0A PC The Location of Jobs 0~~~~~~~ f 0~~~ il aMetroploisg C, Patterns of Growth in Bogota anr Ci , Cl0omibia Kyu SLk Lee 0 ex 0 a ~~~~~A World Bank Research Pu bl cation : I 7 I/- The Location ofJobs in a Developing Metropolis A World Bank Research Publication The Location ofJobs in a Developing Metropolis Patterns of Growth in Bogotd and Cali, Colombia Kyu Sik Lee PUBLISHED FOR THE WORLD BANK Oxford University Press Oxford University Press NEW YORK OXFORD LONDON GLASGOW TORONTO MELBOURNE WELLINGTON HONG KONG TOKYO KUALA LUMPUR SINGAPORE JAKARTA DELHI BOMBAY CALCU`TrA MADRAS KARACHI NAIROBI DAR ES SALAAM CAPE TOWN © 1989 The International Bank for Reconstruction and Development / THE WORLD BANK 1818 H Street, N.W., Washington, D.C. 20433, U.S.A. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Manufactured in the United States of America FiTst printing March 1989 The findings, interpretations, and condusions expressed in this study are the results of research supported by the World Bank, but they are entirely those of the authors and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The maps that accompany the text have been prepared solely for the convenience of readers; the designations and presentation of material in them do not imply the expression of any opinion whatsoever on the part of the World Bank, its affiliates, or its Board or member countries concerning the legal status of any country, territory, city, area, or of its authorities, or concerning the delimitations of its boundaries or its national affiliation. Library of Congress Cataloging-in-Publication Data Lee, Kyu Sik. The location of jobs in a developing metropolis: patterns of growth in Bogota and Cali, Colombia / Kyu Sik Lee. p. cm. - (A World Bank research publication) Bibliography: p. Indudes index. ISBN 0-19-520786-6 : $18.95 1. Labor supply-Colombia-Bogoti. 2. Labor supply-Colombia- -Cali. 1. Title. I. Series. HD5757.B6L44 1989 331.12'098614-dcl9 88-34298 CIP Contents Preface vii 1. Introduction and Summary 1 Urban Policies and the Needfor Research on Employment Location I The City Study 3 Summary of Findings and Policy Implications 4 Notes 10 2. Employment Location by Industry Group 11 Employment Composition by Industry and Firm Size 11 Spatial Distribution of Employment 16 Decentralization of Employment 27 Patterns of Movement and Incidence of Births 29 Summary 31 Appendix: Data Comparisons 33 Notes 37 3. Location Patterns of Manufacturing Employment 39 Comparison of Bogotd and Cali with U.S. Cities 39 Decentralization Trends 43 Components of Change in the Spatial Distribution of Employment 49 Location Patterns of Industries 55 Summary 67 Notes 68 4. Determinants of Manufacturing Employment Location in Bogota 70 The Sample 70 Selected Findings of the Survey 71 v vi Contents Summary 85 Notes 87 5. A Model of Manufacturing Employment Location 88 Theoretical and Empirical Frameworkfor Modeling Employment Location 89 Multinomial Logit Models in Urban Economic Research 90 Results of Multinomial Logit Estimation 91 Summary 95 Appendix: Theoretical and Empirical Specification of the Model 96 Notes 99 6. A Model of Employment Location for Retail Trades and Services 105 Models of Market Potential and Residential Access 106 Empirical Findings 107 Summary 116 Note 116 7. The Spatial Structure of Production 117 Estimating the Elasticity of Capital-Land Substitution 117 Estimating the Elasticity of Substitution between Land and Other Inputs 118 Empirical Evidence on Land Price Gradient 120 Empirical Evidence on Wage Gradient 122 Summary 124 Notes 125 Appendix A: Stastical Tables 126 Appendix B: Questionnaire for the Survey of Manufacturing Establishments 160 Bibliography 174 Index 181 Preface This book is part of a World Bank Research program known as the City Study that examined five urban sectors-employment location, housing, labor markets, transport, and public finance-in Bogota and Cali, Colombia. The goal of the program was to gain a better understanding of how urban policies and projects affect cities in developing countries. I am grateful to Douglas Keare, who invited me to participate in the City Study and gave me encouragement and support throughout the study period. Gregory K. Ingram, who directed the City Study, was responsible for leading me to this previously uncultivated field of employment location. This study owes much to his counsel and professional guidance. Anthony A. Churchill, at that time director of the World Bank's Urban Develop- mnent Department, provided continuous moral support for this research effort and helped to integrate our results with the Bank's operational work. Undertaking this research project would not have been possible without the unstinting assistance and support given by Ramiro Cardona and Jose Fernando Pifieda at the Corporaci6n Centro Regional de Poblacin (CCRP), the collaborating institution for the City Study. My spe- cial thanks go to Maria Clara de Posada, who single-handedly conducted the survey of establishments, visiting scores of firms all over BogotA. I am also grateful to the officials at the national statistical agency (DANE) and the social security agency (Iss) and to local government officials in Cali, who gave us access to the original data files and assisted our work. The studv benefited from valuable comments and suggestions given by the members of the Advisory Committee-Eduardo Aldana, Rodrigo Botero, Pedro G6mez, Miguel Urrutia, and Eduardo Wiesner-and by Rakesh Mohan and other members of the research team. The BogotA Chamber of Commerce sponsored a series of public meetings on the results of the study and published the papers in its Revista. At an early stage I benefited from discussions with Edwin S. Mills, John R. Meyer, vii viii Preface Roger Schmenner, and George Tolley. Many other persons not men- tioned here also provided comments and suggestions during the study period. Cleaning and processing the large microdata files before analyzing them was tedious and time-consuming work. For efficient research assis- tance, I am indebted to Nelson Valverde, Yoon Joo Lee, Wilhelm Wagner, and Leslie Kramer. I would like to thank Morallina Fanwar George for typing and preparing the original manuscript and many tables and Suzana B. Jesus and Armi Felix for retyping and finalizing the material. I dedicate this book to Jai and Cathy. J Introduction and Summary During the past several decades, the population of cities in developing countries has grown rapidly. Even though the growth of many large cities in middle-income developing countries has slowed in recent years, most large cities in Sub-Saharan Africa and other low-income regions are still growing rapidly. The urban population in all developing countries is pro- jected to grow in the next two decades about four times as fast as in the developed countries. Until 1950 Buenos Aires was the only city in the developing world with more than 5 million people; now there are more than fifteen such cities. By the year 2000 the developing world will have forty cities with 5 million or more people; twenty of them are expected to have more than 10 million people. Urban Policies and the Need for Research on Employment Location This rapid urbanization in developing countries has generated enormous pressure for policymakers to control or direct urban growth and to ex- pand public services. In the middle-income countries of Latin America and East Asia (including Brazil, Mexico, Venezuela, the Republic of Korea, and the Philippines) and in a number of countries in Africa and South Asia (such as Egypt, Nigeria, India, and Pakistan), policymakers have used many approaches to cope with the rapid growth of cities. Often the governments of these countries have initiated specific policies to decentralize economic activity and move it to the periphery of met- ropolitan areas. Managing large municipalities is a complex and difficult task, mainly because the process of rapid growth and its effects on the functioning of large cities are not well understood. To deal efficiently with the problems of a rapidly growing city requires a policv framework that links a sound urban development strategy with good investment programs. Having I 2 introduction such a policy framework in turn requires a good understanding of the trends in urban development and how the workings of the land and labor markets and the location behavior of individual households and firms contribute to such trends. Government policies and programs tend to be inefficient and costly if they attempt to reverse observed trends and behavior. Many urban policies have been intended to influence the location pat- terns of employment. This is not surprising since "the true determinants of urbanization and spatial concentration in developing countries are found in the forces that determine the location of employment oppor- tunities: the nature and pattern of industrialization, the pace of agri- cultural development, and the growth of transportation and communica- tion networks" (World Bank 1979, p. 76). Within an urban area, work- place locations tend to influence residential locations and travel patterns. Therefore policies that influence employment location will affect the overall spatial development patterns of the urban area. Particular atten- tion has been given to the location patterns of manufacturing activities on the assumption that manufacturing is the driving force behind regional development. This assumption is in line with the theory that the develop- ment of certain manufacturing activities determines the development of other industries in a region and is followed by population growth and then by the emergence of service sector activities (Lowry 1964). In developing countries, employment location policies for cities have taken various forms: strict zoning regulations, outright prohibition of cer- tain economic activities in particular areas, and various fiscal and finan- cial incentives to induce industries and population to particular areas. The probable impact of these policy measures is uncertain mainly because of the lack of information on trends of employment location pat- terns within cities of developing countries and the lack of understanding of individual firms' location behavior in response to operations of land and other markets. Empirical work on the trends of employment location within cities of developing countries is virtually nonexistent in the literature. Even for developed countries, theoretical and empirical literature on employment location behavior is so rare that it is difficult to anticipate how policies will influence the location patterns of economic activity. In a review of current research in urban economics, Mieszkowski and Straszheim (1979, p. xiii) observe: "Among the most prominent issues not covered in this volume is the relative importance of different factors in employment location and decentralization patterns within metropolitan areas, which deserves far more research." In most urban models-for example, Lowry (1964) and others that followed-the location of manufacturing and other basic sec- Urban Policies and Employment Location 3 tor employment has been assumed to be exogenously determined owing to the paucity of literature on employment location behavior. For predict- ing the future spatial structure of urban areas, however, it is essential to understand the changing patterns of employment location and the factors that influence such changes. The City Study A World Bank research project known as the City Study explored the workings of five urban sectors-housing, transport, employment loca- tion, labor markets, and public finance-to provide analytical tools to assess better the impact of urban policies and investment programs in developing countries. This book presents the results obtained from the research on employment location. The strategy of the City Study was to gather information on the behavior of individual decisionmaking units such as households and firms. In the area of employment location, the main research focus was on (a) documenting the trends in location patterns of employment, (b) analyzing the components of those trends, such as the location patterns of newly established firms and relocating firms, and (c) modeling location decisions of individual firms to understand how the responses of in- dividual firms collectively result in the observed trends.' Undertaking a micro-level multiyear study of employment location seemed more problematic than in the case of the other four sectors. The risk was reduced substantially, however, by our choice of Bogota and Cali as the cities to study. Both offered a rich body of available data as well as local interest and support for the study. Bogota, with a population of nearly 4 million in 1980, had median characteristics of a cross-section of large Latin American cities (Ingram and Carroll, 1981) and was represent- ative of these cities in many respects. Cali, with about 1 million people, was included in the study as a comparator to test the transferability of research results. During the study period in the mid-I 970s Colombia was in a cyclical upswing, and both cities were experiencing rapid growth. Five sets of data were used for the employment location component of the City Study: (a) the annual industrial directory files of DANE (Depar- tamento Administrativo Nacional de Estadistica), Colombia's national statistical agency, for 1970-75; (b) the files of the Colombian social security agency for data on establishments in 1976 and 1978; (c) the World Bank-DANE household survey of 1978 conducted as part of the City Study; (d) the household survey of Phase II of the Bogota Urban Develop- ment Study collected in 1972; and (e) the employment location survey of manufacturing establishments conducted in 1978 as part of the City 4 Introduction Study. The household surveys and the social security data are briefly de- scribed in the appendix to chapter 2. All data files are documented in detail in Y. J. Lee (1979) and Valverde (1978). The most important data base for undertaking this study was DANE'S an- nual industrial directory files, which included the address, the Standard Industrial Classification (sic) code, and employment size for all manufac- turing establishments with 10 or more employees. The City Study em- ployment location survey was conducted for a sample of 126 es- tablishments, randomly selected from more than 2,600 establishment records for Bogota. The survey questionnaire appears as appendix B at the back of the book. Summary of Findings and Policy Implications Although empirical studies on employment location for cities in de- veloped countries are rare, the results for New York obtained by Hoover and Vernon (1959) and those of Struyk andJames (1975) and Schmenner (1982) for other U.S. cities offer a set of stylized facts in the employment location literature. The City Study data on the trends of employment location and on the behavior of firms in Bogota and Cali confirm the stylized facts obtained for North American cities; they are also consistent with and reinforced by the similar results obtained for Seoul (see note I below). The fact that most of the trends and behavioral relations docu- mented in the City Study are similar to those found in developed coun- tries suggests that cities exhibit strong regularities in their development patterns and that our research findings are transferable to cities in other countries. Location Patterns of Employment Are Changing Rapidly The degree of employment location dynamics (that is, changes at the margin) in cities in developing countries was unknown until this study of Colombian cities, which utilized location-specific micro data on es- tablishments. In terms of birth, death, and relocation rates of firms, Bogota and Cali experienced a much higher degree of employment loca- tion dynamics than those of large U.S. cities previously studied (Struyk and James 1975).2 The annual birth rate of manufacturing firms in Bogota at 8.8 percent was much higher than that in large U.S. cities. Phoenix, with a birth rate of 7.6 percent, came closest to that of BogotA and is comparable to that of Cali. The birth rate was higher than the death rate for BogotA, Cali, and Phoenix while the opposite was true for Boston, Minneapolis, and Cleveland. The annual moving rates of firms were Summaerv of Findings 5 similar, ranging from 3 to 5 percent among the North American and Colombian cities compared. This means that in a ten-year period as much as half of all manufacturing firms in these cities would be relocated. The locations chosen by relocating and new firms will influence the spatial structure of an urban area because the location of employment in- fluences residential locations and travel patterns. Therefore understand- ing the location patterns of new, discontinued, and relocating firms will help predict the trends correctly and avoid wrong public investments. Jobs Are Moving Outward from Central Areas The data revealed evidence of employment decentralization in both Bogota and Cali. In Bogota the finance sector alone was not decentraliz- ing, while in Cali neither the commerce nor the finance sector con- tributed much to the decentralization trend. In both cities decentraliza- tion increased progressively in commerce, services, and manufacturing, in that order. For our analysis, Bogota was divided into six concentric rings around the central business district (CBD) and Cali into five rings. In both cities manufacturing employment grew at an accelerating rate with the distance from the CBD. In Bogota the outer ring gained employment by 16 percent a year while the CBD lost by 2 percent a year.3 The strong decentralization of manufacturing employment in these cities is comparable to the trend observed for large U.S. cities during the past several decades (Hoover and Vernon 1959; Leone 1971). Yet the CBD in both cities continues to retain retail and service employment and attracts large commercial and finan- cial establishments. Consequently, the absolute number of jobs in the CBD has remained roughly constant, and the CBD'S share of all employ- ment has fallen over time. In 1978 the CBD in both cities had about a sixth of all city employment. This pattern indicates that cBD-destined travel is decreasing as a share of total travel and that the existing radially oriented transit system should be supplemented with circumferential services. Small Firms Create MoreJobs Than Do Large Ones and Do So in Central Areas In Bogota and Cali the birth of new firms created many more jobs than did the growth of mature firms (60 percent more in BogotA and 80 per- cent more in Cali). This is consistent with findings on North American cities by Birch (1979).' Most new firms were small, with less than 25 em- ployees. In both Bogota and Cali small, new firms tend to locate in central 6 Introduction areas, a tendency that supports the incubator hypothesis: certain central areas of large cities have a special function that is vital to the economy and not easily transferable to outlying areas or to smaller cities. Small, new firms whose spatial needs are still modest choose to locate in established areas despite higher rents in order to benefit from the availability of skilled labor, markets and business services, and shared delivery services. As these new firms grow and expand, spatial constraints become more significant, and thev tend to move out to where more space is available. This hypothesis was developed and tested first for New York by Hoover and Vernon (1959) and for other U.S. cities by Struyk and James (1975). Indeed, our analysis of the data revealed that subareas contiguous with the CBD showed the characteristics of an incubation area in both Bogota and Cali. The data on Seoul also identified such areas near the CBD, the central market, and the old industrial district (K. S. Lee 1985b). Crowded central areas thus seem to be good at "hatching" new industries in these cities. Firms Move, but Not Long Distances In both Bogota and Cali 40 percent of relocating firms stayed within their original subareas; those in Bogota moved an average distance of less than 2 kilometers. The majority of relocating firms were small, and they moved only a short distance from their previous location. Only a few large firms moved long distances. This confirms similar findings on Chicago by Moses and Williamson (1967). The Korean data show that firms rarely moved out of the Seoul region. Only 7 percent of those that relocated went outside the region. Although the moving distance in- creases with the firm size, the firms tend to move without affecting the delivery distance for inputs or output and the commuting distance of employees. Different Types of Firms Respond to Location Attributes Differently The survey data were analyzed to determine the probability that a profit- maximizing firm of a particular type would occupy a site with particular attributes. The results show that the accessibility to the local input and product markets and the commuting distance of production workers are the most important factors in the location choice of small manufacturing firms. For these firms, the benefits from various externalities tend to com- pensate for high rent and congestion costs in the central area, as the in- cubator hypothesis suggests. Large firms tend to be more export-oriented Summars of Findings 7 (from the region) and require more space for modern assemblv-line pro- duction technology. For these firms, land and plant space, available at lower cost in outer areas, are more important than access to local markets. Such locations also offer better access to the road network. It was striking to find that almost all shipments of inputs and final products are made bv truck; rail is seldom used, according to our survey results. Firms Respond to the Land Market and Resulting Location Patterns Are Not Random Regressions based on the same survey data indicate a strong relationship between the intensity with which labor and capital are used and the price of land; that is, land is used with greater intensity when land prices are high. This implies that in rapidly growing cities in developing countries, manufacturing firms respond to the substitutability of land with respect to other inputs in making location decisions. The analvsis of trade and service employment location shows strong regularities between employ- ment densities and the buying power or market potential measured in terms of population, households, and income level. Firms in trade and services choose a location that maximizes their access to customers weighted by purchasing power. Changes in the location patterns of em- ployment in cities such as Bogota and Seoul are therefore bv no means random. Policy Implications The strong decentralization of employment in Bogota and Cali is com- parable to the trend observed for large U.S. cities during the past several decades. This trend was strongest for manufacturing employment as firms relocated away from the CBD and as new firms were established and existing firms expanded in outlying areas. This trend was an outcome of location choices of individual firms in response to operations of land and other markets without explicit government policies for decentralization. Even in the case of Seoul, where strong incentive schemes and relocation orders were implemented, a surprisinglv small proportion of firms moved in response to government actions. Newly established firms ac- counted for nearly 80 percent of new jobs created there, but they were lit- tle affected by the government incentive schemes (K. S. Lee 1985b). The main aim of U.S. policy for intrametropolitan decentralization has been quite different from that in developing countries. In the United States the main objective has been to reduce urban decentralization in the hope of preventing decay in the central city. Although this objective is 8 Introduction based on the belief that decentralization occurs because of deteriorating conditions in the central city, little empirical evidence supports this view. Most decentralization in U.S. cities is attributed to improvements in transport and communication, suburban development programs, and federal subsidies to home ownership (Muth 1969). Many federal pro- grams, such as urban renewal programs that lower housing density in the central city, have expedited rather than reduced the trend toward decen- tralization. Attempting to reverse the tide of decentralization often results in economic inefficiency since improvements in transport and com- munication have reduced the central city's comparative advantages for production and other economic activities (K. S. Lee 1985a). Evidence from Bogota and Cali does not suggest an increasing concen- tration of low-income population in the central city, and there are few signs of central city decay. In fact, attempts to decentralize economic ac- tivitv in developing countries stem mainly from the increasing concentra- tion of economic activity in the central city, accompanied by congestion and pollution as the city's population grows rapidly. But policies to decentralize population and economic activity are probably not good substitutes for better internal management of city growth. For example, reducing the population or employment in a large city by a certain amount is likely to have very little effect on air pollution or traffic conges- tion (Tolley 1979; Henderson 1980). The principal question is how to guard against poorly conceived spatial policies when decentralization is already prevalent because individual firms are making location choices in response to operations of land and other markets. Since employment decentralization can have a significant impact on the patterns of residen- tial location, commuting, and hence overall urban development, inap- propriate spatial policies could seriously reduce the welfare of the economy. A commonly observed but potentially damaging spatial policy is the outright legal prohibition of new manufacturing activities within large cities. Several developing countries, including India, Korea, and Vene- zuela, have implemented such a measure as part of their decentralization policy. Nevertheless, our findings support the-incubator hypothesis that small new firms tend to locate in the central areas where thev have easv access to markets and services. Such a government action to slow further growth of the city would therefore kill the incubator effect of the city's agglomeration economies. Since governments tend to encourage young industries to start up away from large city centers, the important policy question is whether the incubation function could be replicated in outly- ing areas of large cities. The evidence, however, shows that firms do not move long distances and small firms tend to stay within the central area. Summary of Findings 9 For example, only 7 percent of moving firms relocated from Seoul to other regions (K. S. Lee 1985b). An attempt to create the incubator func- tion in outlying areas for small firms could be difficult and prohibitively costly. A related policy question is how to strengthen the incubator func- tion of secondary cities where indigenous manufacturing industries could be "hatched" and helped to grow. This could be a potentially viable policy that would lessen the pressure of growth in the major urban center. The findings from the study on Seoul suggest that the costs of govern- ment subsidies to compensate small firms for moving long distances will be extremely high. In 1978 the Korean government established a new in- dustrial town at Banweol, less than 30 kilometers from Seoul, to draw small and medium-size firms away from the capital. One thousand plant sites were initially prepared, but the occupancy rate remained low until the size restriction on firms was removed several years later. The most serious problems faced by many firms that moved to Banweol were that product markets and input suppliers were not readily accessible, produc- tion workers were reluctant to move to Banweol or commute from Seoul, and day-to-day business information was difficult to obtain because of poor telephone service and the lack of person-to-person contacts. That such a seemingly short distance thwarted the development of Banweol was striking. In contrast, firms pay a penalty of 500 percent of local taxes after moving to Bucheon, a fast-growing industrial city be- tween Seoul and Incheon where manufacturing is discouraged. Many are willing to pay this penalty because the opportunity cost of moving to government-supported areas such as Banweol is even higher (K. S. Lee 1985a). Logit analysis used to study the Bogota data helps explain the Korean experience: small firms prefer central locations. Accessibility to local markets and proximity to production workers are the most impor- tant attributes for them. Analytical results show that the government will induce less dead- weight loss (that is, minimize social cost) by subsidizing inputs that make up a large share of the firm's costs and that are poor substitutes for other inputs. Such a scheme will result in less distortion in the mix of inputs. Similarly, the government will induce less deadweight loss by providing infrastructure that is highly valued by the firms and easily substituted for other inputs used by the firms. Public expenditure on such an item will enable the firm to reduce its private outlay (Murray 1988). From the logit analysis of the BogotA data, it is apparent that government policies inten- ded to influence employment location patterns can be effective and in- duce the least amount of welfare loss if they influence the site attributes that particular types of firms deem important. Sound spatial policies re- 10 Introduzction quire a good understanding of firms' location behavior and the resulting trends of location patterns. Providing infrastructure alone may not attract firms to a newly developed area. Notes 1. Because explicit policy instruments were not implemented in Bogota and Cali, measuring the actual effects of employment location policies is the focus of another World Bank research project that evaluated various industrial location policies in Korea as a sequel to the research reported here (see K. S. Lee 1985a, 1985b, 1985c, and 1986; Choe and Song 1984). 2. A "birth" is the establishment of a new firm, a "death" is the termination of an existing firm, and a relocation is the move of an existing firm. 3. The results for Seoul for 1973-78 show that the change was several times greater there, possibly because of Korea's faster industrialization. In Seoul's CBD manufacturing employment fell by 8 percent a year while in the outermost ring it rose by 34 percent a year (K. S. Lee 1985b). 4. Birch's analysis of the U.S. data showed that roughly 50 percent of new jobs was due to the birth of new firms and that small firms with 20 or fewer employees generated 66 percent of all new jobs in the United States between 1969 and 1976. 2 Employment Location by Industry Group Even though the analyses presented below are descriptive, the results need some qualification since they are based on data sets that are not necessarily consistent. Nevertheless, the analyses reveal a clear pattern of employment decentralization in both Bogota and Cali, a trend also oc- curring in cities in developed countries. The three main data sets used in this chapter are the DANE 1978 household survey for BogotA and Cali, the 1972 Phase II household sur- vey for BogotA, and the 1976 social securitv file for establishments in Cali and the 1978 file for Bogota. Although the social security data are from establishments, they provide poor coverage of small firms. DANE'S 1978 household survey, however, did obtain information on the respondents' firms, including present workplace location, number of employees, the initial year of operation at the present location, the previous location if any, and type of business. By comparing the household survey results with the social security data, it was thus possible to make certain inferen- ces about the characteristics of small firms as well as large ones. Employment Composition by Industry and Firm Size As table 2-1 indicates, manufacturing, commerce, finance, and services accounted for 85 percent of employment in both BogotA and Cali in 1978. The commerce sector consists mainly of retail establishments and wholesalers, while the finance sector includes banking, insurance, and real estate offices. BogotA's share of manufacturing employment (24 percent) is similar to that in other cities in developing countries, as shown in table 2-2. In Cali, however, the share of 31 percent is closer to that of large U.S. cities and reflects the large number of large-scale joint venture manufacturing operations. Employment shares in commerce-about 20 percent for both BogotA and Cali-are comparable to those of other cities in the II Table 2-1. Population and Employment by Industry Group Bogqotd Colombia, 19 7 3' 1975b 1978L Coti, 1978c Number Number Number Number (thousands) Percent (thousands) Percenit (thousands) Perceni (thousands) Percent Population and employmeni Population 19,735d _ 2,894e - 3,439f - 1,055 "° Economically active 5,975 30.28g 1,023 35.35g n.a. n.a. n.a. i.a. Employed 5,118 - 919 - 1,212 - 363 Unemployed 857 14.34" 104 10.17"7 n.a. n.a. n.a. n.a. Employment by sector Agriculture 1,546 30.21 12.6 1.37 13.0 1.07 3.7 1.03 Mining 36 0.70 5.6 0.61 5.0 0.41 1.6 0.44 Manufacturing 678 13.25 201.5 21.92 286.2' 23.62 112.9 31.15 Electricity, gas, water 21 0.41 5.3 0.58 6.2 0.51 3.3 0.91 Construction 200 3.91 72.6 7.90 87.1 7.19 21.6 5.95 Commerce 576 11.25 186.9 20.33 246.3 20.33 78.3 21.59 Transport and communication 167 3.26 51.8 5.64 69.7 5.75 22.4 6.19 Finance 92 1.80 59.7 6.49 98.3 8.11 13.2 3.65 Services 838 16.38 321.8 35.01 394.5 32.56 105.0 28.96 N.i.e. 964 18.83 1.4 0.15 5.8 0.48 0.4 0.11 Total 5,118 100.00 919.2 100.00 1,212.0 100.00 362.6 100.00 -Not applicable. rn.a. Not available. 4 N.i.e. Not included elsewhere. a. The 1973 census, published in DANE (1975a). b. The 1975 household survey, published in DANE (1976a). C. DANE household survey, 1978. d. Not adjusted for the apparent undercoutit. e. The corresponding figure in the 1973 population census was 2,811. f. The sudden increase in manufacturing employment as well as population during 1975-78 iniplies that the two sets of survey results are not directly com- parable, probably because of sampling errors or expansion methods. g. Labor force participation rate with respect to total population. h. Unemployment rate. Source: DANE. Table 2-2. Employment Share of Major Sectors in Selected Cities (percentage of total employimient) Kuala Bogold Cali Seoul Lumpur Manila Abidjan Tunis U.S. small U.S. large Sector (1978) (1978) (1970) (1970) (1970) (1970) (1972) (1960) (1960) Manufacturing 23.6 31.2 22.7 20.5 22.1 22.5 18.9 25.1 30.2 Commerce 20.3 21.6 28.9 17.6 13.8 18.6 17.5 28.4 32.6 Servicesa 40.7 32.6 33.0 35.5 37.3 42.8 39.2 28.8 24.8 a. Includes financial services. Source: Figures for those other than Bogota and Cali are from Renaud (1981). Employment Composition 15 developing world; in contrast, the share of service employment in Bogota is somewhat higher (41 percent including finance), while in Cali it is somewhat lower (33 percent). The data in table 2-3 indicate that a little over half the jobs in both BogotA and Cali are in small firms (less than 10 employees). Across indus- try groups, BogotA has a higher proportion of manufacturing and finan- cial employment in large firms; the reverse is true for commerce and serv- ices. In Cali, however, a higher proportion of financial employment is in small firms. In addition, the proportion of Cali's employment in small firms in the commerce and service sectors-about three-quarters of all jobs-is much larger than Bogota's. Large firms also account for a greater proportion of manufacturing employment in Cali than in BogotA. It is thus likely that large manufacturing establishments play an impor- tant role in shaping overall patterns of employment in Cali, while large- scale financial and service firms play a significant role in BogotA's development. Although many jobs exist in self-employed or family businesses (firms with fewer than 5 employees), a large percentage are in construction and domestic work and therefore tend to be temporarv.' The location patterns of these jobs and the influence of public policy on them are thus difficult to study. Table 2-3. Employment by Firm Size and Industry Group, 1978 (percent) Firm size All' Manufacturing Commerce Finance Services Bogotd Smallb 52.0 42.0 71.6 43.2 59.8 Large' 48.0 58.0 28.4 56.8 40.2 Total 100.0 100.0 100.0 100.0 100.0 Total emplovment (number) 1,211,986 286,245 246,345 98,251 394,519 Cali Smallb 55.2 36.3 77.0 55.6 72.9 Large' 44.8 63.7 23.0 44.4 27.1 Total 100.0 100.0 100.0 100.0 100.0 Total employment (number) 368,273 114,757 79,514 13,412 106,664 a. Includes other sectors. b. Firms with less than 10 employees. c. Firms with 10 or niore employees. Source: DANE household survey, 1978. 