'J I I 3 1293 001 IIHIMWWHWINHMWMWI I L/__ RETURNING MATERIALS: IV‘ESI_J Place in oook drop to LIBRARIES remove thIS checkout from 1.53:,!!_. your record. FINES wiII be charged if book is returned after the date stamped beIow. APW I 2, I 0 7 8 . Quanta“ W 00f ! 7199 "1'39? 1 23 1 @512. 2.5499?” u:’ . ” cxiA’ ~ . .n. _ 9’ JL ,3") -—~‘\ . 7+. KR \. 2 <1? /' $656M MANAGEMENT CHARACTERISTICS, PRACTICES, AND PERFORMANCE IN THE SMALL SCALE MANUFACTURING ENTERPRISES: JAMAICAN MILIEU By Yacob Fisseha A DISSERTATION Submitted to Michigan State University in partial filfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics 1982 ABSTRACT MANAGEMENT CHARACTERISTICS, PRACTICES, AND PERFORMANCE IN THE SMALL SCALE MANUFACTURING ENTERPRISES: JAMAICAN MILIEU By Yacob Fisseha The level of managerial capability among proprietors in the small-scale manufacturing enterprises is an area of concern for researchers and policy makers. This dissertation examines this area by making interfirm comparison of Jamaican small-scale enterpises with respect to the level of efficiency with which available resources of labor, capital, and raw materials are utilized. The measure is commonly called technical efficiency, a name given by Michael Farrell who developed the conceptual and computational technique. The study is enriched with a rigorous examination of (a) the subsector's recent performance in employment, training, and produc- tion within the prevailing economic environment, and (b) the mana- gerial characteristics and practices portrayed by proprietors in the subsector. Flow data collected by enumerators twice a week over a year were used to analyze the technical efficiency measure. Data for the descriptive profiles of the economic scene and managerial attri- butes were collected in Jamaica by the author using a single-visit comprehensive survey questionnaire early in 1980. In all the Yacob Fisseha analysis, special emphasis is given to the tailoring and woodwork enterprise groups. Both LP and OLS techniques are used to analyze the technical efficiency achieved relative to the frontier production curve. Variables important in explaining differences in relative techni- cal efficiency are identified using the OLS technique also. The findings show that the subsector is growing in the num- ber of enterprises, the average size of the labor force and generally in the levels of product demand. It contributes 3.5 percent to GDP, employes 30,000 people, and produces 4,500 trained apprentices every year. The percentage of proprietors who follow approved management practices is, however, usually low. For example, very few pro- prietors keep adequate records, separate business and nonbusiness money or engage in marketing efforts at all. The important findings of this study is, however, the wide differences in relative technical efficiency found among firms. Firms in the subsector scored on average less than 50 percent in efficiency of what they are expected to achieve. Differences in efficiency are explained by variables such as record keeping, apprenticeship duration, marketing effort, educational background, and amount of supervision. DEDICATION To my wife Abeba and our two children Alem and Ziada. Your perseverance in love, patience, and constant encouragement made the completion of this study possible. ii ACKNOWLEDGMENTS I wish to express my thanks to the Department of Agricultural Economics, the Chairman and Graduate Committee of Michigan State University and the African-American Institute for financing my graduate program. For their guidance and assistance, I gratefully acknowledge to the members of my dissertation committee: Carl Liedholm, Lester Manderscheid, Warren Vincent, and Michael Weber. Carl Liedholm, my dissertation supervisor, made along with Warren Vincent critical contributions which greatly improved the content and refinement of this work. His constant encouragement and ever positive attitude were always inspiring. Lester Manderscheid was a source of greatly needed guidance and recommendations which were gratefully incorpo- rated. I am particularly grateful to Warren Vincent who was my major professor throughout most of my graduate career. It was extremely pleasant to work with him for his offer of friendly advice, inspiring guidance and freedom in academic program.‘ In addition to my committee, I would like to acknowledge the numerous important contributions made by Carl Eicher, Omar Davies, and Peter Schmidt. I also extend Iny gratitude to the Off-Farm Employment Research Project network of Michigan State University which financed my field research in Jamaica. I am grateful to the Institute of Social and Economic Research of the University of the West Indies in Jamaica and all the supportive staff that were associated with the research project. Finally, this work would never have been finished without the love, sacrifice, and encouragement of my family to whom it is dedicated. iv TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . . . viii LIST OF FIGURES . . . . . . . . . . . . . . . xi LIST OF APPENDICES . . . . . . . . . . . . . . xii Chapter 1. INTRODUCTION . . . . . . . . . . . . . l 1.0 Background Information . . . . . . . . l 1.1 Problem Setting . . . . . . . . . 13 1.1.1 The Economic Scene: The Last Half of the Seventies . . . . . . 14 1.1.2 The Manufacturing Sector . . . . 16 1. L 3 The Small- Scale Manufacturing Sub- sector (SSI) . . . . . 19 1.2 Objectives . . . . . . . . . 21 1.3 Dissertation Organization . . . . . . . 22 2. RESEARCH METHODOLOGY . . . . . . . . . . . 24 2.1 Introduction . . . . . . . . . . . 24 2.2 Conceptual Framework . . 24 2.2.1 Traditional Methods of Input Produc- tivity Analysis . . . . 27 2.2.2 The Farrell Method of Production Efficiency Analysis . . . . . 31 2.2.3 Extensions to the Farrell Model . . . 46 2.2.4 Empirical Applications of the Models . 58 2.3 Analytical Models Used in This Study . . . . 61 2.4 Sampling Design and Scope of Study . . . . 63 2.4.1 Overall Survey Project Design and Scope . . . . . . 64 3. THE SMALL-SCALE MANUFACTURING SUBSECTOR IN JAMAICA . 68 3.0 Introduction . . . . . 68 3.1 Contribution of the Small- Scale Nonfarm Enterprises (SSE) . . . . . . . . . . 7O Chapter 3.2 3. 3 3.1.1 Scope and Composition of the SSE 3.1.2 Contribution to Employment 3.1.3 Contribution to Worker Training. . 3.1.4 Contribution to Gross Domestic Product Recent Economic Trends 3.2.1 Changes in Product Demand Levels 3.2.2 Changes in Sizes of Enterprises . 3.2.3 Changes in the Price Structure of Key Inputs and Outputs . . . Persistent Problems in the SSI Subsector 4. MANAGEMENT CHARACTERISTICS AND BUSINESS PRACTICES . 4.0 4.1 4.2 4.3 Introduction . Characteristics of the Proprietor 4.1.1 Age and Sex Distributions . 4.1.2 Formal and Informal Training . . 4.1.3 Possible Indicators of Entrepreneurial Disposition or Capacity . Outside Influences . . 4.2.1 Family Influence and Participation . General Business Practices . . 4.3.1 Production and Marketing 4.3.2 Record Keeping 4.3.3 Financial Management Differences Among Proprietors . . 4.4.4 Financial Business Evaluation 5. RELATIVE TECHNICAL EFFICIENCY IN THE SMALL- SCALE MANUFACTURING SUBSECTOR . 5.0 5.1 5.2 5.3 Introduction . Input- Output Relationships . 5.1.1 Data Description . . 5.1.2 Choice of Production Function Model and Results . In te r- Enterprise Relative Efficiency Indices . 5.2.1 Conceptual Considerations . . . 5.2.2 The Linear Programming Approach . 5.2.3 The Corrected Ordinary Least Squares Approach . . . . . 5.2.4 Computational Results . 5.2.5 Distribution of Relative Technical Effi- ciency Indices . Sources of Differences in Technical Efficiency Performance . . 5.3.1 Tailoring: Explaining Technical Effi- ciency . . . . . . . . . vi 169 174 181 188 188 189 193 207 217 218 219 222 226 228 236 237 Chapter 5.4 5.5 6. CONCL 6.0 6.1 6.2 6.3 6.4 6.5 APPENDICES . BIBLIOGRAPHY 5.3.2 Woodworks: Explaining Technical Efficiency 5.3.3 All (SSI) Industries: .Explaining Tech- . nical Efficiency . Adjustment for Capacity Levels and Efficiency Empirical Implications and Conclusions USIONS AND PROGRAM IMPLICATIONS . Introduction . . Contributions of the SSI Subsector . . . Problems of Production Efficiency in the SSI Subsector Management Variables Affecting Firm Production . Efficiency Performance . Specific Recommendations for Program Interven- . tion . 6.4.1 Measures Relevant to All SSI Enterprises. as a Group 6.4.2 Measures Relevant for Smaller SSI Enter-. prises . . 6. 4. 3 Measures Relevant to the Larger SSI Enterprises . . . . Implications for Further Study vii Page 244 247 251 255 258 258 259 262 270 273 274 275 276 278 280 296 Table 10. 11. 12. 13. 14. 15. LIST OF TABLES Economic indicators, 1975-1980: Jamaica Important characteristics of the small-scale nonfarm enterprises in Jamiaca . . . . The employment picture in small scale enterprises (1979- 80). . . . . . . . . . . Indicators used by proprietors to estimate trend changes in demand (percent of proprietors) . Recent economic indicators of the SSI subsector (percent of enterprises) . Industrial differences in change of demand indi- cators (percent of enterprise). . Business condition indicators for 1979 Changes in number of workers and machines between year of business start and 1980 (in percent) April 1980 prices of key items compared with earlier periods (percentage changes) . . Major problems in 1979/80 (percent of proprietors) 1978 problems crosstabulated with 1979 problems (percent of proprietors). . Training background of proprietors Skil; acquisition by proprietors (percent of pr0prie- tors . . . . Training and experience profile of proprietors by industries . . . . . . . . . . Geographical mobility of the labor force (percent of respondents) . . . . . . . . . viii Page 15 71 75 83 85 87 9O 97 107 117 118 126 129 131 136 Table Page 16. Mode of business acquisition (percent of enterprise). 140 17. Entrepreneurial expectations and their fulfillment (percent of proprietors) . . . . . . . . . . 142 18. Proprietors' response to business decline (percent of proprietors) . . . . . . . . . . . . . 152 19. Job distribution of relatives (percent of proprie- tors) . . . . . . . . . . . . . . . . 157 20. Indicators of production and marketing . . . . . 162 21. Proprietors' preference and reason for mode of pro- duction (percent of pr0prietors) . . . . . . . 165 22. Seasonality in production (and sales) . . . . . 167 23. Record keeping among enterprises (percent of pro- prietors). . . . . . . . . . . . . . 171 24. The identification of enterprise costs by proprie- tors (percent of pr0prietors) . . . . . . . . 175 25. Indicators of business performance (percent of pro- prietors) . . . . . . . . . . . . . . . 183 26. Frequency of business performance analysis (percent of proprietors) . . . . . . . . . . . . . 186 27. Annual mean levels of inputs and outputs in produc- tion (no stratum weights applied) . . . . . . . 197 28. Regression results of the Cobb- -Douglas production function model . . . . . . . . . . . . 212 29. Value added regression results of the Cobb-Douglas production function model . . . . . . . . . 213 30. LP and Cobb-Douglas frontier production functions . . 223 31. Summary statistics of efficiency indices (expressed in percentages relative to the most efficient group . . . . . . . . . . . . . . . . 233 32. Simple correlation (r) between the LP efficiency indices and other varaibles . . . . . . . . . 240 ix Table Page 33. Industrial goods price indices (Base Year = 1979/80) . . . . . . . . . . . . . . . 282 34. Averages of capital stock values measured in different ways . . . . . . . . . . . . . 283 35. Value of production regression results of the Cobb-Douglas production function (including vintage indicator) . . . . . . . . . . . . . . 285 36. Value added regression results of the Cobb-Douglas production function (including vintage indicator) . 286 37. Maximum likelihood regression resuTts of the con- stant elasticity of substitution (CES) function . . 290 Figure 01 ab 0) N o o o 0 LIST OF FIGURES Farrell's Unit Isoquant Average and Marginal Production Cost Curves TAILORING Relative Efficiency Indices Histograms WOODWORKS Relative Efficiency Indices Histograms $51 (All Industries) Relative Efficiencey Indices Histograms . . . . . . . . . . xi Page 38 149 230 23l 232 LIST OF APPENDICES Appendix I. Improvised Industrial Goods Price Index . 11. Production Function Regressions Including the Vintage Indicator . . . . . . . . . . . . . . . III. Maximum Likelihood CES Regressions IV. Adjustment for Heteroscedastic Woodwork Data xii Page 281 284 287 291 CHAPTER I INTRODUCTION l.O Background Information The objective of this dissertation is to examine proprietor lcharacteristics, managerial practices, and sources and levels of management performance variations within the small-scale manufactur- ing subsector in Jamaica. Particular emphasis will be given to the relationship between business performance and differences in manage- ment characteristics and practices. The analytical tools that will be used to examine the above relationships are the usual statistical tools such as simple correlation as well as the use of nontraditional aspects of linear programming and regression models. The field survey was done at the end of the first quarter of 1980. It was financed by the small-scale off-farm employment survey network conducted by Michigan State University in a number of devel- oping countries in collaboration with host institutions. The general study on management characteristics and practices is based on a randomly selected sample of enterprises in the small-scale manufac- turing subsector. Specific and detailed analysis of business per- formance and resource efficiency levels will focus, however, on two subgroups of the sample: wearing apparel and woodwork. Through the use of the analytical tools mentioned earlier, a number of performance measures will be utilized to determine the l relative efficiency position of enterprises. Both linear program- ming (LP) and a regression model will be used to explain the inter- enterprise variations in such performance levels. Particular emphasis will be given to a measure of performance which was first developed by Farrell (1957): technical efficiency (vs. allocative or price efficiency). He used the term economic efficiency to describe the combined measure for both. Briefly, by technical efficiency (in the normative or economic sense) of a firm, "one usually means its success in producing as large as possible an output from a given set of inputs" (Farrell, l957, p. 254). A more complete description would be Pthe degree to which producers are achieving the greatest possible output given available resources and techniques" (Pachico, l980, p. l). By allocative or price efficiency, Farrell means "a measure of the extent to which a firm uses the various factors of production in the best pr0portions, in view of their prices" (1980, p. 254). In other words, allocative efficiency refers to the degree a proprietor succeeds in equating factor prices with their respec- tive marginal value products. Technical efficiency is independent of factor prices while allocative efficiency fundamentally requires specific input (and output) prices and contextual or operational assumptions such as the objective function of the proprietor. Efficiency depends mainly on the effectiveness with which inputs are organized. Therefore, the role of the manager, the decision maker and controller of the resources and economic activi- ties, is very crucial for the attainment of higher levels of efficiency. It primarily hinges on the ability and motivation of the pr0prietor to organize effective work schedules, cut waste in inputs and outputs, maintain machinery and tools, cultivate harmonious employee-employer relationship and do an effective work of marketing (see Yotopoulos and Lau, l974, p. 222). The effectiveness of any of these managerial functions will vary, of course, due to economic, social, and policy environments. Historically, economic efficiency was almost taken to be synonymous with allocative efficiency rather than a joint measure both for technical and allocative efficiencies (Pachico, 1980). In many developing countries, the government is increasingly becoming directly involved in the planning, promoting, and controlling the direction of economic development also of the private sector. In many instances, such involvement is a deliberate exercise of power based on a fundamental philosophical conviction of what the role of the state should be in the private economic pursuit of its nationals. In other instances, however, such involvement is imposed by economic constraints,such as the need to allocate scarce resources among competing ends (e.g., foreign exchange). Governmental involvement is particularly common in agriculture and in small-scale industries by setting up programs, agencies, and institutions. After pointing out the importance of the small-scale industry sector in a developing economy, both Petrof (l980) and Nelson (l980) make strong points why governments in developing countries should be actively involved in the promotion and development of small-scale nonfarm enterprises. Petrof (l980, p. 55) says, "Since most of the programs and policies to stimulate small business must be of a long-term nature, it follows that in most instances they can be pursued only through the involve- ment of the government." Nelson (l980, p. 2) adds that government can help in removing handicaps of "lack of production skills, capital, and expertise in cost control, accounting, marketing, and manage- ment." From a slightly different angle, private business investment decisions, and the economic criteria used will, as Schultz (l980) says, determine not only the rate of development but also its direc- tion and sustainability. Thus, government involvement can facilitate development efforts by encouraging economic choices consistent with long-term national development goals and criteria. The objective of this dissertation is not, however, to rationalize or advocate govern- ment involvement in economic spheres, but to try to provide some possible policy hints once it is already involves or plans to be involved in providing services such as credit, technical assistance, management advice, and economic information. Thus, from a policy point of view, it would be economically desirable to identify areas of technical inefficiency among enter- prises and try to increase their output without the need for addi- tional scarce resources. Better still, if the causes of such ineffi- ciencies could be identified, then programs and services could be provided to raise the general management level of all proprietors. Herdt and Mandac (l98l, p. 379) say that, "knowledge of what factors are responsible for technical inefficiency will improve the possibility of their removal through extension education and similar means." However, it is not easy to identify specific managerial char- acteristics and attributes that would help distinguish between poor entrepreneurs and good ones. Questions such as the following have always occupied the upper-most positions in the minds of economic development and social change students: Who are good entrepreneurs? What specific characteristics, attributes, and variables sets them apart from the rest of the population? Where or how do they get their entrepreneurial talents? What is their distribution among the population at large? And how do social, cultural, and economic environments affect their numerical magnitudes (i.e., their supply), their economic effectiveness and their pivotal role to influence, 1 Many people train, and motivate future entrepreneurial generations? have written diverse theories and models describing specific quali- ties of individuals and cultural or social environments that are conducive for entrepreneurial developments. Suffice it to say here that some have tried to identify supply of entrepreneurs with spe— cific traits (Stogdillis, 1948) or specific behaviors (DeCarlo and Lyons, l980, and Greenfield et al., l98l) exhibited by individuals. Others have tried to find conducive environments for the supply of entrepreneurs not so much on the individual as on the culture or 1Since they believe entrepreneurial talent to be inherited or innate, some would object to the idea of individuals being influ— enced, trained, or motivated by others to acquire entrepreneurial qualities. society (McClelland, l96l). Still others such as Kilby (l97l) have emphasized the rooted psychological attitudes of individuals and society as a critical determinant of the supply of good entrepreneurs, thus limiting the role of later education and training. Others such as Leibenstein (l969), Harris (l970), Papenek (l967), Leff (l979), and Schultz (l980) have convincingly argued, however, that there is no supply shortage of entrepreneurs. After presenting a lucid description of the issue, Leff (l979, p. 60) concludes, These analytical interests, however, should not divert atten- tion from an important fact: earlier theoretical concerns that lack of entrepreneurship would prove a serious barrier for economic development have turned out to be much exag- gerated. Not only was a serious identification problem overlooked, but the various reSponses we have discussed permitted the impact of entrepreneurial constraints to be relaxed at the micro and industry level.1 While it is not consistently followed, the literature makes a distinction between management and entrepreneurship (see Leff, 1979, p. 47, for a similar view). Morris (1967, p. 281) says, "The Manager 1What Leff calls identification problem and responses relax- ing entrepreneurial constraints include, l. The supply of entrepreneurship that seems to be highly elastic given favorable incentives from healthy product demarkets; 2. The emergence of a new institution, called the "Group,“ which is large—scale family concern involving extensive vertical integration of procurement, production, marketing and sometimes financing (from its own bank- ing systems) of several product lines; and 3. Government actions such as a. tariffs, and pricing and resource controls that reduce risk and raise returns; and b. the creation of public corporations that pioneer in investment ventures where private investment (domestic and foreign) was not forthcoming. is to be distinguished from the entrepreneur who introduces innova- tions and upsets routine management." DeCarlo and Lyons (1980) also raise a question as to whether creative type activities (entrepreneur- ship) are different from maintaining (i.e., management) type activi- ties.1 Thornton (l964) describes management functions as the "day- to-day execution of the main plan? while "entrepreneurship consists of conceiving the main objectives of organization and its method of Operation, assembling its resources, making the basic managerial arrangements and periodically reviewing the fundamentals" (p. 286). Thus, entrepreneurship is commonly associated with the ownership or provision of business funds for specific goals and the willingness to assume concomitant risks (Knight, l92l). While Frank Knight (l92l) articulated the relationship between risk and the entrepreneur, Schumpeter (l934) was the person who greatly expounded on the role of the entrepreneur. Schumpeter's characterization of the innovative role includes (l) introduction of a new good or service, (2) introduction of a new method of produc- tion, (3) Opening of new markets, (4) finding a new source of supply, and (5) carrying out a new organization of any industry. In short, the above list seems to say that entrepreneurship is what business— men do; however, the key word there is "new” and it is this 1They go on to suggest that the desire just to achieve may be an end itself leading to business failures which may not be due to lack of ability but due to shifting interests. Surely, such behavior must depend on the size of the current asset, the number of outside