A WORLD BANK COUNTRY STUDY PU B-3586 INDONESIA Wages and Employment 0 7 7-0i1 0035 hl.)Ke>ly Bogoasir H f1-001. A WORLD BANK COUNTRY STUDY INDONESIA Wages and Employment The World Bank Washington, D.C., U.S.A. Copyright ©) 1985 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 Manufactured in the United States of America First printing January 1985 World Bank Country Studies are reports originally prepared for internal use as part of the continuing analysis by the Bank of the economic and related conditions of its developing member countries and of its dialogues with the governments. Some of the reports are published informally with the least possible delay for the use of govern- ments and the academic, business and financial, and development communities. Thus, the typescript has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. The publication is supplied at a token charge to defray part of the cost of manufacture and distribution. The full range of World Bank publications is described in the Catalog of World Bank Publications; the continuing research program of the Bank is outlined in World Bank Research Program: Abstracts of Current Studies. Both booklets are updated annually; the most recent edition of each is available without charge from World Bank Publications in either Washington or Paris (see the back cover for addresses). Library of Congress Cataloging in Publication Data Lluch, Constantino. Indonesia--wages and employment. (A World Bank country study) Authors for the study are Constantino Lluch and Dipak Mazumdar. Bibliography: p. 1. Labor supply--Indonesia. 2. Employment forecasting --Indonesia. 3. Wages--Indonesia. 4. Indonesia--Econom- ic conditions--1945- . I. Mazumdar, Dipak, 1932- II. World Bank. III. Title. IV. Series. HD5824.A6L58 1985 331.12'5'09598 84-27133 ISBN 0-8213-0484-4 - iii - CURRENCY EQUIVALENTS Before November 15, 1978 US$1.00 = Rp 415 Annual Averages 1979-83 1979 US$1.00 = Rp 623 1980 US$1.00 = Rp 627 1981 US$1.00 = Rp 632 1982 US$1.00 = Rp 661 After March 30, 1983 US$1.00 = Rp 970 FISCAL YEAR Government - April 1 to March 31 Bank Indonesia - April 1 to March 31 State Banks - January 1 to December 31 - v Table of Contents Page No. SUMMARY AND CONCLUSIONS xiii 1. STRUCTURE OF THE LABOR FORCE, EMPLOYMENT AND LABOR INCOMES IN 1977 ....... ........ ..... ...... ... 1 Educational Attainment of the Population ................. 1 Participation Rates ... .. .. ...... ............... .... . 2 Males ........................................... 0.000... 5 Females .... a ....................................... o........ .. 6 In Sum ................................................. 7 Measured Unemployment in Indonesia ........ .......*....... 7 Duration of Unemployment ...... ... ... ... ......... 8 Underutilization of Urban Youth ...... ...... .......... .... 9 Employed Population ........ ........... ............ . 10 Rural/Urban Earnings Differential ....... ........ .......... 13 Appendix A: On Seasonal Changes in the Labor Force Participation Rate., .... ................ 14 Appendix B: Data ...... .. ................... ......... 16 2. TRENDS IN THE ECONOMIC DEVELOPMENT OF INDONESIA, 1961-1980: EMPLOYMENT, OUTPUT, RELATIVE PRICES AND WAGES ............ 29 Growth and Industrial Composition of Employment ....... 29 Changes in the Sectoral Composition of Employment, 1961-1971, 1971-1980 ....... o..................... ............ 33 Growth in Output Relative to Employment ............... 39 Relative Prices ....o.................. ..... o........ s.......... ... 40 GDP Deflators ... .............................................. 40 Wholesale Price Indexes (WPI)....... ................ 41 Retail Prices ..... ................................. 41 Summary . . o ............... ..................... . 43 Trends in Real Wages..... ....... ... ... ... ... 43 Agricultural Wages in Selected Villages in West Java (Makali Series) ........................... 43 The Agriculture Wage Series of the Central Bureau of Statistics ................................................... 47 The Wage Series from the Padat Karya Program.......... 48 Earnings in Estates...... ...........oo ............ 50 Wages in the Urban Construction Sector ................ 51 Aggregate Trends in Indonesian Development: An Overview .................. 52 This report is based on the findings of two missions to Indonesia, in March 1980 and June 1982. The main authors are Constantino Lluch and Dipak Mazumdar. Roger Key contributed to the first mission. - vi - Appendix A: On the Estimation of the Labor Force in 1971 and 1980 ................................................... 55 Appendix B: Data ................ ..*... 0.0 ........................ 58 3. THE INDONESIAN LABOR MARKETS: AN INTERPRETATION . . 72 Rural Labor Market .......... ....... ..... .... ... 72 Labor Utilization in Javanese Agriculture............... 74 The Multiplicity of Market Activities ............. ..... 75 Interclass Differences in Market Participation ....... 77 Seasonal Variations in Atiivity ...... ........... 78 Patterns of Returns to Labor in the Rural Sector ......... 79 "Surplus" Labor and Seasonality ........................ 82 Technical Change, Commercialization and "Surplus" Labor.. 84 Conclusions on Employment Growth and the Rural Labor Market ..................... .................................. 86 The Urban Labor Market ................................ 86 The Importance of Temporary Migration .... ................. 88 Difference in Earnings of Temporary and Permanent Migrants .............................................. 89 The Wage Ladder in Indonesian Manufacturing Enterprises ............ ............. ............ 89 The Census of Manufacturing, 1974 ....................... 90 Manning's Study of Indonesia Manufacturing Wages ...... 91 The Mission's Field Data..... ........... ....... ..... 95 Conclusions on the Working of Labor Markets ............. .. 98 The Wage Ladder. ....... o ... . ................ oo .... ..... . 98 Surplus Labor and the Elastic Supply of Labor ......... 100 The Meaning of Surplus Labor........................ 100 Elastic Supply of Labor ....... .................... 100 Extent of the "Organized Sector" in the Indonesian Labor Market................ .......... 101 Appendix A: Models of the Rural Labor Markets ... .. 103 The Stylized Facts .................................... 103 A Neo-Classical Model ........ o ...................... .o.. . 104 The Segmented Labor Market Model...................... 106 Testing the Alternative Models: (i) Predictions ....... 107 Predictions (A) ................... 0... 00 ................... ........ 107 The Rice Labor Market with Two Classes of Workers.. ... 108 Predictions (B)....................................... 109 Testing the Alternative Models: (ii) Premises ......... 110 Notes on the Segmented Labor Market Model ....... 111 The Transition from a Surplus Labor to A Neo-classical Model ............................. 114 Appendix B: Shadow Wages inJava......................... 116 The Opportunity Cost of Labor................ 116 The Competitive Model .....*** ................... 116 The Modification of the Competitive Model....*......... 117 The Ratio of the Opportunity Cost of Labor to the Agricultural Wage ................ o............ 118 - vii - The Shadow Wage of Rural Labor ............ ........ 120 The Value of M/W at the Peak Season ................... 122 4. PROSPECTS FOR EMPLOYMENT, OUTPUT AND WAGES ................... 123 Prospects for Real Wages and Employment ............... 125 The Overall Growth Record in Indonesia, Japan and Korea ............................... ........................... 126 The Period of Transition in Japan .................. .... 128 The Korean Record .......... 133 On the Extensive and Intensive Margins of Cultivation in Indonesia ... * ....... ............. .. 134 Long Term Employment Prospects Under Present Trends ... 136 ANNEX A: Long Run Employment Problems in Indonesia: Alternative Views, Implications and Tests ........... 139 Full Employment ....... ................ . 140 Accumulation of Skills .......... ... 144 Sectoral Disaggregation ................. ......... 147 Summary ......................................... 150 Surplus Labor ..................................... 000-00.... 151 Conclusions L ........................................ 154 ANNEX B: References .. ...... .................................. . 156 - viii - LIST OF TABLES Page No. CHAPTER 1 1.1 Percentages of the Population with More Than a Specified Educational Level by Age, Sex and Location, 1977 ...... 2 1.2 Labor Force Participation.Rates by Age, Sex and Location, Indonesia 1977-78 .............. ....... **** 3 1.3 Urban Male Unemployment Rates by Age and Education, 1977... 8 1.4 Rate of Nonemployment of Youth in Urban Areas, 1977 ........ 9 1.5 Proportion of Each Age-Sex Group Which Is Not Employed by Educational Levels, Urban Indonesia, 1977 ................ 10 1.6 Percentage of the Total Employed in Each Cell in the Tertiary Sector, 1977 .................. ............................ . 11 1.7 Relative Mean Earnings in the Urban Areas (Rural = 100).... 13 APPENDIX A IA.l Seasonal Variation in Labor Force Participation Rates, Indonesia, First Quarter 1976 through Fourth Quarter 1978 ..................... . 15 APPENDIX B IB.1 Population (in millions) by Region and Location, 1977 ..... 16 IB.2 Age and Sex Distribution of the Population by Region, 1977 17 IB.3 Age and Sex Distribution of the Population by Region a nd Location ............................................. 18 IB.4 Educational Attainment of the Population by Age Groups 1977 Urban ............ .... -.6 ................ ............... ......... 19 tB.5 Education Attainment of the Population by Age Groups, 1977 Rural ............................................... 20 IB.6 Labor Force Participation Rates by Age, Sex and Location, Indonesia 1977-78 ..0*.0*.*..Oo.ooo..O..................................... 21 IB.7 Urban Labor Force Participation Rates by Age and Educational Attairment, Indonesia ........ ...... ............ ........ 22 IB.8 Rural Labor Force Participation Rates by Age and Educational Attairment, Indonesia ... ... 0*.......................................... 23 IB.9 Measured Open Unemployment Rates by Age, Sex and Location, Indonesia 1977-78 . ........ ...................... O...... . 24 IB.10 Youth Unemployment Rates in Urban Areas by Sex and Educational Attainment in Indonesia 1977 ..... ............ 25 IB.ll Employment by Main Industry, Hours worked, Sex and Location, Indonesia 1977 ............. ... 26 IB.12 Summary Statistics for the Distribution of Monthly Income of Employees from Main Job by Main Industry, Sex and Location, Indonesia 1977 ................. ............ 27 IB.13 Employment by Main Employment Status, Sex and Location, Indonesia 1977 ........................................... 28 - ix - CHAPTER 2 2.1 Growth of Population and Population Indonesia and Java 1961-71, 1971-80 ...................... 30 2.2 Labor Force Participation Rates, Indonesia and Java 1961-71, 1976-80 .................... ............. . ................... 31 2.3 Rate of Growth of Employment by Main Industry, Indonesia and Java 1961-71 and 1971-80 ................... 34 2.4 Increase in Employment and its Distribution by Sex Urban-Rural Location and Main Industry, Indonesia and Java, 1961-71, 1971-80 .................... 36 2.5 Rate of Growth of Employment by Main Industry and by Sex in Rural Areas Indonesia and Java, 1961-1971, 1971-1980 .................. .. .............. 37 2.6 Rate of Growth of Employment by Main Industry and Sex in Urban Areas, Indonesia and Java 1961-1971, 1971-1980 ... ............................................... . 38 2.7 Yearly Rates of Output Growth, By Main Industry and Total Elasticity of Employment with Respect to Output, Indonesia, 1961-1971, 1971-1980 . ......................... 39 2.8 GDP Price Deflators, Relative to Agriculture, Indonesia, 1961-1981 ...................................................... 41 2.9 Relative Price Ratios From Wholesale Price Indexes (1973 = 100), Indonesia, 1971-81 ... .............. 42 2.10 Rice Price, Relative to Kerosene and Textiles, in Jakarta and the Rural Markets of Java and Madura ...... 43 2.11 Average Real Wages per day, by Season, West Java Villages (Kg of milled rice) .................... * ......... 47 2.12 Trends in Agricultural Wage Rates, 1976-81, in Four Provinces of Java (Kg of milled rice per day) ............ 49 2.13 Wages Paid to Men and Women Workers in the Padat Karya Program (Rp per day) .................... 50 2.14 Real Monthly Earnings of Permanent Estate Laborers, Java and Sumatra, 1966-80 (in 1971 Rp) ............... .... 52 APPENDIX A to CHAPTER 2 II.A.1 Jones Estimates of the 1971 Labor Force Through Direct Adjustment ................. .............. 56 IIA.2 Estimate of the 1980 Labor Force through Direct Adjustment (millions of people) ................. 57 APPENDIX B TO CHAPTER 2 IIB.1 Estimates of Population, Population of Working Age, Labor Force and Employment, by Sex, Location and Region, 1961, 1971, 1980 .............................. 58 IIB.2 Labor Force Participation Rates by Age, Sex and Location - Indonesia 1961, 1971, 1976-80 .......... 59 IIB.3 Employment Rate by Sex and Location Indonesia 1961, 1971, 1976-80 ......................... 60 x IIB.4 Percentage Distribution of Employment by Sex, Main Industry and Location, Indonesia, 1961, 1971, 1978, 1979, 1980 ......................................................... 61 IIB.5 Percentage Distribution of Employment by Sex, Main Industry and Location, Java, 1961, 1971, 1978, 1979, 1980 ...... 62 IIB.6 Adjusted Percentage Distribution of Employment and Its Annual Rate of Change Indonesia and Java 1961, 1971, 1980 ............... 63 IIB.7 Adjusted Percentage Distribution of Employment and Its Annual Rate of Change, Urban Male Indonesia and Java 1961, 1971, 1980 ................... 64 IIB.8 Adjusted Percentage Distribution of Employment and Its Annual Rate of Change, Urban Female Indonesia and Java 1961, 1971, 1980.................. 65 IIB.9 Adjusted Percentage Distribution of Employment and Its Annual Rate of Change, Rural Male Indonesia and Java 1961, 1971, 1980 ................... 66 IIB.10 Adjusted Percentage Distribution of Employment and Its Annual Rate of Change, Rural Female Indonesia and Java 1961, 1971, 1980 ................... 67 IIB.ll Cost of Living Index In Jakarta: Food and Clothing (April 77/March 78 = 100) ........... ........... 68 IIB.12 Rural and Urban Prices of Rice, Kerosene and Textiles 1971-81 ........................ ......................0 .... 69 IIB.13 Daily Wages for Unskilled and Skilled Workers in the Construction Industry in 1971 Rupiah, 1971-78 ..... 70 IIB.14 Construction Wages for Daily Paid Unskilled and Skilled Workers by Province, 1976-79 - Base 1976 I ............ 71 CHAPTER 3 3.1 Percentage of Household Working Hours, 1975-76.... ......... 76 3.2 Distribution of Permanent and Temporary Migrants by Employment in Formal and Informal Sectors (14 survey villages, 1973) ................................ . .................. 89 3.3 Mean Hourly Wages by Several Firm Characteristics (All Firms and Weaving, Operator 1) ....... #............... 92 3.4 Hourly Earnings of Operatives (Mean) (Rupiah) .............. 93 3.5 The Indonesian Wage Ladder, 1976 .................. ......... 99 CHAPTER 4 4.1 Projected Employment Growth By Sectors...... ............124 4.2 Output and Employment Growth By Sector Indonesia Japan, and Korea ..... ................................... 127 4.3 Marginal Employment Shares during the Japanese Transition.. 129 4.4 Labor Use in Japanese Agriculture .......................... 132 4.5 Sectoral Allocation of Additional Employment in Korea ...... 134 - xi - 4.6 Ratio of Farm Land to Agricultural Employment, in Indonesia, Japan, and Korea ........... . ............ .*0*0.. 135 4.7 Paddy Yield per Hectare and Labor Input per Hectare in Asian Agriculture.*.**,.,#*#..,. ...... .. 136 ANNEX A Al Yearly Output and Employment Growth Rates, 1971-1980 ....... 139 A2 Yearly Percentage Increase in the Average Rate Wage for Different Rates of Substitution and Technical Progress.,.*.. ......... 143 A3 Rate of Growth of Skilled Wage Times their Share in the Labor Bill in (3) ........................... . 146 A4 Yearly Percentage Increase in Wages in Terms of Sectoral Output for Different Rates of Substitution and Technical Progress .................................... 149 - xii - LIST OF FIGURES CHAPTER 1 1.1 Male/Female Labor Force Participation Rates, Indonesia, 1977/78 977/8........ ..4 CHAPTER 2 2.1 Changes in Peak Season Real Wages. 1966/67-1979/80 for Male (Hoeing) and Female (Weeding) Labor in the Wet Season in Six Villages of West Java (Kg of milled rice per day) ................................................................ 45 CHAPTER 3 3.1 A Neoclassicial Model of the Rural Labor Market ............ 105 3.2 The Rice Labor Market with Two Classes of Workers .......... 107 3.3 The Wage Efficiency Relationship ........................... 112 CHAPTER 4 4.1 Real Wages in Agriculture (Wa) and Non-agriculture (Wn) [Minani (1967), p. 1974 (1934-36 = 100) .................. 130 ANNEX A 1. Moving Equilibrium in the Labor Market: Full Employment in a One Sector Model .............................. ..... . 142 - xiii - SUMMARY AND CONCLUSIONS Introduction 1. This report is part of the Bank's on-going assessment of wages and employment in Indonesia. The last specialized report 1/ examined labor markets and income distribution through 1976. The present report, focusses solely on labor market conditions (resource constraints prevent an in-depth review of the associated factors of poverty and income distribution). The data base includes the quarterly labor force surveys of 1977, 1978 and 1979, as well as the preliminary results of the 1980 Population Census. According to the latter, Indonesia's population grew by 2.3% per year during the seventies instead of the predicted 2%. Consequently additions to the labor force are expected to total some fifteen million people over the next decade, as opposed to twenty-five million in the last two decades. It is thus imperative to monitor and evaluate labor markets and the factors which affect them on a regular basis. 2. Another factor which adds to this urgency is the expected deceleration of output growth in the eighties. How employment and incomes will be affected by a rapidly expanding supply of labor, combined with a slower growth in output and vary with alternative assumptions about how labor markets work. In turn, such alternative views have very different inplications for policy formulation. A primary objective of this report is to point out these implications and to promote further study of labor markets in Indonesia, including a fresh examination of labor use and labor incomes, particularly in rural areas of Java. 3. The previous report took the view that labor markets in Indonesia work efficiently. Given observed wage variations (over time and across regions) and labor mobility, it concluded that, on the whole, supply and demand forces provide an adequate explanation of the level and the structure of wages and employment, and thus, there was no long run, or structural unemployment in Indonesia. Those who supply labor services were bidding wages down to the level where everyone was employed, at a wage which reflected labor productivity and the cost of effort, at the margin. Wages were low, but this was the result of a lack of complementary inputs (capital, skills) rather than the market's failure to reflect the "opportunity cost" of labor - the wage that workers, could get at the margin, in alternative occupations. In addition, the previous report also suggested that observed wage differentials probably resulted from differences in the quality of labor supplied. 4. In the above context, there would be no employment problem per se in Indonesia, and no need for separate policies specifically aimed at employment creation. Also, public investment projects (particularly in rural Java) should be appraised using a price for labor close to the going market wage, i.e. the ratio of shadow to market wages would be near unity. 1/ World Bank (1980), 2788-IND. - xiv - 5. This report takes a different view. First, it casts doubt on the explanation that wage differentials in urban labor markets reflect differences in the quality of labor. A wage ladder is constructed for unskilled labor, and observed wage differences appear too high to be explained by differences in quality within the unskilled labor category. Second, it suggests that low participation rates by young and old males in urban areas, coupled with high unemployment rates for urban youth, add up to a substantial underutilization of young people, particularly the educated, and cannot be explained by models of unemployment highlighting transitional job search. Third, the report argues that social and economic forces, which do not promote a clearing of the labor market, play a large role in the determination of wages. The variability of wages in terms of rice, over time or across villages occurs within a narrow band which is stable over time. The constancy of the agricultural wage within this band in Indonesia is similar to stable agricultural wages observed in Japan and Korea before the exhaustion of surplus rural labor. This report proposes, as a working hypothesis, that labor is in surplus in many Javanese villages: at the going wages there are more people able and willing to work in rice cultivation than can be employed, both during peak and off seasons. In the rural rice labor market, as in urban labor markets, there are mechanisms that prevent such surplus labor from bidding down wages. 6. These three characteristics of Indonesian labor markets have important policy consequences. They highlight the need for policies which address the anticipated fast growth of the labor force and the prevalence of relatively low labor productivity. This report thus points to the need to consider employment creation as a separate policy objective (as opposed to subsuming it under policies promoting maximal output growth). Also, it follows that the appraisal of public investment projects (particularly in rural Java) should not price unskilled labor at the going market wage in rice cultivation, but at the lower return observed in marginal activities down the economic ladder. 7. The evidence supporting this report's interpretation of the labor market comes mainly from village studies, and the labor force surveys. Village studies are of central importance to advance knowledge on the evaluation of the Javanese economy, in particular. The report attempts to use the evidence from those studies to interpret the overall trends in employment and output from aggregate data. This attempt should be pursued further. Unemployment, Participation Rates and Education 8. Chapter 1 provides an overview of open (i.e., measured) unemployment in Indonesia. As in many developing countries, this is a problem in urban areas, particularly among young people. Measured unemployment rates are 3 to 4 times higher in the urban than in the rural sector. Within the urban sector it is 25% for males aged 15-19 and 18% for males aged 20-24 (compared to an overall male urban average of about 7%). - xv - 9. However, measured unemployment does not adequately reflect the real difficulties encountered by urban youth in Indonesia in getting jobs. According to the labor force surveys, among those who declared themselves to be unemployed, the duration of unemployment was not long, mostly one month or less. In contrast, the sluggishness of entry into the urban labor market is revealed quite strikingly by the high rates of nonparticipation among the urban youth. It would then appear that young people do not actively search for jobs until they feel that an acceptable job is likely to be available in the near future. Participation rates were 34% for urban males aged 15-19 and 75% for those aged 20-24. For females the corresponding rates were 23% and 29% respectively - substantially lower than what is observed in the rural sector. The magnitude of the problem can be understood by noting the statistic that even in the age group 20-24, 40% of urban males, and 76% of urban females were not employed (i.e. either unemployed or not in the labor force). SAKERNAS data on the occupation of nonparticipants show that as much as 20% of urban males in this age group said that they were attending school, while the official tertiary enrollment rate at this date (1977/78) was only 2%. Thus it appeared that many nonparticipants were really discouraged workers who were using nonformal education to excuse their inactivity. 10. A question of some importance is the relationship of underutilization of urban youth to educational expansion in Indonesia. It is clear from the data presented in Chapter 1, that the incidence of both open unemployment and nonparticipation increases with educational levels, after controlling for the age group. The participation rate of urban males between 20-24, for example, is around 60% for those with secondary education and around 90% for those with primary education. It should be remembered in this connection, that secondary schooling should be finished well before the age of 20, and in fact, that such a late age for finishing school is itself an aspect of the lack of job opportunities for young people. Figures on open unemployment rates reveal that a sizeable proportion of the Iabor force with secondary education is still searching for a job in their late twenties. At the same time, there are shortages for specific categories of educated labor e.g., technical workers needed in agriculture. Evidently, more research is needed to identify areas of shortages and potential surplus within the educational spectrum. Overall Trends in Output, Employment and Wages, 1961-80 11. The report contains comprehensive aggregate data on the growth of the Indonesian economy over the periods 1961-71 and 1971-80. Particular emphasis is given to the rate of growth of employment (measured by the number of employed persons). Using the best available information, this is estimated at 2.9% per annum during 1971-1980, up from 2.4% during the sixties. These rates fall within the lower end of plausible growth rates given in the previous employment report by the World Bank. In contrast, previous Bank - xvi - economic reports have taken the highest rate in that interval, 4.7%. While the estimate given here is subject to revision as information from the 1980 Demographic Census is more fully analyzed, it would seem that 4.7% is implausibly high. The acceleration in employment (and labor force) growth during the 1970s is associated with an increase in the proportion of the working age population, rather than an increase in labor force participation rates. 12. The growth of output (7.3% per annum) was well above this revised growth rate for employment throughout the last decade. But an examination of real wage trends in the few organized sectors for which data could be obtained - agriculture, plantation and construction - suggests that real wage for unskilled workers were constant over much of this period, despite rapid output growth. 13. The report focusses on the issue of rapid growth in output at constant wages as one of the major characteristics of Indonesia's macro- economic expansion. One possible explanation of this phenomenon within the framework of competitive labor markets and market clearing wages is that the growth of "true" employment (employment measured in terms of efficiency units, after adjustments for skill accumulation, rather than in terms of employed persons) exceeds measured employment growth. Another is that labor saving technical progress has been sufficient to permit fast output growth with constant wages. These possibilities are examined in Annex A to this report. The conclusion from the formal analysis on the first point is that allowance for skill accumulation does not change the overall picture. The numerical results derived from the growth model with technical progress indicate the specific assumptions about the parameter values which would make the observed rate of output growth compatible with constant wages over time. 14. The report proposes that a framework based on the availability of abundant labor at the going wage rates in the major sectors of the economy (including paddy cultivation) would be useful for interpreting the observed trends in productivity and wages in the economy. But, unlike some early versions of the surplus labor model, we cannot accept the hypothesis that labor is in perfectly elastic supply at a wage geared to the average productivity of labor in agriculture. Within the agricultural sector, output grew at an annual rate of 3.6%, significantly above the rate of growth of 1/ See World Bank (1979), 2093-IND and World Bank (1980), 2788-IND. In World Bank (1981), 3307-IND, the quoted employment growth rate is 3.3% per year. In World Bank (1982) 3795-IND, page 95, the rate estimated is 2.65%. - xvii - employment in this sector (1% per annum). -/ The report uses some intensive micro-studies of Javanese villages to examine rural labor markets, and suggests some hypotheses about the determination of agricultural wages. This attempt to link a micro view of the labor market (in Chapter 3) with the macro picture of the economy (in Chapter 2) is only a preliminary effort. Much more work is required on village studies and data collection. Some of the more specific points for further investigation are indicated at the end of this summary. The Working of Indonesian Labor Markets: An Interpretation 15. The two major stylized facts highlighted by the village micro studies in Java are the following: (a) The number of manhours spent in the major agricultural labor market - rice - is only a small part of the total labor time supplied to the market even by landless households. 2I Typically, the proportion of total labor time utilized in agricultural wage labor is 30-40% for landless workers. The rest of the labor time is devoted to a variety of activities in trade, services, handicrafts, and sometimes to nonagricultural wage labor outside the villages. (b) The returns to labor (per manhour) are substantially lower in activities outside the rice labor market. Field studies suggest that this could be on the order of 30% of the prevailing wage rate in rice for similar age/sex groups. Because of the lower return to labor in miscellaneous (largely selfemployed) activities, they are categorized in the following pages as the "marginal sector." Obviously, a larger proportion of time is devoted to marginal 1/ Excluding mining from consideration, the overall rates of growth of output and employment respectively were as follows: agriculture (3.5, 1.0); secondary (12.9, 4.9); tertiary (8.3, 6.1). The secondary sector includes manufacturing, construction, transport and public utilities. The tertiary sector includes trade and services. The percentages of total employment at the end of the period in the three sectors were 56, 15 and 29 respectively. In the World Bank Report (1981) 3182-IND, Annex 1 p. 35 employment in large and medium manufacturing firms is reported to have grown between 6% and 12% over the 1970s. But these firms covered only 13% of total manufacturing employment in 1974/75. 2/ Here, and in the rest of the report "rice" is used as a generic term for the productive agricultural sector which makes extensive use of wage labor. It includes plantation crops, as well as important nonrice crops in the peasant economy like sugarcane. - xviii - activities in the slack season; but even in the busy season of paddy cultivation these activities consume a significant portion of labor time for adult men and women workers. This is partly because rice labor markets are specific to individual villages, and in any particular village, peak demand for labor for rice cultivation is confined to days rather than months. More work is required on the seasonal difference in returns to labor between the rice sector and marginal activities. Not all the field studies report the extent of the differential by season, but the differential seer7 to appear only in studies which are confined to the peak season. - 16. Surplus labor can thus be said to exist in rural Java in the limited and particular sense that there is a pool of labor units (rather than laborers) in marginal activities with earnings per hour significantly less than the rice wage rate. The macro-economic significance of this labor market model is that this pool is potentially available to the rice sector at the prevailing (higher) rice wage. Given this perfectly elastic supply of labor to the rice sector, output per worker can increase in rice over a period of time without any pressure on wage rates even in the absence of labor saving technical changes. The question arises: why do profit seeking landowners pay labor working in their fields a higher wage than workers earn in the marginal activities? The answer probably lies in the wage efficiency relationship i.e., the incentive effect of high wages on the intensity of effort of workers. Starting from a low return to self employed labor in marginal activities, a worker's efficiency increases more than proportionately as the wage offered increases, up to a point of inflexion, after which efficiency increases less than proportionately. Thus, there is a wage level - higher than the return to labor in marginal activities - at which wage cost per unit of labor (in efficiency units) is minimized from the point of view of employers. No profit maximizing employer will offer a wage lower than this, even if there is an abundant supply of labor at this wage. It might appear from this argument that the higher wage is in fact established to attract more efficient labor. But the crucial point is that the direction of causation runs from wage to efficiency, rather than the reverse. At a higher wage (up to the point of inflexion) more or less everybody who is offered a job performs at a proportionately higher efficiency. Since there is an abundant supply of labor at the floor wage which (minimizes wage cost), employers must select some workers more favorably than others. The field studies in rural Java have consistently mentioned the importance of the social network system which establishes preferential hiring rules for workers utilized in the rice sector. 17. It has already been mentioned that rural labor markets in Java (as in other Asian countries) are village specific. Thus, the efficiency wage which establishes the floor wage in agricultural operations, is tied to the 1/ In most village economies there will, in part, be a range of earnings in both rice and marginal activities. The two-tier model with different mean earnings is a stylized fact which underlines the importance of marginal activities as a reservoir of "surplus" labor. - xix - norm for the particular village and reflects its economic conditions. Intervillage variations in agricultural wages for the same season or occu- pation are observed in the rural sector. Thus, the constancy of the real wage over time means that the band of wage rates for a particular occupation has not moved perceptibly upwards, in spite of the increase in output per worker. We do not expect an upward movement of the band until the demand for agricultural labor growth at a sufficiently high rate, relative to the supply, to cause a signficant reduction in the pool of labor working in marginal activities. 18. The urban and rural labor markets are linked by two different types of migrants: the circular migrants who move temporarily in and out of towns without their families; and more permanent migrants who come to the urban areas with their families. The circular migrants are found largely in the informal sector of the urban labor market and, although data on earnings in this sector are very poor, we can expect circular migration to keep returns to labor in the urban informal sector and in the rural marginal activities to be reasonably close, after allowing for cost-of-living differences and transportation costs. Outside the informal sector, the wage-efficiency relationship can be expected to hold for urban wage employment as much as for hired labor in agriculture. In fact, a reasonable hypothesis is that the relationship between wages and efficiency will be stronger, the larger the employment size and the more complex the organization of the hiring unit. It has been observed in many developing countries (including Japan during its period of development) that unskilled manufacturing wages increase with the size of the firm, even in the absence of trade unions or government legislation. Indonesia is no different from the general pattern of urban wage behavior. Evidence is presented in Chapter 3 that there is a wage ladder for unskilled workers in urban manufacturing with nonmechanized small firms at the lower rung, and the "joint venture" firms or multinationals, at the top. The wage spread between these two is as much as 250%. The Shadow Wage 19. The above labor market model implies that the shadow wage for rural projects should be calculated on the basis of the opportunity cost of employing a unit of unskilled labor, namely the return to labor in marginal rural activities. Field studies suggest that in Java the latter can be as low as one third of the agricultural rice wage. Taking into account the social cost of increased consumption by the representative rural household supplying the labor, the shadow wage was calculated to be 28% of the agricultural wage. As discussed below, and in the body of the text, this statement has to be qualified by the fact that seasonal earnings data are still very scant in Indonesia. Prospects for Employment and Wages 20. On the basis of historical trends the report projects the labor force to grow by some 2.6% per annum through 1990. Will this incremental labor be absorbed by increased demand over the next decade, and at what wage level? This is the central question regarding the prospects for employment - xx - growth in Indonesia. This report has utilized the concept of the turning point (related to the exhaustion of a "surplus" of labor, as defined in p.11) in examining the prospects for Indonesian employment. 21. Historical evidence for two Asian rice economies - Japan and Korea - reviewed in Chapter 4 shows that both economies had a period of sustained growth of output per worker in agriculture, with little or no growth in real wages in this sector. This period of growth at nearly constant wages (in agriculture) was prolonged in the case of Japan (50 years or more) with output per worker increasing only at a moderate rate, and substantially shorter for Korea with much faster growth of agricultural productivity during the period of transition to the turning point. But in both cases, the turning point is identified as an historical fact, a clearly recognizable "kink" in the time trend of real agriculture wages. After a relatively long period during which real wages grew very slowly, wages increased six to ten times faster than previously observed. 22. The welfare significance of the turning point is that before this point is reached, the share of output going to wages falls consistently because labor productivity is growing faster than wages. Since mechanization is not a major factor in Asian agriculture before the turning point, much of the difference between labor productivity and wages is reflected in an increase in the share of rents and profits. A second important aspect of the "trickling down" of the increase in agricultural productivity is whether the economy is on a growth path which may soon reach the turning point. If the economy is in such a period of transition, a larger proportion of the labor of workers who depend on agriculture as their primary occupation, will be diverted from marginal activities to agriculture. With hourly earnings higher in the latter, income per worker will increase over time even though wage rates are constant in agriculture. Also in the period of transition, both countries surveyed in Chapter 4 - Japan and Korea - saw a strong diversion of labor from agriculture to manufacturing where wages were higher (with the absolute number of workers in agriculture actually declining during parts of the period of transition). Such reallocation of labor also increases earnings per worker over time for the economy as a whole, up to the turning point, even with constant real wage rates in the individual sectors. 23. The evidence presented in this report suggests that Indonesia has not yet passed such a turning point. There is no evidence of any substantial upward pressure on real wages, in spite of a significant increase in output per worker in agriculture over the last decade. But is there any evidence that Indonesia (or more specifically Java) has been on a growth path which is bringing the economy nearer to the turning point? The evidence on this question is much more difficult to evaluate. We do not have data on the number of hours worked in agriculture and in marginal activities for workers whose primary activity is agriculture. This, we do not know if the proportion of labor devoted to agriculture has been decreasing over time, which would occur if the economy were approaching the turning point. But confining our attention to the major occupation of the labor force, we observed that the major source of additional employment for the growing labor force has been tertiary activities. The peculiarity of Indonesian economic growth in the - xxi - last decade has been the relative insignificance of manufacturing in providing additional employment, while the annual growth of workers primarily dependent on agriculture has grown substantially. This pattern of change in the sectoral allocation of labor over time is in marked contrast to the historical experience of Japan and Korea during their periods of transition, up to their turning points. In Indonesia the ratio of incremental tertiary employment to that in manufacaturing has been 4:1, while in the two "successful" rice economies it was between 1:1 and 2:1. Similarly, Japan and Korea had significantly higher outflows of labor from agriculture even with substantial output growth in this sector. 24. The comparative historical survey suggests that the Indonesian growth path in the last decade is not leading towards the turning point. There are, however, two points which might reduce the strength of this conclusion. First, Indonesia, unlike the other countries, is an oil-rich economy, and the multiplier effect of public sector spending (financed by oil revenue) might be particularly strong on employment creation in tertiary activities. Secondly, there is the possibility that Indonesia is farther away from the extensive and intensive margins of cultivation than the other economies, so that there is more scope for the productive absorption of labor in agriculture. 