16 Employment Location by Industry Spatial Distribution of Employment To analyze the spatial distribution of employment, this study employs two zone systems based on comunas, the administrative units for subareas of Colombian cities. As figure 2-1 illustrates, the thirty-eight comunas in Bogota have been aggregated into six rings and eight radial sectors; figure 2-2 shows how Cali's 28 comunas are divided among five rings and seven radial sectors. The ring system is useful for studying the spatial distribu- tion of employment in terms of distance from the CBD. The radial sector system, in contrast, helps to illustrate certain patterns of land use specialization among different areas of the city. For all the major industry groups, the employment statistics in table 2-4 show thatjobs are more centralized in Cali than in Bogota. The peak concentration in Call occurs in ring 3 and then drops markedly; in Bogota, however, the highest concentration of jobs is in ring 5. Although the CBD'S share of employment is similar in the two cities (and comparable to that in large U.S. cities), the concentration ofjobs varies across industry Table 2-4. Employment Distribution by Ring and Industry Group: Bogota and Cali, 1978 (percent) Ring All' Manufacturing Commerce Finance Senrices Bogota 1 13.95 6.01 15.75 41.43 12.91 2 17.74 13.47 19.77 29.38 18.68 3 16.40 21.54 14.83 11.43 16.87 4 20.60 24.89 19.37 10.88 23.18 5 24.94 28.25 27.72 5.65 21.79 6 3.43 2.19 1.51 0.60 4.41 N.i.. 2.96 3.66 1.05 0.63 2.15 Total 100.00 100.00 100.00 100.00 100.00 Cali 1 16.36 9.83 25.23 67.41 14.09 2 26.16 25.71 23.68 19.69 34.57 3 32.67 39.64 31.64 4.70 30.24 4 14.37 11.81 11.84 1.57 15.26 5 2.78 1.38 4.53 0.00 2.12 N.i.e. 7.66 11.62 3.08 6.64 3.73 Total 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. a. Includes other sectors. Source: DANE household survey, 1978. Spatial Distribution 1 7 groups.' In both Bogota and Cali the concentration of employment in the central area is highest in the finance sector, followed by commerce, serv- ices, and manufacturing. The extent of employment decentralization differs by firm size as well as by industry group. The statistics in table 2-5 indicate that jobs in the small firms of Bogota's commerce and service sectors are decentralized, but those in the large firms are centralized. Small retail stores and service firms are, of course, likely to follow residential locations. In the case of manufacturing and finance, the spatial distribution of employment in both small and large firms is similar, although manufacturing jobs are decentralized while finance jobs are centralized. Small (but not neces- sarily new) manufacturing firms may thus settle near large manufacturing establishments, which tend to locate in outer urban areas where more space is available at lower cost. In contrast, most of the small firms in the financial sector-such as real estate and equipment-renting businesses- tend to locate in the central area where face-to-face contact is easier. Even though Cali is much smaller than BogotA, the data in table 2-6 show that its general spatial patterns of employment are similar.4 In manufacturing and finance, small firms follow large ones while large and small firms in the commerce and service sectors show opposite locational tendencies. As noted above, however, employment is much more cen- trally located in Cali than in Bogota: jobs in large financial firms, for ex- ample, are concentrated in rings 1 and 2, with 75 percent in the CBD alone. The distribution of employment by radial sector reflects the patterns of land use specialization. In Bogota manufacturing employment is concen- trated in sectors 4 and 5, the industrial corridor that extends west from the CBD; sectors 3 (the southwest Bosa area) and 6 (which includes the air- port) have also begun to attract manufacturing activities (table 2-7). The industrial corridor bisects BogotA into the wealthy residential areas in the north (sectors 6, 7, and 8) and the poor residential areas in the south (sec- tors 2 and 3).5 As the data in table 2-7 demonstrate, jobs in financial and service firms are located primarily in the north as well as in the CBD. This fact reflects the northward movement of high-income residences over the past several decades and the subsequent growth of financial and service activities in that area.6 Cali's sector 3, which stretches to the northeast from the CBD, forms the industrial corridor where manufacturing employment is concentrated. Sector 5, a high-density, low-income residential area, contains nearly as much trade activity as the cBD.Jobs in Cali's financial sector, however, are located only in the central area. (Text continuez page 24.) 18 Employment Location by Industry Figure 2-1. Ring qnd qprrtnr Systems for Bogot. Ring System 92 Comuna numbers Comuna boundaries ® Ring numbers Ring boundaries 8 3 ) r)91 f cf)~~9 0 45~~~~~~~~~~~~~~~ %s _,) 23 \21 /X 3253- 8 ~~~~~6 5 <2'4, 3 9~~~~~7 65 62a Spatial Distribution 19 Sector System 92 Comuna numbers Comuna boundaries 8 Sector numbers - Sector boundaries k [ /~~~~9 20 Employment Location by Indusstry Figure 2-2. Ring and Sector Systems for Cali Ring System i~~~~~~~~~~~~~~~~~3 22 1 2 v1 4 t r 52Comunanumber \ ) ( Comun boundar53 << ~ ~ ~ ~ ~~5 Com3u Rna numbers t_ _ @ {_ | ; ~~~~Ring boundaries Spatial Distribution 21 Sector System 52 Comuna numbers \ 6 76 _ 4Comuna boundaries 27 Sector numbers ...- Sector boundaries Table 2-5. Employment Distribution by Ring, Firm Size, and Industry Group: Bogota, 1978 (percent) Manufacturing Commerce Fina ace Services Ring Smalla Largeb Small Large Small Large Small Large 1 4.66 7.00 13.13 22.57 52.14 33.99 8.77 18.60 t'D 2 13.42 13.17 17.69 24.84 14.52 39.77 16.02 22.52 3 16.16 24.60 13.18 17.99 11.23 11.29 16.00 18.19 4 26.70 23.58 19.50 18.68 11.58 11.03 24.75 20.11 5 34.37 24.88 33.74 14.00 8.52 3.17 27.93 13.71 6 4.21 1.09 1.95 0.21 1.51 0.00 5.50 3.18 N.i.e. 0.49 5.68 0.82 1.70 0.49 0.76 1.02 3.69 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. a. Firms with less than 10 employees. b. Firms with 10 or more employees. Source: DANE household survey, 1978. Table 2-6. Employment Distribution by Ring, Firm Size, and Industry Group: Cali, 1978 (percent) Manufactur-ing Comnmerce Firnaace Services Ring Sinall' Largeb Small Large Small Large Small Large 1 7.02 12.6f1 17.02 52.76 57.20 76.04 9.53 23.51 2 28.74 22.81 23.29 23.42 28.51 13.51 35.38 29.87 3 41.14 39.06 36.64 15.03 6.25 0.00 37.76 15.26 4 18.24 9.31 14.42 4.77 3.12 0.00 12.89 22.61 5 2.58 0.63 6.06 0.00 4.91 10.45 2.59 1.53 N.ie. 2.27 15.57 2.57 4.03 0.00 0,00 1.85 7.22 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Nie. Not included elsewhere. a. Firms with less than 10 etmployees. b. Firms with 10 or more employees. Source: DANE housethold survey, 1978. 24 Employment Location by Industry Table 2-7. Employment Distribution by Radial Sector and Industry Group: Bogota and Cali, 1978 (percent) Radial sector All' Manufacturing Cornmerce Finance Serices Bogotd 1 13.95 6.01 15.75 41.43 12.91 2 8.40 8.09 10.65 1.73 8.21 3 13.17 18.87 15.32 3.18 10.61 4 9.07 15.76 10.80 2.81 6.24 5 12.14 20.04 10.48 9.67 9.48 6 12.15 11.71 9.70 3.31 14.26 7 9.96 7.70 9.66 10.24 11.91 8 18.21 8.17 16.60 27.00 24.22 N.i.e. 2.96 3.66 1.05 0.63 2.15 Total 100.00 100.00 100.00 100.00 100.00 Cali 1 16.36 9.83 25.23 67.41 14.09 2 10.48 9.83 4.79 13.23 15.02 3 19.71 33.44 17.52 1.90 11.47 4 12.46 14.99 17.07 1.57 8.91 5 17.65 12.64 23.26 0.00 18.57 6 11.38 6.93 6.11 3.52 19.02 7 4.30 0.73 2.94 5.73 9.19 N.i.e. 7.66 11.62 3.08 6.64 3.73 Total 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. a. Includes other sectors. Source: DANE household sur-vey, 1978. Small establishments in all industry groups tend to be found in Bogota's residential areas. Table 2-8 indicates, however, that sector 8 (the high-income area) contains few manufacturing jobs. In the case of finance, both small and large firms are concentrated in the CBD and sector 8, which includes comuna 81, an area adjacent to the CBD. Known as the International Center, this comuna has been developing as a new CBD with the construction of several high-rise office buildings. In Cali, too, employment in small manufacturing as well as commer- cial firms tends to occur in residential areas. As table 2-9 shows, sector 5 (the low-income residential area) contains a high concentration of small manufacturing establishments. As in the case of Bogota, jobs in small financial firms are concentrated in the CBD. Service employment in both large and small firms in sector 6 reflects the southward development of high-income residential communities in Call. Table 2-8. Employment Distribution by Radial Sector, Firm Size, and Industry Group: Bogota, 1978 (per cenit) Mantufacturing Commerce Finance Services Radial sector Smalla Large' Small Large Small Large Small Large 1 4.66 7.00 13.13 22.57 52.14 33.99 8.77 18.60 2 14.36 4.65 14.91 1.16 2.10 1.57 10.85 4.84 3 24.97 14.89 18.16 8.26 6.52 0.82 12.90 7.98 4 8.75 20.33 12.54 7.46 2.81 2.98 7.60 4.21 NO 5 8.10 27.28 7.36 17.09 4.45 13.59 7.52 12.12 6 18.75 7.77 11.74 4.75 5.07 2.27 11.77 16.87 7 11.50 5.81 8.09 13.53 8.44 11.73 13.92 9.23 8 8.43 6.60 13.26 23.48 17.98 32.29 25.65 22.44 N.i.e. 0.49 5.68 0.82 1.70 0.49 0.76 1.02 3.69 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not iricluded elsewhere. a. Firmns witlh less than 10 employees. b. Firms with 10 or more emnployees. Soairce: DANF. household survey, 1978. Table 2-9. Employment Distribution by Radial Sector, Firm Size, and Industry Group: Cali, 1978 (percent) Manufacturing Commerce Kinance Se.ices Radial sector SmaWll Largeb Small Large Small Large Small Large 1 7.02 12.61 17.02 52.76 57.20 76.04 9.53 23.51 2 8.83 10.41 3.88 7.29 12.50 13.51 15.36 14.23 3 21.54 39.04 16.81 18.45 3.79 0.00 11.63 12.28 4 17.82 14.15 20.38 8.28 3.12 0.00 10.10 7.56 °>1, 5 28.79 4.32 29.25 5.11 0.00 0.00 21.63 10.78 6 12.08 3.59 6.61 4.08 7.03 0.00 18.42 18.39 7 1.66 0.31 3.49 0.00 11.44 0.00 11.47 6.03 N.i.e. 2.27 15.57 2.57 4.03 4.91 10.45 13.09 7.22 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. a. Firmis with less than 10 employees. b. Firms with 10 or more eniployees. Source: DANE household survey, 1978. Decentralization 27 Decentralization of Employment To study changes in employment location patterns requires consistent data sets, such as the census of establishments for the main industry groups, for various vears. Such data, however, are not available in Colom- bia.7 For the analysis of changes in employment location patterns in Bogota, therefore, two sources were used: the 1972 household survey conducted for the BogotA Urban Development Study, Phase II (see the appendix to this chapter), and the 1978 DANE household survey. Like the 1978 survey, the Phase II study included questions about the respon- dent's workplace, including location, size of firm, and type of business. For Cali, the analysis is based on the social security data file for 1976 together with the 1978 DANE household survey. In drawing conclusions from the results on Bogota presented below, the following qualifications should be kept in mind. Because the sample frame was revised and updated for the 1978 survey, the sampling dis- tributions and the expansion factors of the two data sets are not directly comparable. Inconsistent definitions of terms could have introduced biases in the analysis; in the 1972 survey, for example, the CBD is referred to as el centro, the city center-a larger area than the CBD-and as a result economic activity in that zone may have been overestimated. In addition, discrepancies regarding the classification of industries in the two surveys might have biased the changes in industrial composition over the period. These inconsistencies make it impossible to analyze changes in levels of employment; at best, the spatial distribution of employment can be com- pared in percentage terms. Distributions of employment in Bogota are reported in table 2-10 for 1972 and 1978. Although there is evidence of decentralization during this period, the sharp decline of the CBD'S employment share (from 23 to 14 percent) is suspect. As mentioned above, this apparent overstatement is likely due to the problem with the definition of the CBD in the 1972 survey. It should be noted that employment share of the CBD together with ring 2 decreased only moderately from 36.6 percent in 1972 to 31.7 percent in 1978.8 In Bogota the extent of decentralization in manufacturing employment is very similar to that in the commercial sector even though the CBD SUS- tained a much larger share of commercialjobs. Although the service sec- tor also exhibits a clear trend toward decentralization, financial es- tablishments remained primarily in the central area during the six-year period. Ring 2, however, experienced a substantial gain in employment, which reflects the shift of finance jobs from the CBD to the International Center discussed above. Table 2-10. Changes in Employment Location: Bogota, 1972-78 (percenit) All' Manufacturing Cornmmerce Finance Services Ring 1972 1978 1972 1978 1972 1978 1972 1978 1972 1978 1 23.03 13.95 18.20 6.01 19.43 15.75 42.11 41.43 22.62 12.91 2 13.61 17.74 16.07 13.47 12.18 19.77 13.69 29.38 12.74 18.68 3 14.62 16.40 18.94 21.54 13.35 14.83 6.89 11.43 15.88 16.87 4 18.80 20.60 20.27 24.89 21.83 19.37 10.00 10.88 20.74 23.18 5 18.61 24.94 21.76 28.25 21.52 27.72 14.64 5.65 17.89 21.79 6 1.67 3.43 1.04 2.19 3.42 1.51 0.55 0.60 1.88 4.41 N.i.c. 9.67 2.96 3.72 3.66 8.27 1.05 12.12 0.63 8.25 2.15 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. a. Includes other sectors. Sources: DANE household survey, 1978; Phase 11 survey, 1972. Decentralization 29 Evidence of employrment decentralization in Cali also appears in the data presented in table 2-11. As indicated earlier, the 1976 social security data are quite consistent with the subset of the 1978 household survey data representing individuals whose firms are affiliated with the social security system; see the appendix to this chapter for a more detailed com- parison of data sets. Since the social security files have poor coverage of small firms, however, table 2-11 includes only those jobs in firms with 10 or more employees. Although Cali experienced decentralization in the manufacturing and service sectors, the data indicate that the commerce and finance sectors became more centralized. The sharp increase in the CBD'S share of finance jobs, however, is obviously due to sampling errors. The 1978 household survey probably undersampled individuals working in financial firms located in outer rings. Patterns of Movement and Incidence of Births To analyze the various components of the changing patterns of employ- ment location, it is useful to focus on the behavior of four groups of firms: relocating firms (movers), newly established firms (births), defunct firms (deaths), and the expansion or contraction of stationary (nonrelocating) firms. The destination of movers and the location patterns of births are important not only for understanding the changing patterns but also for predicting the future spatial structure. Since establishment-based micro data are available only for the manu- facturing sector, the information analyzed here is derived from the 1978 household survey, which included questions about a firm's years of operation at its present location and the previous location if it had moved. Table 2-12 presents the origin-destination ratios for each industry group, obtained by dividing the number of jobs moving out of an area, or ring, by the number of jobs moving in during 1973-78.9 The results indicate that the CBDS in both BogotA and Cali experienced a net loss ofjobs in all industry groups. The net outflow of jobs from the CBD was much greater in BogotA, however, except in the service sector. In Cali all rings outside the CBD had net employment gains except for service jobs in ring 3, hence there is no evidence that firms were moving out from rings outside the CBD. In the case of BogotA, however, the moving patterns of relocating firms contributed to decentralization. In BogotA's manufacturing sector the origin-destination ratio gradually declines as the distance from the CBD increases; rings 4 and 5, for example, display a net gain of manufac- turing employment (see appendix tables A6 and A8 for origin and des- tination matrices with actual number of jobs). The attraction of jobs into ring 2 reflects the growth of the International Center. Table 2-11. Changes in Employment Location: Cali, 1976-78 (percent) Alla Manufacturing Commerce Finance Services Rinig 1976 1978 1976 1978 1976 1978 1976 1978 1976' 1978 1 31.51 26.19 20.19 14.94 48.38 54.97 45.68 84.90 29.15 25.34 2 37.12 27.61 34.28 27.02 34.50 24.40 16.67 15.10 55.47 32.19 3 28.40 31.33 41.40 46.27 11.09 15.65 37.55 0.00 14.79 16.45 4 2.95 13.82 4.09 11.03 6.03 4.97 0.11 0.00 0.54 24.37 5 0.03 1.05 0.05 00.70 0.00 0.00 0.00 0.00 0.06 1.65 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Note: Based on employment in firms with 10 or more employees. a. Includes other sectors. b. Excludes employment in the government sector. Sources: DANE household survey, 1978; social security data files, 1976. Patterns of Movement 31 Table 2-12. Movement Patterns ofJobs Measured by Origin-Destination Ratio for Rings: Bogota and Cali, 1973-78 Ring .11a Manufacturing Commerce Finance Sernices Bogotd 1 2.37 2.51 3.15 3.32 1.45 2 0.66 1.54 0.45 0.33 0.87 3 1.33 1.58 1.32 1.04 1.36 4 0.55 0.76 0.51 0.06 0.58 5 0.44 0.32 0.20 2.70 0.35 6 0.59 _ b _ b, b,c 0.29 Cali 1 1.85 1.97 1.71 1.58 2.03 2 0.68 0.70 0.61 -' 0.45 3 0.86 0.93 0.84 _ b,. 1.21 4 0.67 0.70 _ b _b, 0.33 5 3.44d _ b c _b,c 1 Note: The origin-destination ratio is the nunmber of jobs moving out of a zone divided by the nuniber of jobs moving into the zone. a. Includies other sectors. b. No firnis moved into the arca. c. No firnms nioved out Of the area. d. Even though the ratio was high. ring 5 lost only 651 jobs and gained 189, whereas the CBD lost 9,479 and gained 5,126. The total numiber ofjobs relocated was 28,170 in Cali and 84,755 in Bogota. See appendix tables A6 and A8. Source: DANE household survey, 1978. Table 2-13 reports the location patterns of jobs created by firms es- tablished over the five-year period (births). It is striking that the employ- ment share of new jobs in both cities increases with distance from the CBD (up to and including ring 5 in Bogota and ring 3 in Cali) for all industry groups except the finance sector. This result suggests that newly es- tablished firms tend to locate at the urban periphery, thus contributing to decentralization. In Bogota's finance sector 43 percent of the jobs created by new firms during this period were located in ring 2-again a reflection of the northward shift of financial activities from the CBD to the Inter- national Center. Summary This chapter describes the spatial distribution of enmployment in Bogota and Cali by four industry groups-manufacturing, commerce, finance, and services-which together accouint for about 85 percent of total em- 32 Employment Location by Indust7y Table 2-13. Location Patterns ofNewJobs: Bogota and Cali, 1973- 78 (percent) Ring Alta Manufacturing Commerce Finance Serices Bogotd 1 13.78 7.55 14.09 29.20 18.61 2 17.80 13.47 20.31 43.04 14.05 3 14.42 19.98 8.15 13.86 17.38 4 20.87 24.66 23.00 9.43 17.55 5 27.67 30.62 31.97 4.47 26.78 6 2.64 2.82 1.72 0.00 2.32 N.i.e. 2.82 0.90 0.75 0.00 3.32 Total 100.00 100.00 100.00 100.00 100.00 Cali 1 14.70 9.27 13.61 0.00 26.27 2 25.62 19.18 24.05 60.00 30.88 3 38.33 44.95 44.76 20.00 32.99 4 13.40 22.24 11.45 0.00 5.16 5 2.60 2.29 1.97 0.00 0.00 N.i.e. 5.35 2.06 4.15 20.00 4.70 Total 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. a. Includes other sectors. Source: DANE household survey, 1978. ployment in each city. The analyses based on the household survey data and social security files showed that for all major industry groups jobs were more centralized in Cali than in BogotA. The central tendency, however, differed substantially among industry groups: it was highest in the finance sector, followed by commerce, services, and manufacturing. For commerce and services, jobs in large firms were more centralized than those in small firms since small retail and service firms tend to follow residential areas. No such differences by firm size were observed for manufacturing and finance. For all industry groups together, the data revealed strong evidence of employment decentralization in both cities. In BogotA the pattern of decentralization was similar for manufacturing, commerce, and services, but the finance sector tended to stay more centralized. In Cali, however, jobs in commerce as well as in finance stayed more centralized. The changing patterns of employment location can be explained by the direction of relocating firms and the births of new firms. The 1978 household survey had questions on the location history of firms where the respondents worked. The analysis of patterns of movement of firms Summary 33 during 1973-78 revealed that the CBD in both Bogota and Cali experi- enced a net loss ofjobs in all industry groups-that is, the outflow ofjobs was greater than the inflow. Outer areas had a net gain ofjobs, and the ex- tent of the gain increases with the distance from the CBD. The data showed that the location patterns of new firms (births) also contributed to the decentralization of employment during the period. In both BogotA and Cali the employment share of new jobs increases with the distance of the ring from the CBD for all industry groups with the exception of finance. Appendix: Data Comparisons Data Sources The three data sets used in this chapter are described here. The 1978 DANE Household Survey. Conducted jointly with the World Bank, the 1978 DANE Household Survev sampled 3,000 households in Bogota and 1,000 in Cali. It covered the characteristics of households and their individual members. For the study of employment location, the sur- vey included questions about the characteristics of establishments where the respondents worked, including the present location of the workplace, the number of employees of the firm, the initial year of operation at the present location, the previous location if any, and the type of industry or business (according to the Standard Industrial Classification, or sic code). (For a detailed description of variables, documentation of data files, and frequency tabulations, see Y. J. Lee 1979.) The total number of workers responding to the set of workplace ques- tions was 5,111 in BogotA and 1,656 in Cali. The workers in the survey were defined as those individuals who reported "work" as their major ac- tivity during the week preceding the interview, or who worked more than fifteen hours a week, or who had some remunerated activity. To "blow up" the sample counts to obtain population figures, expansion factors were based on the total counts of households and dwelling units in each city. The 1972 Phase II Household Survey. Phase II of the Bogota Urban Development Study occurred in 197 1-72 and was funded by the United Nations Development Programme (UNDP) under the auspices of the World Bank. The project included a survey of 4,675 households in 1972 (for a detailed description of the data set, see Valverde 1978). Although the survey focused on demographic and housing characteristics, it also 34 Employment Location by Industry included questions about the respondent's workplace, including its loca- tion and type of industry. The Social Security Files on Establishments. These data provide a complete listing of all establishments affiliated with the social security system. The Colombian Social Security Agency (Instituto de Seguros Sociales) re- leased to the City Study a copy of the complete files of BogotA for 1978 and 1979 and of Cali for 1976. For each firm, the tapes contain the ad- dress, the number of workers, and the sIc; each establishment's location was then geocoded at the barrio level at Corporaci6n Centro Regional de Poblacion in BogotA.'° The main limitation of this data set is its poor coverage of small firms. Even though the law requires that all employees in private and public en- terprises and self-employed individuals register with the social security system, those with small family shops and independent and casual workers tend to avoid membership. Government employees and military personnel have separate social security programs. For the country as a whole, the social security data compare favorably with other information, such as the industrial census and labor force sur- veys, in particular for the manufacturing and the finance sectors (DANE 1975b; further descriptions and analyses of the data appear in DANE 1974 and 1976b). DANE concluded that the coverage is almost complete for large manufacturing establishments (with 10 or more workers) and firms of all sizes in finance, while the coverage is poor in the commerce and ser- vice sectors (which include many small firms). Comparisons of Household Data with Social Security Data, Bogota, 1978 The household survey included a question about whether the respon- dent's firm was affiliated with the social security system. It is therefore possible to compare the distribution of establishment characteristics (such as employment size, industry type, and location) given by the two data sets for the group of firms affiliated with the social security system. The comparisons in table 2-14 indicate that the total number of workers listed in BogotA's 1978 social security file is onlv 45 percent of the total number estimated by the household survey. In the small-firm category (I -9 employees), the social security files cover only 12 percent of the 630,743 workers. But the household survey understates the total number of workers in the three large-firm categories (100 or more em- ployees). For firms with 1,000 or more workers, the household survey's figure of 44,388 is no more than 75 percent of the total given by the social Data Compansons 35 Table 2-14. Comparison of Household Survey and Social Security Data: Employment Distribution by Firm Size, Bogota, 1978 Household surrey Firm size Soczai security (number Social securnty of workers) All workers Percent members Percent All workers Percent 1-9 630,743 52.04 106,066 23.95 74,785 13.74 10-19 118,843 9.81 65,999 14.90 57,960 10.65 20-49 128,021 10.56 75,356 17.02 86,454 15.89 50-99 76,161 6.29 49,399 11.15 64,756 11.90 100-499 126,602 10.44 78,033 17.62 153,188 28.14 500-999 42,181 3.48 24,469 5.53 47,651 8.75 1,000+ 44,388 3.66 19,332 4.37 59,544 10.94 N.i.e. 45,046 3.72 24,205 5.47 18 0.00 Total 1,211,986 100.00 442,858 100.00 544,356 100.00 N.i.e. Not included elsewhere. Sources: DANE household survey, 1978; social security data files, 1978. security file. The social security files thus have poor coverage of small firms while the household survey understates the number of workers in the large firms. (For the large firms, the social security figures should be very close to the population figures.) The understatement in the house- hold survey may be due to the respondent's poor perception of firm size; for example, production workers in a company may not know how many white-collar workers there are. The social security program tends to miss the self-employed, especially those with small independent shops; casual workers such as maids; and other part-time workers, particularly in the construction industry. This poor coverage of the commerce and the service sectors-33 and 35 per- cent, respectively-is evident in the comparisons provided in table 2-15. About 20 percent of the workers not affiliated with the social security sys- tem were in personal services. Government employees are also not in- cluded in the program. According to the household survey, 36.5 percent (442,858) of BogotA's total number of workers are affiliated with the social security system, which is in turn 81 percent of the total social security count of 544,356 workers. For this group, the survey's understatement of workers employed in large firms is substantial. Nevertheless, the two data sets compare well in percentage terms for workers affiliated with the so- cial security system: employment shares by industry group are almost identical, particularly for groups within the manufacturing sector. As Table 2-15. Comparison of Household Survey and Social Security Data: Employment Distribution by Industry Group, Bogota, 1978 Household survey - Social security Social secunty Industry group All workers Percent rnembers Percent All workers Percent Manufacturing 286,245 23.62 158,472 35.78 209,212 38.43 Commerce 246,345 20.33 82,437 18.61 81,707 15.01 Finance 98,251 8.11 67,310 15.20 58,917 10.82 Services 349,519 32.55 74,485 16.82 120,818 22.19 Other 186,628 15.40 60,155 13.58 73,702 13.54 > Total 1,211,986 100.00 442,858 100.00 544,356 100.00 Manufacturing Food, beverages 37,832 13.22 19,702 12.43 24,311 11.62 Textile, apparel 79,662 27.83 34,247 21.61 52,627 25.15 Wood, furniture 22,000 7.69 7,546 4.76 10,669 5.10 Paper, printing 22,262 7.78 12,919 8.15 13,996 6.69 Chemical 33,646 11.75 27,033 17.06 19,779 9.45 Nonmetal 13,908 4.86 8,242 5.20 9,967 4.76 Metal 2,399 0.84 1,446 0.91 6,244 2.98 Machinery 63,487 22.18 43,303 27.33 56,726 27.11 Other 11,049 3.86 4,035 2.55 14,893 7.12 Subtotal 286,245 100.00 158,472 100.00 209,212 100.00 Sources: DANE household survey, 1978; social security clata files, 1978. Data Comparisons 37 Table 2-16. Comparison of Household Survey and Social Security Data: Employment Distribution by Ring, Bogota, 1978 Household )unvey Social security Social security Ring All workers Percent members Percent All workers Percent 1 169,034 13.95 85,339 19.27 122,752 22.55 2 214,953 17.74 92,469 20.88 110,885 20.37 3 198,733 16.40 84,719 19.13 140,008 25.72 4 249,624 20.60 79,759 18.01 81,436 14.96 5 302,234 24.94 74,223 16.76 69,405 12.75 6 41,530 3.43 8,769 1.98 5,661 1.04 N.i.e. 35,878 2.96 17,626 3.98 14,208 2.61 Total 1,211,986 100.00 442,858 100.00 544,356 100.00 N.i.e. Not included elsewhere. Source: DANE household survey, 1978; social security data files, 1978. table 2-16 shows, the spatial distribution of jobs with establishments af- filiated with social security system in the survey is also similar to the dis- tribution provided by the social security data. This analysis of the two data sets helps justify the use of the household survey results in making inferences about the characteristics of respon- dents' workplaces, even though the two data sets are similar mainly in the case of large firms- that is, those that are affiliated with the social security system. Although this analysis is based only on the data for Bogota, analysis of the data for Cali would probably produce a similar finding. Notes 1. For a detailed description of the data, see the appendix to this chapter, which also compares the household survey results with those of the social security data. Selected findings from the analysis of these data appeared in K. S. Lee (1985a). 2. According to the 1978 household survey, about 40 percent of Bogota's total employment was with firms having less than five employees. 3. According to Bronitsky and others (1975), about 10 to 15 percent of total employment in large U.S. cities is in the CBD. 4. In Cali employment is sparse beyond ring 3 and the observed regularities do not occur in rings 4 and 5; this was also the case for ring 6 in Bogota. 5. For detailed descriptions of the labor force and residential location patterns, see Mohan (1980 and 1986) for Bogota, and Terrell (1979) for Cali. 6. The traditional commercial district (Chapinero area) in sector 7 used to be 38 Employment Location by Industry the "Fifth Avenue" of Bogota. In recent years, however, retail stores and res- taurants have sprung up along 15th Avenue northward, and a large shopping cen- ter (called Unicentro) has been developed in the northern part of sector 8. The determinants of commercial and service employment location are tested in chap- ter 6 with the use of a gravity model. 7. DANE, however, has compiled an annual industrial directory that is used for the analysis of manufacturing employment in chapter 3. 8. According to both data sets, the total number ofjobs in the CBD together with those in the International Center (comuna 81, the area directly north of the CBD) remained almost constant during the six-year period: 1972 1978 CBD 201,975 166,878 International Center 24,787 62,565 Total 226,762 229,443 9. Since the number of workers at the previous locations is estimated on the basis of the number of respondents of the survey, the assumption is that the same number of jobs existed at the previous locations as at the 1978 locations. This should not affect our conclusions on patterns of movement. 10. A barrio is the smallest administrative unit and is roughly equivalent to a city district, ward, or precinct. 