opportunities and the seriousness of a failure in the new venture. distinguishing mark which separates the innovative from the follower entrepreneur. Theodore Schultz (l980, p. 437) rejects "the idea of entre- preneurs as risk bearers and getting reward under uncertainty." In fact, his arguments of the entrepreneurial activities are opposite to those of Schumpeter. He does not subscribe, for example, to Morris' (l967) idea that an entrepreneur "introduces innovations and upsets routine management" in an equilibrium situation. Schultz argues that the entrepreneur is required only when routine activities are changed--when there is disequilibria, to bring the system back to equilibrium. If there is a "stationary economy," he says, "then it does not need entrepreneurs and in fact does not have entrepre- neurs" (p. 443). Furthermore he adds, V. . . in a stationary (static) economy there are no entrepreneurs and there is risk,“ therefore, bearing risk is not specific to entrepreneurs only (p. 44l).1 In the United States today, he counters that research and development (R & D) is carried out by the public where 70 percent of all the basic research is funded by the Federal Government and in agriculture it is 75 percent. Hence, the private sector contribution to R & D is much less than expected. The question, then, is: do entrepreneurs spur static econ- omic systems into higher levels of growth through their innovative and enterprising tendencies for new opportunities and profitable 1Contrary to what the above statements may indicate, Schultz actually endorses Frank Knight's elaboration on risk and entrepre- neurship. ventures of investment? Or do they sit back and wait until the sys- tem has been disturbed by some force (e.g., policy, law, nature, etc.) such that a dynamic situation opening new areas of business opportunities are created? In Schultz's analysis, of course, there are no special groups called entrepreneurs to sit back in a static situation--everybody is a potential entrepreneur. (And thus by implication there is no shortage of entrepreneurial supply.) Other issues that have been areas of controversy over the years are whether management abilities are innate and/or learned and whether management can be considered as one of the input factors of production in the neoclassical microeconomic analysis. Schultz (1980) strongly believes in the possibility of training or making individuals more aware of their opportunities by investing in edu- cation, health, and experience. He characterized outlays on such areas as investment in human capital. Others would argue that such management abilities, at least the entrepreneurial aspect of it, cannot be acquired through learning (see Kieruff, 1975). On a dif- ferent aspect, Johnson (1964, p. 120) says that technically speaking management is one of the "nonconventional” inputs (e.g., like tech- nological advance) which should not be "quantified and treated as factors of production." Slater (1980, p. 521), on the other hand, says that the management variable should be included in our models as "a factor (Hi production" and specifically, as a measurable input. Finally, some people make a distinction too between the returns to management and entrepreneurship. Salary is usually associated with the return for managerial service, while profit lO (Thornton, l964), or supernormal profit (Bain, l969, and Makary, l98l) and rent (Schultz, l980) are due to the entrepreneurial input. With respect to the overall role of management, there is less tendency for disagreement. Dillon (l980) lists eight definitions of (farm) management before he presents his own. He describes manage- ment as "the process by which resources and situations are manipu- lated by the farm manager in trying, with less than full information, to achieve his goals" (p. 258). Since this definition of the mana- gerial role explicitly introduces elements involving a dynamic (versus static) process, active (versus passive) manipulation, uncertain (versus certain) environment, goal (versus profit) orientation and a direct confrontation of situations and resources, Dillon thinks his definition is superior to the others he cited there. Although all of the definitions have one central theme running in all of them-- to make business decisions that are consistent with resource endow- ment and the overall objectives of the firm--Dillon's version gives proprietors more scope or responsibility to show their differential managerial capabilities and expertise in decision making. For this reason, his characterization of the managerial process is relatively more realistic and highly relevant to the thinking of policy inter- vention. The main management functions are to perform strategically the basic economic decisions of what, when, where, and how of produc- tion, finance, and marketing. In order to effectively carry out these decisions, it is commonly accepted now that the basic ll management functions Should follow a systematic problem-solving routine. This routine involves the identification of problems, searching and analyzing alternative solutions, and eventually making choices, acting upon them, and evaluating the results (Johnson, l976). Within these broad decision areas, however, some pe0ple stress cer- tain issues or facets of the managerial role than others; for example, some people's approach tend to emphasize the human aspect of it--the planning, organizing, staffing, directing, and motivating the human capital required for a successful business performance (Koontz and O'Donnell, l972). Whether one facet or issue is stressed more than others, the aim is still the same: the efficient use of resources (economic or otherwise) to attain particular ends. What is the relevance of the foregoing discussions to this dissertation and what are the implications for formulations of policy recommendations? The overall objective of this dissertation is the comparative analysis of management endowments and practices and their effect on factor efficiencies. A clear understanding of the concepts involved in the analysis is vital to the successful development of the inquiry. For example, it is important for the recommendations that may come out of it whether one is constrained by the belief that there must be certain social or cultural preconditions that need to be met before the subsector can develop. Equally, a narrow belief that individuals cannot sharpen their entrepreneurial abilities or inclinations and widen their scope of awareness through literacy, extension, general education, seminars and mass-media programs will 12 have very little to contribute to a technical assistance program. On the other hand, this is not to say that such programs are neces- sary to have a successful individual entrepreneur. For the objectives of this dissertation, the positions espoused are the following: l. The owner/operator of a small-scale manufacturing enterprise is both the entrepreneur and the manager; furthermore, at the small-scale family business level, all the managerial roles are essentially performed by this one person, the owner/operator1 (Vincent, 1962). 2. No weight is given to the view that entrepreneurs or managers are "born, not made." 3. On the other hand, while the training and education of individual proprietors are very important for the growth of the sub- sector, the existing economic, social, and political environments are also important for the participation of individuals as entre- preneurs; thus, improvements in these areas could greatly minimize the importance of the so-called shortage of entrepreneurial supply (Leff, l979). 4. With respect to the technical question of whether a management variable can be incorporated explicitly in a production function model, any Specification that will Show a differential con- tribution by management attributes pg: §g_are useful; thus, management 1The entrepreneur/manager or the owner/operator will be referred to as the pr0prietor henceforth. 13 proxy variables such as number of hours spent on specific management activities can be included in the model to get a better specifi- cation. l.l Problem Setting Jamaica has a population of 2.2 million and an area of about 4,400 square miles. It is the largest (excluding Guyana) and most populous English-speaking country in the Caribbean. Among the chief contributors to its l980 gross domestic product were manufacturing (l5 percent), agriculture (9 percent), mining and quarrying (9 percent), and tourism (5 percent).1 The importance of the tourism sector lies not so much on its percentage contribution to GDP as on the liquid foreign earnings it readily pro- vides. Due to external investment funds flow (mainly in mining and tourism), cheaper world prices for oil, relatively trained human power and a relative political stability, many of the sectors in the economy registered high rates of growth in the fifties and sixties. Girvan and colleagues (l980) point out that in the fifties foreign trade increased eightfold, while nominal GDP and per capita national income grew by about sevenfold. In the sixties GDP increased at about 6 percent per year in real terms (GOJ, l979b, p. l7).2 The 1The largest contributors were production of government ser- vices (20 percent) and distributive trade (l5 percent), i.e., whole- sale and retail (Government of Jamaica, l98ld). 2All Government of Jamaica publications will be cited hence- forth as GOJ. 14 next decade was marked, however, by a pervasive decline almost in all sectors. l.l.l The Economic Scene: The Last Half of the Seventies In the seventies, the flow of external funds continually declined either due to certain investment phases having been com- pleted (e.g., mining and tourism) or other prospective ventures were becoming less attractive to investors (e.g., manufacturing and agri- culture); the highly import-based economy was battered by the ever- rising cost of energy; and political friction, coupled with unpre- cedented political awareness and economic expectation, had created an atmosphere of instability, uncertainty, and frustration. The cumulative effect of all these contributed to a crippling shortage of foreign exchange funds, diminished domestic investment sources, created noticeable reduction in resource productivity, induced seri- ous loss of human power from the country and resulted in dangerous unemployment problems. Between l9751 and l980, real aggregate and per capita GDP fell by about l3 percent and ll percent respectively; unemployment rose from 2l percent to 27 percent; the consumer price index almost doubled and the Jamaican dollar fell by almost 50 percent in its exchange rate against the U.S. dollar (see Table l). Among the sec- tors that showed sizeable decline between l975 and l980 (at a l974 constant prices) were manufacturing (-26 percent), construction 1The l974-75 period was chosen as the base in order to make the description here comparable with the special management study survey and to avoid the unprecedented 011 pr1ce Increase 1n the early 1970s. 1!5 TABLE 1.--Econ0mic indicators, 1975-1980: Jamaica 1975-1980 Indicators 1975 1976 1977 1978 1979 1980 Net Change 1. GDP in real terms a a. In millions (J3) 2,154.7 2,012.8 1,900.3 1,973.4 1,933.6 1,848.0 -14.2 Change (:)C -o.7 -6.6 41.6 -0.3 -2.0 -5.4 429d 6. Per Capita (J3) 950.4 932.7 914.7 912.4 899.3 850.7 -10.5d Change (%)C -0.7 -4.9 -1.9 -0.3 -1.4 -5.4 -11.1 2. Contribution (1) to GDP by “productive' Sectors and Annual Change (1) in Each Sectorb a. Manuacturing (x) 18.4 18.7 17.7 16.9 16.3 14.9 -19.0 Change (x)c 2.3 -5.0 -5.3 -4.9 -s.2 -12.a -26.Id b. AgricuIture (1) 7.7 7.9 8.6 9.5 9.0 8.7 13.0 Change (1)c 1.4 -4.1 7.9 9.3 ~6.8 -7.1 -2.46 c. Mining/Quarrying (Z) ' 7.2 6.2 7.4 7.5 7.6 8.8 22.2 Change (1)c -21.4 -19.8 18.4 1.3 -1.5 11.3 -10.5d d. Construction (1) 9.8 8.4 6.7 7.0 6.7 5.2 -46.9 Change (2)c -1.3 -20.0 -20.8 3.6 -5.9 -26.5 -53.2 3. Unemployment (x) 20.9 24.2 23.5 26.0 31.1 26.8 28.2d 4. Consumer Price Index 15.6 8.1 14.1 49.4 19.8 28.6 228.2d Year-to-Year Change (%)C 5. Exchange Rate 1.10 1.10 1.10 0.72 0.61 0.56 ~49.1d (J31 8 USS __) 6. Net Foreign Reserves 56.7 -181.4 -196.0 -447.4 -758.5 -811.8 -2,238.4d (J3 million) Sources: A. All of 1980 figures, Government of Jamaica, 19810. 8. Entries 1 and 2, Government of Jamaica, 1980c. C. Entry 3, Government of Jamaica, 1981d. D. Entry 4, Government of Jamaica, 1981a. E. Entry 5, Government of Jamaica, 1979b. F. Entry 6. Girvan et al., 1980. aIn constant prices (base year - l974). bSince enterprises with a labor force of only 10 or more are covered in the Department of Statistics annual economic survey, that portion of the manufacturing contribution generated by smaller enterprises (less than 10 labor force) is probably a simple estimate. figures are usually revised in coming years. too. The official cThese are year-to-year percentage changes in the gross domestic product value of a sector at constant prices. dNet cumulative-change over the years. 16 (-53 percent), distributive trade, i.e., wholesale and retail, (~34 percent),mining (-10 percent) and transportation and communica- tion (-12 percent). On the other hand, mining had shown some revival by 1980, increasing by 4 percent and agriculture (which fell by about 2 percent) would have shown a much better result if it had not been for the floods of 1979 and the hurricane of 1980 (GOJ, l981e, p. 14). Between the two periods, the population and the labor force grew at rates of 1.3 percent and 3.5 percent, respectively. The corresponding rates of change for the percentages of the labor force employed and unemployed were -l.2 percent and 5.0 percent respec- tively. The official1 average unemployment rate for the period is 25.5 percent (see Table 1). The general conclusion to draw from the above picture is that the period was a time of great economic difficulties, which easily poisoned the political and social life of the country. It is against this economic background that the small-scale manufacturing subsector will be examined in this dissertation. 1.1.2 The Manufacturing Sector The industrial sector which is based on an import-substitution scheme was greatly promoted in the past through the provision of vari- ous industrial incentives. Some of these incentives included extended tax holidays, raw material importation under duty-free concessions and domestic market monopolies (Davies et al., 1979, and Chen-Young, 1Since the definition for labor force includes "persons not actively seeking work" (GOJ, l981b, p. 14.2), the official figures are not likely to be underestimated. 17 1967). The main objectives of the scheme under the different incen- tive laws were to reduce importation of consumer items, increase employment, and eventually produce for export under a sound indus- trial base. The incentive measures did not deliver, however, the full expected results. Although the manufacturing sector grew in real terms by about 9 percent in the fifties and by 5.5 percent in the Sixties,1 the corresponding employment growth in all sectors in the sixties was only 0.5 percent (GOJ, 1978, p. 17); part of the reason for this low growth in employment was the highly capital intensive technique of production in mining which was the major source of growth for the economy. In the seventies, the manufacturing sector was in serious trouble. Between 1975 and 1980, not only did its output decline in real terms by about one-fourth, but its share of GDP fell also from 18.5 percent to 15 percent (GOJ, l980c, p. 15, and 1981b, p. 1.9). The reason its share of GDP did not fall in proportion to its percent- age decline of output is due to the fact that other sectors, particu- larly construction and distributive trade, showed worse records (-53 percent and -34 percent, respectively). The main problem in the seventies was, of course, the shortage of foreign exchange funds to meet import demands whose cost was escalating due to the rising cost of oil and other goods. This resulted in severe restrictions on the importation of raw materials, spare parts, and equipment. 1Calculated from data provided by Girvan et a1. (1980), p. 135. 18 The inescapable outcome was high production cost, dissatisfied labor force, loss of production, and dwindling (external) markets. As a result of the poor competitive position of the larger manufacturing enterprises (due to high production costs), the unabated rural-urban drift of unskilled labor force and the desire and need to expand local expertise and productive participation, the GOJ started a few years ago to pay serious attention to the small-scale manufac- turing sector. This was emphasized in the 1978-82 five-year plan document (GOJ, 1979b, p. 47). It was pointed out there that to util- ize more domestic resources, check the rural-urban migration, spread (geographically and socially) the benefits from available employment opportunities, and to effectively exploit the low capital-labor ratio required there, the small-scale industrial or manufacturing subsector 1 The list of bene- (SSI) should be given special policy attention. ficial aspects accruing to the SSI sector may not, of course, be limited only to the points indicated above; one can also add factors such as the low human capital investment per worker, service pro- vision to the lowest economic strata of society, the development of technical and managerial skills, the social urban integration of the unskilled labor force and the creation of development linkages between the SSI and other sectors, particularly agriculture (see also Anderson and Leiserson, 1980, p. 227; a more comprehensive coverage is given in Chuta and Liedholm, 1979, pp. 2-16). 1Compared with many developing countries, the SSI in Jamaica was not really neglected. This was particularly true in the area of financial services, although there were some problems even here too (see Fisseha and Davies, 1981, p. 117). 19 1.1.3 The Small-Scale Manufactur- ing Subsector (SSI) Typical of many developing countries, information concerning both the small-scale nonfarm enterprises (SEE), in general, and small- scale manufacturing enterprises (SSI), in particular, were usually not available in Jamaica. The industrial statistical figures pro- duced by the Department of Statistics survey works usually refer to the larger manufacturing enterprises which have a labor force of 10 or more pe0p1e. Thus, they include only about 2 percent of the firms in the small-scale manufacturing subsector (SSI) which is defined as enterprises with a labor force of 25 or fewer people1 (Fisseha and Davies, 1981, Table 6; and Davies et al., 1979, Table 4). These 2 percent included in the formal surveys employ about 10 percent of the total labor force employed in the SSI. The overall share of the $51 in the total manufacturing sector employment is nearly 40 per- cent (Fisseha and Davies, 1981, p. 1).2 More will be said later on the contribution of the SSI subsector to the economy. Consistent with the problems that were highlighted earlier for the whole economy, the whole SSI (and the SSE for that matter), have had its share of rough times over the last few years (Fisseha and Davies, 1981). Even then, a high proportion (20 percent) of the 1For a complete definition of the small-scale nonfarm enter- prises (SSE) and the small-scale industrial enterprises (SS1), see Davies et al., 1979, pp. 1, 14-15. 2Since all the firms in the SSI are not included in the Depart- ment of Statistics formal survey works, their share of employment and production are probably rough estimates. 20 enterprises are recent entrants (between 1 and 2 years old) into the subsector. Also, the relative percentage share of the employed labor force in the SSI has been growing over the same period compared to that of the larger manufacturing firms. The indications are that on balance the size of the subsector must have been growing over these difficult times. Such relative growth is influenced by many factors among which is the deteriorating economic picture in manufacturing, as well as in many other sectors of the economy. Thus, the SSE, in general, and the SSI, in particu- lar, may have been serving as a catch-all reservoir of displaced labor from the rest of the economy. What is important and interesting, however, is that during this difficult period, there were small enterprises from all types. of enterprise groups that were growing or at least not declining in spite of the common constraints they were facing (see Fisseha and Davies, 1981, p. 35). Thus, one is tempted to inquire: what are the reasons that some enterprises managed to minimize or completely avoid business declines, while others, seemingly in the same situa- tion, could not? Could it be related to some internal management skills or behavior, or was it due to external factors including pure luck and circumstances beyond management's control? From a policy point of view, and particularly to a government that is trying to encourage and aid the subsector, the answers to such questions are extremely important. For, if by improving the level of management practiced, production costs could be cut, improved production techniques applied, and better marketing schemes 21 followed, then such steps could mean the difference between business failures and successes. Furthermore, if certain management prac- tices and characteristics cOuld be shown to be systematically related to general areas of efficiency or excellence, then steps could be taken to promote such practices and minimize the effects of the undesirable ones. Technical aid programs could be instituted that would improve planning or management decision-making processes and increase overall resource productivity. The aim of this disserta- tion is to contribute information towards the realization of such general objectives vis a vis the prevailing managerial attributes, practices, and behaviors whose cumulative effect shows in the over- all performance levels of each business. 1.2 Objectives The overall objective of this dissertation is part of the Small-Scale Industries (SSI) broader project research objectives.1 More specifically, the dissertation has four major objectives: 1. A description of the recent economic picture of the Jamaican SSI enterprises in order to get a more complete picture of the subsector; 2. An examination of the managerial practices and characteristics with the view of identifying those 1The important project objectives were: (1) to provide a complete descriptive profile of the small-scale manufacturing (SSI) subsector, (2) determine its significant contributions to the rest of the economy, (3) establish a benchmark of data bank against which the subsector may be analyzed in the future, and (4) identify areas of weaknesses and strengths of the subsector and accordingly submit policy recommendations consistent with available resources and pre- vailing constraints. 22 which are most important as sources of variations in relative technical efficiency measures; 3. An analysis of relative technical efficiency varia- tions among enterprises using new models of resource efficiency measures; and 4. An identification of program recommendations based on the outcome of the study. Thus, based on the above objectives, this dissertation will take a closer look at the economic picture of the SSI subsector over the last few years, examine the levels and kinds of managerial practices followed there, and analyze the inter-enterprise differ- ences with respect to the effectiveness with which resources are utilized. It will conclude by suggesting program recommendations. 