of the attributes is prob-a b i l i t y of success, cost, impact on market share and government funding. The discriminant model reverses the order of the f i r s t two variables. d. Executive #253 Description: Subject is a manager of an R&D department of a firm in petroleum and coal sector, not Canadian controlled, Subject's Comments: "Your questionnaire . . . appears to assume that some level of government subsidy is necessary to make the private sector function. My own personal conviction is that such payments either have no effect or distort the market system by encouraging industry to embark on uneconomic unsustainable projects. I believe the main problem facing industrial innovators in Canada today is to define economical viable projects in an environment of rapid inflation, price controls, increasing and variable government regulations and very heavy taxation." 178 Models: For the regression the only significant attribute is the rate of return. For the discriminant model the important attributes are rate of return-, probability of success and impact on market share. Thevmodels are supported by the comments: government funding is not important and p r o f i t a b i l i t y measures are. e. Executive #711 Description: Subject is a Vice President of a chemicals firm with sales of $75-100 million. Market volatile, firm neither a follower nor a leader, R&D of moderate to low importance. Subject's Comments: A model of the procedure used by the firm was provided. The model selects projects to maximize expected rate of return adjusted for resource and time commitment by calculating a discounted cost function as a combination of the payback and the relative cost of the project. Models: In both the regression and discriminant models, the order of significant attributes was: rate of return, probability of success, payback, cost and impact on market share. The models are generally confirmed by the comments with the addition of impact on market share (low coefficient, .15). Analysis of Winning Sets This section deals with project attributes, levels associated with high probabilities of funding. We, therefore, focus upon those projects which received on the average a funding probability of 70% or more in our sample defining them as winning sets. After characterizing the average winning set of projects, we focus upon differences among winning sets for groups of subjects classified on the basis of alternative personal and 179 organizational attributes. One must note, however, that the high degree of f i t obtained in the estimated average models of judgment indicates that differences among individuals are relatively low, and therefore one may expect similarity of judgment patterns among groups. The interpretation of intergroup differences in judgment patterns will focus only on those patterns where differences are significant. The winning set of project profiles is presented in Table 4.6. The required rate of return for successful candidate profiles is high (greater than 35%), payback is less than 5 years, impact on market share is greater than 15% increase, and government funding greater than 34%. Cost-levels did not display a consistent pattern. Cost levels vary over the total range. Probability of success is greater than 74%. The tradeoffs are striking for these profiles: i f the payback is less than 2 years, a probability of success of 74% is acceptable; i f the payback is moderate (about 2-1/2 years) the compensating probability of success must be high (about 90%); the same holds for long paybacks (4-1/2 to 5 years) with one exception,, a project where lower probability of success (83%) was compensated by high market impact and high government funding. Other projects in the portfolio with ROR greater than 30%, generally had low probability of success or long paybacks. Analysis of differences in winning sets (additions or deletions from the total sample average set) identified the following relationship between two project attribute tradeoffs and individual and organizational characteristics: Presidents deleted from the winning sets a l l projects with payback period higher than 2-1/2 years and a l l projects with high (>80%) government 180 Table 4.6 The Winning Set of Projects Payback Period 0 yrs 9 mo % f mkt 37% ROR 35% % Gov't $ 34% Cost R&D 2% Prob. Success 74% 41 probability of funding = 82% Payback Period 1 yrs 9 mo % + mkt 15% ROR 47% % Gov't $ Q6% Cost R&D 23% Prob. Success 74% 60 • probability of funding = 72% Payback Period 2 yrs 4 mo % + mkt 24% ROR 49% % Gov't $ 75% Cost R&D 50% Prob. Success ? 