25. Both these points merit further enquiry. This report cannot give definitive answers on the basis of existing data. Regarding the first point, there is a strong presumption that the large inflow of resources via the oil sector directly and indirectly led to a very high growth in incomes; once terms of trade effects are taken into account, national incomes rose at about 10% per annum in the 1970s - and this would certainly have led to a very bouyant demand for services. However, the final impact on the labor market is not clear. It has already been mentioned that more data - both static and intertemporal - on labor earnings in the tertiary sector are urgently needed. A study of the trends in relative earnings in the tertiary sector and its important subgroups would help to reveal whether labor was "pushed" into this sector by the of lack of opportunities in other sectors. For the moment the report can only draw attention to the ratio of the increase in tertiary employment compared to manufacturing employment in Indonesia, which is extraordinarily high compared not only to Japan and Korea during their periods of transition, but also to other developing countries. However, it is clear that the past pattern of growth will not be repeated in the future, given current prospects for the international oil market. Even if labor was being pulled into service activities by higher wages in the 1970s, this may not occur in the 1980s. There appears to be little or no growth in GDP in 1982 and 1983, and, even with an effective program of structural adjustment to restore medium term growth, overall output for the decade may increase by little more than 5-6% per annum. On the basis of past trends in sectoral labor absorption, this would imply employment growth of only 2% per annum, compared with a labor force growth of 2.6%. In the absence of compensating policies, this would indicate a serious weakening in labor market prospects. 26. Turning to the question of extensive and intensive margins of cultivation in Java, data in Chapter 4 show that the land-man ratio (i.e., the - xxii - ratio of farm land to the agricultural labor force) was substantially lower in Java in 1971, than it was in Korea in the 1960s, or even in Japan through much of its period of transition since the turn of the century. Available farm studies also show that labor input, measured by man-days per hectare (along with yield of paddy per hectare), was higher in Java in 1971 than in Korea during the 1960s. It would appear that, although there might still be some isolated areas in Java in which the addition of more labor could increase productivity and farm profitability, the country as a whole is more intensively cultivated than the other two rice economies during their periods of transition. The report (in Chapter 4) also discusses some calculations, based on typical labor-land coefficients in Java regarding the most important crops grown there in 1979. On very conservative estimates, labor requirements in Javanese agriculture were some 20% below the available supply of farm labor (assuming each worker contributed 200 man-days per hectare). This is consistent with the data from field surveys quoted earlier, which indicated that a considerable portion of labor by agricultural workers is devoted to marginal activities. There are two other relevant questions: (i) how long could the 3.6% annual growth of agricultural output observed during the last decade in Java be sustained; and (ii) will the labor saving practices observed in parts of Java, along with the increased commercialization of agriculture, become more widespread, thus reducing the demand for labor? Policy Implications 27. The report casts considerable doubt on whether employment growth in the 1970s in the Indonesian economy has been moving on a path that would eventually lead to the turning point. With the likelihood of lower oil sector incomes and, thus, lower overall growth, for the remainder of the 1980s, the prospects for labor absorption are gloomy, based on past trends. The policy presumption that follows from this is that measures should be taken to give a more labor-intensive orientation to the growth process. This is a difficult subject, which this report, with its main focus on the analysis of the existing labor markets and trends over time, is not fully equipped to answer. However, some policy implications of the basic conclusion can be outlined. 28. The level and pattern of public expenditure can have a major influence on labor demand. In the short run, labor-intensive public works such as local infrastructure under the INPRES programs, make a direct contribution to employment, and can also channel income toward relatively poor rural areas, with indirect employment creating effects. Maintaining these activities will be of particular importance in the next two to three years when the Government is introducing other expenditure-reducing measures to bring down the external deficit. This will involve expenditure-switching within the public investment program, away from import-intensive activities to relatively labor-intensive ones. In May 1983, the Government announced a major rephasing of large-scale public investment projects, with the rupiah savings to be reallocated to expenditures with a high domestic employment content. - xxiii - 29. Labor absorption by agriculture was low in the 1970s relative to output increases, but its overall weight in employment means that its performance will be critical to the overall labor market. In the 1980s, it is likely that greater diversification away from rice production, to activities such as horticulture, some secondary food crops, small-scale livestock and fisheries, will be necessary to maintain past growth rates of about 4% p.a., and also to provide new employment. In addition, official and spontaneous transmigration to the Outer Islands could have a major impact. At the present rate of about 100,000 families per year, around a fifth of the incremental labor force on Java could be absorbed by this mechanism. 30. Policies relating to the manufacturing sector will play an increasingly important role in the next decade and beyond. As discussed in Chapter 4, this sector historically has been a major source of labor absorption in successful rice economics. Indonesia has been unusual in that this sector has accounted for an extremely small part of total employment and most of this has been in small-scale rural industries. Some reorientation in the policy environment will be necessary to encourage labor absorption. In particular, Indonesia is already near the limits of import-substitution in the labor-intensive consumer goods industries, and, apart from relatively sluggish growth in final demand, further increases in production for the domestic market will have to come largely from the much more capital-intensive producer goods industries. This underlines the need for industrial policies which emphasize export orientation, especially for consumer goods, through measures to reduce the existing import-substitution bias and specific export promotion measures. The development of the small-scale manufacturing sector may also be cause for concern, since there is the possibility that this may suffer declines in output through competition from the more productive medium- and large-scale sectors. This report did not attempt to study measures to support small industries, or the costs of protecting labor-intensive subsectors. Future analysis on this subject should focus attention on the large wage differential that exists between small- and large-scale enterprises, as discussed in Chapter 3. There may an economic choice between employing a small number of workers at a relatively high wage, and a large number at a lower wage. 31. Finally, factor market prices do not give the correct guidance to determine labor intensity in specific investment projects. It will be necessary to develop policy instruments to compensate for this. The appropriate use of the shadow wage in public investment project appraisal has already been mentioned as one such instrument. A point of some importance might be stressed in this connection. In practice, project appraisal studies usually shadow price labor at a fairly late stage in the decision-making process, when the choice between alternatives is only marginally affected by the appropriate accounting price of labor. Thus, direct consideration of sector-specific alternatives must supplement the use of the shadow wage if wider importance is to be given to a labor-intensive strategy of development. - xxiv - 32. An equally important factor - and empirically it might be more significant - which causes market prices to give the wrong signal to the choice of techniques is the underpricing of capital, at least in some major areas of the economy. For example, tractorization might appear to be profitable at existing market prices, but may not be socially productive if appropriate accounting prices for labor and capital were used I', unless specific investigations reveal pockets of labor scarcity in particular areas. Further Work on Employment and Labor Incomes in Indonesia 33. Although the report emphasizes that a model of the labor market based on a hypothesis of surplus labor in the rural sector (and the associated idea of the productive rice sector being a privileged sector from the point of view of job seekers) is directly relevant to understanding Indonesian development, it does not conclude that this model has been definitively tested and verified. Further work and testing of alternative labor market models need to be done, both at the micro and macro level. Appendix A to Chapter 3 presents theoretical ideas on alternative micro models of the rural labor market, and discusses how far existing empirical data can be used to support the "surplus" labor model. The Annex to this report explores whether the broad macroeconomic trends in Indonesian development fit the framework of aggregative neoclassical models, and it also seeks to establish ranges of values for key parameters which are consistent with actual experience. The relevance of the alternative surplus labor model for the interpretation of macro trends is studied in this context. 34. The report recommends further empirical work to observe directly the workings of labor markets. With respect to labor markets in rural areas, the following are important areas for investigation. (i) The demand for labor in Javanese agriculture depends upon the nature of the busy season, as compared to the slack season. How long is the busy season, and how does it affect different aspects of the labor force? What is the proportion of labor time devoted to the productive (rice) sector and to marginal activities, for significant age-sex groups distinguished by economic status? Are earnings per hour in marginal activities substantially lower than in the rice sector for the busy season, as they are for the year as a whole? Much information on many of these questions is already available in completed village studies, but either it has not been analyzed or it is not compiled in a form suitable for answering these questions. A more careful study of the micro data already collected for individual villages is, therefore, a task of high priority. 1/ See Rudolph S. Sinaga, "Implications of Agricultural Mechanization for Employment and Income Distribution," Rural Dynamics Study Series, No. 2, Bogor. - xxv - (ii) Clearly, regional or even village-to-village variations in labor market conditions are of substantial importance in Java, as in all rural Asian societies. Some villages may indeed fit the surplus labor model more than others. Attention should be firmly fixed on this point in the further analysis of village-level data. Even if detailed village studies are not available in sufficient number, a cross section analysis of intervillage variations of wages in selected occupations, and their relationship to the underlying economic conditions in the villages would probably be quite revealing. (iii) At the macro level, the report emphasizes the need to study more series on wage rate movements, particularly in agriculture. Again, the importance of analyzing trends in wages and employment on a regional basis cannot be overemphasized. Data from the 1980 Census should provide a wealth of information on regional employment trends during the 1970s when used in conjunction with other sources of statistics including the 1971 Census. (iv) An equally important area of information is the level and trends in earnings in the very large informal sectors, particularly in service activities. While a fuller analysis of existing household surveys would shed more light on this issue the prevalence of multiple occupations in Indonesia, particularly in the rural sector, requires some careful formulation of questionnaires for future household surveys. An attempt should be made to collect information on hours worked and income earned in the different activities pursued by workers, instead of confining the inquiry to the major activity of the respondent. (v) It is also important to go beyond the linkage among wages, employment and output as discussed in this report. For example, land and credit are key factors, but the present data and state of knowledge have precluded a systematic consideration of the links among markets for factors of production (land, capital, labor) and the role that credit and finance, in a broad sense, play in those links. Gathering further statistical information at the village level, which this report recommends, should be done with a view to facilitating studies on the linkages among markets for factors of production and the role of credit. RESUME ET CONCLUSIONS Introduction 1. Le pr6sent rapport fait partie de l'evaluation permanente, par la Banque, des salaires et de l'emploi en Indonesie. Le dernier rapport specialis6 1/ 6tudiait les march6s du travail et la r6partition des revenus jusqu'en 1976. I1 n'est question ici que des conditions des marches du travail (il n'a pas 6t6 possible, par suite de contraintes financieres, d'examiner de maniere approfondie les facteurs connexes que sont la pauvret6 et la repartition des revenus). Les enquetes trimestrielles de 1977, 1978 et 1979 sur la main-d'oeuvre ainsi que les r6sultats provisoires du recensement de la population de 1980 ont servi de donn6es de base. Ce dernier recensement indique que la population de l'Indonesie a augmente de 2,3 % par an au cours des annees 70, et non de 2 % comme pr6vu. Ce sont donc une quinzaine de millions de personnes qui devaient venir grossir les rangs de la main-d'oeuvre au cours de la prochaine decennie, alors que l'augmentation avait 6t6 de 25 millions pour les deux dernieres. C'est dire combien il est imp6ratif de suivre et d'evaluer en permanence les marches du travail ainsi que les facteurs auxquels ils sont soumis de facon cyclique. 2. La tache est d'autant plus urgente qu'on annonce un ralentissement de la croissance de la production au cours des annees 80. Les effets que peut avoir sur l'emploi et les revenus la conjonction d'une offre de main-d'oeuvre en expansion constante et d'un ralentissement de la croissance varient selon l'idee que l'on se fait de la maniere dont fonctionnent les marches du travail. Les consequences seront tres diff6rentes au niveau des mesures a prendre, selon l'hypothese retenue. Preciser ce que pourraient etre ces consequences et promouvoir l'6tude des marches du travail en Indonesie, notamment par un nouvel examen de l'utilisation et du revenu de la main-d'oeuvre, en particulier dans les zones rurales de Java, tel est, essentiellement, l'objectif du present rapport. 3. Le point de vue adopte dans le rapport pr6cedent 6tait qu'en Indonesie les m6canismes des marches du travail fonctionnent bien. Vu les variations de salaire observ6es (dans le temps et selon les regions) ainsi que la mobilit6 de la main-d'oeuvre, il concluait que, dans 1'ensemble, les forces de l'offre et de la demande expliquent assez bien le niveau et 1/ Banque mondiale (1980), 2788-IND. - xxvi - - xxvi i - la structure des salaires et de l'emploi et que, de ce fait, l'Indonesie ne connaissait pas de chomage permanent ou structurel. Les fournisseurs de main-d'oeuvre faisaient, par le jeu de la concurrence, baisser les salaires jusqu'au niveau du plein emploi, et ce salaire repr6sentait, a la marge, la productivite de la main-d'oeuvre et le coat de l'effort. Les salaires etaient bas, mais cela s'expliquait par l'absence de facteurs de production compl6mentaires (capitaux, comp6tences) et non par le fait que le march6 ne refl6tait pas le coat d'opportunit6 du travail, a savoir la marge de salaire entre diverses occupations possibles. En outre, le rapport pr6c6dent laissait entendre que l'eventail des salaires observe correspondait vraisemblablement a des diff6rences dans la qualite de la main-d'oeuvre fournie. 4. A ce compte la, il n'y aurait pas a proprement parler de probleme de chomage en Indonesie et il n'y aurait donc pas lieu de prendre des mesures specialement destinees a cr6er des emplois. De meme, les projets d'investissement de l'Etat (notamment dans les zones rurales de Java) devraient etre evalues sur la base d'un cout de la main-d'oeuvre proche de son prix de marche, c'est-a-dire que le rapport des salaires de ref6rence aux salaires du march6 devrait etre proche de l'unit6. 5. Tel n'est pas le point de vue adopt6 ici. Tout d'abord, le present rapport met en doute l'idee selon laquelle les differences de salaires, sur les marches urbains du travail, correspondraient a des differences dans la qualit6 de la main-d'oeuvre. II etablit une 6chelle de salaires pour la main-d'oeuvre non qualifi6e, et l1'ventail des salaires observe parait trop ouvert pour pouvoir s'expliquer par des differences de qualite dans cette categorie de main-d'oeuvre. Ensuite, la conjonction d'un faible taux de participation des jeunes et des vieux travailleurs dans les zones urbaines et d'un taux de chomage elev6 parmi les jeunes des villes aboutit, selon le present rapport, a un sous-emploi marqu6 des jeunes, des plus instruits en particulier, et ne peut s'expliquer par des modeles de chomage qui font une grande place a la recherche d'un travail de transition. En troisieme lieu, les forces 6conomiques et sociales qui ne favorisent pas l'equilibre de l'offre et de la demande de main-d'oeuvre sur le march6 du travail jouent un grand role dans la d6termination des salaires. La variabilite des salaires exprim6e en valeur riz, soit dans le temps, soit d'un village a l'autre, se situe dans une 6troite fourchette qui demeure stable. La stabilite du salaire agricole a l'interieur de cette fourchette rappelle celle des salaires agricoles observee au Japon et en Coree avant l'epuisement de l'exc6dent de main-d'oeuvre rurale. L'hypothese de travail adoptee ici est qu'il existe un excedent de main-d'oeuvre dans de nombreux villages de Java : aux salaires actuels, l'offre de main-d'oeuvre capable et desireuse de s'employer a la culture du riz est superieure a l'offre d'emploi dans ce secteur, tant en periode de campagne qu'a la saison creuse. Tout comme dans les villes, le march6 de la main-d'oeuvre, dans les campagnes rizicoles, comporte des m6canismes qui empechent cet exc6dent de main-d'oeuvre de faire baisser les salaires. - xxviii - 6. Ces trois caracteristiques des marches du travail en Indon6sie ont d'importantes consequences au niveau des d6cisions a prendre. Elles mettent en lumiere la necessite d'agir pour repondre a la presence d'une main-d'oeuvre caract6ris6e par une croissance rapide et une productivite relativement faible. II faut donc faire de la creation d'emplois un objectif s6par6 (plutot que de l'inclure dans une strat6gie visant a maximiser la croissance de la production). C'est dire aussi que l'6valuation des projets d'investissement publics (en particulier dans les zones rurales de Java) doit fonder le coat de la main-d'oeuvre non qualifiee non pas sur le salaire pratique dans la riziculture, mais sur la plus faible r6mun6ration obtenue pour une activit6 marginale du bas de 1'echelle 6conomique. 7. L'interpretation qui est donn6e ici du marche du travail s'appuie sur des donnees tir6es principalement des etudes de village et des enquetes sur la main-d'oeuvre. Les 6tudes de village sont indispensables pour mieux evaluer l'economie, de Java notamment. On cherche ici a utiliser les conclusions qui decoulent de ces 6tudes pour degager, a partir de donn6es globales, l'6volution g6n6rale de l'emploi et de la production. Cet effort devrait etre poursuivi. Ch6mage, taux de participation et education 8. Le Chapitre 1 brosse un tableau general du ch6mage d6clar6 (c'est-a-dire chiffre) en Indonesie oui, comme dans nombre de pays en d6veloppement, le probleme se pr6sente plus particulierement dans les zones urbaines, surtout parmi les jeunes. Le pourcentage de ch6meurs d6clar6s est trois a quatre fois plus eleve dans les villes que dans les campagnes. Dans le secteur urbain, le taux de ch6mage masculin est de 25 2 pour le groupe des 15 a 19 ans et de 18 % pour le groupe des 20 a 24 ans (la moyenne generale dans les villes 6tant d'environ 7 % pour la population masculine). 9. Mais le ch6mage chiffre ne donne pas la mesure exacte des difficult6s reelles que rencontrent les jeunes des villes a la recherche d'un emploi. Il ressort des enquetes sur la main-d'oeuvre que, pour ceux qui se disent ch6meurs, la p6riode de ch6mage ne dure gen6ralement pas plus d'un mois. Par contre, les taux 6lev6s de non-participation a la main-d'oeuvre enregistr6s parmi les jeunes des villes revelent, de maniere saisissante, avec quelle lenteur ils abordent le march6 urbain du travail. II semblerait donc que les jeunes ne cherchent pas s6rieusement de travail tant qu'ils n'ont pas l'impression qu'ils pourront assez vite en trouver un a leur convenance. Pour la population urbaine masculine, les taux de participation etaient de 34 % pour les 15 a 19 ans et de 75 % pour les 20 a 24 ans. Les taux correspondants de participation f6minine a la main-d'oeuvre etaient de 23 % et de 29 % - soit sensiblement inf6rieurs aux taux enregistres dans le secteur rural. On peut se faire une id6e de l'ampleur du probleme quand on sait que, dans les villes, meme dans le groupe d'age des 20 a 24 ans, 40 % de la population masculine et 76 % de la population f6minine etaient sans emploi (c'est-a-dire 6taient en chomage ou ne faisaient pas partie de la main-d'oeuvre). Les donn6es SAKERNAS sur l'occupation des non-participants montrent que, dans ce - xxix - groupe d'age, jusqu'a 20 % de personnes de sexe masculin des villes disaient etre a 1'6cole, alors qu'officiellement le taux des effectifs du tertiaire, a cette date (1977/78), n'etait que de 2 %. I1 semble donc qu'un grand nombre de non-participants 6taient en realite des travailleurs decourag6s qui, pour excuser leur inactivite, pr6tendaient suivre des etudes quelconques. 10. Le rapport qui existe entre le sous-emploi des jeunes urbains et 1'expansion de 1'education en Indon6sie est une question d'importance. II est clair, d'apres les donnees present6es au Chapitre 1, que les taux de chomage et de non-participation augmentent avec le degre d'instruction, compte tenu des ajustements en fonction du groupe d'age. Ainsi, dans les villes, le taux de participation des personnes de sexe masculin qui ont de 20 a 24 ans est d'environ 60 % pour ceux qui ont fait des etudes secondaires et d'environ 90 % pour ceux qui n'ont pas depass6 le stade du primaire. II convient de se rappeler, a cet egard, que les 6tudes secondaires devraient normalement avoir pris fin bien avant l'age de 20 ans et que le fait de ne pas finir l'ecole avant un age aussi avanc6 est d6ja une preuve des difficultes qu'ont les jeunes a trouver un emploi. Les chiffres relatifs au ch6mage declar6 font apparaitre qu'une proportion non n6gligeable de ceux qui ont fait des 6tudes secondaires en sont encore a chercher un emploi aux approches de la trentaine. II n'y en a pas moins penurie de main-d'oeuvre instruite dans certains secteurs, comme dans l'agriculture, qui manque de techniciens. I1 faut, de toute 6vidence, poursuivre l'effort de recherche dans ce domaine afin de determiner en quels points de l'6ventail des niveaux d'instruction il y a penurie ou risque d'excedent. Evolution gen6rale de la production, de l'emploi et des salaires de 1961 a 1980 11. Le rapport comprend tout un ensemble de donnees globales sur la croissance de l'6conomie indonesienne de 1961 a 1971 et de 1971 a 1980. Il met particulierement en relief le taux de croissance de l'emploi (mesure par le chiffre de main-d'oeuvre). Ainsi qu'il ressort des meilleurs renseignements disponibles, ce taux aurait ete de 2,9 % par an au cours de la p6riode 1971-80, en augmentation donc, par rapport aux annees 60 (2,4 %). Ces taux se situent au bas de la fourchette des taux de croissance plausibles indiques dans le rapport anterieur de la Banque mondiale sur l'emploi. Par contre, les precedents rapports economiques de la Banque ont retenu le taux le plus 61ev6 enregistre au cours de cette periode, soit 4,7 % 1/. L'estimation donnee ici pourra, naturellement, etre revisee a mesure que seront analysees de facon plus d6taillee les 1/ Voir Banque mondiale (1979), 2093-IND et Banque mondiale (1980), 2788-IND. Dans le document Banque mondiale (1981), 3307-IND, le taux de croissance de 1'emploi cite est de 3,3 % par an. Dans Banque mondiale (1982) 3795-IND, le taux estimatif est de 2,65 %. - xxx - donn6es du recensement de la population de 1980, mais il semble que le chiffre de 4,7 % soit exag6r6ment eleve. L'acc6l6ration de la croissance de l'emploi (et de la main-d'oeuvre) au cours des ann6es 70 est liee a une augmentation de la proportion de la population active plutot qu'a un accroissement des taux de participation de la main-d'oeuvre. 12. La croissance de la production (7,3 % par an) a 6te largement superieure a ce taux revis6 de croissance de 1'emploi pendant la derniere d6cennie. Toutefois, 1'examen de l1'volution des salaires r6els dans les quelques secteurs organises pour lesquels il a ete possible d'obtenir des donn6es - agriculture, plantations et batiment - fait apparaitre que, pour les travailleurs non specialis6s, les salaires r6els sont demeur6s constants pendant la majeure partie de cette p6riode, en d6pit d'une croissance rapide de la production. 13. Cette croissance rapide de la production a salaires constants serait, selon le rapport, l'une des principales caracteristiques de l'expansion macroeconomique de l'Indonesie. Dans une situation oii les marches du travail sont competitifs et oi les salaires sont determines par les forces du marche, ce phenomene peut s'expliquer par le fait que la croissance de l'emploi "effectif" (l'emploi mesur6, non par le nombre de personnes employees, mais par leur efficacit6, avec ajustement au titre de l'accumulation des competences) est sup6rieure a celle de l'emploi "mesure". Une autre explication possible serait que les economies de main-d'oeuvre dues aux progres de la technique ont ete suffisantes pour permettre une croissance rapide de la production a salaires constants. Ces possibilites sont examinees a l'Annexe A. La conclusion qui d6coule de l'analyse m6thodique du premier point est que l'accumulation de competences ne change rien a l'ensemble du tableau. Les resultats chiffres tir6s du modele de croissance avec progres technique indiquent pour quelles hypotheses precises de la valeur des parametres le taux de croissance de la production enregistr6 serait compatible avec des taux de salaire constants. 14. D'apres le rapport, un cadre de r6f6rence fonde sur une main-d'oeuvre abondante, disponible aux taux de salaire du march6, dans les principaux secteurs de l'economie (y compris dans la riziculture) permettrait d'interpr6ter 1'evolution observ6e dans la productivite et les salaires. Mais, a la diff6rence de certaines versions ant6rieures du modele correspondant a un excedent de main-d'oeuvre, l'hypothese selon laquelle l'offre de main-d'oeuvre est parfaitement 6lastique pour un salaire correspondant a la productivit6 moyenne du travail, dans l'agriculture, est inacceptable. Dans le secteur agricole, la production s'est accrue au rythme annuel de 3,6 %, ce qui est nettement superieur au - xxxi - taux de croissance de l'emploi dans ce secteur (1 % par an) 1/. S'appuyant sur des 6tudes microeconomiques approfondies de villages javanais, le rapport examine les march6s de la main-d'oeuvre rurale et emet plusieurs hypotheses sur la determination des salaires agricoles. L'effort qui est ainsi fait pour rattacher un "micro-aspect" du marche du travail (au Chapitre 3) au tableau d'ensemble de 1'6conomie (Chapitre 2) n'est qu'un premier pas. I1 reste encore beaucoup a faire a propos des 6tudes de village et de la collecte de donn6es. Certains des points qui n6cessitent une etude plus pouss6e sont indiqu6s en fin de document. Le fonctionnement des march6s du travail en Indonesie : une interpr6tation 15. Les deux principaux facteurs conventionnels mis en lumiere par les etudes micro6conomiques de village a Java sont les suivants : a) Le nombre d'heures de travail consacrees a l'activite agricole la plus importante - la riziculture - ne represente qu'une faible partie du nombre total d'heures que les m6nages ont fournies au march6 de la main d'oeuvre agricole, y compris les menages sans terre 2/. La contribution aux taches agricoles salariees est g6neralement de 30 a 40 % des travailleurs sans terre. Le reste du temps de travail est consacre a diverses activit6s de commerce, de services, d'artisanat, et parfois a un travail salarie non agricole en dehors des villages. I/ Sans parler de l'extraction miniere, les taux de croissance globale de la production et de l'emploi ont ete les suivants : agriculture (3,5, 1,0); secteur secondaire (12,9, 4,9); secteur tertiaire (8,3, 6,1). Le secteur secondaire comprend les industries manufacturi6res, le batiment, les transports et les services publics. Le secteur tertiaire comprend le commerce et les services. Les pourcentages d'emploi dans les trois secteurs 6taient, a la fin de la periode, de 56 %, 15 % et 29 % respectivement. D'apres le rapport de la Banque mondiale (1981) 3182-IND, Annexe 1, l'emploi dans les grandes et moyennes entreprises aurait augmente de 6 a 12 % au cours des ann6es 70. Toutefois, en 1974/75, ces entreprises ne representaient que 13 % de l'emploi total dans le secteur manufacturier. 2/ Ici, et dans le reste du rapport, le terme g6nerique "riz" est utilise pour designer le secteur agricole productif qui absorbe une part consid6rable de la main-d'oeuvre salariee. IL recouvre les cultures de plantation ainsi que d'autres cultures qui jouent un r6le important dans I'6conomie paysanne, comme la canne a sucre. - xxxii - b) La remun6ration du travail (par homme-heure) est sensiblement inf6rieure en dehors du secteur du riz. II ressort d'6tudes sur le terrain que cette remun6ration pourrait etre de l'ordre de 30 % du salaire pratique dans le secteur du riz pour les personnes de meme age/sexe. Ces diverses activites (exercees le plus souvent au compte de l'int6resse) etant d'un moindre rapport, on les a classees dans la categorie dite "secteur marginal" (voir suite du document). Evidemment, pendant la morte-saison, une plus grande partie du temps est consacr6e a des activit6s marginales. Mais, meme pendant la campagne du paddy, ces activit6s absorbent une part importante du temps de travail des adultes, du fait, notamment, que les marches du travail du riz sont respectivement circonscrits aux villages int6resses et que, dans tout village, meme en saison de pointe, la demande de main-d'oeuvre rizicole se mesure en jours, et non en mois. La disparite saisonniere de remun6ration du travail entre le secteur du riz et les activit6s marginales exige une 6tude plus poussee. L'ampleur du diff6rentiel saisonnier n'est pas mentionnee dans toutes les etudes sur le terrain, mais uniquement, semble-t-il, dans les etudes ne portant que sur la haute saison 1/. 16. On peut donc dire qu'il existe un excedent de main-d'oeuvre a Java dans le sens - limite et particulier - qu'il existe un reservoir d'unites de main-d'oeuvre (plutot que de travailleurs) dans des activit6s marginales oii le salaire horaire est sensiblement inferieur a ce qu'il est dans le secteur du riz. L'importance macro6conomique de ce type de march6 du travail r6side dans le fait que le secteur du riz a la possibilit6 de puiser dans ce r6servoir au taux de salaire superieur pratique dans ledit secteur. Du fait de cette elasticit6 parfaite de l'offre de main-d'oeuvre dont il ben6ficie, le rendement par travailleur peut augmenter dans le secteur rizicole pendant un certain temps sans autre effet sur les taux de salaires, meme en l'absence d'innovations techniques reductrices de main-d'oeuvre. La question se pose donc de savoir pourquoi des propri6taires guides par la recherche du profit payent a ceux qui travaillent leurs champs un salaire plus 6lev6 que n'en per,oivent ceux qui s'emploient a des activites marginales. La reponse est donn6e probablement par le rapport entre salaire et efficacite, un salaire eleve ayant pour effet de stimuler l'ardeur des travailleurs. Par rapport a la faible rentabilite des activites ind6pendantes auxquelles il s'adonne dans le secteur marginal, le rendement du travailleur s'accroit plus vite, en proportion, que le salaire qui lui est offert, jusqu'a un point d'inflexion a partir duquel le gain d'efficacite cesse d'etre 1/ Dans la plupart des villages, il y a generalement un eventail de salaires a la fois pour le riz et pour les activites marginales. Le modele double, avec salaires moyens differents, est une sorte de convention permettant de souligner l'importance des activites marginales comme r6servoir de main-d'oeuvre "exc6dentaire". - xxxiii - proportionnel. II existe donc un niveau de salaire - plus 6lev6 que le prix du travail dans le secteur marginal - qui correspond au coat de salaire unitaire le plus bas pour l'employeur. Nul employeur, fut-il le plus apre au gain, ne proposera un salaire inferieur a celui-la, meme s'il y a abondance de main-d'oeuvre a ce prix. On pourrait donc penser que le but vise en proposant un salaire sup6rieur est d'attirer la meilleure main-d'oeuvre possible. II se trouve - et c'est 1l un point capital - que le sens de la relation de cause a effet va d'u salaire a l'efficacit6, et non pas le contraire. Plus le salaire est 6leve (jusqu'au point d'inflexion), plus le rendement dans presque tous les cas, augmente. Etant donn6 qu'il y a abondance de main-d'oeuvre au salaire plancher qui minimise les coats salariaux, les employeurs doivent choisir certains travailleurs plutot que d'autres. Toutes les etudes faites sur le terrain dans les campagnes javanaises signalent l'importance du systeme d'embauche preferentiel pour la main-d'oeuvre employ6e dans la riziculture. 17. On a deja dit que le march6 de la main-d'oeuvre rurale a Java (comme dans d'autres pays d'Asie) est sp6cifiquement villageois. Ainsi, le salaire 6conomique qui determine le prix plancher pour les travaux agricoles est lie a la norme propre au village et reflete ses conditions 6conomiques. Les salaires agricoles varient d'un village a l'autre, pour la meme saison ou le meme travail. Ainsi, dire que le salaire reel demeure constant revient a dire que la fourchette des taux de salaire correspondant a un travail donn6 ne s'est pas sensiblement elargie, bien que la production par travailleur ait augment6. On ne prevoit pas de hausse tant que la demande de main-d'oeuvre agricole n'aura pas progress6, par rapport a l'offre, dans des proportions suffisantes pour entrainer une r6duction sensible du reservoir de main-d'oeuvre employee a des activites marginales. 18. Le march6 urbain et le marche rural du travail se rejoignent au niveau de deux types de migrants : ceux qui, sans emmener leur famille, vont travailler temporairement dans les villes et ceux qui, migrants plus permanents, vont s'installer dans les zones urbaines avec leur famille. Les premiers se rencontrent surtout dans le secteur non structure du marche urbain du travail et, bien que l'on n'ait que tres peu de donn6es sur la question, on peut penser que ce ph6nomene de va-et-vient migratoire maintient relativement proches l'une de l'autre la remuneration du travail dans le secteur urbain non structure et dans les activites marginales du secteur rural, compte tenu des differences de coat de la vie et du transport. En dehors du secteur non structure, on peut penser que le rapport salaire-efficacit6 s'applique aux salari6s des villes autant qu'aux ouvriers agricoles. En fait, il est raisonnable de supposer que le lien entre salaire et efficacit6 est d'autant plus fort que les effectifs de main-d'oeuvre sont plus nombreux et que l'organisation de l'unite qui embauche est plus complexe. On a pu observer dans maints pays en d6veloppement (y compris au Japon durant sa phase de d6veloppement) que, dans le secteur manufacturier, le salaire des ouvriers non specialises augmente avec la taille de l'entreprise, meme en l'absence de syndicats ou de legislation du travail. Le cas de l'Indon6sie n'est nullement different pour ce qui est de la physionomie generale des salaires urbains. - xxxiv - Le Chapitre 3 traite de 1'6chelle des salaires pour les travailleurs non specialises du secteur manufacturier des villes avec, au bas de l'6chelle, les petites entreprises non m6canisees et, en haut, les coentreprises ou les multinationales. L'ecart des salaires entre ces deux extremes peut atteindre 250 %. Le salaire de reference 19. Le modele de march6 du travail susmentionne suppose que le salaire de r6ference utilise pour les projets ruraux se calcule sur la base du coat d'opportunit6 correspondant a l'emploi d'une unit6 de main-d'oeuvre non sp6cialisee, c'est-a--dire la r6muneration des activites rurales marginales. I1 ressort d'etudes faites sur le terrain qu'a Java, cette remun6ration peut ne pas etre sup6rieure au tiers du salaire pratique dans le secteur du riz. Selon les calculs, si l'on tient compte du coat social de l'accroissement de la consommation du m6nage rural representatif qui fournit la main-d'oeuvre, le salaire de reference repr6senterait 28 % du salaire agricole. Comme on l'explique plus loin, il convient de nuancer cette affirmation, les donn6es relatives aux gains saisonniers etant toujours rares en Indonesie. Perspectives concernant l'emploi et les salaires 20. Se fondant sur l'6volution passee, le rapport prevoit un accroissement de la main-d'oeuvre de 2,6 % par an environ jusqu'en 1990. Sera-t-il absorbe par un accroissement de la demande au cours de la prochaine decennie et a quels niveaux de salaire? Telle est la question primordiale que l'on se pose devant les perspectives de la croissance de l'emploi en Indon6sie. Le rapport examine les perspectives de l'emploi en Indonesie dans l'optique d'un "tournant" (correspondant a l'6puisement d'un "exc6dent" de main-d'oeuvre, tel qu'il est defini a la page 11). 21. Un coup d'oeil r6trospectif sur l'exp6rience de deux pays d'Asie dont l'economie est fond6e sur la riziculture - le Japon et la Coree - dont le cas est 6tudie au Chapitre 4, fait apparaitre que l'un et l'autre ont connu une periode de croissance soutenue de la production par travailleur dans l'agriculture, sans que le salaire r6el ait augmente, sinon legerement. Cette periode de croissance a salaire presque constant (dans l'agriculture) s'est prolong6e au Japon (50 ans ou davantage), la production par travailleur n'augmentant qu'a un rythme mod6r6, et a 6t6 nettement plus courte en Cor6e, oii la productivit6 agricole a augmente beaucoup plus vite durant la periode de transition qui a prec6d6 le tournant. Mais, dans l'un et l'autre cas, le tournant est reconnu comme un fait historique, comme un v6ritable "bond" dans l'evolution temporelle des salaires agricoles reels. Apres une assez longue periode de lente progression des salaires reels, les salaires ont augment6 de six a sept fois plus vite qu'auparavant. - xxxv - 22. Du point de vue de la protection sociale, ce tournant signifie qu'avant qu'il ne soit atteint, la part de la production qui va au salaire diminue r6gulierement, la productivite du travail augmentant plus vite que la masse salariale. Comme en Asie la m6canisation de l'agriculture ne joue pas un grand role avant le tournant, une grande partie de la diff6rence entre productivite du travail et salaire se traduit par un accrois sement de la part des loyers et des profits. Un deuxieme probleme important que souleve la question des "retomb6es" de l'augmentation de la productivit6 agricole est de savoir si l'6conomie suit une ligne de croissance qui pourrait bientot d6boucher sur le tournant. Si tel est le cas, une plus grande proportion du travail de ceux pour qui l'agriculture est l'occupation principale ira, du secteur marginal, a l'agriculture. Les salaires horaires 6tant plus 6leves dans ce secteur, le revenu par travailleur finira par augmenter, meme si les salaires agricoles demeurent constants. Toujours pendant la p6riode de transition, les deux pays examin6s au Chapitre 4 - le Japon et la Coree - ont enregistre un transfert important de main-d'oeuvre de l'agriculture vers le secteur manufacturier, ou les salaires 6taient plus eleves (le nombre de travailleurs agricoles diminuant en fait, en chiffres absolus durant certaines parties de la p6riode de transition). Cette redistribution de la main-d'oeuvre fait egalement progresser le revenu par travailleur, pour l'ensemble de l'6conomie jusqu'au tournant, mame si les taux de salaire reels demeurent constants dans chaque secteur. 23. Il ressort des donn6es present6es ici que l'Indonesie n'a pas encore depasse ce tournant. Rien n'indique qu'il y ait v6ritablement des pressions a la hausse sur les salaires r6els, bien que la production par travailleur agricole ait fortement augment6 pendant la derniere decennie. Mais y a-t-il des signes qui montrent que l'Indon6sie (ou, plus precis6ment, Java) se trouve engagee sur une voie de croissance qui rapproche l'6conomie du tournant? Les elements d'appreciation dont on dispose sur ce point sont plus difficiles a 6valuer. On ne connait pas le nombre d'heures que consacrent a des activites agricoles et marginales les travailleurs dont l'agriculture est la principale activit6. On ne sait pas si la proportion de travail consacree a l'agriculture diminue avec le temps, ce qui serait le cas si l'economie approchait du tournant. Mais, bornant notre attention a l'occupation principale de la main-d'oeuvre, nous avons observ6 que le tertiaire est la principale source d'emplois additionnels pour les effectifs croissants de la main-d'oeuvre. Ce qui a singularise la croissance de l'economie indonesienne pendant la derniere decennie, c'est le role relativement peu important qu'y a jou6 le secteur manufacturier comme source d'emplois additionnels, alors que le nombre annuel de travailleurs tributaires de l'agriculture n'a cess6 de croitre. Cette evolution de la repartition sectorielle de la main-d'oeuvre dans le temps contraste vivement avec le cas du Japon et de la Coree au cours de leur periode de transition, avant le tournant. En Indon6sie, le rapport des augmentations d'emplois dans le tertiaire aux augmentations d'emplois dans le secteur manufacturier a et6 de 4 pour 1 alors que, pour les deux pays rizicoles qui ont "reussi", ce rapport se situait entre 1 et 2. De meme, au Japon et en Coree, les travailleurs agricoles ont ete bien plus nombreux a quitter l'agriculture, ce qui n'a pas empech6 ce secteur de connaitre une forte croissance de sa production. - xxxvi - 24. La comparaison des diverses evolutions suivies indique que la voie suivie par la croissance indonesienne au cours de la derniere decennie ne mene pas au tournant. Deux faits, cependant, pourraient permettre de nuancer cette affirmation. Tout d'abord, a la diff6rence des autres pays, l'Indonesie est riche en p6trole et l'effet multiplicateur des depenses du secteur public (financ6es par les revenus p6troliers) pourrait etre particulierement favorable a la creation d'emplois dans le tertiaire. Ensuite, il est possible que l'Indon6sie ait de plus grandes marges de culture extensive et intensive que les autres pays, si bien que son agriculture aurait une plus grande capacit6 d'absorption productive de main-d'oeuvre. 