1.3 Dissertation Organization The plan of presentation followed in this dissertation will be as follows: Chapter 2 will cover the review of the research methodology, giving special emphasis to comparison of resource pro- ductivity measures between the traditional approach and the Farrell method, and a historical development of extensions and empirical applications of the Farrell method. Sampling design is also elab- orated in this chapter. Chapter 3 gives an overview of the Jamaican small-scale non— farm enterprises (SSE), in general, and of the small-scale manu- facturing subsector (SSI), in particular. T0pics such as economic 23 contributions, historical growth patterns, terms of trade and per- sistent current problems of the SSI group are discussed here. Chapter 4 will present the dominant features of managerial characteristics and practices in the SSI subsector. This is a com- prehensive chapter covering crucial management variables ranging from proprietor geographical mobility to investment patterns of profit earned in the business. In Chapter 5 the identification and description of the relevant production inputs, the determination of both the "average" and the frontier production functions for each enterprise group, the construction of the relative technical efficiency indices and the relationship between these indices and cruCial management variables discussed in Chapter 4 are covered. The final chapter concludes by presenting the findings of the present study, program recommendations consistent with the findings, and suggested areas of further research both to refine and expand the present study. CHAPTER 2 RESEARCH METHODOLOGY 2.1 Introduction This chapter will relate the traditional measures of resource efficiencies, such as labor productivity, to the newer technique or what is called here the Farrell method. Extensions and applications of the Farrell method will also be presented. It concludes by describ- ing the exact analytical technique to be used here and the sampling methodology employed for the survey. 2.2 Conceptual Framework The efficiency with which resources are utilized in the pro- ductive and distributive processes has always been at the heart of economic analysis, both at the micro and macro levels. Marginal analysis, which is the main tool in neoclassical economics, has been dominant in resource utilization studies in the past, so much so that the allocative or price efficiency which it measures was sometimes used as if it indicated overall economic efficiency, i.e., including technical efficiency, see below (see Marshak and Andrews, 1944, p. 145; Pachico, 1980, p. 4). Marginal analysis usually assumes the existence of perfect competition, perfect knowledge, and perfect divisibility of inputs. In reality, however, imperfect markets exist for inputs and outputs; 24 25 there is always a world of uncertainty whose effect can only be mini- mized by incurring increasingly higher information costs; and the indivisibilities inherent in input and output sets limit one's choices of economic alternatives. Given all these common constraints (some face them to a greater extent than others), proprietors will respond differently to different economic situations. Such differences in responses and the accompanying outcomes depend on the degree of market imperfections, miscalculations of benefits and costs, and upon the desired goals to be achieved. By taking the above sources of variations to the extreme, some pe0ple have placed themselves in a position of functional immobility: Pasour (1981, p. 136), for example, claims "In terms of the perfect market norm, then, efficiency is a chimera--the entre- preneur is never efficient since he is never omniscient." This could be true if efficiency is taken to mean in the absolute sense or the best ever possible under given productive system. Cheung (1974, p. 71), on the other hand, goes to the opposite extreme by stating that under traditional economic assumptions, individuals are always efficient since "every individual is asserted to behave con- sistently with the postulates of constrained maximization," and therefore, "economic inefficiency presents a contradiction in terms. Even outright mistakes are traceable to constraints of some type." Behaving in any one of these extreme positions would assume away a number of relevant economic issues and problems. In the real world situations, people don't adhere to any of these two views. Instead, 26 they accept that the world is far from the perfect market norm and this gives the possiblity for seeing differences in business per- formances among proprietors. The imperfect norm then "calls for a shrewd and wise assessment of the realities (both present and future) within the context of which the decision must be taken" (Kinzner, 1980, p. 6). One measurement of performance differences is the efficiency with which all resources are utilized in the business. There are two approaches, as already indicated, to measure resource efficiences: technical efficiency and allocative efficiency (see Section 1.0 for full description). The latter requires input and output prices, while the former does not. In the next two sections, historical problems, developments, and unresolved issues with respect to resource efficiency (particu- larly technical efficiency) will be briefly discussed. It must be noted that resource efficiency is only one indicator of business per- formance. Although they ultimately depend on the level of efficiency with which resources are used, other performance measures, such as economic profitability or return to the business give a much more complete picture of the survivability of the firm. Still others, such as the debt-equity ratio, rate of return to investment, working capital position, and other financial ratios indicate different aspects of business performance (though in a limited sense). How- ever, they are not completely adequate for inter-firm or inter- industry comparison; furthermore, the new technique develooed by 27 Farrell can be applied on them to make them comparable across firms and across industries or between two periods of time for the same firms and industries. (In any nonidentical comparison of economic entities, the question of heterogeneity of subjects is a very crucial and so it is with the newer technique of Farrell too.) 2.2.1 Traditional Methods of Input Productivity Analysis Before Farrell (1957) came up with his overall resource effi- ciency measure consisting of technical and allocative or price effi- ciencies, people were comparing whole sectors, industries, regions, and even countries using partial productivity measures. This was done by dividing the quantity of a single factor into the total pro- duction, however it may be measured. In fact, the concept of "tech- nical" efficiency was not widely used as we know it today. With respect to allocative efficiency, however, there was no problem as to what it meant and how to apply it. As was said earlier, it was, and still is, the main tool of factor uses analysis relative to factor market prices. Schickele (1941) seems to be the first person to use the words technical efficiency, although he actually meant partial or average productivity of individual inputs: ". . . efficiency can readily be expressed in terms of input-output ratios . . . the efficiency which is measured by physical input-output ratios, I shall call 'technical' efficiency" (p. 185). (From the discussion, it is unlikely that Schickele was referring to output coefficients when he wrote "input- output" ratios, instead of output-input ratios as it is conventionally 28 done today.) Schickele had two other concepts of efficiency: 1. "Entrepreneurial" efficiency: "It refers," he said, "to the combination of productive agents devised for the purpose of maximizing income over cost in an individual firm," (p. 185); this would be what is conventionally called today returns to the house- hold firm or to the entrepreneur or just management earnings. 2. "Social-economic" efficiency which referred to "the maximization of social net product," as opposed to individual bene- fits or gains (p. 186). Other people also used the meaning of efficiency in the sense of factor productivity. For example, Johnston (1951, p. 808) says, "The term '1abor efficiency' means the ratio between the amount of available labor and the farm output." Heady and Strand (1955) used efficiency in the average productivity sense, but then recog- nized that it was unsatisfactory as long as it did not account for all agricutlural inputs: "All the Midwest [of the U.S.A.] for example, appears to be efficient relative to all the Southwest when groups are broad and the productivity ratio is value of output divided by physical units of labor or land," (p. 524); then they add, "One of the best measures of average resource productivity and effi- ciency is the relation of production to all resources used in farm- ing. . . . Aggregate productivity of all resources is measured together . . . output for each $1 annual input of labor and capital" (p. 531). Heady and Strand (1955) are not the only ones to realize the deficiency in the average productivity measures of individual factors. 29 Many others realized the weakness of such measures and attempted to correct the problem by trying to get "total productivity" through the use of index numbers by somehow weighting all inputs used (see Working, 1940; Malenbaum, 1941; Hirsh, 1943; more recently, Martin, 1956; Paglin, 1965; and Kendrick, 1961). None came with a satisfac- tory solution because the problem of aggregation or "adding up" has always made such efforts either extremely difficult or else outright inappropriate. The main cricitism against the traditional average factor productivities, such as output per unit of labor or capital, is that they are (in light of the failure to construct acceptable input indices) partial average productivity measures. Each such measure compares output with only one input at a time, without the explicit recognition of the possible changes in the other inputs. Thus, it is possible that any increase in average productivity for labor, say, could simply be brought about by substitution of more capital for it (see Chuta and Liedholm, 1979, p. 34), a process if "pushed beyond a certain point, will lower the price efficiency" of an indus- try or firm (Farrell, 1957, p. 263). In his extensive review on the construction of indices as found in Kendrick's (1961) book, Domar (1962) commented, "If efficiency is understood in the usual sense of a ratio of the actual to some potential output, or of the proximity to some Optimum, clearly the index measures neither" (p. 599). Lau and Yotopoulos (1971, p. 940) add, "The simplest-—and most naive-- measure of economic efficiency is a partial productivity index, usually that of labor although occasionally of land." 30 The danger with substituting average labor productivity meas- ures for efficiency is that an inappropriate technique of production (e.g., capital intensive) may be recommended based on such erroneous measure. (See Chuta and Liedholm, 1979, p. 34, where such measures may be usefully employed but for a different use). White (1978) says the view that high capital-labor ratios as in the DCs are also correct for the LDCs was prOpagated because "the identification of efficiency with 'productivity' (i.e., labor productivity) by many international study groups and productivity missions in the 19505 and 19605 helped to contribute to this view" (p. 30). Such a belief would naturally lead to the conclusion that "efforts to provide assistance to small- and medium-size firms are suspect on efficiency [productivity] grounds" (Bailey, 1981, p. 202). Finally, after labeling the usual dichotomy between capital-intensive and labor-intensive approaches to investment as confusing and inappropriate, Bhalla (1981, p. 19) says, What is more realistic is an optimal degree of total factor intensity . . . it is only an Optimisation of output per unit of all inputs that would lead to cost minimization . . not so much the capital intensity or labour intensity of productive operations, nor the maximization of labour productivity. It is against such reservations against factor productivity and index problems that Farrell (1957) came up with his concept of efficiency measurement which is consistent with economic theory and also free of any indexing problems. "It is the purpose of this paper," he wrote, "to provide a satisfactory measure of productive efficiency 31 --one which takes account of all inputs and yet avoids index number , problems--and to show how it can be computed in practice" (p. 253). 2.2.2 The Farrell Method of Production-Efficiency Analysis The newer technique of efficiency measurement to be discussed in this section was first developed by Farrell in his seminal article of 1957. He and Fieldhouse wrote another article in 1962 making the technique more flexible to handle more complex problems. This impor- tant analytical tool was never taken up by his followers until almost a decade later when Aigner and Chu in 1968 and Timmer in 1970 devel- oped the concept further. Over the last few years, however, the number of professional articles appearing on the subject have been quite numerous. As he pointed out himself, Farrell was not the first to come up with the idea of technical efficiency. Debreu (1951) had a similar conceptualization when he wrote his article "Coefficient of Resource Utilization." A number of people in agricultural economics also had some notion of it earlier. In his article on American agricultural efficiency, Schultz (1947) has ten years earlier (rela- tive to Farrell's) said, The concept of efficiency is applicable to different input-output relationships depending upon the conditions set by the problem. In a certain "technical" setting it may be employed to determine, for example, how to produce the most corn on an acre of land regardless of the cost of the inputs. . . . (p. 646). (Emphasis in the original.) Heady and Strand (1955) were also touching upon it very closely when they were using the amount of output value per one dollar 32 of all-input expenses. So they say “Aggregate productivity of all inputs is measured together. The method cannot indicate which resource is used in excess, however, and which one is used in too small quantities" (p. 531). Furthermore, according to Schultz, when efficiency concept was applied in farm management (i.e., in the economic or evaluative sense), then it was equivalent to the first order conditions for profit maximization and thus was identical with Farrell's allocative efficiency. It was Farrell, however, who gave the concept concrete base of theoretical validity, computational technique, and operational legitimacy by using it himself on 0.5. agriculture. In addition, Heady and Strand's approach uses the amount of output per unit of isocost value while Farrell's is the opposite, namely the amount or ratio of inputs per unit of isoquant value. The definition of production functions in neO-classical econ- omics gave Farrell the starting base for his concept of technical efficiency. The failure of many people to come up with a satisfactory measure of efficiency, according to Farrell (1957) was ". . . partly due to a pure neglect of the theoretical side of the problem" (p. 253). He adds, "When one talks about the efficiency of a firm one usually means its success in producing as large as possible an out- put from a given set of inputs" (p. 254). This is the definition of a frontier production function as opposed to an average one. Henderson and Quandt (1980), for example, say, The production function differs from the technology [i.e., all the available technical information] in that it 33 presupposes technical efficiency and states the maximum output obtainable from every possible input combinations. The best utilization of any input combination is a tech- nical, not an economic, problem“ (p. 66). Thus, the reference point for Farrell's technical efficiency measure- ment is what is possible under the prevailing or existing techniques of production, in the frontier sense. Farrell, and other people who applied the model after him, made the practical distinction between efficiency in the absolute or engineering sense and what is realizable under practical conditions. He called the former the "postulated standard of perfect efficiency" or the "theoretical function? and the latter an "empirical function based on the best results observed in practice." Farrell opted for the second concept because (1) the engineering production function (say for a plant output) would be difficult to accurately specify under variable environments, (2) the theoretical function may be an unobservable one (except being estimated from a sample of firms), and (3) the engineering or theoretical function would "likely be wildly optimistic" in light of human errors and frailties to achieve it. And he concludes, "If the measures are to be used as some sort of yardstick for judging the success of individual plants, firms, or industries, . . . it is far better to compare performances with the best actually achieved than with some unattainable ideal" (p. 255). Thus, Farrell's technical efficiency measure is a relative one, rela- tive to the best in a samplelyffirms generating an estimate of the efficient or frontier production function. 34 What is the relationship of this frontier production to the production function of the industry or the single firm? Aigner and Chu (1968) say that the frontier production function is, in fact, the industry production function: "This maximum output applied not only to the particular firm of interest; conceptually it holds for all other firsm in the industry. We might call the function so defined the industry production function" (p. 828). This industry production function is different from an industry's aggregate production func- tion which shows the relationship between aggregate output and the aggregate inputs used. Aigner and Chu also think that although the form of the production function of each firm is the same as the fron- tier (industry) production function, it is possible some of the para- meters may differ. However, such difference in the parameter values will not be such that a firm's production function lies above that for the industry. Otherwise, there would be an "obvious conflict with theory" (p. 829). Furthermore, such a possibility cannot be entertained by supposing the industry production function is, in some sense, "average." They ask, "Average in the sense of what? a con- ditional median? a mean? or, a mode? lore importantly, average .about what? about output? about some inputs? about technology? or about something else?" (p. 829). The theoretical conceptualization also on the same ground rules out appeals to such concepts as "firm of average sizel and "average technology." One can still speak, how- ever, of "output on the average" from a given set of inputs, or "average" firm production function from aggregate industry production 35 function (i.e., if firm level data are not available) or average func- tion for the industry's aggregate production function using firm level data. How about exogenous random disturbance or statistical "shocks“ that would be confounded with the (management controllable) factors affecting efficiency? Farrell himself was fully aware of this problem. However, because of the difficulty involved in the statis- tical conceptualization and analysis of the error term composed both Of a one-sided error (indicating management controllable inefficiency) and a two-sided one (indicating uncontrollable random effects), there was nothing that he could do at that time except to hope that ". . . if the errors are small compared with the variation in efficiencies, this bias will be negligible" (p. 263). 2.2.2.1 Graphical presentation of the Farrell method.-—The key in Farrell's approach is that the relative efficiency among firms can be measured by simultaneously comparing the amount of each input used to produce one unit of output. Thus, one value or index repre— senting each firm's level of relative efficiency from the use of all the relevant inputs would be generated. Farrell's method can best be described using a graph. Although any number of inputs can concep- tually be used for diagramatic presentation, only two inputs will be considered here. He made two explicit assumptions in his presenta- tion: (1) a frontier production relationship of constant returns to scale and (2) a convex downward sloping isoquant curve. He later relaxed the first assumption. 36 The logic behind Farrell's approach can be explained using a Cobb-Douglas (average) production function (although his LP result is actually a frontier one). If the production system is character- ized by two inputs and one output, then the production function is, y = F(X) (1) where, output ‘< u vector in inputs. Specifically for two inputs in a deterministic model: 1'0 y = a kal (2) where, y = output k = capital used in production 1 = labor used in production a = the Shift constant a = elasticity of output with respect to capital 1-0 = elasticity of output with respect to labor. Constant returns to scale then implies, F(nX) = a(nk)a (nill‘a ,+(.-,) kall'a an n (akall'a) = nF(X) = ny (3) 37 Thus, multiplying each of the inputs by n (where n is any positive number) results in multiplication of the output by the same magni- tude. If n = l/y, then a 1-a . Y/y = a(k/y) (1/y) = 1 un1t of output. (4) The corresponding unit isoquant is drawn in Figure 1. In a two-factor frontier production, Farrell's unit isoquant can be drawn by simply joining adjacent points of factor ratios such that the isoquant does not have a positive slope and no point lies between it and the origin. Farrell used linear programming (LP) algor- ithm to construct his unit isoquant. In a survey article on fron- tier production functions and their relationship to efficiency measurement, Forsund, Lovell and Schmidt (1980) say, PFarrell's approach is non-parametric in the sense that he simply constructs the free disposal convex hull of the observed input-output ratios by linear programming technique" (p. 9). Since the isoquant is drawn (at least in Farrell's model) from a frontier production function, only technically efficient firms (such as B and C) will have their factor ratio on the isoquant curve itself; ratios for the less efficient firms (such as E and F) will lie 38 k/y 1/y where 00' = An envelope of best or efficient unit isoquant A,B,C,E, and F = Scatter of sample firms in the input/ output ratio space k/y l/y PP' and 55' = Factor price ratios (isocost curves). Amount of capital per unit of output Amount of labor per unit of output Figure l. Farrell's Unit Isoquant. 39 on the Opposite side of the unit isoquant curve from the origin (0). No observation will be found between the origin and the unit isoquant. Technical inefficiency is then measured as follows: Firm A is technically less efficient than Firm 8, which uses the same tech- nique of production, since the former uses more of both inputs for 1 for Firm A one unit of output than B. The index of inefficiency would then be measured by the ratio OB/OA. This is to say, only OB/OA (<1) amount of each input used by A would have been necessary to produce one unit of output and thus make Firm A as technically efficient as Firm 8. Looking at it from a different angle, Firm A should have produced OA/OB (>1) amounts of the product with the level of inputs it is using. Similarly, the measure of technical inefficiency for Firm E is OC/OE. Thus, such a measure can be called the index of technical inefficiency with its values ranging from 1 for the most efficient firms (such as Firms 8 and C) to anywhere between 0 and 1 for the inefficient firms (such as A, E, and F). It is clear that to measure technical efficiency, the prices of inputs and outputs are not required. They are necessary, however, to measure allocative efficiency which is not the primary interest of this dissertation. While each firm could have a separate set of prices, Figure 1 shows common factor prices, represented by the budget line PP', faced by all firms. 1Both "index of inefficiency" and "index of efficiency" will be used in the discussion depending on the emphasis given to the measurement in the context. 