3 Location Patterns of Manufacturing Employment In the theoretical literature, the location behavior of firms typically has been examined in the neoclassical framework: given price gradients for rents and wages, transport costs of inputs and outputs, and the location of markets, a firm will tend to locate where it maximizes its profits. In the re- cent empirical studies of employment location, a popular approach has been to investigate not only the growth and decline of stationary (non- relocating) firms but also the location choices of newly established firms and relocating firms and to analyze changes in location patterns at the margin.' According to DANE'S annual industrial directory file, which provides in- formation on location, production, sales, and input uses of individual es- tablishments, decentralization of manufacturing employment occurred in Bogota and Cali during 1970-75.2 The following analysis demonstrates this trend by decomposing changes in the stock of employment by the births and deaths of firms, relocation ofjobs, and stationary growth and decline of employment.3 The choices of newly established firms and the destination of movers are particularly important for understanding loca- tion decisions and for predicting the future spatial structure of an urban area. Comparison of Bogoti and Cali with U.S. Cities According to DANE'S household survey, more than one million people were employed in Bogota in 1978 while more than one-third of a million worked in Cali. Manufacturing jobs accounted for 24 percent of employ- ment in Bogota, second only to the service sector, which provided 33 per- cent of the citv's jobs. In Cali 31 percent of employment was in the manufacturing sector. As table 3-1 shows, large U.S. cities such as Boston, Chicago, and Los Angeles also have about 25 to 30 percent of employ- ment in manufacturing. Manufacturing jobs together with those in com- 39 Table 3-1. Percentage Distribution of Manufacturing Firms and Employment by Firm Size: Bogoti, Cali, and Selected U.S. Cities Bogold Cali Wash., D.C. Boston Chicago Los Angeles Firm size (number of workers) Percent Cumrtulative Percent Cumulative Percent Cusnuilalive Percent Cumulative Percent Cumulative Percent Cumulative Firmsa Less than 20b 57.55 57.55 55.25 55.25 59.11 59.11 50.39 50.39 46.89 46.89 53.03 53.03 20-49 24.17 81.72 22.45 77.70 24.53 83.64 25.08 75.47 24.06 70.95 24.48 77.51 50-99 10.11 91.83 11.95 89.65 8.88 92.52 12.93 88.40 11.60 82.55 11.18 88.69 100-499 7.43 99.26 9.33 98.98 6.54 99.06 10.05 98.45 14.61 97.16 9.84 98.53 500 or more 0.75 100.00 1.02 100.00 0.93 100.00 1.56 100.00 2.84 100.00 1.46 100.00 Employmenta Less than 20b 14.25 14.25 10.78 10.78 12.17 12.17 7.57 7.57 4.95 4.95 8.07 8.07 20-49 18.26 32.51 12.71 23.49 17.47 29.64 12.28 19.85 8.37 13.32 13.48 21.55 50-99 17.31 49.82 15.79 39.28 13.63 43.27 13.92 33.77 9.00 22.32 12.40 33.95 100-499 35.74 85.56 36.32 75.60 31.20 74.47 31.15 64.92 33.42 55.74 30.79 64.74 500 or more 14.45 100.00 24.39 100.00 25.54 100.00 35.08 100.00 44.26 100.00 35.25 100.00 Average firm size (number of workers) Less than 20 9.95 10.07 9.07 9.78 9.60 9.53 20-49 30.35 29.21 31.35 31.86 31.57 34.48 50-99 64.80 68.12 67.61 70.05 70.46 69.47 100-499 193.34 200.81 210.00 201.78 207.66 195.87 500 or more 771.63 1,233.00 1,203.50 1,465.40 1,416.58 1,509.07 Total 40.18 51.58 44.04 65.07 90.80 62.62 Numnber (thousands) Population' 3,439 1,055 2,862 2,754 6,978 7,041 Total employment' 1,212 363 1,110 1,098 2,503 2,596 Manufacturing employmentc 286 113 67 262 782 719 As pcrccntage of total employmrienit 23.60 31.13 6.04 23.86 31.24 27.70 a. Data are for 1970 for Bogota aisd Cali; 1973 for U.S. cities. All U.S. figures ale foi cenitral city. b. For Bogota and Cali this category covcrs 5-19; for U.S. cities, 4-19. c. 1978 estimlate for Bogota and Cali (from table 2-1); 1970 U.S. cenlsuIs figtures for the Standard Mctropolitan Statistical Arcas of U.S. cities. Source: Figuies for U.S. cities are from Ingrans (1976); they are derived from County Busnuess Patterns, 1973, and data on journey to work oftlle 1970 tJ.S. Ceil- stls, table 2. Table 3-2. Birth, Death, and Relocation Rates in Bogota, Cali, and Selected U.S. Cities (percent) Births Deaths Moves' Establishments Employment Establisltments Employment Establishments Employment Percentage Annual Percentage Annual Percentage Annual Percentage Annual Percentage Annual Percentage Annual city of base rate of base rate of base rate of base rate of base rate of base rate Clevclandb 9.97 3.22 2.59 0.86 14.07 4.49 7.75 2.52 13.83 4.41 5.77 1.89 Minneapolis- KD St. Paulb 12.29 3.94 6.17 2.02 18.00 5.67 11.25 3.62 15.93 5.05 8.28 2.69 Boston" 6.10 1.99 1.30 0.43 13.40 4.28 8.00 2.60 9.80 3.17 4.70 1.54 Phoenixb 24.40 7.55 12.10 3.88 20.20 6.32 5.30 1.74 8.90 2.88 4.70 1.54 New Yorkc 10.21 4.98 3.95 1.96 7.56 3.71 3.55 1.76 11.45 5.57 1.24 0.62 Bogot'ad 52.38 8.79 31.96 5.70 27.01 4.90 12.61 2.40 19.12 3.56 16.59 3.12 Calid 43.13 7.44 24.48 4.48 26.88 4.88 11.27 2.16 18.33 3.42 10.40 2.00 a. In the case of Bogota the figures include establishliienits that moved at least to another seccidn (designated by lDANE'S six-digit zone code); in Cali the figures include establishments that moved at least to another barrio (designated by the first four digits of DANE'S zone code). When moves within the same seccion are included, the arnual reloca.ion rate of establishments was 5.12 percent for Bogota and 4.28 percent for Cali. b. From Struyk and Janies (1975) for 1965-68 (1965 was the base year). c. From Leone (1971) for 1967-69 (1967 was the base year). d. The period covered was 1970-75; 1970 was used as the base year. The base-year figures can be seen in K. S. Lee (1978). Comparison with U.S. Cities 43 merce and services represented about 80 percent of all employment in both Bogota and Cali; for the country as a whole, however, these three sectors included only about 40 percent of all jobs (table 2-1). As the comparisons in table 3-1 demonstrate, the percentage distribu- tion of firms by size in Bogota resembles that of Washington, D.C., although the latter has only a small amount of manufacturing employ- ment. Average firm size in Bogota is much smaller than in the U.S. cities, where large firms are twice as large as those of Bogota. Firm size distribu- tion and share of manufacturing employment in Cali, in contrast, looks like that of Los Angeles; Cali, however, has a smaller proportion ofjobs in firms with 500 or more employees. In terms of births, deaths, and relocations of firms, table 3-2 indicates that employment dynamics in Colombian cities are somewhat different from those in U.S. cities. At an annual rate of 8.8 percent, the birth rate of firms in Bogota is higher than that in all five U.S. cities (although this per- centage may include other than genuine births, such as the movement of established firms into the area). The birth rate of 7.6 percent in Phoenix comes closest to the rate in Bogota and is slightly higher than that in Cali. The average death rate of firms in both Bogota and Cali, however, is com- parable to that of Cleveland, Minneapolis, and Boston. The birth rate of firms is greater than the death rate in Bogota, Cali, Phoenix, and New York, whereas the opposite is true in Boston, Minneapolis, and Cleveland.4 In all cities, the birth, death, and relocation rates of establishments are greater than the rates based on employment. This finding provides evidence that the marginal firms are the small ones that have a greater propensity to move, start up, and close down. During 1970-75 nearly 60 percent of firms' births and deaths in both Bogota and Cali were among those with 10-24 employees. Moreover, as of 1970 the median age of es- tablishments was 4.5 years in both cities. (See appendix tables A29 and A30.) Young, small firms therefore seem to contribute significantly to changes in employment distribution in Colombian cities. Decentralization Trends To analyze changes in manufacturing employment location patterns, this section uses the same ring and sector systems described in chapter 2. Tabulations of employment by ring, reported in table 3-3, provide strong evidence of decentralization of manufacturing employment in both cities. Whereas Bogota's ring 1 experienced an absolute decline in em- ployment, each successive ring showed an increase in number of jobs- the annual growth rate reaches 16 percent in ring 5. Although Cali's CBD 44 Manufacturing Employment Table 3-3. Distribution of Manufacturing Employment by Ring: Bogota and Cali, 1970-75 1970 1975 Number Number Annual average Ring employed Percent employed Percent growth rate (percent) Bogota 1 4,538 5.60 4,102 3.47 -2.00 2 11,767 14.53 14,898 12.59 4.83 3 34,351 42.42 47,858 40.44 6.86 4 18,112 22.37 25,958 21.94 7.46 5 11,548 14.26 24,047 20.32 15.80 6 391 0.48 729 0.62 13.27 N.i.e. 266 0.33 741 0.63 - Total 80,973 100.00 118,333 100.00 7.88 Cali 1 2,600 7.65 3,064 6.95 3.34 2 13,836 40.74 14,381 32.60 0.78 3 15,192 44.73 21,704 49.21 7.40 4 1,761 5.18 3,361 7.62 13.80 5 367 1.05 100 0.23 -22.90 N.i.e. 219 0.64 1,499 3.40 - Total 33,965 100.00 44,109 100.00 5.37 -Not applicable. N.i.e. Not included elsewhere. Note: The data are from establishments with 10 or more employees. Source: DANE industrial directory file. did not experience an absolute employment decline, its growth was much lower than that of the outer rings. The exceptional negative growth rate for ring 5 resulted from the shutdown of a large sugar refinery in the southernmost comuna (63). The site has since been converted to a high- income residential and shopping area. In BogotA the zone system of eight radial sectors follows the pattern of land use specialization. The industrial corridor, which bisects the city into north and south, consists of sectors 4 and 5 (see figure 3-1). The residen- tial north is sector 8, and the residential south is sector 2. The residual southwest area is labeled sector 3; the residual northwest is further divided into sector 6, containing the airport, and sector 7, which includes the commercial corridor of Chapinero and the residential northwest. As the statistics in table 3-4 show, the growth of manufacturing employment was lowest in the two main residential sectors (2 and 8). In addition, em- ployment growth in the industrial corridor (4 and 5) was slightly lower Decentralization Trends 45 Figure 3-1. Analysis Zones for Bogota Comuna boundaries O Ring numbers Ring boundaries 8 Sector numbers - ... Sector boundaries 6-8 Analysis zone (ring-sector) ~~C 6 aS S~~~5- S~~~~~~4 46 Manufacturing Employment Table 3-4. Distribution of Manufacturing Employment by Radial Sector, 1970-75 1970 1975 Actual average Number Number growth rate Sector employed Percent employed Percent (percent) Bogota I CBD 4,538 5.60 4,102 3.47 -2.00 2 South 2,451 3.03 3,218 2.72 5.60 3 Southwest 6,255 7.72 12,741 10.77 15.29 4-5 Industrial corridor 57,833 71.42 84,362 71.29 7.84 6 Airport 3,532 4.36 5,333 4.51 8.59 7 Northwest 1,960 2.42 3,493 2.95 12.25 8 North 4,138 5.11 4,343 3.67 0.97 N.i.e. 266 0.33 741 0.63 - Total 80,973 100.00 118,333 100.00 7.88 Cali I CBD 2,600 7.65 3,064 6.95 3.34 2 North 5,368 15.80 6,161 13.97 2.79 3 Industrial corridor 20,239 59.59 24,949 56.56 4.27 4 East 3,566 10.50 5,774 13.09 10.12 5 Southeast 796 2.34 1,110 2.52 6.88 6 South 1,063 3.13 1,275 2.89 3.70 7 West 114 0.34 277 0.63 19.43 N.i.e. 219 0.64 1,499 3.40 - Total 33,965 100.00 44,109 100.00 5.37 -Not applicable. N.i.e. Not included elsewhere. Note: Data are from establishments with 10 or more employees. Source: DANE industrial directory file. than the overall rate; thus the relative shares of these sectors in manufac- turing employment were reduced during the period. In contrast, the shares of the residual sectors (3, 6, and 7) increased. Evidence of employ- ment decentralization from the industrial corridor to the nonresidential sectors is also clear from the extremely high employment growth in sec- tors 3 (15 percent) and 7 (12 percent). Cali's employment pattern by radial sector also exhibits some land use specialization. The industrial corridor, extending to the east of the city (sector 3) had more than half the manufacturing employment in 1975 (see figure 3-2). During 1970-75, however, manufacturing employment in this sector grew at a lower rate than the city's average, which substantially Decentraliza2tion Trends 47 Figure 3-2. Analysis Zones for Cali 3- S - Sconubr = P rs --- Setor boundarie 4-6~ ~ ~~~~- Comnas bouned(riegsetr Table 3-5. Composition of Changes in Manufacturing Employment by Ring: Bogota, 1970-75 Mature' Movers 1970 1975 Birthsa Deaths'a At origin At destination Number Number Number Number Number Number Ring employed Percent ernployed Percent e?nployedb Percent errployed' Percent employed Percent employed Percent 1 2,364 3.95 2,828 3.76 1,011 4.16 581 6.24 1,659 14.80 433 3.22 2 7,136 11.92 8,211 10.93 4,310 17.73 1,701 18.28 2,726 24.32 1,772 13.19 3 26,291 43.93 32,807 43.67 8,393 34.52 3,663 39.36 4,689 41.82 4,535 33.77 4 14,591 24.38 17,701 23.56 4,554 18.73 1,611 17.31 1,697 15.14 3,273 24.37 a1 5 9,149 15.29 13,035 17.35 5,332 21.93 1,677 18.02 358 3.19 3,351 24.95 °° 6 119 0.20 29 0.04 487 2.00 44 0.47 82 0.73 66 0.49 N.i.e. 191 0.32 519 0.69 224 0.92 29 0.31 - - - Total 59,841 100.00 75,130 100.00 24,311 100.00 9,306 100.00 11,211 100.00 13,430 100.00 -Not applicable. N.i.e. Not included elsewhere. Note: The data exclude establishlmleists with less than 10 employees that appeared for only one year in the DANE industrial directory. The accouritirig identity between thle change in stock of employmenit in table 3-3 and the sum of flows and statiorary growth in table 3-5 does not hold, nmainly because of the onsission of 135 firms (out of 2,629) which could not be classified into the four location tenure categories. See K. S. Lee (1978) for details. a. Excluding movers. b. At birth. c. In 1970. Source. DANE iiduistrial diiectory file. Decentralization Trends 49 reduced its share of these jobs. The same changes also occurred in the CBD and the northern residential sector (2). In contrast, the two low-income residential sectors to the south of the industrial corridor (4 and 5) showed relative gains in their employment shares. This pattern thus suggests a decentralization of jobs from the traditional industrial area to the south. Components of Change in the Spatial Distribution of Employment Changes in spatial patterns of employment can best be understood by ex- amining the location history of establishments. A master file for each city was therefore created with the following categories: (a) mature firms, that is, those that appeared with the same address in all six annual directories; (b) births, that is, those that appeared for the first time during 197 1-75 and kept the same address; (c) deaths, that is, those that disappeared from the directory during 1971-75 after having remained at one address; and (d) movers, that is, those that relocated within the city during 1971-75, in- cluding new and defunct firms that changed address during the period. As the data in tables 3-5 and 3-6 demonstrate, the 4.7 percent growth of employment in mature firms in Bogota was much lower than the overall rate of 7.9 percent in table 3-3. (Although the 1974 recession may have af- fected the growth of mature firms, it would not have caused such a large Table 3-6. Summary Statistics: Composition of Changes in Manufacturing Employment by Ring, Bogota, 1970-75 (percent) Annual stationary Annual Annual Origin/destination Ring growth rate' birth rateb death rateb ratio of mozers 1 3.65 4.10 2.44 3.83 2 2.85 6.44 2.74 1.54 3 4.53 4.47 2.05 1.03 4 3.94 4.59 1.72 0.52 5 7.34 7.89 2.75 0.11 6 - 17.56 2.16 1.24 Total 4.66 5.39 2.20 0.83 -Not applicable. Notee See notes for table 3-5. a. Of mature firms. b. Annual average rate based on 1970 msanufacturing employment; see table 3-3. Source: DANE industrial directory file and table 3-5. Table 3-7. Composition of Changes in Manufacturing Employment by Ring: Cali, 1970-75 Mature' Movers 1970 1975 Births' Deaths' At origin At destination Number Number Number Number Number Number Ring employed Percent employed Percent emnployedb Percent employed' Percent employed Percent emnployed Percent I 1,467 5.45 1,612 5.16 393 5.02 446 13.07 627 21.30 1,045 29.58 2 11,224 41.67 11,114 35.60 2,021 25.79 1,444 42.32 1,399 47.52 769 21.77 3 12,644 46.94 16,148 51.73 3,093 39.48 1,363 39.95 654 22.21 1,316 37.25 4 1,600 5.94 2,341 7.50 735 9.38 64 1.88 106 3.60 266 6.40 5 0 0.00 0 0.00 17 0.22 34 1.00 0 0.00 82 2.32 N.i.e. 0 0.00 0 0.00 1,576 20.11 61 1.79 158 5.37 95 2.69 Total 26,935 100.00 31,215 100.00 7,835 100.00 3,412 100.00 2,944 100.00 3,533 100.00 N.ie. Not included elsewhere. Note: The data are from establishments with 10 or more employees. a. Excluding movers. b. At birth. c. In 1970. Source: DANE industrial directory file. Components of Change 51 difference in the employment growth rate.) Mature firms in ring 5, however, expanded at a rate close to the overall average, while those in the CBD and ring 2 grew more slowly. The lowest growth of mature firms occurred in ring 2, where the fairly high concentration of employment may indicate a capacity constraint. The statistics also indicate that newly founded firms accounted for 61 percent of the total number of new jobs, including those created by mature businesses. In all rings more jobs were created by the birth of firms than by the expansion of mature firms; in ring 2, in particular, there were four times as many jobs created by births than by the growth of established businesses. The growth of employment in mature firms in Cali has a pattern similar to that in Bogota. As tables 3-7 and 3-8 show, the 3 percent annual growth rate in mature firms was lower than the overall rate of 5.4 percent (table 3-3). In addition, the number of jobs in mature firms in the outer rings (3 and 4) grew much faster than in the CBD and ring 2, which experi- enced an absolute decline. In Cali new firms created 65 percent of the total number of new jobs; in Bogota, 61 percent. It is striking that Bogota's ring 2 displayed not only the lowest growth of mature firms but also a higher than average birth rate. The relocation statistics indicate that many firms start business in ring 2 and then move out of the area. This finding-based on criteria similar to those used by Struyk andJames (I1975)-may support the modified incubator hypothe- sis that small new firms tend to locate in centralized areas near such es- sentials as production space and financial services.' Ring 2 accounted for Table 3-8. Summary Statistics: Composition of Changes in Manufacturing Employment by Ring, Cali, 1970-75 (percent) Annual stationarn Annual Annual Origin/destination Ring growth rated birth rateb death rateb ratio of movers 1 1.90 2.86 3.22 0.60 2 -0.20 2.76 2.01 1.82 3 5.01 3.78 1.73 0.50 4 7.91 7.23 0.72 0.40 5 - 0.93 1.84 0.00 Total 2.99 4.24 1.93 0.83 -Not applicable. Note: See notes for table 3-7. a. Of mature firms. b. Annual average rate based on 1970 manufacturing employment; see table 3-3. Source: DANE industrial directory file and table 3-7. Table 3-9. Selected Analysis Zones and Incubation Areas in Bogota Employment share (percent) Average firm size (number employed) 1970 base Births Ratio Analysis zone (AZ) (1) (2) (2)1(1) 1970 base Births CBD (AZ 11) 5.60 4.16* 0.74 32.0 37.4 Industrial corridor (AZ 2-4, 2-5) 11.11 15.07* 1.36* 36.0 27.5* Northwesta (AZ 2-7, 3-7, 4-7) 2.16 5.94* 2.64* 21.3 20.3* Industrial corridor (AZ 3-4, 3-5) 37.80 26.47* 0.70 61.3 32.2 Industrial corridor (AZ 4-4, 4-5, 5-4) 22.50 15.66* 0.70 94.4 50.8 Southwestb (AZ 5-3) 3.20 5.85* 1.83* 99.6 94.8 Industrial corridorc (AZ 5-5) 0.01 4.24* 424.00* 11.0 114.4 All AZs listed 82.38 77.39 0.93 55.9 35.5 All Bogota 100.00 100.00 1.00 55.1 34.1 *Meets one of the criteria of an incubation area: a high concentration of new firms, a higher percentage of employnient by births than of base-year employ- ment, or a small average size of new firms (about 25 employees). a. This zone conitains the commercial areas of Chapinero. b. Includes Bosa area. c. Includes Fontibon area. Source: DANE industrial directory file. Components of Change 53 18 percent of the employment created by new firms in Bogota, whereas in the base year (1970) ring 2 contributed 15 percent of total manufacturing employment. (This does not hold for ring 2 in Cali, where the percentages were 26 percent and 46 percent respectively. As discussed below, the in- cubator hypothesis was tested in Cali by smaller subareas.) In Bogota rings 5 and 6 also had a share of new employment from the birth of firms which was larger than their share of total employment in the base year, although these zones are not considered incubation areas because their new firms are not small. Ring 5 is clearly attractive to new firms and movers: it had the highest annual birth rate during the period (except for ring 6, which had a small employment base), with an inflow of jobs more than ten times the outflow. The percentage distribution of firm deaths ac- ross rings is very similar to that of births. To test the incubator hypothesis more specifically, we defined twenty- eight analysis zones for Bogota by intersecting the six rings with eight radial sectors (see figure 3- 1). (An analysis zone is identified by a two-digit number: the first digit refers to the ring and the second refers to the sec- tor.) An incubation area is considered to have the following characteris- tics: (a) a high concentration of new firnms; (b) a higher percentage of em- ployment by births than of base-year employment; and (c) a small average size of new businesses (about 25 employees). The areas listed in table 3-9 were selected because of their high concentration of births during 1970- 75. In the CBD and in the industrial corridor in rings 3 and 4, the share of employment by births was smaller than the share of base-year employ- ment; in the industrial corridor in rings 5 and 2, however, the opposite was true. But in ring 5 the average size of new firms was about 100 em- ployees. Only the industrial corridor in ring 2 and the northwest (Chapinero) area thus meet all three criteria for an incubation area. In a similar test for Cali, table 3-10 indicates that only analysis zone 2-5 (the comuna to the south of the CBD) passes the three criteria. As in the case of Bogoti, the share of employment by births was smaller than the share of base-year employment in the CBD and most of the industrial cor- ridor, with the exception of a segment in ring 4. In Cali, too, the average size of new firms in outer areas was large (92 employees), indicating that the birth of large firms occurs near the urban periphery. To examine the location choices of moving firms, table 3-1 1 presents a matrix of origins and destinations for six subareas of Bogota: the CBD, three segments of the industrial corridor, the north (sectors 6, 7, and 8), and the south (sectors 2 and 3; see figure 3-1). From these statistics, it is clear that nearly 40 percent of the movers relocated within their original subareas; thirty-seven firms (13 percent) moved within analysis zones 3-4 and 3-5 alone. Although only one-third of the thirty-seven firms had 54 Manufacturing Employment Table 3-10. Selected Analysis Zones and Incubation Areas in Cali Employment share (percent) Averagefirm sitze (number employed) 1970 base Births Ratio Analysis zone (AZ) (1) (2) (2)/(1) 1970 base Births CBD 7.65 5.02* 0.66 34.7 24.6* Industrial corridora (AZ 2-3) 26.27 18.93* 0.72 61.5 30.3* South of cBDb (AZ 2-5) 1.80 3.04* 1.69* 27.8 21.6* Northern borderc (AZ 3-2) 5.50 7.52* 1.37* 186.9 73.6 Industrial corridord (AZ 3-3) 28.80 11.13* 0.39 101.9 25.6* East of cBDe (AZ 3-4) 9.92 18.07 1.82* 54.4 47.2 Industrial corridorf (AZ 4-3) 4.45 7.01* 1.58* 88.8 91.5 All AZs listed 84.69 71.74 0.85 66.4 35.8 All Cali 100.00 100.00 1.00 70.8 41.2 *Meets one of the criteria of an incubation area. Only AZ 2-5 passes all three criteria. a. Coniuna 13. b. Comnuna 14. c. Comuna 24. d. Coniunas 31 and 32. e. Coniuna 41. f. Coniuna 33. Source: DANE industrial directory file. more than 50 employees, about half of the firms that moved from ring 3 to rings 4 and 5 in the industrial corridor were large.6 As expected, the CBD experienced a net loss of firms: thirtv moved out while only four moved into the area. The industrial corridor in rings 2 and 3 and the north also lost more establishments than they gained, but the opposite occurred in the south and in rings 4 and 5. Among the six subareas, a relatively large number of firms relocated from the CBD to rings 2 and 3 in the industrial corridor, from ring 2 to ring 3, from ring 3 to rings 4 and 5, and also from the north to rings 4 and 5 of the industrial corridor; only a few firms relocated from the CBD to rings 4 and 5. That a large number of small firms moved out of ring 2 is consistent with the in- cubator hvpothesis tested above. A similar analysis for seven subareas in Cali, roughly corresponding to the radial sectors, is presented in table 3-12. The origin and destination matrix shows that 40 percent of the firms that relocated moved within the same subareas as was the case in Bogota. Because of the small absolute Components of Change 55 Table 3-11. Origin and Destination of Movers: Bogota, 1970-75 (number of rirms) Destination AZ 4-4, AZ 2-4, AZ 3-4, 4-5, 5-4, Origin CBD 2- 5 3-5 5 - 5 North South Total CBD 12 7 10 3 6 4 42 (2) (1) (3) (1) (1) (1) (9) AZ 2-4, 2-5 2 11 18 5 3 6 45 (0) (0) (1) (2) (1) (1) (5) AZ 3-4, 3-5 0 12 37 26 6 7 88 (4) (12) (13) (1) (4) (34) AZ 4-4,4-5, 0 2 6 10 4 2 24 5-4, 5-5 (1) (2) (4) (2) (2) (11) North 2 2 7 13 29 4 57 (1) (0) (2) (5) (2) (1) (10) South 0 3 4 5 3 10 25 (1) (1) (1) (1) (3) (7) Total 16 37 82 62 51 33 281 (3) (7) (21) (26) (7) (11) (75) Note: The data are for establishments with 10 or more enmployees. In parentheses are the nlunber of establishmenrts with 50 or niore emplovees at destination. For the zone systeni, see figure 3-1. Source: DANE industrial directory file. size of the mover group in Cali, however, no clear pattern emerges. Although the outer rings showed a net gain of firms, the CBD also attracted relocating firms. Since most of the firms that chose the CBD were short- distance movers from adjacent areas, however, these results probably reflect the particular subareas used for analysis rather than a trend toward concentration. Location Patterns of Industries About 80 percent of all jobs in both Bogota and Cali were in the twelve largest of the twenty-nine three-digit industries.7 As table 3-13 shows, Bogota's employment in all twelve industries except plastics was concen- trated in ring 3 during 1970-75. A substantial degree of decentralization occurred in all but three industries listed: plastics, fabricated metal, and electric machinery. These durable industries maintained a large share of jobs in rings 4 or 5 also. In the case of plastics, the employment share was 56 Manufacturing Employment Table 3-12. Origin and Destination of Movers: Cali, 1970-75 (number of firms) Destination AZ 2-3, AZ 2-2, AZ 3-3, AZ 3-4, 3-5, AZ 2-6, Origin CBD 2-5 3-2 4-3 4-4, 4-5 3-6, 4-6 AZ 2-7 Total CBD 6 2 3 2 1 1 15 AZ 2-3, 2-5 14 8 1 3 2 1 1 30 AZ 2-2, 3-2 2 5 7 AZ 3-3, 4-3 2 6 2 1 11 AZ 3-4, 3-5, 4-4, 4-5 2 3 7 1 13 AZ 2-6, 3-6, 4-6 2 1 3 AZ 2-7 2 1 2 5 Total 22 15 13 14 12 6 2 84 Note: The data are for establishments that had 10 or more employees and that moved at least between barrios. See figure 3-2 for the zone systenm. Source: DANE industrial directory file. the highest in ring 5. As in most large U.S. cities, employment in the printing industry in Bogota is concentrated in the CBD; although there was a shift ofjobs in this industry to ring 3, 24 percent were still located in ring I in 1975. As table 3-14 indicates, more employment was spatially concentrated in Cali than in BogotA: eight industries sustained two-thirds or more of their work force in a single ring during the five-year period. All twelve in- dustries in Cali, however, also showed a substantial amount of decen- tralization. (This group differs from BogotA's by including the paper and rubber industries rather than the furniture and plastics industries.) The highest concentration ofjobs occurred in ring 3 for most of the industries except for apparel, paper, printing, and rubber, which were concentrated in ring 2. The above description is with respect to the distance from the city cen- ter. To measure the extent of concentration around a centroid of indus- tries, we have computed the "standard distance" for all manufacturing establishments in BogotA and Cali for the six years from 1970 through 1975.' This calculation estimates a standardized distance, weighted by the number of establishments (or jobs), from subareas where thev are located to the centroid of establishments (or jobs) in a particular industry. As table 3-15 shows, standard distances in both cities increased over time with average firm size, indicating greater industrial dispersion. Standard Location Patterns of Industries 5 7 distances of small firms were consistently shorter than those of large firms in both BogotA and Cali, which could indicate that small firms tend to duster in the presence of agglomeration economies. In the absence of in- ternal economies of scale, small firms need to share facilities and services; they are therefore expected to cluster together or locate near the centroid of the parent industry. Although estimates of standard distances at the three-digit level were also computed, they showed no consistent differen- ces between small and large firms. Moreover, the values of standard dis- tances for small firms tended to increase during the five-year period; again the general tendency seems to be for industries to disperse over time. Table 3-16 reports the values of standard distances for the twelve lead- ing industries in Bogota. Between 1970 and 1975 the values increased for all industries except textiles, plastics, and electric machinery. These results support the general trend toward industrial dispersion, but reveal greater concentration in some industries.9 Table 3-16 also compares the location patterns of Bogota's mature firms with those of new and defunct firms. The standard distances of ma- ture firms vary little among industries, and their values are smaller than those of new firms for all industries reported except beverages and plas- tics. This result implies that new firms make different location choices than mature firms; that is, they tend to locate farther from the centroid of a given industry.' It is possible that new firms are unable to compete for the limited space in the central area or that additional space may not be available. This observation holds for the births of small as well as large es- tablishments in most industries. Moreover, small new firms that locate near the CBD-in the incubation area that does not coincide with the in- dustry centroids-gain the advantage of agglomeration economies. The findings in the previous section support this observation: table 3-9 showed that the incidence of births in relation to base-year employment share was low in the centralized industrial districts (rings 3 and 4 in the in- dustrial corridor) where the concentration of manufacturing employment was highest. The plastics, beverages, and electric machinery industries are the only cases where the standard distance of new firms was shorter than that of mature firms." This finding can be interpreted as evidence of the pres- ence of agglomeration economies in the sense that new firms tend to follow the location patterns of the stock of mature firms.'2 That the stan- dard distance of deaths was largest for the plastics industry perhaps in- dicates a high incidence of deaths among firms located far from the in- dustry's centroid. (Text continues page 62.) Table 3-13. Distribution of Manufacturing Employment by Ring and Industry: Bogota4, 1970 and 1975 (percent) Ring Industry 1 2 3 4 5 6 N i.e. Iotal Food (311) 1970 5.0 15.2 62.9 5.0 11.8 0.0 0.2 100.0 1975 2.9 15.5 53.8 9.2 18.1 0.5 0.0 100.0 Beverages (313) ',, 1970 0.0 30.9 54.2 12.7 2.2 0.0 0.0 100.0 °° 1975 0.5 4.3 52.9 16.7 25.6 0.0 0.0 100.0 Textiles (321) 1970 2.7 15.4 46.1 25.8 9.2 0.8 0.0 100.0 1975 1.1 11.0 44.0 25.9 17.2 0.3 0.7 100.0 Apparcl (322) 1970 18.5 17.2 49.6 3.3 11.4 0.0 0.0 100.0 1975 9.5 23.2 46.7 12.5 8.1 0.0 0.1 100.0 Furniture (332) 1970 3.1 19.6 36.8 32.3 8.2 0.0 0.0 100.0 1975 2.0 5.1 48.6 31.7 12.6 0.0 0.0 100.0 Printing (342) 1970 28.0 15.4 29.6 22.6 4.4 0.0 0.0 100.0 1975 23.6 10.7 38.7 20.2 5.9 1.1 0.0 100.0 Oiher chemicals (352) 1970 0.3 17.5 38.8 17.7 25.8 0.0 0.0 100.0 1975 0.2 14.6 34.5 19.3 31.1 0.3 0.0 100.0 Plastics (356) 1970 1.1 4.3 26.6 13.4 48.1 0.3 6.2 100.0 1975 0.4 5.3 30.8 12.1 40.3 0.0 11.1 100.0 Fabricated metal (381) 1970 8.2 12.5 33.9 32.3 13.2 0.0 0.0 100.0 1975 0.1 13.2 34.5 33.2 18.5 0.5 0.0 100.0 Nonclectric machinery (382) 1970 2.7 5.8 42.6 28.1 20.8 0.0 0.0 100.0 1975 1.2 7.4 36.9 25.4 28.7 0.4 0.0 100.0 Electric machinery (383) 1970 4.4 16.5 33.2 34.4 11.5 0.0 0.0 100.0 1975 0.6 11.0 48.0 27.8 12.6 0.0 0.0 100.0 Transport equipment (384) 1970 1.2 7.3 39.7 28.8 22.9 0.0 0.0 100.0 1975 0.1 8.3 28.5 29.5 33.5 0.0 0.0 100.0 N.i.e. Not inicluded elsewhere. Note: The data are for establishments with 10 oi more employees. sic in parenitheses. Soune: DANE industrial directory file. Table 3-14. Distribution of Manufacturing Employment by Ring and Industry: Cali, 1970 and 1975 (percent) Ring Industry 1 2 3 4 5 N i.e. 7otal Food (311) 1970 15.6 12.6 56.8 2.0 12.0 1.0 100.0 1975 7.7 12.6 49.6 0.7 22.3 7.1 100.0 Beverages (313) 1970 23.0 1.1 74.2 1.7 0.0 0.0 100.0 1975 0.0 23.7 67.2 0.7 0.0 8.3 100.0 Textiles (321) 1970 2.3 23.3 72.4 0.9 0.0 1.2 100.0 1975 5.0 14.5 68.6 11.9 0.0 0.0 100.0 Apparel (322) 1970 39.8 53.7 2.6 3.9 0.0 0.0 100.0 1975 32.7 45.8 15.0 4.5 0.0 2.0 100.0 Paper (341) 1970 0.7 70.6 21.1 7.6 0.0 0.0 100.0 1975 1.5 48.6 41.9 7.9 0.0 0.0 100.0 Printing (342) 1970 10.2 87.2 2.6 0.0 0.0 0.0 100.0 1975 11.3 76.9 10.8 0.0 0.0 1.1 100.0 Other chemicals (352) 1970 0.0 51.3 46.8 1.9 0.0 0.0 100.0 1975 0.2 38.4 53.4 8.0 0.0 0.0 100.0 Rubber (355) 1970 0.0 93.7 6.3 0.0 0.0 0.0 100.0 1975 0.0 60.6 5.5 0.0 0.0 34.0 100.0 Fabricated metal (381) 1970 2.0 23.3 60.5 14.2 0.0 0.0 100.0 1975 1.3 17.3 61.6 18.1 0.0 1.7 100.0 Nonclectric machinery (382) 1970 0.0 13.7 65.3 21.0 0.0 0.0 100.0 1975 5.9 18.7 61.2 14.2 0.0 0.0 100.0 Electric machinery (383) 1970 0.0 24.0 76.0 0.0 0.0 0.0 100.0 1975 0.0 29.0 66.6 4.3 0.0 0.0 100.0 Transport equipment (384) 1970 0.0 10.5 82.0 7.6 0.0 0.0 100.0 1975 0.0 5.3 86.7 8.0 0.0 0.0 100.0 N.i.e. Not included elsewhere. Note: The data are for establishments with 10 or more employees. sic in parentheses. Source: DANE industrial directory file. 62 Manufacturing Employment Table 3-15. Standard Distances of Establishments by Firm Size: Bogota and Cali, 1970-75 Standard distance (kilometers) Average firm sizea Proportion of Less than 25 or more (number of small firmsb Year 25 employees employees Allfirmsa employees) (percent) Bogotd 1970 3.54 3.69 3.66 55.1 49.0 1971 3.65 3.70 3.72 57.3 47.2 1972 3.54 3.85 3.76 62.1 44.4 1973 3.83 4.02 3.97 60.8 45.3 1974 3.94 4.02 4.02 63.6 43.7 1975 4.01 4.04 4.06 64.7 42.7 Cali 1970 1.64 1.87 1.78 70.8 46.8 1971 1.60 1.81 1.74 78.7 41.7 1972 1.84 1.87 1.88 78.8 41.6 1973 1.78 1.95 1.89 76.6 43.9 1974 1.95 2.07 2.03 80.9 41.5 1975 1.98 2.06 2.05 81.3 39.7 Note: The data are for establishments with 10 or more employees. a. All firms with 10 or more employees. b. Less than 25 employees. Source: DANE industrial directory file. The standard distance of all firms in an industry will decrease whenever the standard distance of births is shorter than that of the stock of mature firms-that is, when the degree of concentration increases. In Bogota agglomeration economies thus occur in the textile industry, whose stan- dard distance decreased over time, as well as in plastic products, beverages, electric machinery, leather, paper, instruments, and "other manufacturing." Together, however, these eight industries accounted for only 32 percent of Bogota's total manufacturing employment in 1975. As the data in table 3-17 show, the standard distances also increased for eight of Cali's twelve leading industries between 1970 and 1975. Cali ex- perienced an increasing dispersion of manufacturing industries during the period. As in Bogota, the standard distances of mature firms are smaller than those of new firms for most industries. The standard distance as a measure of the degree of concentration does not allow for the possibility of multiple centers of employment. To take account of such a possibility an "index of contiguity" was con- structed as follows. First, a measure of proximity (P) of thejth industry's employment to the sth subarea is defined in terms of employment (E) Table 3-16. Standard Distances of Establishments by Firm Type for Selected Industries: Bogota, 1970-75 (kilometers) Birlhs All Industry Mature Srnall" Large All Death) 1970 1975 Nondurable Food (311) 3.04 4.78 5.66 5.18 3.97 3.22 3.97 Beverages (313) 2.83 2.53 0.00 2.53 0.86 3.25 3.32 Textiles (321) 3.88 4.28 3.60 4.26 4.74 4.09 3.96 Apparel (322) 2.96 3.74 3.74 3.81 3.89 3.31 3.58 Other chemicals (352) 3.36 4.91 5.78 5.37 4.73 3.33 4.25 Durable Furniture (332) 3.79 4.08 3.25 3.93 4.12 4.02 4.29 Printinig (342) 2.94 2.67 5.80 4.19 2.00 2.74 3.50 Plastics (356) 3.48 2.89 3.63 3.25 5.97 3.98 3.01 Fabricated mctal (381) 3.54 3.66 5.61 4.41 2.41 3.29 3.72 Nonelectric machinery (382) 2.65 3.83 3.35 3.66 3.19 2.85 3.36 Electric machinery (383) 2.67 2.62 2.54 2.65 3.55 2.94 2.83 Transport equipment (384) 3.26 3.94 4.94 4.35 2.35 3.10 4.16 Note: The data are for establishmenits with 10 or niore employees. sic in parentheses. a. Less than 25 employees. Soarce: DANE industrial directory file. 64 Manufacturing Employment Table 3-17. Standard Distances of Establishments by Firm Type for Selected Industries: Cali, 1970- 75 (kilometers) All Industry Mature Births Deaths 1970 1975 Nondurable Food (311) 1.79 2.37 0.89 2.13 2.45 Beverages (313) 1.64 1.26 0.84 1.73 1.56 Textiles (321) 1.74 1.86 1.85 1.78 2.68 Apparel (322) 1.21 2.16 1.11 1.19 1.57 Other chemicals (352) 1.82 2.28 1.69 1.92 2.23 Durable Paper (341) 1.19 1.44 0.66 1.11 1.38 Printing (342) 0.67 1.06 0.65 0.83 0.90 Rubber (355) 1.53 0.00 0.00 1.51 1.15 Fabricated metal (381) 1.83 1.44 1.52 1.94 1.79 Nonelectric machinery (382) 1.41 3.21 1.02 1.59 1.98 Electric machinery (383) 1.24 2.20 2.23 1.83 1.76 Transport equipment (384) 1.53 2.05 1.33 1.52 2.14 Note: The data are for establishrments with 10 or nmore employees. sic in parentheses. Source: DANE industrial directory file. and distance (d) between the subareas s and t (comunas are used as subareas): = 2v j = 1, 2, ... J; s, t = 1, 2,... T 1s The value of Pj, will therefore be high when a large amount of the jth in- dustry's employment is located in subareas near the sth area. The con- tiguity index for thejth industry (CJ) is then defined as the correlation coefficient between Pp and E), across all subareas, namely Cj = C(Pjs, Ej,), where 0 < C) < 1. A high value of C1 implies that when a subarea has a large (small) amount of thejth industry's employment, the surrounding sub- areas also have a large (small) amount of that industry's employment. The contiguity index has been computed for the twelve major industries for 1975, and their values are plotted against the corresponding values of the standard distances in figure 3-3 for Bogota and in figure 3-4 for Cali. With these two measures the spatial characteristics of the employment distribution can be summarized as follows (page 67): Location Patterns of Industries 6 5 Figure 3-3. Contiguity Index and Standard Distance for Manufacturing Employment: Bogota, 1975 0.6 0.5 9383 *381 - 0.4 -*9311 *382 *321 0v 3 - *322 9352 0.2 356 *e342 *313 0.1 * _332 384 0 I I I 1 2 3 4 5 6 Standard distance (kilometers) sic codes: 311 = food, 313 = beverages, 321 = textiles, 322 = apparel, 332 = furniture, 342 = printing, 352 = other chemicals, 356 = plastics, 381 = fabricated metal, 382 = nonelectric machinery, 383 = electric machinery, 384 = transport equipment Source: DANF industrial directorv file. 66 Manufacturing Employment Figure 3-4. Contiguity Index and Standard Distance for Manufacturing Employment: Cali, 1975 0.6 .322 *383 0.5 - .321 342 0.4 _ * 382 V 0.3 _ * 381 0.2 *341 .352 0.1 X 3130 0384 0 ,3551 1 1 1 0311 1 1 2 3 4 5 Standard distance (kilometers) s,c codes: 311 = food, 313 beverages, 321 = textiles, 322 = apparel, 341 = paper, 342 - printing, 352 = other chemicals, 355 = rubber, 381 = fabricated metal, 382 = nonelectric machinery, 383 = electric machinery, 384 = transport equipment Source: DANF. industrial directorv file. Location Patterns of industries 67 Standard distance Contiguity index Small Large High Single-centered Multi-centered Low Primate Dispersed It is possible that an industry with a high value on the contiguity index can have a large standard distance or a small one; the former represents a multi-centered industry and the latter, a single-centered one. An industry with a low value on the contiguity index tends to be widely dispersed if the standard distance is large; if the distance is small, the industry will be a primate case in which most employment is in a single subarea. Figure 3-3 indicates that the contiguity index is inversely related to the standard distance, implying that manufacturing employment is single- centered in Bogota. Nevertheless, there is weak evidence of multi- centeredness in the fabricated metal and nonelectric machinery indus- tries, while the beverage industry has the primate characteristic. For most of the industries in Cali, figure 3-4 shows that standard distances are short without much variation, and some industries (rubber, paper, other chemicals, beverages, and transport equipment) with a low value on the contiguity index have primacy. In contrast, the food industry is widely dispersed in Cali. Summary This chapter presents the spatial distribution of manufacturing employ- ment in Bogota and Cali and its changes from 1970 to 1975 based on DANE's annual industrial directory data. In terms of birth, death, and relocation rates of manufacturing firms, the degree of employment dynamics was high in both cities. The annual birth rate of 8.8 percent for firms in Bogota was higher than that for all five U.S. cities compared; only Phoenix, with a birth rate of 7.6 percent (about the same as that of Cali), came close to Bogota. The birth rate was greater than the death rate in BogotA and Cali as well as in fast-growing Phoenix, while the opposite was true in large U.S. cities such as Boston, Minneapolis, and Cleveland. The incidence of births and deaths was concentrated in small firms in both BogotA and Cali. The increase in the number ofjobs in new firms ac- counted for more than 60 percent of the new employment created in both Bogota and Cali. The annual growth rate of employment in mature firms was far lower than the overall employment rate in both cities. Births and deaths of small firms contributed significantly to changes in the spatial distribution of employment. 