25. L'un et l'autre points meritent reflexion. Le present rapport ne peut y apporter de r6ponse cat6gorique au vu des donnees existantes. Pour ce qui est du premier point, il y a tout lieu de penser que le gros apport de ressources da au secteur p6trolier a provoqu6, directement et indirectement, une tres forte croissance des revenus. Compte tenu des effets des termes de l'6change, le revenu national s'est accru d'environ 10 % par an dans les ann6es 70 - ce qui ne pouvait manquer de provoquer une tres forte demande de services. Mais on ne voit pas bien quelle en aura 6t6 finalement l'incidence sur le marche du travail. On a d6ja dit qu'il est urgent d'avoir davantage d'informations - tant statiques que chronologiques - sur la r6muneration du travail dans le secteur tertiaire. Une 6tude comparative de l'evolution des salaires dans le tertiaire et dans ses importants sous-groupes permettrait de savoir si la main-d'oeuvre a ete "pouss6e" vers ce secteur par l'absence de possibilites d'emploi dans les autres. On ne peut, pour l'instant, que se borner a appeler l'attention sur le taux plus rapide d'accroissement du nombre d'emplois dans le tertiaire par rapport au secteur manufacturier. II est extremement eleve si on le compare, non seulement aux taux enregistres au Japon et en Coree durant leurs p6riodes de transition, mais aussi dans d'autres pays en developpement. Toutefois, il est clair que dans le tertiaire, la croissance future ne suivra pas le meme trace que dans le passe, vu les perspectives actuelles du march6 international du petrole. Dans les annees 70, les salaires plus eleves qui y 6taient pratiques ont attire la main-d'oeuvre. Mais il est peu probable que ce phenomene se reproduise pendant les annees 80. Il n'y a pas - ou guere - eu, semble-t-il, de croissance du PIB en 1982 et 1983 et, meme avec un bon programme d'ajustement structurel pour retablir la croissance a moyen terme, il est peu probable que l'ensemble de la production, pour la decennie, augmente de plus de 5 a 6 2 par an. Compte tenu des tendances pass6es de l'absorption de main-d'oeuvre par secteur, cela signifierait que 1'emploi n'aurait progress6 que de 2 Z par an, alors que la main-d'oeuvre aura augment6 de 2,6 %. S'il n'est pas pris de mesures compensatoires, il faut s'attendre a un affaiblissement grave des perspectives du marche du travail. 26. En ce qui concerne la question des marges de cultures extensive et intensive a Java, les donnees du Chapitre 4 indiquent que le coefficient terre-homme (c'est-a-dire le rapport entre terres agricoles et main-d'oeuvre agricole) 6tait considerablement plus faible a Java en 1971 - xxxvii - qu'en Cor6e dans les ann6es 60, ou meme qu'au Japon pendant une grande partie de sa periode de transition depuis le d6but du siecle. On voit aussi, d'apres les etudes dont on dispose sur les exploitations agricoles, que l'apport de main-d'oeuvre, mesure en hommes-jours par hectare (avec le rendement de paddy a l'hectare) 6tait plus 6lev6 a Java en 1971 qu'en Coree dans les ann6es 60. I1 semblerait que, bien qu'il puisse y avoir encore a Java quelques zones isolees ou un nouvel apport de main-d'oeuvre serait susceptible d'accroitre la productivite et la rentabilite des exploitations, le pays, dans son ensemble, soit cultiv6 de maniere plus intensive que les deux autres pays rizicoles durant leur p6riode de transition. Au Chapitre 4, le rapport examine aussi certains calculs, faits sur la base de coefficients terre-homme propres a Java, concernant les plus importantes cultures pratiquees dans l'lle en 1979. Selon une estimation tres prudente, les besoins de main-d'oeuvre agricole, a Java, etaient inferieurs d'environ 20 % aux effectifs disponibles (en prenant pour hypothese une contribution par travailleur de 200 hommes-jours par hectare). Ces chiffres sont conformes aux donnees des enquetes susmentionnees selon lesquelles les travailleurs agricoles consacrent une part considerable de leur travail a des activit6s marginales. Deux autres questions se posent a ce sujet : i) Pendant combien de temps la croissance annuelle de la production agricole pourrait-elle etre maintenue au taux de 3,6 % enregistre a Java au cours de la derniere decennie? ii) Les techniques d'economie de main-d'oeuvre qui ont fait leur apparition dans certaines parties de Java, parallelement a la commercialisation accrue de l'agriculture, vont-elles se g6n6raliser, r6duisant ainsi la demande de main-d'oeuvre? Action a envisager 27. Les auteurs du rapport doutent beaucoup que la croissance de l'emploi dans l'6conomie indonesienne, dans les annees 70, ait suivi une voie susceptible de conduire, a la longue, au tournant. Avec une baisse probable des revenus du secteur petrolier et, par consequent, une diminution de la croissance globale pendant le reste de la decennie, les perspectives d'absorption de la main-d'oeuvre sont mauvaises si l'on considere l'evolution des dernieres ann6es. On en d6duit que des mesures devront etre prises pour orienter davantage la croissance vers la cr6ation d'emplois. C'est la une question difficile a laquelle le rapport, qui cherche surtout a analyser la situation actuelle et les perspectives des march6s du travail, n'est pas en mesure de repondre. On peut, neanmoins, tenter de d6finir brievement un certain nombre de cons6quences qui decoulent, sur le plan pratique, de la conclusion fondamentale du rapport. 28. Le montant des depenses publiques et leur repartition peuvent avoir une grande influence sur la demande de main-d'oeuvre. A court terme, les travaux publics qui font appel a une nombreuse main-d'oeuvre, comme les travaux d'infrastructure locale des programmes de 1'INPRES, favorisent directement l'emploi et peuvent aussi drainer les ressources vers les zones rurales relativement pauvres, entrainant la creation d'emplois. Le maintien de ces activit6s sera particulierement important pendant les deux ou trois ann6es a venir, l'Etat devant adopter d'autres - xxxviii - mesures de r6duction des depenses afin de ramener la dette exterieure a des limites plus modestes. I1 faudra donc redistribuer les depenses au sein du programme d'investissements publics, les activit6s a fort 61ement d'importation faisant place a celles qui sont plus fortes consommatrices de main-d'oeuvre. En mai 1983, le Gouvernement a annonce un vaste plan de reamenagement des grands projets d'investissements publics devant permettre de reaffecter les rupiahs 6conomises a des investissements nettement cr6ateurs d'emplois. 29. L'absorption de main-d'oeuvre par l'agriculture a et6 faible dans les annees 70 par rapport a l'accroissement de la production, mais, vu son r6le important en matiere d'emplois, le comportement de ce secteur sera determinant pour l'ensemble du marche du travail. II sera probablement n6cessaire, au cours des ann6es 80, si l'on veut maintenir les taux de croissance passes qui etaient de 4 % par an environ et creer de nouveaux emplois, de diversifier la production agricole en favorisant l'horticulture, quelques cultures vivrieres secondaires, ainsi que l'elevage et la peche a petite echelle, aux depens du riz. En outre, le mouvement de transmigration officielle et spontanee vers les autres iles pourrait avoir une influence considerable. Au rythme actuel d'environ 100.000 familles par an, c'est pratiquement un cinquieme de la main-d'oeuvre javanaise excedentaire qui pourrait ainsi etre absorb6. 30. Les orientations du secteur manufacturier joueront un role de plus en plus important au cours de la prochaine d6cennie et au-dela. Comme il est dit au Chapitre 4, ce secteur a, dans le pass6, absorbe un fort pourcentage de main-d'oeuvre dans les pays rizicoles qui ont reussi. L'Indon6sie fait exception dans la mesure oul ce secteur n'y a represente qu'une infime partie de l'offre totale d'emplois, et encore s'agissait-il surtout de petites industries rurales. Une r6orientation des options sera necessaire pour favoriser l'absorption de la main-d'oeuvre, d'autant que l'Indonesie en a deji presque fini avec la substitution des importations dans les industries de biens de consommation a forte intensit6 de main-d'oeuvre et que, a part la croissance relativement lente de la demande finale, les nouvelles augmentations de production pour le marche interne devront etre surtout le fait des industries de biens de production, qui absorbent beaucoup plus de capitaux. C'est dire la necessite d'une politique industrielle axee sur les exportations, surtout de biens de consommation, par des mesures tendant a reduire la pref6rence accord6e a l'heure actuelle aux substitutions d'importations et par des mesures destin6es a encourager les exportations. II y aura peut-etre a se preoccuper aussi du d6veloppement du petit secteur manufacturier, dont la production peut se ressentir de la concurrence des secteurs a moyenne et grande echelles plus productifs. On n'a pas cherch6, ici, a 6tudier les mesures de soutien aux petites industries ou le prix a payer pour proteger des sous-secteurs qui sont gros consommateurs de main-d'oeuvre. Les prochaines analyses devraient porter sur la question du grand ecart de salaires entre les petites et les grandes entreprises, comme il est dit au Chapitre 3. Il peut y avoir un choix 6conomique a faire entre l'option consistant a employer un petit nombre de travailleurs a un salaire relativement elev6 et celle consistant a employer une main-d'oeuvre nombreuse a un moindre salaire. - xxxix - 31. Enfin, les prix du marche des facteurs ne permettent pas de determiner correctement les besoins de main-d'oeuvre pour tel ou tel projet d'investissement. I1 faudra donc trouver des mecanismes d'intervention appropries, notamment comme on l'a deja dit, l'utilisation appropri6e du salaire de r6ference dans l'evaluation des projets d'investissements publics. On pourrait, a ce propos, souligner un point qui a son importance. Dans les 6tudes sur l'evaluation des projets, le prix de ref6rence de la main-d'oeuvre est generalement fixe a un stade relativement avanc6 du processus de d6cision, lorsque l'utilisation du prix comptable approprie n'intervient plus que marginalement dans le choix entre telle ou telle option. Aussi, 1'6tude directe des diverses options sectorielles doit-elle venir completer l'utilisation du salaire de r6ference si l'on veut donner plus d'importance a une strategie de d6veloppement ax6e sur l'utilisation de la main-d'oeuvre. 32. I1 est un facteur non moins important - et peut-&tre meme, davantage - qui fait des prix du march6 un indice trompeur pour le choix des techniques: la sous-evaluation du capital, du moins dans certains grands secteurs de l'economie. Ainsi, l'achat de tracteurs peut paraitre rentable au prix du marche, mais il peut ne pas etre socialement productif si l'on utilise, pour la main-d'oeuvre et le capital, des prix comptables appropri6s 1/, sauf si des recherches precises ont permis de d6celer des poches de penurie de main-d'oeuvre dans certains domaines. Travaux ulterieurs sur l'emploi et la remuneration du travail en Indonesie 33. Bien qu'il insiste sur le fait qu'un modele de marche du travail fond6 sur l'hypothese qu'il existe un exc6dent de main-d'oeuvre dans le secteur rural (et sur l'idee connexe que le secteur du riz est un secteur privilegie pour les chercheurs d'emploi) permet de comprendre le d6veloppement de l'Indon6sie, le rapport ne conclut pas pour autant que ce modele ait ete definitivement teste et verifie. D'autres travaux sont necessaires, et d'autres hypotheses sont ai envisager tant au niveau micro6conomique qu'au niveau macroeconomique. L'Appendice A du Chapitre 3 presente divers micromodeles de marche de la main-d'oeuvre rurale et examine dans quelle mesure les donnees empiriques existantes justifient le choix du modele fonde sur l'existence d'un "excedent" de main-d'oeuvre. Dans l'Annexe au rapport, on cherche a savoir si les grandes tendances macroeconomiques du developpement indon6sien sont conformes au cadre des modeles n6oclassiques globaux et a etablir pour les parametres cl6s, des fourchettes de valeur qui soient conformes aux faits. On examine, dans ce contexte, la pertinence du modele fonde sur l'existence d'un exc6dent de main-d'oeuvre pour l'interpr6tation des grandes tendances. 1/ Voir Rudolph S. Sinaga, "Implications of Agricultural Mechanization for Employment and Income Distribution", Rural Dynamics Study Series, No 2, Bogor. - xl - 34. Le rapport recommande d'autres travaux pratiques pour observer directement le fonctionnement des march6s du travail. Pour les zones rurales, voici quelques points importants qu'il conviendrait d'approfondir : i) La demande de main-d'oeuvre agricole, a Java, d6pend de la nature de la saison active par rapport a la morte-saison. Combien de temps dure la saison active et quels en sont les effets sur tel ou tel aspect de la main-d'oeuvre? Quelle est la proportion du temps de travail consacree au secteur productif (du riz) et aux activites marginales, par principaux groupes d'age/sexe classes en fonction de leur situation economique? Les salaires horaires dans le secteur marginal sont-ils nettement inf6rieurs a ce qu'ils sont dans le secteur du riz pendant la saison active, comme ils le sont pour l'ensemble de l'ann6e? Les 6tudes qui ont ete faites sur les villages fournissent pas mal de reponses a ces questions, mais, ou bien ces donn6es n'y ont pas 6t6 analys6es, ou bien elles ne se pr6sentent pas sous une forme qui permette de degager une r6ponse. Une 6tude plus attentive des microdonn6es d6ja recueillies pour les diff6rents villages s'impose donc. ii) Naturellement a Java, comme dans toutes les soci6t6s rurales d'Asie, les conditions des marches du travail varient beaucoup d'une region a l'autre ou meme d'un village a l'autre. Il pourra arriver, sans doute, que le modele de l'exc6dent de main-d'oeuvre s'applique a certains villages plus qu'a d'autres. C'est la un point qu'il ne faudra pas perdre de vue dans l'analyse ulterieure des donnees relatives aux villages. Meme s'il n'existe pas assez d'6tudes detaill6es de villages, il est probable qu'une analyse crols6e des variations de salaires entre villages pour un certain nombre d'occupations ainsi que de leur rapport avec les conditions economiques fondamentales des villages serait tout a fait rev6latrice. iii) Au niveau macro6conomique, le rapport souligne la necessit6 de poursuivre l1'tude des mouvements salariaux, en particulier dans l'agriculture. Ici encore, on ne saurait trop attacher d'importance a une analyse r6gionale des tendances observees en matiere de salaires et d'emploi. Les donnees du recensement de 1980 devraient, en utilisant conjointement d'autres sources de statistiques, y compris le recensement de 1971, constituer une mine d'informations sur les tendances regionales de l'emploi au cours des annees 70. iv) Tout aussi importants, comme source d'information, sont le niveau et l'evolution des revenus dans les tres grands secteurs non structur6s, en particulier celui des services. Une analyse plus approfondie des enquetes sur les m6nages apporterait, sans doute, de nouvelles lumieres sur la question; mais les Indon6siens ayant souvent plusieurs occupations, en particulier dans le secteur rural, il importe de formuler soigneusement les questionnaires a utiliser pour les futures enquetes sur les m6nages. I1 faudrait s'efforcer de recueillir des renseignements sur le nombre d'heures de travail consacrees a chacune et la r6mun6ration correspondante au lieu de se borner a poser des questions sur l'activit6 principale. - xli - v) I1 est important aussi d'aller au-dela des rapports entre salaires, emploi et production, tels qu'ils sont exposes ici. La terre et le credit, par exemple, sont des facteurs cles, mais il n'a pas ete possible, vu les donnees dont on disposait et l'6tat des connaissances sur la question, d'etudier syst6matiquement les rapports entre les marches de facteurs de production (terre, capital, main-d'oeuvre) et le role que le cr6dit et le financement, pris au sens large, jouent dans ces rapports. La collecte de nouvelles donn6es statistiques au niveau des villages, que le pr6sent rapport recommande, devrait viser a faciliter l'etude des rapports entre les marches de facteurs de production et le r6le du credit. - xlii - RESUMEN Y CONCLUSIONES Introducci6n 1. El presente informe es parte de la evaluaci6n que viene realizando el Banco sobre los salarios y el empleo en Indonesia. En el ultimo informe especializado 1/ se examinaron los mercados laborales y la distribuci6n del ingreso hasta 1976 inclusive. El informe actual se concentra solamente en las condiciones del mercado laboral (las limitaciones de recursos impiden hacer un analisis en profundidad de los factores asociados de la pobreza y la distribuci6n del ingreso). Los datos basicos incluyen las encuestas trimestrales de la fuerza laboral de 1977, 1978 y 1979, asi como los resultados preliminares del censo de poblaci6n levantado en 1980. De acuerdo con este dltimo, la poblaci6n de Indonesia creci6 en el 2,3% por anio durante la decada de 1970, en lugar del 2% que se habia previsto. Por consiguiente, se espera que las adiciones a la fuerza laboral alcancen a un total de aproximadamente quince millones de habitantes en el curso de la pr6xima decada, en contraposici6n a los veinticinco millones registrados en los dos uIltimos decenios. Asi, pues, es imprescindible observar y evaluar regularmente los mercados laborales y los factores que influyen en ellos. 2. Otro factor que se suma a esta urgencia es la desaceleracion que se preve en el crecimiento de la producci6n en la decada de 1980. ZC6mo resultarAn afectados el empleo y los ingresos por una oferta laboral en rApida expansi6n, combinada con un crecimiento mas lento de la producci6n, y c6mo variarain con diversos supuestos acerca de c6mo funcionan los mercados laborales? Tales puntos de vista diversos tienen, a su vez, diferentes repercusiones en la formulaci6n de politicas. El objetivo primordial de este informe es senalar esas repercusiones y promover el estudio adicional de los mercados laborales en Indonesia, incluido un nuevo examen de la utilizaci6n de la mano de obra y del ingreso de 6sta, en particular en las zonas rurales de Java. 3. En el informe anterior se asumia el punto de vista de que los mercados laborales en Indonesia funcionan de manera eficiente. En 6l se llegaba a la conclusi6n de que, en conjunto, dadas las variaciones observadas en los salarios (en el curso del tiempo y entre las regiones) y la movilidad de la mano de obra, las fuerzas de la oferta y la demanda proporcionan una explicaci6n suficiente del nivel y la estructura de los salarios y el empleo y, en consecuencia, no habia desempleo a largo plazo 1/ Banco Mundial (1980), 2788-IND. - xliii - ni estructural en Indonesia. Quienes suministraban servicios de mano de obra estaban abaratando los salarios a un nivel al que todo el mundo estuviera empleado, a un salario que reflejaba la productividad de la mano de obra y el costo del esfuerzo, en el margen. Los salarios eran bajos, pero ese era el resultado de la falta de insumos complementarios (capital y aptitudes) y no la falla del mercado en cuanto a reflejar el "costo de oportunidad" de la mano de obra, el salario que los trabajadores podian obtener en el margen, en ocupaciones opcionales. Ademas, en el informe anterior tambi6n se indicaba que las diferencias observadas en los salarios era probable que fuesen consecuencia de las diferencias en la calidad de la mano de obra suministrada. 4. En el contexto que acaba de exponerse, no habria problema de empleo per se en Indonesia, ni tampoco necesidad de formular politicas separadas dirigidas de manera especifica hacia la creaci6n de empleos. Asimismo, los proyectos de inversi6n puiblica (en particular en la zona rural de Java) deberian evaluarse utilizando un precio para la mano de obra pr6ximo al salario vigente del mercado, es decir, la relaci6n entre salarios de cuenta y de mercado seria cercana a la unidad. 5. En el presente informe se asume un punto de vista diferente. En primer lugar, se pone en tela de juicio la explicaci6n de que las diferencias salariales en los mercados urbanos de la mano de obra reflejan diferencias en la calidad de esta. Se elabora una escala de salarios para la mano de obra no especializada y las diferencias salariales observadas parecen demasiado elevadas para que se puedan explicar por diferencias en calidad dentro de la categoria de mano de obra no especializada. En segundo termino, en el documento se indica que las bajas tasas de participaci6n de hombres j6venes y de edad en las zonas urbanas, aunadas a las elevadas tasas de desempleo de la juventud urbana, se traducen en una subutilizaci6n sustancial de la gente joven, en particular de la instruida, y no se puede explicar mediante modelos de desempleo en los que se pone de relieve la bAsqueda de empleo transicional. En tercer lugar, en el informe se argumenta que las fuerzas sociales y econ6micas, que no promueven una compensaci6n del mercado laboral, desempenian una funci6n amplia en la determinaci6n de los salarios. La variabilidad de 6stos en terminos de arroz, en el curso del tiempo o entre poblados ocurre dentro de una faja angosta que se mantiene estable en el tiempo. La constancia del salario agricola dentro de esta faja en Indonesia es similar a los salarios agricolas estables observados en el Jap6n y Corea antes del agotamiento del excedente de mano de obra rural. En este informe se formula, como hip6tesis de trabajo, que en muchos poblados javaneses hay un excedente de mano de obra: a los salarios vigentes hay mas gente capaz de trabajar en el cultivo del arroz y dispuesta a hacerlo de la que se puede emplear, tanto en la temporada punta como fuera de ella. En el mercado rural de la mano de obra arrocera, al igual que en los mercados urbanos laborales, hay mecanismos que impiden que tal excedente de mano de obra abarate los salarios. 6. Esas tres caracteristicas de los mercados laborales indonesios tienen importantes consecuencias en materia de politica. Ponen de relieve la necesidad de formular politicas que aborden la cuesti6n del rapido - xliv - crecimiento previsto de la fuerza laboral y de la prevalencia de la productividad relativamente baja de la mano de obra. El presente informe, por consiguiente, hace notar la necesidad de considerar la creaci6n de empleos como un objetivo separado de politica (en contraposici6n a incluirlo bajo politicas en las que se promueve el crecimiento maximo de la producci6n). Del informe se desprende, asimismo, que en la evaluaci6n de proyectos de inversi6n publica (en particular en la zona rural de Java) no debe valorarse la mano de obra no especializada al salario vigente en el mercado para el cultivo de arroz, sino al rendimiento mas bajo observado en las actividades marginales en la parte inferior de la escala econ6mica. 7. Las pruebas en que se apoya la interpretaci6n de este informe del mercado laboral provienen principalmente de estudios de poblados y de encuestas de la fuerza laboral. Los estudios citados son de importancia fundamental para mejorar los conocimientos acerca de la evaluaci6n de la economia javanesa en particular. En el informe se intenta utilizar las pruebas que aportan esos estudios para interpretar las tendencias generales en materia de empleo y producci6n a partir de datos agregados. Ese intento debe llevarse mas adelante. Desempleo, tasas de participaci6n y educaci6n 8. En el Capitulo 1 se presenta una visi6n global del desempleo manifiesto (es decir, medido) en Indonesia. Al igual que ocurre en muchos paises en desarrollo, este es un problema que se plantea en las zonas urbanas, en particular entre la poblaci6n joven. Las tasas de desempleo medido son de tres a cuatro veces mas elevadas en el sector urbano que en el rural. Dentro del sector urbano es del 25% para los varones de 15-19 ainos de edad, y del 18% en lo que se refiere a los de 20-24 anios (comparadas con un promedio global de alrededor del 7% para los varones del sector urbano). 9. Ahora bien, el desempleo medido no refleja en medida adecuada las dificultades reales que encuentra la juventud del sector urbano en Indonesia para obtener empleo. Seguin las encuestas relacionadas con la fuerza laboral, entre los que declararon hallarse sin empleo, la duraci6n de la falta de ocupaci6n no fue larga, ya que en la mayoria de los casos fue de un mes o menos. En contraste, la lentitud del ingreso en el mercado laboral urbano la revelan en forma que llama la atenci6n las elevadas tasas de falta de participaci6n entre la juventud del sector urbano. Pareceria, entonces, que la gente joven no busca empleo en forma activa hasta que es probable que en un futuro cercano haya disponible un empleo aceptable. Las tasas de participaci6n fueron del 34% en el caso de los varones del sector urbano de 15-19 anios de edad y del 75% entre los de 20-24 ainos. En lo que se refiere a las mujeres, las tasas correspondientes fueron del 23% y del 29%, respectivamente, porcentaje bastante mas bajo del que se observa en el sector rural. La magnitud del problema se puede comprender si se hace notar el dato estadistico de que incluso en el grupo de edad de 20-24 anios, el 40% de los hombres del - xlv - sector urbano y el 76% de las mujeres del mismo sector no estaban empleados (es decir, se encontraban desempleados o bien no figuraban en la fuerza laboral). Los datos de SAKERNAS con respecto a la ocupaci6n de los no participantes muestran que hasta el 20% de los varones del sector urbano de ese grupo de edad declar6 que estaba asistiendo a la escuela, en tanto que la tasa oficial de matricula en instituciones de nivel terciario en esa fecha (1977/78) era de s6lo el 2%. Parecia, asi, que muchos de los no participantes eran en realidad trabajadores desalentados que estaban utilizando la educaci6n no formal para excusar su inactividad. 10. Una cuesti6n de cierta importancia es la relaci6n existente entre la subutilizaci6n de la juventud del sector urbano y la expansi6n educacional en Indonesia. Por los datos que se presentan en el Capitulo 1 es manifiesto que la incidencia del desempleo evidente y la falta de participaci6n aumenta con los niveles educacionales, tomando como referencia el grupo de edad. La tasa de participaci6n de los hombres del sector urbano de 20-24 anios de edad, por ejemplo, es de alrededor del 60% con respecto a los que han recibido educaci6n secundaria, y del 90% aproximadamente para los que tienen educaci6n primaria. Debe recordarse a este prop6sito que la ensenianza secundaria debe terminarse bastante antes de los 20 anios y, en realidad, tal edad tardia para terminar la escuela ya constituye en si un aspecto de la falta de oportunidades de empleo para la juventud. Las cifras relativas a las tasas de desempleo manifiesto revelan que una proporci6n apreciable de la fuerza laboral con educaci6n secundaria auin sigue en busca de un empleo ya bien cumplidos los veinte afios. Al mismo tiempo hay escaseces para categorias especificas de mano de obra instruida, por ejemplo, los trabajadores tecnicos que se necesitan en la agricultura. Es evidente que se precisa investigar mas con objeto de identificar campos de escasez y de excedente potencial dentro de la gama educacional. Tendencias generales en materia de producci6n, empleo y salarios, 1961-80 11. El informe contiene datos agregados completos y detallados acerca del crecimiento de la economia indonesia durante los periodos de 1961-71 y 1971-80. Se presta atenci6n particular a la tasa de crecimiento del empleo (medida por el nulmero de personas empleadas), la cual se estima, utilizando la mejor informaci6n disponible, en el 2,9% por anio en el curso de 1971-80, lo que representa un aumento con respecto a la del 2,4% registrada durante la d6cada de 1960. Esas tasas caen dentro del extremo inferior de las de crecimiento plausible dadas en el informe anterior del Banco Mundial sobre el empleo. En contraste, en informes econ6micos previos del Banco se ha tomado la tasa mas elevada en ese intervalo, 4,7% 1/. Si bien la estimaci6n que se presenta aqui esta 1/ V6anse los documentos del Banco Mundial (1979), 2093-IND y Banco Mundial (1980), 2788-IND. En el documento Banco Mundial (1981), 3307-IND, la tasa de crecimiento del empleo que se cita en 6l es del 3,3% anual. En el documento Banco Mundial (1982), 3795-IND, pAg. 95, la tasa estimada es del 2,65%. - xlvi - sujeta a revisi6n una vez que se haya analizado a cabalidad la informaci6n del censo demografico de 1980, pareceria que esa tasa del 4,7% es elevada en grado inverosimil. La aceleraci6n en el crecimiento del empleo (y de la fuerza laboral) durante la d6cada de 1970 esta asociada con el incremento en la proporci6n de la poblaci6n en edad de trabajar, y no con un incremento en las tasas de participaci6n de la fuerza laboral. 12. El crecimiento de la producci6n (7,3% anual) estuvo bastante por encima de esa tasa revisada de crecimiento del empleo durante toda la d6cada pasada. Pero el examen de las tendencias de los salarios en los pocos sectores organizados con respecto a los cuales se pudieron obtener datos --agricultura, plantaci6n y construccion-- indica que los salarios reales correspondientes a los trabajadores no especializados fueron constantes durante gran parte de ese periodo, pese al raipido crecimiento de la producci6n. 13. El informe se concentra en la cuesti6n del rapido crecimiento de la producci6n a salarios constantes como una de las caracteristicas importantes de la expansi6n macroecon6mica de Indonesia. Una posible explicaci6n de este fen6meno en el marco de los mercados laborales competitivos y de los salarios de compensaci6n de mercado es que el crecimiento del empleo "verdadero" (el empleo medido en terminos de unidades de eficiencia, despues de introducir ajustes para tener en cuenta la acumulaci6n de aptitudes, mAs bien que en t6rminos de personas empleadas) excede del crecimiento del empleo medido. Otra es que el progreso tecnico economizador de mano de obra ha sido suficiente como para permitir el rapido crecimiento de la producci6n con salarios constantes. Estas posibilidades se examinan en el Anexo A del presente informe. La conclusi6n que cabe inferir del anAlisis formal del primer argumento es que el tener en cuenta la acumulaci6n de aptitudes no modifica el cuadro global. Los resultados numericos derivados del modelo de crecimiento con progreso tecnico indican los supuestos especificos acerca de los valores de los parametros que harian compatible la tasa observada de crecimiento de la producci6n con los salarios constantes en el curso del tiempo. 14. En el informe se expone que para interpretar las tendencias observadas en la economia en cuanto a productividad y salarios seria util un marco basado en la disponibilidad de mano de obra abundante a las tasas de salarios vigentes en los principales sectores de la economia (incluido el cultivo del arroz con cascara). Pero, a diferencia de algunas versiones del modelo de mano de obra excedente, no podemos aceptar la hip6tesis de que la mano se encuentra en una situaci6n de oferta perfectamente elastica a un salario dirigido hacia la productividad media del trabajo en la agricultura. Dentro del sector agropecuario la producci6n creci6 a una tasa anual del 3,6%, por encima en medida apreciable de la tasa de crecimiento del empleo en ese sector (1% - xlvii - anual) 1/. En el informe se utilizan algunos microestudios intensivos de poblados javaneses a fin de examinar los mercados laborales rurales y se sugieren algunas hip6tesis relacionadas con la determinaci6n de los salarios agricolas. Este intento de vincular un microestudio del mercado laboral (en el Capitulo 3) al macrocuadro de la economia (en el Capitulo 2) es s6lo un esfuerzo preliminar. Es menester hacer un trabajo mucho mas amplio en estudios de poblados y en la recopilaci6n de datos. Al final de este resumen se indican algunos de los puntos mias especificos para investigarlos detenidamente. Los mercados laborales indonesios: Interpretaci6n de su funcionamiento 15. Los dos hechos importantes convencionales realzados por los microestudios de poblado efectuados en Java son los siguientes: a) El numero de horas-hombre empleadas en el principal mercado laboral agricola --el del arroz-- es s6lo una pequenia parte del total del tiempo laboral suministrado al mercado incluso por unidades familiares sin tierras 2/. En forma caracteristica, la proporci6n del tiempo total laboral utilizado en el trabajo agricola asalariado es del 30% al 40% para los trabajadores sin tierras. El resto del tiempo laboral se dedica a diversas actividades de comercio, servicios, artesanias y, en ocasiones, al trabajo asalariado no agricola fuera de los poblados. 1/ Si no se considera la mineria, las tasas globales de crecimiento de la producci6n y el empleo, respectivamente, son como sigue: sector agropecuario (3,5; 1,0), secundario (12,9; 4,9), terciario (8,3; 6,1). El sector secundario incluye manufacturas, construcci6n, transporte y servicios publicos. El sector terciario incluye comercio y servicios. Los porcentajes de empleo total al final del periodo en los tres sectores fueron 56, 15 y 29, respectivamente. En el Informe del Banco Mundial (1981) 3182-IND, Anexo 1, pag. 35, se manifiesta que en las empresas manufactureras grandes y medianas el empleo creci6 entre el 6% y el 12% durante la decada de 1970. Pero esas empresas cubrieron s6lo el 13% del total del empleo en el sector manufacturero en 1974/75. 2/ Aqui, y en el resto del informe, la palabra "arroz" se utiliza como termino generico para el sector agropecuario productivo que utiliza el trabajo asalariado en escala extensiva. El termino incluye cultivos de plantaci6n, asi como cultivos no arroceros importantes en la economia campesina, como el de la cania de azucar. - xlviii - b) Los rendimientos de la mano de obra (por hora-hombre) son sustancialmente mas bajos en actividades ajenas al mercado laboral del arroz. Los estudios realizados sobre el terreno indican que estos podrian ser del orden del 30% de la tasa de salarios prevaleciente en el mercado del arroz para grupos similares de edad y sexo. Debido al rendimiento mas bajo para la mano de obra en actividades diversas (en gran parte de empleo por cuenta propia), en las paginas siguientes se las clasifica como el "sector marginal". Obviamente, una mayor proporci6n del tiempo se dedica a actividades marginales en la temporada de poco movimiento, pero incluso en la estaci6n dinamica de cultivo del arroz con cascara, esas actividades consumen una porci6n significativa del tiempo laboral de trabajadores adultos de ambos sexos. Esto se debe en parte a que los mercados laborales arroceros son especificos de poblados individuales, y en cualquier poblado particular la demanda punta de mano de obra para el cultivo del arroz se limita a dias en lugar de meses. Es menester realizar un trabajo mAs amplio con respecto a la diferencia estacional en los rendimientos de la mano de obra entre el sector arrocero y las actividades marginales. No todos los estudios hechos sobre el terreno informan acerca de la magnitud de la diferencia por temporada, pero la diferencia da la impresi6n de que aparece s6lo en los estudios que se confinan a la temporada punta 1/. 16. Puede decirse, asi, que en la zona rural de Java existe excedente de mano de obra en el sentido limitado y particular de que hay un conglomerado de unidades laborales (en lugar de trabajadores) en actividades marginales con ingresos por hora apreciablemente menores que la tasa salarial arrocera. La importancia macroecon6mica de este modelo de mercado laboral es que ese conglomerado es potencialmente asequible al sector arrocero al salario arrocero prevaleciente (mis alto). Dada esta oferta perfectamente elastica de mano de obra para el sector arrocero, la producci6n por trabajador puede incrementarse en arroz durante un periodo sin que se ejerza presi6n en las tasas de salarios aun cuando no haya cambios tecnicos que economicen mano de obra. Entonces se plantea la cuesti6n: ipor que los terratenientes que buscan obtener utilidades pagan a la mano de obra que trabaja en sus campos un salario mas alto del que perciben los trabajadores en las actividades marginales? Es probable que la respuesta estribe en la relaci6n de eficiencia en funci6n del salario, es decir, el efecto de incentivo que ejercen los salarios altos en la intensidad del esfuerzo de los trabajadores. A partir de un rendimiento bajo para la mano de obra empleada por su propia cuenta en actividades 1/ En la mayoria de las economias de poblado habrA, en parte, una escala de ingresos tanto en las actividades arroceras como en las marginales. El modelo de nivel doble con diferentes ingresos medios es un hecho convencional que subraya la importancia de las actividades marginales como dep6sito del "excedente" de mano de obra. - xlix - marginales, la eficiencia de un trabajador se incrementa en grado mas que proporcional a medida que aumenta el salario ofrecido, hasta llegar a un punto de inflexi6n, pasado el cual la eficiencia se incrementa en escala menor que proporcional. Por consiguiente, hay un nivel de salario --mas alto que el rendimiento de la mano de obra en actividades marginales-- al cual el costo salarial por unidad de mano de obra (en unidades de eficiencia) se minimiza desde el punto de vista de los empleadores. Ningun empleador que busque la maximizaci6n de utilidades ofrecera un salario mAs bajo que 6se, aun cuando haya una oferta abundante de mano de obra a ese salario. Pudiera parecer por este argumento que de hecho se establece el salario mas alto para atraer mano de obra mas eficiente. Pero la cuesti6n fundamental es que la direcci6n de la causalidad va del salario a la eficiencia, en lugar de ser a la inversa. A un salario mas alto (hasta el punto de inflexi6n) a todo el mundo, mas o menos, que se le ofrece un empleo lo desempenia a un nivel de eficiencia proporcionalmente mas alto. Toda vez que hay una oferta abundante de mano de obra al salario minimo (lo cual minimiza el costo salarial), los empleadores deben seleccionar algunos trabajadores de preferencia a otros. En los estudios sobre el terreno llevados a cabo en la zona rural de Java se ha mencionado en forma constante la importancia del sistema de la red social que establece reglas de contrataci6n preferencial para los trabajadores empleados en el sector arrocero. 17. Ya se ha expuesto que los mercados laborales rurales en Java (al igual que en otros paises asiaticos) son especificos de los poblados. Asi, el salario de eficiencia que establece el salario minimo en las operaciones agricolas esta vinculado a la norma correspondiente al poblado particular y refleja sus condiciones econ6micas. En el sector rural se observan variaciones entre poblados en los salarios agricolas con respecto a la misma temporada u ocupaci6n. Por lo tanto, la constancia del salario real en el curso del tiempo significa que la banda de tasas salariales correspondientes a una ocupaci6n particular no se ha desplazado de manera perceptible en sentido ascendente, pese al incremento en producci6n por trabajador. No esperamos un movimiento alcista de la banda hasta que la demanda de crecimiento de la mano de obra agricola a una tasa suficientemente elevada, en relaci6n con la oferta, cause una reducci6n significativa en el conglomerado de mano de obra que se ocupa en actividades marginales. 18. Los mercados laborales urbano y rural estan vinculados por dos tipos diferentes de migrantes: los circulares que de manera temporal entran a las ciudades y salen de ellas sin sus familias, y los inmigrantes mas permanentes que llegan a las zonas urbanas con sus familias. Los primeros se encuentran en gran parte en el sector no estructurado del mercado laboral urbano y, aunque los datos acerca de los ingresos en ese sector son muy deficientes, podemos esperar que la migraci6n circular mantenga razonablemente cercanos los rendimientos de la mano de obra en el sector no estructurado urbano y en las actividades marginales rurales, despues de tener en cuenta las diferencias en el costo de vida y los costos de transporte. Fuera del sector no estructurado, puede esperarse que la relaci6n de eficiencia en funci6n del salario se conserve tanto para el empleo asalariado urbano como para la mano de obra contratada en la agricultura. En realidad, una hip6tesis razonable es que la relaci6n entre los salarios y la eficiencia sera mas firme cuanto mayor sea la magnitud del empleo y mas compleja la organizaci6n de la unidad de contrataci6n. Se ha observado en muchos paises en desarrollo (incluido el Jap6n durante su periodo de desarrollo) que los salarios en el sector manufacturero de la mano de obra no especializada aumentan con el tamaiio de la empresa, aun cuando no haya sindicatos ni legislaci6n gubernamental. Indonesia no es diferente de la norma general de comportamiento del salario urbano. En el Capitulo 3 se presentan pruebas de que hay una escala de salarios para los trabajadores no especializados en las entidades manufactureras urbanas en que las empresas pequenias no mecanizadas se encuentran en el escal6n inferior y las "empresas conjuntas" o multinacionales figuran en la parte superior. El margen salarial entre ambas llega hasta el 250%. El salario de cuenta 19. El modelo del mercado laboral anterior implica que el salario de cuenta para los proyectos rurales debe calcularse sobre la base del costo de oportunidad de emplear una unidad de mano de obra no especializada, es decir, el rendimiento de la mano de obra en las actividades rurales marginales. Los estudios realizados sobre el terreno indican que en Java ese rendimiento puede ser de apenas un tercio del salario agricola arrocero. Teniendo en cuenta el costo social del mayor consumo por la unidad familiar representativa rural que suministra la mano de obra, se calcul6 que el salario de cuenta seria el 28% del salario agricola. Seguin se examina abajo, y en el texto principal, esta declaraci6n tiene que ser condicionada por el hecho de que los datos acerca de los ingresos estacionales son todavia muy escasos en Indonesia. Perspectivas en relaci6n con el empleo y los salarios 20. Fundamentado en tendencias hist6ricas, en el informe se proyecta un crecimiento de la fuerza laboral de alrededor del 2,6% anual hasta 1990. ZSera absorbido este incremento de la mano de obra por la mayor demanda en el curso de la pr6xima d6cada, y a que nivel salarial? Esta es la cuesti6n central que se plantea a prop6sito de las perspectivas de crecimiento del empleo en Indonesia. En el presente informe se ha utilizado el concepto del punto de inflexi6n (relacionado con el agotamiento de un "excedente" de mano de obra, tal como se define en la pagina 11) al examinar las perspectivas de empleo en Indonesia. 