4O Allocative efficiency implies that factors of production are used up to the point where their market prices (including any inci- dental expenses) are equal to the value of their marginal products and each dollar of factor expense on any one input brings in the same returns in production as in any other input. These two conditions imply that the ratio of factor prices is equal to their marginal pro- ductivity ratio which is the same as the slope of the isoquant curve 00'. This equality would hold along the ray OE for any parallel movement of the isoquant as long as the two factor prices did not change or if they change, they do so in equal proportions. Therefore, any firm that lies on the line 0E would be allo- catively efficient--but only if it falls on the isoquant curve would it be technically efficient as well. Hence, Firms C and E are allo- catively efficient but of the two, only Firm C is technically effi- cient. By this criteria, Firms A, B, and F are allocatively inefficient. The index of allocative inefficiency for B (and for A) is given by 00/08. In other words, by not combining resources in the proper proportion, B is incurring a higher private cost of production per unit of output as represented by the broken budget line 53' than is necessary as shown by the budget line PP'. It fails to fulfill the first order conditions for profit maximization or cost manimiza- tion, although it is technically efficient. With respect to E, the case is the reverse (a parallel movement of both the isocost and the isoquant would still result in a point of tangency at E). Firms 41 such as A and F are neither technically nor allocatively efficient. Thus, technical efficiency is independent not only of the factor prices faced, but of the factor proportion used. Finally, Farrell measures economic efficiency (which in his approach is the product of technical and allocative efficiencies) as 00/08 for 8 (since 08/08 = l) and OD/OA for A since OB/OA X 00/08 = OD/OA. (5) It is not necessary that an observation such as 8 actually exist in order to measure the efficiency levels such as for A. Any firm can be compared with its corresponding hypothetical efficient firm that has the same technique of production or factor ratio. 2.2.2.2 Advantages and disadvantages of the model.--Compared with the traditional methods of "efficiency" (productivity) analysis, the Farrell technique is consistent with the concept of production function or frontier; it accounts for "all" inputs used in production without the indexing number problem; it can accommodate firms using "different" production techniques; and compared even with other models based on it, the Farrell method has the advantage that it does not require an explicit specification of the functional form for the frontier production. The major disadvantage of the technique lies in the fact that the frontier unit isoquant is determined by a sub-set of supporting observations in the sample. These sub-set of observations are the ones that lie on the unit isoquant itself. This makes the technique 42 particularly susceptible to extreme observations and measurement errors which will bias the isoquant Optimistically. Collecting more observations will not improve it; it would be like collecting more observations "to make" a sample range value narrower. The second disadvantage of the model is the restrictive assumption of constant returns to scale, although it can be handled using cumbersome calcu- lations (Forsund et al., 1980, p. 9).1 A third disadvantage of the model is that being a nonparametric approach, no statistical estima- tion is possible. It will be seen later that extensions to the Farrell method have taken care of some of these shortcomings. 2.2.2.3 Farrell's inefficiency index and management.--Since a firm's measure of technical inefficiency is constructed relative “u: the set of firms from which the isoquant is estimated, Farrell's technique in a way measures the efficiency of management against a realizable standard. This is possible, however, if the inputs are correctly measured both with respect to quality and quantity. As Farrell says, however, it is not possible to completely isolate the input effect from the management effect: "Thus the technical effi- ciency of a firm must always, to some extent, reflect the quality of its inputs, it is impossible to measure the efficiency of its manage- ment entirely separately from this factor" (p. 260). 1One easy way to handle this problem would be to divide the observations (if there are enough of them) into a number of size groups and then compare a group's production frontier with another group's (Farrell, 1957, p. 259). 43 Other writers too (see, for example, Page, 1980, and Pachico, 1980) equate management efficiency with technical efficiency. Pachico (1980, p. 4) says, "Differences among firms in their abili- ties to be technically efficient are essentially differences in management." Page (1980, p. 319) speaks of his objective as being to “clarify the relationship between technical (or managerial) effi- ciency, the choice of technique and economic performance." Shapiro and Muller (1977, p. 293) also say, "If policy makers know why some farmers are better managers (i.e., why there are technical effi- ciency differentials), they might have firmer grounds for choosing among such an array of programs." Similarly, Tyler (1979, p. 478) makes the same fundamental relationship between "firm specific measures of technical efficiency" and the "exercise of managerial capabilities." Finally, Page (1980) and Shapiro and Muller (1977) have equated Leibenstein's (1966) X-efficiency or "organizational slack" with technical efficiency. Leibenstein (1977) has maintained, however, that they are distinct concepts. His argument runs, Two underlying neoclassical notions are retained in the notion of T. E. [Technical Efficiency]: the notion of maxi- mizing decisions and the view of the firm as a unified and integrated decision making unit in the same sense in which an individual can be such a decision making unit (p. 313). The second point raised by Leibenstein is not applicable in the case of small-scale enterprises, which are usually completely run and controlled by the proprietor. The first point of maximizing decisions will be covered later in this dissertation with respect to proprietors' goals, opportunities, and constraints. 44 A much more fundamental issue is raised though by the fact that the level of inefficiency measured by the index includes other unmeasured sources of efficiency variation besides management capa- bilities (see last part of Section 2.2.2 here). For example, pro- prietors may have different stock of knowledge, they may face differ- ent techniques of production (so that the frontier function would not then be common to all),and random disturbances could cause or contribute to deviations from the frontier production function. All these are valid points and their effect will depend mainly on two factors: (1) the extent of their presence in the sample or popula- tion and the attempts made to control them, and (2) the degree of accuracy present in identifying and measuring the variables. For the purpose of this study, it is expected that there would be fewer sources of extreme variation among small-scale enterprises of the same type, such as tailoring, than, say, in larger manufacturing or in agriculture. In fact, among the enterprises picked for this study, it is unlikely that there would be great variations in the technique of production that would cause the common unit isoquant to have a significantly different shape. With respect to the differences in the stock of information or technical knowledge among proprietors, they are assumed here to be part of management's characteristics and attributes and to the extent that they can be measured, they will be used to help explain sources of inefficiency. As for the "residual" deviation being entirely associated with technical or managerial inefficiency (which was suggested by Heady (1946) ten 45 years earlier), no satisfactory way has been found to handle it in the frontier production function technique. Attempts have been made (see, for example, Afriat, 1972; Richmond, 1974; Greene, 1980a; and Aigner et al., 1977) to isolate the random effect of the residual from the systematic variation due to management; however, models that employ this conceptualization require certain (usually compu- tationally convenient) assumptions about the residual or one-sided error term. Since such assumptions may not reflect the underlying distribution, they are not fully satisfactory. A more acceptable alternative is to minimize measurement errors and to control (or incorporate) as many important exogeneous variables as possible in the model. In this, the technique is no different from many stochas- tic models used in economic analysis. Pachico (1980, p. 8) also says, "Hence, the very usefulness of frontier production function analysis is to identify first, the best practice firms, and secondly, what characterizes them as a group." And the more refined data one uses, the closer one gets to making exact indexing of individual differences from the frontier function. Finally, with respect to the allocative or price efficiency of the firm, it was said earlier that it was the main tool of analy- sis for "economic" efficiency in the past. Farrell says that not only is price efficiency estimation very complex, but its use is also much limited. The reason is, of course, that it is very difficult to discover the exact prices of inputs. Even when discovered, it is very likely that they may be related to some future prices. A firm's price 46 efficiency will "provide a good measure of its efficiency in adapting to factor prices only in a completely static situation" (Farrell, 1957, p. 261). Further, he adds, "Thus price efficiency is a measure that is both unstable and dubious of interpretation; its value lies in leaving technical efficiency free of these faults, rather than in any intrinsic usefulness" (p. 261). Just as Leibenstein (1966) has argued that the loss to society from X-efficiency is more than from misallocation of resources, both Timmer (1970) and Pachico (l980) think that there is a greater wast- age of resources from technical inefficiency than from allocative inefficiency. Noting the fact that technical inefficiency has received much less theoretical attention in the economic literature, Timmer (1970, p. 99) says that it is, relative to allocative ineffi- ciency, "potentially more important quantitatively (in terms of wasted resources)." Hence, due to its relatively limited potential importance, the problem of getting correct prices and problems of interpreting the results, no attention is given to allocative ineffi- ciency in this study. 2.2.3 Extensions to the Farrell Model While employing the basic model outlined by Farrell, many people have extended the technique to accommodate some of the points raised in the last section. Most of these extensions are summarized in a survey article on frontier production functions by Forsund, Lovell, and Schmidt (1980). The most important ones will be summar- ized briefly here. 47 As Forsund et a1. (1980) said, frontier production studies can be classified according to the way the frontier or function is specified or estimated: (1) Parametric or nonparametric function of inputs, (2) deterministic or random fronteir function, and (3) sta- tistical frontiers; finally, two other approaches (develOped by Lau and Yotopoulos, 1971, and Toda, 1976, 1977) which do not require the frontier approach will be briefly mentioned. Farrell's original model is deterministic and nonparametric. The frontier was completely and without the disturbance term deter- mined by the best observed firms and there was no need to estimate parameters. 2.2.3.1 Deterministic parametric models.--Aigner and Chu (1968) extended Farrell's model by making it parametric but at the expense of Specifying a functional form. Timmer (1970) introduced a probablistic element into their model. Aigner and Chu's basic model is cast in a homogeneous Cobb-Douglas production function (without an error or disturbance term). For a model with two inputs and one out- put, the functional form would be (note the number of inputs can be as many as one wants): 48 0‘1 0(2 where, y = output k = capital used 1 = labor used 0 = shift constant 01 = output elasticity with respect to capital 02 = output elasticity with respect to labor. Converting the model into log linear: 1n (y) =1n(oco) + C11 1n(k) + 021n(1) (7) or, v = A + 01K + 02L (8) where, 1n = natural logarithm operator Y = log value of average production function K,L = log values of capital and labor respectively A = log value of 0 . 0 The model is then converted into a frontier production (and designated by 9) by requiring that all n observations (yi's) lie on or below the efficient or frontier production function (yi's) such that for each enterprise i, A+01K1+02L1=Y13Yi i=1,...,n (9) or, A + alKi + azLi - Ui = Yi - U, = Yi (10) and, U1 3_O where, Y = the technically efficient production level Y. = actual or observed output U. = the difference between the frontier and the observed (log) values. Since the aim is to have all the observations lie on or below the frontier function, an infinite number of values for A, <01 and 02 will satisfy the equation, thus producing a large number of unreal- istic and impossible frontier functions. So, consistent with Farrell's model that firm efficiency be judged against the best in the sample, the values of the parameters are limited to those values that make the frontier curve closer to the observed values by requiring that the sum of the (positive) errors (ZU) be minimum.1 1In order to make the outcome (or values of the parameters) somewhat comparable with the results of an OLS average production function, one could have minimized U2 (using quadratic programming), but the squared errors would accentuate any extreme values or errors in measurement (Timmer, 1971). 50 The programming problem then becomes: Minimize EU (11) Subject to A + 01K + 02L 3_Y (12) and A, 01, 02 3_O. (13) (For ease of writing, the i's have been dropped. Note also that the parameters are the unknowns here.) Summing equation 10 over all n enterprises and solving for EU: 2 A + 01 2 K + 022 L - 2U = ZY (14) 2 U = Z A + 012 K + 02 Z L - ZY (15) Since ZY is constant for any given sample, it can be dropped from the equation without affecting the minimum value of EU; any set of 01 that minimizes EU for any constant will do so for any other constant including zero (see Timmer, 1971, p. 780). After dividing and expanding Equation 15 by the number of enterprises, n, the full model becomes,1 1This LP model represents an hypothetical firm (sample-Firm) with three activities (the unknown parameters) n constraints (observed outputs from n firms) and the nonnegativit constraints (for the parameters). The known coefficients (C.'s) in the objective function are the sample means of capital and labOr. Minimize A + 01 R + oz'L (16) Subject to A + 01 Kn + 02 Ln 3_ Y (17) and, A,OL1, 0‘2 _>_ 0 (18) The model can then be solved using simple LP model and once A. 01, and 02 have been estimated, they can be used to compute the attainable maximum output Y for any given set of K and L of an enter- prise. The observed Y is compared with the predicted frontier value Y. The index of efficiency for the enterprise is then the ratio between the actual Y and the predicted Y or Y/Y which will always be between 0 and 1 inclusive: 0 §_Y/Y 5_1. (Except for the conven- ient form of the ratio lying between 0 and 1, there is no theo-- retical problem why the one-sided disturbance term U cannot be used as a measure of efficiency instead of the ratio.) All efficient enterprises will have a ratio of 1 or U equal to 0. Depending on the sizes or levels of inefficiency, the rest will have ratios less than one. 52 The advantage of this parametric LP model compared with Farrell's is that it uses a simple mathematical form to determine the frontier (unless one has only two inputs in which case the graphi- cal approach could be used); also it accommodates nonconstant returns to scale, i.e., 01 + 02 is not required to add up to 1. The dis- advantage of the model is that the number of observations that can fall on the frontier function are limited by the number of para- meters to be estimated. Forsund et a1. (1980) say, ". . . there will in general be only as many technically efficient observations as there are parameters to be estimated" (p. 10). Furthermore, there can be no statistical estimation made on the parameters as there is no distribution assumed for U. 2.2.3.2 Deterministic statistical frontiers.--Aigner and Chu's model of Equations 6 and 10 can be written as: 01 02 -U y=akl e (19) where, U a one-sided error term e base of natural logarithm All other variables are as given earlier. To convert this into a statistical form, all that is required would be to make an assumption about the distributions of U and between U and the inputs. 53 When the model is converted into a log linear, it becomes: ln(y) = ln(ao) + 01 ln(k) + 02 ln(l) - U (20) or as before, .< l - A + 01 K + 02 L ' U (21) where, U C | v < 1. O and thus 0 §_e' By assuming that the observations on U are independently and identically distributed (iid) and that the inputs are independent of U (i.e., exogenous), many peOple have specified a number of dis- U). tributions for U (and thus for e' Afriat (1972) assumed a two- ” and suggested a maximum likeli- parameter beta distribution for e- hood method for estimation. Richmond (1974) showed that this amounts to a gamma distribution for U. Schmidt (1976) has also shown that Aigner and Chu's LP model would be maximum likelihood if U has exponential distribution. The usefulness of such an approach is that the efficiency indices (or e'U) can then have statistical prOperties such as mean and variance. The disadvantage with the model is that the maximum likelihood estimation of the parameters (that determine the functional form of the distribution) depend on what distribution is assumed for U. Forsund et a1. (1980) say, YThis is a problem because there do not appear to be good a priori arguments for any particular distribu- tion" (p. 11). Furthermore, unless one specifies a particular 54 statistical distribution for U (e.g., a gamma distribution as Greene, 1980a.did), Schmidt says that the maximum likelihood estimates will not be consistent and asymptotically efficient. But then, if such a choice is not based on the underlying distribution of the data, it becomes a convenient, if not an arbitrary, choice just for statisti- cal convenience. 2.2.3.3 Corrected ordinary least squares (COLS).--In this model, which was first suggested by Richmond (1974), the constant or intercept term is "corrected" by the mean of the disturbance error, U. If'U is the man of U, the model in the last section would become, Y=A-U+01K+02L-(U-U) (22) where, U still satisfies U 3_0 and the mean of the new error term (U-U) = O. The new error term will satisfy all the necessary conditions, except normality. Ordinary least squares (OLS) can now be used to get consistent estimates of (A-U), 01 and<12. Given the right dis- tribution for U, the value of U can be estimated from higher order (second, third, etc.) central moments of U. One possible outcome or disadvantage of this approach is that some of the observations will fall above the "frontier“ curve and this is not only contrary to the initial conceptualization, it will 55 also give an inconvenient base for making individual technical effi- ciency indices. Furthermore, the size of the correction factor or mean of U, U, depends on what distribution is assumed for U. For example, the mean of a one-parameter gamma distribution is equal to the variance of the distribution; hence, the corrector factor is the variance of the distribution. However, if an exponential distribu- tion is assumed for the same observations, then the mean is the positive square root of the variance of the observations. Thus, one gets two totally different corrector factors for the same set of observations (unless the variance is equal to 1). One simple approach, suggested and proved to provide con- sistent estimate of "A" by Gabrielson (1975) and Greene (l980a), is not to correct the constant term as above, but to shift it up such that all the observations are below the frontier and one is on it. Thus, for the average production function, the function is raised to a frontier one by raising "A" such that only one observation lies on it and the remaining are below it. That is, all the residuals are greater than zero, but one is zero. A second approach (suggested by Aigner et al., 1977) to handle the stochastic frontier problem would be to consider the residual U to be composed of two parts: (1) symmetric or random elements that are beyond the control of the proprietor affecting U, and (2) systematic or one-sided effects (inefficiency) which are under the control of the pr0prietor. The old model of Equation 19 would be written now as: 56 y = aokallazeW-E) (23) or in log linear, Y = A + a] K + 02 L + (V-E). (24) The frontier would then be, Y = A + a] K + 02 L + V (25) where, V = a random element with symmetric distribution E = "efficiency disturbance" or a one-sided effect, showing technical inefficiency. In this format observations can fall above or below the frontier production function. However, there is no way to decompose or isolate that part of U due to firm specific technical inefficiency and that due to random effects beyond management's control. Aigner et a1. (1977) say, ". . . it is not possible to decompose individual residuals into their two components, and so it is not possible to estimate technical inefficiency by observation" (p. 14). It is useful though, to estimate mean inefficiency over the sample. For that purpose, either maximum likelihood or the corrected ordinary least squares (COLS) can be used to estimate the parameters. In any case, the distribution of U must be specified to make the necessary statis- tical statements. 57 2.2.3.4 Nonfrontier models.--Finally, there are two models that do not use the frontier approach (i.e., they don't force a one-sided error) which can be used to compare simultaneously both technical and price efficiency between groups of firms. The ineffi- ciencies can also be individually compared between the groups. The first one which uses the constrained profit function was developed by Lau and Yotopoulos (1971) and Yotopoulos and Lau (1973). It can be used to test equal price efficiency, equal technical effi- ciency, and equal economic efficiency between groups of firms. The main problem here is that it cannot be used to test firm-by-firm efficiencies and also it requires detailed input prices, even for the technical efficient check. Furthermore, since it uses a con- strgined profit function, the production function must be of the Cobb- Douglas type in order to derive such a profit function. Forsund et a1. (1980) say, "This practically restricts use of the model to a homogeneous Cobb-Douglas specification" (p. 18). The second model was developed by Toda (1976, 1977) and uses a cost function. Its main use is to check price efficiency. It is not restricted to any functional form as the Lau and YotOpoulos model. Again, this model is ill-equipped to handle firm-to-firm comparison. In conclusion, the models that seem to have less conceptual, computational, and functional limitations are the original technique developed by Farrell, the extension to it of a parametric format developed by Aigner and Chu and the corrected ordinary least squares (COLS). This is particularly true if the comparison is made on 58 firm-by-firm basis. For comparing firms in the same industry (e.g., small vs. large ones), the Lau and Yotopoulos model and the Toda model seem well suited. Timmer's probabilistic modification is useful too. 2.2.4 Empirical Applications of the Models A number of studies have employed either the basic Farrell model (either modified ala_Aigner-Chu or Timmer or as it is) or the constrained profit function of Lau and Yotopoulos. Some of the most relevant ones will be briefly mentioned here. For the sake of com- pleteness, Toda's application of his own model, the cost function, will also be mentioned. All the studies that will be mentioned here refer to the efficiency or frontier production function. Empirical studies deal- ing with functional models attempting to explain sources of ineffi- ciencies will be mentioned under relevant sections later in Chapter 5. The distribution of the applications mentioned here between agriculture and nonagricultural firms is about the same. Except Page's (1980) study in Ghana, none of the studies deals with small- scale industries. Farrell (1957) himself used his model on U.S. agriculture to compare efficiency indices among 48 states. His 1950 observation on each state consisted of aggregate outputs and inputs and thus each state was a "farm firm." He tested the inclusion of a number of inputs and their explanatory power of the residual from the frontier. He found substantial differences in efficiencies among the states. 