9 0 % 15 probability of funding = 76% Payback Period 2 yrs 7 mo % t mkt 20% ROR 50% % Gov't $ 7jl% Cost R&D 45% Prob. Success 93% 23 probability of funding = 81% Payback Period 4 yrs 5 mo % + mkt 49% ROR 37% % Gov't $ 88% Cost R&D 26% Prob. Success 83% 59 probability of fundi ng = 70% . Payback Period 4 yrs 7 mo % t mkt 29% ROR 36% % Gov't $ 60% Cost R&D *2% Prob. Success 92% 50 ". probability of funding = 73% Payback Period 5 yrs 0 mo % + mkt 20% ROR 42% •% Gov't $ 64% Cost R&D 1% Prob. Success 95% 55 probability of funding = 78% 181 funding. Senior managers seemed to favour the winning set identified for the sample as a whole while staff executives were making more liberal tradeoffs between rates of return and risks, expanding the winning set with two riskier investment alternatives. Executives in stable markets differed from those in vol a t i l e markets in the tradeoff they made between probability of success and longevity of payback period. Those in volatile markets eliminated projects from their winning set with long payback periods except for those with high probability of success. Large companies [with sales above $50 million) tended to make more liberal tradeoffs between rates of return and payback periods and between risk and rates of return. They added to the winning l i s t projects with moderate rates of return and shorter payback periods as well as projects with higher risks and higher rates of return. Small companies (with less than $1 million in sales) tended to accept high risks for high rates of return i f cost commitments are low and expected impact upon market shares is high. Size measured by employment is positively related to the range of risk tradeoffs. Executives from small companies (having below 100 employees) tended to focus upon safe projects only, high ROR (>36%), high probability of success (>90%) and short payback period (>5 years). Those from medium companies added to the winning set safe projects with high expected rates of return but with longer payback horizons, while those from large companies (with above 1000 employees) permited a l l the range of tradeoffs identified in our discussion of the average winning set. 182 The impact of ownership patterns on funding preferences suggests that public companies with widely held shares have greater concern with market shares for which they are willing to take higher risks. Concen-trated ownership patterns (both in private and public corporations) are associated with no such attention to market share performance. No differences were discovered among the winning sets of Canadian and non-Canadian controlled companies. However, when R&D project managers are removed from the sample, and comparisons are made between the winning sets of executives in Canadian and non-Canadian controlled firms, i t is noted that Canadian executives tended to choose conservatively, imposing higher safety margins upon project selection. Policy Implications and Conclusions The study has several implications for government policy with respect to R&D stimulation in the private sector. F i r s t , i t suggests the usefulness of general compensatory actuarial models for predfcting prefer-ences among R&D investment opportunities. While Chapter 3 had demon-strated the selective impacts.of environmental economic variables upon R&D investment, this chapter has demonstrated the existence of a high consensus in executive judgments of projects by their specific attributes. This consensus was reflected in the high correspondences between predicted, values and observations for the models representing the sample as a whole, as well as the similarity of winning sets of alternative executive and firm g r o u p i n g s . 