21. Las pruebas hist6ricas relativas a dos economias arroceras asiaticas --el Jap6n y Corea-- examinadas en el Capitulo 4 muestran que ambas economias tuvieron un periodo de crecimiento sostenido de la producci6n por trabajador en la agricultura, con escaso o ningun crecimiento real de los salarios reales en ese sector. Ese periodo de crecimiento a salarios casi constantes (en la agricultura) se prolong6 en - li - el caso del Jap6n (50 aiios o mas) incrementandose la producci6n por trabajador s6lo a una tasa moderada, y fue sustancialmente mas breve en lo que se refiere a Corea, con un crecimiento mucho mas rApido de la productividad agricola durante el periodo de transici6n hasta el punto de inflexi6n. Pero 6ste, en ambos casos, se identifica como un hecho hist6rico, como un "retorcimiento" claramente reconocible en la tendencia temporal de los salarios reales de la agricultura. Despu6s de un periodo relativamente largo durante el cual los salarios reales crecieron con mucha lentitud, 6stos se incrementaron a un ritmo de seis a diez veces mayor que el observado con anterioridad. 22. La significaci6n del punto de inflexi6n en lo relativo al bienestar es que antes de que se liegue a 61, la parte de la producci6n que va a los salarios desciende en forma constante debido a que la productividad de la mano de obra crece con mAs velocidad que los salarios. Dado que la mecanizaci6n no es un factor de importancia en la agricultura asiatica antes del punto de inflexion, gran parte de la diferencia entre la productividad de la mano de obra y los salarios se refleja en un incremento en la proporci6n de las rentas y las utilidades. El segundo aspecto importante de la "filtraci6n gradual" del incremento de la productividad agricola es determinar si la economia se encuentra en un sendero de crecimiento que tal vez llegue pronto al punto de inflexi6n. Si la economia se halla en tal periodo de transici6n, una mayor proporci6n de la mano de obra de los trabajadores que dependen de la agricultura como su principal ocupaci6n se desviarA de las actividades marginales hacia la agricultura. Al percibir en 6sta ingresos por hora mAs elevados, la remuneraci6n por trabajador aumentarA con el tiempo aun cuando las tasas de salarios se mantengan constantes en la agricultura. Asimismo, en el periodo de transici6n los dos paises examinados en el Capitulo 4 --el Jap6n y Corea-- experimentaron una acentuada desviaci6n de la agricultura hacia las manufacturas donde los salarios eran mas elevados (reduci6ndose de hecho el nuimero absoluto de trabaiadores en la agricultura durante partes del periodo de transici6n). Tal reasignaci6n de la mano de obra tambi6n incrementa los ingresos por trabajador en el tiempo para la economia en conjunto, hasta el punto de inflexi6n, incluso con tasas constantes de salarios reales en los sectores individuales. 23. Las pruebas presentadas en este informe indican que Indonesia no ha pasado todavia tal punto de inflexi6n. No hay pruebas de presi6n sustancial alguna de signo ascendente sobre los salarios reales, pese a un apreciable incremento en la producci6n por trabajador en la agricultura durante la d6cada pasada. ePero hay alguna prueba de que Indonesia (o mas concretamente Java) se haya encontrado en un sendero de crecimiento que este acercando a la economia al punto de inflexi6n? Las pruebas acerca de esta pregunta son mucho mas dificiles de evaluar. No poseemos datos acerca del nunmero de horas trabajadas en la agricultura ni en las actividades marginales con respecto a trabajadores cuya actividad primaria es la agricultura. Desconocemos, pues, si la proporci6n de mano de obra dedicada a la agricultura se ha venido reduciendo con el tiempo, lo cual ocurriria si la economia estuviera acercandose al punto de inflexi6n. - lii - Pero al limitar nuestra atenci6n a la ocupaci6n principal de la fuerza laboral, observamos que la fuente principal de empleo adicional para la creciente fuerza laboral la han constituido las actividades terciarias. La peculiaridad del crecimiento econ6mico indonesio en la d6cada pasada ha sido la relativa insignificancia del sector manufacturero en cuanto a proporcionar empleo adicional, en tanto que el crecimiento anual del n6unero de trabajadores que dependen primordialmente de la agricultura se ha elevado en grado sustancial. Esta modalidad de cambio en la asignacion sectorial de la mano de obra en el curso del tiempo ofrece un marcado contraste con la experiencia hist6rica del Jap6n y Corea durante sus periodos de transici6n hasta sus respectivos puntos de inflexi6n. En Indonesia la relaci6n entre el empleo terciario incremental y el de la manufacturaci6n ha sido de 4:1, en tanto que en las dos economias arroceras "pr6speras" fue entre 1:1 y 2:1. De manera anAloga, el Jap6n y Corea tuvieron salidas significativamente mas elevadas de mano de obra de la agricultura, aun cuando el crecimiento de la producci6n en ese sector fue sustancial. 24. El estudio hist6rico comparado indica que el sendero de crecimiento de Indonesia en la decada pasada no estA conduciendo hacia el punto de inflexi6n. Hay dos argumentos, sin embargo, que pudieran debilitar la fuerza de esta conclusion. Primero, Indonesia, a diferencia de los demAs paises, es una economia rica en petr6leo, y el efecto multiplicador del gasto del sector puiblico (financiado por los ingresos petroleros) pudiera ser particularmente vigoroso en la creaci6n de empleo en actividades terciarias. Segundo, hay la posibilidad de que Indonesia est6 mas alejada de los margenes extensivos e intensivos de cultivo que las demAs economias, de modo que hay mas Ambito para la absorci6n productiva de la mano de obra en la agricultura. 25. Ambos argumentos merecen indagarse mas a fondo. En el presente informe no pueden darse respuestas definitivas con base en los datos existentes. En lo que se refiere al primer punto, hay la acentuada presunci6n de que la cuantiosa entrada de recursos a traves del sector petrolero dio lugar directa e indirectamente a un crecimiento muy elevado en los ingresos. Una vez que se tienen en cuenta los efectos de la relaci6n de intercambio, el ingreso nacional se elev6 en alrededor del 102 anual en la decada de 1970, y esto sin duda alguna hubiera dado lugar a una demanda sumamente dinamica de servicios. Sin embargo, el impacto final causado en el mercado laboral no esta claro. Ya se ha mencionado que se necesitan con carActer de urgencia mas datos --tanto estaticos como intertemporales-- en lo que se refiere a los ingresos de la mano de obra en el sector terciario. El estudio de las tendencias de los ingresos relativos en el sector terciario y en sus importantes subgrupos ayudaria a revelar si la mano de obra fue "empujada" al seno de este sector por la falta de oportunidades en otros sectores. Por el momento el informe s6lo puede senialar a la atenci6n la relaci6n del incremento en el empleo terciario comparado con el empleo en el sector manufacturero en Indonesia, el que es extraordinariamente elevado comparado no s6lo con el Jap6n y Corea durante sus periodos de transici6n, sino tambien con otros paises en - liii - desarrollo. Es claro, sin embargo, que la modalidad pasada de crecimiento no se repetira en el futuro, dadas las perspectivas actuales que se le ofrecen al mercado petrolero internacional. Aun cuando la mano de obra fuese atraida a las actividades de servicio por los salarios mas altos en la d6cada de 1970, puede que esto no ocurra en la de 1980. Parece que hubo poco o ninguin crecimiento del PIB en 1982 y 1983 y, aun con un programa eficaz de ajuste estructural para restaurar el crecimiento a plazo medio, es posible que la producci6n global correspondiente a la decada se incremente en poco mas del 5-6% anual. Sobre la base de tendencias pasadas en la absorci6n sectorial de la mano de obra, esto supondria un crecimiento del empleo de s6lo el 2% anual, comparado con un aumento de la fuerza laboral del 2,6%. En ausencia de politicas compensadoras, esto indicaria un grave debilitamiento en las perspectivas del mercado laboral. 26. Volviendo a la cuesti6n de los margenes de cultivo extensivo e intensivo en Java, los datos que se presentan en el Capitulo 4 muestran que la relaci6n hombre-tierra (es decir, la relaci6n entre la tierra de cultivo y la fuerza laboral agricola) fue sustancialmente miAs baja en Java en 1971, que en Corea en la decada de 1960, o incluso en el Jap6n durante gran parte de su periodo de transici6n desde comienzos de siglo. Los estudios agricolas disponibles muestran tambien que el insumo de mano de obra, medido por dias-hombre por hectarea (junto con el rendimiento de arroz con cascara por hectArea), fue mias alto en Java en 1971 que en Corea en el curso de la d6cada de 1960. Pareceria que, aunque todavia pudiera haber algunas zonas aisladas en Java en que la adici6n de mas mano de obra podria incrementar la productividad y rentabilidad agricolas, el pais en conjunto es cultivado mas intensamente que las otras dos economias arroceras durante sus periodos de transici6n. En el informe (Capitulo 4) tambi6n se examinan algunos calculos basados en coeficientes tipicos de tierra-mano de obra en Java con respecto a las cosechas mas importantes cultivadas alli en 1979. Con base en estimaciones muy conservadoras puede indicarse que las necesidades de mano de obra en la agricultura javanesa fueron inferiores en alrededor del 20% a la oferta diponible de mano de obra agricola (dandose por supuesto que cada trabajador aport6 200 dias-hombre por hectarea). Esto concuerda con los datos obtenidos de estudios realizados sobre el terreno citados con anterioridad, los que indicaron que una porci6n considerable de mano de obra de los trabajadores agricolas se dedica a actividades marginales. Hay otras dos cuestiones pertinentes: i) icuAnto tiempo podria sostenerse el crecimiento del 3,6% anual de la producci6n agricola observado durante la ultima decada en Java?, y ii) illegaran a generalizarse mas las practicas de economia de mano de obra observadas en algunas partes de Java, junto con la comercializaci6n acrecentada de la agricultura, reduciendose asi la demanda de mano de obra? Repercusiones en materia de politica 27. En el informe se proyectan grandes dudas en cuanto a si el crecimiento del empleo en la decada de 1970 en la economia indonesia ha estado siguiendo un camino que eventualmente conduciria al punto de - liv - inflexi6n. Dada la probabilidad de que los ingresos del sector petrolero sean mas bajos y, por consiguiente, de un menor crecimiento global para el resto de la decada de 1980, las perspectivas, basadas en tendencias pasadas, de absorci6n de la mano de obra son sombrias. La presunci6n de politica que se infiere de esto es que deben adoptarse medidas a fin de dar al proceso de crecimiento una orientaci6n hacia la utilizaci6n mas intensiva de mano de obra. Este es un tema dificil y el presente informe, con su enfoque principal puesto en el analisis de los mercados laborales existentes y de las tendencias en el curso del tiempo, no cuenta con los medios cabales para dar respuesta. Ahora bien, de la conclusi6n basica se pueden esbozar ciertas repercusiones en materia de politica. 28. El nivel y modalidad del gasto publico pueden ejercer una influencia de importancia en la demanda de mano de obra. A corto plazo, las obras publicas con utilizaci6n intensiva de mano de obra, como en el caso de la infraestructura local de conformidad con los programas de INPRES, aportan una contribuci6n directa al empleo y tambien pueden canalizar el ingreso hacia zonas rurales relativamente pobres, con efectos indirectos de creaci6n de empleo. El mantener esas actividades revestira importancia particular en el curso de los dos o tres anios venideros cuando el Gobierno introduzca otras medidas de restricci6n de los gastos a fin de reducir el d6ficit externo. Esto llevara consigo el desplazamiento de gastos dentro del programa de inversiones publicas, desviandolos de las actividades en que se utilizan importaciones en escala intensiva a las de empleo relativamente intensivo de mano de obra. En mayo de 1983 el Gobierno anunci6 una reprogramaci6n importante de los proyectos de inversi6n publica en gran escala, en que las economias obtenidas en rupias se reasignarian a gastos con un elevado contenido de empleo interno. 29. El nivel de absorci6n de mano de obra por la agricultura fue bajo en la decada de 1970 en relaci6n con los incrementos logrados en producci6n, pero su peso global en el empleo significa que su desempeno sera esencial para el mercado laboral en general. En la decada de 1980 es probable que, para mantener tasas anteriores de crecimiento de alrededor del 4% anual y tambien para proporcionar nuevo empleo, sea necesaria una mayor diversificaci6n alejAndose de la producci6n de arroz hacia actividades como la horticultura, algunos cultivos alimentarios secundarios, la ganaderia y las pesquerias en pequeiia escala. Ademas, la transmigraci6n oficial y espontAnea hacia las islas exteriores podria ejercer un efecto importante. Al ritmo actual de traslado de unas 100.000 familias por anio, alrededor de una quinta parte del incremento de la fuerza laboral en Java podria ser absorbida por este mecanismo. 30. Las politicas relacionadas con el sector manufacturero desempeniaran una funci6n cada vez mas importante en la pr6xima decada y las ulteriores. Segun se examina en el Capitulo 4, este sector ha sido hist6ricamente un elemento importante de absorci6n de mano de obra en economias arroceras pr6speras. Indonesia ha sido poco comuin en el sentido de que ese sector ha representado una parte sumamente pequena del empleo total y casi todo el se ha registrado en las industrias rurales de pequeiia - lv - escala. A fin de alentar la absorci6n de mano de obra, sera menester introducir cierto grado de reorientaci6n en el ambiente de politica. Cabe senialar, en particular, que Indonesia ya se encuentra cerca de los limites de la sustituci6n de importaciones en las industrias de bienes de consumo en que se utiliza mano de obra en escala intensiva y, aparte del crecimiento relativamente lento de la demanda final, el logro de nuevos incrementos en la producci6n para el mercado interno tendra que provenir de las industrias de bienes de producci6n que emplean capital en grado mucho mAs intensivo. Esto subraya la necesidad de formular politicas industriales en las que se haga hincapie en la orientaci6n hacia la exportaci6n, en especial de bienes de consumo, mediante la aplicaci6n de medidas para reducir el actual sesgo hacia la sustituci6n de importaciones y de medidas especificas de promoci6n de las exportaciones. El desarrollo del sector manufacturero en pequeiia escala tambi6n puede ser motivo de preocupaci6n, ya que existe la posibilidad de que este sufra descensos en la producci6n como consecuencia de la competencia de los sectores mas productivos que operan en escala mediana y grande. En el presente informe no se trat6 de estudiar medidas para apoyar a las pequeiias industrias, ni los costos de proteger a los subsectores que utilizan mano de obra en escala intensiva. El anilisis futuro de este tema deberia concentrar la atenci6n en la gran diferencia salarial existente entre las empresas que operan en pequenia y en gran escala, como se examina en el Capitulo 3. Puede que haya que hacer una elecci6n econ6mica entre emplear un pequenio nuimero de trabajadores a un salario relativamente elevado o un gran numero de ellos a un salario mas bajo. 31. Finalmente, los precios de factores en el mercado no dan la guia correcta para determinar la intensidad de la mano de obra en proyectos especificos de inversi6n. Sera necesario elaborar instrumentos de politica para compensar esto. Ya se ha mencionado como uno de esos instrumentos la utilizaci6n apropiada del salario de cuenta en la evaluaci6n de proyectos de inversi6n puiblica. A este prop6sito cabe subrayar un punto de cierta importancia. En la practica, los estudios de evaluaci6n de proyectos usualmente asignan un precio de cuenta a la mano de obra en una fase un tanto tardia en el proceso de formulaci6n de decisiones, cuando la elecci6n entre alternativas es afectada s6lo marginalmente por el precio contable apropiado de la mano de obra. Por lo tanto, la consideraci6n directa de alternativas especificas del sector debe suplementar la utilizaci6n del salario de cuenta si desea darse importancia mas amplia a una estrategia de desarrollo de utilizaci6n intensiva de mano de obra. 32. Un factor igualmente importante --y que desde el punto de vista empirico pudiera ser mas significativo-- que hace que los precios de mercado transmitan una senial err6nea a la elecci6n de t6cnicas es la fijaci6n de precio inferior al capital, por lo menos en algunas zonas importantes de la economia. Por ejemplo, la utilizaci6n amplia de tractores pudiera parecer rentable a los precios de mercado vigentes, pero no ser productiva desde el punto de vista social si se utilizaran precios - lvi - contables apropiados para el trabajo y el capital 1/, a menos que investigaciones especificas revelaran segmentos de escasez de mano de obra en zonas particulares. Trabajo ulterior acerca del empleo y el ingreso de la mano de obra en Indonesia 33. Aunque en el informe se subraya que un modelo del mercado laboral basado en una hip6tesis de mano de obra excedente en el sector rural (y en la idea asociada de que el sector arrocero productivo es un sector privilegiado desde el punto de vista de quienes buscan empleo) es directamente pertinente para comprender el desarrollo indonesio, no se llega en 6l a la conclusi6n de que ese modelo ha sido en verdad sometido a prueba y verificado. Es necesario trabajar mAs y someter a prueba modelos alternativos de mercado laboral, a niveles tanto micro como macro. En el Apendice A del Capitulo 3 se presentan ideas te6ricas acerca de micromodelos alternativos del mercado laboral rural, y se examina hasta qu6 punto se pueden utilizar los datos empiricos existentes para apoyar el modelo del "excedente" laboral. En el Anexo del presente informe se explora si las amplias tendencias macroecon6micas del desarrollo indonesio encajan en el marco de los modelos neoclasicos agregativos y tambi6n se busca establecer gamas de valores con respecto a parametros clave que esten en consonancia con la experiencia real. En este contexto se estudia la aplicabilidad del modelo alternativo del excedente laboral para la interpretaci6n de las tendencias macro. 34. En el informe se recomienda que se haga mAs trabajo empirico con objeto de observar en forma directa el funcionamiento de los mercados laborales. Con respecto a estos en las zonas rurales, las siguientes son esferas importantes de investigaci6n: i) La demanda de mano de obra en la agricultura javanesa depende de la indole de la temporada dinamica, comparada con la de menor actividad. ZCual es la duraci6n de la temporada dinamica y c6mo afecta a diferentes aspectos de la fuerza laboral? ZCual es la proporci6n de tiempo de trabajo dedicado al sector (productivo) arrocero y a las actividades marginales, para grupos significativos de edad y sexo distinguidos por su situaci6n econ6mica? ZLos ingresos por hora en actividades marginales son sustancialmente mas bajos que en el sector arrocero en la temporada dinamica, al igual que lo son con respecto al anio en conjunto? Ya se dispone de mucha informaci6n acerca de estas preguntas en estudios de poblado terminados, pero no se ha analizado o bien no esta compilada en forma adecuada para dar respuesta a esas preguntas. Por lo tanto, un estudio mas cuidadoso de los datos micro ya recopilados en relaci6n con poblados individuales es una tarea de alta prioridad. 1/ V6ase el estudio de Rudolph S. Sinaga, "Implications of Agricultural Mechanization for Employment and Income Distribution", Rural Dynamics Study Series, No. 2, Bogor. - lvii - ii) Es claro que las variaciones regionales o incluso de un poblado a otro en lo que se refiera a las condiciones del mercado laboral revisten importancia sustancial en Java, lo mismo que en todas las sociedades rurales asiaticas. Algunos poblados pueden en realidad ser mas a prop6sito que otros para el modelo del excedente laboral. En el analisis ulterior de los datos a nivel de poblado debe fijarse la atenci6n con firmeza en este punto. Incluso si no se dispone de estudios detallados de poblados en numero suficiente, un analisis representativo de variaciones entre poblados de salarios en ocupaciones seleccionadas, y su relaci6n con las condiciones econ6micas subyacentes en los poblados, seria probablemente muy revelador. iii) A nivel macro, en el informe se hace hincapie en la necesidad de estudiar mas series sobre las fluctuaciones en las tasas de salarios, en particular en la agricultura. De nuevo en este caso, no se puede insistir demasiado en la importancia de analizar las tendencias de los salarios y el empleo sobre una base regional. Los datos del censo de 1980 deben proporcionar un gran acervo de informaci6n acerca de las tendencias del empleo regional durante la decada de 1970 cuando se utiliza en conjunci6n con otras fuentes de datos estadisticos, incluido el censo de 1971. iv) Una zona igualmente importante de informaci6n la constituyen el nivel y tendencias de los ingresos en los sectores no estructurados muy grandes, en particular en las actividades de servicios. Si bien un analisis mas completo de las encuestas de unidades familiares existentes proyectaria mas luz sobre esta cuesti6n, la prevalencia de ocupaciones multiples en Indonesia, sobre todo en el sector rural, exige la preparaci6n cuidadosa de cuestionarios para futuras encuestas de unidades familiares. Debe hacerse un intento por recopilar informaci6n acerca de las horas trabajadas y los ingresos percibidos en las diferentes actividades desplegadas por los trabajadores, en lugar de limitar la indagaci6n a la principal actividad de la persona encuestada. v) Tambi6n es importante ir mas alla de la vinculaci6n existente entre salarios, empleo y producci6n tal como se examinan en el presente informe. Por ejemplo, la tierra y el credito son factores clave, pero los datos actuales y el estado de conocimientos han impedido la consideraci6n sistematica de los vinculos entre mercados de factores de producci6n (tierra, capital y trabajo) y de la funci6n que el credito y el financiamiento desempeiian, en un sentido amplio, en esos vinculos. El recopilar informaci6n estadistica adicional a nivel de poblado, que es lo que se recomienda en este informe, debe llevarse a cabo con la mira de facilitar estudios de las vinculaciones entre los mercados de factores de producci6n y la funci6n del credito. CHAPTER 1 STRUCTURE OF THE LABOR FORCE, EMPLOYMENT AND LABOR INCOMES IN 1977 1.01 The purpose of this chapter is to set the scene for the analysis of labor markets and employment in subsequent chapters. A few of the salient features regarding the use of labor in economic activity in Indonesia around 1977 will be discussed. 1977 was selected because it represented the latest completed and revised data from the annual Labor Force Survey (SAKERNAS) when this report was being written. Examination of later data from the SAKERNAS of 1979 or the Census of 1980 revealed that the differences were not large enough to merit a complete revision of this Chapter. 1.02 The population of Indonesia was not very urbanized by 1977. According to SAKERNAS data, only 18% of the population lived in urban areas. Along with other Asian countries, Indonesia has a young population - about 30% are under 10. But unlike some other developing countries, the age structure of the population does not differ markedly between the urban and rural areas. This is because rural/urban migration does not consist of large numbers of people of prime age living temporarily in the urban areas. 1.03 A significant feature, both in the analysis of the labor market and in the study of trends in labor earnings, is the expansion of formal education. We begin, therefore, by looking at the quantitative picture of the educational profile of the population in 1977. The Educational Attainment of the Population 1.04 Appendix Tables IB.4 and IB.5 present the educational distribution of the population by age and sex. There has been, as might be expected, a marked upgrading of education, but the tables reveal important differences between the urban and rural areas. The major points can be underlined by the figures set out in Table 1.1 which contrast the educational profile of the younger age groups with that of the 30-49 age group. 1.05 The figures reveal a very large increase in primary education, resulting in a jump in the proportion of young people with a primary education to more than 75% in urban areas, and 50% in rural areas. The expansion of secondary education is very much an urban phenomenon. A substantial proportion of the young urban population - a third of the males and a fifth of the females - had senior high school education or above. But the comparable populattfons in the rural areas continued to be very small. Table 1.1: PERCENTAGES OF THE POPULATION WITH MORE THAN A SPECIFIED EDUCATIONAL LEVEL BY AGE, SEX AND LOCATION, 1977 Junior Senior Completed high high primary school school & above & above & above Urban Males 15-19 80.2 38.4 5.0 20-24 82.1 52.0 32.3 30-49 64.0 31.9 6.5 Urban Females 15-19 73.5 32.1 4.5 20-24 70.3 38.1 21.4 30-49 41.5 19.7 8.7 Rural Males 15-19 60.5 14.4 7.7 20-24 60.9 19.3 5.0 30-49 35.8 6.5 3.2 Rural Females 15-19 51.5 18.0 2.9 20-24 43.2 7.7 1.9 30-49 15.2 2.4 0.6 Source: SAKERNAS, 1977. 1.06 As we shall see in the following sections, the higher educational level of the urban population has a pronounced effect on participation and unemployment. Participation Rates 1.07 Table 1.2 gives the participation rates by age, sex and location for all Indonesia. In view of the importance of the quantitative data on participation rates for the subsequent analysis of labor free growth and employment in the next chapter, two sets of figures are given - one from the Labor Force Survey of 1977-78, and the other from the sample population of the Population Census of 1980. The pattern of participation rates reported by the two sources are very similar; differences by age-group, sex and location are much the same. But the participation rates derived from the Census tabulations are systematically lower. This difference between the Population Census and Labor Force Surveys in counting the economically active has been noted in many countries, and can be traced in part to the fact that sample surveys will generally make a more concentrated effort to locate specific groups in the labor force. Table 1.2: LABOR FORCE PARTICIPATION RATES BY AGE, SEX AND LOCATION, INDONESIA, 1977/78 Urban Rural Age group Male Female Male Female A. Labor Force Survey, 1977/78 10-14 4.8 4.7 19.4 11.4 15-19 33.9 23.2 65.4 39.1 20-24 74.9 29.4 90.9 39.4 25-29 92.8 30.1 97.8 44.4 30-49 97.3 37.6 98.7 53.8 50-59 80.6 34.4 94.6 51.3 60+ 53.6 21.7 75.7 12.1 Total 63.2 26.3 76.1 40.5 B. Population Census, 1980 10-14 3.3 4.7 15.2 10.9 15-19 27.2 22.3 54.8 34.4 20-24 67.5 27.0 84.2 35.4 25-29 89.5 28.0 93.1 38.6 30-49 95.1 32.7 94.9 45.4 50-59 79.6 32.6 90.0 45.6 60+ 47.6 17.5 67.0 25.5 Total 60.0 24.0 70.6 34.7 Source: SAKERNAS, 1977. 1.08 Figure 1 charts the age-specific participation rates, separately for the sexes, and for rural and urban areas. The difference in the profiles for males and females is readily apparent. The graph of age-specific LFPRs for females in the rural areas lies more or less parallel, although above, the one in the urban sector. Female participation rates are higher by roughly the same percentage in the rural areas for all age groups. But for the males the rural-urban divergence occurs mainly at age groups below 30 and above 44. - 4- FICURP 1.1 MALE - FEMALE LABOR FORCE PARTICIPATION RATES. INDONESIA. 1977-78 900 4 4... Male Rural z / t < 0 A 0 0 ~~~~~~~~~~~~~~~~~tiale Urban 0 ~ ~ ~ ~ ~ ~ ~ 0. K I *C C;f .; i 2 3 4 S 6 } t 9 10 11 12~~~~~100 ~~~. ~ Fmale Urban AGE GROUP See next p age for Age Croup Index _o_l_ank _ - 21 _ Source: SAKERNAS, 1977. - 5 - Age Group Index 1 ........ 10-14 years 2 *....... 15-19 years 3 ........ 20-24 years 4 ........ 25-29 years 5 ........ 30-34 years 6 .. . 35-39 years 7 ...... 40-44 years 8 ....... 45-49 years 9 ........ 50-54 years 10 ....... 55-59 years 11 ....... 60-64 years 12 ....... 65+ years Males 1.09 The relatively low participation rates of young males in urban areas is important. In particular, attention should be drawn to the fact that, even for the age group 20-24, only 75% of the population participated in market activity. This phenomenon represents the sluggishness of entry into the labor market for first-time job seekers 1/ and has been noticed in other Asian economies. This is a vital aspect of youth unemployment (see below), which is such an important part of the labor market in these economies. The difficulty of getting an acceptable first job is revealed partly in the high rate of open unemployment and partly in the high incidence of the discouraged worker's nonparticipation. As might be expected, this phenomenon is particularly important for the more educated members of the labor force. Relatively high expectations, as well as higher socioeconomic family conditions, of the educated tend to increase their minimum acceptable standards for a first job. By and large, the Indonesian data bear out this hypothesis. It will be seen from the Appendix Table IB.7 for males aged 20-24, that participation rates decrease drastically between primary and secondary education. It should be remembered that, with reasonable performance, secondary education should be completed well before age 20. In some cases, completing school education at such a late age is itself an aspect of "discouragement." 1.10 The relationship between education and LFPR for young males is, however, not monotonically decreasing. In particular, males aged 20-24 with no schooling have a substantially lower participation rate than those with elementary education - almost as low as those with junior secondary education. But the proportion of people with no schooling at all in this age group was very small, about 2%. Data on the occupation of the nonparticipants given in SAKERNAS show that as much as 20% of the urban males in the age group were attending school. But the official tertiary enrollment rate was only 2%. Thus we can conclude that many of the nonparticipants were really discouraged workers who were using nonformal education as a means of covering up their inactivity. -6 - 1.11 While the lower participation rates for urban males between 20-24 can largely be ascribed to the impact of educational expansion, the substantial rural/urban difference in the 15-19 age group cuts across all educational categories. Even for the numerically important group with some elementary education, LFPRs in the urban areas are some 20 percentage points lower. Evidently, the urban labor market works differently from the rural one and requires that young people, wait longer before getting their first job. 1.12 A sizeable proportion of the rural population in the 10-14 age group participates in economic activity. This reflects the nature of work organization among peasant families, in which all members of the household from an early age contribute to making a living. 1.13 At the other end of the age spectrum, early retirement is a feature of Indonesian urban labor markets. For males aged 50 and over, the LFPR is 17% lower in urban than rural areas (the rate in the latter being 84%). The difference is observed almost equally for all educational groups - except those with Academy or University qualifications. Females 1.14 The graph of age-specific participation rates for females is relatively flat for both rural and urban areas. This profile contrasts not only with the male profile, but also with the typical profile for females in developed countries. In the United States for instance, before the First World War, there was a peak in the participation rates for females between 20- 24, reflecting the withdrawal of females from the labor market following marriage. This has been replaced in the last decade or two by a double-peaked profile, the second peak coming in the age group 35-39, reflecting the return to work by married women after children have grown up. For various reasons which cannot be analyzed in detail here, the impact of the marriage and child- rearing cycles on females' LFPR is minimal in Indonesia, as in several other Asian countries. 1.15 The higher female LFPR in all ages observed in Indonesia in the rural areas, compared to the urban sector, is not a prevalent feature of all Asian economies. It does not seem to be the case in Thailand or the Philippines. - It is, however, a feature of India. 1/ The female participation rates for female (all ages taken together) for these countries were as follows: Rural Urban Philippines 35.0 37.7 Thailand 41.8 40.5 (Source: Bureau of Census and Statistics Labor Force Survey, 1976, Philippines. Final Report of the Labor Force Survey, 1974, Thailand.) -7- In Sum 1.16 The lower rate of participation in the urban areas by both males and females (for specific ages for the former and for all ages for the latter) implies a higher dependency burden in the towns. Permanent migration from the rural to the urban sector (which is continuing in Indonesia) would require a wage gap in favor of the urban areas to compensate for the typically higher dependent-earner ratio in the urban sector. This will be important in discussing the urban wage structure in Chapter 3. 1.17 We turn now to a discussion of selected characteristics of the labor force in Indonesia - the unemployed and the employed. Measured Unemployment in Indonesia 1.18 The SAKERNAS surveys of 1977/78 reported unemployment characteristics in Indonesia similar to those in neighboring countries. r Open unemployment is very much an urban phenomenon. In the rural areas, unemployment was only 1.25% for females and about 2% for males. The respective unemployment rates for the two sexes in the urban sector was 4.6% and 6.7%, respectively. Secondly, in both sectors, but especially in the urban areas, open unemployment is concentrated in the younger age groups. As can be seen from the Appendix Table IB.9, the highest unemployment rate is observed for the 15- 19 age group but the rates for the 10-14 and 20-24 age groups are also high. The figures are as follows: Rate of Unemployment Age group Males Females 10-14 17.4 7.1 15-19 25.3 11.8 20-24 18.0 12.5 Source: Appendix Table IB.9. 1.19 Youth unemployment in urban Indonesia is intimately connected with education. Table 1.3 shows that the unemployment rates for males with no schooling in the 10-14 and 15-19 age groups are quite high, but the absolute numbers in these cells are not large. Leaving aside these two observations, the unemployment rate increases systematically with higher educational levels for each age group. The same is true for females. Remarkably high 1/ The definition of those who were working was that they were at work for at least one hour a day during the past week. So the unemployed were those who were in the labor force and not working in this sense, and excluding those temporarily out of work. - 8 - unemployment rates are observed for those with senior high school qualifications. For both sexes, half of those between 15 and 19 participating in the labor force with this level of schooling are unemployed; and even in the 20-24 group the unemployment rate is more than 30%. The general phenomenon is common to many Asian economies, but the magnitude of unemployment rates in Indonesia is high. It is interesting to note that unempoyment rates for those with post elementary education do not drop to "normal" levels until the 30+ age group. A sizeable proportion of the labor force with a secondary education or above is still searching for a job through their late 20s. Table 1.3: URBAN MALE UNEMPLOYMENT RATES BY AGE AND EDUCATION, 1977 (%) Age group 10-14 15-19 20-24 25-29 30-34 35-39 All ages No school 22.7 29.7 4.1 3.2 2.6 0.8 2.2 Not finished elementary 15.0 19.7 10.3 4.6 0.8 0.6 4.8 Elementary 38.4 24.1 13.0 3.0 1.2 0.3 6.2 General junior high - 36.6 20.5 8.4 2.3 0.8 8.5 Vocational junior high - 36.4 30.0 7.8 2.2 1.0 8.8 General senior high - 51.9 29.7 8.7 2.3 2.3 9.0 Vocational senior high - 58.5 32.5 5.1 2.0 0.9 12.5 Academy - - 21.5 14.4 1.0 0.8 3.8 University - - 0.0 10.6 2.5 0.0 1.7 All levels 21.8 25.8 18.4 5.7 1.6 0.7 6.4 Source: SAKERNAS, 1977. Duration of Unemployment 1.20 A significant dimension of the unemployment problem is its duration. The data show that it is not very high: 84% and 70%, respectively, for rural and urban males are unemployed for less than one month; correspondingly 76% and 72% of rural and urban females are unemployed for a month or less. - Thus nonparticipation would seem to be the major indicator of underutilization of labor. Young persons are actively unemployed only for a short period once they enter the labor market. In fact, it seems likely that they do not actively search for a job until they feel that acceptable employment is likely to become available soon. 1/ SAKERNAS, 1977 (Average), pp. 205-213. -9- Underutilization of Urban Youth 1.21 The preceding paragraphs have emphasized a problem of underutilization of the young population in Indonesia - particularly in the urban areas. We should stress that young people between 15 and 24 constitute around 30% of the total population of 15 and above in Indonesia. Their inadequate utilization in gainful activity in the urban economy must be an important problem of employment in Indonesia. It has also been mentioned that the process of educational upgrading of the urban labor force might have had a significant influence on the degree of underutilization. To put the problem in perspective we have calculated the total rate of nonemployment (adding nonparticipation and active unemployment together) for the critical age groups 15-19 and 20-24 by different educational levels. Table 1.3 shows the total rate of nonemployment for each cell, and Table 1.4 gives the distribution of the total not employed in each age-sex group in the broad educational levels. Table 1.4: RATE OF NONEMPLOYMENT /a OF YOUTH IN URBAN AREAS, 1977 Junior Senior Academy No Elementary school high high & uni- Sex/age schooling Incomplete Complete school school versity Total Males 15-19 52.4 51.8 70.7 92.0 84.4 n.a. 75.9 20-24 34.0 17.2 20.7 49.5 60.9 68.6 40.2 Females Nearly 15-19 53.3 67.9 83.0 100.0 92.6 n.a. 81.6 20-24 74.7 74.1 78.4 77.9 65.3 68.0 75.8 /a Nonparticipants plus unemployed as a proportion of the population of the particular cell. Source: SAKERNAS, 1977. 1.22 It is clear that males in both age groups with secondary education have a higher rate of nonemployment than those with primary education. But this relationship is not as clear for females. A second important conclusion from Table 1.5 is that only among males in the 20-24 age group is a substantial proportion of the nonemployed accounted for by persons with a reasonably high education (i.e., senior high school). For the other age - 10 - sex/groups, those with primary education or less constitute the majority of young people without employment. Table 1.5: PROPORTION OF EACH AGE-SEX GROUP WHICH IS NOT EMPLOYED BY EDUCATIONAL LEVEL, URBAN INDONESIA, 1977 Junior Senior Academy No Elementary school high high & uni- Sex/age school Incomplete Complete school school versity Males 15-19 1.0 12.6 38.9 40.5 5.6 - 20-24 0.7 6.2 15.5 25.8 47.1 2.0 Females 15-19 3.0 19.6 40.4 33.9 5.1 n.a. 20-24 5.8 23.3 33.2 17.1 17.4 1.1 Source: SAKERNAS, 1977. 1.23 The substantial underutilization of youth - even males in the 20-24 age groups - in urban Indonesia is an important feature of the labor market scene. Analytically, it suggests a market in which the pressure of new job seekers is only slowly relieved by market adjustments. In socioeconomic terms (apart from being a waste of human resources) it is a significant and potentially serious social problem. This phenomonen is not due entirely to the rapid increase of secondary education in urban areas. Also it must be stressed that the existence of a large "informal" sector in urban Indonesia does not lead to the quick absorption of freshers in the labor market. The Employed Population 1.24 The major characteristic of the composition of industrial employment in Indonesia in the late 1970s is the small proportion of persons employed in manufacturing - around 10% (see Table IB.ll). Agriculture accounted for 60% of total employment - somewhat more in outside Java. The importance of the tertiary sector - accounting for nearly a third of all employment - is also noteworthy. As we shall see in Chapter 2 the tertiary sector has been increasing in relative importance during Indonesia's recent growth. 1.25 The status composition of the employed population by industry is given in Table IB.13. Attention should be drawn to the following points. Wage earners (employees) constituted a quarter of the employed in agriculture. The importance of nonwage employment in the economy is brought out by the fact that, even in the manufacturing and the tertiary sectors, employees were about half of the total employed. A characteristic of tertiary employment is the large population of own account workers (a third of the total), underlying the importance of autonomous single workers in this sector. - 11 - 1.26 The major importance of the tertiary sector in the Indonesian economy calls for further enquiry about the characteristics of this sector. Within the limits of the data available to the mission, only a few additional points can be made. Table 1.6 shows the percentage employed in this sector, separately for males and females, and for urban and rural areas. Table 1.6: PERCENTAGE OF THE TOTAL EMPLOYED IN EACH CELL IN THE TERTIARY SECTOR, 1977 (All Sectors) Urban Rural Males 76.6 22.0 Females 74.6 22.0 Source: Appendix Table IB.ll. 1.27 The differences in the importance of the tertiary sector vary with the location of employment, but not much with the sex of the person employed. Leaving agriculture aside, the tertiary sector accounted for 85% of total employment in the urban areas, and 74% in the rural. 1.28 Neither does the proportion of female workers within the tertiary sector vary much between the rural and the urban areas: it is 29% in the urban areas, and 34% in the rural. In sum: outside agriculture, tertiary employment is extremely significant in the Indonesian economy - equally in the rural and the urban areas, and for males as for females. 1.29 Some idea of the composition of the tertiary sector might be observed from Appendix Table IB.14 given for the urban areas. Almost all of the females employed in this sector fall into two broad categories - trade and services. Most of the males also fall into these two groups, but, in contrast to the females construction and transport provide a fair number of additional jobs for males (of the order of 20% of the total in the sector as a whole). 