59 Aigner and Chu (1968) modified Farrell's model, as already discussed earlier, and used Hildebrand and Liu's (1965) 1957-1958 data on the U.S. primary metal industry. Again, their 28 observa- tions were state aggregations. Although their main interest was to develop a model, rather than test one empirically, they found that good use of capital was highly associated with "good" management. Timmer (1971) introduced a subjective probability element into the Aigner-Chu model and applied again to agriculture in the 48 states. In order to introduce the probability element, he discarded (a few at a time) the most efficient firms until the model was stable with respect to the values of the parameters. His 384 observations consisted of 48 cross-sectional contiguous state values of 8 years (1960-67). He found that firms (states) at the frontier were more capital intensive and hence, the marginal productivity of labor was higher there than elsewhere. Also, the frontier firms were technically efficient, but less price efficient compared with the nonfrontier ones. His final conclusion is that 75 percent of the states were within 10 percent of the efficient or frontier production function levels. Tyler (1979) used both the original Farrell model and the parametric Aigner and Chu (1968) modification. Using 1971 data, he examined both Brazilian plastic and steel industries. He had 16 cross-sectional observation for the first group and 22 for the second. He found widespread inefficiencies in both industries: for example, the Farrell model showed that 50 percent of the plastic 60 firms were below 70 percent of the efficient frontier; the Aigner- Chu model also showed about three-fourths of the firms in this industry were below 60 percent of the frontier production level. Using 1971-73 data, Page (1980) examined efficiency indices and other topics on three industrial groups in Ghana: logging, sawmilling, and furniture manufacturing. He used the Aigner-Chu model as modified by Timmer (i.e., by discarding efficient observa- tions until the model [parameters] were stable). He discovered that on the average, firms were achieving 70 percent of the predicted frontier level. Furthermore, he found that in all the 28 logging, 36 sawmilling, and 11 furniture manufacturing firms, frontier firms showed greater capital productivity than average (OLS results) firms. He concludes that this shows ". . . improvements in technical effi- ciency are capital augmenting." Two authors used the contrained profit function (nonfrontier approach) of Lau and Yotopoulos (1971). Using 1977 cross-section farm data, Leddin (1980) used it in Ireland to compare 23 small-sized and 26 medium-sized farms. He found that both were price efficient, but the medium-sized farms were more technically efficient. Trosper (1978) also used the model on American Indian ranch farms compared with non-Indian ranchers. He used 1967 data for 43 ranches. He concluded that there was no difference in technical efficiency between the two groups. As was indicated earlier, Lau and Yotopoulos' model which they applied on agriculture in India compares efficiency between groups and not between individual farms. 61 Finally, Toda (1976), who developed the cost function, applied it to eight industrial sectors in the Soviet Union. He used 1958-71 time series aggregate data of variables. He found there were sig- nificant differences between the shadow and actual prices in each sector. He concluded from this that there were price inefficiencies in all the sectors. This was particularly so with respect to capital usage.1 2.3 Analytical Models Used in This Study Consistent with the objectives stated earlier, the models used in this dissertation will serve two purposes: (1) to measure enterprise technical efficiency performance with respect to the frontier production function, and (2) to explain the variations in efficiency performance among enterprises. Altogether, two types of models will be used to construct two frontier functions: (1) the linear programming model, as elaborated by Farrell, Aigner and Chu, and Timmer; and (2) the corrected or shifted ordinary least squares (COLS) technique. The need for more than one curve is necessary both for comparative purposes and because the COLS has statistical properties that one can make some hypothesis testing of the para- meters. For purpose of comparing frontier firms with those on the 1There are a number of other studies that used the frontier approach to examine efficiency differences among firms and the degree to which such differences are a function of certain management vari- ables. However, because it was felt that the above review is ade- quate for a sample and because some of the studies were conducted on specialized firms, e.g., regulated firms, they have not been included here. 62 average production function, one can simply use the results of the Linear Programming (LP) model and compare it to the results of the Ordinary Least Squares. This will help to compare the factor margi— nal productivities of firms at the frontier with those represented by the average production function or the OLS curve. The variables used and their measurements are fully described in Chapter 5. It will be noted there that the effect of particularly two variables on the production functions will be examined: (1) the weighted average age of an enterprise equipment, and (2) the level or capacity utilized of fixed capital. The former is based on the assumption that newer capital equipment is more productive either because it embodies new technological progress or the older machines may be less efficient due to simple wear and tear. To account for this, each firm will have a single vintage indicator variable con- structed from the age and purchase price of each equipment. Adjust- ing for the level of capital equipment utilization is based on the rationale that since efficiency relates achieved production to avail- able resources, firms producing at substantially less than full capacity will appear less efficient than those which are producing at higher levels of capacity. This will be so only if the over- capitalization initially resulted from nondemand related factors, (e.g., hoarding of capital to beat inflation). Another reason to use capacity adjusted capital is to treat it equally with labor where only the actual labor flow has been included, thus excluding idle and holiday hours. Such an approach will avoid the underesti- mation of factor productivities. However, since under capacity 63 production can result from inefficient management, a regression model with capacity unadjusted capital will also be checked. 2.4 Sampling Design and Scope of Study The small-scale nonfarm enterprise (SSE)1 survey project in Jamaica was a collaborative effort between the University of the West Indies (U.W.I.) and Michigan State University. It was spon- sored by the Small Enterprise Development Corporation (SEDCO) which was changed into the Small Industrial Finance Company (SIFCO) late in 1980. The University of the West Indies (U.W.I.) participated through one of its bodies the Institute of Social and Economic Research (I.S.E.R.) where the project was housed. Michigan State University participation included on field personnel and computer data analysis in East Lansing. The survey field work started in late August 1978 and ended May 1980. The survey project was divided into three main phases, plus some special studies. The objectives of the three phases are given in the first two project reports (Davies et al., 1979; and Fisseha and Davies, 1981). A brief summary of each will be pre- sented here. 1The definition of an SSE employed in the study is those enterprises that employ 25 people or fewer including the pr0prietor and household members. Thus, a better word to use would be a work force or labor force of 25 or fewer. This definition excludes enterprises in transport, hotel, and higgling activities; it also excludes chain stores (see Davies et al., 1979, p. l). 64 2.4.1 Overall Survengroject Design and Sc0pe The aim of the first phase of the project (Phase I) was both to identify and describe the small-scale nonfarm enterprises (SSE) in Jamaica and to prepare a sampling frame for subsequent studies. It described the number, composition, location, and size distribution of these enterprises, as well as additional information on the size and composition of labor force, the number and kind of machinery used and the workshop structure housing these enterprises. Close to 9,500 enterprises (3,500 manufacturing and 5,900 nonmanufacturing, i.e., distribution and services) were contacted during this phase. The sample design used was a two-stage stratified sampling. The first stage of stratification was at the parish (i.e., administrative area or province) level; thus, including the metropolitan Kingston, the whole country of 14 parishes was covered. The second level of stratification was the population size distribu- tion of cities, towns, and localities within these parishes. There were four population size strata.1 (These strata are sometimes referred to as locations here and in the project reports.) 1The general sampling methodology is given in Davies et al., 1979, p. 9-13. The pOpulation size strata or locations consisted of the following: 1 Greater than 100,000 (Kingston only) 2 20,000 - 100,000 (the Major Towns: Montego Bay, Spanish Town, and May Pen) 3. 2,000 - 20,000 (the Smaller Towns: about 60 rural towns) 4 2,000 or below (about 2,250 rural localities or Enumera- tion Districts called here EDS). For the purposes of the 1960 and 1970 population census surveys, Jamaica was divided by the Department of Statistics into such 65 The percentage of sampling coverage for each stratum is as follows: 100 percent coverage of the areas in the first two strata; 50 percent coverage of the third stratum consisting of 60 smaller or rural towns; and 4 percent from the last stratum which had about 2,250 enumeration districts or EDS. For all the localities that fell in the sample, a street-by-street canvassing of areas was conducted to complete the Phase I questionnaire. The compiled list of enterprises was the sampling frame for the subsequent Phase 11 survey.1 The rest of the information has been reported in Davies et a1. (1979). The aim of the Phase 11 part of the survey was to describe the socioeconomic characteristics and constraints of the small-scale manufacturing subsector (SSI). Data were collected for a sample size of 710 enterprises randomly selected from the list compiled in Phase I.2 The main tOpics covered in this phase were descriptions of the proprietors and the enterprises, identification, and classifi- cation of major problems faced by the subsector and some explanation on managerial practices and characteristics. This phase was a single visit (one-shot) survey and the idea was to collect policy-oriented Enumeration Districts (EDS). The boundaries and physical sizes of these EDS were clearly defined in special maps which were used in the survey. 1As already indicated, close to 9,500 enterprises were enum- erated and described; to do this, about 25 enumerators and four field supervisors were directly involved. 2Because it was assumed there would be problems of business closures (failure), site changes, refusals, migration, and even wrong addresses, an initial sample size close to 1,000 was picked. A weighting procedure among strata and among enterprises within stratum was used to pick the sample. 66 data in a shorter time. The report for this Phase has been already completed (Fisseha and Davies, 1981). The third phase dealt with the collection of flow data on inputs and outputs for 13 months. Data for the first month were dis- carded as it would have been less reliable during this intial learn- ing and adjustment stage. Close to 300 enterprises were selected for this Phase from the Phase II respondents in a similar procedure as employed in the Phase 11 survey; by the end of this twice-a-week visit of longitudinal study, the number of respondents with adequate amount of data was close to 200. Out of the Phase II reSpondents, a sample of 80 was randomly selected for the management study of this dissertation. A two-to- three hour management questionnaire was administered by the author on each enterprise at the end of its flow type study (the Phase III) in April and May, 1980. Thus, the detailed analysis on management practices and characteristics and a historical profile of the SSI presented in the first four chapters will employ the data from these 80 respondents. The relative business efficiency analysis in Chapter 5 employs information both from the flow data for the construction of the frontier production functions and from the one-shot management questionnaire to analyze causes of inefficiency. The number of cases used to construct the frontier production functions for tailor- ing (i.e., tailors and dressmakers) and woodworks are 50 and 29 67 respectively. Garment manufacturing from tailoring and lumber pro- duction from woodworks are excluded in the analysis.1 Since it is generally believed that memory recall could be a problem with respondents, data for the flow questionnaire were collected twice a week by asking respondents for the previous three or four days only.2 Either on respondents expressed preference or for administrative purposes, some respondents were visited only once a week and they were also asked for the previous three or four days only; during the data cleaning and preparation process, the data base for this particular group were updated to reflect full weekly flow of inputs and outputs. Similarly, data for missing half weeks for both groups of respondents were adjusted by using an enter- prise's own half-week mean values for a specific yearly quarter within which the missing period happened to fall. 1It was not possible to keep distinct the functions of the cabinet maker and the carpenter in woodworks; the majority in the sample are, however, mainly cabinet makers. 2One of the reasons for collecting the single visit manage- ment survey data at the end of the Phase III survey (the twice-a- week visit) was to see how annual values for certain variables com- pare under the two systems of data collection. The general conclu- sion is that the one-shot management survey tended (1) to underesti- mate moderaly the annual values for firm labor hours and value of production and (2) to overestimate grossly all expenses. However, only 55 percent of the proprietors were able to give the complete information (i.e., both expenses and income). The rest (ranging from 15 percent in the urban areas to 51 percent in the rural) said they cannot supply the information, i.e., they can't tell. CHAPTER 3 THE SMALL-SCALE MANUFACTURING SUBSECTOR IN JAMAICA 3.0 Introduction This chapter will deal with three main topics: (1) a review of the static environment wihin the small—scale nonfarm enterprises (SSE), i.e., including those that are nonmanufacturing enterprises;1 (2) an examination of the dynamic changes that have been taking place over the years in the small-scale manufacturing enterprises (SSI); and (3) a description of the persistent problems that have been hindering production and growth in the subsector. These topics will be analyzed both at the locational (strata) and industrial (enterprise group) levels. Emphasis will be placed on the garment and woodwork industries since the model used in Chapter 5 is applied to the them. In this study garmet refers to tailoring and dressmaking only and will be referred also as wearing apparel (exclusive of footwear). The static descriptions will specifically deal with (1) the size, type composition, and geographical spread of the small-scale nonfarm enterprises (SSE), and (2) the contributions to employment, 1The small-scale nonfarm enterprises was defined earlier as those that employ 25 people or fewer. This definition does not include enterprises involved in transport activities, hotels, higgling, and chain stores (whose combined employment exceeds 25). 68 69 worker training and production particularly by the small-scale manufacturing enterprises ($51), a subdivision of the SSE. In describing the static characteristics of the SSE, a brief review will also be made of the salient findings observed in the first two project surveys (Davies, et al., 1979, and Fisseha and Davies, 1981). Such brief review of the SSE and particularly of the SSI from the earlier findings (surveys) will hopefully make the description of their static characteristics more meaningful and complete. The dis- cussion will sometimes be cast in an urban-rural1 dichotomy; and some parameters derived from the SSE and SSI groupings will be compared with those found in the large-scale establishments. The dynamic changes that will be discussed in this chapter deal with the global or industry demand (mainly during the last half of the 1970), labor force size, number and composition of machinery and the market prices of key inputs and outputs. Inasmuch as change in industry demand may be reflected in the changes of either the number of firms in the industry or the output size of individual enterprises, these two indicators will also be fully discussed. Finally, certain problems associated with production, market- ing, and employment will be examined and their interactions noted. In all of these discussions, emphasis at the enterprise group level will be given to wearing apparel and woodwork; this will minimize the amount of review necessary on these industries in Chapter 5. 1Rural is used here according to U.N. definition of locali— ties with population size of 20,000 or fewer. 70 3.1 Contribution of the Small Scale Nonfarm Enterprises (SSE) The first section here starts by discussing the size, sc0pe, and composition of the small-scale nonfarm enterprises (SSE) (i.e., including those that are in the nonmanufacturing subsector). This will be followed by describing the employment1 and training contribu- tions of the SSE and the SSI subsector. Finally, the economic con- tribution (i.e., to GDP) of small-scale manufacturing enterprises (SSI) is briefly presented. 3.1.1 Scope and Composition of the SSE There are nearly 38,000 small-scale nonfarm enterprises in Jamaica about 65 percent of which are in the nonmanufacturing group (Table 2); their combined employment of 80,000 peOple are also shared in an indentical manner between the two groups of enterprises. The overwhelming majority (96 percent) of the enterprises in both groups have a labor force size of five or fewer people. A complete percent- age distribution of all the enterprises are given in Table 1 and Appendix I of the Phase I report (Davies et al., 1979). In each of the subsectors, the majority of the enterprises are found in the rural areas or localities with population sizes of 20,000 or fewer: 81 percent and 84 percent, respectively, for manu- facturing and nonmanufacturing enterprises. Altogether, nearly 90 types of enterprises were identified during the Phase I survey. They 1The term labor force, work force, and employment of an enter- prise will always include all the peOple working in it. This includes proprietors, family workers, permanently hired, apprentices, etc. 71 TABLE 2.--Important characteristics of the small-scale nonfarm enterprises in Jamaica Subsector Grouping Both Groups Variable Manufac- Nonmanu- Jamaica turing facturing 1. Enterprise number 13,340 24,400 37,740 2. Employment 29,360 50,000 79,360 3. Employment/enterprise 2.2 2.1 2.1 4. Percentage of enterprises with work force :_5 93.8 96.9 95.8 5. Number of machines per enterprise 1.1 0.5 0.7 6. Percent of machines that are powered 51.0 93.9 78.7 7. Percent of enterprises keeping records 9.1 20.1 16.2 8. Percent of enterprises with permanent workshop 70.0 97.3 87.6 9. Percent of enterprises in rural areas 81.2 84.2 83.1 10. Percent of employment in rural areas 67.5 77.4 73.9 11. Percent of enterprises accounted by the largest two industries 72.8 85.8 78.3 SOURCE: Compiled from the Phase I report (Davies et al., 1979). 72 were later classified into nine major enterprise groups or industries (see the Phase I report, Davies et al., 1979, p. 14). Except for craft work and auto repairs (and possibly a few of the other manufacturing categories), no distinct pattern of nationwide geographical distribution exists among the SSE enterprises. Kingston naturally accounts for a very large number of the auto repairs and manufacturing enterprises. With respect to craft enter- prises, however, although a fairly large number are found in Kingston, the majority are found in the rural areas of three parishes (St. Andrew, St. Catherine, and St. Mary) and in the tourist towns of Ocho Rios and Montego Bay and their surrounding areas. At the national level of aggregation, the average number of machines per enterprise for all enterprises is less than one. This average is one for manufacturing enterprises, however. Nearly 75 percent of the machines are powered; again, there is a difference for the manufacturing enterprises where the percentage there is only about 50. On the average, about one-fourth of all the enterprises have at least one powered machine. The distribution giving rise to such an average ranges,however, from 1 percent in craft to more than 90 percent for metal work. The average number of workers per machine (whether powered or nonpowered) for all enterprises ia about two. The corresponding average for powered machines alone is four. 3.1.2 Contribution to Employment As indicated earlier, expect for the Phase 1 survey, all subsequent phases and the special studies have as their subject 73 matter only manufacturing enterprises. For this reason, the emphasis for the rest of this review will be on the manufacturing sector and particularly on the small-scale industrial or manufacturing subsector, whose acronym here is SSI. Toward the end of 1978, at the time that the project census of small-scale nonfarm enterprises (SSE) was conducted, the total labor force in Jamaica was about 940,000. This was out of a popula- tion of 2.1 million. By the end of 1980, the labor force had grown close to one million, i.e., an increase of 6.4 percent over 1978 (GOJ, 19800, p. 14.3). For these two points in time, the average rates of unemployment were 26 percent and 27 percent (GOJ, l981d, p. 3). In 1978, the manufacturing sector accounted for about one- tenth of the total labor force and had a 21 percent rate of unemploy- ment.1 Of those employed in manufacturing, a little more than 40 percent were found in the small-scale manufacturing subsector (SSI). Between 1976 and 1980, the total employment in large manufacturing establishments had been continually falling. For example, this decline in 1976/77 and in 1977/78 was 7 percent and 8 percent, respectively (Fisseha and Davies, 1981, p. 2 and GOJ, 1978a, p. 1). At the same tine the SSI subsector was growing, both in absolute and relative terms. Thus, during the 1976/77 period, its labor force grew by 12 percent (Fisseha and Davies, 1981, p. 2) improving its 1The Department of Statistics annual labor force survey includes sectoral unemployment rates too (GOJ, 1981d. p. 83). 