68 Manufacturing Employment The data revealed strong evidence of decentralization of manufactur- ing employment in both cities. In Bogota, although the CBD experienced a net loss of manufacturing jobs during 1970-75, manufacturing employ- ment grew at an accelerating rate with the distance from the city center. Although Cali's CBD did not experience such an absolute decline in em- ployment, the overall decentralization was comparable to that of BogotA. Jobs were moving outward from central areas within as well as outside the industrial corridor. During the five-year period, the industrial corridor in both cities was losing its share of employment, which indicates decen- tralization from the traditional industrial areas as well. The analysis of the data shows that in both Bogota and Cali the sub- areas contiguous with the CBD displayed the characteristics of an incuba- tion area in which small new firms tend to locate for externalities such as easy access to markets and business services. The study of origins and des- tinations of moving firms indicates that nearly 40 percent of them relocated within their original subareas in both BogotA and Cali. The ma- jority of relocating firms were small and moved only a short distance from the center. Only a small number of large firms moved to outer areas. The estimates of standard distance for individual industries confirm the general trend of increasing dispersion of manufacturing establishments over time in both cities. Notes I. Some of the information on Bogota in this chapter previously appeared in Lee (1981a). For studies of the theoretical framework, see Alonso (1967), Mills (1972), and Ingram (1977a). A comprehensive review of the literature on intraur- ban manufacturinglocationappears in Kemper(1973);seealso StruykandJames (1975) and Hanushek and Song (1978). 2. This data set of DANE is very similar to the Dun and Bradstreet Market Iden- difier (DMI) data in the United States, which have been used in recent employment location studies. The DANE files, however, include only establishments with 10 or more employees, or more than 70 percent of total manufacturing emplovment in BogotA. 3. This approach was also used by Leone (1971), Schmenner (1973 and 1982), Struyk and James (1975), Kemper (1973), and Cameron (1973). 4. Although this was true for New York in the late 1960s, the situation could have been reversed by now. Both studies for U.S. cities cited in table 3-2 covered a two- to three-year period in the late 1960s. A study of such a short period may reflect only what happened in that particular portion of a business cycle. In this respect, our data set is somewhat more attractive because it covers a six-year period. 5. Hoover and Vernon (1959) focused their analysis on the incubation phe- Notes 69 nomenon in the city center; Struyk and James, however, extended the hypothesis to other centralized areas and traditional manufacturing districts. 6. The finding that long-distance movers tend to be large firms is consistent with results reported for Chicago by Moses and Williamson (1967). 7. It was 81 percent for Bogota and 83 percent for Cali in 1975. Each of these industries had at least 3 percent of the total employment. 8. The standard distance was computed using the following statistic: | LE (x, -X )2 + -E, (y, 1)2 d,= where , YE E, = the number employed in thejth industry in the ith area; (xi, yj) = the center of the ith area in terms of x, y coordinates; and (x,, -y) = the location of the cen- troid ofthejth industry. Isard (1960) had attributed its origin to Bachi (1957). The expression of this measure appearing in Isard (1960), however, has an error. The form of the squared standard distance is actually the same as that of the moment of inertia. The value of standard distance will show the degree of concentration relative to the industry's centroid. 9. Among those industries not reported in table 3-16, the value of standard dis- tance also declined for leather, industrial chemicals, and "other manufacturing" categories. 10. Cameron (1973) obtained similar results for the Clydeside conurbation in Scodand, but his measure was a simple average distance from the core of the cen- tral city. 11. Of the industries not reported in table 3-16, this was also true for leather, paper, instruments, and "other manufacturing"; these four industries together, however, had only 6 percent of total manufacturing employment in 1975. 12. In his Clydeside study Cameron (1973) concluded that new plants tend to "replicate the pattern of their parent industries with regard to their access to the center." This was interpreted as "evidence for the existence of significant agglomerative forces that mold the locational advantages for manufacturing ac- tivity," as stated by Struyk and James (1975). Cameron's test was based on the rank correlation coefficient between the average distance from the city center of new firms and that of existing firms in eighty-four industries. Determinants of Manufacturing Employment Location in Bogoti To explain the changing location patterns of manufacturing employment in Bogota, a survey of manufacturing establishments was conducted in 1978 with the industrial directory as the sample base.' This chapter pro- vides a brief description of the sampling strategy and summarizes three categories of findings: (a) characteristics of firms, (b) site characteristics, and (c) the factors considered to be important in the respondent's choice of location. These findings are broken down by the size, location history, and type of industry of the sample firms. The underlying theoretical framework is that a firm will choose a site with attributes that are optimal in terms of profits, costs, or other criteria of the firm. The relevant characteristics of the firm include type of prod- uct, production process, type of building, lot size, floor space, and labor skills. Important site attributes include proximity to product markets and suppliers, commuting distance of employees, transport modes, as well as the quality and availability of public utilities and municipal services. The Sample The sample of 126 establishments interviewed in the survey was drawn from DANE'S records of 2,629 firms in the industrial directory file for 1970-75.2 The sample of firms was stratified by (a) location history (that is, whether they were new or mature firms or movers);3 (b) zone of loca- tion defined by thirty-eight comunas; (c) type of industry defined by three-digit sic codes; and (d) size defined by the number of employees. To minimize the cost of sampling while obtaining a sufficient number of observations for econometric estimation, the survey focused on textiles and fabricated metals. These industries have no particular locational re- quirements, such as proximity to a river or mines, and should be more amenable to spatial policies than some others such as cement or steel. Moreover, together these two industries accounted for 50 percent of total 70 The Sample 71 manufacturing employment in Bogota. The homogeneity of firms within each industry also makes it possible to test behavioral hypotheses with sufficient degrees of freedom. A third group of "other industries" has been added, however, to allow descriptive studies of other kinds of firms. The second major consideration in the sampling process was to over- sample large firms in order to maximize the number of workers included. Moreover, there was an attempt to cover a wide geographic area that would allow estimation of rent and wage gradients for the entire city. The goal was to obtain a sample of at least 120 firms, with equal representa- tion of the three types of firm. As table 4-1 shows, the actual sample con- sists of 126 establishments. (Of the 128 interviews completed successfully, two cases had to be dropped from the final sample: one firm was located outside BogotA, and the other had recently moved to another city.) There were 58 mature firms, 50 movers (including two firms that moved to Bogota from outside the city), and 18 new firms. The sample coverage ac- ross zones was satisfactory; twenty-seven comunas were represented and allowed a fairly even spread over the three rings with high densities of manufacturing employment. Only a small number of establishments, however, was selected from ring I (CBD) and ring 6 (three residential com- unas in the north). In some cases the four-way stratification severely limited the possibility of drawing sample establishments from a specific population category. For example, not enough textile firms were located in certain comunas. Additional establishments were therefore selected from the apparel in- dustrv to supplement the textile industry sample and from the nonelec- tric machinery industry to supplement the fabricated metal industry sam- ple. As shown in table 4-2, the final sample has fairly even shares among the three industry groups: about 35 percent each for the two main indus- try groups and 30 percent for the "other" category. The data in table 4-3 show that the average size of mature firms in the sample is about four times larger than that of new firms, and more than twice that of movers. This results from the oversampling of large firms; the average size of firms in the sample (135 employees) is about twice as large as the average of BogotA's manufacturing establishments.4 Selected Findings of the Survey The results of the survey presented here focus on the characteristics of es- tablishments and sites and on important factors for location choices. Par- ticular attention is given to the results for movers. 72 Determinants of Manufacturing Location Table 4-1. Number and Percentage of Establishments in the Sample by Zone and Type Movers Movers within from Zone Mature New Bogold outside Total Ring 1 Number 0 2 2 0 4 Percentage of ring total 0.00 50.00 50.00 0.00 100.00 Percentage of column total 0.00 11.11 4.17 0.00 3.17 Ring 2 Number 7 3 5 0 15 Percentage of ring total 46.67 20.00 33.33 0.00 100.00 Percentage of column total 12.07 16.67 10.42 0.00 11.90 Ring 3 Number 17 6 13 1 37 Percentage of ring total 45.95 16.22 35.14 2.70 100.00 Percentage of column total 29.31 33.33 27.08 50.00 29.37 Ring 4 Number 16 3 13 1 33 Percentage of ring total 48.48 9.09 39.39 3.03 100.00 Percentage of column total 27.59 16.67 27.08 50.00 26.19 Ring 5 Number 16 4 12 0 32 Percentage of ring total 50.00 12.50 37.50 0.00 100.00 Percentage of column total 27.59 22.22 25.00 0.00 25.40 Ring 6 Number 2 0 3 0 5 Percentage of ring total 40.00 0.00 60.00 0.00 100.00 Percentage of column total 3.45 0.00 6.25 0.00 3.97 All rings Number 58 18 48 2 126 Percentage of ring total 46.03 14.29 38.10 1.59 100.00 Percentage of column total 100.00 100.00 100.00 100.00 100.00 Source: City Study establishnient survey. Characteristics of Establishments As the data in table 4-4 demonstrate, the majoritv of sarmple firms used a line-flow type production process housed in a single-storied plant. There is, however, an interesting contrast between newly established firms on the one hand, and mature firms and movers on the other. The new firms are much smaller (about one-third of the sample average) and have little land space for expansion. Compared with the other two types of firms, a Selected Findings 73 Table 4-2. Number and Percentage of Establishments in the Sample by Zone and Industry Fabrzcated Nonelectric Zone Textzles Apparel metal machinery Other Total Ring I Number I I I 0 1 4 Percentage of ring total 25.00 25.00 25.00 0.00 25.00 100.00 Percentage of column total 3.03 10.00 2.86 0.00 2.56 3.17 Ring 2 Number 3 1 4 1 6 15 Percentage of ring total 20.00 6.67 26.67 6.67 40.00 100.00 Percentage of column total 9.09 10.00 11.43 11.11 15.38 11.90 Ring 3 Number 6 6 13 4 8 37 Percentage of ring total 16.22 16.22 35.14 10.81 21.62 100.00 Percentage of column total 18.18 60.00 37.14 44.44 20.51 29.37 Ring 4 Number 12 1 9 2 9 33 Percentage of ring total 36.36 3.03 27.27 6.06 27.27 100.00 Percentage of column total 36.36 10.00 25.71 22.22 23.08 26.19 Ring 5 Number 10 1 6 2 13 32 Percentage of ring total 31.25 3.13 18.75 6.25 40.63 100.00 Percentage of column total 30.30 10.00 17.14 22.22 33.33 25.40 Ring 6 Number 1 0 2 0 2 5 Percentage of ring total 20.00 0.00 40.00 0.00 40.00 100.00 Percentage of column total 3.03 0.00 5.71 0.00 5.13 3.97 All rings Number 33 10 35 9 39 126 Percentage of ring total 26.19 7.94 27.78 7.14 30.95 100.00 Percentage of column total 100.00 100.00 100.00 100.00 100.00 100.00 Source: Citv Study establishment survey. Table 4-3. Numboer and Percentage of Establishments in the Sample by Type and Size Size (number of emaployees) 100- T-ype offirm 1-4 5 9 10-19 20-49 50- 99 9,999 To/l Mature Number 0 1 8 13 4 32 58 Percentage of row total 0.00 1.72 13.79 22.41 6.90 55.17 100.00 Percentage of colun-rl total 0.00 25.00 38.10 34.21 23.53 72.73 46.03 Avcragc size' 0.00 6.00 16.25 33.54 81.75 324.72 194.66 New Number 1 2 3 9 1 2 18 Percentageofrowtotal 5.56 11.11 16.67 50.00 5.56 11.11 100.00 < Percentage of column total 50.00 50.00 14.29 23.68 5.88 4.55 14.29 Average size' 3.00 6.00 13.00 26.56 63.00 174.00 39.11 Movers Number I I 10 16 12 10 50 Percentage of row total 2.00 2.00 20,00 32.00 24.00 20.00 100.00 Percentage of column total 50.00 25.00 47.62 42.11 70.59 22.73 39.68 Average size' 3.00 7.00 13.50 31.94 78.75 335.60 99.14 All types Number 2 4 21 38 17 44 126 Percentage of row total 1.59 3.17 16.67 30.16 13.49 34.92 100.00 Percentage of column total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Average size' 3.00 6.25 14.48 31.21 78.53 320.34 134.53 a. Average number of employees of tirrrls in each cell. Souirce: City Study establishment survey. Selected Findings 75 Table 4-4. Plant Characteristics by Type of Firm (percentage of total in each category) Plant characteristzc Mature New Moven All Production process Batch 29 44 24 29 Line flow 45 33 50 45 Both 16 11 10 13 Other 10 12 16 13 Total 100 100 100 100 Average age of buildings (years) 19 24 12 17 Land for expansion Yes 41 17 30 33 No 57 78 56 60 No answer 2 5 14 7 Total 100 100 100 100 Number of floor levels One 64 67 64 64 Two 19 17 14 17 Three 5 0 8 6 Four or more 2 0 2 2 Combination 10 17 12 12 Total 100 100 100 100 Number of eight-hour shifts One 48 83 70 62 Two 33 6 14 21 Three 16 11 16 15 No answer 3 0 0 2 Total 100 100 100 100 Average number of employees 195 39 99 135 Number of establishments 58 18 50 126 Source: City Study establishnment survey. large proportion of the new businesses tend to use single shifts in batch- process type of production. The buildings occupied by new firms are comparatively older (twice as old as those of movers) and tend to be more complex with a combination of single and multilevel structures that are characteristic of manufacturing plants in the central area. These findings are thus consistent with the incubator hypothesis that small new firms tend to choose centralized locations where various externalities are readily available. The plant characteristics of the recent movers are quite similar to those of the mature firms, except that the latter have larger operations and make greater use of multiple shifts. It is possible that the level of produc- 76 Determinants of Manufacturing Location tion technology of movers tends to approach that of mature firms, which may represent the best available. Only fifteen firms in the sample reported multiplant operations: eleven establishments were the company headquarters with the main plant, and the remaining four were branch plants. Eight firms had two branch plants, one firm had seven. Of a total of thirty-three branches reported in the survey, twenty were located in BogotA. Only two establishments of the multiplant firms specialized in a particular part of the manufacturing process-an indication of the near absence of the vertical integration of production process. The manufacturing industry in Colombia thus ap- pears to be dominated by single-plant operations with a moderate scale of production. According to the industrial directory data for BogotA, in the largest category of firms (500 or more employees) the average size was about 700 employees, only about half the average for the same category in large U.S. cities. Site Characteristics In response to the question of what value of goods was shipped by dif- ferent modes, sample firms reported that trucks were by far the dominant means of transporting input materials and finished goods. Table 4-5 shows that 81 percent of those responding to the question used trucks for shipping more than 95 percent of their final products, and 79 percent received more than 95 percent of their material inputs by truck. New firms, however, used trucks much less often than the other two groups, perhaps because roads and trucking facilities are not easily accessible in the central area. The use of trucks is a well-documented factor in the decentralization of employment in U.S. cities (for example, see Hoover and Vernon 1959; Moses and Williamson 1967). Although the highway network in Colom- bia is by no means comparable to that of the United States, it seems clear that the extensive use of trucks (or the unimportance of rail) has con- tributed to the dispersion of manufacturing jobs in Bogota. In view of the central role of the rail station near the CBD for shipment of goods only two decades ago, it is remarkable that none of the 126 sample firms used rail for shipping their products; only three firms indicated that they received as much as 20 percent of their inputs by rail. On the average, sample firms exported about 43 percent of their out- put to areas outside BogotA and imported about the same percentage of inputs (table 4-5). Compared with the extent of exporting by manufactur- ing industries in large U.S. cities, the percentage of exports from BogotA is small. According to Schmenner (1982), for example, manufacturing firms Selected Findings 77 Table 4-5. Shipments of Output and Inputs by Type of Firm (percentage of total in each category) Shipments Mature New Movers All Output shipped by truck Less than 50 percent 1.7 0.0 2.0 1.6 50-94 percent 6.9 0.0 12.0 7.9 95-100 percent 84.5 77.8 78.0 81.0 No answer 6.9 22.2 8.0 9.5 Total 100.0 100.0 100.0 100.0 Averagea 89.7 77.8 88.8 87.7 Inputs shipped by truck Less than 50 percent 1.7 11.1 2.0 3.2 50-94 percent 13.8 5.6 6.0 9.5 95-100 percent 81.0 66.7 80.0 78.6 No answer 3.5 16.7 12.0 8.7 Total 100.0 100.0 100.0 100.0 Averagea 92.4 75.3 84.6 86.8 Output sold in Bogota Less than 25 percent 15.5 16.7 22.0 18.3 25-49 percent 19.0 0.0 20.0 16.7 50-74 percent 25.9 33.3 34.0 30.2 75-100 percent 39.7 50.0 22.0 34.1 No answer 0.0 0.0 2.0 0.8 Total 100.0 100.0 100.0 100.0 Averagea 58.8 71.7 50.6 57.4 Inputs bought in Bogota Less than 25 percent 31.0 5.6 12.0 19.8 25-49 percent 13.8 11.1 24.0 17.5 50-74 percent 15.5 11.1 12.0 13.5 75-100 percent 32.8 66.7 44.0 42.1 No answer 6.9 5.6 8.0 7.1 Total 100.0 100.0 100.0 100.0 Averagea 50.7 75.7 58.1 57.2 Number of establishments 58 18 50 126 a. Average percentage of the value of output or inputs for each type of firm. Source: City Study establishment survey. in Cincinnati exported between 70 and 90 percent of their output, depending on the size of the firm; the mover firms with I 00 or more em- ployees in that city exported 95 percent of their output. BogotA's new firms are more oriented to the local markets for both output and inputs than the other two groups. 78 Determinants of Manufacturing Location Evaluation of Present Location Respondents were asked to state whether they were very satisfied, more or less satisfied, or dissatisfied with various attributes of their current locations. Reported in table 4-6 are the percentage of firms of each type that were "very satisfied" with particular attributes. For example, 43 per- cent of the mature firms responded that they were very satisfied with plant capacity at the present site. On the whole, the sample firms were verv satisfied with road access, proximity to clients and suppliers, and the availability of unskilled workers, but dissatisfied with the quality of municipal services, zonal amenities, the availability of skilled workers, and the cost of land for expansion. A much larger proportion of the movers felt very satisfied with plant capacity than did the mature firms. Only a few movers, however, were very satisfied with the cost of land for expansion. This may indicate their needs for on-site expansion at a new location or for further relocation. Another noteworthy point is that movers felt dissatisfied with the avail- ability of skilled workers. Since the majority of employees stay with a firm after it moves (table 4-9), this need for skilled workers must be in addition to the current work force. The new location may therefore be too far from the high-income residential areas or have other undesirable attributes. Table 4-6. Firms Very Satisfied with Present Location by Type of Firm (percentage of total in each category) Attribute of present location Mature New Movers All Plant capacity 43 39 62 50 Cost of land for expansion 31 17 16 23 Availability of skilled workers 31 22 10 21 Availability of unskilled workers 62 39 60 58 Quality of public utilitiesa 26 50 36 33 Quality of municipal servicesb 14 11 16 14 Road access 81 72 88 83 Proximity to clients 78 72 74 75 Proximitv to suppliers 72 28 52 58 Amenities of zone' 24 6 18 19 Number of establishments 58 18 50 126 a. For example, electricitv and water supplv. b. For example, police and fire protection. c. For example, parks, recreation facilities, and air quality. Source: City Study establishment survey. Selected Findings 79 The new firms seem to have difficulty attracting even unskilled workers- perhaps because the average monthly wage paid by this group of firms (2,700 pesos) was much lower than the sample average (3,400 pesos). All three types of firms expressed dissatisfaction with the quality of munici- pal services at their present locations. The new firms showed the highest satisfaction with the quality of public utilities and the lowest for zonal amenities relative to the other two types of firm, again reflecting their cen- tralized locations. The respondents' evaluations of their present sites are reported in table 4-7 by ring. These results indicate that the level of satisfaction with most attributes-except the quality of public utilities and municipal services- increases with distance from the center. This may indicate that infrastruc- ture investments have been lagging behind the demand for better public services at the urban periphery. The survey also asked respondents more specific questions about the quality of public services. As table 4-8 demonstrates, spatial variations occur in the frequency with which services are interrupted. For all four items, the quality tends to decline with the distance from the center, es- Table 4-7. Firms Very Satisfied with Present Location by Zone (percentage of total in each ring) Ring Ring Ring Ring Ring Ring Attribute of present location 1 2 3 4 5 6 All Plant capacity 75 47 38 52 59 60 50 Cost of land for expansion 0 27 5 33 38 0 23 Availability of skilled workers 25 7 14 15 47 0 21 Availability of unskilled workers 25 53 58 55 66 80 58 quality of public utilitiesa 50 40 41 36 19 20 33 Quality of municipal servicesb 0 27 14 18 9 0 14 Road access 50 47 89 85 94 80 83 Proximity to clients 75 80 89 76 59 60 75 Amenities of zone' 0 7 14 21 25 60 19 Number of establishments 4 15 37 33 32 5 126 a. For example, electricity, water supply. b. For exanmple, police and fire protection. c. For example, parks, recreation facilities, and air quality. Source: City Study establishnient survey. 80 Determinants of Manufacturing Location Table 4-8. Quality of Public Service by Zone (percentage of total in each ring) Ring Ring Ring Ring Ring Ring Quality ofpublic service 1 2 3 4 5 6 All Electricity never interrupted 75.0 66.7 73.0 63.6 62.5 40.0 65.9 Excellent fire protection 0.0 20.0 8.1 15.2 6.3 20.0 11.1 Sewerage and garbage collection seldom interrupted 75.0 73.3 32.4 33.3 31.3 60.0 39.7 Road never interrupted 100.0 100.0 70.3 90.9 96.9 80.0 87.3 Number of firms 4 15 37 33 32 5 126 Source: City Study establishment survey. pecially in the case of sewerage and garbage collection. In general, the sample firms were least satisfied with fire protection and most satisfied with road services. Important Factors for Location Choice As table 4-9 shows, the proportion of movers in all three main industries that had "still serviceable" conditions at the previous plant was greater than that of firms whose previous plants were cramped, obsolete, or worn out. In the case of the textile industry and the "other industry" category, the relocation tended to accompany substantial changes in technology. The fabricated metal industry, however, experienced only moderate changes in technology; as many as 40 percent of the firms in this industry kept the same production method after the move. The fabricated metal industry was also able to retain the highest percentage of workers after relocation. For the sample as a whole, 80 percent of the employees stayed with the same establishment after moving. Table 4-9 also shows that moving distance increases with firm size. This is consistent with the results presented in chapter 3 and confirms similar findings by Moses and Williamson (1967) for Chicago. Most small firms moved between one and two kilometers, whereas the large firms chose sites more than five kilometers away. For the movers as a group, however, the average distance moved was small. (Schmenner 1982 found similar evidence.) Movers' evaluations of various attributes of the firm's new plant site are presented in table 4-10. Plant space had the largest median percentage in- Selected Findings 81 Table 4-9. Characteristics of Movers (percentage of total in each category) Fabricated Nonelectlic Characteristic Textiles Apparel metal machinery Other All Condition of previous plant Good but cramped 18 25 30 0 31 24 Good but obsolete 24 50 20 33 19 24 Still serviceable 35 0 50 67 31 36 Worn out 23 25 0 0 13 14 No answer 0 0 0 0 6 2 Total 100 100 100 100 100 100 Change in technology after move Considerable 47 100 0 67 38 40 Moderate 41 0 60 33 31 38 No change 12 0 40 0 19 18 No answer 0 0 0 0 13 4 Total 100 100 100 100 100 100 Percentage of workers stayed after move Less than 50 percent 18 0 0 33 19 12 50-99 percent 29 0 50 0 25 30 100 percent 53 100 50 67 44 54 No answer 0 0 0 0 13 4 Total 100 100 100 100 100 100 Averagea 79 100 90 90 67 80 Number of establishments 17 4 10 3 16 50 Szze (number of emploYees) Distance of move Less than 25 25-99 100 or more All 1-2 kilometers 67 32 30 44 3-5 kilometers 11 18 20 16 6-10 kilometers 6 36 30 24 More than 10 kilometers 11 9 20 12 No answer 6 5 0 4 Total 100 100 100 100 Average kilometers 3.5 5.3 46.3b 12.9 Number of establishments 18 22 10 50 a. Average percentage of the number of workers who stayed after niove. b. Includes two firms that nioved to Bogota from other cities. Source: City Study establishment survey. 82 Determinants of Manufacturing Location Table 4-10. Experience after Relocation Percentage of moversa Median After After After percentage Item < Before = Before > Before increaseb Production 12.2 8.5 79.6 29 Plant space 10.4 2.1 87.5 60 Land space 12.5 2.1 85.4 51 Number of skilled workersc 8.7 39.1 52.2 22 Wage of skilled workers' 0.0 43.4 56.5 30 Number of unskilled workers 10.4 25.0 64.6 25 Wage of unskilled workersd 0.0 34.0 66.0 31 Distance to work by managers 18.8 43.8 37.5 9 Distance to work by workers 22.9 52.1 25.0 1 Distance of product shipment 12.5 68.8 18.8 2 Distance of input delivery 18.8 62.5 18.8 0 Profitsd 17.0 29.8 53.2 11 Local tax payments 2.1 29.2 68.8 32 Cost of public servicesc 6.5 26.1 67.4 51 a. There were 48 movers in the sample that moved within Bogoti. b. Median of actual percentage increase after move among all firms that responded. c. Two firms did not respond. d. One firm did not respond. Source: City Study establishment survey. crease, followed by land space and the cost of public services (obviously the total, not the unit, cost of public services). There were virtually no changes, however, in the shipping distances for inputs and output or in the commuting distance of workers. This does not mean that these factors were unimportant in the location decision; indeed, it is likely that these factors were so important that firms chose sites that did not significantly alter proximity to markets, suppliers, and labor. Of the factors that may influence location choices listed in table 4-1 1, plant capacity was considered the most important by 75 percent of the movers. Rent payments, proximity to suppliers, and amenities of the zone were also important. The proximity of clients, the quality of public utilities, and the cost of land for expansion were moderately influential in the location decision. It is significant that the majority of firms did not consider the quality of public utilities and municipal services better after relocation. Indeed, as table 4-12 indicates, 63 percent felt that waste removal actually became worse. Electricity was the only service that a substantial number of firms Selected Findings 83 Table 4-1 1. Importance of Factors in Choosing Present Location (percentage of nmovers for each order of importance) Order of importance Factor First Second Third Plant capacity 74.5 6.1 13.0 Rent 4.3 18.2 0.0 Availabilitv of skilled workers 0.0 3.0 0.0 Cost of skilled workers 0.0 0.0 4.3 Availability of unskilled workers 0.0 0.0 0.0 Cost of unskilled workers 0.0 0.0 0.0 Cost of public utilities 0.0 0.0 0.0 Quality of public utilities 0.0 6.1 13.0 Cost of land for expansion 6.4 6.1 13.0 Proximity of suppliers 2.1 18.2 4.3 Proximity of clients 6.4 6.1 17.4 Proximity of competitors 0.0 0.0 0.0 Road access 2.1 9.1 8.7 Rail access 0.0 0.0 0.0 Proximity to business services 0.0 3.0 0.0 Property taxes 0.0 0.0 0.0 Municipal services 0.0 3.0 4.3 Security 0.0 0.0 0.0 Amenities of zonea 4.3 18.2 17.4 Community attitude 0.0 0.0 4.3 Other 0.0 3.0 0.0 Total 100.0 100.0 100.0 Number of movers responding 47 33 23 a. For example, parks, recreation facilities, and air quality. Source: City Study establishment survey. Table 4-12. Changes in Quality of Public Services after Relocation (percentage of movers) Substantially Somewhat Became Service improved improved Unchanged worse Total Electricity 29.2 10.4 52.1 8.3 100.0 Water 18.8 6.3 72.9 2.1 100.0 Fire protection 8.3 16.7 68.8 6.3 100.0 Police service 2.1 8.3 66.7 22.9 100.0 Sewerage 12.5 6.3 70.9 10.4 100.0 Waste removal 2.1 6.3 29.2 62.5 100.0 Road maintenance 10.4 8.3 54.2 27.1 100.0 Source: City Study establishment survey. 84 Determinants of Manufacturing Location felt improved after relocation. These results may imply that the public sector has little influence on the location choices of Bogota's manufactur- ing firms; it may also reflect the lack of specific policies to induce indus- tries to choose certain areas. Future plans to expand and relocate reflect the extent of firms' need for further adjustment to attain an optimal location in an urban area. About one-third of the mature firms in the sample indicated that they had seriously considered relocating during the eight-year period since 1970, and 16 percent had considered opening branch plants. As the statistics in table 4-13 show, the mature firms that considered relocation were mostly of medium size. Although only a few firms opened branch plants during this period, the majority of establishments experienced on-site expansion. Table 4-14 shows that 22 percent of all sample firms intended to relo- cate within the next five years, which amounts to an annual relocation Table 4-13. Relocation, Branch Operation, and On-Site Expansion of Mature Firms by Size (percentage of total) Number of employees Less than 100 or Item 25 25-99 more All Relocation considered since 1970 No 91 27 81 69 Yes 9 73 19 31 Total 100 100 100 100 Branch considered since 1970 No 91 93 78 84 Yes 9 7 22 16 Total 100 100 100 100 Branch opened since 1970 No 100 93 84 90 Yes 0 7 16 10 Total 100 100 100 100 Expanded on present site since 1970 No 45 60 9 29 Yes 55 40 91 71 Total 100 100 100 100 Average number of employees 16 48 325 195 Number of establishments 11 15 32 58 Source: City Study establishment survey. Selected Findings 85 Table 4-14. Expansion and Relocation Plans by Type and Size of Firm (percentage of total) Movers Movers Plan Mature New within from oz/side All Plan to expand in next 5 years No 60 61 50 0 56 Yes 40 33 50 100 44 No answer 0 6 0 0 1 Total 100 100 100 100 100 Plan to relocate in next 5 years No 66 72 73 50 69 Yes 22 17 23 50 22 No answer 12 11 4 0 9 Total 100 100 100 100 100 Number of establishments 58 18 48 2 126 Size (number of emplo'vees) Less than 25 25-99 100 or more All Plan to expand in next 5 years No 68 60 41 56 Yes 30 40 59 44 No answer 2 0 0 1 Total 100 100 100 100 Plan to relocate in next 5 years No 65 52 89 69 Yes 20 38 9 22 No answer 15 10 2 9 Total 100 100 100 100 Average number of employees 15 54 320 135 Number of establishments 40 42 44 126 Source: City Study establishment survey. rate of about 4 percent. Of all types of firm, medium-size establishments (with an average of 54 employees) showed the highest propensity to move. Summary Results of the City Study establishment survey revealed that Bogota's manufacturing industries are primarily single-plant operations with a moderate production scale. The majority of sample establishments used a line-flow production process housed in a single-storied plant and 86 Determinants of Manufacturing Location operating with one shift. Newly established firms are small and tend to be located in much older and more complex building structures than those of mature firms; they also tend to employ a batch-process production technique. Most shipments of inputs and final products are made by truck; rail is seldom used. Among the new firms, however, the use of trucks is more limited, which perhaps indicates the difficulty of using trucks in a congested area. These characteristics of small new firms sup- port the incubator hypothesis that their birth place tends to be in an old district near the city center where various externalities are readily available. Sample firms as a whole exported less than 50 percent of their products to areas outside Bogota. The lack of multiplant operations and the mod- erate production scale seem consistent with the local market orientation of manufacturing firms in BogotA. It is therefore difficult to consider these manufacturing establishments as the export sector. The majority of firms expressed satisfaction with the availability of un- skilled workers, road access, and proximity to clients and suppliers, but dissatisfaction with the cost of land for expansion, the availability of skilled workers, the quality of public services, and zonal amenities. The level of satisfaction with a particular location increased with distance from the center for all attributes except the quality of public utilities and municipal services and the proximity to clients. In general, plant relocation tends to be associated with changes in pro- duction technology. In the case of the fabricated metal industry, however, 40 percent of the movers kept the same production method after reloca- tion. For the majoritv of firms that moved, the distance from the old to the new site was fairly short (about one or two kilometers); large firms, however, moved longer distances. Although more than 80 percent of a firm's workers remained with the company after it relocated, many movers found it difficult to attract new skilled workers. The movers sub- stantially increased both plant and land areas at the new location and ex- perienced a moderate increase in production costs. For the majority of firms, however, both input and output delivery distances and the com- muting distance of production workers stayed about the same after the move; the commuting distance of administrative workers increased only slightly. The increase in plant space was the most important factor in the firm's location choice, followed by rent, proximity to suppliers, amenities of the zone, road access, proximity to clients, the quality of public serv- ices, and the cost of land for expansion. The survey findings in this chapter support the employment location patterns observed in chapter 3 and provide some explanations for them. Summary 87 In many respects the results for Bogota are similar to those for U.S. cities obtained by Schmenner (1982). Notes 1. Roger Schmenner's suggestions were very helpful in designing the survey questionnaire and selecting the sample establishments. A copy of the question- naire appears in appendix B. 2. An earlier version of this section appeared in K. S. Lee (1982). Although the DANE files actually contained 3,388 records for the six-year period, firms with Icss than 10 employees or those that appeared only for one year were not included in the master file. The basic structure of the industrial directory data was docu- mented in K. S. Lee (1978). 