1.30 An important issue is that of labor earnings in the tertiary sector compared to the goods-producing sector. The question is critical to the issue of whether labor is being "pulled" into the tertiary sector by the process of Indonesian development, or being "pushed" into it because labor supply for traditional activities is outstripping demand. If it is the latter, we would expect labor earnings to be significantly lower in the tertiary sector compared to the other sectors, particularly agriculture. No detailed evidence exists on this question. The only source of data is the information on the earnings of employees in different sectors produced by the National Labor Force Survey (SAKERNAS). The summary satistics on the distribution of earnings in the different industries, by sex and location, are given in Appendix Table IB.12 for what they are worth. They show that at all points in the distribution for both males and females, and in the urban and rural areas the monthly income for employees in trade and services is higher than in - 12 - agriculture by a considerable margin, and it is more or less on a par with that in manufacturing. The fact that the income per employee at the first quartile of the distribution in the tertiary sector is twice that at the first quartile of the distribution in agriculture would, at first constitute clear evidence in support of the hypothesis that labor is being pulled to more lucrative jobs in trade and services. But unfortunately no such firm conclusion can be derived from the SAKERNAS data. Apart from the fact that the earnings data refer only to employees, the reported monthly income per employee is derived from the "main industry." As is well known - and we go into this point in greater detail in Chapter 3 - employees working for wages in agriculture devote only a portion of their time, perhaps 50% or less, to agriculture even when it is their main occupation. Multiplicity of occupation, is much more important for those whose principal source of livelihood is agriculture, particularly the landless or the nearly landless. It follows that the monthly income from the main occupation in agriculture reported in the SAKERNAS survey will not be comparable with the incomes of those whose main occupation is in the nonagricultural sector. The latter represents returns to labor from a larger amount of activity in the market. Verification of this point is provided by another set of data from the SAKERNAS survey reproduced in Appendix Table IB.ll which gives the distribution of the number of hours worked per employee by the main industry of employment. It will be seen that a quarter of the rural males, whose main activity in agriculture worked less than 24 hours per week, and as much as 37% of the rural females worked less than 24 hours. By contrast, in both manufacturing and tertiary sectors, the proportion of employees wy7rking less than 24 hours was 8% for rural males and 16% for rural females. - The difference between agriculture, on the one hand, and the nonagricultural sectors on the other, is equally striking for the urban areas. 1.31 The conclusion from the SAKERNAS survey data on the measurement of interindustrial income differences from is thus largely negative. But it is an important point, since at the moment in Indonesia such survey data are the only countrywide sources of earnings information. Without more careful design of the surveys, paying particular attention to multiple activities and hours worked, no inference on differences in returns to labor in different industries is possible. Yet, as we have seen, an assessment of the relative earnings in the tertiary sector is of utmost importance for the interpretation of recent Indonesian development. 1.32 In Chapter 3, we will be referring to some data from intensive field surveys by individual research workers. These studies shed some light on differences in returns to labor in different activities for those whose principal activity is agriculture. But even the individual field studies have little to say on earnings in the tertiary sector for those who depend on it as the major source of income. 1/ The data presented refer to hours worked in the main industry. This interpretation, although not explicitly stated in the SAKERNAS survey, is consistent with field workers' evidence cited in Chapter 3. - 13 - Rural/Urban Earnings Differential 1.33 A question of some importance which comes up at various points is that of the rural-urban wage differential. As already mentioned, the SAKERNAS earnings data pertaining to agriculture cannot be used for this purpose. But in other industries, employees' earnings, from activities other than the main one are presumably much smaller. Some comparison of earnings in rural and urban areas in the nonagricultural industries can be made tentatively from the SAKERNAS data. The mean urban earnings relative to those in the rural areas are as given in Table 1.7. Table 1.7: RELATIVE MEAN EARNINGS IN THE URBAN AREAS (RURAL = 100) Industry Males Females Manufacturing 171 171 Construction 139 153 Trade 221 175 Transport 145 371 Services 137 139 Source: Appendix Table IB.12. 1.34 There would seem to be a sizeable differential in labor earnings in favor of urban areas. On the one hand, the importance of circular migration and "commuter" labor tends to keep the urban-rural differential low compared to many other economies. (This point is discussed more fully in Chapter 3.) But the higher burden of dependence in the urban economy for permanent migrants, and the evidence already presented regarding youth unemployment are consistent with a significant rural-urban wage gap. 1.35 But the differences in mean earnings between rural-urban areas in Table 1.7 do not correct for differences in skill-mix. For males, the urban- rural differential is the least in construction and services. This is probably because the rural-urban differential in the skill mix is least in these two sectors. In trade, the earnings differential is the highest. Here, the structure of the activity and, hence the skill-mix differs substantially between rual and urban areas. Finally, the figures on earnings differentials do not take into account the cost of living difference between the two sectors. - 14 - Chapter 1 Appendix A Page 1 Seasonal Changes in the Labor Force Participation Rate 1. Previous World Bank reports have emphasized seasonal changes in agricultural employment in Indonesia, as an indication of the "high degree of responsivenes.i of the labor force to seasonal changes in labor market conditions." - The evidence cited is the decline by about 6 million people in agricultural employment between March 1976 (information from SUPAS, the Intercensal Labor Force Survey) and October 1976 (Information from SAKERNAS, the labor force survey for September-December 1976). Attention was concentrated on whether the supply of labor at the peak season was drawn from unpaid family labor or hired labor. Some attention should have been devoted to whether the recorded decline in employment could have been attributed to definitional changes between the two sources rather than to seasonal variation in the labor force. The reason to look for definitional changes is the reported decline of 4 million workers in the rural female labor force between March and October 1976 - the group for which work and housekeeping are difficult to distinguish. 2. A clue to the effect that definitional changes may have had, is contained in the SUPAS and SAKERNAS data. The reported decline in agricultural employment occurred during the reference week in both sources. In an attempt to eliminate seasonal fluctuations, both sources asked respondents about usual activity (i.e., the activity during the year preceding the enumeration) as opposed to current activity (i.e., the activity during the previous week). Data on usual activity should be free from seasonality effects. But, in fact, the rural female labor force according to usual activity is still hi P er, by 5 million, in SUPAS (March 1976) than in SAKERNAS (October 1976) -/ 3. This appendix throws additional light on seasonal changes in participation rates in Indonesia, using available information for the years 1976, 1977, 1978 and 1979. Table IA.1 contains the participation rates by sex, in rural and urban areas, for each of the three years, at the peak and the slack seasons. The main point is that the measured change in the participation rate for rural females in 1976 is way out of line relative to the change in 1977, 1978 and 1979. In these three years, the maximum differ- ence between the peak and slack rates is 4 percentage points in 1978, and the minimum just over 1 percentage point in 1979. The differences reported in SAKERNAS are nowhere near the difference of 11 percentage points given for 1978. 1/ /See World Bank (1979), Report No. 2093-IND, p. 43, paragraph 2.52. 2/ This has been pointed out by R.M. Sundrum, Australian National University, unpublished manuscript, November 1979. - 15 - Chapter 1 Appendix A Page 2 Table IA.1: SEASONAL VARIATIONS IN LABOR FORCE PARTICIPATION RATES, INDONESIA, FIRST QUARTER 1976 THORUGH FOURTH QUARTER 1978 (%) Year/Quarter Urban Rural Male Female Male Female 1976 I 65.2 26.5 79.8 51.0 IV 63.2 25.1 76.3 39.5 1977 I 62.7 24.8 76.0 40.6 IV 61.4 24.4 74.5 37.0 1978 I 65.5 28.6 77.6 44.2 IV 62.9 28.1 75.3 40.3 1979 I 64.4 26.8 77.6 40.3 IV 61.9 23.8 76.5 39.1 Sources: First quarter 1976, SUPAS, VP78-01, Table 02; fourth quarter 1976, SAKERNAS 1976, VUS 78-22, Tables 01.1, 01.2, 01.4, 01.5; first quarter 1977 through fourth quarter 1979, SAKERNAS 1977, 1978, 1979. 4. A second point is that seasonal change in the participation rate for rural males is of the order of 1-2 percentage points, rather than 3.5, as in 1976. 5. A third point, much less important quantitatively, is that the rates for urban males show some seasonal change, slightly more pronounced than the changes for urban females. 6. Information about seasonal changes in the age-specific participation rates for each of the four labor force groups can be obtained from Chapter 2, Table IIC.1. - 16 - Chapter 1 Appendix B Page 1 DATA Table IB.1: POPULATION (IN MILLIONS) BY REGION AND LOCATION, 1977 Java Outside Java Indonesia Urban All ages 15.6 9.0 24.6 10 and over 11.4 6.3 17.7 Rural All ages 69.5 39.9 109.3 10 and over 50.5 27.7 78.2 Total All ages 85.1 48.9 133.9 10 and over 61.9 34.0 95.9 Source: Appendix Tables IB.2 and IB.3. Chaptr 1 Appendix B Page 2 Table IB.2: AGE AND SEX DISTRIBUTION OF TE POPULATION BY REGION, 1977 Age Group Indonesia Java Outside Java Male Female Male Female Male Female 0-4 14.9 14.2 14.1 13.3 16.3 15.9 5-9 14.0 '13.7 13.7 13.3 14.4 14.3 10-14 12.8 12.5 12.8 12.3 12.8 12.8 15-19 11.1 10.6 11.1 10.6 11.0 10.6 20-24 9.2 9.2 9.2 9.0 9.4 9.5 25-29 6.5 6.8 6.3 6.7 6.8 7.1 30-34 5.7 5.8 5.6 6.0 5.7 5.3 35-39 5.9 5.9 6.0 6.3 5.8 5.3 40-44 5.2 5.1 5.6 5.5 4.4 4.4 45-49 4.2 4.2 4.6 4.6 3.5 3.7 50-54 3.4 3.5 3.7 3.7 2.9 3.2 55-59 2.6 2.9 2.8 3.0 2.4 2.6 60-64 1.9 2.2 1.9 2.3 1.8 2.1 6 i+ 2.6 - - - - - Total 100 100 100 100 100 100 (66221) (67719) (41791) (43284) (24430) (24435) Note: Numbers in parenthesis denote population, in thousands. Source: BPS, Proyeksi Penduduk Indonesia, 1976-20001, VP78-06, Tables 6.1, 6.2 and 6.3. Chapter 1 Appendix B Page 3 Table IB.3 AGE AND SEX DISTRIBUTION OF THE POPULATION, BY REGION AND LOCATION Age Group Urban Rural Urban Rural Male Female Male Female Male Female Male Female <10 28.0 26.1 28.8 27.8 31.1 29.2 31.5 30.8 10-14 13.0 12.4 13.3 11.4 14.0 13.0 14.1 12.2 15-19 11.5 11.8 10.5 8.8 11.1 12.1 10.8 10.5 20-24 8.7 9.6 5.9 7.2 7.6 9.5 6.6 7.7 25-29 6.6 7.2 6.1 7.5 7.0 7.9 5.9 7.4 30-34 5.7 6.4 5.3 6.4 5.8 5.9 5.2 6.5 35-39 6.5 6.8 7.1 7.6 6.2 6.0 6.3 6.8 40-44 5.4 5.3 5.8 6.3 4.6 4.5 5.2 5.1 45-64 12.4 11.7 14.4 13.7 10.6 9.6 11.9 10.6 65+ 2.1 2.7 2.7 3.2 1.9 2.2 2.6 2.5 Total 100 100 100 100 100 100 100 100 (7601) (7819) (33788) (34896) 4512 4478 18386 18289 Note: Numbers in parenthesis denote population, in thousands. Source: BPS, Labor Force Situation in Indonesia, The Average, 1977, VUS 79-28, Tables 26.1, 26.2, 26.4 and 26.5. Chapter I PVage 4 Table 1B.4: EOUCATIONAL ATTAIN1ENT OF THE POPULATION BY AGE GROUPS, 1977 URBAN Age group No School Elemnetary School Junior H1gh Scho SA Sior High School (SLP) AcadeVy University Total incomplete Loiiiplete General Voc Itonal lotaT liGneral Yocational lotal hale 10-14 1.94 75.20 21.87 0.82 0.13 0.95 100.00 I 15-19 1.42 18.39 41.76 29.34 4.05 33.40 3.29 1.75 5.04 100.00 A 20-24 2.15 14.55 30.10 16.64 4.30 20.93 19.07 12.02 31.08 1.13 0.06 100.00 25-29 2.54 16.06 32.04 13.88 3.94 17.82 16.11 10.31 26.42 4.40 0.71 100.00 30-49 7.55 22.64 32.10 12.55 3.63 16.18 10.16 5.58 15.73 3.37 2.43 100.00 50. 24.68 27.69 30.33 8.04 2.72 10.76 3.36 1.45 4.81 0.84 0.83 100.00 Total 6.68 30.98 31.24 13.03 3.02 16.06 7.84 4.57 12.41 1.69 0.93 100.00 Fe_ale 10-14 2.93 73.10 23.12 0.70 0.12 0.82 100.00 IS-19 4.63 23.53 39.68 24.65 3.00 27.64 3.23 1.25 4.48 100.00 20-24 5.89 23.80 32.13 13.97 2.71 16.68 11.86 8.33 20.20 1.11 0.09 100.00 25-29 8.69 24.40 32.77 13.80 2.88 16.68 8.14 7.16 15.30 1.50 0.51 100.00 30-49 28.98 28.52 22.80 7.84 3.0S 10.92 4.01 3.66 7.67 0.79 0.21 100.00 50 63.36 19.88 11.68 2.20 1.07 3.27 0.63 0.86 1.49 0.09 0.05 100.00 Total 19.79 33.26 26.44 10.06 2.22 12.28 4.21 3.24 7.45 0.55 0.15 100.00 Source: PS, Labor Force Situation in Indonesia, The Average, 1977, National Labor Force Survey, SAKERNAS, YUS79-28, 1977; Tables 4.1. 4.2 Chapter I App?edlx B Page 5 Tab]^ 13.5: EDUCATI0 ATTAINMNENT OF THE POPULATION BY ACE GROUPS, 1977 RURAL Age Croup No School Elementary School Junior High School Senior High School Academy University Total Inacomplete Completed General Vocational Total General Vocational Total 10-14 6.21 76.57 16.32 0.56 0.04 0.60 100.00 15-19 6.87 32.60 46.02 11.46 1.66 13.12 0.72 0.57 1.29 100.00 20-24 7.14 31.99 41.59 9.29 2.35 11.64 3.39 4.10 7.49 0.16 100.00 25-29 10.86 35.0,; 41.83 5.59 1.62 7.21 1.70 3.08 4.78 0.25 100.00 r 3G,49 25.60 38.62 28.93 2.45 1.14 3.59 0.74 2.19 2.93 0.23 0.08 100.00 0 5C* 52.93 31.50 13.71 0.70 0.65 1.36 0.12 0.27 0.39 0.07 0.01 100.00 Total 20.45 43.05 28.88 4.03 1.08 5.11 0.81 1.47 2.28 0.12 0.03 100.00 Fpeale 30-14 8.64 76.20 14.39 0.49 0.09 0.57 100.00 15-19 12.39 36.09 41.25 7.81 1.54 9.35 0.38 U.50 0.88 100.00 20-24 17.1JA 39.65 35.41 3.42 1.45 4.87 1.03 1.76 2.79 0.08 100.00 25-29 26.61 40.13 27. L4 2.52 0.89 3.41 0.62 1.23 1.85 0.08 100.04 30-49 52.99 31.78 13.43 0.66 0.48 1.14 0.13 0.44 0.58 0.04 0.01 100.00 50* 83.83 13.27 i.69 0.04 0.07 0.11 0.00 0.03 0.03 100.00 Total 38.60 38.51 19.39 1.96 0.63 2.59 0.27 0.54 0.03 0.00 100.00 Urc.: 5AUMS, 1977 Chapter 1 Appendix D5 Page 6 - 21 - Table IB.6: LABOR FORCE PARTICIPATION RATES BY AGE, SEX AND LOCATION, INDONESIA, 1977-78 Age Group Urban Rural Male Female Male Female 10-14 4.7 4.6 19.3 11.3 15-19 33.9 23.1 65.4 39.1 20-24 74.8 29.4 90.8 39.4 25-29 92.7 30.0 97.8 44.3 30-34 98.1 30.5 98.7 49.5 35-39 98.1 36.4 98.9 52.6 40-44 97.7 42.4 98.8 57.1 45-49 94.1 41.5 98.1 56.3 50-54 84.9 36.3 95.9 52.9 55-59 72.7 31.0 92.3 48.3 60-64 62.5 26.4 85.2 41.0 65- 43.5 15.0 65.8 25.2 Total 63.1 26.2 76.1 40.4 Note: Rates are arithmetic averages of eight quarterly observations. Source: SAKERNAS, 1977, 1978. Chapter 1 Appendix B Page 7 INDONESIA Table IB.7: URBAN LABOR FORCE PARTICIPATION RATES BY AGE AND EDUCATIONAL ATTAINMENT Age Group No School Elementary Iligi School Aczidenv and - Incomplete CcipleLc Junior Senior Uni versity Total Male 10-14 20.00 3.74 5.31 1.96 4.38 15-19 67.69 59.86 38.48 12.93 31.17 33.92 20-24 69.91 92.34 91.11 63.45 55.58 40.00 73.39 25-29 85.02 95.57 97.01 94.42 87.74 63.22 91.81 30-49 94.90 97.56 96.79 "7.48 98.21 96.38 97.14 1 50+ 66.17 76.30 63.03 57.53 67.27 87.96 67.54 Total 74.41 48.07 67.23 59.11 77.62 '6.51 62.26 Female 10-14 12.75 3.34 4.83 0.00 3.93 15-19 43.68 35.41 19.86 0.42 17.93 21.37 20-24 26.98 26.60 23.06 27.20 40.09 34.75 28.42 25-29 31.71 24.66 18.22 25.52 50.75 56.76 27.90 30-49 39.48 32.65 25.81 28.94 43.44 49.47 33.67 50+ 27.89 26.69 21.19 22.49 12.79 80.00 26.51 Total 33.32 20.18 19.50 20.01 40.48 49.04 24.29 Source: SAKERNAS, 1977 Chapter 1 Appendix B Page 8 INDONESIA Table IB.8: RURAL LABOR FORCE PARTICIPATION RATES BY AGE AND EDUCATIONAL ATTAINMENT Age Group No School Elementary High School Academy and Incomplete Complete Junior Senior University Total Male 10-14 56.68 13.00 23.77 16.27 17.45 15-19 88.53 79.92 67.30 22.91 21.91 66.41 20-24 93.68 96.61 95.75 73.47 65.85 41.18 90.96 25-29 94.17 98.75 98.68 94.58 94.34 83.75 97.68 30-49 98.03 99.12 98.84 97.31 99.71 87.77 98.68 50+ 81.65 88.43 85.68 73.31 78.13 33.64 84.21 Total 88.13 65.87 81.67 60.62 81.76 83.21 75.06 Female 10-14 24.47 7.70 12.92 5.60 9.89 15-19 53.99 47.72 32.65 9.15 29.84 38.49 20-24 39.81 40.78 35.03 27.35 59.05 50.00 38.45 25-29 43.95 43.99 33.83 31.93 62.79 84.38 41.10 30-49 55.56 48.03 36.99 38.46 81.45 59.68 50.61 50+ 38.45 38.23 28.46 29.41 100.00 38.41 Total 47.06 33.12 31.81 20.21 61.50 63.71 38.13 Source: SAKERNAS, 1977 - 24 - Chapter 1 Annex B Page 9 Table IB.9: MEASURED OPEN UNEMPLOYEENT RATES BY AGE, SEX AND LOCATION, INDONESIA, 1977-78 Urban Rural Age Group Male Female Male Female 10-14 17.36 7.09 3.89 2.44 15-19 25.34 11.84 6.68 4,66 20-24 18.03 12.54 5.30 3.26 25-29 6.11 4.36 1.74 0.71 30-34 2.41 0.99 0.39 0.26 35-39 1.30 0.63 0.31 0.23 40-44 0.66 0.40 0.23 0.16 45-49 0.88 0.00 0.43 0.03 50-54 0.91 0.06 0.45 0.06 55-59 1.94 0.04 0.14 0.15 60-64 0.26 0.00 0.11 0.28 65+ 0.10 0.53 0.01 0.25 Total 6.7 4.64 1.96 1.25 Note: Rates are arithmetic averages of eight quarterly observations. Source: SAKERNAS, 1977, 1978. Chapter I Apelidix R Table TB.10: YOUTH UNEXPLOYMENT RATES IN URBAN AREAS BY SEX AND EDUCATIONAL ATTAIN-ENT, INDONESIA, 1977 Age Group No School Eleoentary School Junior High School Senior Hfigh School Ince.eoplete Cumpleted Generdl Vocational General Vocattonal Academy Universtty Total MAIC 10-14 22.7 15.0 38.4 0.0 21.8 15-19 29.7 19.7 24.1 36.6 36.4 51.9 58.5 25.9 20-24 4.1 10.3 13.0 20.5 30.0 29.7 32.6 21.5 0.0 18.4 25-29 3.2 4.6 3.0 8.5 7.8 8.7 5.1 14.4 10.6 5.7 1 Total 2.2 4.8 6.2 8.5 8.8 9.0 12.5 3.8 1.7 6.4 Female 10-14 9.9 10.8 18.5 12.9 15-19 2.4 9.3 14.5 26.6 21.8 58.8 40.5 13.9 20-24 5.9 2.9 6.6 18.6 16.7 38.4 26.5 8.0 0.0 14.7 25-29 2.9 2.4 2.6 9.2 6.4 7.1 3.5 19.9 12.0 4.7 Total 0.7 3.4 6.0 11.9 6.6 20.3 11.2 9.5 4.3 5.4 Source: SAfERNAS. 1977 Chapter I Appendix A - 26 - Pop It Table 1B.11: - D 0.50 ha. Household & annual crop 4.4 3.9 1.6 Rice farm 20.0 10.2 3.8 Secondary crop 6.8 4.2 2.3 Farm labor 10.9 16.4 25.2 Trading 16.9 15.0 30.9 Handicraft 10.2 26.9 20.0 Fish pond & livestock 19.2 15.6 8.7 Other nonagriculture 7.1 5.3 6.3 Mutual help 4.7 2.5 1.8 a/ Originally White and Makali classified Class I households as having more than 0.50 ha of land, Class II between 0.25 ha and 0.50 ha, and Class III less than 0.25 ha (including same landless). The numbering of three classes given here is the reverse of that in the above paper. Source: Benjamin White and Makali (1979), p.38. 3.15 Even if we assume that all "farm labor" is expended on rice cultivation, the proportion of man-hours spent in rice is less than 30% for all three classes, even though these villages generally grew two crops of rice. 3.16 Gillian Hart's intensive study of labor utilization gave the following percentage distribution of time spent in different labor markets by apult men and women workers in her sample (all households taken together). - 1/ Gillian Hart, op. cit., Table V-9, p. 146. - 77 - Rice Rice inside outside village village Sugarcane Fish pond Tobacco Other Total Women 49.2 16.0 32.8 - - 2.0 100.0 Men 28.1 11.4 13.0 32.2 7.1 8.2 100.0 3.17 The importance of fish pond labor (in which women do not participate) gives Hart's coastal village a certain peculiarity. In White's villages a substantial amount of labor time was taken up in trading, animal care and feeding, and in handicrafts. 3.18 By concentrating exclusively on the rice labor market, observers may sometimes have underestimated the time typically spent by an agricultural laborer in gainful activity. This mistake has been made in other parts of the world because of the preoccupation with the main agricultural crop. 1/ 3.19 Interclass Differences in Market Participation. Important relationships between the market activity of household members and the economic status of households have been revealed by recent field research in Java. Since decisions on participation by different age-sex groups in a household are interdependent, it is appropriate that the household should be used as the unit of analysis. Gillian Hart divides her sa7ple of households into three economic groups, much as do Makali and White. 2 Her data on labor use in different activities by different demographic groups in the three classes show that there is relatively little difference in the working time in economic activity for adult males across classes (though naturally Class I male workers spend more time working with their own productive assets). But the hours worked in market activity increase sharply both for wom73n and children under 15 as the asset base of the household decreases. - SJ Cf. Bent Hansen reports that male "farmers" spent only 53% of their time in field work, while male farm laborers spent 58% in such work. For women the proportions were 19% and 31% of their total market activity, op. cit, p. 300. 2/ Hart's principle of classification is the control over productive assets (mainly rice land) and the ability of households to meet their consumption needs by the use of these assets. Class I households can attain a net income of at least 300 kg of milled rice per consumer unit p.a. from the land they control, Class II can cover "stable food needs" (150 kg of rice per consumer unit), while Class III housebolds cannot even meet these needs. 3/ See, in particular, Hart, ibid., Table V-5, p. 131. - 78 - 3.20 The proportion of time devoted to housework by women decreases from two thirds for Class I females to somewhat less than one half for Class II, and to as little as one third for Class III. At the same time, the proportion of time devoted to wage labor increases substantially for women as we go down the economic ladder. Apart from work on their own farms, the major amount of off-farm work done by Class I women is in trading which with some access to capital can be very lucrative. Class III women spend most of their work outside the household in wage employment, often in fields beyond the village (involving a significant amount of travelling time.) 3.21 A striking feature of the Hart's data is that the amount of time devoted by girls aged 10-15 to wage labor is almost as much as women in Class III households: it is three times the number of hours of wage labor contributed by girls in the next higher economic class. Boys aged 10-15 also devoted larger amounts of time to market activity in the lower classes, but with two qualifications with respect to the activity pattern of young girls: between Class II and Class III households the increase in the number of hours worked is not very large; and boys participate more in "fishing and gathering" than in wage labor. 3.22 Seasonal Variations in Activity. There is a well-known seasonal cycle, in the cultivation of the main crop (rice) in Javanese agriculture, the slack period extending between planting and harvesting of each season's crop. (There are in many areas two crops, a dry season and a wet season one.) Generally speaking,the slack months are from January through March, and from June through August. But although the fall in rice employment in the slack season is substantial, the evidence from available field studies suggests much less fluctuation in directly productive work. Thus Benjamin White reports: "Comparing the five markedly "busy" months and seven markedly "slack" agricultural months for men and women, we find that while the percentage of directly "productive" working time devoted to agricultural work declines from 48% to 29% (for men) and 36% to 11% (for women), there is no significant chInge in the total daily input of "directly productive" work for either sex." - Thus seasonal fluctuations do not take the form of substantial variation in unemployment, but in the proportions of different kinds of market activity, particularly for agricultural and nonagricultural work. Similar results emerge from Gillian Hart's field work in a different type of village, in which the proximity sugar cane and fishing provides clear alternatives to rice cultivation. _ 3.23 Labor's response to this seasonal variation in demand is largely manifested changes in participation in market activity, particularly in the case of women and children of higher economic classes. Hart reports that in the slack season, with the large decrease in wage rates, small landowning females withdraw completely from wage labor. Landless girls and women, 1/ Benjamin White (1976), p. 281. 2/ Gillian Hart, op. cit., Figures V-6 and V-7, pp. 147 and 149. - 79 - however, cannot afford to do this. Thus, the observed difference in the total time worked in wage labor between the classes is partly accounted for by the different seasonal patterns of participation in the labor market. Patterns of Returns to Labor in the Rural Sector 3.24 The last point brings us to the variations in returns to labor from different activities. Households from different economic strata participate in different ways in the labor market and thus have varying levels of labor earnings. Two previous points will be better understood within a standard framework of "perfect labor markets." These are (i) the differences in returns to labor for clearly distinguished age-sex groups; and (ii) the seasonal variations in wage rates. Households of lower economic status, because of the lower reservation price of potential workers, contribute more women and children to the labor market, and relatively more in the slack season than higher income households. 3.25 Recent research in Javanese labor markets is, however, beginning to highlight another reason for the lower return to labor supplied by poor households. This is the existence of differential wage rates within the same village's labor market for a given age-sex group and season. Wage rates in some types of activities are systematically higher than in others, and, insofar as workers from households with a higher economic status have better access to the "good jobs" in the rural sector, the returns to such labor are higher. This view of the rural labor market is, as we shall see, significant in the interpretation of trends in output and real wages in Indonesia in the last decade, as well as in the analysis of shadow wages. 3.26 It will be recalled that labor time devoted to rice cultivation takes up only a small part of the total market activity in rural Java, and that there is a marked seasonal variation in labor use in rice. But the returns for an hour of work seem to be the highest in the labor market for rice. A number of field studies in the 1970s in Java have established the pervasive tendency for returns to nonagricultural work in the rural areas to be at a lower level than the returns to agricultural labor, particularly rice. In White's field study of a poor village in Yogjakarta, the wages of agricultural wage labor were Rp 20/hour for men, Rp 24/hour for women in transplanting, and Rp 36/hour in harvesting. The various marginal activities - handicrafts, minor trading, sharecropping palm trees to make coconut sugar, etc7 - earned between Rp 6-10/hour. (The figures are given in 1976 prices). -7 Apart from estimating returns to labor, White's direct observation confirmed that wage labor in rice was regarded by workers in the village as the top of the hierarchy of income-earning activities. White writes: "If we were to rank the various productive opportunities in order of their returns to labor (with harvesting at the top, mat-weaving jj Benjamin White quoted by Gordon Hughes, World Bank draft, p. 20. - 80 - at the bottom and so on) we would expect to find that households would whenever possible choose the available combination of activities with the highest total returns to labor. Thus for example, women will often stop or reduce their trading or mat-weaving activity during harvest time to take advantage of the better returns in harvesting. Men may remain at home, cooking and babysitting to free their wives for the harvest; young children may herd livestock or cut fodder when there are wage-labor opportunities for their fathers, or they may cook and babysit while their father cuts fodder and their mother is planting rice, are so on. Mat-weaving is normally done only at times when there is no more productive activity available, particularly at night, or when one cannot leave the house (weaving, cooking and babysitting are often combined). It is hardly surprising that there are virtually no 'full time' mat-weavers (exceptions are elderly women who can do no other kind of work), for at some times of the year when rice prices are high and mat prices low it would requir7 20 hours of weaving per day to provide one adult with rice. ''L 3.27 A similar view was recorded by Penny and Singarimbun (1973) in another village in Yogjakarta. "The rice enterprise ... is regarded as being by far the most important single way of making a living - by the rural people themselves, and by the vast majority of other Indonesians.... Most of the "other work" mor eover, is low paying and distinctly not preferred." 2 3.28 Most intensive micro studies in rural Java have established that nonagricultural activities in rural areas are on average, a marginal activity for farm workers. References include Hart (1977), Kabul (1977), Rozany (1978), Sawit et al. (1979), White and Makali (1980), for villages in Central, East and West Java for the first three, r7spectively, and for six villages in different provinces for the last study. - 3.29 The segmentation of rural labor market is only one aspect of the operation of Javanese labor markets. The typically easier access to higher wage employment by workers of richer rural households is a second important 1/ Benjamin White, op. cit., p. 280. 2/ Penny and Singarimbun (1973). 3/ Hart, Gillian, op. cit. Kabul, Santoso: The Income Distribution and Employment in Desa Surbarajo Pasunam (M.S. thesis, Bogor, 1977). Rozany, Nurmanaf A. et al.: Report III, Analysis of Rural Labor Utilization in West Java (Bogor, 1978). Sawit, M. Husein et al.: Rural Dynamics Study, Report 1977/78, Bogor 1979. White B. and Makali, H: Wage Labor and Wage Relations in Javanese Agriculture: Some Preliminary Notes from the Agro- Economic Survey (The Hague, May 1979). - 81 - feature which field workers have recently shown. The most careful and elaborate evidence on this point comes from Gillian Hart. Hart studied the earnings function separately for males and females, and also for each month of the year in which the monthly wage labor income by sex was related (i) to the value of productive assets controlled by the household, and (ii) the hours per month spent in wage labor. The first independent variable assesses the influence of status per se on the return to wage labor, and the second tests the hypothesis that workers who put in longer hours (as those from poorer households might be expected to do) will tend to get a lower return to their labor. For males the asset variable turned out to be significant in 8 of the 12 months. Hart suggests that the four exceptional months are those of peak demand for land preparation and for fishing. The earnings function for females, on the other hand, shows a much smaller effect of the assets variable, except in the harvesting months. As already mentioned, females from the richer households tended to withdraw from the labor market in the slack seasons when wage rates declined. This is consistent with Hart's result that the wage effect of the number of hours worked was significantly negative for females in most months of the year, but not for males. It should be noted that the positive relationship between the possession of assets and the returns to labor in Hart's equation tells only part of the story. The analysis does not include the income from nonwage labor or self employment, and this part of labor income may be even more strongly related to economic classes. 3.30 The association between high returns per unit of labor and the household economic status, while consistent with the hypothesis of preferential selection of higher status workers for better jobs, is not conclusive evidence of the theory. An alternative explanation of the relationship would be that higher status workers are drawn into the labor market at particular times of the year, and have to be paid a higher rate to match their higher supply price. Two pieces of evidence would tend to favor this hypothesis: (i) Hart's evidence shows that the incidence of higher returns to workers from higher economic classes is typically in the slack seasons, while the alternative hypothesis would require that marginal workers with higher supply prices are utilized when there is additional demand for labor in the busy season; and (ii) field workers' direct observations on preferential treatment to some classes are really crucial on this point. Hart describes that in her study village there were opportunities for work in sugarcane fields, and some harvesting for cash wages (tebasan contracts) outside the village, in the slack season of rice cultivation. But the returns to labor in these activities were low; sometimes the workers had to undertake commitments to work for a specified period with the mandum (supervisor), and wage rates were further reduced by having to spend two to three hours a day walking to and from work. Hart observed that workers from landless families participated in these types of wage labor, while the more remunerative jobs within the village (although s Yradic and intermittent) went to workers from the higher status households. - Further field evidence on preferential treatment of different classes of workers in the important operation of harvesting is given in the next subsection. 1/ See Hart (1978), pp. 166-68. - 82 - 3.31 The factors causing the differentiation of the high wage rice labor markets from other rural labor markets, as well as the differential access to the rice market by different economic classes cannot yet be fully explained by available research. It is, however, unlikely that economic factors alone could provide a satisfactory explanation. The market for wage labor in rice is very much like the "formal" sector of the urban labor market which has attracted much attention in the literature. It is very likely that labor used preferentially in these high wage organized markets in the rural areas would be seen to be "superior" by employers. For example, they might be more productive because of their better nutritional standards. It has also been mentioned by some field workers that families with some land often provide one another with opportunities for wage labor in their respective rice fields. It will probably be difficult in practice to pinpoint the "superiority" of labor supplied by landholding families. The substantive point is that their real or assumed superiority stems from their economic class which is higher than that of landless families. "Surplus" Labor and Seasonality 3.32 It is possible to suggest a model of the Javanese rural labor market in which surplus labor in the sense described above (i.e., earning lower wages than the rice sector wage) exists only in the slack seasons. The fact that additional workers from relatively high-income households are drawn into the labor force in the busy seasons (at higher wage) might suggest a scenario of "full" employment in peak periods. But of course, although wages, as well as the supply of labor are higher in the busy season, it does not follow that wages in the rice sector clear the market, in the sense of equating the supply and demand for labor. Certain additional evidence would be useful: (i) is there a substantial amount of labor in marginal activities, earning significantly less than the rice wage even in the busy season; and (ii) is there evidence of discrimination against particular groups of workers in the peak period? 3.33 Clearly, further micro studies of rural labor markets are required to examine these issues. 1 Some of the data available from existing reports on the village studies refer to the whole year. White's work in a village in Yogjakarta, already cited, - refers specifically to the winter busy season of 1973. Another relevant question is: how long is the peak? White makes the point that a peak activity like harvesting in a typical field in his study village lasts no more than a few hours. 1/ The mission's recommendations on further empirical work are summarized in p. (viii) above. 2/ See p. 79 above. - 83 - 3.34 A field study by Ann Stoler of labor recruitment for harvesting in a village in South-Central Java sheds important light on the availability of abundant labor for this remunerative task and on the nature of discrimination between different groups of jobseekers. - The two harvests in the village (in the summer and in the winter) are periods of peak labor demand as the tasks must be done precisely and at the right time, with little possibility of staggering. Traditionally, it is a female occupation. Small farmers, in addition to employing household members, typically invite females from neighboring households to undertake harvesting. "By allowing members of a neighboring household to participate and paying them in kind with a relatively large share, a farmer is insured that female members of his or her household will be given reciprocal employment opportunities. Such arrangements are as much a means of spreading the risks of cultivation as they are a means of meeting labor requirements."2/ Large farmers employ harvesters from outside this network of kin and close neighbors. "As many as 50 to 150 women are used to reap the padi on fields of half a hectare and above. A large proportion are orang bin (literally "other people") from within and outside the village who seek out Shese larger fields where their harvesting chances are greater."'2! 3.35 Stoler makes the crucial point that "the notion that there is one set wage that prevails throughout a region or that applies to all harvesters on all types of land imposes homogeneity on a complex distribution system". Rather, the size of the bawan (the share of the harvested crop paid out) typically depends on the social proximity of the harvester to the host household. "Close relatives receive from one fourth to one half of what they harvest shares of one sixth to one eighth are given to women from neighboring households (tetangga), defined not only by their physical proximity to the host household, but by their mutual participation in gotong-royang activities. Shares of one tenth to one twelfth are given to orang bin, that is distant villagers and nonvillagers who fall outside the first two categories. In other villages the bawan for this group may be as little as one twentieth." 1/ Ann. L. Stoler (1977). 2/ Ibid., p. 682. 3/ Ibid., p. 683. - 84 - 3.36 Harvest labor by landless families, who operate outside the social network system, thus get a significantly lower harvest wage than landown,ng families, although they might work more harvesting days per household. ! Wages are not equalized for harvesting labor from different economic groups as in a perfectly competitive market. On the other hand, in terms of a model of "competitive monopsony", with each farmer facing an upward sloping supply curve of labor to himself, the larger landowners would pay higher wages since, because of their larger demand, they have to attract labor from landed families with a higher marginal supply price. But the outcome of the social network system is the opposite. It is the larger landowners who pay an effectively lower harvest wage, as they employ a larger population of harvest labor from landless families. Technical Change, Commercialization and "Surplus" Labor 3.37 The description of the bawan harvesting system in the previous section suggested that Javanese social relations ensured that some farms would employ more labor in harvesting at a higher effective wage than the wage harvesters from landless families were willing to accept. That is to say "surplus" labor existed within the body of workers involved in harvesting. The social determination of harvest share meant that the wage of harvest labor was higher than its marginal product. 3.38 Recent field studies of Javanese agriculture have pointed to what could be called "rationalization of labor use" in industrial economics, namely the tendency to reduce the complement of labor for a given task, bringing the marginal product of that labor close to the wage paid. The following developments have accompanied higher labor productivity associated with the new rice technology: increased interest in rice production as a commercial proposition (e.g., by new landowners from outside the village community acquiring land as a capital asset); and the change in the structure of political power within the village, which has reduced the importance of traditional relationships between landlords and laborers. 3.39 A major institutional change along these lines has been the shift in many parts of Java from the traditional bawan system of harvesting to the tebasan system. Instead of the wide sharing of output among community members, under the new system, farmers sell their standing crops to middlemen called penebas, sometimes before the harvest. The latter, who often come from outside the village, are not bound by traditional obligations. They can foreclose the opportunity for harvest employment to the majority of the villagers, and hire only a small number of more regular workers to harvest their crops. - 1/ In Stoler's study 15 landless households had 43.8 harvesting days per household with 2.5 kg of rice for harvesting day. By contrast 39 households with some land (< 0.2 ha) contributed 25.3 harvesting days for households with a payment of 3.1 kg per day (see ibid., p. 685). 2/ See, for example, W.A. Collier et al. (1974). - 85 - 3.40 Field workers have reported that in some areas, even though there has not been a pronounced shift to the tebasan system, landowners in recent years have tried to rationalize the harvesting system with the dual objective of limiting the number of workers in harvesting, and reducing the share of wages in total output. An example is the seblokan system reported by a team of field workers in a West Java village. I/ "This system limits participation in harvesting to workers who performed extra services (e.g., transplanting and weeding) without pay. The system reduces the real wage rate of harvester7 because the same share of output is paid for a larger amount of work." 2 The field survey reports that by 1978, the shift from the bawan to ceblokan system was complete, with the larger farmers taking the lead. The authors argue that in the past with lower yield and a higher man-land ratio, the customary bawan one-sixth share might have resulted in an effective marginal product of harvest labor close to the market wage. But with increasing population and rising yields, if the traditional share of harvest were maintained, the marginal product of labor would have been much above the market wage. Landowners had the option of switching to a system of harvesting using casual daily-paid workers. But this would have represented too much of a break from the traditional practices in the village. The institutional compromise was to reduce the number of harvesters and also cut their effective wage by asking them to perform additional agricultural operations for the landowners without pay. An additional feature of the new system was that it helped strengthen the patron-client relationship between the employers and the select body of workers who obtained exclusive harvesting rights. 