74 relative share from 36 percent to 40 percent of the total employment in manufacturing. The average per enterprise employment figures both for the SSE and the SSI are 2.1 and 2.2, respectively. Both for the SSE and the SSI, enterprises in the urban areas employ almost twice as many people per enterprise as their counterparts in the rural areas (i.e., localities whose population is 20,000 or fewer). For example, in the urban areas the average employment for a SSI enterprise is neary four, while the corresponding figure for the rural areas is only two. About two-thirds of the SSI enterprises are one-person (the proprietor) Operations, although proprietors as a whole account for fewer than 50 percent of the labor force there. In the case of the small-scale nonfarm enterprises (SSE) as a whole, however, the corresponding percentage both for enterprises and proprietors is about the same, 50 percent. At the national level, the proportion of SSE enterprises with a labor force size between 10 and 25 is less than 1.5 percent; the corresponding proportion just for the SSI subsector is about 2 per- cent. The corresponding figures for the urban and rural SSI group- ings are 7 percent and 0.8 percent, respectively. About half of the labor force in the SSE consists of the pro- prietors or owners (in the urban areas, however, this number falls close to one-third). Permanently hired (actually "permanent" job or piece workers) and family members account respectively for about one-fourth and one-fifth of the larbor force--a large number of the family workers (26 percent) are found in the nonmanufacturing 75 TABLE 3.--The employment picture in small-scale enterprises (1979-80) Location Item Urban Rural Jamaica 1. The SSE a. Average employment per enterprise 3.3 1.9 2.1 b. Proprietors as percent of total employment 35.7 58.2 51.0 c. Hired workers (%) 43.7 13.1 24.3 d. Apprentices (%) 7.7 3.4 4.3 e. Family members (%) 12.9 25.3 20.3 2. The SSI (averages) a. Workers per enterprise 3.8 1.8 2.2 b. Labor force age (years) 29.6 36.6 33.3 c. Females in labor force (%) 14.7 43.0 32.0 d. Apprentices trained per enterprise 13.3 2.3 4.2 Source: The Phases I and II reports (Davies et al., 1979, and Fisseha and Davies, 1981) except the last entry which comes from the management study survey (1980). 76 subsector as opposed to 12 percent in the SSI subsector. Apprentices represent only 4 percent of the SSE labor force, although the per- centage goes as high as 10 percent for the SSI group. The number of workers in the SSI labor force who are hired on a "permanent" basis is small; only 22 percent of all the skilled workers are paid on time rate. The rest are paid as job workers or on the basis of piece rate. The case for the apprentices is, however, the reverse: more than 80 percent of them are paid on time rate. SSI females represent nearly half (49.3 percent) of the pro- prietors, but only about one-third (32 percent) of the labor force. Their share in the SSI labor force ranges from 14.7 percent in the urban areas to 45.0 percent in the rural localities (where more than 80 percent of the enterprises are found). The high proportion of females (it is close to 60 percent among the proprietors) in the rural areas is due to the large number of dressmakers and straw craft makers there, activities almost entirely dominated by women. The average age of the labor force (including the pr0prietors) is about 33 years. If the proprietors are excluded, the average drOps to 27 years. Except for the unskilled groups, males are gen- erally older than females. The average age for the "permanently" hired (i.e., including job or piece rate workers) is 28.5 years with a median of 25. In the case of this group, however, the mean number of years worked in an enterprise is nearly four years. The average age of apprentices is only 20 and the median is 18.3 years. They have also worked in the enterpsie for a mean period of 2.5 years (median is 2). 77 3.1.3 Contribution to Worker Training One of the important contributions by the SSI enterprises is the training of apprentices for future skilled workers and proprie- tors. It will be noted in the next chapter that more than three- fourths of the proprietors acquired their skill through some kind of participation in an apprenticeship scheme. The average duration of their apprentice training was about 21 months. The national visibility of apprentices is insignificant in the SSE: they account for only 4.3 percent of the labor force there. Their share of the labor force in the SSI, however, rises to 10 per- cent. It is even more Significant to note that each SSI has trained about four (the median is 2) apprentices on the average. Since the average age of a SSI enterprise is about 13 years, this amounts to training one apprentice every three years by each enterprise. The highest rate of apprentice training occurs in the urban areas, although the average enterprise age there is only 8 years. Among the important enterprise groups, those that show the highest rate of apprentice training are woodwork (8 apprentices per enterprise), repairs (6), metal works (6), shoemaking (5), garment (3), and craft (3). Foods has the lowest rate for apprentice train- ing with an average of less than one per firm. If the average number of months of apprenticeship for each pr0prietor in a given industry is taken as a proxy for the usual duration of apprenticeship training, then the major industries rank as follows: auto repairs--three and one-quarter years; woodwork-- 78 three full years; metal works--a little under three years (33 months); garments and shoemaking--each one and one-half years; and craft work--only two months. The policy implications of such depth and breadth of training will be discussed in the concluding chapter. 3.1.4 Contribution to Gross Domestic Product (GDP) During the survey year, the SSI subsector generated about J$l48 million to Gross Domestic Product (GDP) at purchasers prices compared to $682 million for all manufacturing sector, and $4,289,000,000 for the whole economy (see GOJ, 1981e, p. 13). Thus, the SSI subsector contributed about 3.5 percent to GDP or about 21.7 percent that of the manufacturing sector as a whole. The $51 con- tribution to GDP of 3.5 percent is quite high compared to the 2.9 per- cent contribution by Sierra Leone's SSI enterprises, particularly since the Sierra Leone study includes also enterprises that employ between 25 and 50 people, compared to the Jamaican study with a labor force of 25 or fewer only (see Chuta, 1977, p. 50).1 Given their large number, the contribution by the Jamaican SSI enterprises to GDP is relatively modest. This is not surprising 1For example, if enterprises with employment sizes of up to 50 had been studies, it is likely that the contribution to GDP by the Jamaican SSI would have risen to about 30 to 35 percent. This would have been impressive given that the level of industrial development in Jamaica is quite high. 79 since the average annual value of production in the subsector is about J$10,000. In fact, about 45 percent of the SSI contribution comes from about 800 firms (only 5 percent of the total) with a gross annual value of production ranging between J$50,000 and $325,000. About 45 percent of the SSI contributions also comes from the rural areas--the EDs along contribute about a third of the total. Of that contributed by the urban enterprises, about 30 per- cent comes from Kingston whose enterprises average about $26,000 a year in value of production. The corresponding annual value of pro- duction for the remaining locations (strata) are $44,000, $15,000, and $5,000, respectively for the Major Towns, the Rural Towns, and the £05. The manufacturing contribution to GDP by the SSI subsector must be seen relative to its capital labor ratio gig a_yj§_the large-scale manufacturing firms. In 1979, the value of fixed capi- tal (at replacement value) per unit of labor was J$2,041 in the SSI subsector. The corresponding value for all the manufacturing sector was $8,605 (see Ayub, 1981, p. 58).:1 Thus, the corresponding value for just the large-scale firms must be much higher than the overall average indicated here. In fact, for a sample of selected groups of firms with gross annual sales in excess of half a million dollars, Ayub gives this ratio close to $11,000 as of 1973 (p. 24). It may be concluded, therefore, that the capital-labor ratio in 1979 for the 1Ayub's capital value is not replacement value. Most probably it is book value. 8O large-scale manufacturing firms may be at least five to seven times as much as in the SSI subsector.1 In conclusion, when the necessary calculations are made for the total employment and dependency,2 the number of people who are directly supported fully or substantially by the small-scale non- farm enterprises (SSE) in Jamaica could be anywhere from one-quarter of one million to 300,000 people. This is more than one-eighth of the national population. When the other contributions, such as indirect employment creations, the training of apprentices, the generation of foreign exchange (e.g., the craft industry) and the social and political benefits are considered, the role of the small- scale nonfarm enterprises subsector in the national economy holds an important place. For general descriptions of such contributions to an economy,.see Chuta and Liedholm (1979, pp. 2-7). 1Because of data paucity or incomparability, no such compara- tive analysis could be made in terms of value added. For the SSI subsector alone, however, the rate of value added in gross value of production was about 80 percent. The fact that some clients (e.g., in tailoring) bring some of their own raw materials will probably tend to alter this rate. The distortion will be small, however, as the main input in the SSI subsector is (own) labor. In fact, for this reason out of every dollar of sales, about 60 ¢. accrues to pr0prietor and family labor (and normal profit). 2Dependency refers to the number of people or family members who get more than one-half of their support needs for more than half of the year from the person who is working in a SSI enterprise. The working person could be the proprietor, a permanently hired worker or an apprentice. The mean number of dependents for each of these three labor categories was 5, 2.5, and 0.3, respectively. This infor- mation was collected along with the management study data; these averages were assumed to be the same for the SSE enterprises also. Only about 6.4 percent of the total income to support these "depend- ents" comes from sources other than the small-scale nonfarm enter— prises (see Section 4.1.3.4). 81 Regarding skill development through apprenticeship training, the SSI are far more important than the nonmanufacturing enterprises: The percentage of apprentices out of the total labor force among the nonmanufacturing group of enterprises in the SSE is only 0.8 (10 per- cent in the SSI enterprises). Although their number is half as many, the SSI enterprises support nearly 90 percent (125,000 people) of 1 what the nonmanufacturing group of the SSE as a whole do. They also contribute about 3.5 percent to GDP. 3.2 Recent Economic Trends Earlier in the problem-setting section of Chapter 1, the plight of the Jamaican economy was reviewed. During roughly the last half of the 19705, manufacturing output declined by about 26 percent and the national GDP fell by 13 percent. It was noted that the small-scale manufacturing subsector (SSI) seemed to be growing, number-wise at least, relative to the larger scale subsector. In this section, a closer focus will be applied to the SSI to see what changes have taken place over the years and particularly during the second half of the 19705. Special attention will be given to changes in the global or industry-wide product demand and the extent to which such demand changes are due to changes in number of firms in the industry and demand (own output) changes among existing firms. Furthermore, changes in the price structure of key inputs and outputs will be closely examined with the view that (a) they 1Unless indicated otherwise, discussion henceforth will deal with the small-scale manufacturing or industrial (SSI) enterprises. When both manufacturing and nonmanufacturing are included, they will still be referred to as SSE. 82 might have influenced the level of production at the industry level or at the individual firm levels and (b) they may shed some light on the terms of trade for the subsector. The direction of changes both in industry demand, number of industry firms, and own output were supplied by proprietors who were asked to state the general business trend in their own industry and their own enterprises over the last year and the last five years. the proportioncniproprietors who responded with an increase are then compared with those who responded with a decrease. The difference between these two proportions is taken as a guide to explain trend changes in the respective area for a particular period of time. Such differences Show only the net percentage of respondents who claim the demand or firm size to have improved or deteriorated over the period and not the actual rates of business growth or decline (see Haggblade et al., 1979, p. 37). This will be made clearer as the economic trend indicators of industry demand, number of firms, and own output are anlayzed in the following pages. Changes in the size of an enterprise are also examined using changes in initial labor force size and number of all machines and specifically of powered machines over the years. As reSpondents were required to remember the numbers for these variables when the business was established and at the time of the interview (1980) only, rates of changes for them are relatively more reliable--i.e., there is less dependence on personal judgment compared with some of the other indicators. 83 3.2.1 Changes in Product Demand Levels In attempting to describe the general economic picture of the small-scale manufacturing subsector over the last few years, the attention in this section will be on the demand situation. The basis for such analysis will be proprietors' perception of both demand levels and other external influences. Their responses are shown in Table 5. Before discussing that table, however, it would be useful to review the basis or criteria on which such responses were made. In other words, when respondents describe the demand level as being weak or strong, what are the criteria used to measure such weaknesses or strength? Attempts were made to find answers for this question. The results are shown in Table 4. TABLE 4.--Indicators used by proprietors to estimate trend changes in demand (percent of proprietors) Indicators Percent 1. Total Sales (or Work) 91.1 2. Total Cash Received 5.4 3. "Profit" 2.5 4. Periodic Withdrawals 0.6 5. Other __Jlg4 100.0 Source: Management Study Survey, 1980. 84 It is important to keep in mind that the indicators shown in Table 4 refer to description of the general business condition or state. It does not necessarily imply that the same variables are used by proprietors to analyze the periodic financial performance of their respective businesses. It will be shown in Chapter 4 that variables or criteria used for the latter purpose are different (except for total sales and profit) than those shown in Table 4. Table 4 shows then that the demand responses given in Table 5 are based mostly on total sales or total amount of work done. When close to three-fourths of the total business (production) in the SSI enterprises is done on customer order basis, as opposed to production for stock or inventory (see Chapter 4), sales and work assume almost the same meaning. The respondents who gave "total cash received" as criteria are those 85 percent of whom have reported in Chapter 4 quitting extending credit (presumably because they may have been losing money from bad debts or credit sales). Looking at Table 5, respondents' perception of what happened to demand over the previous year is not clear. Equal percentages thought business has improved (35.9 percent) and declined (34.7 per- cent), with the rest sensing no change. However, the analysis of such an aggregate response is misleading since there is a cancelling effect of answers across different industries. When the responses are examined within each industry, the existence of some trend is more obvious. Table 6 shows this industry-by-industry difference which will be discussed next. 85 TABLE 5.--Recent economic indicators of the SSfisubsector (percent of enterprises) Percent of Proprietors Indicators 1 2 3 4 b Decreased Same Increased (3-1) Period 1. Over the lastyearC a. Industry Demand 32 31 33 l 0. Number of Firms 33 34 33 O c. Own Output 39 29 32 -7 d. 1978 Output Value d Compared with 1977 53 30 16 -37 e. 1979 Output Value Compared with 1978 64 7 26 -38 2. Over the Last 5 YearsC a. Industry Demand 20 28 43 23 b. Number of Firms 24 22 45 21 c. Own Output 31 14 51 20 Source: Management Study Survey (1980). aSSI stands for the small scale industrial (manufacturing) enterprises subsector. bPercentage increase minus percentage decrease. cThe balance from 100 percent is accounted for by people who didn't know the direction of the trend. dThis information was collected during the Phase 11 survey. 86 3.2.1.1 Over the oneeyear period.--An examination of the one-year period reveals the following from Table 6. Metal works, shoemaking, and wearing apparel (tailoring) seem to have their indus- try demand increased over the year. The remaining industries men- tioned in Table 6 had a decline with the largest percentage of enter- prises reporting such a decline found in woodwork. The increased industry demand level in wearing apparel is reflected in increased number of new firms and expanding output of existing firms. Table 6 reveals there is great variation at the industry level. Thus, wearing apparel (except dressmakers which are not shown in Table 6), metal works and shoemaking had increased output demand while this is not true for woodworks, craft, and repairs. Both for craft and repairs, the industry demand is described as declining over the year. In craft, this is accompanied by declines both in the number of enterprises and in own output. In the case of repairs, the declining industry demand is not only accompanied by an increased number of new firms joining the industry, but the existing ones were losing slightly, although the perception of pro- prietors seem to be less definitive for this particular group. In the case of woodworks, there was both a general industry demand decline and an increase in the number of firms. Both of these resulted in a decline of the output in existing enterprises. The main problem with the woodwork industry was the lack of lumber (Fisseha and Davies, 1981). 87 .muowcmn cmmxum>ww 0:0 cmmzlmco 050 com >~0>wuumammc 0:000 m use H ochm .Aommfiv am>csm zuzgm acmemmmcmz ”00czom am no m m - we- mm. 0 AN oe mm- NH em “n-0v .0 mm Hm we ma oN m Nm we we NH Nm om mmmmgocg .0 o HN m cm 0 mm mm mN m Nm m0 0a 020m .n Nm NH me NN «N Nm mN mN NH om oN mN 0000;00o .0 usapeo :30 .m we- we- Hm om «N we- w mN 0 NH NH oH Aniov .0 NN 0N mm me No ON Nm MN om N0 Nm mN mmmmeocH .0 NH o HN mm 0 0H mm NN HH am No Nm 050m .n no «N 0H 0H mm em mN o eN vN ON mH mmmmc0mo .0 weer; Lo consaz .N mN mm we mN- mN NH- w mm N mm- we ow Aeu0v .0 em mm mm HN mm 0N 0m me we ON mm me mmemc0cH .0 o o oH Hm «N me 0 mm m N oN mm 020m .n om HN 0N me «N Nm me c we wN m m 0mmmc00o .0 cameos xgumaucH .F m P m P m _ m P mm _ m P mmozm mc_egmm pLecu xcoz_muoz xcozuooz pewsceo mcopmu_0cH nausea mmcwazocw xcumavcH Aomwcacmpcm Lo pcmucmav mcoumuwccw ucmsmu Lo mmcmg0 cw mmococmmmpu pmwcumzncHii.m m4m

m>c=m xuzum pcmswmmcmz "moczom m.NH m.Nm o.H N.mN 0.0 o.m v.0 m.o ucmm m.moH N.mNH o.HN o.Hm N.©N N.Hm N.o N.N mmcwnumz zmz o.ooH ¢.¢mH m.cm o.mm m.m¢ m.N¢ m.o o.H mmcwgumz vow: ¢.ooH N.NmH o.om o.mm o.ON m.NN H.o m.H Loam; 00—wam «.mefi o.NmH w.oo m.Ho N.mH N.MN o.o m.N Lone; umppwxmca m.omH m.mmH N.ow o.mHH 0.0N H.N¢ H.o m.HH 00300;; hwy o.oom o.omm m.mm N.NPF F.mN m.wm N.¢ m.NH megmumz 30m Amx :mwumz cmmz cmwumz :00: cmvvwz :00: :mwumz :00: mNmH NNmH ommfi EmuH Lo Agmzcww Ammmcmc0 mmmgcwocmav mvomema Lowpcmm 50?: 00200500 mama? max Lo moovca ommfi chaL0> m0 xmwu=00w mcou0wcmogq ;0_;z ~0>0F 0 :op0a 0300 _F0m 00 0003 0Eouc_ 0000030 0:0 L_m Nmm0=_m:a 0:0 500» mc_uu0m 0L0 a050 0200:_ Lo _0>0F Aomm_v 0000030 0:» gap: 00PLm_u0m mcou0wgaoca 0e<0 .zoc mp0om xuwco_cm 0:0 000 000cpmzn 0:» Easy u:0E>o_mE0 Lo 0500:? 50:: so» mcou0_caogm 00 00000000 N0mooc0 x050 0pzoz omwcac0uc0 Lo 0:_x 00:3 .000003 x050 0m_gac0uc0 00 00>» 0:» 000000 0» omm— cw 000000 0 00; mcou0wcnogm mmu .A00poz0 0mmsgc0uc0 Low 000000 ..0._V _ asapcu cw :0>_m 003000 0pnwmmoq 0000 Lo» 0000000 5000 mm0=_m:n 000:» 0000000 0;: 00000000000 00 0m0uc0000ma .0m_cmc00:0 co mm0=_m=a c0_00_uc0m L_0;u 000:0 »—_0cpm_co >000 ms: 00300 0003 mcou0_gaoca0 .Aowmpv x0>cam x0=0m uc0s0m0c0x "00caom N.Ne 0.00 0.00 ".mm 0.0N 0.00 e.m0 0.00. mo<000 mommm mommm momll mommm mom]: mommm mommm .momli 00:00 .0 N.em m.Hm 0.0 0.00 0.0 m.mN 0.00 H.m amoe_m=e »__sea .0 0.30 e.~0 0.0 m.~0 _.NH 0.02 0.00P _.0 suaeseeoeao ume_e .e 0.00 “.00 0.N0 N.NH 0.0 0.00 o.co_ 0.00 0e60e_ 0660 .m 0.00 0.00 0.03 N.Nm 0.00 0.03 N.Nm 0.0N »0_espeoeao »_eo .N N.me 0.00 0.0N 0.00 0.0m 0.00 e.NN 0.00 amoeooe_ ego .— 00” 1.0.0 90“...“ in...“ .mmmw 000cwmam -m_umm mp0ow 0:02030 0:0>wu 00wozo L mm0cwm=w 0030800:U Lo 00wocu ("111.111 l.liUNrHiHu. u N1 111.11 lilfillnfluilli UnirlHlu. uni {1.3.01.0 .111 (l.l ”I. "I. u I "It'll-ll" H n I. (I "Illa a. ‘(IIIOIHI 14 .” J n“! 1! «5171) II.. H II 1.": .II ”(511”)“ MI I. l. I'lflln “Till-1115.171.“ In ,1: .111“! "Ill. "l.llllIIIIIll II 1.4 Amcou0_caogg Lo 0:00000v 0:0eppwwpzw c_0;u 0:0 meowu0000ax0 _0_L:0:0LQ0L»:NII.NH mmm0F0m 00 000502 .0 m.mH o.mm 0.00 0.00 w.mH H.0m mZMHH hmou 00< mom z c000: 00000 000>0P0m 050 00000000 00 0000000 00000000000 00 0000000 A00000000000 00 0000000V 00000000000 00 00000 0000000000 00 00000000000000 0:0--.0N 00000 176 named what they identified as their costs) if they would produce an item for sale if they were paid the costs they mentioned. About 40 percent of them said they could not, indicating they had left out some costs. After they added whatever costs they considered, they were again asked if they were willing to sell now after the price covers the latest costs as well. If the response to the second ques- tion is positive, then the number of all relevant costs and those left out by the respondent were checked and tallied. The results of these two approaches will be discussed now. The average number of cost items (excluding taxes and inter- est) in the SSI enterprises is about 5.6 (median is 5.7) out of which about 2.2 (median is 2.3) were left out in the expense approach. Thus, about 40 percent of the cost items were left out. (When indi- vidual proprietor's percentage of costs left out are added and aver- aged, the resulting percentage is also 40 percent.) When the profit approach was used, the percentage of costs left out is 20 percent. The 20 percent is a very conservative level, since proprietors were already made aware of some of the costs from the expense approach question which was asked first. Ranking the costs according to the percentage of proprietors that failed to mention them in the expense approach, fuel and trans— portation expenses are in first place followed by repair-maintenance- depreciation. Almost a fourth of the proprietors also failed to include their own labor as cost. By contrast, the number of people who failed to include raw materials and hired labor is very small. With the profit approach there is improvement with every cost item. 177 This, as already explained, is due to the fact that the expense approach was administered first and as a result of the probing that occurred, proprietors had their minds refreshed and thus were able to count or include more costs in the profit approach. The percentage shown for taxes are not actually percentage of proprietors who failed to count them as costs when they should. They show percentage of pr0prietors who never mentioned them in the discussion. Thus, nobody raised them during the expense approach while about 14 percent mentioned them during the profit approach. Hence, it could be inferred that at least 14 percent of the proprie- tors probably pay taxes. Because of the sensitivity of the topic, respondents were never directly asked whether they pay taxes or not. Table 24 also shows the percentage of proprietors who failed in the two approaches to include a number of valid costs. For example, at the national level and using the expense approach, 43.4 percent of the pr0prietors failed to include at least three cost items. At the industry level, wearing apparel showed the highest rate (5l.33 percent) of failure to include all costs under the expense approach. It is distantly followed by shoes (39.8 percent), repairs (33.6 percent), and woodwork (31.0 percent). When the profit approach is used, there is not much difference although woodwork and repairs take the lead now over shoes. The percentage of cost items "forgotten" to be included probably depends on the number of purchased inputs, on the frequency and size of purchases and on the differences of meanings for costs. 