183 Differences in judgment formation are realized mainly in ranges of tradeoff between risks and rates of return. Government subsidies and participation in funding in the private sector do not have a high direct impact upon R&D decisions. On the basis of comments received from a variety of executives there is fear of increased level of government interference in managerial decisions associated with receipt of government funds. This observation coupled with the observation that project evalua-tion for many companies is marked by conjunctive selection, once certain risk thresholds are exceeded, points to a new role for government in the f i e l d . The role proposed is that of an independent insurance agency. Insurance permits firms to trade rates of return and risks. This provides an expanded choice space and many candidate projects rejected now as too risky (though with high expected payoffs) may join the winning sets. POSTSCRIPT This dissertation has focused upon some important aspects of R&D decision making in Canada. A review of the state of the art led to the identification of the following four areas of information which are deficient: 1. i n f o r m a t i o n a b o u t t h e n a t u r e o f s e l e c t i v e p e r -c e p t i o n p r o c e s s e s o f R&D d e c i s i o n m a k i n g ; 2. t h e o b j e c t i v e f u n c t i o n s ( e x p l i c i t a n d l a t e n t ) w h i c h g u i d e c h o i c e s among a l t e r n a t i v e R&D i n -v e s t m e n t o p p o r t u n i t i e s ; 3. t h e i m p a c t o f R&D u p o n t h e p o s i t i o n s o f p r i m e b a r g a i n i n g u n i t s i n o r g a n i z a t i o n s ; a n d 4. t h e i m p a c t s o f o r g a n i z a t i o n a l s t r u c t u r e a n d p r o c e s s e s u p o n i m p l e m e n t a t i o n o f i n v e s t m e n t d e c i s i o n s . Attempts have been made to contribute to the f i r s t three areas. Some of the major findings of the present research are: 1. A c c e p t i n g a n e o c l a s s i c a l f r a m e w o r k o f a n a l y s i s , i n m o s t s e c t o r s , R&D h a s h a d n o i m p a c t o n i n p u t s h a r e s , i n d i c a t i n g t h a t s c a l e a n d p r i c e e f f e c t s d o m i n a t e t h e s t r u c t u r e o f t e c h n o l o g y . W h e r e R&D h a s had i m p a c t o n t h e s t r u c t u r e , i t h a s s o m e t i m e s had a l a b o u r u s i n g a n d s o m e t i m e s a c a p i t a l u s i n g b i a s . 2 . S i g n i f i c a n t d i f f e r e n c e s i n p a t t e r n s o f a t t e n t i o n t o e n v i r o n m e n t a l c o n d i t i o n s w e r e i d e n t i f i e d . T h e s e d i f f e r e n c e s a r e r e l a t e d t o e x e c u t i v e a t t r i b u t e s a n d f i r m c h a r a c t e r i s t i c s . 184 185 3. H i g h c o n c e n s u s w i t h r e s p e c t t o t r a d e o f f s among p r o j e c t a t t r i b u t e s a c r o s s a l l f i r m - e x e c u t i v e g r o u p i n g s e x i s t s . 4. C o m p e n s a t o r y a c t u a r i a l m o d e l s p r o v i d e a g o o d f i t w i t h o b s e r v a t i o n s o f R&D i n v e s t m e n t j u d g m e n t s . Some normative implications of the study for public policy include the following: 1. A s R&D i m p a c t u p o n t h e e c o n o m i c o b j e c t i v e s o f t h e m a j o r b a r g a i n i n g u n i t s i n f i r m s i s n e u t r a l i n m o s t c a s e s , p e r h a p s a n e f f o r t s h o u l d be made t o e l i m i n a t e t e c h n o l o g i c a l d e v e l o p m e n t a s p a r t o f t h e t r a d i t i o n a l a r e n a o f l a b o u r - m a n a g e m e n t b a r g a i n i n g . 2. In c r e a t i n g f a v o u r a b l e R&D i n v e s t m e n t c l i m a t e s , g o v e r n m e n t o u g h t t o d e v e l o p a s e n s i t i v e s t r a t e g y w h i c h r e c o g n i z e s e x p l i c i t l y t h e s e l e c t i v e i m p a c t o f s i n g l e d i m e n s i o n i n t e r v e n t i o n s o n a l t e r n a t i v e t a r g e t p o p u l a t i o n s . 3. T h e r o l e o f g o v e r n m e n t a s a n i n d e p e n d e n t i n s u r a n c e , a g e n t f o r R&D v e n t u r e s i s r e c o m m e n d e d t o r e p l a c e d i r e c t p a r t i c i p a t i o n i n p r o j e c t f u n d i n g . >' Promising areas for future research are to: 1. i n v e s t i g a t e R&D d e c i s i o n s a s g r o u p p r o c e s s e s w i t h e m p h a s i s o n i n t e r n a l b a r g a i n i n g a n d t h e r e l a t i o n -s h i p b e t w e e n o r g a n i z a t i o n a l s t r u c t u r e a n d s t a n d a r d o p e r a t i n g p r o c e d u r e s o n t h e s u c c e s s f u l i m p l e m e n t a -t i o n o f R&D a c t i v i t y ( i t e m 4 i n t h e a g e n d a o f a r e a s i d e n t i f i e d a b o v e ) ; 2. d e s i g n s o c i a l m a r k e t i n g s t r a t e g i e s t o s t i m u l a t e R&D a c c o r d i n g t o t h e d i f f e r e n c e s i n a t t e n t i o n p a t t e r n s t o e n v i r o n m e n t a l v a r i a b l e s a s i d e n t i f i e d i n C h a p t e r 3; d e v i s e an i n s u r a n c e s c h e m e t o p e r m i t f i r m s