3.41 A similar development il described by Frans Husken in a village on the north coast of Central Java. - Husken draws a sharp distinction between agricultural laborers in the village who sell their labor rather indifferently to all landowners, and sharecroppers "who have rather a close and practically exclusive relationship with a single landowner." Husken reports that in this village, with commercialization and intensification of cultivation, landowners have come to depend much more on the sharecroppers than on the traditional system which gave the women from laborers' households the right to participate in harvesting. Since the sharecroppers do virtually all the agricultural operations and also often other odd jobs for their patron landowners, the effective cost of labor to the landowners is lower, while total employment per unit of output is reduced. Thus, far from being an institutional form for sharing poverty, as Geertz suggested, sharecropping in some areas appears to be a method whereby landowners are cultivating rice on a more commercial basis, with the help of a smaller, exclusive body of 1 /M. Kikuchi et al. (1980). 2 /Ibid. 3/ Frans Husken (1979). - 86 - workers. The security of employment is attractive to those who are accepted as sharecroppers. Husken reports that in his study village, "compared with agricultural laborers, sharecroppers rarely have important additional sources of income." Their tied relationship with particular landlords frees them - even with a reduced share of output from the necessity of putting together a basket of income from a variety of marginal activities outside the rice sector. Husken also confirms the evidence of the field workers given above that returns per hour of labor are much lower in the marginal activities compared to the rice sector. Conclusions on Employment Growth and the Rural Labor Market 3.42 We concluded earlier that the evidence on the rural labor market pointed to the existence of a pool of labor in marginal activities which was potentially available to the rice sector at the higher returns prevailing in the latter. Thus, high output and employment growth could take place in the rice sector without an upward pressure on wages. The evidence on rationalization of labor in rice given in the last section tends to suggest that the growth rate of employment (in man-days or man-hours) need not have been as great as might have been expected from the difference between the growth of output in agriculture and the growth of the labor force dependent on that sector. Surplus labor was, in other words, shed from the rice sector, as landowners were induced toward a position of profit maximization. 3.43 In the absence of data on man-hours of labor devoted to rice production - or its rate of growth - it is not possible to assess the relative importance of these two factors in preventing upward pressure on wages in the face of significant output growth. But this question is important for the evaluation of employment aspects of Indonesian economic growth. Rationalization of labor, as described, tends to create a marked dualistic structure in the rural labor market, in which participation in the high wage rice sector is restricted to a limited number of workers, employed reasonably full time, while the majority of job seekers is forced even more than before to meet their subsistence needs from the low-paid nonrice activities. Clearly, the emergence of such a "protected" sector of employment within the rural labor market, hinted at by the field studies, needs much more thorough investigation. An essential step towards fuller understanding would be the systematic collection of data on hours of work in different activities by the rural labor force, and, in particular, by those who specify agriculture as their major occupation. The Urban Labor Market 3.44 We turn now to a discussion of the urban labor market, and the determination of wage levels in this sector. It might be useful to present some preliminary thoughts about the factors influencing urban wages in an economy like Java's before reviewing the available empirical data. We shall be mainly concerned with unskilled workers or production workers requiring only a low level of skill (i.e., what can be acquired in on-the-job training of about three months). - 87 - 3.45 The starting point for the determination of urban wages is the recognition that the minimum urban wage will be approximately equal to the alternative income of rural-urban migrants in agriculture. The supply price of migrants thus defined is, however, ambiguous. We must distinguish between different types of migrants to the urban area - in particular temporary and permanent migrants - who have different supply prices. In the Javanese context temporary migrants consist both of "circular migrants" who return to the rural areas after a spell of urban work, and "commuters" who maintain their rural residence but work in town for at least part of the year. In either case, these temporaries are absent from their rural activity only for a short period (or periods), and migrate to the towns without their families. Their supply price will, be below that of permanent migrants for two reasons: (a) the net income loss to the family farm due to the migrant's temporary absence will be small, either because the migrant is away in the slack season, or this loss can be partly offset by other family members for a short period; and (b) the cost of supporting a family in town will be substantial for the permanent migrant, particularly as the opportunity for females and children to supplement the family income is less in the urban areas. 3.46 The existence of two basic types of migrants with different levels of supply prices means in effect that the permanent migrants will not be employed by urban employers, unless they are convinced that their higher wage cost is offset by higher productivity. Thus, we will find that firms in the formal sector of the urban labor market, where the relationship between stability and labor productivity is strong, will generally set their wages at a high enough level to attract the permanent migrants. In the informal sector, where the link between stability and productivity is weak, wages will be determined by the lower supply price of temporaries. 3.47 This process, however, does not fully explain the difference in wage levels for unskilled or low-skilled labor between the formal and informal sectors. Once firms develop a stable body of workers, further upward pressures on wages are generated. Labor attached to particular employers creates, in a sense, firm specific labor markets, in which wages may be divorced from the supply price of such labor. The durability and smoothness of employer/employee relationships enters the objective function of employers, at least as much as the desire for cost-minimization. Thus, the rent or surplus, created within the firm through technical progress or on-the-job training, tends to get shared between employers and employees. That is to say, the wage level, is related to the level of average labor productivity in the firm, and is higher if the latter is higher. Clearly this phenomenon is more likely if the share of the wage bill in total costs is small (as it tends to be in capital intensive firms with modern technology); also it will be especially important in foreign-owned or multinational firms in which there is considerable social/political pressure to share profits with the workers. 3.48 In Indonesia the factors discussed in the last two paragraphs cause inter-firm differences in wages in the urban market for labor of low skill. - 88 - The impact of minimum wage legislation or trade unions is virtually negligible. The Importance of Temporary Migration 3.49 In some Asian countries like India, the large role of temporary migrants without their families is revealed in the abnormally low ratio of females to males in the large cities. Urban Java exhibits no such markedly different sex ratio. There is, however, evidence of temporary migration which does not show up in the urban sex ratio partly because of the nature of the census, and partly because of the importance of "commuters" in the temporary migration stream. 3.50 The 1971 census classified 6.4% of the population of Indonesia as migrants. But the figure includes only those who had changed places across provincial bounderies, and who had been away from their place of residence for six months or more. Hugo gives three sets of information of the importance of circular migration: 1/ (a) A fifth of all intraprovincial migrants recorded in the census were "lifetime return migrants", i.e. persons born in the province of enumeration whose immediately previous residence had been elsewhere. This was a surprisingly large population considering it referred only to migrants crossing provincial boundaries and staying more than six months. (b) The census of 1971 records that in West Java the number of males over 10 years old who had worked in agrriculture during the last season was 39% greater than those who were employed in the census week. The percentage difference between the "seasonal" and ''icensus'' agricultural work force was more than twice as great in the urban areas of West Java and Jakarta. Part of this difference is most likely due to the persistence of migrants who divide their time between rural agricultural and urban nonagricultural activities. (c) The most direct evidence on the importance of circular migrants and commuters comes from Hugo's survey of 14 villages in West Java in 1973. Recording only migrants who had moved to seek or engage in work or formal education, Hugo found that no less than two thirds of his sample of migrants were nonpermanent (i.e., not meeting the census criteria of being absent for six months or more). Of the temporary migrants, about 20-25% were "commuters" who returned to their villages nightly but worked daily in towns. 1/ Hugo, Graeme, "Circular Migration", Bulletin of Indonesian Economic Studies, Volume 13 (3), 1977, p. 59. - 89 - Difference in Earnings of Temporary and Permanent Migrants 3.51 Hugo also collected data in the occupational distribution of permanent and temporary migrants. His figures are reproduced in Table 3.2. The hypothesis of the concentration of temporaries in irregular wage employment or low grade self employment is borne out by the data. It should be noted that casual employment on daily contracts is a feature of many well- established large formal sector enterprises in urban Indonesia. Table 3.2: DISTRIBUTION OF PERMANENT AND TEMPORARY MIGRANTS BY EMPLOYMENT IN FORMAL AND INFORMAL SECTORS (14 SURVEY VILLAGES, 1973) Temporary Permanent Income Earning Activity migrants migrants Formal Income Opportunities Public sector, wages 6 22 Private sector wages (permanent) 13 42 Private sector wages (day labor) 15 1 Informal Income Opportunities Primary and secondary activities 3 1 Tertiary enterprises with large capital inputs 1 2 Small-scale distribution 41 28 Transport 16 1 Other services 5 3 Total 100 100 Source: Hugo (1977). 3.52 Hugo's wage data showed that at the time of the survey (1973) that the mean weekly earnings of a private sector day laborer were Rp 1,700, as contrasted with mean earnings of Rp 2,400 for permanent wage laborers, Rp 3,800 for Pegwai Negari in the Public Sector and Rp 3,000 for an airline worker, thus, confirming the hypothesis of temporary migrants working for lower wages in irregular jobs, compared to higher earnings in more permanent work. The Wage Ladder in Indonesian Manufacturing Enterprises 3.53 The propositions advanced earlier in this section would tend to support the hypothesis that there will be a hierarchy of enterprises paying various wage levels for similar workers, the highest being multinational firms; somewhat lower would be domestic large firms, and lowest would be small - 90 - firms. This wage ladder would probably exist for service sector firms as well as manufacturing enterprises. The available evidence in Indonesia, however, relates only to the manufacturing industry. (a) The Census of Manufacturing, 1974 3.54 The major official source of data for the manufacturing sector of Indonesia is the Industrial Census of 1974/75, published by the BPS in 1978. The census covered cottage (and household) industries (CIs); small-scale industries employing less than 20 persons (SIs); medium industries employing between 20 and 99 persons (MIs); and large industries (LI). The CIs employed 80% of the total persons although 95% of their workers were unpaid family workers. The productivity of labor in CIs was very low, so that they accounted for only 14% of total value added. At the other end of tIhe scale, LIs employing 9% of the workers accounted for 62% of value added. _' 3.55 Our interest in the results of the census is in the earnings differential per paid employee. Of the 1.1 million paid employees recorded by the ce2sus, 16% were in CIs, 24% in SIs, 18% in MIs, and the remaining 42% in LIs.- The average wage/salary per day per person employed was as follows (in Rupiah): _ cI SI MI LI 255 285 390 650 3.56 The differentials by size of establishment reached in these figures are, of course, not corrected for the composition of the labor force. In particular they are exaggerated due to the inclusion of salaried employees, who are much more important in large firms. 1/ World Bank, Report No. 2490-IND, November 79, Volume I pp. 9-12. 2/ Hugo, Graeme, "Circular Migration," Bulletin of Indonesian Economic Studies, Volume 13 (3), 1977, p. 59. 3/ Ibid., p. 12. - 91 - (b) Manning's Study of Indonesian Manufacturing Wages 3.57 The second source of wage data comes from a diss rtation by Professor Manning of Gaja Mada University in Yogjakarta. 1 Manning surveyed a sample of firms in three industries: weaving, kretek and cigarettes during the period December 1975 to August 1976. Altogether 95 firms were successfully covered with a stratified random sampling to include four criteria: location, size of firm, degree of mechanization and ownership. Several of the major producing areas in each industry were represented. This is the only systematic study of this type available for Indonesia. In spite of one glaring deficiency - that the wage rates are averaged for selected occupational groups in the firms, and not for each individual worker - several important insights into the wage structure of Indonesia manufacturing can be obtained from this work. 3.58 The overall picture of inter-firm wage differentials can be obtained from Table 3.3, in which the data on hourly wages are presented for the firms, categorized by one characteristic at a time. The wage data are for "machine operators" representing the major semi-skilled occupation in each industry: loom operators in weaving; cigarette rollers in kretek; and machine tenders in the cigarette industry. 3.59 It is evident that foreign firms will have higher K/L ratio and also will be larger in size. Does ownership play an independent role in causing higher wages apart from the capital intensity factor? The data in Table 3.4 show that it does - and the magnitude of the effect is very large. 3.60 Table 3.4 also contains data for unskilled workers and white collar employees (clerks). A significant point is that there continued to be a hierarchy of wages for both these labor categories in the three types of firms surveyed: domestic firms with low K/L ratio, domestic firms with high K/L ratio, and foreign firms. (The figure for clerks in non-mechanized domestic firms is probably biased by the small number of observations in the cell.) 3.61 Manning employed multiple regression analysis to determine the parameters of hourly earnings using the characteristics of the firms in his sample. Some of the independent variables were undoubtedly correlated. 1/ Manning, Christopher G., Wage Differentials and Labour Market Segmentation in Indonesian Manufacturing, Ph.D. Thesis, Australian National University, October 1979. - 92 - Table 3.3: MEAN HOURLY WAGES BY SEVERAL FIRM CHARACTERISTICS (all firms and weaving, operator 1) All Firms Weaving Rp No. Rp No. Firm Characteristics firms firms (1) (2) (3) (4) 1. Capital Intensity Non-Mechanized 30 32 33 12 Mechanized: low K/L ( < Rp 1.5 m.) 55 29 55 26 high K/L ( > Rp 1.5 m.) 107 21 88 15 2. Ownership Domestic: pribumi 45 26 49 22 non-pribumi 42 39 50 21 state/coop. 82 4 82 4 Foreign 130 13 103 7 3. Number of Employees 100 39 24 43 19 100 - 499 56 31 54 20 500 + 80 27 87 15 4. Industry Kretek 29 20 - - Weaving 59 54 Cigarrettes 133 8 - - 5. Location Jakarta 89 14 85 12 Bandung 65 13 65 13 Other 50 55 46 29 6. Years of operation 5 or less 75 28 80 18 5 - 20 47 27 21 More than 20 55 27 41 15 59 82 59 54 Source: Java Wages Survey, 1975-76, Manning, op. cit. pp. 240 and 242. - 93 - Table 3.4: HOURLY EARNINGS OF OPERATIVES (MEAN) (Rupiah) Domestic Non-mechanized Low K/L High K/L Foreign Weaving firms 33 57 78 103 All firms 30 52 74 130 Unskilled 38 37 63 95 Clerks 154 97 131 171 Source: Manning, op. cit., p. 242. The correlation coefficient for example between the dummy variables for high K/L and foreign ownership was 0.59 for the sample. In spite of this, it was precisely these two variables which were both significant and had reasonably high coefficients. 3.62 Size of firm (as measured by number of workers employed) was significant for the log-linear form of the equation, but not the linear. Thus, Manning concluded that it was the degree of mechanization, and not firm size per se which was important in explaining wage differences. An important conclusion was that several variables which might have been expected to cause wage differentials were not significant: rural-urban location; industry; years of operation; and pribumi or non-pribumi ownership. 3.63 How much of the very large wage differential between firms of different capital intensities, as well as between foreign and domestic firms, is explained by different "qualities" of labor attracted to these enterprises? It should be stressed that the above analysis relates to fairly homogeneous skill groups. Thus, certain personal characteristics of the workers - for example, sex ratio, education, seniority, etc. may possibly contribute to wage differentials. We have already mentioned that the earnings variable in Manning's regression model was the average earnings of the group in the firm in question. The personal characteristics were also measured by the mean value for the group. When these were included along with the firm characteristics in a multiple regression analysis, sex ratio and seniority (but not, surprisingly, education) were significant. But controlling those variables in the case of foreign ownership continued to exert a strong "net" effect on earnings: the coefficient of the variable was reduced from 0.27 to just 0.20. 3.64 Does stability of the labor force explain the difference in "net" earnings between more and less capital intensive firms, or between domestic and foreign firms? It is, of course, not possible to pinpoint the direction - 94 - of the casual relationship between stability and wage levels, but is there an association bet ween the two? Manning's data for his sample of weaving firms is as follows:1 Type of Mean rate of Rate of enterprise turnover /a absenteeism /b Domestic Nonmechanized 42.6 13.3 Low K/L 22.0 7.0 High K/L 24.3 7.1 Foreign 18.3 3.5 All 26.8 8.0 /a The turnover rate is the ratio of the total number of separations (quits plus dismissals) per annum to the average number of workers employed during 1975. /b Absenteeism is the ratio of mandays lost (excluding annual leave) to the product of the number of employees and the number of days worked in the months before the survey. Source: Manning (1979). 3.65 The general relationship between stability and wage levels holds, but the magnitude of the difference does not seem to be as much as might be expected from the observed wage differentials. There is a substantial difference in the stability of the labor force between nonmechanized and mechanized firms; and it is more than likely that the low supply price of temporary migrants mentioned earlier holds down the wage level in the nonmechanized firms. But, it would appear from the quantitative data that the substantial wage difference between high and low capital intensity firms, on the one hand, and between the foreign-owned and domestic firms on the other are due to factors other than the stability of the labor force. The considerations mentioned at the beginning of this section are probably significant. 1/ Manning, op. cit., p. 351 - 95 - (c) The Mission's Field Data 3.66 The employment mission undertook visits to a number of industrial units in Jakarta, Surabaya and Yogjakarta to get a first hand impression about hiring processes and wage levels. A conscious attempt was made to sample firms of different sizes and ownership classes. 3.67 The mission found that the minimum wage for a typical time rated worker (who would generally be unskilled) in the small-scale manufacturing sector at the time of the mission (March 1980) would be around Rs 500 per day in the Jakarta area and Rs 400 per day in Yogjakarta. Sometimes this wage would be supplemented by a meal (costing say about Rs 50). These rates were sufficient to attract an adequate supply of unskilled labor on a regular basis, as evidenced by the fact that a State-run industrial estate near Jakarta operated a three month probationary system for new recruits. Workers came both from the immediate vicinity and from the areas outlying Jakarta. On the other hand, the unskilled wage in these units was rather lower than that of unskilled construction workers requiring heavy labor. In Jakarta the BPS reports a daily wage level of Rs 650 for construction workers in the last quarter of 1979, but wages for construction workers do vary by firm size. One large firm quoted the rate in Jakarta at Rs 1,250. 3.68 The small-scale manufacturing sector makes use of a lot of labor with some skills which can be acquired in a short time, but which are nevertheless job specific. Many of those semi-skilled workers are paid a piece rate, and there is clearly an advantage for both parties to improve their productivity through continued attachment to the firm. There is a facet of wage differentials which is related to the acquisition of skills. The high return is, however, not clearly a return to "investment in on-the-job training" as some extreme views suggest. The relevant point is that there are many more potential semi-skilled workers, able to perform the job, than those who in fact get selected, and the return to their firm specific skill has an element of rent - which is their share of the surplus created in the firm. The mission found that semi-skilled workers in small and medium-size industrial firms in the Jakarta area could earn Rp 700-800 per day. 3.69 The larger multinational and joint venture firms occupy a distinctive part of the labor market. Wage rates are clearly set out with distinct scales for different grades of skill. Typically, the scales contain a basic wage, provision for family allowances, and annual increments. A peculiar feature of all the firms visited was that men and women had the same scales of pay for the same grades. Wages varied from firm to firm within this sector, but even the lowest wage which the mission found in this group was well above the wages in native firms elsewhere in Jakarta (a finding fully in agreement with Manning's conclusions already reported). A few examples will suffice. 3.70 Firm A - was a joint venture between an international British firm, a Chinese Indonesian firm which has already existed, and the Commonwealth Development Corporation. It employed about 300 workers with a 2:1 ratio of males and females. Most workers were grade 2 "production workers", having had - 96 - 3 months of training. The daily earnings of a married person with two children with about 1 year's service was Rp 1,200 with allowances, plus Rp 170 for working an evening shift. 3.71 Firm B - was a joint venture firm with a parent Dutch company. The minimum monthly wage was Rp 31,500 with a maximum of Rp 58,000 after 30 years service. Training did not take more than 1 month. There was a 40-hour week, and allowances of various kinds could be as much as 40% of basic wages. 3.72 Firm C - was an American multinational established only 5 years, geared to assembly-line production in Southeast Asia for turning out sophisticated electronic products for markets in industrialized countries. It employed 6,000 young girls who required minimal training. The daily wage was quoted as Rp 1,500 (including the value of free meal and transportation), with a maximum potential of Rp 1,800 with merit increase. 3.73 Firm D - was a well-established American firm producing a major commodity for the domestic market. It employed 1,700 workers and the wage for an unskilled worker was quoted as Rp 50,000 per month, with another 40-50% in fringe benefits. 3.74 Why do these firms pay such high wages? Management is quite aware of the potentially "unlimited" supply of labor from the surrounding countryside willing to work at the (lower) prevailing wage. Recruitment is usually done by intermediaries who have direct contact with the workers. Generally, there seems to be a 3-month probationary period for all workers recruitetd, and turnover is high during this period. By contrast once "confirmed" the rate of turnover is very low. 3.75 Such a large difference between the wage offered and the supply price of potential recruits, should produce a massive surplus of applicants over available jobs, and is consistent only with a rigorous screening procedure, in which perhaps only one in a hundred is found suitable for employment. However, the low level of skill required of a typical worker in most of these firms scarcely deserves this type of preselection. 3.76 Sometimes objective criteria are established which narrow the population of applicants. For example, Firm D, which had the highest wage level, among the firms surveyed by the mission, now requires its new recruits to be high school graduates. But, it is clear that this condition has been established ex post, after the wage scales had been established at a high level. 3.77 The number of new recruits to regular jobs in these high wage enterprises is not very large since turnover of permanent workers is very low - not more than 5% p.a. Also, increases in output are largely met by increased labor productivity. Firm A quoted its elasticity of employment with respect to output at 0.5 over the last five years. Firm D has had a static work force for several years, with incremental labor productivity coming partly through increased mechanization. - 97 - 3.78 One incentive for management to restrain the size of the permanent work force is the increasing number of firm-specific workers' organizations that are being established with Government support, although the trade union movement as a whole is still weak. The scales of pay and the associated conditions of work are subject to the scrutiny of the Government Labor Department. In Firm A the labor intermediary has the specific talent, contacts and responsibility to mediate between management and the Labor Department. But, in some other firms this role is gradually being filled by an enterprise union, concerned with protecting job security for the regular employees. The growing difficulty of dismissing permanent workers is likely to slow down the growth of the established work force in these firms. Increased demand for labor has been, and is likely in the future to be, met partly through increased productivity (with or without mechanization), and partly through larger use of probationary or casual workers. 3.79 We should emphasize that the institutional influences just mentioned are a consequence, and not the cause, of the existence of an elite high wage labor force. Foreign capital, operating with new technology in a less developed region, creates its own kind of labor - a labor aristocracy set apart from the income (and cultural) level of the general labor force in the area. This same tendency to establish a firm-specific high wage level exists in modern large-scale firms of domestic origin. But foreign capital and ownership seem to add an extra dimension. The process of formation of this class of elite workers cannot be described in detail within the confines of this report, and certainly cannot be handled by the standard tools of traditional economists. But this mission will have made a contribution, if it can focus, wide-spread attention on this special segment of the labor market which has become a major feature during the recent period of growth. The magnitude of the differential in wages between this sector and the rest of the economy is very large - a multiple of perhaps 2 or 3. 3.80 One feature of this type of firm specific labor which represents a change from traditional Indonesian practice is that there is an implicit contract with management (quite apart from the secondary effect of institutional factors) which protects the employee and his wage level. A relationship between the employer and his employees particularly in the mind of the employer is an integral part of the development of a firm-specific labor force. For example the mission confirmed that wages in the large-scale sector had increased substantially following devaluation. - In all the large firms visited, wages were increased between 40% and 55%, generally in three phases, between January 1979 and early 1980. In some cases, the wage increase seemed even to anticipate the cost of living increase following devaluation. 3.81 Finally, we should mention that, in spite of the relatively high wage observed in the modern sector firms, labor costs were often only about 1/ This was previously reported by a Bank Economic Report, World Bank (1980), 3.06, p.11. - 98 - 10% of total costs. (This would be somewhat higher if the total salary bill were included.) This is probably one reason that management follows a high wage policy. Also Firm C, which has several plants in Southeast Asia, reported that the share of wages in total production costs in Indonesia is the lowest in the region. According to their records in Singapore the wage share was 3-1/2 times, higher, in Korea 6 times, and the Philippines about one third more than in Indonesia. Conclusions on the Working of Labor Markets (i) The Wage Ladder 3.82 The views on Indonesia labor markets presented in this chapter differ in some respects from conventional models of "perfect" labor markets. In the latter, market clearing wages in different parts of the market minimize wage differentials for labor of equivalent quality. Generally, labor market imperfections will not be significant if there is true mobility of labor and institutional influences, such as minimum wage legislation or trade unions, are not significant. Certainly, in Indonesia such institutional rigidities are slight, and there is evidence of a highly mobile labor force - often dividing its time between rural and urban areas to eke out a livelihood. But, in spite of conditions favoring a perfect labor market, returns to low skilled labor differ widely for different activities in both the rural and urban labor markets. The reasons for such different wage ladders are only now beginning to be investigated. Wage labor for rice cultivation occupies the top of the hierarchy in the rural sector, and it appears that workers from households of better economic classes have preferential access to this market. In the urban areas, the upper rungs of the wage ladder are occupied by large modern firms, especially joint ventures or multinationals. One difference between the urban and rural markets, stemming from variations in the incidence of multiple occupations, is that only a small proportion of the urban labor force has access to the high wage markets, while, in the rural areas, all workers can participate. But for the latter, the proportion of income earning activity devoted to these markets probably averages less than half, although within a fairly wide range depending on economic class. Outside the high wage sectors, in both rural and urban areas, large amounts of labor participate in the "informal" sectors, in a variety of activities, usually at fairly low wages. 3.83 It is thus easy to see that one of the traditional tests of labor market imperfection in LDCs - the extent of the rural-urban wage gap is of limited significance. With unskilled or semi-skilled labor facing a wage ladder in each sector, the comparison depends on the particular segments of the market being examined. A summary statistical picture of the Indonesian wage ladder is given in Table 3.5. It has been mentioned that circular migrants or commuters - commonly found in Indonesia - could be expected to keep the return to labor in the "informal" sectors of the urban and the rural areas fairly close together. But accurate data on incomes in these activities are particularly hard to get. - 99 - Table 3.5: THE IHIDONESIAN WAGE LADDER, 1976 Agriculture Labor Daily Earnings Rp. 1. Hart (Men, average for all 210 seasons, Central Java). 2. Makali and White, West Java, 275 average both sexes. 3. Padat Karya, both sexes, West Java. 200 4. White, Jogjakarta, Hoeing, Weeding 96-116 Harvesting 169-211 Plantation Average Earnings, Permanent Workers, All Java 302 West Java 360 Construction, Urban Centers BPS, Unskilled workers: Jakarta 409 West Java 408 Central Java 293 3IS, Unskilled workers: Jakarta 500 Jogjakarta 250 Manufacturing, Unskilled Workers Non-mechanized firms 304 Domestic firms with low capital labor ratio 296 Domestic firms with high capital labor ratio 504 Foreign firms 760 Sources: Agriculture: Gillian Hart (1976). Appendix Tables Padat Karya: Original data obtained from the files. White and Makali (1979). Appendix Tables. White (1976). 1973 data corrected to 1976 prices by Gordon Hughes (1980), p.20. Plantation: BPS Wages on Estates, 1976, 1977. (Average earnings of all permanent plantation workers in all crops taken together). The data for West Java are 1977 average earnings corrected to 1976 prices by using the Rural Price Index for 12 food articles. The reported earnings for 1976 seem to be abnormally low for West Java. Construction: Data collected by the BPS (Statistics Department). Data collected by the Building Information Center of the Provincial Government. Manufacturing: Data collected from a sample of firms by Manning (1976). Manning data on hourly earnings have been converted to daily earnings assuming an .8-hour day. - 100 - (ii) Surplus Labor and the Elastic Supply of Labor 3.84 The view of the labor markets presented here has two important implications. These are (a) the definition and concept of "underemployment" or surplus labor; and (b) the macro-economic implications for economic growth and labor supply. (a) The Meaning of Surplus Labor 3.85 In the market clearing model of the labor market, there can be no involuntary unemployment. It is possible, however, to have less than "normal" hours of employment, in equilibrium, if we attach the concept of the casual labor market to that of market clearing wage. In peasant agriculture, labor is hired on daily contracts. If the market functions perfectly for casual labor, labor demand will be distributed randomly among the job seekers, so that, over a period of time, every worker shares (more or less) equally in the total man-hours of work. The average number of hours secured by an individual job seeker would, with this mechanism, easily be less than what might reasonably be considered normal full employment, thus leading to a situation of underemployment for all workers. 3.86 We suggest the following modifications to this market clearing model of the rural labor market: (a) there are differentiated wages within the labor market, with wages in the organized market being higher than elsewhere, and consequently the supply of labor desiring to work in the sector exceeds demand; and (b) rationing of jobs in the rural rice labor market is partly determined by economic class, and workers from all classes participate in different segments of the informal rural market at lower returns to labor. 3.87 Underemployment does not take the form of strikingly short hours of work in economic activity. On the contrary, workers from poor households put in very long hours, much longer than workers from more prosperous households. "Surplus" labor can still be said to exist in the limited and particular sense that, with the returns to labor being low in marginal activities in rural areas, there exists a potential supply of labor hours to the rice labor market which is perfectly elastic at the going rate. 3.88 Adding to this picture both the urban wage structure, and the differentiation of urban labor markets, the result is a perfectly elastic supply of labor to the high wage modern sector. (b) Elastic Supply of Labor 3.89 The macro economic implications of this labor market model are obvious. It permits the economy to achieve a high rate of growth (and employment) with little impact on real wages in the organized sectors. 3.90 The perfectly elastic supply of labor rests on maintaining the organized sector above the level of labor earnings in marginal activities where wages are presumably determined competitively. While this report cannot answer fully what determines wages in the organized sector, we can point to - 101 - the wage ladder as an empirical fact, and suggest the need for systematic research in this area. On the basis of what has been said, a few points do stand out. First, the wage in the organized sector is determined neither by the average productivity of labor on family farms, nor by a notion of subsistence. If the former were true, wages would have increased pari passu along with the rapid growth in labor productivity in agriculture during the last decade. On the other hand, the notion of a subsistence norm does not conform with the existence of a wide variety of wage rates observed, even in the same occupation, (e.g. hoeing), in different village labor markets. Secondly, the organized sector wage, even in the same occupation or industry, should be viewed, not as a single level, but as a band of wage rates. The constancy of the wage over time thus refers to the evidence that the band has remained relatively stable; but wage rates are free to, and do, fluctuate within the band while maintaining their same relative position. Thirdly, the concept of the band of wage rates implies that supply and demand forces do affect wage rates but within the limits of the bands which are institutionally determined. Thus, observed intervillage or seasonal variations in wage rates are partly the result of "market" forces operating within the band. Fourthly, while we have little research on the nature of these "institutional" forces it does appear that the hierarchy of wage rates (i.e. the wage ladder) do correspond quite closely to the hierarchy of labor productivity levels in different activities within the organized sector. Thus, we suggest that the starting point for investigating the determination of wages in the organized sector might be the classical idea of a "just wage" related to some perception of normal productivity in the occupation in question. Extent of the "Organized Sector" in the Indonesian Labor Market 3.91 One final question remains. What proportion of the total Indonesian (or Javanese) labor market is represented by the "organized" sector in which earnings are significantly above those in alternative sectors where wages are determined more competitively by supply and demand? The "marginal" activities within agriculture, and the "informal" sector outside agriculture (i.e. small- scale manufacturing and services) would constitute the competitive part of the labor market. 3.92 Numerically, the most important part of the organized sector covered by our wage data is agricultural labor. Landless workers as well as small farmers participate in the market for wage labor, but the data on employment does not disaggregate the time devoted to wage labor by different classes of agriculturists. According to the SAKERNAS figures for 1977, 25.7% of the total labor force had agricultural wage employment as their main occupation. Obviously, a proportion of "own account workers" in agriculture would also be spending some of their time in wage labor, and similarly not all of the time of those reporting agricultural wage labor as their main occupation would be used in this sector. If these two points balance each other out, it seems plausible that the proportion of total labor time used in agricultural wage labor in Indonesia is 25%. - 102 - 3.93 The plantation and construction sectors/are not big employers. In 1977, the former employed about 135,000 w?rkers 1 and the number of "employees" in construction was 779,000 2' out of a total estimated labor force of 48 million. Thus, together they constituted no more than 2% of the labor force. 3.94 The data on the Indonesian wage ladder show that wages in the large- scale manufacturing sector are at a relatively high level, and seemed to have been more or less unchanged in real terms during the 1970s. The factors affecting wage determination in large-scale manufacturing firms would also pertain to large units in the tertiary sector. A crude estimate of the likely proportion of employment in the high wage formal sector could be determined by the fraction of wage earners receiving more than a certain level of income in these activities. If we arbitrarily define the lower bound of the high wage sector in 1977, by a monthly income of Rp 20,000 (or Rp 650 per day) then the SAKERNAS information on the employee earnings by industry suggests that about 20% of the employees in manufacturing and 45% of the employees in social and personnel services earned more than this amount. The percentages of high wage earners in other industries fell between these two limits. Manufacturing and services were, in fact, the two numerically largest industries outside agriculture, accounting for 66% of all nonagricultural wage employment. Thus, as a broad conclusion, the high wage in nonagriculture formal sector in Indonesia probably accounted for 30-40% of wage employment outside agriculture. 1/ BPS: Wages Paid on Estates (1977), Table 1, p. 4. 2/ SAKERNAS (1977), Table 10.9, p. 90. - 103 - Chapter 3 Appendix A Page 1 Models of the Rural Labor Markets 1. It should be apparent from the report thus far that the diagnosis of the employment problem in Indonesia (identified as a situation with a great deal of "surplus" labor in marginal activities) depends crucially on the model of the rural labor market which one accepts. The choice of the most appropriate model will also influence the interpretation of past trends in key macroeconomic variables (particularly constant real wage in agriculture with significant trend growth in output per worker), and also future prospects of employment growth and wages (particularly the approach to the"turning point" identified in Chapter 4). It is important, therefore, to set out the model more formally (and starkly), and also pinpoint the difference between, it and the neo-classical model of full employment in the rural labor market. 