3. As in chapter 3, new firms are those that appeared for the first timc during 1971-75; mature firms are those that appeared in all six annual directorics with the same address; movers are those that relocated within BogotA during 197 1- 75. 4. According to the industrial directory file of 1975, 1,829 establishments with 10 or more employees had an average size of 65 employees. 5 A Model of Manufactuiing Employment Location Although research on various approaches to location theory began at the turn of the century,' empirical studies have tended to use an optimization framework that emphasizes transport costs (Isard 1960). According to the transport cost theory, firms choose locations that minimize transport costs per unit of output, given production costs that are the same everywhere, constant returns to scale, and fixed location of product and materials (inputs) markets. Other than characterizing location choices as either "materials oriented" or "market oriented," this theory does little to explain the actual decision. If the assumption of identical production costs is relaxed, however, variations in input costs among different sites become important determinants of employment location. Moses and Williamson (1967), for example, have suggested that a land price gradient and other input price gradients determine a firm's location choice relative to the central city. Mills (1972) and Solow (1972) provided the basic theoretical foun- dations for economic analysis of residential and employment locations in an urban area. Although research on housing and residential location choices advanced quickly during the 197 Os,2 comparable progress in the field of employment location has been impeded by the lack of data. Nevertheless, studies by Leone (1971), Schmenner (1973), and Kemper (1973), followed by Struyk and James (1975)- have been significant in developing an empirical basis for analyzing employment location deci- sions. Moreover, recent modeling efforts have been quite promising: Hanushek and Song (1978) developed a framework for analyzing the spa- tial structure of employment in the Boston metropolitan area; Erickson and Wasylenko (1980) estimated a model of relocating firms in the Milwaukee metropolitan area; and Schmenner (1973, 1982) provided em- pirical results from his econometric work on Cincinnati and New Eng- land. Carlton (1979, 1983) has also analvzed the location decisions of new firms with the use of U.S. metropolitan area data. 88 Theoretical and Empirical Framework 89 Theoretical and Empirical Framework for Modeling Employment Location Residential location studies usually assume that a household will choose a dwelling unit at a location that maximizes utility within the constraints of the household's budget. In a similar optimization framework, the em- ployment location model described here assumes that the firm, as a price taker, locates where it maximizes profits. (For a formal presentation of the model, see the appendix to this chapter.) The attributes of the particular plant site and the lot size directly influence the optimum combination of inputs for production and thus directly enter into the firm's production function in the following way: (5-1) Q = (L, X; Z), where Q = output, L = lot size, X = a vector of variable inputs such as labor, plant, and equipment; and Z = a vector of site characteristics. Site characteristics are independent of lot size and represent local public goods available at a particular location.3 In the firm location model, local public goods include the quality of public utilities such as el- ectricitv and water, the quality of municipal services such as police and fire protection, and zonal amenities such as air quality and level of congestion. In calculating the optimal combination of inputs-or in selecting a par- ticular location-the relevant cost components are wages, capital costs, input costs, delivery costs of inputs and output, and land rent. Following standard urban economic theory, a particular plant site is then occupied by the firm willing to pay the highest price. The bid price will depend on the attractiveness of the site to a particular firm. In a locational equilibrium, all firms of a particular type in an urban area make the same profit and no firm has an incentive to relocate. In this case, the cost tradeoffs calculated by individual businesses determine the spatial distribution of firms. For example, a large manufacturing plant may choose a site in a low-rent area near the urban periphery to meet its need for more space, but at the expense of a greater delivery distance. Small firms, in contrast, may prefer a central location where the avail- ability of various externalities more than compensates for high rents. The above theoretical framework leads to an empirical framework for predicting the probability that a firm of a particular type will occupy a site with particular attributes (Z). Since the firm with the highest bid will oc- cupy a given site, the relevant random variable for determining this prob- ability is the maximum amount that similar firms would pay (that is, the maximum bid). The probability distribution of a random variable 90 A Model of Manufacturing Location associated with the maximum bid leads to a multinomial logit specifica- tion for the firm location model. (For more details of the stochastic specification, see the appendix to this chapter.) Multinomial Logit Models in Urban Economic Research The application of multinomial logit models to urban economic research became popular with McFadden's work on travel demand (1973, 1974, 1976). Such a model was used to predict an individual's choice of travel mode among a finite number of alternatives-such as car, bus, taxi, metro, or foot-given the characteristics of the individual or household. Subsequently, the multinomial logit framework was applied to studies of housing and residential location; for instance, Friedman (1975), Lerman (1977), and Quigley (1976). A utility-maximizing consumer type t chooses a house of type Z, which is analogous to the consumer's choosing a travel mode (Z) among several alternatives. This specification can be written in probabilistic terms as follows, wherep refers to the probability and g(Z) is the utility function: (5-2) p(Z It) = p[g(Z)] Ellickson (1977, 1981) offered an alternative multinomial logit specifi- cation for the residential choice model in an important departure that used the bid-rent theory. According to Ellickson (1981, p. 63), "the most natural way to interpret such models is in terms of a prediction of what sort of consumer is most likely to occupy a house with a specified set of characteristics." Then the probability that a house with characteristics Z will be occupied by a household of type t can be written as: (5-3) p(tjZ) = p[h(Z)] where h(Z) is the bid-rent function. (For a fully specified version of this equation, see equation 5-18 in the appendix to this chapter.) Ellickson points out that his approach has several advantages, such as en- dogenously specifying the properties of the disturbance term as having the Weibull distribution. Ellickson's approach is particularly suitable for modeling employment location. As discussed above, our problem is to predict the probability that, given a site with particular characteristics (Z), a firm of a particular type will occupy that site. The most important advantage of this approach is that the model allows us to observe a wide range of spatial variations in site characteristics. In contrast, the McFadden type of approach requires specifying a small number of subareas of a city from which a firm of a par- ticular type is supposed to choose; it therefore restricts the number of Multinomial Logit Models 91 variations in site characteristics. Furthermore, defining subareas in this fashion tends to be arbitrary and introduces idiosyncracies that will mar the subsequent analysis, since the results will be affected by the wav sub- areas are partitioned as altemative locations to choose from. Another advantage of the Ellickson type of specification is that the model allows policy analysis. Since it predicts the probability that a firm Qf a particular type will occupy a site with particular attributes and these attributes are subject to manipulation by government policy, the model can be used to analyze the effects of industrial location policies. In prac- tice, such policies would direct selected industries to specified areas, often to sites prepared by government programs. Results of Multinomial Logit Estimation The multinomial logit model, specified as equation 5-3, has been es- timated with data from the Bogoti City Study establishment survey. As discussed in chapter 4, the survey collected information on plant charac- teristics, the composition of the firm's work force, access to transporta- tion, proximity to markets, the quality of local public services, and the respondent's evaluation of the plant location. Firms can thus be stratified in many ways: by variables related to output, such as product type and an- nual sales; by those related to technology, such as type of production pro- cess and building structure; and by those related to inputs, such as plant space, lot size, and the number of production workers. The site charac- teristics to be used as independent variables include those associated with accessibility to various types of markets (product, material inputs, and labor) and those associated with the quality of local public services. Of the 126 firms in the sample, 87 are in the textile and the fabricated metal industries. For the dependent variable, the 87 firms are grouped into two plant sizes according to floor space (see table 5-1). The indepen- dent variables include access to local markets for output and material in- puts, measured by the proportion of output sold and inputs bought in Bogota; proximity to the residential areas of production and administra- tive workers; an index of the quality of local public services, measured by the frequency with which electricity is interrupted; the presence of agglomeration economies, measured by the employment location quo- tient of individual industries in the zone; and the intensity of economic activity and degree of congestion, measured by the population density in the zone. Distance to the CBD is also included as a measure of accessibilitv to the city center. (See table 5-2.) Ideally, the dependent variable should be stratified by more than the two-way (and four-cell) classification used here; the small sample size, 92 A Model of Manufacturing Location Table 5-1. Stratification of Firrms to Define the Dependent Variable Number of Industry Floor space observations Group I Small textile firmsa Less than 1,000 square 17 meters Group 2 Large textile firmsa 1,000 square meters or 26 more Group 3 Small fabricated metal Less than 1,000 square 27 firmsb meters Group 4 Large fabricated metal 1,000 square meters or 17 firmsb more Total 87 Note: Floor space is used for the stratification. a. Includes both textiles (sic 321) and apparel (sic 322). b. Includes both fabricated metal (sic 381) and nonelectric machinery (sic 382). however, precludes this possibility. Therefore the two additional stratifi- cation variables are included on the right-hand side of the equation: the year of initial operation at the present location (which discriminates ma- ture establishments from new firms and recent movers) and the ow- nership dummy variable (to distinguish renters from owners). All independent variables enter the model as "group-specific" (that is, alternative-specific) except for the location quotient variable and the ow- nership dummy variable; the former is specified as "generic" within the same industry group, and the latter within the same size group. In es- timating this multinomial logit model, large fabricated metal firms (group 4) are the reference group. Since the coefficients of group-specific variables should therefore be interpreted as relative differences, their signs do not necessarily mean the direction of causation; they merely reflect the relative orders of magnitude of individual coefficients with re- spect to the reference group. The t statistics in table 5-3 indicate that the differences in coefficients between two size groups (small and large firms) are significant, especially within the same industry (groups 3 and 4). None of the coefficients of group 2 (large textile firms) is statistically significant. The likelihood ratio index of 0.29 suggests that the overall goodness-of-fit is adequate. These patterns also hold true when the model is specified with lot size or employ- mentvariables in place ofthe floor space variable. (See tables 5-5 to 5-8 at the end of this chapter for the results of these alternative specifications.) The elasticities of probabilities reported in table 5-4 measure the per- centage change in the probability of being in the ith group with respect to Multinomial Logit Estimation 93 Table 5-2. Definition of Independent Variables CONSTANT = Group-specific constants PRODSOLD = Percentage of product sold in BogotA INPUTBT = Percentage of inputs bought in BogotA DISTCBD = The airline distance in kilometers from the CBD (the center of comuna 31) to the center of the comuna in which the establishment is located WKSOUTH = Percentage of production workers living in the south ADMNORTH = Percentage of administrative workers living in the north ELECINT = Frequency of electricity interruption (1 = never, 2 = once a week, 3 = twice a week, 4 = more than twice a week) POPDENS = Population per hectare of the comuna in which the establishment is located LOCQT = Location quotient defined as comunaj's share of industry i relative to its share of total manufacturing employment (separate values are used for the two industry groups) YRINOP = Year of initial operation at the present location RENTER = Ownership dummy (I if renter, 0 if owner); this dummy variable has been assigned to establishments with floor space of less than 1,000 square meters in both industry groups. a 1 percent change in a given independent variable. For example, if the distance from the CBD (DISTCBD) increases by I percent, the probability of the firm's being in group I increases by 0.05 percent, that of being in group 2 by 0.16 percent, and that of being in group 3 by 0.58 percent. This implies that the small fabricated metal firms (group 3) tend to locate farther from the CBD than the other two groups; large textile firms (group 2) tend to locate about three times farther away from the CBD than small textile firms (group 1). Small metal-fabricating firms (group 3) have the highest elasticities for most variables but are least sensitive to the rate of electricity interruption (ELECINT) and the location quotient (LOCQT). The variable that most influences the probability of being in group 3 is the measure of access to local input markets (INPUTBT), followed by the measure of access to local product markets (PRODSOLD). For small textile firms (group 1), the measure of access to local input markets is also the most important variable, followed by proximity to production workers' residential areas (WKSOUTH). The weakest variable in this case is distance from the CBD, indicating that small textile firms tend to locate nearer the center than the other two groups. As noted above, the probability of being in group 2 becomes three times higher than that of being in group I as distance from the CBD increases. Small metal- fabricating firms, however, tend to locate farther from the CBD than do both small and large textile firms. Table 5-3. Logit Estimation of Firms' Location Choice (Dependent Variable: Industry and Floor Space) Coe]fcients t statistics' Independent variable Group I Group 2 Group 3 Group 4 Group I Group 2 Group 3 Group 4 CONSTANT -15.680 -2.128 -12.880 - 2.09' 0.57 2.07*- PRODSOLD 0.011 0.008 0.028 - 0.74 0.60 1.83 - INPUTBT 0.019 -0.010 0.027 - 1.39 0.89 2.05*- I)ISTCBD 0.012 0.032 0.151 - 0.07 0.21 0.92 - WKSOUTH 0.014 0.003 0.022 - 0.80 0.20 1.33 - ADMNORTH -0.010 -0.016 -0.020 - 0.64 1.12 1.40 - ELECINT 0.501 0.448 0.115 - 1.05 1.11 0.24 0 POPDENS 0.008 0.002 0.012 0 1.11 0.35 1.89_ LOCQT 0.749 0.738 1.69 1.71 YRINOP 0.159 0.033 0.095 1.63 0.60 1.20 RENTER 2.069 - 2.069 - 2.67 - 2.67* Number of observations 17 26 27 17 Percentage correctly predicted 54.02 Likelihood ratio index 0.2903 Likelihood ratio statistic 70.02 -Not applicable. Note: Definitions of the variables are given in tables 5-1 and 5-2. Group 4 is used as the base. a. Trhe coefficients with a single asterisk are significant at the 5 percent level; those with a double asterisk are significant at the 2.5 percent level. Source: City Study establishment survey. Multinomial Logit Estimation 95 Table 5-4. Elasticities of Probability: Logit Estimation of Location Choice (Dependent Variable: Industry and Floor Space) Independent variable Group I Group 2 Group 3 Group 4a PRODSOLD 0.515 0.272 1.367 - INPIJTBT 1.182 -0.293 1.455 - DISTCBD 0.052 0.155 0.584 - WKSOITTH 0.808 0.128 1.120 - ADMNORTH -0.496 -0.538 -0.689 - ELECINT 0.711 0.556 0.123 - POPDENS 0.794 0.124 1.233 - LOCQT 0.544 0.722 0.468 - YRINOP 9.264 1.585 4.630 - RENTER 1.665 - 1.467 - Share 0.1954 0.2989 0.3103 0.1954 -Not applicable. Note: Definitions of the variables are given in tables 5-1 and 5-2. The elasticity of probabil- ity is defined as eib = (I - p,)bi X , wherep= the share of the zth group, b, = jth logit coeffi- cient of the ith group, and XI, = sample mean of thejth independent variable for the zth group. The logit coefficients reported in table 5-3 are the differences with respect to the coef- ficients of the base group. Therefore, the values of elasticities in this table are the results based on (b, - b*) instead of by, where b14 is the coefficient of the base group. a. Group 4 is used as the base. Source: City Study establishment survey. In the case of large textile establishments (group 2), the most important variable is the location quotient (LOCQT), followed by the electricity in- terruption rate (ELECINT) and proximity to the residential areas of adminis- trative workers (ADMNORTH). For this group of large firms, the measure of access to local markets and the proximity to production workers' residen- tial areas seem negligible, reflecting the fact that these companies tend to be export-oriented and use capital-intensive production facilities. The fact that large firms are less likely to locate in densely populated areas (POPDENS) is consistent with the finding that they tend to locate at some dis- tance from the CBD. Summary The results of the establishment survey conducted in Bogota are used to test a multinomial logit specification of bid-rent function, following the approach used by Ellickson (1981). The model was estimated with a two- 96 A Model of Manufacturing Location way stratification of the dependent variable by industry type and floor space; independent variables included measures of access to output and input markets, indexes of concentration of economic activities, and an index of the quality of public utility services. Even though the sample size was not large, the goodness-of-fit was satisfactory, and the estimated model was capable of predicting, in probabilistic terms, which types of firms are likely to occupy a site with particular characteristics. The location patterns that the model predicts are consistent with those expected. Among small firms, accessibility to local input and output markets is the most important factor in the location decision; the benefits of accessibility to the central area tend to compensate for the high rents and congestion costs. Large establishments, which are more export- oriented and require more plant space with modern production technol- ogy, tend to locate in outer areas where more space is available at lower cost. The estimation results also indicate that the quality of public utility services is very important for large firms and that proximity to adminis- trative workers' residences is more important than proximity to produc- tion workers' residences. Appendix: Theoretical and Empirical Specification of the Model Consider T types of manufacturing firms in an urban area.4 The firm maximizes profits as a price taker in both product and factor markets. The firm uses a set of variable and fixed inputs to produce an output. The problem to be solved is to determine the optimal combination of inputs, including the lot size and the plant location, to attain locational equilib- rium profits in an urban area. The production function is specified as equation 5-1, Q= f (L, X; Z), where Q= output, L = lot size, X = a vector of other inputs such as labor, plant, and equipment; and Z = a vector of site characteristics that are in- dependent of lot size and can be considered as "local public goods,"5 such as the quality of public utility services, accessibility to markets, and amenities of the zone of plant location. The firm's profit is (5-4) n = pf(L, X; Z) -RL -wX where nT = profit,p = output price, R = land rent per unit, and w = other input prices such as wage rate and price of capital. From the first-order conditions for profit maximization, one obtains the following demand equations for variable inputs: Spec ifcation of the Model 97 af _R (5-5) aL p af w (5-6) 8x p ax p Solving equations 5-5 and 5-6 for the optimal input quantities V and X*, and substituting them into equation 5-4, the "profit function," based on the duality theorem,6 is obtained as (5-7) n* =pf(L*, X*; Z) - RL* - wX* = n*(p, R, w; Z) Let v = unit transport cost for shipment of output; thenp - v is the fac- tory price of output. Usingp as the numeraire and introducing the loca- tion variable (u), equation 5-7 becomes (5-8) n*(u) = g[l - v(u), R(u), W(u); Z(U)], where -n, v, R, and wv are values normalized byp, and u refers to the dis- tance to the product market. In locational equilibrium, for a given u every firm should have the same profit, and there is no incentive for any firm to relocate. An equilib- rium rent profile must satisfy (5-9) in *(u) = g[1 - v(u), R(u), wv(u); Z(u)] = constante As with residential location, this formulation of a firm's choice of location is in terms of the firm's bid-rent function, which gives the price for a site with characteristics Z that yields a profit n*. Let R*(u) denote the bid rent, then (5-10) R*(u) = h[ - I(u), w(u); Z(u); ffn (u)] h z 98 A Model of Manufacturing Location For convenience, suppose the unit transport cost is site-invariant within an urban area and include it as an element in the constant term. Also suppress n*(u), which is constant. Hence equation 5-10 can be writ- ten as (5-11) R*(u) = h[v(u); Z(u)] where (5-12) aR* < 0; 'az > O For illustration, consider the case of labor input. As the labor-land ratio increases, the marginal product of land increases relative to that of labor, and the relative price of land with respect to labor also rises. This argu- ment supports the empirically observed rent gradient in an urban area in the sense that as the distance to the CBD becomes shorter the intensity of a variable input such as labor increases and the land rent rises. (A measure of the land price gradient using the City Study survey data appears in chapter 7.) In other words, producers respond to differences among areas in the price of inputs to obtain optimal combinations of inputs, including lot size. Also the value of land increases with improvements in the desir- able characteristics, such as public services and accessibility. Since wis the input price vector normalized by output price, equation 5-6 can rewritten as (5-13) af' (u) = "W(u) af Substituting equation 5-13 into 5-11, we have the bid-rent function ex- pressed in terms of characteristics of afIaX and site characteristics Z. For expository reasons, rewrite equation 5-11 as (5-14) R*(u) = h[x(u), Z(u)] where x(u) = [af/aX(u)] now represents a vector of a firm's characteristics, or combination of inputs, which in turn depend on the technology as characterized, for example, by type of production process and building structure. As mentioned earlier Z(u) is a vector of site characteristics. Now suppose that there are T types of firms defined by x and S types of site defined by Z. Let N be the number of type t firms in the market. Then using equation 5-14, the bid-rent for a site with characteristics Z by the nth firm of type t is given by (5-15) = h,,(z.), n E Nt Specification of the Model 99 We have now suppressed the vector x(u), which is used to define the firm type t. For example, all firms of type t have a similar output, combination of inputs, and technology; that is, they have an identical production func- tion. Following Ellickson's (1977, 1981) work on residential location we can interpret this model as predicting the probability that a certain type of firm (t) will locate at a site with a specified set of characteristics (Z). The stochastic version of equation 5-15 is (5-16) Rl = h,,(Z4) + e, n C N, where e is a random disturbance term reflecting unaccounted variations in the characteristics of type t firms. Since a given site is occupied by the firm with the highest bid, the rele- vant variable for determining the probability that a given site is occupied by a firm of type t is the maximum bid given by firms of type t. (5-17) max(R,n) h,(Z) + e,, t C T where e, = max (e,n), n C N, If the e, are identically and independently distributed in the Weibull manner,8 the specification of a logit model follows. In other words, the probability that a firm of type t occupies a site with characteristics Z takes the logit specification9 (5-18) p(tIZ) = exp [h,(Z)I E exp [h, (Z)] The above discussion shows that the basic theoretical approach used in the study of residential location can provide a useful analytical framework for the study of employment location.'" The firm's optimizing behavior is postulated as location specific; that is, the firm's choice of a specific site is partofthe production decision. Furthermore, the location-specificequilib- rium position of individual firms is extended to the locational equilibrium situation of all firms in an urban area. The theoretical model is easily ex- tended to the stochastic specification of the model in an estimable form. Notes 1. Weber (1929) was originally published in 1909. 2. Major work in this area included Muth (1969), Kain and Quigley (1975), Straszheim (1974), and Ingram (1977b). 100 A Model of Manufactuing Location 3. In a recent work on housing, Burstein (1980) explicitly introduces local public goods into the consumer's utility function; Ellickson (1981) also integrates the theory of residential location choice with that of local public goods by includ- ing in the consumer's utility function the attributes of the surrounding neighbor- hood and public schools. 4. An earlier version of this appendix appeared in K. S. Lee (1982). 5. Burstein (1980) included this variable in the household utility function in her housing demand study. 6. For the duality relations between the production function and the profit function, see Diewert (1974), and Lau and Yotopoulos (1971). 7. Solow (1972) shows an equilibrium rent profile of households in an urban area. 8. For example, the maximum value of an identically and independently dis- tributed normal variate has the Weibull distribution. 9. Ellickson (1977, 1981) derived this variation of the logit model in his residential location study. 10. Theoretical and empirical work is rare in this area. Mills (1972) and Solow (1972) offer basic micro foundations; the work by Hoover and Vemon (1959), Struyk andJames (1975), and Schmenner (1982), although descriptive, serves as the empirical basis in the field. Table 5-5. Logit Estimation of Firms' Location Choice (Dependent Variable: Industry and Lot Size) Coefficients t statisticsa Independent variable Group I Group 2 Group 3 Group 4 Group I Group 2 Group 3 Group 4 CONSTANT -15.100 -1.475 -11.050 - 1.98** 0.40 1.81 - PRODSOLID 0.019 0.009 0.042 - 1.29 0.73 2.68*- INPUTBT 0.009 -0.014 0.016 - 0.67 1.36 1.24 - DISTCBD -0.082 -0.043 -0.042 - 0.55 0.32 0.26 WKSOUTH 0.005 -0.001 0.016 - 0.35 0.003 0.99 - ADMNORTH -0.004 -0.009 -0.013 - 0.33 0.75 0.96 - ELECINT 0.513 0.487 0.282 - 1.14 1.30 0.62 - POPDENS 0.0071 0.0 .1 1.11 1 0.14 1.97*_ LOCQrT 0.729 0.001 0 01 - 1.67 1.780 YRINOP 0.166 0.029 0.075 ) - 1.67 0.55 0.97 RENTER 1.223 - 1.223 - 1,751 - 1.751 _ Number of observations 16 27 23 21 Percentage correctly predicted 54.02 Likelihood ratio index 0.2639 Likelihood ratio statistic 63.67 -Not applicable. Note: Industry categories are defined in table 5-1; lot size is the second variable used for grouping: groups I and 3 have lots of less than 1,000 square meters. Independent variables are the same as in table 5-2. Group 4 is used as the base. a. The coefficients with a single asterisk are significant at the 5 percent level; those with a double asterisk are significant at the 2.5 percent level. Source: City Study establishment survey. 102 A Model of Manufacturing Location Table 5-6. Elasticities of Probability: Logit Estimation of Location Choice (Dependent Variable: Industry and Lot Size) Independent variable Group I Group 2 Group 3 Group 4 PRODSOLD 0.930 0.296 2.333 - INPUTBT 0.558 -0.424 0.911 - DISTCBD -0.336 -0.202 -0.158 - WKSOUTH 0.286 -0.002 0.879 - ADMNORTH -0.197 -0.305 0.476 - ELECINT 0.733 0.597 0.325 _ POPDENS 0.729 0.060 1.313 - LOCOT 0.546 0.678 0.492 - YRINOP 9.814 1.373 3.883 - RENTER 0.998 0.900 0.632 - Share 0.1839 0.3103 0.2644 0.2414 -Not applicable. Note: See note to table 5-5 for definitions of variables and note to table 5-4 for the defini- tion of the elasticity of probability. Source: City Study establishment survey. Table 5-7. Logit Estimation of Firms' Location Choice (Dependent Variable: Industry and Employment Size) Coefficzents t statistzcs' Independent variable Group I Group 2 Group 3 Group 4 Group I Group 2 Group 3 Group 4 CONSTANT -26.690 -2.669 -14.020 2.87* 0.57 2.23*- PRODSOIL) 0.009 0.003 0.015 - 0.52 0.19 0.96 - INPUl'Bl' 0.016 0.003 0.027 - 1.04 0.20 1.94* - DIS1'CBD 0.047 -0.051 0.080 - 0.24 0.27 0.43 - WKSOtJTH 0.017 -0.001 0.015 - 0.88 0.09 0.87 - ADMNORTH -0.011 -0.011 -0.015 - 0.62 0.65 0.87 ELECINT 0.590 0.781 0.455 - 0.96 1.40 0.78 - POPDENS 0 -0.002 0.011 - 1.45 0.27 1.44 _ 1.ocolI 0.110.6901 0.653 _ 1.60 16 YRINOP 0.320J 0.048 0.144 - 2.59 0.69 1.68 1 RENTER 1.195 - 1.195 - 1.59 - 1.59 Number of observations 18 25 33 11 Percentagc correctly predicted 55.17 Likelihood ratio index 0.2878 Likelihood ratio statistic 69.42 -Not applicable. Note: Industry categories are defined in table 5-1; number of employees is the second variable used for grouping: groups l and 3 have less than 50 employees. Independent variables are the same as in table 5-2. Group 4 is used as the base. a. The coefficients with a single asterisk are significant at the 5 percent level; those with a double asterisk are significant at the 2.5 percent level. Source: City Study establishment survey. 