t o t r a d e o f f r a t e s o f r e t u r n a n d r i s k t o i n d u c e i n c r e a s e d R&D a c t i v i t y ; e s t i m a t e c o s t f u n c t i o n s u s i n g d a t a f r o m i n d i v i d u a l f i r m s t o i n v e s t i g a t e i f t h e a p p a r e n t n e u t r a l i t y o f R&D e x p e n d i t u r e i s a n a g g r e g a t i o n p h e n o m e n o n , a n d : 186 5. c o n d u c t a n i n t e r n a t i o n a l c o m p a r i s o n o f t h e f i n d i n g s r e p o r t e d ; s p e c i f i c a l l y i t w o u l d be f r u i t f u l t o i n v e s t i g a t e w h e t h e r : a ) t h e p a t t e r n s o f a t t e n t i o n t o e n v i r o n -m e n t a l c o n d i t i o n s a r e c u l t u r e - d e t e r m i n e d , b) t h e same c o n c e n s u s w i t h r e s p e c t t o t r a d e o f f s among p r o j e c t a t t r i b u t e s a c r o s s a l l f i r m - e x e c u t i v e g r o u p i n g s e x i s t s i n o t h e r b u s i n e s s e n v i r o n m e n t s , a n d , c ) t h e i m p a c t o f R&D o n i n p u t s h a r e s i s e s s e n t i a l l y n e u t r a l i n a l l e c o n o m i e s i n d e p e n d e n t o f t h e p r o p o r t i o n o f t h e - GNP d e v o t e d t o R&D. REFERENCES Allen, J.M. (1970) A survey into the R&D evaluation and control procedures currently used in industry. Journal of I n d u s t r i a l Economics, 18, 161-181. Allen, R.G.D. (1938) Mathematical Analysis for Economists. MacMillan, London. Ansoff, H.I. (.1965) Corporate Strategy: An Analytic Approach to Business Policy for Growth and Expansion. McGraw-Hill, New York. Ansoff, H.I. and J.M. Stewart (1967) Strategies for a technology-based business. Harvard Business Review, 45, 71-83. Arrow, K.J. (1962) Economic welfare and the allocation of resources for invention in N.B.E.R. The Rate and Direction of Inventive A c t i v i t y . Princeton University Press, pp. 609-625. Arrow, K.J., H.B. Chenery, B.S. Minhas and R.N. Solow (1961)., Capital and labour substitution and economic efficiency. Review of Economics and Statistics, 43, 225-250. Atkinson, A.C. and A.H: Bobis (1966) A mathematical basis for the selec-tion of research projects. IEEE Transactions on Engineering Management, EM-16, 2-8. Bacharach, M. (1970) Biproportional Matrices and Input-output Change. Cambridge University Press, Cambridge. Baguley, R.W. and J.J. Booth (1965) Royal Bank Trendicator Report, Vol.. 2, No. 8, Economics Department, The Royal Bank of Canada. Baldwin, W.L. and G.L. Childs ( 1 9 6 9 ) The fast second and rivalry in research and development. Southern Economic. Journal, 36, 18-24. 187 188 Beattie, C.J. (1970) Allocating resources to research in practice, in E.M.L. Beale (ed.), Applications of Mathematical Programming Techniques. . American Elsevier Publishing Co., Inc., New York, pp. 281-292. Becker, G.T. (1964) Human Capital. National Bureau of Economic Research, Columbian University Press, New York. Beged-Dov, A.G. (1965) Optimal assignment of R&D projects in a large company using an integer programming model. IEEE Transactions on Engineering Management, EM-12, 138-142. Bell, D.C. and A.W. Read (1970) The application of a research project selection method. R&D Management, 1, 35-42. Berndt, E.R. and D.O. Wood (1975) Technology, prices and the derived demand for energy. Review of Economics and Statistics, 57, 259-268. Binswanger, H.P. (1974) The measurement of technical change biases with many factors of production. American Economic Review, 64, 964-976. Birnbaum, M.H. (1973) The devil rides again: correlation as an^index of f i t . Psychological Bulletin, 79, 239-242. Birnbaum, M.H. (1973) Reply to the deyil's advocates: don't confound model testing and measurement. Psychological B u l l e t i n , 8, 854-859. Bobis, A.H., T.F. Cooke and J.H. Paden (1971) A funds allocation method to improve the odds for research success. Research Management, 14, 34-49. Bowman, M.J. (1964) Schultz, Denison and the Contribution of 'Eds' to national income growth. Journal of P o l i t i c a l Economy, 72, 450-464. Bright, J.R. (1970) Evaluating signals of technological change. Harvard Business Review, 48, 62-70. ' . B r o o k s , H. (1972) What's happening to the U.S. lead in technology? Harvard Business Review, SO, 110-118. Brown, M. and J.S. de Cani (1963) Technological change and the distribu-tion of income. International Economic Review, 4, 289-309. 