2. In this Appendix, after recounting the key facts about the rural labor market (given in Chapter 3), the neo-classical model which might "explain" these stylized facts is outlined.l/ The predictions of this neo-classical model are then compared with those of the segmented labor market model to see if there are some stylized observations which might help us to discriminate between the applicability of alternative models to the Javanese rural labor market. In the second section of the Appendix the determinants of the wage level in the rice sector and of earnings in marginal activities - as as well as the implications of their relative movements over time - are explored. The Stylized Facts 3. The following are the more important stylized facts which were discussed in the section on rural labor markets in the text. (a) The amount of time spent by rural labor in marginal activities (i.e., outside the village rice labor market) is substantial and could be as much as one-half of the total working time of low income households. (b) Returns to labor in marginal activities for low income households are typically no higher than one-third of the average returns in the privilege rice labor market 1/ This neo-classical model is due to Mr. Trent Bertrand. Memorandum to C. Lluch and D. Mazumdar dated June 3, 1981. - 104 - (c) Seasonal fluctuations in the activities of the rural sector do not take the form of substantial variations in unemployment, but in the proportions of different kinds of market activity, particularly work in the rice sector and in marginal activities. (d) Total labor supplied to the market by a household is inversely related to the household income level. Low income households typically contribute more hours by women and children to the labor market, particularly in the slack season. 4. Apart from these basic stylized facts, one other feature might be mentioned because it has been discussed in the text, and it follows as a deduction from the above. will be recalled that, in the study by Gillian Hart, return to wage labor - was significantly associated with the household asset size, particularly in the slack season. Lower income households, who typically supply more labor to the market, have to devote a greater proportion of the supply to marginal activities in the slack season, and thus end up with a significantly lower return. A Neo-classical Model 5. A model with equilibrium market clearing wages in both the rice labor market and the marginal activities - but with a positive differential between the two wage levels - can be constructed if we introduce the following premises: (a) the relatively high wage rice sector involves harder work than the marginal activities, so that the supply curve of labor to the former is, for all types of labor, at a higher level than the supply curve to the latter; and (b) the supply curves of labor for poorer (or landless) households are at a lower level in both activities than those for rich households (or households with some land). 6. The upper panel of Figure 3.1 graphs the equilibrium in the rice labor market (Type A) and the lower panel in marginal activities (Type B) Sr and S are the supply curve of labor to the Type A labor market for rich and poor household respectively. There are seasonal variations in the demand for Type A labor, so that we have DA and D'A , the demand curves in the two 1/ It is important to remember that Hart considers only returns to wage employment. Thus her set of marginal activities leave out the large sector of self employment. In her terms all wage labor markets outside the privileged village labor market is the marginal sector. - 105 - Figure 3.1 A Neo-Classical Mcdel of the Rural Labor Markets REPRESENTATIVE WORKER TYPE A LABOR AGGREGATE LABOR MARKETS Woae wo9e~ Sr S Wage WA 4 -- -- ---WA 3 DAI WA…I~~~~~~~W WVA - - - - - - - - T -- - 2 DA I I' I I - I II I II ltL I I I 1 I i 2 3 4 5 6 7 8 9 10 11 12 13 HFourso Work Wok inme TYPE B LABOR Wage Wage 5 4 ~~~Sr ~s W 820- m <__~~~~S __W SP s o 2 8 f / // --w~ 1 2 3 4 56b7 8 9 1011 12 13Hoursot WaIk W;oiI lme Wokld Bonlc-24891 - 106 - seasons confronting the aggregate labor supply curve in rice S. Thus wage rates in Type A labor vary substantially between the seasons, and affect the supply curves of Type B labor in the two seasons. The supply curves for Type B labor, consistent with the peak season wage rate for Type A labor, are shown by the heavy line supply curves, consistent with the slack season wage in the top panels, given by the demand curves in the lower panels. 7. Equilibrium wages are WA (W'A) and WB (W'B) (slack) seasons in the two types of activities respectively and are the same for labor from either type of household. The difference between Type A and Type B wage in this model reflects the difference in the marginal disutility of labor in the two activities; or alternatively, the transaction costs involved in transforming Type B labor into Type A labor. Type B activity is less arduous, it is said, partly because it involves working "around the house", or at more convenient hours. There are constraints on transforming Type B labor into Type A labor imposed by the amounts of energy already expanded on work during the day and given by the limited number of daylight working hours. Clearly these con traints will be more operative in the busy than in the slack season. _ 8. The other premise of this model - the different levels of the supply curves for rich and poor households - explains the larger labor input over both seasons from poor (or landless) families. It also implies that the poor households allocate a larger proportion of their labor time in the slack season to Type B activities, and hence the average returns to their labor (taking Type A and Type B together) is less than the average wage obtained by the rich households in the slack season. (This confirms Hart's result on the significance of the asset variable). The Segmented Labor Market Model 9. The segmented labor market model suggested in this report does not dispute the second of the two premises of the neo-classical model. The income effect does result in a lower supply curve of labor for poor households. But the wage difference between the productive and the marginal activities in the segmented model is not due to a difference in the disutility prices of the two types of labor - since this model does not accept premise (a) (See above para. 5) of the neo-classical model. Wages in the rice sector are maintained at a relatively high level due to other factors which will be more fully discussed in section 2. A limited amount of employment is available in the rice sector at this wage, so that jobs are rationed according to some hiring rules. Both poor and rich households supply a substantial amount of labor to the marginal activities where earnings are driven down to low levels, particularly for poor households. (The mechanics of this model are discussed in section 2 so as not to interrupt the flow of the argument.) The immediate task is to see if the two models can be tested in a simple way - first with reference to their 1/ This point about seasonal difference in constraints is important and will be referred to again later on. - 107 - predictions, and second with respect to the premise of the neo-classical model which the alternative model rejects. Testing the Alternative Models: (i) Predictions 10. The four stylized facts mentioned earlier are consistent with either model. This is, of course, partly because one of the two premises - the income effect on the supply curve of labor from rich and poor households - is common to both models. The difference in wages between the productive (rice) and the marginal sectors is explained by alternative assumptions in the two models. Prediction (A) 11. There is, however, one prediction about the nature of preferential hiring in the rice sector which is different in the two models and which can be tested against observed facts. 12. The point can be best understood if the supply curves of the Panel A of Figure 1 are redrawn to show that there are minimum reservation prices for both rich and poor households below which no labor supplied' 1/ More people will agree that the supply curves should look like those shown in Figure 2 (with the rich having a higher reservation price). igure 3.2 The Rice Labor Ma rket W\th Two Classes of Workers Sr~~~~~~~~~~~~~~~~~~~~~~~ DSlock 0 ~~~~~~~~~~~~~x 1/ The argument does not depend on this point. Even if the two supply curves start from the same point (e.g. the origin) the result will hold as long as SR lies above Sp . - 108 - The Rice Labor Market with Two Classes of Workers 13. When labor from both types of households is aggregated, the summed supply curve will be affected by the weights attached to the two types of labor. Since there are more poor households, the aggregate supply curve S will be more elastic than Sp. This means that if there is no preferential hiring of workers from the rich households, no rich worker will be hired at all in the rice sector until the demand for labor reaches a point where a repesentative worker supplies OX number of hours. The message of the neo- classical model, in other words, is that the rich households will typically supply the marginal amount of labor demanded in the busy season, and probably nothing at all in the slack season. The prediction of the segmated labor market model with preferential hiring is exactly the opposite. When labor demand is small in the slack season, apparently the limited number of jobs in the high wage rice sector is retained by richer members of the community. 14. Field workers' evidence tend to support the predictions of the latter model. In her village study of the slack period of the dry season, Hart notes: "Men, women and children from landless households entered into long duration sugarcane and tebasan harvesting contracts (outside the village) in August and September/October respectively. The strategy adopted by small landowning households was quite different. By virtue of their relationships with large landowners, Class II (landed) men and women had preferential access to the more restricted and remunerative types of wage labor within the village'. 1/ 15. Stoler's study, 2/ throws light on the participation of different groups of female workers in harvesting (an intensely seasonal activity), and of the preferential treatment accorded to labor from landed (rich) households. Small farmers, in addition to employing household members, typically invite females from neighboring households to form part of the team of harvesters. "By allowing members of a neighboring household to participate and paying them in kind with a relatively large share, a farmer is insured that female members of his or her household will be given reciprocal employment opportunities. Such arrangements are as much a means of spreading the riskI/of cultivation as they are a means of meeting labor require- ments." - Large farmers employ harvesters from outside this network of kin and close neighbors. "As many as 50 to 150 women are used to reap the padi on fields of half a hectare and above. A large proportion are orang bin (literally "other people") from outside the village w4h9 seek out these larger farmers where their harvesting chances are greater." _ 1/ Gillian Hart, p. 168. 2/ Ann Stoler (1977). 3/ Ibid., p. 682. 4/ Ibid., p. 683. - 109 - 16. Stoler also discusses different rates of wages paid to different classes of workers in harveting. According to her study the size of the bawan (the share of the harvested crop paid out) typically depends on the social proximity of the harvester to the host household. "Close relatives receive from one-fourth to one-half of what they harvest ... shares of one-sixth to one-eight are given to women from neighboring households (tetangga), defined not only by their physical proximity to the host household, but by their mutual participation in gotong-royang activities. Shares of one-tenth to one- twelfth are given to orang-bin, that is distant villagers who fall outside the first two categories. In other villages the bawan for this group may be as little as one-twentieth." 17. Harvest labor from landless families, who operate outside the social network system, thus get a significantly lower harvest wage than lando yping families, although they might get more harvesting days per household. - Wages are not equalized for harvesting labor from different economic groups, as in a perfectly competitive market. On the other hand, if we think in terms of a model of "competitive monopsony" with each farmer facing an upward sloping supply curve of labor to himself, we would expect the larger landowners to pay higher wage since their larger demand requires that they attract labor from landed families with higher marginal supply prices. But the outcome of the social network system is the opposite of this prediction. It is the larger landowners who pay an effectively lower harvest wage, as they employ a larger population of harvest labor from landless families. 18. In sum, in the segmented labor market model, labor supplied by the poor households constitutes the marginal supply of labor to the rice sector, whereas in the competitive model, analyzed in Figure 1, it is the labor supplied by the rich households which is the marginal labor supply. Consequently, the preferential treatment of rich households in the rice labor market is more operative in the slack than in the busy season. Prediction (B) 19. A second prediction of the neo-classical model as set out in Figure 1 is that the wage rate in the village rice sector should be much nearer to the wage in marginal activities in the slack season than in the busy season. As already mentioned, the transaction costs of transforming Type B labor to Type A labor are substantially smaller in the slack season. In the segmented labor market model, there is no clear prediction about the seasonal behavior of the wage differential, but the probalility is that the wage differential will widen in the slack season - the opposite of the prediction of the neo- 1/ In Stoler's study 15 landless household had 43.8 harvesting days per household with 2.5 Kg of rice per harvesting day. By contrast 39 households with some land (< 0.2) contributed 25.3 harvesting days per household with a payment of 3.1 Kg per day. (See ibid., p. 685). - 110 - classical model. This is because if there is wage maintenance in the rice sector to ensure a minimum efficiency wage cost (see section II below), it is likely to be more operative when earnings in marginal activities are falling. 20. We are unable to test for this difference in predictions because comparable data on hourly wage earnings in the two types of activities, controlling for the demographic and the economic status variables, are not available by season in the material from the village studies which have been analyzed. But such data do exist, and it is recommended that they be analyzed to throw light on this issue, as well as on the issue of appropriate shadow wages. Testing the Alternative Models: (ii) Premises 21. The neo-classical model makes definite assumptions about the nature of the tasks involved in Type B jobs. The disutility of effort involved in these tasks is substantially less than in Type A jobs, so that the supply curve of labor to the Type B market is lower for all volumes of work supplied and for either income class. The model of Figure 1, therefore, conceives Type B jobs to be self-employment in or near the home, or outside regular "daylight" hours, largely in petty trade or handicrafts. 22. The distinction between the "productive" rice sector and marginal activities in the segmented labor market model does not essentially follow this line. It would appear from the discussion in the last section that the social network system establishes rice production within the village economy as the primary labor market to which some groups have preferential access. Marginal activities thus extend beyond petty trade in the village, or handicrafts in the workers' household. It includes some agricultural work outside the social network system of the village economy, e.g., harvesting for cash wages (the tebasan system). It also includes wage labor in non-rice sectors, e.g. sugarcane, and tobacco, fishpond labor and other non- agricultural work. In the slack season men, women and children from low- income (i.e. landless) families participate in these forms of wage labor outside the village, since the limited amount of work in the rice sector of the village is preferentially given to workers from landed families. Wage rates in these activities outside the village are lower than in the village rice market, and the average returns per hour of labor are further ryduced by having to spend two or three hours a day walking to and from work. 1 23. It would be hard to describe such work in the "marginal" sector, which involves longer hours of toil and search, as involving less disutility of labor, so that it could be represented by a supply curve lower than that in the village rice sector (Type A work of Figure 3.1). 1/ Gillian Hart, p. 166. - 1ll - 24. Some work in petty trade and handicrafts on the other hand, does give the impression of slow work, involving longer hours at a low return to labor per unit of time. But to conclude, as the neo-classical model would do, that the participants in these activities are doing so by choice rather than necessity, is to prejudge the question about the existence of involuntary underemployment in Javanese villages. Taken in conjunction with the elastic demand curve for labor in Type B jobs (of Figure 3.1), it does imply the doubtful proposition that poor workers are poor through choice. Notes on the Segmeneted Labor Market Model 25. The basic point of the segmented labor market model in Javanese rural areas is the maintenance of higher wages in the village rice labor market than in marginal activities. The text, as well as this Appendix, has suggested that institutional factors were significant in determining the rewards to labor. For example, the share in harvesting, tied to the social network system, has a strong content of customary value. But there is no uniformity of wages across villages for the same agricultural operation even within a limited region. 1/ While no systematic work has been done in Indonesia on the determinants of inter-village wage variation, one presumes that economic conditions in the village do influence the agricultural wage. One would also like to reconcile the maintenance of the wage in the rice sector at a level higher than marginal earnings with the profit maximizing behavior of small farmers. While a complete theory of rural wages cannot be provided here, it may be interesting to suggest a framework for conceptualizing the determination of wages in the rice sector, and of the existence of wage differentials with respect to marginal activities: (a) the use of a significant amount of fixed capital - farm land - in the rice sector determines a minimum level of labor productivity in this sector which is higher than productivity in marginal activities; and (b) the wage efficiency relationship which holds for individual workers ensures that there is a wage level which determines the minimum wage cost of a unit of labor to the employer and the productive sector, and it will not pay an employer to offer a wage below this level. 26. The determinants of earnings, at least in the large27elf employment marginal sector are well described in the following passage. - 27. "When there is persistent unemployment in a stagnant economy the redundant workers may take to employing themselves with tiny quantities of capital (say as shoe-blacks and pedlars) or by selling their services directly to consumers (as domestic servants, porters, odd-jobmen etc.). This kind of occupation is usually described as disguised unemployment. From a formal point of view it may be regarded as an extreme form of demechanization of technique. The possibility of making a living in this way sets a bottom to 1/ See White's data on wages in six villages in West Java, p. 47, above. 2/ Joan Robinson, "The Accumulation of Capital," pp. 157-158. - 112 - the fall in real wages and so to the level of technique in regular capitalist industry". 28. The productivity sector in the rural economy is differentiated from the marginal sector by its having a "higher" level of technique because of the availability of substantially more fixed assets (farm land). The productivity of labor is higher, and there is scope for labor to claim a share of this output to attain a higher real wage than in marginal activities. Why do owners of productive land concede to profit sharing and the claims of labor? The answer most likely lies in the incentive effect and the wage efficiency relationship which has been discussed in the literature for a long time. Starting from the low level of real wage in marginal activities, the efficiency of a worker increases more than proportionately as the wage offered increases, up to a point of inflexion after which efficiency increases less than proportionately. Thus, there is a wage level - higher than the real wage in marginal activities at which wage costs per unit of labor (in efficiency units) is minimized from the point of view of employers. No profit maximizing employer will offer a wage lower than this, in the productive sector. 29. The hypothesis is portrayed in Figure 3.3. The left-hand side of the diagram shows the increase in efficiency units supplied by a representative worker corresponding to varying levels of wage per man. To the left of the point of inflexion X, the elasticity of the function is greater than 1 showing a more than proportionate increase in efficiency with increased wages; and to the right of the point X, the elasticity is less than 1. Xgure 3.3 The Wage Efficiency Relationship Wage Cost Par Woage Per Effcienc, Mzndoy Unit .~~~~~~~ x x E / Number of Nunlm cif / /~~~~~i Efficiency EM6knc Units Units Vdd Bw*-24889 - 113 - Thus, this efficiency wage function yields a U-shaped cost curve, shown in the right hand panel, with wage cost per efficiency unit falling up to X' (corresponding to the wage X), and rising for higher values of wage per man. Let the earnings per man day in marginal activities be E, less than the critical wage X. Then workers will be available to the productive sector in unlimited quantity at the wage X, and no employer will offer a wage below X. The wage X sets a floor at which wage costs per efficiency unit of labor are minimized and profit maximizing employers are able to meet their total demand for labor (in efficiency units) by hiring as many man-days as required at this minimum average wage cost. Thus, in spite of an abundant potential supply of labor in the marginal sector at the wage X, a positive wage differential is maintained in favor of the productive sector. The wage in the latter will not increase until enough labor has been absorbed in the productive sector to cause earnings in the marginal sector to increase to X. 30. This wage efficiency relationship has a superficial similarity with the differentiation between Type A and Type B labor of the model of Figure 1. But there is a crucial difference. In the neo-classical analysis the causation runs from efficiency to wages, whereas in this view the causation runs the other way. At a higher wage (up to the point of inflexion) more or less anyone who is offered a higher paying job performs at proportionately higher efficiency. Thus, at the wage at which wage costs per unit of effort are minimized there is an abundant supply of labor. 31. Note the following points about the wage-efficiency theory: (a) Since there is an abundant supply of labor at the floor wage X, employers must (select some workers more favorably than other.) The preferential hiring rules discussed by the field workers strengthen the social network system without adding extra economic cost to the hiring. (b) The wage-efficiency relationship should be interpreted in terms of incentive effects rather than in nutritional terms as has sometimes been done in the literature. The nutritional interpretation predicts uniformity of wages across villages (and across seasons) which is contradicted by the evidence. The incentive effect ties the least cost wage to the real wage in marginal activities, and thus allows for the economic conditions of the particular village economy to influence its level. (c) The efficiency wage, we would expect, would be tied to a "norm" for the village economy, and would have considerable stability in real terms. It would change when conditions change more than marginally - in the upward (as after the turning point in Japan, and Korea) or downward (Bangladesh?) direction. Field - 114 - workers in Java and elsewhere have often commented on the "social" pressures underlying the maintenance of wage levels in the rice sector 1, 32. The idea of an efficiency wage as a "norm" for a particular village is consistent with the existence of the apparent paradox of widely differin real wage for the same operation in the same season in adjacent villages. -f The interdependence of the landowners and the workers seeking wage employment in a village ensures that the norm is established within the boundaries of the village. Professor Ashok Rudra writes on the results of3/a field survey of wage formation in a group of villages in Eastern India. _ "It is easy to anticipate the answers we would get if one were to pose the question, as we have done, in a survey as to why a laborer does not go to a neighboring village for higher earning and why a property owner does not employ a laborer from a neighboring village at a lower rate. The answers would be like the following: "Even if the employers in village give us lower wages, it is they who provide us with whatever employment we get round the year. Can we afford to do without that?" "If I should forsake the laborers of this village for saving a little money on wage what guarantee is there I should have labor when I need it most? It is these laborers who see us through the entire year." 33. Migrant labor from outside the village society is, in this view, a "foreign body" which is used sparingly to meet particular short periods of heavy labor demand. We have already quoted Ann Stoler's study of harvesting labor in her village when it was found that "outsiders" were typically hired only by large landowners and at a significantly lower share of the harvest than those part of the social network system in the village. 1/ See Maso Kikuchi (et al.) "Class Differentiation, Labor Employment and Income Distribution in a West Java village," The Developing Economies, March 1980. Ashok Rudra: "Local Power and Farm Level Decision Making, "mimeo. Viswavarati University, West Bengal. 2/ See, White's wage data, Figure 2.1, p. 47 above. 3/ Ashok Rudra, op. cit., p. 10. - 115 - The Transition from a Surplus Labor to a Neo-classical Model 34. The surplus labor model obtained above has two characteristics: (a) there is an abundant supply of labor at the "floor" rice wage W which minimizes labor cost per unit of effort. Employment at this wage is preferentially given to workers from certain types of households who are close to the property owning group in the villages; (b) the low-income (or landless) households have to meet their minimum income needs by devoting long hours to marginal activities. 35. While this mechanism drives down the wage in marginal activities, there is no "distortion" in the rice sector. The rice wage is at the equilibrium value consistent with minimizing the wage cost of an efficiency unit of labor. 36. The labor market will move towards the neo-classical model described in Figure 1 as more demand for labor is created in the rice sector. With sufficient labor demand in rice, the traditional upward sloping supply curve of labor will come to exist in both the rice labor market and in marginal activities. The conditions for this to happen are related but different in the two sectors. In the rice sector, the demand for labor has to be high enough to require that the number of efficient units supplied to the floor wage W is insufficient to meet the demand when all available workers have been employed at this wage. In marginal activities, the supply curve of labor will be the normal upward sloping type when an individual worker gets enough employment in rice to meet the minimum subsistence needs of his household. 37. Preferential hiring in the rice sector might slow down the transition to the neo-classical world. Reference might be made here to the "rationalization of labor" in the rice sector described in the text. As we saw, in some areas there has been an increase in labor-tying arrangements with economic development. As labor productivity increased after the green revolution, employers have sought to stabilize the effective wage of labor, not by interfering with the traditional share of hired labor in output, but by asking selected workers to perform other agricultural operations as a price for participating in the harvest. This development in the labor market increases the employment (in man-days) of a select body of workers at the cost of limiting the employment opportunities of other landless workers who are not preselected for such labor typing arrangements. In other words, the increase in demand for labor is not randomly distributed among the job seekers, thus limiting the affect on the pool of "surplus" labor. - 116 - Chapter 3 Appendix B Page 1 Shadow Wages in Java 1. To discuss the principles which determine the opportunity cost of rural labor in Java, we must find an answer to the question: what is the marginal product of a unit of unskilled labor which is withdrawn from agricultural occupations to work on a rural project? Shadow wage calculations will require additional judgements on the social cost of the wage bill of the project and on distribution weights. These topics are considered later in the Appendix. The Opportunity Cost of Labor 2. The opportunity cost of labor withdrawn from the rural sector is determined by the value (at social prices) of the loss of output caused by this withdrawal. The relevant loss of output, however, cannot be directly observed; its probable magnitude has to be deduced from the theory of the rural labor market which seems most acceptable. We will contrast the conclusions derived from the model of the rural labor market outlined in Chapter 3 with those derived from a purely competitive labor market model. (i) The Competitive Model 3. In the purely competitive model of the labor market, the loss in output due to the withdrawal of a unit of labor is measured by its marginal product which, in turn, is approximated by the wage rate for agricultural labor. A complication immediately arises in peasant agriculture because of the prevalence of small farmers who work the land themselves without the use of hired labor. But, in economies like Indonesia's, where landless labor, as well as small farmers, participate significantly in the market for wage labor, the conclusion from the competitive model would be that the marginal product of peasants on their own farms should be equated to the wage in the market for hired labor. 4. A second complication is that the loss of agricultural output due to the withdrawal of a unit of labor will be equal to its marginal product only if the elasticity of supply of effort by an individual worker is zero. Otherwise, some of the loss in labor time due to the withdrawal of unit of labor will be made up by extra effort supplied by those remaining in agriculture. The loss in output will, therefore, be less than the marginal product (= the wage) of labor. But this complication will not modify the calculation of shadow wage very much if we value the disutility cost of the extra effort which is being supplied. In equilibrium, the marginal supply price of effort is equated to the wage rate, and hence the value of the extra output produced by the workers remaining in agriculture will be more or less offset by the disutility cost of the extra effort. - 117 - Chapter 3 Appendix B Page 2 5. A third point is that the peak season wage of agricultural labor is so much higher than the slack season wage. The appropriate opportunity cost of labor depends upon the period for which it is withdrawn. If labor is withdrawn for the whole of the year, then the loss in output will equal the marginal product of a unit of labor for the agricultural year as a whole - taking the slack and the busy seasons together - which in turn, will equal the weighted average of the wage rates in the two seasons (the weights being the hours of work secured by a typical worker in the different seasons). (ii) Modification of the Competitive Model 6. The two points about the functioning of the rural labor market in Java (discussed in detail in Chapter 3 and its Appendix A) which require modification of the competitive model for the purposes of calculating the shadow wage are the following: (a) The existence in the rural labor market of activities pursued by laboring households in which the returns for an hour's work are much lower than in the rice labor market. Agricultural workers devote a significant proportion of their time to these "marginal" activities if they do not get enough work in the high wage rice labor market. Clearly more labor hours are devoted to these activities in the "slack" season, but they continue to provide an opportunity for extra income for some groups even in the "busy" season. (b) The importance of household economic status in the allocation of the limited amount of work in the high wage rice market between different socio-economic groups. Workers from higher economic groups are more likely to be employed in the high wage market than workers from landless families. 7. The implication of the last point for the evaluation of the shadow wage can best be understood by contrasting it with the situation which would exist if the rice labor market functioned in a fairly competitive manner. Workers from higher status families have a higher supply price (because of their superior asset position). Consequently, in a fairly competitive market, workers from such households would be employed in the rice market only after the available pool of labor from lower income groups has been exhausted. Thus, the wage in the rice market would in a competitive model, equal the relatively high marginal supply price of higher status workers. But, in a segmented labor market, such as described in Chapter 3, the high wage jobs are preferentially given to job seekers from the higher economic groups. A potential supply of workers with lower supply prices exist in the households of landless workers, but their opportunities for employment in the high wage market are limited. 8. The opportunity cost of labor in such a segmented labor market will not be given by the going wage of rice labor, but by the lower earnings in the "marginal" activities. For, if a unit of labor is withdrawn from the rice labor market, it will be replaced by a unit moving out of marginal - 118 - Chapter 3 Appendix B Page 3 activities. The loss in output in the rural sector as a whole would then be the loss due to the withdrawal of a unit of labor from the marginal activities - which is approximated by the earnings per hour of work in these activities. The Ratio of the Opportunity Cost of Labor to the Agricultural Wage 9. Let us now turn to the estimation of the ratio of labor earnings in marginal activities to the going agricultural wage. A numerical estimate for this ratio, however rough, is the first step in calculating the shadow wage. At the same time, it is more useful to work with this ratio than to try to estimate the absolute value of the opportunity cost of labor in rural Java. We can then apply this ratio to the most relevant going wage in the region, for the particular year (or years) of the project. 10. Makali and White have provided data on the distribution of hours worked and income derived from different activities for different classes of households in,7ix villages in West Java for the wet season 1975/76 and the dry season 1976. The households were divided by their land holdings into three groups: small (0-0.24 ha); medium (0.25-0.50 ha); and large (more than 0.50 ha). The relevant data for small households are as follows: Distribution of household Earnings for hours of working hours devoted work in the activity Activity to the activity (Rp per hour) Farm labor 25.15 49.92 Trader 30.94 17.72 Handicraft 19.59 9.71 Source: White and Makali (1979). 11. Trading and handicraft are the principal "marginal" activities for "small" agriculturists. They are also important activities for higher asset groups - though own cultivation occupies a significant proportion of the working time of the latter in addition to the three activities mentioned above. The "medium" and "large" cultivators have higher returns to labor in all three activities than the "small" cultivators, but their returns in trade and handicrafts increase relatively more than in farm labor. This is because the richer households are able to mobilize more complementary factors (particularly capital) in these activities. They also, on the average, have 1/ Benjamin White and Makali, "Wage Labor and Wage Relations in Javanese Agriculture",: unpublished, The Hague, May 1979, Table A6, p. 38. - 119 - Chapter 3 Appendix B Page 4 better access to the more highly paid jobs in agriculture, but their higher returns to labor on this account is of less importance. 12. It appears from the data given above that the returns to labor in "marginal" activities for the "small" households are between one-fifth and one-third of their average returns in farm labor. (Note that the amount of time spent in these "marginal" activities is fully one half of the total working time of these households and double the amount of time spent on farm labor). 13. Some economists might be tempted to calculate the opportunity cost of labor in the agricultural sector by concentrating exclusively on the lower returns secured by low income households in farm labor. From the data provided by White and Makali,l/ the marginal returns to farm labor can be calculated for the extra hours devoted to this activity by "small" cultivators as against "medium" cultivators. The marginal returns per hour of farm labor for the former, work out to be Rp 38.11, or rather more than 25% below the average wages secured by "small" cultivators. Estimating the opportunity cost of rural labor exclusively from the returns to farm labor is thus an overestimate. 14. A second field study which could throw light on the opportunity cost of rural labor in Java - and which has in fact, been already used by Gordon Hughes for this purpose - is by Gillian Hart. 2/ The study also refers to the agricultural year 1975-76 (the same period as the study referred to in the last paragraph), but is for a coastal village in which off-farm wage labor in fish ponds and non-agricultural activities outside the village seems to be more important than trading or handicrafts as non-rice activities. Hart's study provides figures for average returns to wage labor as a whole (including farm and off-farm labor). For the year as a whole, adult men earned a mean hourly wage of Rp 35.7, and adult women Rp 20.0, working an average of 156 hours and 100 hours per month respectively (using these hours as weights) the weighted average hourly wage comes to Rp 30.7 - which is as much as 30% lower than the average hourly returns to farm labor in the Rural Dynamics Study Villages (last paragraph). A part of the difference is due to regional variations in wage levels, but it is clear that the lower wage in Hart's villages is partly due to the inclusion of non-farm wage labor in the calculation along with farm labor. 15. As in the Rural Dynamics villages, returns to labor in Hart's study are given separately for three economic classes and also for peak and slack seasons. The difference in hourly wages for the medium (Class II) and small (Class III) households is considerable, especially in the slack season. The figures are Rp 32.7 and Rp 27.8 respectively in the peak season, and Rp 27.3 and Rp 21.8 in the slack season. It might seem appropriate to use the hourly wages of Class III households as the opportunity cost of rural labor in the two seasons, but in a segmented labor market, the marginal return to labor is 1/ White and Makali (1979), op.cit. _/ Gillian Hart, (1978), op.cit. - 120 - Chapter 3 Appendix B Page 5 the relevant concept - and this will be below the average return which is pulled up by the higher wages which some privileged job-seekers manage to get. 16. Gordon Hughes has tried to calculate the marginal return to wage labor separately for the peak and slack seasons in Hart's sample. For the slack season, he uses the average returns to female labor - which is Rp 15 per hour - since "similar work is available to men as well". For the peak season, Hughes calculated the "marginal return for the hours which the adult men in Class III household work in addition to the number worked in Class II households". This was calculated to be Rp 20 per hour and was lower than the marginal returns for a female worker for Class III household (Rp 23.5). Hughes accepted Rp 20 as the marginal return for wage labor for both sexes. 17. According to Hart's data, peak season wage labor accounts for about 40% of the yearly total. Thus, the year round weighted marginal return for wage labor is Rp 17. We do not have the data from Hart's village to know how much below the mean wage in rice labor this figure is. But compared to the returns to farm labor in the Rural Dynamics Village (according to White and Makali discussed earlier), this figure is about one-third of the farm wage. We also saw that this level was the upper limit of the returns to marginal activities in the White/Makali villages. 18. Benjamin White I/ had studied a poorer village in Yogjakarta at an earlier data. He collected information on earnings and hours of work in different activities "for early 1973", presumably the latter half of the wet season. The earnings per hour for male agricultural laborers (in hoeing) were Rp 9-11, and for women workers Rp 6-7 in transplanting, and Rp 16-20 in harvesting. In non-agricultural activities in the preparation of food items for sale, or in handicrafts - the returns per hour of labor were around Rp 3.5, and perhaps somewhat lower for women. These figures give the general picture that earnings per hour in "marginal" activities are one-third or less of what they are in farm labor. 