104 A Model of Manufacturing Location Table 5-8. Elasticities of Probability: Logit Estimation of Location Choice (Dependent Variable: Industry and Employment Size) Independent variable Group I Group 2 Group 3 Group 4 PRODSOLD 0.422 0.101 0.611 - INPUTBT 0.853 0.102 1.285 - DISTCBD 0.201 -0.243 0.287 - WKSOUTH 0.983 -0.043 0.661 - ADMNORTH -0.497 -0.398 -0.492 - ELECINT 0.780 1.024 0.454 - POPDENS 1.172 -0.113 0.935 - LOCOT 0.562 0.641 0.511 - YRINOP 18.527 2.323 6.308 - RENTER 0.948 - 0.742 - Share 0.2069 0.2874 0.3793 0.1264 -Not applicable. Note: See note to table 5-7 for definitions of variables and note to table 5-4 for the defini- tion of the elasticity of probability. Source: City Study establishment survey. A Model of Employm en t Location for Retail Trades and Services This chapter presents a simple model of employment location in retail trades and services and reports empirical findings based on the 1978 household survey for BogotA. Since micro-establishment data were not available for modeling the location behavior of individual retail and ser- vice firms, an alternative approach based on social physics was used to study aggregate location patterns. A gravity model (explained below) was used to summarize statistical regularities observed for the location pat- terns of trade and services employment in Bogota. Although little progress has been made in modeling export-oriented industries such as manufacturing, attempts to model the location of em- ployment in retail trades have been relatively successful. In Lowry's pioneering work (I1964, pp. 2-3), for instance, production activities were divided into two groups: (a) The basic sector includes industrial, business, and administrative establishments whose clients are predominantly out- side the local area. These "export" industries are relatively unconstrained by problems of access to local markets, and their employment levels de- pend primarily on events outside the local economy. (b) The retail sector includes business, administrative, and other establishments (most schools and local government agencies, as well as retail trades and serv- ices) that deal primarily with the local population. Access to local residents is therefore a powerful factor in their choice of site, and their employment levels are closely related to local population growth. In the Lowry model, employment locations in the basic sector are assumed to be predetermined, while those in the retail sector are deter- mined endogenously. Only recently have efforts been made to model the locations of employment in the basic sector. (See chapter 5 for references on modeling manufacturing employment locations.) Because of a lack of data, empirical work on the retail sector has bypassed the modeling of individual firms' location decisions and instead used a "gravity" measure to represent the potential of a location as a site 105 106 A Model for Retail Trades and Services for a particular activity. This measure is "a simple analogue to Newton's law of gravity: The level of interaction is directly proportional to the mass of the interacting bodies (size of the groups of actors) and inversely pro- portional to the distance between them" (Lowry 1964, p. 21).' Even though this approach does not address directly the optimizing behavior of individual decisionmakers, the statistical regularities that emerge should represent the aggregate outcome. In fact, the implicit behavioral assumptions of the gravity model used in this chapter can be interpreted as an analogue to the multinomial logit framework developed for manufacturing firms in chapter 5. The model thus assumes that the retail firm locates where it maximizes profits; that is, its choice of a location with certain attributes is an integral part of achieving an optimal combination of inputs. In developing a gravity type model of employment location, Lakshmanan (1964, p. 3) states: The factors that influence the locations of various activities in the urban area are assumed to be "site" or area attributes. Thus the model framework assumes that each subarea in the metropolis attracts the various activities in relation to its relevant locational attributes. These attributes will vary with the activity group considered but will include both "policy" attributes such as available land for that activity and "status" attributes such as current population or employment levels and land value. Models of Market Potential and Residential Access In accordance with the work by Lowry (1964), Lakshmanan (1964), and W. G. Hansen (1959), it is postulated that the most important determi- nant of location for retail and service firms is the market potential-that is, the strength of the market-in any given subarea. The market potential is defined by a gravity measure, a weighted index of the number of actors (customers) in the zone, as follows: ( 6 - 1 ) Gi l ri)g + d g j i , j = 1 ... N where G = gravity measure for a subarea; Z = a proxy for market poten- tial such as the number of households, population, and household in- come; r = radius of a subarea; d = the distance between two subareas; N = the total number of subareas; and g = distance weight. The employment location model, then, is (6-2) Ei = J(Gi, ET) Market Potential 107 where Et = the number of the ith subarea's jobs in the kth activity; G, = a measure of the ith subarea's market potential; and El' = the total number of jobs in all activities in the ith subarea. The second term in equation 6-2 is included since shopping trips can originate from workplace as well as homes, but the model assumes that only short (walking distance) trips originate from workplaces; only the total number of jobs in that subarea (ET) instead of a gravity measure is therefore relevant. In contrast, since shopping trips originating from homes can be longer trips by car, bus, or metro, the probability of mak- ing a shopping trip from the ith to thejth subarea will diminish with dis- tance (dj). The measure of market potential (G) thus captures both ac- cessibility (trip distance, dij) and competition (level of activity, Z,,) as determinants of the probability that a particular customer will shop in a particular subarea. The model depicted in equation 6-2 thus represents the hypothesis that the retail or service firms in the ith subarea can attract customers from all subareas, but the extent of attraction diminishes with distance. This hypothesis also implies that retail firms of a particular type will locate in a subarea with particular attributes. Empirical Findings The model specified in equation 6-2 was fitted with DANE'S 1978 house- hold survey data for the twelve two-digit sic groups of the three major nonmanufacturing sectors in Bogota (sic numbers in parentheses). Commerce: wholesale (61), retail (62), restaurants and hotels (63) Finance: financial establishments (81), insurance (82), real estate and business services (83) Services: government services (91), sanitary services (92), social and other community services (93), recreation and entertainment (94), personal and household services (95), and international and other foreign organizations (96) Strictly speaking, most wholesalers, some financial establishments such as large banks, some insurance companies, and national government agencies should be treated as part of the basic (export) sector. The data used here, however, do not justify further disaggregation of industries. Therefore, although the model was specified primarily for retail trades and services, it also tested the extent of spatial variations in employment for this wide range of nonretail and nonservice industries with respect to the market potential of subareas. Table 6-1. Log-Linear Regression Estimates of the Residential Access Model Number of Industry group (sic code) Constant Coefficient t statistic 2 comunasa Wholesale (61) -6.47 0.79 1.19 0.0379 18 Retail (62) -9.56 1.47 7.76 0.6259 38 Restaurants and hotels (63) -12.37 1.51 3.08 0.2081 33 Finance (81) -14.63 1.55 2.21 0.1192 18 Real estate and business services (83) -15.94 1.77 2.72 0.1709 27 Government services (91) -10.27 1.25 1.82 0.0841 26 Sanitary service (92) -1.15 0.22 0.41 0.0047 11 a) Social and community services (93) -8.04 1.25 5.50 0.4562 38 Recreation and entertainment (94) 6.71 0.87 1.38 0.0504 24 Personal and household services (95) -4.99 1.03 5.35 0.4431 38 Commerce (6) -10.00 1.53 8.09 0.6452 38 Finance (8) -18.24 2.03 3.23 0.2248 30 Services (9) -5.83 1.18 5.67 0.47 15 38 Note: Weight (g) = 2; independent variable = household gravity. a. The number of comunas that had nonzero values of employment. Only four comunas had employment in insurance business (sic 82), and only three comunas had employment in international organizations (sic 96); the results for these two groups are not reported here. Source: DANE household survey, 1978. Empirical Findings 109 Bogota's thirty-eight comunas are used as subareas. Since the comunas are not the same size, the employment variable enters the regression as a density measure. The number of comunas with nonzero values of em- ployment for each industry is shown in table 6-1. All thirty-eight comunas were included in all regressions, however, even though some of them had no jobs in some industries. The estimated regression equation takes the following form: (6-3) (EVA), =f [G , (ET/A),J i =1, . .. 38 where A = area. The regression includes three proxies for market potential: the number of households, population, and household income. The first two repre- sent the market potential in terms of number of customers; the income variable represents the "purchasing power potential," which does not necessarily vary with the number of households or population subareas. In all cases, the log-linear specification had a better fit than the linear version. The log-linear regression results reported in tables 6-1 through 6-3, however, exclude the total employment variable (ET), which was highly collinear with the market potential variable; regressions with both variables thus produced statistically insignificant coefficients, often with incorrect signs. As expected, table 6-1 shows that the model fits best for retail trades, followed by social and community services, and personal and household services. It is evident that accessibility to residences is not important to the location decisions of wholesalers, entertainment businesses, and government and sanitary services. Compared with the commerce and the service sectors, the finance sector has a poor fit; this may reflect the fact that large financial establishments are located in or near the CBD. Nevertheless, the elasticity for the finance sector has the highest value, which indicates that the location choices of financial establishments are much more sensitive to changes in accessibility to residences than are the decisions of commercial or service establishments. The regressions that used population as a measure of market potential gave results (presented in table 6-2) that are similar to those derived with the use of households as a proxy. In fact, the above interpretation of the data in table 6-1 is equally appropriate to these results although they are statistically less robust. Using a measure of income gravity as an explanatory variable signifi- cantly improves the fit for the finance sector, especially for real estate and business services. The coefficients in table 6-3 also indicate that the per- sonal and household services and social services show a much better fit than do retail trades, which suggests that certain specialized service es- Table 6-2. Log-Linear Regression Estimates of the Population Access Model Number of Industry group (SIC code) Constant Coefficient t statistic R2 comunasa Wholesale (61) -5.56 0.62 0.85 0.0197 18 Retail (62) -12.43 1.52 6.84 0.5653 38 Restaurants and hotels (63) -14.58 1.50 2.75 0.1740 33 Finance (81) -13.43 1.26 1.61 0.0670 18 Real estate and business services (83) -17.28 1.66 2.29 0.1270 27 Government services (91) -8.99 0.99 1.30 0.0451 26 Sanitary service (92) -2.00 0.26 0.45 0.0057 11 Social and community services (93) -10.07 1.26 4.80 0.3907 38 Recreation and entcrtainment (94) -4.40 0.58 0.83 0.0188 24 Personal and household services (95) -6.92 1.06 4.85 0.3948 38 Commerce (6) -12.77 1.56 6.91 0.5701 38 Finance (8) -19.20 1.85 2.61 0.1593 30 Services (9) -7.29 1.15 4.70 0.3804 38 Note: Weight (g) = 2; independent variable = population gravity. a. The number of comunas that had nonzero values of employment. only four comunas had employment in insurance business (sIC 82), and only three comunas had employment in international organizations (sic 96); the results for these two groups are not reported here. Source: DANE household survey, 1978. T'able 6-3. Log-Linear Regression Estimates of Purchasing Power Potential Model Number of Industry group (sic code) Constant Coefficient t statistic R 2 comunas' Wholesale (61) -31.61 1.64 3.02 0.2024 18 Retail (62) -19.75 1.28 7.13 0.5853 38 Restaurants and hotels (63) -26.33 1.49 3.45 0.2480 33 Finance (81) -40.97 2.11 3.67 0.2725 18 Real estate and business services (83) -48.59 2.53 5.21 0.4301 27 Government services (91) -19.21 1.10 1.78 0.0806 26 Sanitary service (92) -6.90 0.40 0.84 0.0192 11 Social and community services (93) -21.80 1.34 8.02 0.6409 38 Recreation and entertainment (94) -30.37 1.61 3.14 0.2149 24 Personal and household services (95) -19.36 1.25 12.00 0.8000 38 Commerce (6) -21.14 1.36 7.78 0.6272 38 Finance (8) -47.63 2.51 5.16 0.4250 30 Services (9) -20.96 1.36 11.34 0.7814 38 Note: Weight (g) = 2; independent variable = income gravity. a. The number of comunas that had nonzero values of employment. Only four comunas had employment in insurance business (sic 82), and only three comunas had employment in international organizations (sic 96); the results for these two groups are not reported here. Source: DANE household survey, 1978. Table 6-4. Changes in Goodness-of-Fit (R2) with Increasing Gravity Weight (g): Residential Access Model (R2 values from log-linear regressions) Industry group (Sic code) g = 1.0 g = 1.5 g - 2.0 g = 2.5 g - 3.0 Wholesale (61) 0.0484 0.0401 0.0379 0.0393 0.0432 Retail (62) 0.6220 0.6278 0.6259 0.6197 0.6123 Rcstaurants and hotels (63) 0.1825 0.1947 0.2081 0.2154 0.2162 Finance (81) 0.1203 0.1199 0.1192 0.1176 0.1166 Real estate and business services (83) 0.1588 0.1655 0.1709 0.1733 0.1742 Government services (91) 0.1070 0.0961 0.0841 0.0754 0.0703 Sanitary service (92) 0.0166 0.0081 0.0047 0.0041 0.0046 Social and community services (93) 0.3999 0.4252 0.4562 0.4830 0.5042 Recreation and entertainment (94) 0.0384 0.0451 0.0504 0.0526 0.0525 Personal and household services (95) 0.3767 0.4099 0.4431 0.4652 0.4773 Commerce (6) 0.6232 0.6370 0.6452 0.6461 0.6430 Finance (8) 0.2052 0.2192 0.2248 0.2226 0.2184 Services (9) 0.4279 0.4485 0.4715 0.4885 0.5001 SouTce: DANE household survey, 1978. Empirical Findings 113 tablishments tend to locate in high-income areas. Moreover, regression results for wholesalers also improved markedly with the income variable: the coefficient is statistically significant, while the goodness-of-fit is mod- erate. This finding may reflect wholesalers' tendency to locate in Bogota's outer areas near high-income neighborhoods. Similar improvements oc- curred in the results for the recreation and entertainment group, busi- nesses that also tend to be oriented toward high-income households in Bogoti. The regression results in tables 6-1 to 6-3 were obtained with the dis- tance weight (g) of 2 in equation 6-1. The larger the value of g, the more weight the distance factor (d,,) has in the gravity measure; that is, a larger value of g implies a higher travel cost for a given distance between two subareas. It is possible that a higher value of g would alter the goodness- of-fit for the regression model. If the fit (correlation) improved, it would mean that potential customers who are some distance from the ith sub- area are rather unimportant to business establishments in that zone. This type of business establishment would thus exhibit a tendency toward local clustering. If the fit worsens as g increases, however, the implication is that such firms tend to be uniformly distributed over all subareas. The goodness-of-fit measure for the residential access model is re- ported in table 6-4 for five different values of g. In five of the ten industry groups-recreation and entertainment, personal services, social services, restaurants and hotels, and real estate and business services-the value of R2 increases as g rises. The value decreases, however, for financial in- stitutions, government services, and sanitary services. Table 6-5 shows comparable results obtained from the population access model. As discussed above, the amount of positive association between g and R2 should indicate the extent of local clustering of an industry group. When industries are listed in descending order by the percentage change in R2, the rankings in tables 6-4 and 6-5 are almost identical: (1) recreation and entertainment, (2) social and community services, (3) personal and household services, (4) restaurants and hotels, (5) real estate and business services, (6) retail trades, (7) finance, (8) government services, and (9) sanitary services. (The last three categories had a negative change in R2.) In the purchasing power potential model, the value of R2 decreases as g rises for the wholesale and the recreation and entertainment groups as well (table 6-6). On the whole, nonmanufacturing establishments in Bogota exhibit some local clustering with respect to market potential, with the exception of firms in government services, sanitary services, finance, and wholesale trades. Most of these establishments are likely to be export-oriented, with residential access unimportant to their location choices. Table 6-5. Changes in Goodness-of-Fit (R2) with Increasing Gravity Weight (g): Population Access Model (R2 values from log-linear regressions) Industry group (sic code) g = 1.0 g = 1.5 g = 2.0 g = 2.5 g = 3.0 Wholesale (61) 0.0274 0.0198 0.0197 0.0236 0.0285 Retail (62) 0.5468 0.5542 0.5653 0.5714 0.5724 Restaurants and hotels (63) 0.1478 0.1569 0.1740 0.1867 0.1918 Finance (81) 0.0642 0.0618 0.0670 0.0739 0.0800 Real estate and business services (83) 0.1097 0.1147 0.1270 0.1389 0.1475 Government services (91) 0.0648 0.0523 0.0451 0.0424 0.0422 Sanitary service (92) 0.0169 0.0087 0.0057 0.0053 0.0059 Social and community services (93) 0.3162 0.3443 0.3907 0.4338 0.4670 Recreation and entertainment (94) 0.0118 0.0138 0.0188 0.0238 0.0272 Personal and household services (95) 0.3062 0.3452 0.3948 0.4310 0.4514 Commerce (6) 0.5362 0.5489 0.5701 0.5852 0.5923 Finance (8) 0.1344 0.1449 0.1593 0.1695 0.1755 Services (9) 0.3241 0.3432 0.3804 0.4145 0.4392 Source: DANE household survey, 1978. Table 6-6. Changes in Goodness-of-Fit (R2) with Increasing Gravity Weight (g): Purchasing Power Potential Model (R' values from log-linear regressions) Industry group (sic code) g = 1.0 g = 1.5 g - 2.0 g = 2.5 g = 3.0 Wholesale (61) 0.2129 0.2138 0.2024 0.1871 0.1733 Retail (62) 0.5477 0.5671 0.5853 0.5934 0.5950 Restaurants and hotels (63) 0.2400 0.2439 0.2480 0.2500 0.2489 Finance (81) 0.2756 0.2803 0.2725 0.2585 0.2441 Real estate and business services (83) 0.3987 0.4279 0.4301 0.4141 0.3922 Government services (91) 0.0944 0.0865 0.0806 0.0773 0.0756 Sanitary service (92) 0.0273 0.0216 0.0192 0.0193 0.0201 Social and community services (93) 0.6021 0.6250 0.6409 0.6442 0.6430 Recreation and entertainment (94) 0.2325 0.2360 0.2149 0.1864 0.1617 Personal and household services (95) 0.7299 0.7790 0.8000 0.7908 0.7666 Commerce (6) 0.5798 0.6041 0.6272 0.6386 0.6419 Finance (8) 0.3824 0.4141 0.4250 0.4174 0.4013 Services (9) 0.7492 0.7764 0.7814 0.7663 0.7450 Source: DANE household survey, 1978. 116 A Modelfor Retail Trades and Services Summary To test the market potential model of employment location, a gravity measure for ten industry groups was used with DANE'S 1978 household survey data. The number of households, population, and household in- come were used as proxies for market potential. As expected, the model achieved the best fit for retail trades, social and community services, and personal and household services. Market poten- tial is not important in the location decisions of wholesalers, government services, and sanitary services. In the case of the purchasing power poten- tial model, the fit improved markedly for the finance sector, which in- dicates the importance of its being accessible to high-income households. The purchasing power potential model also performed better for per- sonal and household services, which tend to be highly income-oriented in Bogota, than for retail trades. The sensitivity analysis performed with different values of the distance weight (g) indicated that the R' increased with a higher value ofg for all in- dustry groups except for financial institutions, government services, sanitary services, and wholesalers. This result implies that BogotA's non- manufacturing firms tend to exhibit some local clustering, while the loca- tion choices of establishments in the above four groups attached little im- portance to market potential. Note 1. Among the early applications of this approach are W. G. Hansen (1 959), W. B. Hansen (1961), Pendleton (1963), Huff (1961), and Lakshmanan (1964). 7 The Spatial Structure of Production The underlying assumption of the employment location model described in chapter 5 is that land prices and possibly other input prices vary among subareas and that these variations affect a firm's choice of site. This chap- ter presents empirical evidence on both the land price and wage gradients in Bogota and tests the firms' responsiveness to the substitutability of land for other inputs in attaining optimal locations. Estimating the Elasticity of Capital-Land Substitution Although there is a rich body of theoretical and empirical literature on the elasticity of substitution between capital and labor in production, analysts have only recently begun to study the elasticity of capital-land substitution. Of the dozen or so analyses, most deal with urban housing, although Fallis (1975) has written about manufacturing and commercial establishments and Clapp (1979) about office buildings.1 The majority of these studies are based on the constant-elasticity-of-substitution (CES) pro- duction function with the standard assumptions of profit maximization, perfectly competitive markets, and constant returns to scale. (See Arrow and others 1961 for the original work on the CES production function.) Two studies, however, employ a variable-elasticity-of-substitution pro- duction function, and one a translog cost function. The elasticity of capital-land substitution can be defined as (7-1) 1= dln(K/L)/dln(R/r) where K = the stock of capital, L = the stock of land, R = the value of land, and r = the price of capital. The basic CES expression for estimating the elasticity can be derived from equation 7-1 as (7-2) ln(rK/L) = c + cylnR + (1 - cy)lnr 117 118 Spatial Structure of Production or (7-3) ln(rK/RL) = c + (1 - o)lnr - (I - (Y)lnR These most commonly used specifications imply that the intensity of land use is determined by the relative prices of land and other inputs. The elasticity of substitution of land for other inputs is an important parameter for explaining patterns of urban land use and for predicting probable changes in spatial structure. Relative factor prices, particularly with respect to land prices, will cause capital-land ratios to differ among locations (as exemplified by high rises near the city center) and thus in- fluence population and employment densities (Fallis 1979). The trend toward employment decentralization is therefore sensitive to the value of this parameter (Muth 1971). Moreover, since the substitution parameter determines not only the factor shares of lot size and housing space but also the elasticity of housing supply, it has important implications for urban housing policies (Muth 1969; Kau and Lee 1976). In his comparison of empirical studies, McDonald (1981, pp. 200-01) notes that estimates of the elasticity of capital-land substitution range from 0.36 to 1.13; nine out of twelve estimates are significantly less than 1; none was significantly greater than 1; and five fall in the interval between 0.4 and 0.6. The results based on data from one metropolitan area vary more widely (seven values ranging from 0.38 to 1.13) than those based on data for a cross-section of metropolitan areas (five values ranging from 0.36 to 0.55). It is possible that the elasticity of substitution may vary among metropolitan areas: by excluding three studies of Chicago, for ex- ample, the range of estimates is only 0.36 to 0.83. Chicago may thus have a relatively large elasticity value. Like estimates of the elasticity of capital-labor substitution, which show wide variation in the literature, the range of the elasticity estimates of capital-land substitution needs further investigation.2 Because of the na- ture of the data, as well as the particular theoretical assumptions, model specifications, or estimation technique used (or some combination of all these factors), biases are likely (see Reedy 1985). Sirmans, Kau, and Lee (1979), for instance, obtained alternative estimates of 0.664 to 0.925 with a variable elasticity of substitution (VEs) specification, compared with an elasticity of 0.766 using equation 7-2. Estimating the Elasticity of Substitution between Land and Other Inputs Of the 126 firms included in the City Study establishment survey, 88 owned land and structures while 38 were renters. Estimates of the elas- Substitution between Land and Other Inputs 1 19 ticity of land for capital using equation 7-2 are based on the 84 owners that reported land values. The dependent variable in the regression is the market value of the plant and equipment divided by the land area oc- cupied by the plant. Since the portion of land not used for production purposes varied widely among sample firms, using total land area owned as the denominator could have introduced a bias to the measure of land use intensity. The land price alone enters as the independent variable. Since the price of capital (plant and equipment) can be assumed invariant with respect to location (Muth 1969, pp. 52-53), the capital price term in equation 7-2 was therefore dropped in the regression. The estimated equation is thus: (7-4) ln(rK/L) = 0.2623 + 0.3058 InLR R = 0.0835 (2.73) The estimate of the elasticity of substitution is 0.3058 and statistically significant; it therefore provides solid evidence that the intensity of land use in Bogota indeed depends on land value. Compared with estimates obtained in earlier empirical studies, this elasticity estimate for manufac- turing establishments is smaller than the lowest value (0.36) obtained for urban housing. This result supports the assumption that the elasticity of capital-land substitution for manufacturing is likely to be smaller than that of housing since the possibilities for locating manufacturing activity in multistoried structures are limited. Moreover, the estimate is unlikely to contain the same downward biases that occurred in the housing studies because of errors in measuring land values (McDonald 1981). The land values in this study were provided by the managers of the sample es- tablishments rather than estimated. The measure of the dependent vari- able is also actual values of plant and equipment given by the respon- dents, rather than a residual measure, which is often used in housing studies for the value of structures.3 In his study of metropolitan Toronto, Fallis (1975) obtained the only previous estimate of the elasticity of capital-land substitution for manu- facturing (a value of 0.69). His more recent work (1979) analyzes the relationship between employment density and the value of the substitu- tion elasticity. Using the same specification as that in equation 7-2, the elasticity of substitution of land for labor has been estimated as follows (the wage term is suppressed): (7-5) ln(wN/L) = 2.9114 + 0.3037 lnR R2 = 0.1085 (3.16) 120 Spatial Structure of Production where w = wage rate, and N = the number of production workers. The actual data were available for the number of skilled and unskilled workers and the wages paid for both groups; wN was calculated as the weighted av- erage of the two groups. The elasticity of labor-land substitution is statis- tically significant, and its value is about the same as that of capital-land substitution. As land price increases, the intensity of land use rises in terms of both capital and labor. Empirical Evidence on Land Price Gradient The land values of eighty-four sample establishments are fitted to a nega- tive exponential function as follows: (7-6) lnR = 8.029 -0.1126D R2 = 0.1093 (3.17) where D = the air distance from the CBD in kilometers. Equation 7-6 can be written as (7-7) R = 3,069e-126D indicating that the land price for industrial use is 3,069 pesos per square meter at the city center and declines by 11.26 percent at a distance of one kilometer.4 With a strong t statistic, the slope coefficient is statistically significant. Based on data provided by the real estate firm of Wiesner and Cia. Ltd., table 7-1 presents the estimated values of the land price gradient in Bogota.s The slope estimate of the most recent years (based on the weighted unit price) closely resembles the estimate based on the es- tablishment survey data reported in equation 7-6 above. Although the es- timates in table 7-1 fluctuate slightly during the last five-year period, they are smaller than the equation 7-6 estimate; this probably reflects the fact that the gradient based on all types of lots, including commercial and residential, is flatter than that based on industrial sites only. In chapter 5 we postulated that land prices depend on site attributes such as the quality of public services, local amenities, and accessibility to markets. If site attributes vary directly with distance from the city center, it is possible that such variations explain differences in land prices. The gradient, in terms of distance, may therefore not be as statistically signifi- cant when site attributes are included in the regression. Although such a test has not been performed here, it is unlikely that all site attributes will vary with the distance from the city center. Amenities and road access, for example, tend to improve outside the central area in Bogota, whereas the opposite is true for the quality of public services (table 4-7). Substitution between Land and Other Inputs 121 Table 7-1. Parameter Estimates of Negative Exponential Land Price Functions for Bogota, 1955-78 Based on weighted averagesa Based on unweighted averagesb Year Constant Gradient R2 Constant Gradzent R2 1955 4.89 -0.234 0.464 5.07 -0.226 0.509 1956 4.79 -0.187 0.418 4.81 -0.163 0.369 1957 4.84 -0.197 0.477 4.99 -0.195 0.496 1958 4.96 -0.227 0.449 4.99 -0.207 0.412 1959 4.99 -0.194 0.573 5.06 -0.180 0.620 1960 4.74 -0.135 0.436 4.92 -0.121 0.479 1961 4.99 -0.204 0.564 4.98 -0.154 0.531 1962 4.91 -0.159 0.502 4.96 -0.146 0.456 1963 4.48 -0.086 0.129 4.65 -0.085 0.252 1964 5.07 -0.162 0.505 5.11 -0.135 0.508 1965 4.87 -0.178 0.514 5.06 -0.162 0.572 1966 4.80 -0.178 0.378 4.87 -0.160 0.432 1967 5.00 -0.147 0.501 4.98 -0.125 0.507 1968 5.12 -0.185 0.576 5.09 -0.156 0.568 1969 4.52 -0.087 0.142 4.76 -0.096 0.332 1970 4.63 -0.109 0.303 4.56 -0.075 0.383 1971 4.93 -0.154 0.501 4.87 -0.118 0.398 1972 4.60 -0.088 0.294 4.48 -0.057 0.252 1973 4.98 -0.133 0.386 4.81 -0.082 0.396 1974 4.77 -0.092 0.388 4.63 -0.057 0.250 1975 4.58 -0.082 0.388 4.38 -0.043 0.226 1976 4.80 -0.108 0.354 4.43 -0.047 0.166 1977 4.77 -0.110 0.335 4.71 -0.075 0.336 1978 4.74 -0.072 0.224 4.50 -0.028 0.090 Note: All coefficients are statistically significant at the 99 percent level of confidence. a. Comuna unit prices are calculated by weighting lot prices by their relative share in the total area of the comuna. b. Comuna unit prices are calculated by taking the simple arithmetic average of lot prices in each comuna. Source: Wagner (1984). In his study of rent gradients for manufacturing firms in Cincinnati, Schmenner (1981) included thirteen variables for site and structure characteristics in the regression and found that the distance variable was statistically not significant. Schmenner's empirical results are too weak, however, to draw a definite condusion that the rent gradient is flat for manufacturing. Indeed, not only is the coefficient of the distance variable statistically insignificant but so are most other variables. This result sug- 122 Spatial Structure of Production gests that the regression may suffer from severe multicollinearity OT that the sample contains little variation in land prices. Empirical Evidence on Wage Gradient Using a subsample of production workers in private establishments at fixed locations from the 1978 household survey, we tested for the pres- ence of a wage gradient for manufacturing jobs in Bogota and Cali. As table 7-2 indicates, for Bogota two of the three specifications (quadratic and double-log) result in statistically significant slope coefficients. In the case of Cali, the coefficient of the semi-log version is the most robust. Even though the estimates are statistically significant, however, their magnitudes seem small. The slope coefficient of the semi-log version for Bogota indicates that labor income decreases by less than 0.5 percent as distance from the center increases by one kilometer; the implication is that monthly income decreases by about 15 pesos per additional kilo- meter from the city center.6 Nevertheless, the results demonstrate that a negative wage gradient exists in both cities. To test the stability of the estimated wage gradient, additional re- gressions were performed including variables associated with characteris- tics of workers (their human capital) and of firms (such as number of em- ployees and presence of unionism). These variables are defined in table 7-3. The regression results for Bogota in table 7-4 indicate that the coeffi- Table 7-2. Wage Gradient for ManufacturingJobs: Bogota and Cali, 1978 Bogotd Cali Variable' Semi-log Quadratic Double-log Semi-log Quadratic Double-log DIST -0.0043 -0.0143 -0.0083 -0.0353 (1.30) (2.18) (1.95) (1.55) DISTSQ 0.0001 0.0003 (1.77) (1.21) LNDIST -0.0692 -0.0989 (2.46) (1.86) CONST 3.0037 3.0570 3.0881 2.9983 3.0507 3.0433 R2 0.0028 0.0079 0.0099 0.0147 0.0204 0.0134 Note: t statistics are in parentheses. a. See table 7-3 for the definitions of variables. Source: DANE household survey, 1978. Substitution between Land and Other Inputs 123 Table 7-3. Definition of Variables Dependent variable = Monthly labor income divided by hours worked DIST = Air distance from the CBD in kilometers DISTSQ = Square of DIST LNDIST = Log of DIST YRSED = The number of years of education EXP = The number of years of experience, defined as (age - years of education - 6) EXPSQ = Square of EXP SEX = 1 if male; 0 if female CONTR = 1 if employment contract exists; 0 otherwise LNSIZE = Log of the total number of workers at the establishment UNION = I if unionized; 0 otherwise CONST = Constant cient of the distance variable remains as statistically significant as in table 7-2. Worker attributes (education, experience, and sex), however, explain most of the variations in labor income. Although number of employees, unionism, and use of employment contracts were also statistically signifi- cant, they accounted for only about 20 percent of the explained vari- ations. The results for Cali in table 7-5 are similar to those for Bogoti except that including the additional variables reduced the level of signifi- cance of the distance variable and increased the R2 substantially. These tests do provide evidence on the existence of wage gradients for manufac- turing in the two cities. The presence of a negative wage gradient supports the hypothesis that manufacturingjobs in Bogota and Cali are sufficiently concentrated in the central area to attract workers to commute from outer areas. Studies on intra-urban wage gradients are rare in the literature. Eberts (1981) estimated wage gradients for five groups of municipal public em- ployees in the Chicago Standard Metropolitan Statistical Area. In four of the five cases, the coefficients of the distance variable were negative and statistically significant, and the magnitudes of the coefficients with a double-log specification varied from -0.09 to -0.38. Eberts found only a few previous studies on intra-urban wage differentials: Segal (1960) studied spatial wage patterns in New York City and found both posi- tive and negative gradients, depending on the industry; a study con- ducted by the Institute of Office Management (1962) for the London met- ropolitan area also showed that the wages of clerical workers tended to decrease with distance from the CBD. These studies, however, lack suffi- cient observations and the results are not based on rigorous statistical analysis. 124 Spatial Structure of Production Table 7-4. Wage Gradient for ManufacturingJobs in Bogota, 1978 Alternative regression specifications Variable' (1) (2) (3) (4) (5) DIST -0.0037 -0.0030 (1.31) (1.06) LNDIST -0.0588 -0.0492 (2.43) (2.02) YR.SED 0.0739 0.0752 0.0731 0.0747 0.0878 (8.45) (8.53) (8.39) (8.49) (10.07) EXP 0.0406 0.0411 0.0398 0.0405 0.0518 (7.16) (7.10) (7.01) (7.00) (9.29) EXPSQ -0.0006 -0.0006 -0.0005 -0.0005 -0.0007 (5.09) (5.05) (4.96) (4.95) (6.58) SEX 0.2283 0.2192 0.2292 0.2200 0.2263 (5.77) (5.45) (5.82) (5.50) (5.55) CONTR 0.0970 0.1383 0.0946 0.1378 (2.34) (3.46) (2.29) (3.46) LNSIZE 0.0523 0.0537 (4.76) (4.90) UNION 0.1543 0.1544 (3.38) (3.39) CONST 1.7084 1.8254 1.7881 1.8942 1.7330 R2 0.2826 0.2694 0.2875 0.2729 0.2311 Note: The data are for production workers at private establishments with a fixed location. There were 606 workers in this category out of a total of 1,186 manufacturing workers in the sample. t statistics are in parentheses. a. See table 7-3 for the definitions of variables. Source: DANE household survey, 1978. Summary The above estimates of the elasticity of substitution of land for capital and labor indicate that the intensity of land use in BogotA and Cali is deter- mined by land prices. Together with the well-shaped land price gradients, this fact indicates that the spatial structure of production activity in Bogota is monocentric. Moreover, the presence of a wage gradient for manufacturing firms supports the hypothesis that jobs in this sector are sufficiently concentrated in the central area to attract commuters. These findings further imply that manufacturing firms in Bogota do indeed re- spond to the relation of land prices to other input prices in making their location choices. Notes 125 Table 7-5. Wage Gradient for ManufacturingJobs in Cali, 1978 Alternative regression specifications Variable' (1) (2) (3) (4) (5) DIST -0.0047 -0.0046 (1.48) (1.42) LNDIST -0.0636 -0.0572 (1.59) (1.40) YRSED 0.0830 0.0827 0.0812 0.0810 0.1048 (6.11) (5.92) (5.96) (5.78) (7.94) EXP 0.0501 0.0536 0.0487 0.0524 0.0693 (5.38) (5.68) (5.20) (5.51) (7.81) EXPSQ -0.0006 -0.0007 -0.0006 -0.0007 -0.0009 (3.63) (3.87) (3.50) (3.75) (5.14) SEX 0.3558 0.3289 0.3606 0.3327 0.3363 (6.21) (5.56) (6.32) (5.64) (5.65) CONTR 0.1104 0.2109 0.1156 0.2176 (1.56) (3.26) (1.64) (3.36) LNSIZE 0.0594 0.0612 (3.59) (3.70) UNION 0.1063 0.1127 (1.64) (1.74) CONST 1.3970 1.5281 1.4391 1.5690 1.3665 R 2 0.4673 0.4456 0.4680 0.4455 0.4056 Note: The data are for production workers at private establishments with a fixed location. There were 256 workers in this category. t statistics are in parentheses. a. See table 7-3 for the definitions of variables. Source: DANE household survey, 1978. Notes 1. In a review article McDonald (1981) attributes the first detailed empirical study of the elasticity of capital-land substitution in urban housing to Muth (197 1), followed by Koenker (1972). He then notes that no research followed these two studies until the work by Rydell (1976), but that nine empirical studies have appeared since. 2. Morawetz (1976) reviews a large number of studies that vary widely in their estimates of the elasticity of capital-labor substitution. 3. Price of the house minus the value of land (pQ - RL) was often used as a proxy for rK in equation 7-2. See Koenker (1972), for example. 4. The average land price per square meter was 2,579 pesos in the survey. 5. For.detailed analyses of this data set, see Mohan and Villamizar (1982), Villamizar (1981), and Wagner (1984). The data file contained about 6,000 trans- actions of vacant lots for all uses during 1955-78. 