189 Brown, M. and A. Conrad (19671 The influence of research and education in CES production relations in M. Brown, ed., The Theory and Empirical Analysis of Production. National Bureau of Economic Research, Columbia University Press, New York. Charnes, A. and A.C Stedry (1966) A chance-contrained model for real-time control in research and development management. Management Science, 12, B-353 - B-362. Christensen, L.R. and D.W. Jorgenson (1969) The measurement of U.S. real capital input 1929-1967. Review of Income and Wealth, 15, 293-320. Christensen, L.R., D.W. Jorgenson and L.J. Lau (1973) Transcendental logarithmic production frontiers. Review of Economics and S t a t i s t i c s , 55, 28-45. Cobb, C.W. and P.M. Douglas (1928) A theory of production. American Economic Review Papers and Proceedings, 18, 139-165. Cochran, M.A., E.B. Pyle III, L.C. Greene, H.A. Clymer and D. Bender Investment model for R&D project evaluation and selection. IEEE Transactions on Engineering Management, EM-18, 89-100. (1971) Collcutt, R.H. and R.D. programme for 219-242. Reader (1967) Choosing the operational research B.I.S.R.A. Operational Research Quarterly, 18, Comanor, W.S. (1967) Market structure, product differentiation and industrial research. Quarterly Journal of Economics, 81, 639-657. Cooper, A.C. (1966) Small companies can pioneer new products. Business Review, 44, 162-179. Harvard Cranston, R.W. (1974) First experiences with a ranking method for portfolio selection in applied research. IEEE Transactions on Engineering Management, EM-21, 148-152. Cummings, E.D. and W.E. Diewert (1973) Manpower in a general equilibrium context: the model specification. Department of Manpower and Immigration, Canada. Cyert, R.M. and J.G. March (1963) A Behavioral Theory of the Firm. Prentice Hall, New Jersey. 190 Dawes, R.M. (1971) A case study of graduate admissions: application of three principles of human decision making. American Psychologist, 26, 180-188. Dawes, R.M. and B. Corrigan (1973) Linear models in decision-making. Psychological Bulletin, 80, 96-106. Dean, B.V. and L.E. Hauser (1967) Advanced materiel systems planning. IEEE Transactions on Engineering Management, Em-14, 21-43. Denison, E.F. (1962) The sources of economic growth in the U.S. and the a l t e r n a t i v e s before us. Committee for Economic Development, New York. Denison, E.F. (1964) The unimportance of the embodied question. American Economic Review, 54, 90-94. Denison, E.F. (1966) Why growth rates differ. Unpublished. Diewert, W.E. (1971) An application of the Shepard duality theorem: A generalized Leontief production function. Journal of P o l i t i c a l Economy, 79, 481-507. Diewert, W.E. (1974) Applications of duality theory in M.D. I n t r i l l i g a t o r " and D.A. Kendrick, Eds., Frontiers of Quantitative Economics, Vol. II. North Holland, Amsterdam. Diewert, W.E. (1975a) Exact and superlative index numbers,'Department of Manpower and Immigration, Ottawa (forthcoming, Journal of Econometrics). Diewert, W.E. (1975b) Annual employment, industry by occupation in Canada, Department of Manpower and Immigration, Ottawa. Disman, S. (1962) Selecting R&D projects for profit. Chemical Engineering, 69, 87-90. Drucker, P.F. (1975) Quoted in R. Perry, Drucker formula, information is power. The Financial Post, April 5. Einhorn, H.J. (1971) Use of nonlinear, noncompensatory models as a function of task and amount of information. Organizational Befiavior and Human Performance, 6, 1-27. 191 Einhorn, H.J. and R.M. Hogarth (1975) Unit weighting schemes for decision making. Organisational Behavior and Human Performance, 17>, 171-192. Fellner, W. (1971) Empirical support for induced innovation. Quarterly Journal of Economics, 85, 580-605. Foster, R.N. (1971) Organize for technology transfer. Harvard Business Review, 49, 110-120. Franchet, Y., R.A. Inman and A.S. Manne (1973) Numerical data for multi-sector planning in L.M. Goreux and A.S. Manne, eds., M u l t i - l e v e l planning: ease studies in Mexico. North Holland, Amsterdam. Frank, R.E., W.F. Massy, and D.G. Morrison (1965) Bias in multiple discriminant analysis. Journal of Marketing Research, 2, 250-258. Galbraith, J.K. (1973) Economics and the Public Purpose. Houghton M i f f l i n , Boston. Gear, A.E., J.A. Gillespie and J.M. Allen (1972) Applications of decision trees to the evaluation of applied research projects. Journal of Management Studies, 9, 172-181. Gear, A.E., A.G. Lockett, and A.W. Pearson (1971) An analysis of some portfolio selection models for research and