178 It has not been proven here that failure to identify certain costs as expenses means proprietors also ultimately fail to make provisions for all of them. Some proprietors lump some costs with others (e.g., transportation with returns to own labor). Failure to identify a relevant cost may result from (1) not considering it as cost, (2) lumping it with other costs, or (3) considering it unim- portant. Whatever the reason may be, failure to explicitly identify and account for all costs could result in ineffective cost control measures and thus in inefficient business performance. The relation of the present discussion to that in Chapter 5 is very important. For, if respondents did not know what their costs consist of, then it would be difficult to see how they can achieve price or allocative efficiency. The performance analysis of the business and the decisions that come out of such analysis may be misdirected at least. Calculated technical efficiency could be affected also via the value added approach if respondents don't state all their cost items. The implication of the discussion strongly points also the importance of paying greater attention to how field data should be collected. It is not enough to simply ask for certain values without making sure there is a common understanding of concepts between the interviewer and the respondent. 4.3.3.2 Handling of funds.--Here topics such as the separa- tion of business and nonbusiness funds, the mode and frequency of payments for the proprietor's services and factors affecting cash withdrawals and budget allocations will be briefly discussed. 179 Fifty percent of the proprietors don't separate household and business funds (particularly long-term funds). The other half tries to separate them through bank accounts, bookkeeping, and physically separating them in storage. The two most popular ways to separating the two funds (home and business funds) are bank accounts and physical separate storage of each. As will be seen below, sometimes priority is given either to the home or the busin- ess. About half of those who try to separate the two funds maintain bank accounts; this would be about 33 percent of the total number of proprietors. At the industry level, except for repairs and wood- works about 30 to 40 percent of the enterprises keep bank accounts. Repairs proprietors have the highest percentage (61 percent) for bank accounts while woodworks have the lowest (7 percent). Proprietors who don't try to keep the two funds separate were asked how they allocate funds between business and other needs. About 50 percent said that the one which has the greatest need takes priority in withdrawing from the funds. This depends strictly on need. About 40 percent said they usually have no specified way of fund allocation. Cash needs for the business, the home, personal use, etc. would all be drawn from the same pocket, so to speak, depending on current necessities. About 9 percent said that they give first priority to business; only 3 percent would give first priority to the home. Another 2 percent would treat them equally, i.e., split the available funds between the two. About 4 percent of the wearing apparel pr0prietors would give priority to nonbusiness 180 need in sharing the current sales funds; the corresponding percentage for woodworks is 0. About 8 percent of the wearing apparel group would given priority to business need to 7 percent for the woodwork group. Only 16 percent of the pr0prietors have a fixed salary which they withdraw regularly. The remaining depend on regular withdrawals from sales. For 99 percent of the respondents, payment is done every week. For a third of those who depend on this method, regular withdrawals depend on the size of the total sales. At the industry level, 43 percent of the wearing apparel group base withdrawals on the size of total sales; the bigger the sales, the bigger the withdrawal. The corresponding percentage for wood- works is only 14 percent. Proprietors were also asked how they allocate what they call profit. (It was pointed in the last section that this profit may contain at least 20 percent of the cost items undeducted.) The attempt here is to see what other nonbusiness related activities have a claim also on the funds generated from the business. Since the analysis both for those who keep the business and nonbusiness funds separate and for those who don't was done together the picture is not clear. However, what is strongly shown is that about 58 percent of the pr0prietors would use money generated from this business to support other businesses, such as farming, or build a home. In light of what was said in the previous subsection concerning proprietors awareness of future needs of investment funds, this could create 181 critical financial problems when the time comes to replace some of the equipment or machinery. Another 18 percent would reinvest it in the business in the form of working capital while another 12 percent just put away in the bank for future long-term investment. To conclude this subsection then, about 50 percent of the pr0prietors don't separate business and nonbusiness funds. About a third of the proprietors keep bank accounts which they use to sepa- rate the funds. About 84 percent of the proprietors are paid on withdrawal basis every week. It seems a large portion of the busin- ess funds are siphoned to other businesses or activities. Finally, looking at the effect of such financial management on the technical efficiency of resources, it seems that the avail- ability of funds to replace old machinery or to buy new ones for expansion, has an important implication for the average productivity of labor and thus for the enterprise's overall technical efficiency. This has also an implication for the potential scale of production and the concomitant input productivities. 4.3.4 Financial Business Evaluation The subject matter in this section is fundamentally related to the objectives of the proprietor. It has been pointed out in Section 4.1.3 that the overwhelming majority of proprietors want income and employment from their businesses. Therefore, the criteria used in business evaluation should reflect these objectives. For two reasons, the discussion here will be brief. A sub- stantial portion of it has been discussed in the Phase 11 report 182 (Fisseha and Davies, 1981, p. 90) and some of it has already been touched upon in the previous section and in Chapter 3. . It was indicated in Chapter 3 that proprietors look at total sales to see overall trends and business environments. When it comes to examining the financial or income performance of each business, however, the indicators used vary with some industries using more of some indicators than others. Table 25 shows that less than a third of the pr0prietors check the returns made on given raw material pur- chase per period, usually a week or two. For example, a shoemaker may buy a piece of leather for $50, then at the end of a week, he would sit down and compare the total sales or repairs with the value of leather he has already used up. If sales has covered the purchase value of the used leather and some is left for his own labor and other expenses, then he may be doing good. The percentage of craft proprietors who check return to raw material is very high (94 percent). This is partly due to the fact that they don't commonly have any input other than raw material, usually straw, to buy and that only occasionally. About 28 percent of the pr0prietors check sales and expenses for key items. A dressmaker may, for instance, check the number of dresses she has sewn and the expenses that go with them (rubber bands, thread, belts, etc.) and not pay attention to the amount of repair work she has done. About a fifth to a fourth of the pr0prie- tors cost out each item made making sure there would be enough in the charge to pay for other expenses besides the raw materials involved. In fact, due to the lack of standardization of products, 183 .000000 00>000 00000 000000000: ”000000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 00<000 mqml MAW. 040.! .040! .03.! .0401! film: 000.05 .0 0.0 0.0m 0.0m 0.0 0.0 0.0 0.0 000000 .0 0.~N 0.0m 0.00 0.0 0.00 0.00 0.00 0000 00000>0000 00 000000 .m 0.0m 0.0m 0.0m 0.0 0.00 m.0~ 0.00 000000x0 000 000 00000 0500 .N 0.00 0.0m 0.00 0.00 0.00 N.m~ 0.00 00000000 300 00 000000 .0 00000 0000000 00000 000002 00003000: 0000000 0000000 0000000000 0000000000 000000000000 00 00000000 00000000000 00000000 $0 0000000000nn.0m 00m000 00000 0000000002 "000000 000000000 00000000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 000000 0401: .0401. .mqmnu .040.1 .0401. 0000:. 0400:. 000000 .0 0.00 0.00 0.00 0.0 0.00 0.00 0.00 00 0wfimww00mn0wowmmm .0 0.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0000 .0 0.0 0.0 0.00 0.0 0.0 0.0 0.0 0000000 .0 0.00 0.00 0.0 0.00 0.0 0.00 0.0m 000000000000 .0 0.00 0.00 0.0 0.0 0.00 0.0 0.00 000003 .N 0.00 0.0 0.0 0.00 0.0 0.0 0.0 00000000 oz .0 0000000 00000 000000 00000 00000: 00003000: 0000000 00000000 00 000000000 000000000000 00 00000000 00000000 00000000000 00000000 00 000000000--.0~ 000 .u ..a g". .n_ ~.. on. oN m.m ow. on A“. “segue _auoh .o o.o~ mo.- end m.- m~.- GN ..- _~.- om .V 4.36: coca; .a o.m_ n~.- .8 ~.nm ”H. ._ 6.6. ow. ON ..V deco. cone; .. u~.m om_gnsuu:u .— —o>u4 L mouse —o>oa L momma pw>mg L mamau uu:ou.o_cu—m we we cocoa—u'cm_m we we oucau.u_cm_m ‘0 Go a o~_m ences: a o~_m Loose: u o~_m cones: xoeo_o_ccu e. copuo.go> noa.cocouem _.< nxcozcoo: oe.co__ap co outsom mo*goouuuu po_gum:uc_ mmpnuvgo> cacao vac mmuvuc* xocm.u¢cum 84 oz» comxuoa “CV co_uo—osgoo a—asvm-.~n m4m

so» Q\>-AqiuA8o 1 Hy + a-x8cvom n 8> unquzo mo mapm> Low Q\>iAa-EA8o-8o-HV + 8-888 + Q-x8ovom n 8 "88888858888 o>8888888888 888 888888888888 uaqcpv 88838—888 on» 888 888888888 8888888888 8:» .mom 8:8 mom 88888 88.8 888 aw 8:8 mm 88_888 888 .88888888 o>888888588 8888 a” an 88888888888 Rm 88 88888888888 88 uoH an 88888888888 « .om\mmmfi Aw>g=a 8889 3888 8888588 "wugaom «ct 88 888 88 88 8888888888 88 888888 .8 888 88 88 88 8888888 88 8888888 .8 8888.88 8888.88 8888.88 888888.88 8 8888.88 8888.88 8888.8 8888.8 A 88 8888888888 .8 8888.88 8888.88 8888.88 8888.88 8888.8 8888.88 8888.8 8888.8 888 888888_88888 .8 8888.88 «8888.88 8.88888.88 8888.88 8888.8 8888.8 8888.. 8888.. Asv 88888 .8 «88888.88 - - -- 8888.8 88 888 .8888882 .8 ##«Amom.mv «¥«Aoo.mm~v «#«A~H~.mv «Aves-my 88888.8 8888.8 8888.8 8888.8 8888 8.8 L888 .8 «##ANNO.NV Amoo.ov «88Ammv.mv ttflcoo.mv 8888.8 8888.8 8888.8 8888.8 88 888 _88_888 .8 88888 88.88 88888 88888 88888 88_8> 88_8> .8888 11111111111. 11111111111. L88858888 ._.8.8 8888 888888_88 8888 88 88888888888 mmc_8:ogu xcumscc~ o-.I- III-1| 1 -1111-111I 111-ll11 1 1 1 ".u.‘ 111-oqr‘1 011 1.11111111litl.l.. 1fl11-.11..1|1I11.11111.1l1111lru1. 1 1 .. . 1 0.11-11 11111 11! 411-1.!1- 1111.111 8:8888838 Ammuv 888888888888 88 8888_888_8 8:888:88 8:8 88 88—8888 888888888; 88858—8888 E:E_xmz-.8m w8m<8 APPENDIX IV ADJUSTMENT FOR HETEROSCEDASTIC WOODWORK DATA 291 APPENDIX IV ADJUSTMENT FOR HETEROSCEDASTIC NOODWORK DATA The possible problem of heteroskedasticity of the woodwork data discussed on pages 210 and 211 was corrected using the pro- cedure suggested by Park (1966). He hypothesizes that the variance of each Observations error term is proportional to the values of the independent variable causing the heteroscedastic problem. Thus the relationship, 8: = 02kgev (A.1) where 02 = variance of an observation error term Ui 02 = the true regression variance k. = independent variable causing the problem, in this case capital at replacement value adjusted for capacity levels r = a constant e = base to the natural logarithm v = a well behaved error term. The equation implies that if the variance of the i'th error is divided (deflated) by K raised to the power of r, then the mean of the new value will be a consistent and asymptotically efficient estimate of the true constant regression variance or 02. (It can 292 293 easily be converted into an unbiased estimator also, see Kmenta, 1971, p. 259). Thus, resulting in unbiased regression variance: E [oa./kr] = E[02ev] = 02 (a.2) 1 Similarly, if the observation error term variance is deflated by k raised to the power of r/2 (thus extracting the square root on both sides), then 0U. will estimate the true regression standard deviation. It is thIs last procedure which is used to correct for the heteroscedasticity. In order to get a good estimate of the 03., repeated observa- tions of the dependent;variable must be taken for 8 fixed value of capital. Since this is not usually possible or reliable in socio- economic research, he assumes that the variance of the error term can be estimated by using the square of the error term itself generated from an OLS regression of the untransformed or initial equation. Thus, each residual at a given capital value when squared estimates, its own variance at that capital level. The unknown parameters (02 and r) can then be estimated using the log-linear model. Thus, ln(U?) = ln (02) + r ln (ki) + v (a.3) The result (from the OLS regression) is,1 1n(u§) = 7.3557 - 5.0729 (k1) (a.4) (3.4325) OY‘ 1The value in bracket is T-value. 294 = e3.6779 ( -2.5365 U. 1 k') 1 (a.5) OY‘ U./(k-.2.5365) = e3.6779 = O 1 1 (a.6) The coefficient of capital (k) is significant at 1% with 27 degrees of freedom (although the Value of the adjusted Résquared is small). The sign of the coefficient is also as already indicated negative. The correction factor for the residual or Ui (i.e., capital value raised to the power of -5.0729/2 = -2.5364) is then applied also to the remaining variables in the woodwork regression (i.e., to the constant or shift parameter, capital itself, labor and raw materials). Since, the constant is also similarly affected, there will be no intercept for the new regression of transformed variables. And thus the usual computer generated adjusted R-square cannot be used to check goodness of fit (see Kennedy, 1979, p. 25). It can be calculated however by finding the ratio between the sum of squares of the difference between the observed and the predicted value of Y(RSS) and the total sum of squares of Y(TSS); it should be adjusted for degrees of freedom. In the present case, adjusted R-squared improved from 0.70 in the original equation to 0.78 under the new one. but there was no improvement in the coefficient sizes. After the variables were corrected for the heteroscedastic problem, both the OLS regression and CES maximum likelihood were 295 estimated. The result was that the raw material continued to swamp the effect of both labor and capital in the OLS; the CES maximum likelihood estimate again failed to converge. The result for the OLS value-added regression was V = 2.9118 + 0.3119(k) + 0.5486(L) (a 7) (3.131)*** (2.082)** (3.781)*** - R2 = 0.780 where variables are the gnng as those in Table 29. Thus, the approach did not substantially improve the results. Both the labor and capital coefficients are lower (although of the same significance levels) and the size of the constant is 27% higher than in the original (untransformed) regression. This is in spite of the highly significant coefficient for the correcting variable, capital, shown in Equation a.4. Thus, the correction for heterosce- dasticity was considered unnecessary and was dropped. BIBLIOGRAPHY 296 BIBLIOGRAPHY Afriat, S. N. 1972. "Efficiency Estimation of Production Functions." International Economic Review 13 (October): 568-598. Aigner, D. J., T. Amemiya, and D. J. Poirier. 1976. "On the Esti- mation of Production Frontiers: Maximum Likelihood Esti- mation of the Parameters of.a Discontinuous Density Func- tion." International Economic Review 17 (June): 377-396. Aigner, D. J., and S. F. Chu. 1968. ”On Estimating the Industry Production Function.“ American Economic Review 58 (September): 826-598. Aigner, D. J., C. A. Know Lovell, and Peter Schmidt. 1977. "Formu- lation and Estimation of Stochastic Frontier Production Models." Journal of Econometrics 6 (August): 21-37. Anderson, Dennis, and Mark W. Leiserson. 1980. "Rural Non-Farm People in Developing Countries." Economic Development and Cultural Change 18 (January): 227-248. Arrow, K. J., H. Minkas Chenery, and R. M. Solow. 1961. "Capital Substitution and Economic Efficiency." Review of Economics and Statistics 43 (August): 225-250. Artus, Jacques R. 1977. "Measures of Potential Output in Manufac- turing for Eight Industrial Countries, 1955-78." Staff Papers (International Monetary Fund) 24 (March): 1-35. Atkinson, A. B. 1976. The Personal Distributions of Income. London: George Allen and Unwin Ltd. Baily, Mary Ann. 1981. "Brick Manufacturing in Colombia: A Case Study of Alternative Technologies." World Development 9 (February): 201-213. Bain, Joe S. 1969. "Survival Ability as a Test of Efficiency." Pnpers and Proceedings, American Economic Review 59 (May): 99-104. Bakan, David. 1966. "The Test of Significance in Psychological Research." Psychological Bulletin 66 (December): 423-437. Baumol, William. 1968. "Entrepreneurship in Economic Theory." American Economic Review 58 (May): 61-70. 297 298 Bernhardt, Irwin. 1981. "Sources of Productivity Differences Among Canadian Manufacturing Industries." The Review of Economics and Statistics 63 (November): 503-512. Bhalla, A. 5., ed. 1981. Technolpgy and Enplnyment in Industry. 2nd ed. Geneva: International Labour Office. Boles, J. N. 1967. "Efficiency Squared-Efficiency Computation of Efficiency Indexes." Proceeding§ of the 1966 Meetings of the Western Farm Economics Associations.’ Pullman, WEshington: Washington State University, pp. 137-142. Byerlee, Derek, Carl Eicher, Carl Liedholm, and Dunstan Spencer. 1977. "Rural Employment in TrOpical Africa." African Rural Employment Working Paper No. 20. East Lansing: MTcthan State Universtiy, Department of Agricultural Economics. Carter, H. 0., and H. 0. Hartley. 1958. "A Variance Formula for Marginal Productivity Estimates Using the Cobb-Douglas Function." Econometrica 26 (April): 306-313. Chambers, Edward J., and Raymond L. Gold. 1963. A Pilot Study of Successful and Unsuccessful Small Business Enterpnises Within Montana. Bozeman: Montana State Universtiy. Charnes, A., W. W. Cooper, and E. Rhodes. 1978. "Measuring Efficiency of Decision Making Units." European Journal of Operational Research 2 (November): 429-339. 1981. "Evaluating Program and Managerial Efficiency: An Application of Data Envelopment Analysis to Program Follow Through." Management Science 27 (June): 668-699. Chen-Young, Paul. 1966. "An Economic Evaluation of the Tax InCentives Programme in Jamaica." Ph.D. dissertation, University of Pittsburgh. . 1967. " A Study of Tax Incentives in Jamaica." National Tax Journal 20 (September). Cheung, Steven N. G. 1974. "A Theory of Price Control." Journal of Law and Economics 17 (April): 53-71. Chow, G. C. 1960. "Tests of Equality Between Subsets of Coeffi- cients in Two Linear Regressions." Econometrica 28 (July): 591-605. 299 Christiano, Lawrence J. 1981. "A Survey of Measures of Capacit Utilization." Staff Papers (International Monetary Fund) 28 (March): 144-197. Chuta, Enyinna. 1977. "A Linear Programming Analysis of Small Scale Industries in Sierra Leone." Ph.D. Dissertation, Michigan State University, East Lansing. 1978. “The Economics of the Gara (Tie-Dye) Cloth Indus- try in Sierra Leone." African Rural Employment Working nger No. 25. East Lansing: Michigan State University, Department of Agricultural Economics. , and Carl Liedholm. 1979. "Rural Non-Farm Enployment: A Review of the State-of-the-Art." Michigan State University Rural Development Paper No. 4. East Lansing: Michigan State University, Department of Agricultural Economics. ____,____, Ola Roberts, and Joseph Tommy. 1981. "Employment Growth and Change in Sierra Leone Small Scale Industry: 1974-1980." Africal Rural Economy Program Working Paper No. 37. East Lansing: Michigan State University, Deparment OTTAgricul- tural Economics. Coffey, Robert E., and John F. Herrmann. 1976. "Is Formal Education Related to Success?" Journal of Small Business Management 14 (April): 1-5. Davies, Omar, Yacob Fisseha, and Claremont Kirton. 1979. "Small- Scale, Non-Farm Enterprises in Jamaica: Initial Survey Results." Michigan State University Rural Development Series, Working Paper No. 8. East Lansing: Michigan State University, Department of Agricultural Economics. Debreu, G. 1951. "The Coefficient of Resource Utilization." Econometrica 19 (July): 273-292. DeCarlo, James F., and Paul R. Lyons. 1980. "Toward a Contingency Theory of Entrepreneurship.“ Journal of Small Business Management 18 (July): 37-42. Dhrymes, P. J. 1965. "Some Extensions of Tests for the CES Class of Production Functions." Review of Economics and Statistics 47 (November): 357-366. Dillon, John L. 1980. "The Definition of Farm Management." Journal of Agricultural Economics (U.K.) 31 (May): 257-258. Dixon, W. J., and M. B. Brown. 1981. BMDP-Bl Biomedical Computer Programs P-Series. Los Angeles: University of California. 300 Domar, E. 1962. "On Total Productivity and All That." Journal of Political Economy 70 (December): 597-608. Douglas, Merrill E. 1976. "Relating Education to Entrepreneurial Success.“ Business Horizon 19 (December): 40-44. Duft, Kenneth D. 1979. Principles of Management in Agro Business. Ruston, Va.: Ruston Publishing Co. Eicher, Carl, and Lawrence Witt, ed. 1964. Agriculture in Economic Development. New York: McGraw-Hill Book Company. Fare, R., and Lovell, C. A. K. 1978. "Measuring the Technical Effi- ciency of Production." Journal of Economic Theory 19 (October): 150-162. Farrell, Michael J. 1957. “The Measurement of Productive Effi- ciency." Journal Royal Statistical Society (Serial A General), 120, Pt. 3: 253-281. , and M. Fieldhouse. 1962. "Estimating Efficient Production under Increasing Returns to Scale." Journal of Royal Statistical Society (Serial A General), 125, Pt. 2: 252-267. Feldstein, Martin S. 1967. "Alternative Methods of Estimating a CES Production Function for Britain." Economica 34 (November): 384-394. Ferguson, E. C. 1963. "Cross-section Production Functions and the Elasticity of Substitution in American Manufacturing Indus- try." Review of Economic and Statistics 45 (August): 305-313. Fisher, Franklin M. 1969. "The Existence of Aggregate Production Functions." Econometrica 37 (October): 553-577. , Robert M. Solow, and James M. Kearl. 