6. The average hourlywage was about 20 pesos, or about 3,500 pesos per month. Appendix A Statistical Tables 126 Statistical Tables 127 Table Al. Employment Distribution by Size of Firm, Bogota and Cali, 1978 (percent) Fzrm size (number employed) Bogotd Cali 1-9 52.04 55.17 10-19 9.8] 8.52 20-34 8.22 6.05 35-39 2.34 1.68 50-74 4.19 3.76 75-99 2.10 1.76 100-199 4.58 3.50 200-499 5.86 5.08 500-999 3.48 2.97 1000 + 3.66 3.48 N.i.e. 3.72 8.03 Total 100.00 100.00 Total number employed 1,211,986 362,577 N.i.e. Not included elsewhere. Source: DANE household survey, 1978. Table A2. Distribution of Employment by Ring, Bogota, 1978 All Smalla Largeb Ring Persons Percent Persons Percent Persons Percent 1 169,034 13.95 73,390 11.64 89,996 16.78 2 214,953 17.74 94,088 14.92 110,880 20.68 3 198,733 16.40 88,097 13.97 101,407 18.91 4 249,624 20.60 136,795 21.69 102,762 19.16 5 302,234 24.94 199,955 31.70 94,464 17.62 6 41,530 3.43 30,126 4.78 10,983 2.05 N.i.e. 35,878 2.96 8,292 1.31 25,707 4.79 Total 1,211,986 100.00 630,742 100.00 536,198 100.00 N.i.e. Not included elsewhere. Note: Employment figures for small and large firms do not add to the total because some workers did not respond to the question about size of firm. a. Establishments with less than 10 employees. b. Establishments with 10 or more employees. Source: DANE household survey, 1978. 128 Appendix A Table A3. Distribution of Employment by Radial Sector, Bogota, 1978 All SmalP Largeb Radial sector Persons Percent Persons Percent Persons Percent 1 169,034 13.95 73,390 11.64 89,996 16.78 2 101,859 8.40 80,936 12.83 20,042 3.74 3 159,595 13.17 107,642 17.07 47,343 8.83 4 109,895 9.07 52,785 8.37 53,469 9.97 5 147,113 12.14 44,834 7.11 93,359 17.41 6 147,235 12.15 84,424 13.38 57,184 10.66 7 120,663 9.96 71,480 11.33 46,171 8.61 8 220,716 18.21 106,961 16.96 102,927 19.20 N.i.e. 35,878 2.96 8,292 1.31 25,707 4.79 Total 1,211,986 100.00 630,742 100.00 536,198 100.00 N.i.e. Not included elsewhere. Note: Employment figures for small and large firms do not add to the total because some workers did not respond to the question about size of firm. a. Establishments with less than 10 employees. b. Establishments with 10 or more employees. Source: DANE household survey, 1978. Table A4. Distribution of Employment by Ring, Cali, 1978 Total Smalla Largeb Ring Persons Percent Persons Percent Persons Percent 1 60,254 16.36 25,083 12.35 30,839 22.75 2 96,338 26.16 54,826 26.99 32,505 23.98 3 120,298 32.67 75,178 37.00 36,880 27.21 4 52,929 14.37 33,350 16.41 16,266 12.00 5 10,248 2.78 8,631 4.25 1,239 0.91 N.i.e.' 28,205 7.66 6,101 3.00 17,798 13.13 Total 368,273 100.00 203,170 100.00 -135,528 100.00 Note: Employment figures for small and large firms do not add to the total because some workers did not respond to the question about size of firm. a. Establishments with less than 10 employees. b. Establishments with 10 or more employees. c. Not included elsewhere. Reflects the number of workers who live in Cali but work in Yumbo, an industrial park outside the city limit. Source: DANE household survey, 1978. Statistical Tables 129 Table A5. Distribution of Employment by Radial Sector, Cali, 1978 Total SmalP La rge0 Radial sector Persons Percent Persons Percent Persons Percent 1 60,254 16.36 25,083 12.35 30,839 22.75 2 38,610 10.48 18,485 9.10 16,313 12.04 3 72,590 19.71 30,069 14.80 35,307 26.05 4 45,883 12.46 30,764 15.14 13,648 10.07 5 64,986 17.65 53,695 26.43 8,312 6.13 6 41,908 11.38 25,963 12.78 11,003 8.12 7 15,836 4.30 13,009 6.40 2,307 1.70 N.i.e. 28,205 7.66 6,101 3.00 17,798 13.13 Total 368,273 100.00 203,170 100.00 135,528 100.00 N.i.e. Not included elsewhere. Note: Employment figures for small and large firms do not add to the total because some workers did not respond to the question about size of firm. a. Establishments with less than 10 employees. b. Establishments with 10 or more employees. Source: DANE household survey, 1978. 130 Appendix A Table A6. Origin and Destination ofJobs by Ring, Bogota, 1973-78 Destination Origin (ring) 1 2 3 1 Number of jobs 8,570.9 15,370.2 3,151.6 Percentage of row total 27.76 49.78 10.21 Percentage of column total 65.70 55.19 23.09 2 Number of jobs 1,121.7 8,244.6 2,872.6 Percentage of row total 6.07 44.65 15.56 Percentage of column total 8.60 29.60 21.05 3 Number of jobs 2,529.6 2,890.7 6,117.2 Percentage of row total 13.96 15.95 33.75 Percentage of column total 19.39 10.38 44.83 4 Number of jobs 369.0 352.0 949.4 Percentage of row total 4.43 4.23 11.40 Percentage of column total 2.83 1.26 6.96 5 Number of jobs 171.0 318.0 201.2 Percentage of row total 3.24 6.03 3.82 Percentage of column total 1.31 1.14 1.47 6 Number of jobs 0.0 0.0 0.0 Percentage of row total 0.00 0.00 0.00 Percentage of column total 0.00 0.00 0.00 N.i.e. Number of jobs 282.7 676.6 354.6 Percentage of row total 9.38 22.44 11.76 Percentage of column total 2.17 2.43 2.60 Total Number of jobs 13,044.9 27,852.1 13,646.6 Percentage of row total 15.39 32.86 16.10 Percentage of column total 100.00 100.00 100.00 N.i.e. Not included elsewhere. Note: The table includes all firms that relocated during 1973-78. The values for number of jobs are estimates based on sample counts. Source: DANE household survey, 1978. Statistical Tables 131 (ring) 4 5 6 N.i.e. Total 1,957.0 1,540.6 0.0 284.3 30,874.6 6.34 4.99 0.00 0.92 100.00 12.83 12.86 0.00 15.54 36.43 3,385.4 2,653.0 0.0 188.1 18,465.4 18.33 14.37 0.00 1.02 100.00 22.20 22.15 0.00 10.28 21.79 4,619.6 1,683.6 0.0 282.7 18,123.4 25.49 9.29 0.00 1.56 100.00 30.29 14.06 0.00 15.46 21.38 3,487.4 2,293.4 327.1 548.9 8,327.2 41.88 27.54 3.93 6.59 100.00 22.87 19.15 28.35 30.01 9.83 667.0 3,184.0 359.5 369.5 5,270.2 12.66 60.42 6.82 7.01 100.00 4.37 26.58 31.16 20.20 6.22 0.0 212.1 467.0 0.0 679.1 0.00 31.23 68.77 0.00 100.00 0.00 1.77 40.48 0.00 0.80 1,135.5 420.4 0.0 155.5 3,015.3 37.66 13.61 0.00 5.16 100.00 7.44 3.43 0.00 8.50 3.56 15,251.9 11,977.1 1,153.6 1,829.0 84,755.2 18.00 14.13 1.36 2.16 100.00 100.00 100.00 100.00 100.00 100.00 132 Appendix A Table A7. Origin and Destination ofJobs by Radial Sector, Bogota, 1973-78 Origin (sector) 1 2 3 1 Number of jobs 8,570.9 0.0 195.4 Percentage of row total 27.76 0.00 0.63 Percentage of column total 65.70 0.00 3.28 2 Number of jobs 280.5 1,913.0 679.0 Percentage of row total 7.41 50.54 17.94 Percentage of column total 2.15 47.59 11.38 3 Number of jobs 0.0 1,412.0 2,306.8 Percentage of row total 0.00 28.46 46.50 Percentage of column total 0.00 35.13 38.67 4 Number of jobs 0.0 0.0 1,109.8 Percentage of row total 0.00 0.00 16.21 Percentage of column total 0.00 0.00 18.60 5 Number of jobs 475.7 0.0 1,297.3 Percentage of row total 4.59 0.00 12.51 Percentage of column total 3.65 0.00 21.75 6 Number of jobs 0.0 0.0 0.0 Percentage of row total 0.00 0.00 0.00 Percentage of column total 0.00 0.00 0.00 7 Number of jobs 198.0 0.0 0.0 Percentage of row total 3.93 0.00 0.00 Percentage of column total 1.52 0.00 0.00 8 Number of jobs 3,237.1 485.1 376.8 Percentage of row total 18.01 2.70 2.10 Percentage of column total 24.82 12.07 6.32 N.i.e. Number of jobs 282.7 209.8 0.0 Percentage of row total 9.38 6.96 0.00 Percentage of column total 2.17 5.22 0.00 Total Number of jobs 13,044.9 4,019.9 5,965.1 Percentage of row total 15.39 4.74 7.04 Percentage of column total 100.00 100.00 100.00 N.i.e. Not included elsewhere. Note: The table indudes all firms that relocated during 1973-78. Source: DANE household survey, 1978. Statistical Tables 133 Destination (sector) 4 5 6 7 8 Ni.e. Total 1,219.6 5,051.1 1,549.9 1,878.6 12,124.8 284.3 30,874.6 3.95 16.36 5.02 6.08 39.27 0.92 100.00 20.08 37.15 24.39 24.18 46.45 15.54 36.43 0.0 377.1 171.0 193.8 171.0 0.0 3,785.4 0.00 9.96 4.52 5.12 4.52 0.00 100.00 0.00 2.77 2.69 2.49 0.66 0.00 4.47 583.4 0.0 280.5 0.0 206.9 171.0 4,960.6 11.76 0.00 5.65 0.00 4.17 3.45 100.00 9.61 0.00 4.41 0.00 0.79 9.35 5.85 2,138.8 1,887.1 743.7 188.1 777.8 0.0 6,845.3 31.24 27.57 10.86 2.75 11.36 0.00 100.00 35.22 13.88 11.70 2.42 2.98 0.00 8.08 1,189.2 4,168.3 1,131.7 374.5 1,175.3 557.6 10,369.6 11.47 40.20 10.91 3.61 11.33 5.38 100.00 19.58 30.66 17.81 4.82 4.50 30.49 12.23 313.5 0.0 307.3 518.4 545.1 206.9 1,891.2 16.58 0.00 16.25 27.41 28.82 10.94 100.00 5.16 0.00 4.84 6.67 2.09 11.31 2.23 171.0 857.2 178.8 2,898.5 732.1 0.0 5,035.6 3.40 17.02 3.55 57.56 14.54 0.00 100.00 2.82 6.31 2.81 37.31 2.80 0.00 5.94 280.5 694.3 1,581.8 1,538.7 9,329.6 453.7 1 7,977.6 1.56 3.86 8.80 8.56 51.90 2.52 100.00 4.62 5.11 24.89 19.81 35.74 24.81 21.21 177.3 559.8 410.4 177.3 1,042.5 155.5 3,015.3 5.88 18.57 13.61 5.88 34.57 5.16 100.00 2.92 4.12 6.46 2.28 3.99 8.50 3.56 6,073.3 13,594.9 6,355.1 7,767.9 26,105.1 1,829.0 84,755.2 7.17 16.04 7.50 9.17 30.80 2.16 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Table A8. Origin and Destination ofJobs by Ring, Cali, 1973-78 Destination (nng) Onigin (ring) 1 2 5 4 5 N.i.e. Total I Number ofjobs 4,548.7 3,616.6 1,103.9 210.0 0.0 0.0 9,479.2 Percentage of row total 47.99 38.15 11.65 2.22 0.00 0.00 100.00 Percentage of column total 88.73 40.55 12.57 7.57 0.00 0.00 33.65 2 Numbcr ofjobs 367.5 3,028.6 1,816.5 0.0 0.0 812.5 6,025.1 Percentage of row total 6.10 50.27 30.15 0.00 0.00 13.49 100.00 Percentage of column total 7.17 33.96 20.68 0.00 0.00 34.18 21.39 3 Number of jobs 0.0 822.5 4,623.4 1,284.4 189.0 653.3 7,572.6 Percentage of row total 0.00 10.86 61.05 16.96 2.50 8.63 100.00 Percentage of column total 0.00 9.22 52.64 46.27 100.00 27.48 26.88 4 Number of jobs 0.0 443.3 819.0 609.0 0.0 0.0 1,871.3 Percentage of row total 0.00 23.69 43.77 32.54 0.00 0.00 100.00 Percentage of column total 0.00 4.97 9.33 21.94 0.00 0.00 6.64 5 Number of jobs 0.0 231.0 0.0 210.0 0.0 210.0 651.0 Percentage of row total 0.00 35.48 0.00 32.26 0.00 32.26 100.00 Percentage of column total 0.00 2.59 0.00 7.57 0.00 8.83 2.31 N.i.e. Number ofjobs 210.0 776.3 420.0 462.5 0.0 701.6 2,570.4 Percentage of row tota] 8.17 30.20 16.34 17.99 0.00 27.30 100.00 Percentage of column total 4.10 8.70 4.78 16.66 0.00 29.51 9.12 Total Number of jobs 5,126.2 8,918.3 8,782.8 2,775.9 189.0 2,277.4 28,169.6 Percentage of row total 18.20 31.66 31.18 9.85 0.67 8.44 100.00 Percentage of column total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. Note: The table includes all firms that relocated during 1973-78 Source: DANE household survey, 1978. 136 Appendix A Table A9. Origin and Destination ofJobs by Radial Sector, Cali, 1973-78 Origin (sector) 1 2 3 I Number of jobs 4,548.7 2,097.9 1,992.6 Percentage of row total 47.99 22.13 21.02 Percentage of column total 88.73 46.30 30.19 2 Number of jobs 0.0 654.5 210.0 Percentage of row total 0.00 43.24 13.87 Percentage of column total 0.00 14.44 3.18 3 Number of jobs 367.5 1,140.3 2,264.6 Percentage of row total 6.00 18.61 36.96 Percentage of column total 7.17 25.16 34.31 4 Number of jobs 0.0 0.0 595.0 Percentage of row total 0.00 0.00 25.40 Percentage of column total 0.00 0.00 9.01 5 Number of jobs 0.0 192.5 840.0 Percentage of row total 0.00 6.81 29.71 Percentage of column total 0.00 4.25 12.73 6 Number of jobs 0.0 210.0 0.0 Percentage of row total 0.00 6.83 0.00 Percentage of column total 0.00 4.63 0.00 7 Number of jobs 0.0 236.3 0.0 Percentage of row total 0.00 100.00 0.00 Percentage of column total 0.00 5.21 0.00 N.i.e. Number of jobs 210.0 0.0 698.8 Percentage of row total 8.17 0.00 27.19 Percentage of column total 4.10 0.00 10.59 Total Number of jobs 5,126.2 4,531.5 6,601.0 Percentage of row total 18.20 16.09 23.43 Percentage of column total 100.00 100.00 100.00 N.i.e. Not included elsewhere. Note: The table includes all firms that relocated during 1973-78. Source: DANE household survey, 1978. Statistical Tables 137 Destination (sector) 4 5 6 7 N.i.e. Total 210.0 420.0 0.0 210.0 0.0 9,479.2 2.22 4.43 0.00 2.22 0.00 100.00 8.18 14.05 0.00 47.05 0.00 33.65 0.0 0.0 439.1 0.0 210.0 1,513.6 0.00 0.00 29.01 0.00 13.87 100.00 0.00 0.00 12.44 0.00 8.83 5.37 420.0 256.7 210.0 236.3 1,232.5 6,127.9 6.85 4.19 3.43 3.86 20.11 100.00 16.35 8.59 5.95 52.84 51.84 21.75 1,072.9 441.0 0.0 0.0 233.3 2,342.2 45.81 18.83 0.00 0.00 9.96 100.00 41.77 14.76 0.00 0.00 9.81 8.31 445.7 1,138.7 210.0 0.0 0.0 2,826.9 15.77 40.28 7.43 0.00 0.00 100.00 17.35 38.10 5.95 0.00 0.00 10.04 0.0 462.4 2,400.7 0.0 0.0 3,073.1 0.00 15.05 78.12 0.00 0.00 100.00 0.00 15.47 68.01 0.00 0.00 10.91 0.0 0.0 0.0 0.0 0.0 236.3 0.00 0.00 0.00 0.00 0.00 100.00 0.00 0.00 0.00 0.00 0.00 0.84 420.0 270.0 270.0 0.0 701.6 2,570.4 16.34 10.50 10.50 0.00 27.30 100.00 16.35 9.03 7.65 0.00 29.51 9.12 2,568.6 2,988.8 3,529.8 446.3 2,377.4 28,169.6 9.12 10.61 12.53 1.58 8.44 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Table AIO. Distribution of NewJobs by Ring and Industry, Bogota, 1973-78 Ring Manufacturing Commerce Finance Services Other Total I Number ofjobs 2,610.9 6,472.3 2,799.8 5,283.8 726.8 17,893.6 Percentage of row total 14.59 36.17 15.65 29.53 4.06 100.00 Percentage of column total 7.55 14.09 29.20 18.61 6.38 13.78 2 Number ofjobs 4,661.2 9,329.2 4,127.5 3,990.3 1,008.1 23,116.3 Percentage of row total 20.16 40.36 17.86 17.26 4.36 100.00 Percentage of column total 13.47 20.31 43.04 14.05 8.86 17.80 3 Number ofjobs 6,913.8 3,473.4 1,329.5 4,934.5 1,815.3 18,736.5 Percentage of row total 36.90 19.98 7.10 26.34 9.69 100.00 Percentage of column total 19.98 8.15 13.86 17.38 15.95 14.42 4 Number ofjobs 8,531.6 10,564.0 904.3 4,983.4 2,128.6 27,111.9 Percentage of row total 31.47 38.96 3.34 18.38 7.85 100.00 Percentage of column total 24.66 23.00 9.43 17.55 18.70 20.87 5 Number ofjobs 10,592.7 14,683.8 428.5 7,605.0 2,634.4 35,944.4 Percentage of row total 29.47 40.85 1.19 21.16 7.33 100.00 Percentage of column total 30.62 31.97 4.47 26.78 23.14 27.67 6 Number of jobs 975.2 788.5 0.0 660.0 1,005.3 3,429.0 Percentage of row total 28.44 23.00 0.00 19.25 29.32 100.00 Percentage of column total 2.82 1.72 0.00 2.32 8.83 2.64 N.i.e. Number ofjobs 309.8 342.2 0.0 942.7 2,065.7 3,660.4 Percentage of row total 8.46 9.35 0.00 25.75 56.43 100.00 Percentage of column total 0.90 0.75 0.00 3.32 18.15 2.82 Total Number ofjobs 34,595.2 45,923.4 9,589.6 28,399.7 11,384.2 129,892.1 Percentage of row total 26.63 35.36 7.38 21.86 8.76 100.00 Percentage of column total 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. Note: The table includes all firms established during 1973-78. Source: DANE household survey, 1978. Table Al 1. Distribution of NewJobs by Radial Sector and Industry, Bogota, 1973-78 Radial sector Manufacturing Commerce Finance Seroices Other Total I Number of jobs 2,610.9 6,472.3 2,799.8 5,283.8 726.8 17,893.6 Percentageofrowtotal 14.59 36.17 15.65 29.53 4.06 100.00 Percentage of column total 7.55 14.09 29.20 18.61 6.38 13.78 2 Number ofjobs 2,400.5 1,663.7 0.0 2,342.3 967.3 7,373.8 Percentage of row total 32.55 22.56 0.00 31.77 13.12 100.00 Percentage of column total 6.94 3.62 0.00 8.25 8.50 5.68 3 Number ofjobs 8,541.2 5,404.9 341.2 3,188.4 1,561.0 19,036.7 Percentage of row total 44.87 28.39 1.79 16.75 8.20 100.00 Percentage of column total 24.69 11.77 3.56 11.23 13.71 14.66 4 Number of jobs 5,628.6 6,400.5 0.0 706.8 769.4 13,505.3 Percentage of row total 41.68 47.39 0.00 5.23 5.70 100.00 Percentage of column total 16.27 13.94 0.00 2.49 6.76 10.40 5 Number ofjobs 4,075.3 5,081.5 362.1 2,790.5 1,167.6 13,477.0 Percentage of row total 30.24 37.70 2.69 20.71 8.66 100.00 Percentage of column total 11.78 11.07 3.78 9.83 10.26 10.38 Table All (continued) Radial sector Manufacturing Commerce Finance Services Other Total 6 Number of jobs 4,596.2 8,534.3 397.6 3,720.3 342.0 17,590.4 Percentage of row total 26.13 48.52 2.26 21.15 1.94 100.00 Percentage of column total 13.29 18.58 4.15 13.10 3.00 13.54 7 Number ofjobs 4,060.4 3,947.3 2,723.8 2,393.4 1,503.2 14,628.1 Percentage of row total 27.76 26.98 18.62 16.36 10.28 100.00 Percentage of column total 11.74 8.60 28.40 8.43 13.20 11.26 8 Number of jobs 2,372.3 8,076.7 2,965.1 7,031.5 2,281.2 22,726.8 Percentage of row total 10.44 35.54 13.05 30.94 10.04 100.00 Percentage of column total 6.86 17.59 30.92 24.76 20.04 17.50 N.i.e. Number ofjobs 309.8 342.2 0.0 942.7 2,065.7 3,660.4 Percentage of row total 8.46 9.35 0.00 25.75 56.43 100.00 Percentage of column total 0.90 0.75 0.00 3.32 18.15 2.82 Total Number of jobs 34,595.2 45,923.4 9,589.4 28,399.7 11,384.2 129,892.1 Percentage of row total 26.63 35.36 7.38 21.86 8.76 100.00 Percentage of column total 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. Note: The table includes all firms established during 1973-78. Source: DANE household survey, 1978. Table A12. Distribution of NewJobs by Ring and Industry, Cali, 1973-78 Ring Manufacturing Commerce Finance Services Other Total I Number of jobs 848.8 1,447.7 0.0 2,443.0 0.0 4,739.5 Percentage of row total 17.91 30.55 0.00 51.55 0.00 100.00 Percentage of column total 9.27 13.61 0.00 26.27 0.00 14.70 2 Number ofjobs 1,755.8 2,558.5 630.0 2,871.8 446.3 8,262.4 Percentage of row total 21.25 30.97 7.62 34.76 5.40 100.00 Percentage of column total 19.18 24.05 60.00 30.88 21.15 25.62 3 Number of jobs 4,114.2 4,761.0 210.0 3,067.6 210.0 12,362.8 Percentageofrowtotal 33.28 38.51 1.70 24.81 1.70 100.00 Percentage of column total 44.95 44.76 20.00 32.99 9.95 38.33 - 4 Number ofjobs 2,036.0 1,218.0 0.0 480.0 588.0 4,322.0 Percentage of row total 47.11 28.18 0.00 11.11 13.60 100.00 Percentage of column total 22.24 11.45 0.00 5.16 27.86 13.40 5 Number of jobs 210.0 210.0 0.0 0.0 420.0 840.0 Percentage of row total 25.00 25.00 0.00 0.00 50.00 100.00 Percentage of column total 2.29 1.97 0.00 0.00 19.90 2.60 N.i.e. Number ofjobs 189.0 441.0 210.0 437.5 446.3 1,723.8 Percentage ofrow total 10.96 25.58 12.18 25.38 25.89 100.00 Percentage of column total 2.06 4.15 20.00 4.70 21.15 5.35 Total Number ofjobs 9,153.8 10,636.2 1,050.0 9,299.9 2,110.6 32,250.5 Percentage of row total 28.38 32.98 3.26 28.84 6.54 100.00 Percentage of column total 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. Note: The table includes all firms established during 1973-78. Source: DANE household survey, 1978. Table A13. Distribution of New jobs by Radial Sector and Industry, Cali, 1973-78 Radial sector Manufacturing Commerce Finance Services Other Total I Number of jobs 848.8 1,447.7 0.0 2,443.0 0.0 4,739.5 Percentage of row total 17.91 30.55 0.00 51.55 0.00 100.00 Percentage of column total 9.27 13.61 0.00 26.27 0.00 14.70 2 Number ofjobs 1,408.8 535.0 630.0 460.9 236.3 3,271.0 Percentage of row total 43.07 16.36 19.26 14.09 7.22 100.00 Percentage of column total 15.39 5.03 60.00 4.96 11.20 10.14 3 Number ofjobs 2,105.1 1,717.1 0.0 1,267.5 0.0 5,089.7 Percentage of row total 41.36 33.74 0.00 24.90 0.00 100.00 Percentage of column total 23.00 16.14 0.00 13.63 0.00 15.78 4 Number ofjobs 1,656.1 1,775.4 0.0 1,270.6 210.0 4,912.1 Percentage of row total 33.71 36.14 0.00 25.87 4.28 100.00 Percentage of column total 18.09 16.69 0.00 13.66 9.95 15.23 5 Number of jobs 2,502.7 2,716.0 0.0 2,049.2 420.0 7,687.9 Percentage of row total 32.55 35.33 0.00 26.65 5.46 100.00 Percentage of column total 27.34 25.54 0.00 22.03 19.90 23.84 6 Number of jobs 443.3 1,125.8 0.0 1,142.1 588.0 3,299.2 Percentage of row total 13.44 34.12 0.00 34.62 17.82 100.00 Percentage of colunmnl total 4.84 10.58 0.00 12.28 27.86 10.23 7 Number of jobs 0.0 878.2 210.0 229.1 210.0 1,527.3 Percentage of row total 0.00 57.50 13.75 15.00 13.75 100.00 Pcrcentage of column total 0.00 8.26 20.00 2.46 9.95 4.74 N.i.c. Number of jobs 189.0 441.0 210.0 437.5 446.3 1,723.8 Percentage of row total 10.96 25.58 12.18 25.38 25.89 100.00 Percentage of column total 2.06 4.15 20.00 4.70 21.15 5.35 Total Number ofjobs 9,153.8 10,636.2 1,050.0 9,299.9 2,110.6 32,250.5 Percentage of row total 28.38 32.98 3.26 28.84 6.54 100.00 Percentage of column total 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. Note: The table includes all firms established during 1973-78. Source: DANS; household survey, 1978. 144 Appendix A Table A14. Distribution of Employment by Ring, Bogota, 1972-78 1972 1978 Annual average Ring Persons Percent Persons Percent growth rate (percent) 1 203,936 23.03 169,034 13.95 -3.08 2 120,478 13.61 214,953 17.74 10.13 3 129,429 14.62 198,733 16.40 7.41 4 166,446 18.80 249,624 20.60 7.00 5 164,751 18.61 302,234 24.94 10.64 6 14,782 1.67 41,530 3.43 18.79 N.i.e. 85,657 9.67 35,878 2.96 Total 885,479 100.00 1,211,986 100.00 5.37 -Not applicable. N.i.e. Not included elsewhere. Sources: Phase 11 household survey, 1972; DANE household survey, 1978. Table A15. Distribution of Employment by Radial Sector, Bogota, 1972-78 1972 1978 Radial Annual average sector Persons Percent Persons Percent growth rate (percent) 1 203,936 23.03 169,034 13.95 -3.08 2 61,187 6.91 101,859 8.40 8.87 3 107,931 12.19 159,595 13.17 6.74 4 75,483 8.52 109,895 9.07 6.46 5 86,640 9.78 147,113 12.14 9.23 6 78,076 8.82 147,235 12.15 11.13 7 79,433 8.97 120,663 9.96 7.22 8 107,134 12.10 220,716 18.21 12.80 N.i.e. 85,657 9.67 35,878 2.96 - - Total 885,479 100.00 1,211,986 100.00 5.37 -Not applicable. N.i.e. Not included elsewhere. Sources: Phase II household survey, 1972; DANE household survey, 1978. Table A16. Distribution of Employment by Radial Sector and Industry, Bogota, 1972-78 (percent) Manufacturing Commerce Finance Services Radial sector 1972 1978 1972 1978 1972 1978 1972 1978 1 18.20 6.01 19.43 15.75 42.11 41.43 22.62 12.91 2 9.30 8.09 8.22 10.65 2.22 1.73 6.97 8.21 3 17.28 18.87 12.89 15.32 7.72 3.18 11.08 10.61 4 14.03 15.76 7.36 10.80 7.40 2.81 5.96 6.24 5 15.26 20.04 9.04 10.48 8.11 9.67 7.53 9.48 6 6.84 11.71 9.99 9.70 8.82 3.31 10.13 14.26 7 6.61 7.70 9.86 9.66 4.83 10.24 11.77 11.91 8 8.77 8.17 14.95 16.60 6.66 27.00 15.68 24.22 N.ie. 3.72 3.66 8.27 1.05 12.12 0.63 8.25 2.15 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. Sources: Phase 11 household survey, 1972; DANE household survey, 1978. 146 Appendix A Table Al 7. The Origin-Destination Ratio ofJobs by Radial Sector, Bogota, 1973-78 Radial sector All' Manufacturing Commerce Finance Services 1 2.37 2.51 3.15 3.32 1.45 2 0.94 1.30 0.27 - 0.84 3 0.83 0.30 1.53 3.44 1.41 4 1.13 1.48 0.65 - 1.34 5 0.76 0.94 1.14 0.32 0.65 6 0.30 0.53 - 0.19 0.17 7 0.65 1.03 0.36 0.71 0.62 8 0.69 1.22 0.48 0.33 1.10 -Not applicable. Note: The origin-destination ratio is the number of jobs moving out of a zone divided by the number of jobs moving into the zone. a. Includes all industry groups. Source: DANE household survey, 1978. Table Al 8. The Origin-Destination Ratio ofJobs in Small and Large Firms by Ring, Bogota, 1973- 78 AIl Manufacturing Commerce Finance Services Ring Smale Large' Small Large Small Large Small Large Small Large 1 1.46 2.84 1.57 2.39 2.56 3.47 1.22 8.76 1.07 1.65 2 0.73 0.66 1.16 2.17 0.86 0.20 0.34 0.34 0.89 0.86 3 0.99 1.51 0.98 2.00 0.82 1.88 1.04 1.03 1.35 1.31 4 0.86 0.48 1.20 0.77 0.55 0.46 - 0.06 0.94 0.59 5 0.88 0.23 0.52 0.18 - 0.23 1.77 - 1.45 - 6 0.56 0.64 - - - - - 0.64 -Not applicable. Note: The origin-destination ratio is the number of jobs moving out of a zone divided by the number of jobs moving into the zone. a. All industry groups. b. Establishments with less than 10 employees. c. Establishments with 10 or more employees. Source: DANE household survey, 1978. Table A19. The Origin-Destination Ratio ofJobs in Small and Large Finns by Radial Sector, Bogota, 1973-78 All' Manufacturing Commerce Finance Services Radial sector Smalt Large' Small Large Small Large Small Large Small Large I 1.46 2.84 1.57 2.39 2.56 3.47 1.22 8.76 1.07 1.65 2 0.71 1.80 1.21 1.69 0.27 - - - 0.61 0.98 3 1.42 0.44 0.46 0.22 2.40 0.60 3.44 - 3.24 0.99 4 0.60 1.42 0.75 1.63 0.76 0.54 - - 0.53 2.27 5 0.70 0.71 0.83 0.96 0.35 1.06 - 0.32 - 0.51 6 0.73 0.18 0.76 0.36 - - - 0.19 1.65 - 7 0.52 0.85 1.00 1.04 0.51 0.41 0.33 1.35 0.26 0.83 8 1.26 0.63 4.60 1.45 0.89 0.37 1.00 0.26 2.00 1.07 -Not applicable. Note: The origin-destination ratio is the number of jobs moving out of a zone divided by the number of jobs moving into the zone. a. All industry groups. b. Establishments with less than 10 employees. c. Establishments w.ith 10 or more employees. Source: DANE household survey, 1978. Table A20. Distribution of Employment by Radial Sector and Industry, Cali, 1976-78 All5 Manufacturing Commerce Finance Services Radial sector 1976 1978 1976 1978 1976 1978 1976 1978 1976' 1978 1 31.51 26.19 20.19 14.94 48.38 54.97 45.68 84.90 29.15 25.34 2 21.56 13.86 13.19 12.33 13.04 7.60 40.10 15.10 36.14 15.33 3 29.93 30.00 45.00 46.24 27.33 19.23 10.16 0.00 15.92 13.23 4 7.74 11.59 14.94 16.75 2.24 8.63 0.30 0.00 0.58 8.15 5 3.43 7.06 2.86 5.11 5.37 5.32 0.30 0.00 4.88 11.62 6 2.95 9.35 2.53 4.25 1.93 4.25 0.45 0.00 7.13 19.82 7 2.90 1.96 1.28 0.37 1.70 0.00 3.02 0.00 6.22 6.50 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Note: Data are for establishments with 10 or more employees. a. All industry groups. b. Excludes employment in the governmenit sector. Sources: Social security data file, 1976; DANE household survey, 1978. 150 Appendix A Table A21. The Origin-Destination Ratio ofJobs by Radial Sector, Cali, 1973-78 Radial sector All' Manufacturing Commerce Finance Services 1 1.85 1.97 1.71 1.58 2.03 2 0.33 0.36 0.21 - 0.19 3 0.93 0.67 1.02 - 1.63 4 0.91 1.68 0.72 - 0.75 5 0.95 1.06 2.22 - - 6 0.87 0.91 - - 0.48 7 0.53 - - - 0.53 -Not applicable. Note: The origin-destination ratio is the number of jobs moving out of a zone divided by the number of jobs moving into the zone. a. All industrv groups. Source: DANE household survey, 1978. Table A22. The Origin-Destination Ratio ofJobs in Small and Large Firms by Ring, Cali, 1973-78 All' Manufacturing Commerce Finance Services Ring Smallb Large' Small Large Small Large Small Large Small Large 1 2.30 1.53 2.20 1.75 2.48 1.26 - 1.00 1.47 2.62 2 0.46 0.97 0.52 0.97 0.35 1.91 - 0.52 0.40 3 0.77 1.15 0.52 1.60 1.21 - - 1.80 - 4 1.27 0.28 1.64 0.21 - - - - 0.33 - 5 1.11 - - - - _ _ -Not applicable. Note: The origin-destiination ratio is the number of jobs moving oiut of a zone divided by the number of jobs movinlg into the zone. a. All industry groups. b. Establishments with less than 10 employees. c. Establishments with 10 or more employees. Source: DANF. household survey, 1978. Table A23. The Origin-Destination Ratio ofJobs in Small and Large Firms by Radial Sector, Cali, 1973- 78 All' Manufacturing Commerce Finance Services Radial - sector Small5 Large' Small Large Small Large Small Large Small Large 1 2.30 1.53 2.20 1.75 2.48 1.26 - 1.00 1.47 2.62 2 0.31 0.39 - 0.90 - 0.48 - 0.45 3.44 3 0.83 1.01 0.51 0.71 1.03 1.00 - - 1.89 - 4 0.54 3.94 - 3.94 0.72 - - - 0.75 5 0.67 3.00 0.59 3.00 2.22 - - 6 0.91 0.75 1.13 - - - - 0.48 -Not applicable. Note: The origin-destination ratio is the number of jobs moving out of a zone divided by the number of jobs moving into the zone. a. All industry groups. b. Establishments with less than 10 employees. c. Establishments with 10 or more employees. Souirce: DANE household survey, 1978. Statistical Tables 153 Table A24. Distribution of Firns by Size and Industry Group, Bogota, 1978: Social Security Data Percentage offirms by size' Industry group 1-4 5-9 10 or more Total Agriculture 54.66 20.98 24.46 100.00 Mining 27.27 6.82 65.91 100.00 Manufacturing 39.49 21.85 38.66 100.00 Construction 42.35 16.76 40.89 100.00 Utilities 27.91 13.95 58.14 100.00 Commerce 57.58 18.70 23.72 100.00 Transport and communication 59.04 15.29 25.67 100.00 Services 57.88 19.52 22.60 100.00 Other 57.06 14.50 28.44 100.00 All 51.92 19.58 28.50 100.00 Number of firms 16,374 6,175 8,987 31,536 Percentage of employment by size offirma 1-4 5-9 10 or more Total Agriculture 5.72 7.24 87.04 100.00 Mining 2.84 2.73 94.43 100.00 Manufacturing 3.83 6.29 89.88 100.00 Construction 4.34 5.73 89.93 100.00 Utilities 1.40 1.69 96.91 100.00 Commerce 8.19 8.52 83.29 100.00 Transport and communication 4.06 3.77 92.17 100.00 Services 9.25 10.14 80.61 100.00 Other 5.08 4.74 90.18 100.00 All 6.23 7.51 86.26 100.00 Number of workers 33,918 40,867 469,571 544,356 a. Number of employees. Source: Social security data file, 1978. Table A25. Percentage Distribution of Workers Affiliated with Social Security System by Ring, Bogota, 1978: Comparison of Two Data Sets Alla Manufacturing Commerce Finance Services Ring Survey' SS. Survey SS Survey 55 Survey SS Survey SS 1 19.27 22.55 6.30 9.85 25.78 29.29 41.95 45.37 20.69 29.06 2 20.88 20.37 13.33 15.47 25.19 19.57 31.67 33.40 24.63 22.79 3 19.13 25.72 26.73 31.01 14.54 25.10 12.02 12.25 16.88 25.15 4 18.01 14.96 24.27 19.41 18.63 12.21 10.19 4.45 15.40 12.29 41 5 16.76 12.78 23.37 19.86 14.07 12.30 3.01 2.45 13.35 6.09 6 1.98 1.04 0.59 0.74 1.12 0.58 0.88 0.11 4.62 2.05 Ni.e. 3.98 2.61 5.41 3.66 0.68 0.96 0.29 1.98 4.42 2.56 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. a. All industry groups. b. Workers who reported social security affilitation in the DANE 1978 household survey. c. All employment in the social security data file on establishments. Sources: DANE household survey, 1978; social security data file, 1978. Table A26. Percentage Distribution of Workers Affiliated with Social Security System by Radial Sector, Bogoti, 1978: Comparison of Two Data Sets Alta Manufacturing Commerce Finance Services Radial - -_- - sector Surveyb SS, Survey SS Survey SS Survey SS Survey SS 1 19.27 22.55 6.30 9.85 25.78 29.29 41.95 45.37 20.69 29.06 2 3.37 1.54 4.73 2.04 2.18 1.01 1.58 0.28 4.20 1.47 3 8.54 5.23 15.52 9.05 6.46 3.49 1.06 0.88 5.54 2.48 _ 4 10.35 16.09 18.38 28.06 8.72 19.25 2.56 1.71 6.89 4.89 5 18.18 17.97 28.58 28.40 17.44 15.59 11.99 13.73 13.15 5.81 6 7.53 3.80 7.09 4.15 6.16 2.81 2.36 0.68 9.91 2.63 7 8.37 8.91 5.84 6.88 9.05 10.09 8.29 2.58 12.68 16.38 8 19.92 21.30 8.16 7.91 23.54 17.52 29.92 32.80 22.52 34.73 N.i.e. 3.98 2.61 5.41 3.66 0.68 0.96 0.29 1.98 4.42 2.56 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. a. All industry groups. b. Workers who reported social security affiliation in the DANE 1978 household survey. c. All employment in the social security data file on establishments. Sources: DANE household survey, 1978; social security data file, 1978. Table A27. Percentage Distribution of Employment in Large Firms by Ring, Bogota, 1978: Comparison of Two Data Sets Alla Manufacturing Commerce Finance Services Ring Survey' SS, Survey SS Survey SS Survey SS Survey SS 1 16.78 22.20 7.00 9.70 22.57 29.22 33.99 46.09 18.60 28.32 2 20.68 20.29 13.17 14.79 24.84 18.62 39.77 34.98 22.52 23.58 3 18.91 26.14 24.60 31.56 17.99 26.09 11.29 11.44 18.19 26.08 4 19.16 14.17 23.58 18.72 18.68 11.13 11.03 3.16 20.11 10.88 5 17.62 13.37 24.88 20.59 14.00 13.64 3.17 2.22 13.71 6.00 6 2.05 1.00 1.09 0.65 0.21 0.52 0.00 0.06 3.18 2.19 N.i.e. 4.79 2.84 5.68 3.99 1.70 0.78 0.76 2.05 3.69 2.96 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. Note: The data are for establishments with 10 or more employees. a. All industry groups. b. Workers who reported social security affiliation in the DANE 1978 household survey. c. All employment in the social security data file on establishments. Sources: DANE household survey, 1978; social security data file, 1978. Table A28. Percentage Distribution of Employment in Large Firms by Radial Sector, Bogota, 1978: Comparison of Two Data Sets All' Manufacturing Commerce Finance Services Radial sector Surveyb Ss Survey SS Survey SS Survey SS Suroey SS 1 16.78 22.20 7.00 9.70 22.57 29.22 33.99 46.09 18.60 28.32 2 3.74 1.29 4.65 1.60 1.16 0.73 1.57 0.19 4.84 1.30 3 8.83 4.99 14.89 8.72 8.26 2.49 0.82 0.80 7.98 2.24 4 9.97 16.75 20.33 28.97 7.46 20.53 2.98 1.56 4.21 4.96 _ 5 17.41 19.38 27.28 30.14 17.09 17.16 13.59 14.71 12.12 5.77 6 10.66 3.62 7.77 3.56 4.75 2.73 2.27 0.41 16.87 2.56 7 8.61 8.29 5.81 5.97 13.53 9.48 11.73 1.93 9.23 17.32 8 19.20 20.63 6.60 7.34 23.48 16.88 32.29 32.26 22.44 34.58 N.i.e. 4.79 2.84 5.68 3.99 1.70 0.78 0.76 2.05 3.69 2.96 Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 N.i.e. Not included elsewhere. Note: The data are for establishments with 10 or more employees. a. All industry groups. b. Workers who reported social security affiliation in the DANE 1978 household survey. c. All employment in the social security data file on establishments. Sources: DANE household survey, 1978; social security data file, 1978. Table A29. Distribution of Firms by Size: Mature Firns, Births, and Deaths, Bogota, 1970-75 (percent) Mature Births Deaths Size (number employed) Establishments Employment Establishments Employment Establishments Employment 1-4 0.18 0.01 0.26 0.02 1.26 0.19 5-9 2.95 0.35 3.12 0.74 14.86 4.21 10-14 15.94 3.02 26.49 9.21 37.78 16.84 15-19 13.95 3.74 18.57 9.26 13.10 8.41 20-24 11.06 3.82 13.77 8.96 6.80 5.77 25-34 14.38 6.58 14.42 12.27 7.81 8.79 35-49 12.07 7.85 9.87 11.77 6.55 10.38 50-74 10.60 10.36 6.10 10.40 5.54 13.28 75-99 4.88 6.68 1.95 4.93 2.27 7.24 100-199 8.29 18.69 3.25 13.46 3.27 17.38 200-499 4.24 20.70 2.08 16.35 0.76 7.57 500 and more 1.47 18.21 0.13 2.63 0.00 0.00 Total 100.00 100.00 100.00 100.00 100.00 100.00 Source: Industrial directory file, 1970-75. Statistical Tables 159 Table A30. Manufacturing Establishments by Initial Year of Operation, Bogota and Cali, 1970 Bogotd Cali Cumulative Cumulative Year Age Percenta percent Percenta percent 1970 0 11.30 14.69 1969 1 9.61 20.91 7.84 22.53 1968 2 10.34 31.25 8.40 30.93 1967 3 6.24 37.49 6.02 36.95 1966 4 5.70 43.19 5.88 42.83 1963-65 5- 7 14.39 57.58 13.71 56.54 1960-62 8-10 11.30 68.88 10.77 67.31 1955-59 11-15 12.57 81.45 13.29 80.60 1950-54 16-20 6.93 88.38 7.14 87.74 1945-49 21-25 3.97 92.35 5.46 93.20 1940-44 26-30 1.78 94.13 2.24 95.44 1935-39 31-35 1.32 95.45 0.84 96.28 1931-34 36-39 0.55 96.00 0.70 96.98 1930 or before 40 or older 0.87 96.87 1.54 98.52 N.i.e. - 3.13 100.00 1.48 100.00 Number of establishments 2,196 715 -Not applicable. N.i.e. Not included elsewhere. a. Percentage of firms that started operations in the year or period indicated. Source: Industrial directory file, 1970. Appendix B Questionnaire for the Survey of Manufacturing Establishments Name of person responding to survey Position PART I. ESTABLISHMENT CHARACTERISTICS (Part I is to be completed for all firms.) 1. Name of establishment (pre-coded) Other name used: 2. Address of establishment (pre-coded) Barrio name (pre-coded) 3. This establishment is: (1) a single-establishment operation (2) headquarters of a multi-establishment operation (3) a branch of a multi-establishment operation If (3), answer 4. 4. Name of parent company Address of parent company Other name used A. Location Tenure (All questions from here on are with respect to the establishment located at this address as specified in question no. 2. First, I will ask you about location history of your establishment.) Al. When was this establishment founded? - Year A2. When did this establishment first operate at this location? Year 160 Questionnaire for the Survey of Manufacturing Establishments 161 (1) It began operation at this location in 1970 or before. (Ask Part III after completing Part 1.) It began operation at this location after 1970, and (2) was newly founded. (Ask Part IV after completing Part 1.) (3) was relocated from another location in Bogota. (Ask Parts II, III, & IV after completing Part I.) (4) was relocated from another location outside Bogota. (Ask Parts II, III, & IV after completing Part 1.) B. Plant Characteristics (Now, I will ask you a series of questions about characteristics of your establishment including those of the plant, outputs, and inputs. Let me start with the characteristics of your plant.) B1. What are the major products manufactured at this establish- ment? List up to three in order of importance (for example, women's clothes, rubber shoes, automobile parts, etc.) (1) (2) (3) B2. Which of the following production processes does your plant most closely resemble? (1) many different kinds of products produced, short produc- tion runs, general purpose equipment, constant attention to scheduling operations. (2) few kinds of products, long production runs, special pur- pose equipment, relatively easy scheduling. (3) combination of both. (4) other, please specify: B3. Roughly, can you give information on the following? Annual sales: Col$ Annual purchase of raw materials: Col$ Annual wage bill (including fringe benefits): Col$ Replacement value of plant and equipment: Col$ B4. Roughly, how many square meters of work space does your plant have? Rent square meters (Ask B6) Own square meters Total square meters 162 Appendix B B5. Roughly, how many square meters is the land area of plant site? Rent square meters (Ask B6) Own square meters Total square meters B6. Roughly, how much rent do you pay per year? Plant Col$ Site Col$ Total Col$ B7. How old are the main buildings of this establishment? ______ Years B8. Roughly, what proportion of the land is occupied by buldings on this site? _ % B9. Do you have some land space reserved for plant expansion next to the present plant site? yes no If yes, substantial _ _, modest B 10. Is the building used for your primary operation single-story or multi-story? (1) single story (2) multi-story (3) mixed single & multi-story Bl 1. Do you store your products and raw materials indoors or out- doors? (indoors = 1, outdoors = 2) (1) products (2) raw materials B 12. Where is the warehousing for your finished goods done? (Check all applicable) (1) on this site (2) within Bogota; roughly km from here. (3) outside Bogota; roughly km from here. C. Employment (Now, I will ask some questions about employees of your estab- lishment.) Cl. Roughly, how many people work at this establishment? Full-time Part-time (1) management (2) skilled workers (3) unskilled workers (4) total Questionnaire for the Survey of Manufacturing Establishments 163 C2. Roughly, what was the maximum number of employees working at one time in this establishment during the past five years (in- cluding part-time workers)? C3. Roughly, what are the average hourly wage rates (excluding fringe benefits)? (1) skilled workers Col$ (2) unskilled workers Col$ C4. On the average, how many (8-hour) shifts are run at the plant? (1) one shift (2) two shifts (3) three shifts C5. Roughly, what proportion of workers are unionized? % C6. Where do most of your employees live? Please give rough per- centage guesses to the following categories, if possible. Management Non-management Immediate neighborhood % Within this barrio _ % Adjacent barrios % % Northern part of the city %_ _ Southern part of the city %% Outside the city but within metro area % % Total 100% Total 100% C7. What proportion of your employees travel to work by: Foot % Commercial bus % Company bus % Car _ Total 100% D. Transportation Access and Proximity to Markets (The following are concerning transportation access for inputs and outputs of your establishment.) DI. Roughly, the percent of the value of output shipped within Colombia by: Truck only Rail only Truck-rail combination A 164 Appendix B Truck-water combination % Air (during any part of trip) % Other, please specify % Total 100% D2. Roughly, the percent of the value of raw material inputs shipped within Colombia by: Truck only % Rail only % Truck-rail combination % Truck-water combination % Air (during any part of trip) % Other, please specify Total 100% D3. If you use rail for making and/or receiving deliveries, does your plant use a railroad siding? yes no If no, how far is the nearest railway station? _ _ km D4. If you use trucking for making and/or receiving deliveries, how far is the nearest highways? 