development. IEEE Transactions on Engineering Management, EM-18, 66-76. Gerstenfeld, A. (1970) E f f e c t i v e Management of Research and Development. Addison-Wesley, Reading, Mass. Goldberg, L.R. (1971) Five models of c l i n i c a l judgment: an empirical comparison between linear and nonlinear representations of the human inference process. Organizational Behavior and Human Performance, 6, 458-479. Green, P.E. (1968) Descriptions and explanations: a comment on papers by Hoffman and Edwards in B. Kleinmuntz, editor, Formal Representa-tions of Human Judgment. John Wiley and Sons, New York, pp. 91-98. - ... Green, P.E. and F.J. Carmone (1974) Evaluation of multiattribute alterna-tives: additve versus configural u t i l i t y measurement. Decision Sciences, 5, 164-181. 192 Griliches, Z. (1963) The sources of measured productivity growth: U.S. agriculture 1940-60. Journal of P o l i t i c a l Economy, 71, 331-346. Griliches, Z. (1964) Research expenditures, education and the aggregate agricultural production function. American Economic Review, 54, 961-974. Griliches, Z. (1968) Production functions, technical change and a l l that. Netherlands Econometrics Institute Report 6328. Grossman, D. and S.N. Gupta (1974) Dynamic time-staged model for R&D • portfolio planning - a real world case. IEEE Transactions on Engineering Management, EM-21S 141-147. Halm J. (1973) U:B.C. FAN: Factor Analysis, Computing Centre, The University of British Columbia, Vancouver, Canada. Hamberg, D. (1966) R&D: Essays on the economics of research and development. Random House, New York. Hamien, Y. and V.W. Ruttan (.1970) Factor prices and technical change in;, agricultural development: the United States and Japan, 1880-1960. Journal of P o l i t i c a l Economy, 78, 1125-1135. Helin, F.A. and W.E. Souder (1974) Experimental test of a Q-Sort procedure for prioritizing R&D projects. IEEE Transactions on Engineering Managementa EM-21, 159-164. Hertz, D.B. (1968) Investment policies that pay off. Harvard Business Review, 49, 96-108. Hess, S.W. (1962) A dynamic programming approach to R&D budgeting and project selection. I.R.E. Transactions on Engineering Mangement, EM-9, 170-179. Jorgenson, D.W. ( 1 9 6 6 ) The embodiment hypothesis. Journal of P o l i t i c a l Economy, 74, 1-17. Jorgenson, D.W. and Z . Griliches ( 1 9 6 7 ) The explanation of productivity change. Review of Economic Studies, 34, 249-283." Kami en, H . I . and N.L. Schwartz ( 1 9 6 8 ) Optimal "induced" technical change.. Econometrica, 36, 1-17. Kami en, M.I. and N.L. Schwartz ( 1 9 7 1 ) Expenditure patterns for risky R&D projects. Journal of Applied Probability, 8, 60-73. 193 Kamien, M.I. and N.L. Schwartz (J974). Risky R&D with rivalry. Annals of Economic and Social Measurement, 3, 267-277. Kennedy, C. and A.P. Thirlwall 0972) Technical Progress: a survey. Economic Journal, 82, 11-72. Kepler, CE. and A.W. Blackman (1973) The use of dynamic programming techniques for determining resource allocations among R&D projects: an example. IEEE Transactions on Engineering Management, EM-20, 2-5. Keynes, J.M. (1964) The General Theory of Employment, Interest and Money. Harcourt, Brace and World, New York. Kim, J. (1975) Factor Analysis in Nie, N.H., C.H. Hull, J.G. Jenkins, K. Steinbrenner and D.H. Bent (eds.), S t a t i s t i c a l Package for the Social Sciences, 2nd Ed., McGraw-Hill, New York. Klecka, W.R. (1975) Discriminant Analysis in Nie, N.H., C.H. Hull, J.C. Jenkins, K. Steinbrenner, and D.H. Bent (ed.), S t a t i s t i c a l Package for the Social Sciences, 2nd Ed., McGraw-Hill, New York. Kort, F. (1968) A nonlinear model for the analysis of judicial decisions. The American P o l i t i c a l Science Review, 62, 546-555. , Kotler, P. (1967) Operations research in marketing. Harvard Business Review, 45, 30-44. Leonard, W.N. (1971) Research and development in industrial growth. Journal of P o l i t i c a l Economy, 79, 232-255. Levitt, T. (1966) Innovative imitation. Harvard Business Review, 44, 63-70. Lithwick, N . H . (1969) Canada's Science Policy and the Economy. Methuen Publications, London. Lockett, A.G. and P. Freeman (1970) Probabilistic networks and portfolio selection. Operations Research Quarterly, 21, 353-360. Lockett, A.G. and A.E. Gear (1973) Representation and analysis of multi-stage problems in R&D. Management Science, 19, 947-960. Lucas, Jr., R.E. (1971) Optimal management of a research and development project. Management Science, 17, 679-697. 