1977. "Aggregate Production Functions: Some CES Experiments." The Review of Economic Studies 44 (June): 305-320. Fisseha, Yacob, and Omar Davies. 1981. "The Small-Scale Manufactur- ing Enterprises in Jamaica: Socioeconomic Characteristics and Constraints." Michigan State University Rural Develop- ment Series, Working Paper No. 16. East Lansing: MiChigan State University, Department of Agricultural Economics. Fitchett, Delbert A. 1976. "Capital-Labor Substitution in the Manu- facturing Sector of Panama.“ Economic Develgpment and Cul- tural Change 24 (April): 577-592. 301 Forsund, F. R., and E. S. Jansen. 1977. "On Estimating Average and Best Practice Homothetic Production Functions via Cost Functions." International Economic Review 18 (June): 463- 476. Forsund, F. R., C. A. Know Lovell, and Peter Schmidt. 1980. "A Survey of Frontier Production. Functions and of Their Relationship to Efficiency Measurement." Journal of Econo- metrics 13 (May): 5-25. Gabrielson, A. 1975. "On Estimating Efficient Production Functions.“ Working_Paper No. A-85. (Chr. Michelsen Institute, Depart- ment of Humanities and Social Science, Bergen, Norway.) Girvan, Norman. 1980. “Swallowing the IMF Medicine in the 'Seven- ties.'" Development Dialogue 2 (July): 55-74. , Richard Bernal, and Wesley Hughes. 1980. "The IMF and the Third World: The Case of Jamaica, l974-80." Development Dialogue 2 (July): 111-155. Gold, Bela. 1979. Productivity, Technology and Capital. Lexington, Mass.: 0. C. Heath and Company. Green, W. H. 1980a. "Maximum Likelihood Estimation of Econometric Frontier Functions." Journal of Econometrics 13 (May): 27-56. . 1980b. "On the Estimation of a Flexible Frontier Produc- tion Model." Journal of Econometrics 13 (May): 101-115. Greenfield, S. M., and A. Strickon. 1981. "A New Paradigm for the Study of Entrepreneurship and Social Chan e." Economic Development and Cultural Change 29 (April?: 467-499. Griliches, A., and V. Ringstad. 1971. Economics of Scale and the Form of the Production Function. Amsterdam: North-Holland. Gujarati, Damodar N. 1978. Basic Econometrics. New York: McGraw- Hill Book Company. Haggblade, S., Jacques Defay, and Bob Pitman. 1979. "Small Manu- facturing and Repair Enterprises in Haiti: Survey Results." Michigan State University Rural Development Series, Working Paper No. 4. East Lansing: Michigan State UniversTty, Depart- ment of Agricultural Economics. Harcourt, G. C. 1972. Some Cambridge Controversies in the Theory of Capital. .Cambridge: Cambridge University Press, England. 302 Harper, Malcolm. 1978. Consultanoy for Small Business. London: Intermediate Technology Publications, Ltd. Harris, John R. 1970. "Nigerian Entrepreneurship in Industry." In Growth and Development of the Nigerian Economy. Edited by Carl Eicher and Carl Liedholm. East Lansing: Michigan State University. Heady, Earl O. 1946. "Production Functions from a Random Sample of Farm." Journal of Farm Economics 28 (November): 989-1004. , and Edwin G. Strand. 1955. "Efficiency within American Agriculture.” Journal of Farm Economics 37 (August): 524- 537. Henderson, James M., and Richard E. Quandt. 1980. Microeconomic Theory: A Mathematical Approach. 3rd ed. New York: McGraprill Book Company. Herdt, R. W., and R.EL Mandac. 1981. "Modern Technology and Economic Efficiency of Philippine Rice Farmers." Economic Development and Cultural Change 29 (January): 375-399. Hildebrand, G., and T. Liu. 1965. Manufacturing Production Func- tionsir1the United States, 1957. Ithaca: Cornell University Press. Hoard, Donald W. 1964. Management Factors Contributing to the Success or Failure of New Small Manufacturers. Ann Arbor: The University of Michigan. Hirsh, H. G. 1943. "Crop Yield Index Number." Journal of Farm Economics 25 (August): 583-598. Hock, Irving. 1958. "Simultaneous Equation Bias in the Context of the Cobb-Douglas Production Function." Econometrica 26 (October): 566-78. 1976. "Returns to Scale in Farming: Further Evidence." American Journal of Agnicultural Economics 58 (November): ' 745-749. Hoffman, L., and B. Weber. 1976. "Economies of Scale, Factor Intensities and Substitution: Micro Estimates for Malaysia's Manufacturing Industries." Review of World Economics 112 (June): 111-35. Hrems, Hans. 1968. Quantative Economic Theory: A Synthetic Approach. New York: JOthNiley and Son. 303 Huffman, W. E. "Decision Making: The Role of Education." American Journal of Agricultural Economics 56 (February): 85-97. Hull, 0. L., John T. Bosley, and Gerald G. Udell. l980. "Renewing the Hunt for the Heffalump: Identifying Potential Entre- preneurs by Personality Characteristics." Journal of Small Business Management 18 (January): 11-18. Hymer, S., and S. Resnick. 1969. "A Model of An Agrarian Economy with Non-Agricultural Activities." American Economic Review 59 (September): 493-506. Jamaica, Government of. 1979a. Employment, Earnings and Hours in Large Establishments, 1978. Kingston: Department of Sta- tistics. 1979b. Five Year Development Plan 1978-82, Main Document. Kingston: National Planning Agency. . 1979c. Statistical Abstract 1978. Kingston: Department of Statistics. 1980a. Consumer Price Indices, Annual Review 1979. Kings- ton: Department of Statistics. 1980b. The Labor Force. Kingston: Department of Sta- tistics. 1980c. National Income and Product. 1979. Kingston: Department of Statistics. 1981a. Consumer Price Indices, Annual Review 1980. Kingston: Department Of Statistics. 1981b. Economic and Social Survey, Jamaica 1980. Kingston: National Planning Agency. 1981c. Employment Earnings and Hours in Large Establish- ments, 1979. Kingston: Department of Statistics. 1981d. The Labor Force, 1980. Kingston: Department of Statistics. . 1981e. National Income and Product. Kingston: Department of Statistics. Johnson, Glenn L. 1960. "The State of Agricultural Supply Analysis." Journal of Farm Economics 42 (May): 441-442. 304 Johnson, Glenn L. 1964. "A Note on Nonconventional Inputs and Conventional Production Functions.: In Agriculture in Economic Development. Edited by Carl Eicher andTLawrence Witt. New York: McGraw-Hill Book Company. , Albert N. Halter, Harold R. Jensen, and D. Woods Thomas. 1961. A Study of Managerial Processes of Midwestern Farmers. Ames, Iowa: The Iowa State University Press. , and Lewis F. Zerby. 1976. "The Contribution of Bio in Ethics to Research on Problems Involving Food, Health, and Reproductive Biology." Agricultural Economics Staff Ropers, No. 76-2. East Lansing: Michigan State University, Depart- ment of Agricultural Economics. , and C. L. Quance, eds. .1972. The Ovenproduction Trap in U.S. Agriculture: A Study of Resource Al1ocation from World War I to the Late 1960s. Baltimore: The Johns Hopkins. Johnston, P. E. 1951. "Improving Labor Efficiency Through Improving Farm Organization." Journal of Farm Economics 33 (November): 808-817. Johnston, J. 1972. Econometric Methods. New York: McGraw-Hill Book Company. Junankar, P.11. 1980. "Tests of the Profit-Maximization Hypothesis: A Study of Indian Agriculture." The Journal of Development Studies 16 (January): 186-203. Kadivala, K. R. 1972. "Production Functions and Elasticity of Sub- stitution.“ Southern Economic Journal 38 (January): 281-284. Kearl, J. M. 1977. "Aggregate Production Fucntions: Some CES Experiments." Review of Economic Studies 44 (June): 305-20. Kelley, John C., and Luis Coronado. 1979. A Profile of Rural Small Industries in Selected Areas of Honduras. Tegucigalpa, Honduras, C.A.: The Agency for International Development. Kendrick, John W. 1961. Productivity Trends in the United States. Princeton, N.J.: Princeton University Press. Kennedy, Peter. 1979. A Guide to Econometrics. Cambridge: The MIT Press (MA.). Kieruff, Herbert E. 1975. "Can Entrepreneurs Be Developed?" .MSU Business Topics. East Lansing: Michigan State University, Graduate School of Business (Winter). 305 Kilby, P. 1962. The Development of Small Scale Industry in Eastern Nigeria. Lagos: United States Agency for International Development. , ed. 1971. Entrepreneurship and Economic Development. New York: The Free Press. King, John, and Philip Regan. 1976. Macmillan Studies in Economics: Relative Income Shares. New York: The Macmillan Press. Kirzner, I. M. 1973. Competition and Entrgpreneurship. Chicago: University of Chicago Press. Klein, Lawrence R. 1962. An Introduction to Econometrics. Engle- wood Cliffs, N.J.: Prentice-Hall, Inc. Kmenta, Jan. 1967. I'On Estimation of the CES Production Function." International Economic Review 8 (June): 180-189. 1971. Elements of Econometrics. New York: Macmillan Publishing CO., Inc. Knight, Frank H. 1921. Risk Uncertainty and Profit. New York: Houghton Mifflin. Koontz, Harold and Cyril O'Donnell. 1972. Principles of Management: An Analysis of Managerial Functions. New York: McGraw Hill Book Company. Lau, L. J. and P. A. Yotopoulos. 1971. "A Test for Relative Effi- ciency and Application to Indian Agriculture." American Economic Review 61 (March): 94-109. Lazarsfeld, Paul F. 1958. "Evidence and Inference in Social Research." DAEDALUS. (Journal of the American Academy of Arts and Sciences).87 (Fall): 99-130. Lecrew, D. 1978. "Determinants of Capacity Utilization by Firms in Low-Wage Countries." Journal of Development Economics 5 (June): 139-53. 1979. "Choice of Technology in Low-Wage Countries: A Nonneoclassical Approach." QuarterlynJournal of Economics 93 (November): 631-654. Leddin, Anthony. 1980. "Inefficiency in Irish Agriculture." Economic and Social Review (Dublin, Ireland) 11 (January): 127-136. 306 Lee, L. F., and M. M. Pitt. 1978. "Pooling Cross-section and Time- Series Data in the Estimation of Stochastic Frontier Produc- tion Function Models." Unpublished manuscript. , and W. G. Tyler. 1978. "The Stochastic Prodcution Function and Average Efficiency: An Empirical Analysis." Journal of Econometrics 7 (June): 385-389. Leff, Nathaniel. 1979. "Entrepreneurship and Economic Development: The Problem Revisited." Journal of Economic Literature 17 (March): 46-64. Leibenstein, H. 1966. "Allocative Efficiency vs. X-efficiency." American Economic Review 56 (June): 392-415. 1976. Beyond Economic Man. Cambridge, Mass.: Harvard University Press. 1977. "X-Efficiency, and Incomplete Information Use: A Comment." Economic Development and Cultural Change 25 (January): 311-316. 1978. “X-Inefficiency Xists--Reply to an Xorcist." American Economic Review 68 (March): 203-11. Levin, Richard I., and Charles A. Kirkpatrick. 1978. Quantative Approaches to Management. International Student Edition. Tokyo: McGraw-Hill Kogakusha, Ltd., Japan. Liedholm, Carl, and Enyinna Chuta. 1976. "The Economics of Rural and Urban Small-Scale Industries in Sierra Leone." African Rural Economy Paper No. 14. East Lansing: Michigan State University, Department of Agricultural Economics. Lockheed, Marlaine, Dean T. Jamison, and Lawrence J. Lau. 1981. "Farmer Education and Farm Efficiency: A Survey." Economic Development and Cultural Change 29 (April): 37-76. Mabawonku, Adewale F. 1978. "Incomes and Factor Productivity in Western Ni eria Small-Scale Industries." The Developing Economies CInstitute of Developing Economics, Tokyo, Japan) 16 (September): 269-282. Maddala, G. S. 1977. Econometrics. New York: McGraw-Hill Book Company. Mahony, T. A. 1961. The Identification of Management Potential. Minneapolis: University of Minnesota. 307 Makary, S. R., and H. Rees. 1981. "An Index of Management Effi- ciency for Egyptian Agriculture: A Case Study for Large Farms." Journal of Agricultural Economics 32 (May): 189-196. Malenbaum, Wilfred. 1941. "Indexes on A Type-Farm Basis." Journal of Farm Economies 23 (Auguat): 584-606. Marris, Peter, and Anthony Somerset. 1971. African Businessman: A Study of Entrepreneurship and Development in Kenya. London: Routledge and Kegan Paul. Marschak, J., and W. Andrews. 1944. "Random Simultaneous Equations and the Theory of the Production Function." Econometrica 12 (July-October): 143-205. Martin, Joe A. 1956. "Some Myths of Southern Economic Growth: A Study of Comparative Growth Rates in the Manufacturing Economy of the 11 South-Eastern States." Proceedings Journal of Farm Economics 38 (December): 1363-1374. McClelland, David C. 1961. The Achieving Society. Princeton, N.J.: D. Van Nonstrand Company. McFadden, 0. "Review of Hildebrand and Liu." 1967. Journal of American Statistical Association 62 (June): 295-300. . 1973. "Cost Revenue and Profit Functions." In An Econometric Appnoach to Production Theony. Edited By D. L. McFadden. Amsterdam: North-Holland Publishing Company. Meeusen, W., and J. van de Broeck. 1977.. "Efficiency Estimation from Cobb-Douglas Production Functions with Composed Error." International Economic Review 18 (June): 435-444. Meller, P. 1976. "Allocative Frontiers for Industrial Establish- ments of Different Sizes." Exploration in Economic Research. Occasional Papers of the NatTOnal Bureau of EtonomTC Research 3 (Summer): 379-407. Mijindali, Ndanusa Braimah. 1980. "Production Efficiency on Farms in Northern Nigeria." Ph.D. dissertation, Cornell University, Ithaca, New York. Moock, Peter R. 1981. "Education and Technical Efficiency in Small- Farm Production." Economic Development and Cultural Change 29 (July): 723-739. Morawetz, David. 1974. "Employment Implications of Industrializ- tion in Developing Countries: A Survey." Economic Journal 84 (September): 491-542. 308 Morris, Bruce R. 1967. Economic Growth and Development. New York: Pitman Publishing Corporation. Morse, Stanley. 1965. Modern Small Industry for Developing Coun- tries. New York: McGraw-Hill Book Company. MUller, Jergen. 1974. "On Sources of Measured Technical Efficiency: The Impact of Information." American Journal of Agricul- tural Economics 56 (November): 730-38. Mundlak, Yair. 1961. "Empirical Production. Function Free of Management Bias." Journal of Farm Economics 43 (February): 44-56. Nadiri, M. Ishaq. 1970. "Some Approaches to the Theory and Measure- ment of Total Factor Productivity: A Survey." Journal of Economic Literature 8 (Debember): 1137-77. Nafziger, E. Wayne. 1977. African Capitalism: A Case Study in Nigerian Entrepreneurship. Stanford: Hoover InstitutiOn Press. Nelson, Robert E. 1980. "Training Needs of the Small Enterprise Sector in Asian Countries." Journal of Small Business Manoge- ment 18 (October): 1-7. Pachino, Douglas. 1980. "Applying Efficiency Analysis to Small Farms in Low Income Countries: Some Theoretical and Impirical Considerations." Department of Agricultural Economics, Cornell University, Ithaca, New York, pp. 1-16. (Mimeographed.) Page, John M., Jr. 1979. "Firm Size, the Choice of Technique and Technical Efficiency: Evidence from India's Soap Manu- facturing Industry." Studies in Employment and Rural Development No. 59. Development Economics, Princeton University, Princeton, New Jersey. Page, John N., Jr. 1980. "Technical Efficiency and Economic Per- formance: Some Evidence from Ghana." Oxford Economic Papers 32 (July): 319-339. Paglin, M. 1965. "'Surplus' Agricultural Labor and DevelOpment-- Facts and Theories." American Economic Review 55 (September): 815-34. 1967. "Surplus Agricultural Labor and Development-~Facts and Theories: Reply." American Economic Review 57 (March): 202-O9. 309 Papanek, G. F. 1962. "The Development of Entrepreneurship." Ameri- can Economic Review 52 (May): 46-58. 1967. Pakistan's Development, Social Goals and Private Incentives. Cambridge, Mass.: Harvard University Press. Park, R. E. 1966. "Estimation with Heteroscedastic Error Terms." Economitrica 34 (October): 888. Pasour, E. C., Jr. 1981. "A Further Note on the Measure of Effi- ciency and Economics Of Farm Size." Journal of Agricultural Economics 32 (May): 135-145. , and J. B. Bullock. 1975. "Implications of Uncertainty for the Measurement Of Efficiency." American Journal of Agricultural Economics 57 (May): 335-339. Petrof, John V. 1980. "Small Business and Economic DevelOpment: The Case for Government Intervention." Journal of Small Business Management 18 (January): 51-56. Phillips, Alamari. 1963. "An Appraisal of Measures of Capacity." American Economic Review 53 (May): 275-292. Ramsey, James E., and Paul Zarembka. 1971. "Specification Error Tests and Alternative Functional Forms of the Aggregate Producation Function." Journal of the American Statistics Association 66 (September): 471-477. Rao, Potluir, and Roger L. Miller. 1971. Applied Econometrics. Belmont, Cal." Wadsworth. Richmond, J. 1974. "Estimating the Efficiency of Production." International Economic Review 15 (June): 515-521. Ringstad, Vidar. 1978. "Economies of Scale and the Form of the Production Function: Some New Estimates." The Scandinavian Journal of Economics 80 (September): 251-164. Rottenberg, Simon. 1953. "Notes on 'Economic Progress and Occupa- tional Distribution.'" Review of Economics and Statistics 35 (May): 168-170. Sahota, Gian S. 1968. "Efficiency of Resource Allocation in Indian Agriculture.“ American Journal of Agricultural Economics 50 (August): 584-605. Sato, Kazuo. 1975. Production Functions and Aggregation. Amsterdam: North-Holland PUBlishing Company. 310 Schickele, Rainer. 1941. "Effect of Tenure Systems on Agricultural Efficiency." Proceedings, Journal of Farm Economics 23 (February): 185-207. Schmidt, P. 1976. "On the Statistical Estimation of Parametric Frontier Production Functions." Review of Economics and Statistics 58 (May): 238-239. 1978. "On the Statistical Estimation of Parametric Frontier Production Functions: Rejoinder." Review of Economics and Statistics 60 (August): 481-482. , and C. A. K. Lovell. 1979. "Estimating Technical and Allocative Inefficiency Relative to Stochastic Production and Cost Functions." Journal of Econometrics 9 (February): 343-366. Schultz, Theordore W. 1947. "How Efficient is American Agriculture." Journal of Farm Economics 29 (August): 644-658. . 1964. Transforming Traditional Agriculture. New Haven, Conn.: Yaié University Press. 1975. "The Value of the Ability ot Deal with Disequili- bria." Journal of Economic Literature 13 (September): 827- 46. Schumacher, E. F. 1975. Small is Beautiful: Economics as if People Mattered. Perennial Library. New YOrk: Harper anleow. Schumpeter, Joseph A. 1934. The Theory of Economic Development: An Inquiry into Profits, Cgpital, Credit, Interest, and the Busihess Cycle. Cambridge, Mass.: Harvard university Press. Reprint Ed. 1959. Seitz, W. D. 1970. I'The Measurement of Efficiency Relative to a Frontier Production Function." American Journal of Agricul- tural Economics 52 (November): 505-511. Shapiro, K. H., and J. Muller. 1977. "Sources of Technical Effi- ciency: The Roles of Modernization and Information." Economic Development and Cultural Change 25 (April): 293-310. Slater, M. 1980. "The Managerial Limitation to the Growth of Firms." The Economic Journal 90 (September): 520-528. Solow, Robert M. 1957. "Technical Change and the Aggregate Produc- tion Function." Review of Economics and Statistics 39 (August): 312-20. 311 Solow, R. M. 1964. "Capital, Labor, and Income in Manufacturing." In National Bureau of Economic Research Conference on Research in Income and Wealth, The Behavior of Income Shares (Studies in Income and Wealth, 26). Princeton: Princeton University Press, pp. 101-128. Spencer, Dunstan S. C., Ibi I. May-Parker, and Frank S. Rose. 1976. "Employment, Efficiency and Income in the Rice Processing Industry in Sierra Leone.“ African Rural Employment Paper No. 15. East Lansing: Michigan State University, Department of Agricultural Economics. Stanley, Eugene, and Richard Morse. 1965. Modern Small Scale Industry for Developing,Countries: Evidence from Ghana. New York: McGraw-HaiiiBook Company. Stigler, G. J. 1976. "The Xistence of X-efficiency." American Economic Review 66 (March): 213-216. Stogdillis, Ralph M. 1948. "Personal Factors Associated with Leader- ship: A Survey of the Literature." Journal of Psychology 25 (January): 35-71. Thornton, D. S. 1964. "The Organization of Production in the Irrigated Areas of the Sudan." Journal of Agricultural Economics 16 (December): 286-295. Tidrick, Gene M. '1975. "Wage Spillover and Unemployment in a Wage-Gab Economy: The Jamaican Case." Economic DevelOpment and Cultural Change 23 (January): 306-324. Timmer,C. P. 1970. "On Measuring Technical Efficiency." Food Research Institute Studies 9 (#2): 98-171. 1971. "Using a Probablistic Frontier Production Function to Measure Technical Efficiency." Journal of Political Economy 79 (July/August): 776-794. Toda, Y. 1976. "Estimation of a Cost Function When the Cost is not Minimum: The Case of Soviet Manufacturing Industries, 1958-71." Review of Economics and Statistics 58 (August): 259-268. . 1977. "Substitutability and Price Distortion in the Damand for Factor Of Production: An Empirical Estimation." Applied Economics 9 (June): 203-217. Todd, D. 1971. "The Relative Efficiency of Small and Large Firms." Committee of Inquiry on Small Firms, Research Report No. 18 ‘(HMSO, London). 312 Trosper, R. L. 1978. "American Indian Relative Ranching Efficiency." American Economic Review 68 (September): 503-516. Tsang, H. H., and Joseph, J. Persky. 1975. "On the Empirical Con- tent of CES Production Functions." The Economic Record 51 (December): 539-548. Tyler, William G. 1979. "Technical Efficiency in Production In a Developing Country: An Empirical Examination of the Brazilian Plastics and Steel Industries." Oxford Economic Papers 31 (November): 477-495. Upton, Martin. 1964. "A Development of Gross Mar in Analysis." Journal of Agricultural Economics 16 (June?: 111-119. . 1979. "The Unproductive Production Function.” Journal of Agricultural Economics 30 (May): 179-191. Uzawa, H. 1962. "Production Functions with Constant Elasticities of Substitution." The Review of Economic Studies 29 (October): 291-299. Vincent, Warren H. 1962. Economies and Manngement in Agriculture. Englewood Ciffs, N.J.: Prentice Hall, Inc.. Wallace, T. and V. Asher. 1972. "Sequential Methods in Model Con- struction.“ Review of Economics and Statistics 54 (May): 172-78. Wallace, Dudley T. 1977. "Pretest Estimation in Regression: A Survey." American Journal of Agricultural Economics 59 (August): 431-443. Walters, Alan A. 1963. "Production and Cost Functions: An Econ- ometric Survey." Econometrica 31 (January-April): 1-66. Watanabe, Susumu. 1970. "Entrepreneurship in Small Enterprises in Japanese Manufacturing." International Labour Review 102 (December): 531-576. Welch, F. 1970. "Education in Production." Journal of Political Economy 78 (January/February): 35-59. White, Lawrence J. 1978. "The Evidence on Appropriate Factor Pro- portions for Manufacturing in Less DevelOped Countries." Economic Development and Cultural Chonge 27 (October): 27-59. Winston, Gordon C. 1974. I'The Theory of Capital Utilization and Idleness." Journal of Economic Literature 12 (December): 1301-20. 313 Working, E. J. 1940. "Crop-Yield Index Number." Journal of Farm Economics 22 (November): 701-713. Yotopoulos, P. A. 1967. "From Stock to Flow Capital Inputs for Agricultural Functions: A Microanalytic Approach." Journal of Farm Economics 49 (May): 476-491. , and Lawrence J. Lau. 1973. "A Test for Relative Economic Efficiency: Some Further Results." American Economic Review 63 (March): 214-223. , and Jeffrey B. Nugent. 1976. Economics of Development: Empirical Investigations. New York: Harper and Row. Zak, Thomas A., Cliff J. Huang, and John J. Siegfried. 1979. "Production Efficiency: The Case of Professional Basketball." The Journal of Business 52 (July): 379-392. Zellner, A. 1962. "An Efficient Method for Estimating Seemingly Unrelated Regressions and Tests for Aggregation Bias." Journal of American Statistical Association 57 (June): 348-368. , J. Kmenta, and J. Dreze. 1966. "Specification and Estima- tion of Cobb-Douglas Production Function Models." Econ- ometrica 34 (October): 784-795. Zerembka, P. 1970. I'On the Empirical Relevance of the CES Produc- tion Function." Review of Economics and Statistics 52 (February): 47-53.