2-lane highways 4-lane highways Within 1 km Within 5 kms Within 10 kms More than 10 kms D5. What percent of your products are sold in the following areas? Within Bogota % Within Cundinamarca % National % International _ % Total 100% D6. Within BogotA, on the average, how far are your products delivered from this location? (1) less than 1 km (2) 1 to 5 kms (3) 5 to 10 kms (4) 10 to 20 kms (5) more than 20 kms D7. Roughly, what proportion of your output is used mainly as in- puts of other industries? Questionnaire for the Survey of Manufacturing Establishments 165 (1) less than 20% (2) 20-40% (3) 40-60% (4) 60-80% (5) more than 80% D7.1 What industries are they? (1) (2) (3) D7.2 If at least half of your output is used by these industries, how far are they located from your establishment? (1) less than I km (2) 1 to 5 kms (3) 5 to 10 kms (4) 10 to 20 kms (5) more than 20 kms D8. What percent of your raw materials and/or input components are in the following areas? Within Bogota Within Cundinamarca % National % International % Total 100% D9. Within Bogoti, on the average, what is the distance of input deliveries to your establishment? (1) less than I km (2) 1 to 5 kms (3) 5 to 10 kms (4) 10 to 20 kms (5) more than 20 kms DIO. Roughly, what proportion of your inputs is output of other industries? (1) less than 20% (2) 20-40% (3) 40-60% (4) 60-80% (5) more than 80% D1 0.1 What industries are they? (1) 166 Appendix B (2) (3) D 10.2 If at least half of your inputs is output of other industries, how far are these industries located from your establishment? (1) less than I km (2) 1 to 5 kms (3) 5 to 10 kms (4) 10 to 20 kms (5) more than 20 kms DI 1. Are there any public (or private) transportation problems for employees at this location? (1) yes (2) no If yes, what are these problems? E. Public Services and Taxes (The following questions have to do with the quality of public services and local taxes.) El. How much public utility service does your plant require? (substantial = 1, modest = 2, very little = 3, almost none = 4) (1) electricity (2) water (3) natural gas E2. On the average, how frequently are public utility services inter- rupted? (almost never = 1, once a week = 2, twice a week = 3, more than twice a week = 4) (1) electricity (2) water (3) natural gas E3. How many private security personnel (guards) do you have to prevent criminal attacks on employees and thefts of company property? persons Do you consider the cost of keeping these persons (1) substantial (2) modest (3) very little E4. What do you think of the city's fire protection service in this area? Questionnaire for the Survey of Manufacturing Establishments 167 (1) excellent (2) good enough (3) not enough (4) very poor E5. How often are the city's sewerage and waste removal services interrupted? (1) rarely (2) once a month (3) twice a month (4) more than twice a month E6. How often is your business disrupted by the lack of adequate road maintenance service by the city? (1) almost never (2) once a month (3) twice a month (4) more than twice a month E7. On the average, how much local taxes do you pay per year on the property of this establishment (including plant, buildings, and land)? Col$ _ _ Do you consider this amount excessive _ , about right , or on the low side _? E8. What local taxes that you pay affect your business operations most seriously? (1) (2) (3) (4) none E9. Can you think of any incentive systems provided by the local or national government that are important for your business operations? (1) (2) (3) F. Past Trends and Future Growth Fl. What has been the average growth rate of your establishment in terms of output (or sales) over the past five years? (1) declined (4) 5-10% peryear 168 Appendix B (2) no growth (5) 10-20% per year (3) - less than 5% per year (6) over 20% per year F2. How has the nature of the products manufactured changed over the past five years? (1) no fundamental change (2) introduced new product lines (3) different product mix but in the same industry (4) changed to entirely new kind of industry F3. Do you anticipate that your industry will be rapidly expanding in future years or is it likely to decline? (1) rapidly expanding (2) expanding somewhat (3) no growth (4) declining F4. Do you anticipate that the sales of this establishment at this loca- tion will increase or decrease in the next five years? (1) increase substantially (2) increase slightly (3) remain the same (4) decrease slightly (5) decrease substantially G. Summary Evaluation of Present Location GI. In sum, please evaluate your present location by the factors listed below (satisfactory = 1, not satisfactory = 2, very important 3, somewhat important = 4, not important = 5): (1) plant capacity (2) rent payment (3) availability of skilled workers (4) cost of skilled workers (5) availability of unskilled workers (6) cost of unskilled workers (7) cost of utilities (8) proximity to suppliers (9) proximity to customers (i.e., market) (10) proximity to competitors (I 1) highway access (12) railroad access (13) cost of nearby land or expansion space Questionnairefor the Survey of Manufacturing Establishments 169 (14) proximity to subcontractors, repair and maintenance services, and accounting, legal, and other business services (15) property tax rates (16) municipal services (police, fire, street maintenance, etc.) (17) security (18) pleasant surroundings, nearby recreational facilities, etc. (19) local community attitudes toward this business PART 11. COMPARISONS WITH FORMER LOCATION (This part is to be completed for recent movers to check how conditions changed after moving to the present location.) A. Former Location Al. The address of former location: Street City (if not Bogota) Barrio name A2. Roughly what is the distance between the old and the new locations? - km A3. How would you best describe the condition of your former plant just prior to your move from it? (1) in good condition, but cramped (2) in good condition, but obsolete (3) still serviceable (4) worn out B. Experiences after Relocation (Try to answer the following as resulting solely from the change in location at the time of move. Put down + sign if increased and - sign if decreased.) After moving to the present location, Changed by Stayed Not More about more than the than 10- 20% same 5% 5-10% 20% (specify) Bl. Production (or sales) 170 Appendix B B2. Work space of plant B3. Land area of the site B4. Rent payment per square meter B5. Number of skilled workers B6. Hourly wages of skilled workers B7. Number of un- skilled workers B8. Hourly wages of unskilled workers B9. Commuting distance for managers BIO. Commuting distance for workers B 11. Output delivery distance B12. Input delivery distance B13. Tax payment B 14. Utility costs B15. Any changes in the quality of public utility servicesP Substantially Somewhat Became improved improved Unchanged worse Electricity Water Natural gas B16. Any changes in the city services? Substantially Somewhat Became improved improved Unchanged worse Fire protection Police service Sewerage Questionnaire for the Survey of Manufacturing Establishments 171 Waste removal Road maintenance C. Important Factorsfor Relocation Cl. List five items that you considered to be most important when you made the decision to relocate your establishment to the pre- sent location. (List in order of importance. If needed, the inter- viewer should assist the respondent by referring to GI in Part I.) (1) (2) (3) (4) (5) C2. Roughly what percent of your labor force moved with you to the new location? _ _ % C3. Was crime or vandalism at the former location an important con- sideration in your relocation? yes, no C4. Was your relocation forced because of public action such as highway construction, urban renewal, or zoning regulations, etc.? yes, no C5. Was the need for larger plant size one of the main reasons for relocation? yes, no PART III. PLANS FOR CAPACITY EXPANSION OR RELOCATION (This part is to be completed for stationary firms and movers about the future expansion and relocation possibilities.) Al. Do you have any plans for expanding your operations within the next five years, such as leasing more space, adding more buildings, or utilizing now vacant space? yes, no If yes: (1) Is this expansion likely to take place: here at this location? another location in Bogoti? (name of barrio ) another location outside Bogota? (name of city (2) Roughly how many new workers will you hire as a result of this expansion? (3) How much additional floor space will you require? square meters 172 Appendix B A2. Do you have any plans for relocating the entire operation from the present site to another location in the next five years? yes, no If yes: (1) Is the new location likely to be: in Bogota? (name of barrio ) outside Bogota? (name of city (2) Roughly how would your labor force change? will increase by persons will decrease by persons (3) How much additional floor space will you require? square meters A3. (If needed, the interviewer should assist the respondent by referring to GI in Part I.) If you have plans for relocating your establishment or for expanding your operations at another location, what factors would you consider to be most important in selecting the new site? List five in order of importance: (1) (2) (3) (4) (5) PART IV. ALTERNATIVE LOCATION CONSIDERATIONS OF NEWLY ESTABLISHED FIRMS (This part is to be completed for newly established firms since 1970 and for movers, to learn about alternative locations considered before select- ing the present location.) Al. Did you consider locations other than this one when you were look- ing for a plant site? yes, no If yes, list the barrio of those locations. (1) (2) (3) A2. What were the main reasons for not choosing these locations? Specify reasons for each location. Questionnaire for the Survey of Manufacturing Establishments 173 Barrio name Reason (1) (2) (3) A3. Were there any government incentive schemes that induced your es- tablishment to the present location? yes, no If yes, list those schemes (1) (2) (3) Bibliography Alonso, William. 1969. "A Reformulation of Classical Location Theory and Its Relation to Rent Theory." In GeraldJ. Karaska and David F. Bramhall, eds., Lo- cational Analysisfor Manufacturing. Cambridge, Mass.: MIT Press. Arrow, K. J., and others. 1961. "Capital-Labor Substitution and Economic Effic- iency." Review of Economics and Statistics 43(3):225-50. Bachi, Roberto. 1957. Statistical Analysis of Geographical Series. Jerusalem: Kaplan School, Hebrew University, and Israel Central Bureau of Statistics. Birch, D. L. 1979. "TheJob Generation Process." MIT Program on Neighbor- hood and Regional Change. Prepared for U.S. Department of Commerce, Economic Development Administration. Blackley, Paul R. 1985. "The Demand for Industrial Sites in a Metropolitan Area: Theory, Empirical Evidence, and Policy Implications." Journal of Urban Econ- omics 17(2):247-61. Bronitsky, Leonard, and others. 1975. Urban Data Book. U.S. Department of Transportation, Report DOT-TSC-OS5-75-45-I. Washington, D.C. Burstein, N. R. 1980. "Voluntary Income Clustering and the Demand for Hous- ing and Local Public Goods." Journal of Urban Economics 7(2):175-85. Cameron, Gordon C. 1973. "Intra-Urban Location and the New Plant." Papers of the Regional Science Association 31:125-44. Carlton, Dennis W. 1979. "Models of New Business Location." In William C. Wheaton, ed., Interregional Movements and Regional Growth. Washington, D.C.: Urban Institute. _ 1983. "The Location and Employment Choices of New Firms: An Economic Model with Discrete and Continuous Endogenous Variables." Review of Economics and Statistics 65(3):440-49. Choe, S. C., and B. N. Song. 1984. "An Evaluation of Industrial Location Policies for Urban Deconcentration in the Seoul Region."Journal of Environmental Studies 14:73-116. Clapp,J. M. 1979. "The Substitution of Urban Land for Other Inputs."Journalof Urban Economics 6(l):122-34. 1 74 Bibliography 1 75 . 1980. "The Elasticity of Substitution for Land: The Effects of Measure- ment Errors."Journal of Urban Economics 8(2):255-63. -____ 1981. "The Elasticity of Substitution of Nonland for Land: Reconciliation of Diverse Estimates." Journal of Regional Science 21(1):123-25. DANE (Departamento Administrativo Nacional de Estadistica). 1974. Statistical Bulletin 279, October. . 1975a. Statistical Bulletin 289, August. . 1975b. Statistical Bulletin 293, December. . 1976a. Statistical Bulletin 299, June. _ .____ 1976b. Statistical Bulletin 301, August. Dennis, Robert. 1978. "The Decline of Manufacturing Employment in Greater London." Urban Studies 15(l):63-73. Diewert, W. E. 1974. "Applications of Duality Theory." In M. D. Intriligator and D. A. Kendrick, eds., Frontiers of Quantitative Economics. Vol. 2. New York: Elsevier-North Holland. Eberts, R. W. 1981. "An Empirical Investigation of Intraurban Wage Gradients." Journal of Urban Economics 10(1):50-60. Ellickson, Bryan. 1977. Economic Analysis of Urban Housing Markets: A New Approach. Santa Monica, Calif.: Rand Corporation. -_____ 1981. "An Alternative Test of the Hedonic Theory of Housing Markets." Journal of Urban Economics 9(l):56-79. Erickson, R. A., and M. Wasylenko. 1980. "Firm Relocation and Site Selection in Suburban Municipalities." Journal of Urban Economics 8(1):69-85. Fallis, G. B. 1975. "The Technology of Production and the Location of Employ- ment in Urban Areas." Ph.D. dissertation, Princeton University. . 1979. "Factor Substitution, Employment Density, and Suburbaniza- tion."Journal of Urban Economics 6(2):156-75. Firn, J. R. 1976. "Economic Micro Data Analysis and Urban-Regional Change: The Experience of Gurie." In J. K. Swales, ed., Establishment-Based Research Con- ference Proceedings. Department of Social and Economic Research, University of Glasgow. Friedman,Joseph. 1975. Housing Location and the Supply of Local Public Services. Santa Monica, Calif.: Rand Corporation. Hamer, Andrew M. 1985. Decentralized Urban Development and Industrial Location Behavior in Sdo Paulo. World Bank Staff Working Paper 732. Washington, D.C. Hansen, Willard B. 1961. "An Approach to the Analysis of Residential Exten- sion." Journal of Regional Science 3(1):37-55. Hansen, Walter G. 1959. "How Accessibility Shapes Land Use." Journal of American Institute of Planners 25(1):73-76. Hanushek, E. A., and B. N. Song. 1978. "The Dynamics of Postwar Industrial Location." Review of Economics and Statistics 60(4):515-22. 1 76 Bibliography Henderson, J. Vernon. 1980. "A Framework for International Comparisons of Systems of Cities." Urban and Regional Report 80-3. World Bank, Develop- ment Economics Department. Hoover, Edgar M., and Raymond Vernon. 1959. Anatomy of a Metropolis. Cam- bridge, Mass.: Harvard University Press. Huff, P. L. 1961. Determination of Intra-Urban Retail Trade Areas. Real Estate Research Program, University of California, Los Angeles. Ingram, Gregory K. 1976. "Reductions in Auto Use from Carpools and Improved Transit." Harvard University, Department of City and Regional Planning Dis- cussion Paper D76-10. .. 1977a. "Transportation-Land Use Interactions and Future Patterns of Urban Growth." Joint Center for Housing Studies of MIT and Harvard University/National Bureau of Economic Research Conference, Cambridge, Mass. _____, ed. 1977b. Residential Location and Urban Housing Markets. Cambridge, Mass.: Ballinger. Ingram, Gregory K., and Alan Carroll. 1981. "The Spatial Structure of Latin American Cities."Journal of Urban Economics 9(2):257-73. Ingram, Gregory K., A. Pachon, andJ. F. Pifieda. 1982. "Summary of Results and Policy Implications of the City Study." World Bank, Urban Development Department Discussion Paper. Institute of Office Management. 1962. Clerical and Salaries Analysis: 1962. London. Isard, Walter. 1960. Methods of Regional Analysis. Cambridge, Mass.: MIT Press. Kain, J. F. 1967. "The Distribution and Movement of Jobs and Industry." In James Q. Wilson, ed., The Metropolitan Enigma: Inquiries into the Nature and Dimen- sions of America's Urban Crisis. Rev. ed. Cambridge, Mass.: Harvard University Press. Kain, John F., and John M. Quigley. 1975. Housing Markets and Racial Discrimina- tion: A Microeconomic Analysis. New York: National Bureau of Economic Re- search. Kau, J. B., and C. F. Lee. 1976. "Capital-Land Substitution and Urban Land Use.".Journal of Regional Science 16(l):83-92. Keeble, David, and Egbert Wever, eds. 1986. New Firms and Regional Development. London: Croom Helm. Kemper, Peter. 1973. "The Location Decisions of Manufacturing Firms within the New York Metropolitan Area." Ph.D. dissertation, Yale University. Kemper, Peter, and Roger W. Schmenner. 1974. "The Density Gradient for Manufacturing Industry."Journal of Urban Economics 1 (4):410-27. Koenker, Roger. 1972. "An Empirical Note on the Elasticity of Substitution be- tween Land and Capital in a Monocentric Housing Market."Journal of Regional Science 12(2):299-305. Bibliography 1 77 Lakshmanan, T. R. 1964. "A Model for the Distribution of Urban Activities: For- mulation, Evaluation, and Reformulation." Paper presented at the seminar on Models of Land Use Development, Institute of Urban Studies, University of Pennsylvania, October 22-26. Alan M. Voorhees & Associates, Washington, D.C. _- 1965. "A Theoretical and Empirical Analysis of Intraurban Retail Loca- tion." Ph.D. dissertation, Ohio State University. Lau, L. J., and P. A. Yotopoulos. 1971. "A Test for Relative Efficiency and Ap- plication to Indian Agriculture." American Economic Review 61(1):94-108. Lee, K. S. 1978. "Distribution of Manufacturing Establishments and Employ- ment in BogotA and Cali." City Study Workshop Paper. World Bank, Develop- ment Economics Department. _- 1981 a. "Intra-Urban Location of Manufacturing Employment in Colom- bia."Journal of Urban Economics 9(2):222-41. -_____ 1981b. "Determinants of Intra-Urban Location of Manufacturing Em- ployment: An Analysis of Survey Results for BogotA, Colombia." Urban and Regional Report 81-3. World Bank, Development Economics Department. _- 1982. "A Model of Intra-Urban Employment Location: An Application to Bogota, Colombia."Journal of Urban Economics 12(3):263-79. -_____ 1985a. "Decentralization Trends of Employment Location and Spatial Policies in LDC Cities." Urban Studies 22(2):151-62. -_____* 1985b. "An Evaluation of Decentralization Policies in Light of Changing Location Pattems of Employment in the Seoul Region." Water Supply and Urban Development Department Discussion Paper 60. World Bank. -_____. 1985c. "A Model of Intra-Urban Employment Location: Replicating the Bogota Experiment in Seoul." Water Supply and Urban Development Depart- ment Discussion Paper 90. World Bank. -_____* 1986. "Industrial Location Policies: They're Costly and Inefficient in Korea." World Bank Research News 7(1), Summer. Lee, YoonJoo. 1979. "The City Study: The Available Data." Vol. 2. Urban and Regional Report 79-13. World Bank, Development Economics Department. -_____ 1985. The Spatial Structure of the Metropolitan Regions of Brazil. World Bank Staff Working Paper 722. Washington, D.C. Leone, Robert. 1971. "Location of Manufacturing Activity in the New York Met- ropolitan Area." Ph.D. dissertation, Yale University. Leone, Robert, and Raymond Struyk. 1976. "The Incubator Hypothesis: Evi- dence from Five SMSAs." Urban Studies 13(3):325-31. Lerman, S. R. 1977. "Location, Housing, Automobile Ownership, and the Mode to Work: Ajoint Choice Model." Transportation Research Board Record 610:6-11. Washington, D.C. Linn, Johannes F. 1983. Cities in the Developing World: Policies for Their Equitable and Efficient Growth. New York: Oxford University Press. 1 78 Bibliography Lowry, I. S. 1964. A Model of Metropolis. Santa Monica, Calif.: Rand Corporation. Lubell, Harold, and Douglas McCallum. 1978. Bogotd Urban Development and Em- ployment. Geneva: International Labour Office. Mason, C. M. 1980. "Industrial Decline in Greater Manchester, 1966-1975: A Components of Change Approach." Urban Studies 17(2):173-84. Mayo, S. K. 1975. "Local Public Goods and Residential Location: An Empirical Test of the Tiebout Hypothesis." InJohn E. Jackson, ed., Public Needs and Private Behavior in Metropolitan Areas. Cambridge, Mass.: Ballinger. McDonald, John F. 1981. "Capital-Land Substitution in Urban Housing: A Sur- vey of Empirical Estimates." Journal of Urban Economics 9(2):190-21 1. . 1984. Employment Location and Industrial Land Use in Metropolitan Chicago. Champaign, Ill.: Stipes. McFadden, Daniel. 1973. "Conditional Logit Analysis of Qualitative Choice Behavior." In Paul Zarembka, ed., Frontiers in Econometrics. New York: Academic Press. . 1974. "The Measurement of Urban Travel Demand." Journal of Public Economics 3(4):303-28. . 1976. "The Theory and Practice of Disaggregate Demand Forecasting for Various Modes of Urban Transportation." Paper presented at the Seminar on Emerging Transportation Planning Methods, Daytona Beach, Florida, Decem- ber 5-9. Washington, D.C.: U.S. Department of Transportation. Mieszkowski, Peter, and Mahlon Straszheim. 1979. Current Issues in Urban Eco- nomics. Baltimore, Md.: Johns Hopkins University Press. Mills, Edwin S. 1972. Urban Economics. Glenview, Ill.: Scott, Foresman. Mills, Edwin S., and Charles M. Becker. 1986. Studies in Indian Urban Development. New York: Oxford University Press. Mohan, Rakesh. 1980. The People of Bogotd: Who They Are, What They Earn, Where They Live. World Bank Staff Working Paper 390. Washington, D.C. . 1984. "The City Study: Understanding the Developing Metropolis." World Bank Research News 5(3), Fall/Winter. . 1986. Work, Wages, and Welfare in a Developing Metropolis: Consequences of Growth in Bogotd, Colombia. New York: Oxford University Press. Mohan, Rakesh, and Rodrigo Villamizar. 1982. "The Evolution of Land Values in the Context of Rapid Urban Growth: A Case Study of Bogota and Cali." In Matthew Cullen and Sharon Woolrey, eds., World Congress on Land Policy: Nine- teen Eighty Proceedings. Lexington, Mass.: D. C. Heath. Morawetz, David. 1976. "Elasticities of Substitution in Industry: What Do We Leam from Econometric Estimates?" World Development 4(l):11-15. Moses, Leon, and H. F. Williamson, Jr. 1967. "The Location of Economic Ac- tivity." American Economic Review 57(2):211-22. Murray, Michael P. 1988. Subsidizing Industrial Location: A Conceptual Framework with Application to Korea. World Bank Staff Occasional Paper, New Series, 3. Balti- more, Md.: Johns Hopkins University Press. Bibliography 1 79 Muth, Richard F. 1969. Cities and Housing. Chicago, Ill.: University of Chicago Press. -_____ 1971. "The Derived Demand for Urban Residential Land." Urban Studies 8(3):243-54. Oakland, W. H. 1978. "Local Taxes and Intraurban Industrial Location: A Sur- vey." In George F. Break, ed., Metropolitan Financing and Growth Management Policies. Madison: University of Wisconsin Press. Pendleton, W. C. 1963. "The Value of Highway Accessibility." Ph.D. dissertation, University of Chicago. Quigley, J. M. 1976. "Housing Demand in the Short Run: An Analysis of Polytomous Choice." Explorations in Economic Research 3(l):76-102. Reedy, D. E. 1985. "Empirical Studies of Input Substitutability in Production: A Survey of Literature." Water Supply and Urban Development Department Dis- cussion Paper 66. World Bank. Renaud, Bertrand M. 1981. National Urbanization Policy in Developing Countries. New York: Oxford University Press. Rydell, C. Peter. 1976. Measuring the Supply Response to Housing Allowance. Santa Monica, Calif.: Rand Corporation. Schmenner, Roger W. 1973. "City Taxes and Industry Location." Ph.D. disserta- tion, Yale University. -____ 1977. "Urban Industrial Location: An Evolutionary Model." Journal of Regional Science 17(2): 179-93. -_____ 1981. "The Rent Gradient for Manufacturing."Journal of Urban Economics 9(l):90-96. -_____ 1982. Making Business Location Decisions. Englewood Cliffs, NJ.: Prentice- Hall. Schmenner, Roger W.,J. C. Huber, and R. L. Cook. 1987. "Geographic Differ- ences and the Location of New Manufacturing Facilities." Journal of Urban Economics 21(l):83-104. Segal, Martin. 1960. Wages in the Metropolis. Cambridge, Mass.: Harvard Uni- versity Press. Shieh, Y. N. 1985. "A Note on Lee's Model of Intraurban Employment Loca- tion." Journal of Urban Economics 18(2):196-98. Sirmans, C. F., J. B. Kau, and C. F. Lee. 1979. "The Elasticity of Substitution in Urban Housing Production: A VES Approach." Journal of Urban Economics 6(4):407-15. Solow, Robert M. 1972. "On Equilibrium Models of Urban Location." In Michael Parkin with A. R. Nobay, eds., Essays in Modern Economics. Proceedings of the Association of University Teachers of Economics. London: Long- mans, Green. Storey, D. J. 1982. Enterpreneurship and the New Firm. London: Croom Helm. Straszheim, Mahlon R. 1974. An Econometric Analysis of the Urban Housing Market. New York: National Bureau of Economic Research. 180 Bibliography Struyk, Raymond J., and Franklin J. James. 1975. Intrametropolitan Industrial Loca- tion. Lexington, Mass: Lexington Books. Terrell, Katherine. 1979. "Workers of Cali: Who They Are, What They Do, and Where They Live." City Study Intermediate Paper 37. World Bank, Develop- ment Economics Department. Thomas, Vinod. 1985. "Evaluating Pollution Control: The Case of Sao Paulo, Brazil." Journal of Development Economics 19(1-2):133-46. Tolley, George S., and others. 1979. Urban Growth Policy in a Market Economy. New York: Academic Press. Townroe, P. M. 1979. "Employment Decentralization: Policy Instruments for Large Cities in Less Developed Countries." In D. R. Diamond and J. McLoughlin, eds., Progress in Planning. Vol. 10, pt. 2. Elmsford, N.Y.: Perga- mon Press. . 1981. "A Framework for the Study of New Industrial Locations within the City Region." Urban and Regional Report 81-7. World Bank, Development Economics Department. . 1984a. "Employment Decentralization Policy for a Major Metropolis: The Case of Sao Paulo, Brazil." Water Supply and Urban Development Depart- ment Discussion Paper 26. World Bank. . 1984b. "Spatial Policy and Metropolitan Economic Growth in Sao Paulo, Brazil." Geoforum 15(2). Valverde, Nelson. 1978. "The City Study: Available Data." Vol. 1. Urban and Regional Report 79-6. World Bank, Development Economics Department. Villamizar, Rodrigo. 1981. "Land Prices in BogotA between 1955 and 1978: A Descriptive Analysis." InJ. Vernon Henderson, ed., Research in Urban Economics. Vol. 2. Greenwich, Conn.: Jai Press. Wagner, M. Wilhelm. 1984. Market Prices and Assessed Values in the Urban Land Market in Bogota, Colombia: The Role of Quantity Premiums and Discounts. World Bank Staff Working Paper 651. Washington, D.C. Weber, Alfred. 1929. Theory of the Location of Industry. Translated by C. J. Friedrich. University of Chicago Press. World Bank. 1979. World Development Report 1979. New York: Oxford Univer- sity Press. . 1984. "Evaluating Government Intervention: Industrial Location Poli- cies," Urban Edge 8(9), November. Index Access of firms: to markets, 93, 95, Diewert, W. E., 100 n6 96; to residential areas, 95, 109, 113 Eberts, R. W., 123 Alonso, William, 68 nl Ellickson, Bryan, 9, 90-91, 95-96, Arrow, Kenneth J., 117 100 n3, n9 Employment composition, 11, 15, Birch, D. L., 5, 10 n4 39, 43 Birth rates of firms, 4, 5, 33, 43, Employment decentralization: in 51, 53, 57, 62, 67 Cali and BogotA, 5-6, 7; Bronitsky, Leonard, 37 n3 contributions to, 29, 31, 33; Burstein, N. R., 5, 100 n3, n5 effect of transport modes on, 76; effect of urban policy on, 7-8; Cameron, Gordon C., 12, 68 n3, for manufacturing, 7, 27, 68; 69 nlO, n12 pattern of, 11, 16-17, 24; trend Capital-land substitution (estimates toward, 39, 43-44, 46, 49. See of elasticity of), 117-20, 124 also Employment distribution; Carlton, Dennis W., 88 Firms; Market location; Transport Carroll, Alan, 3 modes Central business district (CBD): Employment distribution, 55, 57, employment in, 5, 17, 24, 27; as 62, 123; degree of, 39, 43, 68; incubator for small, new firms, 6; by radial sector, 44, 46, 49; outflow of industiy jobs from, spatial characteristics of, 16-17, 29, 33 24, 49, 51, 53-55, 64, 67 Centralization, 32 Employment growth rate (new and Choe, S. C., 10 nl mature firms), 51, 53 Clapp,J. M., 117 Employment location: model for Contiguity index, 62, 64, 67 retail and service sectors in BogotA, 105-06, 109, 113, 116; Death rates of firms, 4, 43, 57, 67 pattern of, 27, 29, 32-33, 43-44, Decentralization of employment. 55-57, 62, 64, 67-68; policies See Employment decentralization for, 2-3; satisfaction with, 78-80. Departamento Administrativo See also Relocation Nacional de Estadistica (DANE), 3-4 Erickson, R. A., 88 181 182 Index Fallis, G. B., 117, 118, 119 Kau,J. B., 118 Firms: births and location of, 5-6, Kemper, Peter, 3, 68 nl, n3, 88 29; characteristics of, 11, 15, 70, Koenker, Roger, 3, 125 nl, n3 72, 75-76, 85-86; distribution in Bogota and Cali of, 11, 15, 24; Lakshmanan, T. R., 106, 116 nl employment growth rate of, 51, Land price gradient, 120-21 53; factors affecting relocation of, Land use, 7, 17, 44, 46, 49, 124 80, 82, 84-85, 86, 89; location Land values, 7, 98, 120, 124. See choice for mature, 57; location also Capital-land substitution; choice of new, 31, 33, 57; Land price gradient location choice of small, 9, 57, Lau, L. J., 100 n6 68; location history of, 49, 51; Lee, C. F., 118 size of, 43. See also Access of Lee, Kyu Sik, 6, 7, 8, 9, 10 nl, firms; Birth rates of firms; Death n3, 37 nl, 68 nl, 87 n2, 100 n4 rates of firms; Manufacturing Lee, YoonJoo, 4, 33 firms; Relocation Leone, Robert, 5, 68 n3, 88 Friedman, Joseph, 90 Lerman, S. R., 90 Location choice, 9, 53-55; factors Gravity measure, 105-06 influencing, 17, 80, 82, 84-85, 86, 89, 124; model for, 89; Hansen, Walter G., 106, 116 nl satisfaction with, 78-80, 86; Hansen, Willard B., 116 nl theories for, 88. See also Firms; Hanushek, E. A., 68 nl, 88 Relocation; Site characteristics Henderson,J. Vernon, 8 Lowry, I. S., 2, 105-06 Hoover, Edgar M., 4, 5, 6, 68-69 n5, 76, 100 nlO McDonald,John F., 118, 119, 125 nl Huff, P. L., 116 ni McFadden, Daniel, 90-91 Manufacturing firms: characteristics Incubator hypothesis, 6, 8, 9, 51, in Bogoti of, 70, 72, 75-76, 85; 75, 86; test for, 53 decentralization of employment Industry groups: employment in Bogota for, 7, 27, 68; factors composition of, 11, 15; in choice of location for, 17, 80, employment concentration for, 82, 84-85, 86, 89, 124; ratio to 55-56, 57, 62, 64, 67. other firms in BogotA and Cali See also Contiguity index; of, 11, 15 Standard distance Manufacturing industry, Colombia, Ingram, Gregory K., 3, 68 nl, 99 n2 76 Institute for Office Management, 123 Market location: effect on International Center, Cali, 24, 27, employment decentralization, 29, 31, 38 n8 76-77 Market potential: model for, James, Franklin J., 3, 4, 6, 51, 106-07, 109, 113, 116 68 nl, n3, 68-69 n5, 69 nI2, Mieszkowski, Peter, 2 88, 100 nlO Mills, Edwin S., 68 nl, 88, 100 nlO Mohan, Rakesh, 37 n5, 125 n5 Kain, John F., 99 n2 Morawetz, David, 125 n2 Index 183 Moses, Leon, 6, 69 n6, 76, 80, 88 Song, B. N., 10 nl, 68 nl, 88 Multinomial logit analysis: estimation Spatial policy. See Urban policy using, 91-93, 95-96; model for, Standard distance: as measure of 90-91, 96-99 degree of industry concentration, Murray, Michael P., 9 56-57, 62, 67, 68. See also Muth, Richard, 7, 99 n2, 118, 119, Contiguity index 125 nl Straszheim, Mahlon, 2, 99 n2 Struyk, RavmondJ., 3, 4, 6, 51, Pendleton, W. C., 116 nl 68 nl, n3, 68-69 n5, 69 n12, 88, Policy decisions: on employment 100 nlO location, 2-3; minimizing social Subsidies, effect of, 9-10 cost, 9-10. See also Urban policy Production technology (effect on Terrell, Katherine, 37 n5 relocation decision of), 80, 86 Tolley, George S., 8 Quigley, John M., 90, 99 n2 Transport modes (effect on employment decentralization of), Reedy, D. E., 118 76, 86 Relocation: as contribution to decentralization, 29; factors related Urban development: in developing to, 80, 82, 84-85, 86, 89; of countries, 1-2; in Korea, 9 firms, 4-5, 6, 7, 43, 68 Urban policy: and effect on Rydell, C. Peter, 125 nl development, 2, 7-10; in United States, 7-8 Satisfaction with location choice, 78-80, 86 Valverde, Nelson, 4, 33 Schmenner, Roger W., 4, 68 n3, Vernon, Raymond, 4, 5, 6, 68-69 76, 80, 87, 88, 100 nlO, 121 n5, 76, 100 nlO Segal, Martin, 123 Villamizar, Rodrigo, 125 n5 Service sector: employment in CBDS for, 29; location pattern of firms Wage gradient, 122-23, 124 in, 27. See also Employment Wagner, M. Wilhelm, 125 n5 location Wasylenko, M., 88 Sirmans, C. F., 118 Weber, Alfred, 99 nl Site characteristics (as factor in Williamson, H. F., Jr., 6, 69 n6, location choice), 76-80, 82, 76, 80, 88 84-85, 120-21 World Bank, 2 Social cost, minimizing of, 9-10 Solow, Robert M., 10, 88, 100 n7, nlO Yotopoulos, P. A., 100 n6 The complete backlist of World Bank publications is shown in the annual Index of Publications, which is of value principally to libararies and institutional purchasers. The latest edition is available free of charge from Publications Sales Unit, The World Bank, 1818 H Street, N. W., Washington, D.C. 20433, U.S.A., or from Publications, The World Bank, 66, avenue d'1ena, 75116 Paris, France. The World Bank This book is the first to establish in detail, for a large city in the developing world, the trends in the location patterns of jobs and the factors that determine the location choices of individual firms Governments of developing countries have often tried to control rapid urban growth through policies that influence the location of firms and thus of employment Such policies tend to be ineffi- cient and costly, however, because they attempt to reverse trends that are poorly understood and well entrenched through the operations of markets The empirical findings in this book offer insight into the probable effects of location policies They should also help in devising investment programs for urban housing and transport that reflect the location dynamics of the demand for these services Kyu Sik Lee documents changing location patterns of employment in Bogota and Cali with the use of several large files of data on firms and households, he analyzes the components (such as the location of newly established firms and those that moved or failed) to predict future trends, and he presents estimated econometric models to explain location choices of different types of manufac- turing and commercial firms The models mclude an extension of the bid-rent theory to the multinomial logit specification for manufacturing firms' location choices, gravity models for trade and service firms, and aggregate models to determine the elasticity of substitution between land and other factors of production and to estimate land price and wage gradients The empirical findings confirm that trends widely observed in developed countries, such as the decentralization of employment in large cities, also occur in developing countries This suggests that the study's findings in Colombia are applicable to cities in other countries Kyu Sik Lee is senior economist in the Infrastructure and Urban Develop- ment Department of the World Bank Of related interest from Oxford and The World Bank Rakesh Mohan, Work, Wages, and Welfare in a Developing Metropolis Conse- quernes of Growth in Bogota, Colombia Oxford UnrvosO Prtss W- 51078166 Coverdesion byif &asef