194 Maclay, W.N. (1974) Appendix to "Experimental test of Q-sort procedure for prioritizing R&D projects." IEEE Transactions on Engineering Management, EM-21, 163-164. Mansfield, E. (1964) Industrial research and development expenditures: determinants, prospects and relation to size of firm and inventive output. Journal of P o l i t i c a l Economy, 72, 319-340. Mansfield, E. (1965) Rates of return from industrial research and developr-ment. American Economic Review, 55, 310-322. Mansfield, E. (1968) I n d u s t r i a l Research & Technological Innovation. Norton & Co., New York. Mansfield, E. (1969) Industrial research and development: characteristics, costs and diffusion of results. American Economic Review, Papers and Proceedings, 59, 65-71. Mansfield, E. (1972) R&D's contribution to the economic growth of the nation. Research Management, 15, 30-46. March, J.G. and H.A. Simon (1958) Organizations. Wiley, New York. McGlauchlin, L.D. (1968) Long-range technical planning. Harvard Business Review, 46, 54-64. McLeod, Young & Weir Co., Ltd. Forty bond yield averages. Toronto, Mimeo. Minasian, J. (1969) Research and development, production functions and rates of return. American Economic Review, 59, 80-85. Ministry of State for Science and Technology (1973) Directory of Research and Development Establishments in Canadian Industry. Information Canada, Ottawa. Moore, J.R. and N.R. Baker (1969) Computational analysis of scoring models for R&D project selection. Management Science, 16, B-212-B-232. Moore, R.L. (1974) Methods of determining priorities in a programme of research. IEEE Transactions on Engineering Management, EM-21 126-140. Morgan, P. (1975) Inflation helps raise the top 100 to new heights. Financial Post, July 26. Morrison, D.G. (1969) On the interpretation of discriminant analysis. Journal of Marketing Research, 6, 156-163. 195 Hosteller, F. and R.R. Bush (1954) Selective quantitative techniques in A. Lindsey, ed. Handbook of Social Psychology, Vol. 1, Addison-Wesley, Reading, Mass. Mottley, CM. and R.D. Newton (1959) The selection of projects for industrial research. Operations Research, 7, 740-751. Mueller, R.K. (1971) The Innovation Ethic. American Management Association. Muleuk, S.A. (1972) The Foundations of Factor Analysis. McGraw-Hill, New York. Nelson, R.R. (1959) The simple economics of basic s c i e n t i f i c research. Journal of P o l i t i c a l Economy, 67, 297-305. Nutt, A.B. (1965) An approach to research and development"effectiveness. IEEE Transactions on Engineering Management, EM-12, 102-112. Peterson, R.W. (1967) New venture management in a large company. Harvard Business Review, 45, 68-76. Phi l l i p s , A. (1966) Patents, potential competition and technical progress. American Economic Review, Papers and Proceedings, 56, 301-310. Preston, R.S. (1975) The Wharton long term model: Input-output within the context of a macro forecasting model. International Economic Review, 3-19. Quinn, J.B. (1966) Technological competition: Europe vs. U.S. Harvard Business Review, 44, 113-130. Quinn, J.B. (1967) Technological forecasting. Harvard Business Review, 45, 89-106. Raiffa, H. (1968) Decision Analysis. Addison-Wesley, Reading, Mass. Reader, R.D., M.F. James and R.W. Goodsman (1966) The evaluation and selection of research projects - A progress report and appendices. BISRA, The Inter-Groups Laboratories of the British Steel Corp., London. Rosen, E.M. and W.E. Souder (1965) A method of allocating R&D expenditures. IEEE Transactions on Engineering Management, EM-12, 87-93. Rosenberg, N. (1974) Science, invention and economic growth. Economic Journal, 84, 90-108. 196 Rummel, R.J. (1970) Applied Factor Analysis. Northwestern University Press, Evanston. Samuelson, P.A. (1953) Prices of factors and goods in general equilibrium. Review of Economic Studies, 21, 1-20. Scherer, F.M. (1971) I n d u s t r i a l Market Structure and Economic Performance. Rand McNally & Company, Chicago. Schmookler, J. (1966) Invention and Economic Growth. Harvard University Press, Cambridge. Schultz, T.W. (1960) Capital formation by education. Journal of P o l i t i c a l Economy, 68, 571-583. Schumpeter, J. (1971) The instability of capitalism, reprinted in N. Rosenberg. The Economics of Technological Change. Penguin Books, Middlesex. Schwartz, S.L. (1964) Note on the availability of interindustry tables. Information on Economic Growth Technical Report #5, University of California, Berkeley. Schwartz, S.L. (1973) Second-hand machinery in development or how to recognize a bargain. Journal of Development Studies, 9, 544-555. Schwartz, S.L., I. Vertinsky, W.T. Ziemba and M. Bernstein (1976) Some behavioural aspects of information use in decision "making: a study of c l i n i c a l judgments in H. Thiriez and S. Zoints (eds.), M u l t i p l e C r i t e r i a Decision Making, Springer-Verlag, Heidelberg, 378-389. Shepard, R.W. (1953) Cost and Production Functions. Princeton University Press, Princeton. Shephard, R.W. (1970) Theory of Cost and Production Functions. Princeton. University Press, Princeton. Slovic, P. (1969) Analyzing the expert judge: a descriptive study of a stockbroker's decision processes. Journal of Applied Psychology,. S3, 255-263. Smith, V.K. (1973) A review of models of technological change with reference to the role of environmental resources. Socio-Economic Planning Sciences, 7, 489-509. 197 Solow, R.M. 0957) Technical change and the aggregate production function. Review of Economics and Statistics, 35, 312-320. Solow, R.M. (.1962) Technical progress, capital formation and economic growth. American Economic Review, Papers and Proceedings, 52, 76-92. Souder, W.E. (.1972) A scoring model methodology for rating management science models. Management Science, 18, B-526 - B-543. Souder, W.E. (.1974) Autonomy, gratification and R&D outputs: a small-sample f i e l d study. Management Science, 20, 1147-1156. Souder, W.E. (1974b) Field studies with Q-sort/nominal-group process for selecting R&D projects. Forthcoming in Research. Policy. Souder, W.E. (1975) Achieving organizational consensus with respect to R&D project selection c r i t e r i a . Management Science 21y 669-681. Souder, W.E. (1975b) Effectiveness of Delphi and other group processes for integrating strategic R&D market planning decisions. Forthcoming in Management Science. Souder, W.E., P.M. Maher and A.H. Rubenstein (1973) Two successful experiments in project selection. Research Management, 15, 44-54. Terleckyi, N.E. (1960) Sources of productivity advance, a pilot study of manufacturing industries. Ph.D. Dissertation, Columbia University. Theil, H. (1967) Economics and Information Theory. North-Holland, Amsterdam. Thurston, R.H. (1971) Make TF serve corporate planning. Harvard Business Review, 49, 98-102. T i l l e s , S. (1966) Strategies for allocating funds. Harvard Business Review, 44, 72-80. T i l ton, J.E. (1971) International Diffusion of Technology: The Case of Semi-Conductors. The Brookings Institution, Washington, D.C. Toffler, A. (1971) Future Shock. Bantam, Toronto -U.S. Department of Commerce (1975) Survey of Current Business, Washington. 198 Uzawa, H. (1962) Production functions with constant e l a s t i c i t i e s of substition. Review of Ecanomic Studies, 29, 291-299. Wets, R. (1976) "An Algorithm for Stochastic Programs with Simple Recourse: Forthcoming in Mathematical Programming. Wilde, D.J. and CS. Beightler (1967) Foundations of Optimization. Prentice-Hall, Englewood C l i f f s , N.J. Williams, D.J. (1969) A study of a decision model for R&D project selec-tion. Operational Research Quarterly, 20, 361-363. Williamson, O.E. (1965) Innovation and market structure. Journal of P o l i t i c a l Economy, 73, 67-73. Woodland, A.D. (1972) The construction of price and quantity components of inputs for Canadian industries, 1927-69. Department of Manpower and Immigration, Canada. Zangwill, W.I. (1969) Nonlinear Programming: A Unified Approach. Prentice-Hall, Englewood C l i f f s , N.J. Zoppoth, R.C (1972) The use of systems analysis in new product develop-ment. Long Range Planning, S, 23-36. Data Sources Statistics Canada. Aircraft and Aircraft-Parts Manufacturers. General Review of the Manufacturing Industries of Canada. Household Furniture Manufacturers. Industrial Research and Development Expenditures. Iron and Steel Mills. Manufacturers of Small Electrical Appliances Miscellaneous Chemical Industries. 199 Statistics Canada. Miscellaneous Food Processors. (Continued) Miscellaneous Machinery and Equipment Manufacturers Miscellaneous Metal Fabricating Industries. Miscellaneous Non-metallic Mineral Products Industries. Miscellaneous Petroleum and Coal Products Industries. Miscellaneous Textile Industries. Miscellaneous Wood Industries Pulp and Paper Mills. Rubber Products Industries. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0094327/manifest

Comment

Related Items