19. The conclusion from these three studies is that it is probable that the marginal opportunity cost of rural labor is about one-third of the going agricultural wage for the year as a whole. The agricultural wage in this connection means an average wage paid for the main operations in paddy cultivation, suitably weighted by the proportion of man-days used in each operation. The Shadow Wage of Rural Labor 20. The estimation of the opportunity cost of rural labor discussed so far is only one building block in the calculation of the shadow wage. Gordon Hughes has recently provided an extensive discussion of the theory of the shadow wage and has also provided an estimate of the various parameters needed 1/ Benjamin White: "Population, Involution and Employment in Rural Java," Development and Change, Vol. 17, 1976, 267-290. Chapter 3 - 121 - Appendix B Page 6 for its estimation in the Javanese context. The theory is by now a fairly standard one, accepted by most economists in this field. The mission did not go into a detailed review of Hughes' estimates of the relevant parameters other than the opportunity cost of labor discussed here. We provide below a summarv of Hughes' theory. (1) The opportunity cost of labor defined as the earnings foregone measured at accounting prices (Vm) defines the traditional shadow wage. Rural labor employed in the project, however, may not be attracted at this price. For reasons given in Appendix A to Chapter 3, the wage in the organized rice sector is established at a level to minimize the cost of efficiency unit. It is likely that this floor wage will influence the views of both labor and the project authorities about the appropriate wage to offer to workers used in the public sector project. But to keep the argument prefectly general let Wp be the wage at which labor is typically attracted to the project at a level of efficiency appropriate to the organized sector of the rural market. Workers from rural households employed in these projects, therefore, experience a gain in income equal to (Wp - M), where Wp is the project wage and M is the output foregone of the worker. (2) This increase in consumption is valued at the accounting ratio based on the average rural expenditure pattern. Call this ratio Ver. Then the social value of the increase in consumption is (Wp - M) Ver. (3) But there is a social cost aspect to the increase in consumption. First, there is the loss in potential investment funds available to the economy. Secondly, depending on the relative economic position of the rural households which experience an increase in consumption, the distribution weight attached to the additional consumption might be less that unity. Both effects are taken into account by noting the income class of the households supplying the workers relative to the "critical income level" at which the government attaches zero social cost to increments in consumption. Let the income weight attached to the increase in consumption of households supplying labor be Wr* Then combining (1), (2) and (3): SWR VmM + (Ver Wr) (Wp M) 1/ 1/ It might help some readers to define the key parameters in the shadow wage equation: Vm: this accounting ratio is measured by the ratio of the shadow price of the basket of commodities produced by agricultural labor to its market price. Ver: measures the ratio of the shadow price of the basket of commodities consumed by the average rural workers to its market price. Vr: is the weight attached by society to the extra consumption occuring to the rural household due to the additional person working on the public project. It really reflects the society's (or government's) evaluation of the income level of the household which benefits relative to the critical rural income level at which the weight is unity - 122 - Chapter 3 Appendix B Page 7 The expression for the shadow wage as a proportion of the rice wage (W) is then: vr V + (V W) (W - M) W m W er r ) w We have discussed above the value of M- It will be about 0.3. W Hughes' values for the other parameters are V - 1.14, Ver = 1.01,W = 1.09 Hence SWR _ 34 - .08 (Wp - M) 21. It is seen that the value of the second term is small, because of the value of (Ver - Wr). Thus whatever the wage which the public sector project pays to unskilled rural labor, the shadow wage as a proportion of the rice wage is fairly well calculated by M1 22. It might be relevant here to draw attention to Hughes' comment on his derived income weight: "The income weight exceeds the accounting ratio for rural expenditure, so that the shadow wage rate at social prices is lower than that at traditional efficiency prices. This is relatively unusual and it emphasizes the very low income of households in this category relative to rural incomes in general."' The Value of M/W at the Peak Season 23. Because of the lack of data, the report is only able to give an approximate value of the ratio of M/W for the year as a whole. It is possible that the ratio has a lower value in the busy season. If labor is withdrawn from agriculture to work in the public sector project in the busy season the value of M/W in the busy season will be appropriate. The report recommends that an attempt be made to collect more information of the earnings of prime- age workers and their hours of employment in marginal activities in the busy periods of the agricultural cycle in Java. Such data could indeed be culled from existing village studies as a first step. A related and essential q.4estion is: how long is the busy season in the village economy? Further &ta on hours worked and on returns to labor in different activities by month or by season, are also necessary to answer questions on the working of the rural labor market raised in Chapter 3, and the Appendix A to this chapter. - 123 - CHAPTER 4 PROSPECTS FOR EMPLOYMENT, OUTPUT AND WAGES 4.01 The purpose of this chapter is to relate the expected increase in labor force in Indonesia to the anticipated growth in output; to inquire about prospects for real wage growth, in particular, for unskilled workers. The timing of unskilled wage growth is, of course, difficult to pinpoint. The approach taken here towards finding an answer consists of examining the historical record of the successful economies of Japan and Korea. In these (originally) rice economies, real wages in agriculture increased sharply after a period of structural transformation of output and employment of varying length and intensity. Along these growth paths, a point came when the rate of growth of real wages rose six to ten times the historically very low rate. It is useful to examine their employment and output structure at the beginning of their modern growth periods; the corresponding changes over a transition period with relatively constant real wages in agriculture; and the characteristics of their "turning point" in terms of the timing and magnitude of sharp real wage growth in agriculture. Then, a comparison is made with conditions in Indonesia where both the initial conditions and the characteristics of the transformation up to the present time are very different from the other cases. So much so, in fact, that if the process of modern growth is similar across rice economies, Indonesia may not reach a turning point along its present growth path. 4.02 Projected labor force growth in Indonesia over the next ten years is a relatively simple task, barring fundamental changes in the labor market scene. Most of the population which will be seeking work in the labor market in the next ten years has been b o7n. The population aged 10 and over is projected to grow at 2.62% p.a. - It will be recalled that the conclusion in Chapter 2 on participation rates was that, although the labor force surveys indicated participation rates a few percentage points higher than the censuses, the two alternative sets of data failed to reveal any significant trend. It is probably advisable to accept the growth rate of the population aged 10 and over as the best approximation to the likely growth of the labor force over the next decade. 4.03 On the basis of employment elasticities for the past decade obtained in Chapter 2, the anticipated increase in the labor force requires output growth to be around 7% p.a. to prevent a rise in unemployment. The latest World 2Bnk projections on output growth are well below this required rate. - Although the growth rate of GDP is expected to pick up from its expected low level during 1983, the projected growth rate for the later years of the decade is only 5.3%. These projected sectoral growth rates of 1/ World Bank (1982), Annex II, Table 5, p. 192. 2/ See World Bank (1983), p. 27. - 124 - employment corresponding to the World Bank's expectations of output growth are given in TabIe 4.1. It is seen that the overall growth of employment at 2.03% falls significantly below the expected labor force growth of 2.62%. This projection is on the basis of constant participation rates. Some World Bank work has suggested that there might be an upward trend in participation ra yes in the 1980s primarily due to the increase in market activity of females _ If this happens the employment situation will be worse than that projected in Table 4.1. 4.04 Another major point emerging from this rather limited numerical exercise is that, in spite of the rather high output growth projected for both agriculture and the secondary sector for the next decade, the tertiary sector will be the dominant sector for the absorption of labor, and the share of this sector in total employment will continue to increase. Table 4.1: PROJECTED EMPLOYMENT GROWTH BY SECTORS, 1982-1990 Annual growth Employment Projected rate of rate of output elasticity employment growth Agriculture 3.23 0.28 0.90 Secondary/a 9.73 0.40 3.90 Tertiary/b 5.13 0.75 3.90 Total 5.20 0.39 2.03 Notes: /a Manufacturing Transport and Public Utilities and Construction. /b Trade and Services. This prospect raises many questions about the role of this sector in the overall problem of employment. First, since entry into this sector is easy, the relevant question is not just the absorption of labor, but absorption at what level of income. Earlier chapters have pointed out that there is no firm knowledge about the relative levels or about trends of labor earnings in this sector. Do the anticipated growth rates in output (at 5.13% p.a) and employment at (3.9% p.a.) of Table 4.1 represent a state of affairs in which labor earnings will be growing pari passu with labor productivity in tertiary activities? The growth of tertiary output is itself conjectural. With better knowledge of wages and labor shares in this sector, it might be possible to deduce the growth of incomes (or output). Again, we recommend that high priority be given to such a study. 1/ World Bank (1982), Annex II, p. 186. - 125 - 4.05 A related question - demanding investigation of linkages between sectors - is how the growth rate in agriculture and in the secondary sector in the rural and the urban areas are related to the growth rate in the tertiary sector. Another important question is whether, in fact, the rate of growth of labor absorption given in Table 4.1 can be sustained, and if so, how much of it will take place in Java, and how much in the Outer Islands. This question is examined, in part below. Prospects for Real Wages and Employment 4.06 If the labor force grows at 2.6% per year and output at about 5.2% per year over the next decade, what are the likely changes in real wages and, in particular, the wages of the unskilled? Again, this question cannot be answered on the basis of current evidence about labor demand and supply schedules in Indonesia, and how they would shift over time as investment and output shift (both in amount and in composition). Such evidence is not available. All one can say is that, on the basis of the experience of output and employment growth in the last decade, the expected rate of growth of the Indonesian economy is not likely to lead to a tightening of the labor market in the 1980s. This conclusion is quite compatible with the increase in real wages in 1980/81. As discussed in Chapter 2 this upward pressure on wages might have been caused by a sharp increase in rice harvest well above its trend rate of growth, and as such, could well be a temporary event. 4.07 The lack of evidence on the process of wage formation in Indonesia calls for two tasks. The first is, as recommended in this report, the accumulation of wage data, over time and across regions, with particular emphasis on wages in rice at the seasonal peak, and the determination of whether comparable labor which, if not employed in rice cultivation has to alternatively work at lower wages somewhere else. 4.08 The second task is to examine the historical experience of other Asian rice economies who have undergone modern economic growth. In these countries, real wages in agriculture did not rise until the process was well advanced. An examination of that historical experience is a useful first step in assessing the Indonesian situation. 4.09 The focus on agricultural wages can be justified on two grounds. The first is positive: to study how the price of unskilled labor changes with economic growth. The second is normative: what are the welfare implications of unskilled real wage growth. At early stages of development, agriculture is the main activity for a large fraction of the labor force and the main source of livelihood for their dependents. Income per person may be well below the agricultural wage, because of the small number of days worked per year and high dependency ratios. In this situation, the use of the agricultural wage as a crude welfare index is justified. As development proceeds, the share of agricultural employment drops, as does its absolute amount (after some point). Income from agricultural activities becomes a smaller fraction of the total income of farm households. By the time the real wage in agriculture starts rising sharply, it has already lost importance as a welfare index for a substantial fraction of the population. - 126 - 4.10 Indonesia is at an early stage of this process. It is questionable whether its high output growth over the seventies has been associated with the changes in the structure of output and employment that characterized the early stages of growth of the successful rice economies of Japan, and Korea. Admittedly, the parallel is limited since they did not have oil revenues. But, nevertheless, the agricultural wage is an important welfare index in Indonesia. 4.11 In the next section, the overall growth records of Indonesia, Japan and Korea are examined, in terms of their broad output and employment characteristics. Such an overview is followed by brief surveys of the experiences in Japan and Korea regarding agricultural wages, with emphasis on their turning points. Then, wage and employment prospects in Indonesia are examined in the light of the experience in the other economies. (i) The Overall Growth Record in Indonesia, Japan and Korea 4.12 Table 4.2 gives the structure of output and employment and the corresponding growth rates in Indonesia, Japan and Korea. The last two have gone through the process of modern economic growth and passed the turning point (the period of a few years when the rate of growth of real wages in agriculture becomes six to ten times higher than the historically, very low rate). The turning point in Japan occurred during the fifties; and in Korea in the late sixties. Before this, these two economies went through a period of economic transformation. For Japan, this was a very long period (since the 1870s) of gradual growth in employment (from 0.6 to 1.5%), faster growth in output (from 2.3 to 5.4%), and great changes in the structure of both, always with an absolute decline in agricultural employment (except for a few years after WorId War II). For Korea, the period of transformation was much shorter, with very high rates of growth in output and employment (about twice those of Japan) and a constant agricultural labor force. Table 4.2: OUTPUT AND EMPLOYHMENT GROWTH BY SECTOR, INDONESIA, JAPAN AND KQREA Output (percent per year) Employment (percent per year) Agriculture Mining Secondary Tertiary Total Agriculture lining Secondary Teritary Total Indontesia 1961-1971 3.2 8.9 4.2 5.7 4.5 1.4 - 4.4 7.6 2.4 1971-1980 3.6 6.5 12.9 8.3 7.3 1.0 - 4.9 6.1 2.9 Japan 1887-1902 1.5 5.5 1.9 2.3 -0.1 2.6 1.9 0.7 1905-1917 4.0 7.1 5.1 5.2 -0.3 2.6 1.7 0.6 1917-1937 1.1 6.1 1.8 3.1 -0.3 2.2 2.0 0.8 1937-1962 2.1 -0.2 2.5 2.6 1.5 Korea I 1960-1970 4.4 8.1 17.6 7.7 8.4 -0.3 - 7.1 2.6 1.7 1970-1978 4.0 6.7 16.9 8.0 10.2 o.7 - 11.8 4.2 4.1 Notes and Sources: Indonesia, Chapter 2, Table 2.3 and Table B.4; Table 2.6. Yearly growth rates for output are trend rates. Japan, as in Tabe 4.2. Korea, as in Table 4.2. - 128 - 4.13 Japan and Java had very sp7ilar economic and demographic characteristics during the 1860s. - Japan at the 2urn of the century had characteristics similar to Korea during the 1960s. - The comparisons across economies are therefore justified, up to a point. On the basis of these comparisons, what can be said about the present situation of Indonesia, relative to its turning point? About 56% of Indonesian employment in 1980 was in agriculture. Agricultural employment since 1961 has been growing at between 1.0 and 1.5% per year - a rate much higher than the other economies during their transition. But the most important difference is in the growth of tertiary employment. In the other two economies, the ratio of the growth in the secondary and tertiary sectors was above unity, at times about 2 or 3. In Indonesia, it has always been below unity. 4.14 To interpret these historical differences, and to assess the prospects for wages and employment in Indonesia relative to its turning point, more information is needed about the experiences of the other economies. This is the purpose of the brief summaries that follow. The Indonesian experience will then be examined. (ii) The Period of Transition in Japan 4.15 Table 4.3 presents the sectoral allocation of incremental employment during the Japanese transition. About half was absorbed by tertiary activities, and slightly more than half by the secondary sector, whose contribution increased with greater reductions in agricultural employment. 4.16 During this long transition pe3iod, how did real wages change? Figure 4.1 shows the evolution of real _ wage in agriculture and nonagriculture for the period 1920-1965. The agricultural wage was constant 1/ For an elaboration on this point, see Geertz (1966). 2/ This point is emphasized by De Bever (1976), and used by him to justify the transformation of employment in terms of the number of employed persons into a measure of the flow of labor services in Japan, applying information from Korea. 3/ The money wage in each sector divided by the sectoral price deflator. - 129 - Table 4.3: MARGINAL EMPLOYMENT SHARES DURING THE JAPANESE TRANSITION 1887-1962 Total increase in the labor Percent into Period force (millions) Agriculture Secondary Tertiary 1887-1902 2.1 -13 65 48 1905-1917 1.9 -36 81 55 1917-1937 4.8 -21 66 55 1937-1962 14.0 - 6 54 52 Source: As in Table 4.2. from 1920-1950, with an increasing trend during the fifties and an acceleration of this trend around 1960. By 1920, the Japanese economy had been growing for fifty years. There is no estimate comparable to Figure 1 dating all the way back to 1870, however, there is an estimate of the agricultural wage for males dating to 1895, given in Figure 4.2. This gives the male daiJy wage at 1934-36 prices, using two different deflators. 1 There is a cycle in real wages over the fifteen year period 1915-1930, with an upsurge after the First World War. Wages maintained a relatively high plateau for about ten years, but then declined to the level of 1915. The long swing nature of this change, and the acceleration since the fifties, together with the data in Figure 4.1, suggest that the turning point took place well into the post World War II period. 4.17 The dating of the turning point in Japan, with its associated notion of the exhaustion of surplus labor, has been a matter of considerable debate. There are three main positions: (a) the turning point is a post World War II phenomenon; 2- (b) the turning point is a post World War I 1/ Figure 4.2 is given in Minami (1968), p. 386. For both deflators, 1934-36 = 1.00. 2/ Minami (1968), p. 381; Blumenthal (1980), p. 556; Ohkawa and Rosovsky (1973), p. 42, and Ch. V. - 130 - Flgure 4.1 Real Wages in Agriculture (WC) and Non Agriculture (Wn) (Minami (1967), p. 1974, (1934-36 =100)1 400 wn / 2 100 _w2 _ 9w - 111 i i - i 1920 1925 1930 1935 1940 1950 1955 1960 Year Figure 4.2 Reol Daily Wage for Males In Agriculture (MinamI (1968), p. 386, (1934-36 = 11 20- 10 - ~ ~ ~ ~ ~ ~ ~ 10 \ I 6 - p I I I fi I I , , , I 1890 1900 1910 1920 1930 1940 1950 1960 Yeor - Defcltod by Consumer Price Index Deficited by Aodrcultural Price Index Woud Bonk-24888 - 131 - phenomenon; - there was no turning point, since there never was any surplus labor. - Part of this debate reflects differences on how to define surplus labor and what are its implications for wage trends and other economic magnitudes in addition, part of the debate can be t aced to differences in the interpretation or measurement of facts. _ 4.18 The evidence suggests that, in Japan, it took a lmyg period (at least 75 years) of sustained growth in agricultural output - and of steady decline in the agricultural labor force, before the price of agricultural labor rose sharply. Between 1951 and 1963, real daily wages for male workers in agriculture grew at 5% per year, six tim s the rate of growth observed during the period 1894-1939, which was 0.74%. 5 During the sixties, with mechanizati ns, the agricultural real wage grew even faster, at about 7.2% per year. 62 The changes in labor flows in agricultural activities can be broadly described thanks to the recent contribution by Kaneda (1980); these are summarized in Table 4.4. The number of labor hours applied to a unit of land declined steadily, at a faster rate than the reduction of the agricultural labor force, since 1935. (Figures for 1935 refer to paddy cultivation only. All other years, refer to total agriculture.) 4.19 Thus, the demand for agricultural labor, derived from the demand for output, probably kept increasing, as shown in the increase in days per worker (although the figures before and after the War are not comparable). For a long time, though, such increases did not affect agricultural wages. This strongly suggests a flat supply curve at going wages. 4.20 Labor saving devices in agriculture (that would reduce the growth of labor demand, or make it negative) did not occur until after the turning point. Mechanization did not start until about 1960, after real wages had been growing for about ten years: "The growing shortage of farm labor in the sixties prompted mechanization of an increasing number of agri- cultural operations. The use of machines in harvesting, even in transplanting, has spread to all parts of Japan, 1/ Fei and Ranis (1964). 2/ Jorgenson (1966), pp. 59-60; Taira (1976), p. 431. 3/ See Minami (1968), p. 398-400, for a critique of the statement by Jorgenson [(1966) p. 54] on upward trends in agricultural labor income per person from 1878 to 1917. 4/ The arithmetic mean of growth rates in agricultural output, Table 6, is 2.2% per year. For total output, the rate is 4%. 5/ Minami (1968), p. 385. 6/ Hayami et. al. (1975), Table 2.10, p. 35. - 132 - particularly to rice farms. The process of mechanizing most field operations...(is) the prima facie evidence of a gradual shift to emphasis from the land-productivity growth to the labor-productivity growth in Japanese agriculture." (Kaneda (1980), p. 478). Table 4.4: LABOR USE IN JAPANESE AGRICULTURE Hours per Days per Tan Worker 1935 487 146 1953 379 181 1960 317 181 1970 275 226 Notes and Sources: Hours per tan (1 tan = 0.09917 hectares): 1935, Kaneda (1980), Table 1, p. 473, assuming that the labor input was provided only by paddy owner cultivators and tenant farmers; 1953 and 1960, Kaneda (1980), Table 4A, p. 478, computed as the mean of columns 1 and 2, using the weights in columns 5 and 6; 1970, Kaneda (1980), Table 4B, p. 478, computed as simple extrapolation over the labor input measures for 1965 and 1975, by farm size. Days per worker (1 day = 8 hours); 1935, computed from hours per tan, 13,750 workers and 3,290 hectares of paddy land from Hayami and Ruttan (1971), t.C.3, p. 340; 1953, computed from hours per tan, 15,639 workers and 5,931 hectares, from Hayami and Ruttan, ibid; 1960, computed from hours per tan, 13,390 workers and 6,071 hectares, from Hayami and Ruttan, ibid; 1970, computed from hours per tan, 9,296 workers and 6,071 hectares. The labor force figure is obtained by applying to 13,390 (the 1960 figure), the rate of decline in the labor force for the primary sector reported by Ohkawa and Rosovsky (1973), Basic Statistical Table 15, p. 311. 4.21 With mechanization, labor use in agriculture fell, and the rate of growth of the marginal product of labor must have been very high. The ratio of non-labor toWlabor inputs probably increased by 10 to 12% annua ly from 1960 to 1970. - With an output elasticity between 0.3 and 0.6, 2 1/ See Hayami et. al. (1975), t. 2.9 p. 33 2/ Estimated using Cobb-Douglas production functions. See Akino (1979),Table 4, p. 109, first line; Minami (1969), Table 1, p. 382, footnote. - 133 - this level of growt9 is quite consistent with a 7% growth in the marginal product of labor, 1 which was also the rate of increase of real wages. 4.22 In summary, the process of Japanese economic growth can be characterized by a long period of transition, at roughly constant real agricultural wages, during which the structure of output and employment changed substantially; agricultural development was based upon increases in the productivity of land. During this transition period, there was surplus labor, in the sense that demand/supply force7 did not determine unskilled wages, either in agriculture or elsewhere. 2 The transition ended somewhere during the fifties, when agriculture contributed about one-tenth of value added and absorbed about one-third of the labor force. Mechanization in agriculture and sharp increases in the productivity of labor occured at that point. By 1970, agriculture absorbed only one-fifth of the labor force and agricultural income was only one-third of VIe total income of farm households, down from 64% in 1953-55 and 50% in 1960. " Under these conditions, the agricultural wage had clearly ceased to be an adequate welfare index; and demand/supply forces in the labor market had become useful predictors of wage and employment trends. (iii) The Korean Record 4.23 The period of transition in Korea was short and intense. This and usual data deficiencies make it more difficult to date the Korean turning point. Available evidence suggests that it occurred in the late sixties. Between 1962 and 1968, the wage per farm employee (in 1970 US$) gre% by 0.74% per year. Between 1968 and 1973, the corresponding rate was 8.8%. - The series of monthly earnings of production workers in agriculture from 1959 to 1962 shows movement roughly in accordance with changes in rate of thc cost of living index in Seoul and the index of prices received by farmers. 5 4.24 There exist two sources of information about changes in the structure of employment during the sixties: the population censuses of 1960, 1966 and 1970; and the quarterly sample surveys carried out since 1963. These 1/ The rate of change in the marginal product of labor is equal to the rate of exogenous technical change plus the product of the output elasticity of labor and the rate of growth of the non-labor/labor input ratio, in the Cobb-Douglas case. See Annex A. 2/ The implication of surplus labor for the distribution of income outside agriculture have been examined by Minami and Ono (1980). 3/ For these figures, see Ohkawa and Rosovsky (1973), Basic Statistical Table 15, p. 311; and Kaneda (1980), Table 8, p. 484. 4/ See Hong (1976), Table 5.2, column 4, p. 37. 5/ See Economic Planning Board (1970), Table IX. 4, column 2, p. 86; Table X. 5, column 1, p. 93; Table X. 6, column 1, p. 94. - 134 - show different rates of growth in employment over the period 1966-70 - 6.3% per year from the census, 3.7% from the sample surveys. It is generally believed that the sample surveys underestimate. According to t7em, the open unemployment rate declined from 8.2% in 1963 to 4.1% in 1974. 1 4.25 For the purpose of this report (the comparison with the Indonesian experience), rough estimates of marginal employment shares by sector have been calculated; these are presented in Table 4.5. It is to be emphasized that: (a) both sources indicate an increase in the marginal employment share in agriculture after the turning point; (b) the ratio of the marginal employment shares in "other," relative to manufacture varies between 1 and 2.2. (iv) On the Extensive and Intensive Margins of Cultivation in Indonesia 4.26 The overall comparisons of Tables 4.5 and 4.6, together with the summaries of the growth experience in two Asian rice economies which have passed the turning point, bring into focus the main questions to be asked about wage and employment prospects in Indonesia: For how long can the growth in agricultural employment go on? Has Indonesia more scope now for labor absorption in agriculture than the other Asian rice economies during their transition periods? Some preliminary answers to these questions can be obtained by examining the intensive and extensive margins of cultivation in Indonesian agriculture. The remarks that follow are limited, on the whole, to Java and to paddy cultivation. Table 4.5: SECTORAL ALLOCATION OF ADDITIONAL EMPLOYMENT IN KOREA Additional employment Percent into (millions) Primary Manufacture Other A. Census 1960-66 0.93 -3 52 52 1966-70 2.18 28 23 50 B. Sample Surveys 1963-68 1.49 1 38 61 1968-73 1.99 36 30 34 Source: Hong (1976), Table 4.3, p. 27. The primary sector includes Agriculture, Fishery and Mining. "Other" includes services, transport, public utilities and construction. 1/ See Hong (1976), Table 4.2, column 9, p. 25, and p. 26. - 135 - 4.27 Table 4.6 contains the man/land ratio for Java (1971), and for the other economies at different points in their transition periods. Ignoring the incidence of multiple cropping, the low figure for Java contrasts sharply with the other economies. Arable land per worker was much lower in Java in 1971 than in Japan or Korea during the 1960s. Given this situation, with respect to the extensive margin, is Javanese agriculture significantly far from the point of intensive cultivation attained by Japan and Korea during their transition periods? 4.28 In Asian agriculture there is a strong positive coryelation between output and labor use, (expressed in units of land). - This holds within countries by farm size, and across countries, Japan and Korea had very high values for both variables, relative to India, / during the 1950s. What can be said about the values for Indonesia? Some comparisons Table 4.6: RATIO OF FARM LAND TO AGRICULTURAL EMPLOYMENT, INDONESIA, JAPAN AND KOREA Ha/worker Indonesia 1971 0.487 Java 1971 0.316 Japan 1878-1882 0.326 1918-1922 0.433 1960 0.430 Korea 1960 0.450 Sources: Indonesia, Employment in Agriculture from Apendix Tables IIB.3, IIB.4. Agricultural employment in Java is 0.608 of total employment in agriculture (see Population Census, Series C, Table 44, p. 222). Japan, Ohkawa, Johnston and Kaneda (1970), Table 1, p. 108, LTES estimates for 1878-1922. Korea, Korea, ibid. 1/ See Ishikawa (1967), chart 3-3, p. 230-231. After a point, the correlation turns negative, as in Japan during the fifties. 2/ The cross country relationship estimated by Ishikawa (1967), pp. 235 using the proportion of irrigated land, G/L as a shift variable was log (Q/L) = -.026 + 1.079 log (N/L) - - 0.531 (log N/L)2 + 0.0062 (G/L), R2 , 0.845, where (Q/L, N/L) are output and labor flow per hectare. - 136 - for paddy are ayailable, (see Table 4.7). Paddy yields in Java, at least in some regions, 1 comparable to yields in Korea (1960), and Japan (1950). They are definitely above yields in the Philippines and India. Correspondingly, labor input (in man-days per hectare) in rice cultivation is already high. (vi) Long Term Employment Prospects Under Present Trends 4.29 The lesson from the previous sections is that the turning point (a sharp increase in the agriculture wage) came late in the growth process of two Asian rice economies which we have examined. In each case, the turning point was preceded by significant structural transformations. Subsequently more changes occurred which caused the agricultural wage to cease to be a relevant indicator of welfare because of a shrinking agricultural labor force Table 4.7: PADDY YIELD PER HECTARE AND LABOR INPUT PER HECTARE IN ASIAN AGRICULTURE Paddy yield Labor input (metric ton/ha) (days per ha) Japan 1950 National 4.249 255.6 1962 National 5.798 190.0 Korea 1960 3.271 139 China East Central 1921-25 2.559 146 Philippines Central Luzon-Lagra 1966 2.2 60 Central Luzon-Lagra 1974 2.2 82 India Hoogly 1956-57 1.800 132.9 24-Pagamas 1956-7 1.541 103.4 Indonesia Sidoarjo, E. Java 1971 4.5 256 Subang, W. Java 1971 3.0 170 Source: Ishikawa (1978), Table 1, p. 4; p. 30 1/ Overall paddy yield for Indonesia has been computed as 3.6 tons/hectare World Bank (1979), Annex, Table 7.3, p. 199. - 137 - and a decline in the proportion represented by agricultural income in the total iVcome of farmers. Change after the turning point also takes a long time. 17, 4.30 Indonesia is still far from the turning point. The immediate question is whether the economy is moving toward or away from it, given the present trends in output and employment. A preliminary analysis of this question can be made using fixed input labor coefficients in agricultural activities, setting them at values that systematically err on the conservative side: too high on the side of demand and too low on the side of supply. 4.31 The experience of other Asian rice economies at their turning points provides a conservative definition of full employment labor input (man-days per hectare) in agricultural activity. Consider three workers per hectare, the situation in Indonesia around 1971, and let 200 be the number of man-days per worker in agriculture at full employment (i.e., around the turning point). To err conservatively, let the full employment figure be 180 man- days, i.e., 540 man-days per hectare. This hypothetical supply flow can be matched with a demand flow by specifying the typical use of a hectare and its associated la27or input coefficients. Such rough calculation has been made by J. Goldberg - for Java. Under his assumptions, again designed to establish upper bound to demand, the labor use per hectare comes to 450 man-days. 3/ 1/ In Korea, for example, 70% of total farm income still comes from agri- cultural activities, in 1977. See Korea, Ministry of Agriculture and Fisheries (1978), Table 2.1, p. 52-3. In Japan, it was pointed out above that such proportion was one third in the seventies. 2/ See J. Goldberg (AEPIA), World Bank memorandum, December 20, 1979. 3/ A hectare is used as follows: 0.40, sawah; 0.40, tegal; 0.20, tree crops All 0.80 hectares of field crops are assumed to have 200% cropping intensity. All tegal is doubled cropped with dry field crops; dry season sawah is half utilized for irrigated rice and half for rice field. Tree crops are worked all year long. The labor input coefficients are: Wet rice: 200 man-days/ha Dry field crop: 180 man-days/ha Tree crops: 120 man-days/ha The demand for labor services is then: Sawah: 0.4 x 1.5 x 250 = 150 0.4 x 0.5 x 180 = 36 Tegal: 0.4 x 2 x 180 = 144 Tree Crops: 0.2 x 1 x 120 = 120 Total 450 man-days. 4/ In the field surveys quoted above (p. 25) the man-days per hectare in paddy cultivation ranged between 170 and 256. - 138 - 4.32 Under these assumptions, demand falls short of "full employment" supply in agriculture by about 20%. Thus, in spite of the sustained output growth in agriculture during the last decade, the Javanese workers dependent on this sector are significantly away from anything approaching full employment in agricultural activities, even under the most conservative estimates. We have no data on the elasticity of hours (or man-days) of work demanded in agriculture with respect to the growth in output. So we do not know whether the man-days per agricultural worker have been increasing or decreasing. Assuming a figure of 0.39 for the elasticity of demand for labor measured in man-days, which seems reasonable for Indonesia, a 3.6% yearly rate of output growth implies a growth rate for labor demand (in man-days) of 1.4% per worker. But, as we have seen, growth of the labor force in agriculture was around 1% p.a. Thus, it seems that any tendency for the flow of man-days per agricultural worker to increase has been very modest at best. This is, of course, in keeping with the observed stagnation of real wages in agriculture. 4.33 If the growth prospects for output and employment continue the trends of the last decade, we would expect little movement towards the turning point of full employment in agriculture. But actually there may be no movement at all, for two reasons: (a) How long could a 3.6% rate of agricultural output growth be sustained? As we have indicated, rice cultivation in Java is probably close to the intensive margin, and the impact of high yielding varieties and the commercialization of agriculture will gradually be exhausted. (b) There has been much discussion in Indonesia about labor-saving innovations and mechanization which tend to reduce the elasticity of demand for labor in agriculture. These changes are partly a product of the commercialization of agriculture, and yartly induced by a policy of subsidizing the price of capital. - 4.34 All these points emphasize the importance of further systematic examination of the likely magnitudes of the elasticity of demand for labor in Indonesian agriculture, in the light of the changing structure of outputs and inputs in this sector. In the absence of more information, the tentative position of this report is that if the number of workers primarily dependent on agriculture continue to increase at its previous rate, there will not be any perceptible movement towards the turning point, with the current trends in output growth. 4.35 The ability to provide non-agricultural jobs to a larger proportion of the incremental labor force depends upon the successful development of non- agricultural sectors. In this context, this report has noted a peculiarity of Indonesian development in the last decade: the proportion of the additional labor force going into tertiary activities is much higher than that observed in the successful development process of Japan and Korea. This suggests that a sustained effort should be made to collect information about earning opportunities in the tertiary sector. 1/ See, on this point, Sinaga (1978), pp. 2-3 - 139 - ANNEX A Page 1 Long Run Employment Problems in Indonesia: Alternative Views, Implications and Tests 1. Table Al gives the rates of growth of output and employment in Indonesia during the seventies. It also gives the employment growth rates for Java separately, to be compared with output growth rates in Java when these become available. The labor force is expected to grow at about 2.6% per year during the eighties, that is, two million entrants per year. Against this background, the main questions about trends in wages and employment are: (a) how much longer can agricultural employment, particularly in Java, keep growing? (b) what are the trends in the structure of pay and how is labor used in activities outside agriculture? (c) when will wages in agriculture and, unskilled wages in general, start to rise? (d) what are the forces producing fast output growth with roughly constant wages for a large fraction of the labor force? 2. While no final solutions are at hand, groping towards an answer to these questions requires a study of alternative views of how markets, especially that for labor, function in Indonesia. A formalized presentation of alternative micro models of rural labor markets was given in Appendix A to Chapter 3. The purpose of this Annex is to set out and compare the various macroeconomic implications of these alternative views and to recommend field work which might test their validity to the Indonesian context. Table Al: YEARLY OUTPUT AND EMPLOYMENT GROWTH RATES, 1971-1980 Percent per year Output Employment Indonesia Indonesia Java Agriculture 3.6 1.0 0.5 Secondary 12.7 5.3 5.2 Tertiary 8.3 6.2 5.6 Mining 6.5 Total 7.3 2.9 2.6 Source: Chapter 2, Tables 2.3 and 2.6. - 140- ANNEX A Page 2 I. Full Employment 3. On the simplest economic theory of wage determination, Indonesia employers would hire labor until the wage is equal to the value of the marginal product; workers would offet labor until the wage is equal to the marginal disutility of effort; and, in the long run, the market mechanism would ensure that observed wages reflect both. A study of wage and employment trends in Indonesia would then review changes in the forces affecting labor supply and demand. 4. Employment status would be immaterial. With well integrated labor markets, everyone could become a wage laborer, and 36% of the labor force in 1977 chose to, while 45% preferred self-employment or unpaid family work. Also, wages might be very low, and indeed they are in marginal activities in rural Java. Wage differentials across workers may be high. But all this would result from workers' lack of skills or inadequate complementary nonlabor inputs, and from differences in skills among workers. In this view, the causation would run from efficiency and skill to wages. A worker making five times per hour what another does, would be viewed as five times more efficient, as having five times more "labor units" than the other. The only exception to this general rule would occur if wages were administered (as in plantations in the government sector); or if there were restrictions on the freedom to hire and fire workers (like those facing foreign firms). For the economy as a whole, these are in fact, probably unimportant. Trade unions or employers' associations would also represent "market distortions" but these are fairly unimportant in Indonesia at the present time. 5. This view of the Indonesian labor market can be represented formally !n aggregate terms. Such representation could be used to organize existing data and to determine new data which would be needed to answer the questions about wage and employment trends listed above. The most aggregative representation will be considered first, in order to bring out the main ideas more clearly, and to prepares the ground for the sectoral disaggregation that follows. 6. Consider the aggregate production function X = F(K, L, t) relating all non labor inputs, K, and all labor inputs, L, to aggregate output. The passage of time affects output in two ways, at a constant input use. First, through the rate of technical progress, a , defined at Ft/F. Second, through the bias of technical progress, a , defined as log (FL/FK)/t. If a <0, technical progress is biased against labor use: the marginal product of capital increases, relative to that of labor, at a constant input yue. With the aggregate production function written in intensive form, _ X = Lf(k, t), the rate and bias of technical progress are a ' f /f- ' (1-v) (c-f't/ft), 1/ Where f(k, t) = F(k, 1, t), and k = K/L, assuming constant returns to scale for the economy as a whole. ANNEX A - 141 - Page 3 where v(k,t) - kfl/f, the share of capital, and ft(k,t) - af/at, f (k,t) - / I aft' - af,/at. 7. The rate of growth in output, g, is linked to the rate of technical progress a, to the rate of employment growth X, and to the rate of change in the capital labor ratio, gk, by the expression g . X+a+vgk * (1) 8. The marginal product of labor is h(k, t), defined as h(k,t) - aX/aL - f(k,t) - kf' (k,t), and it changes over time at the rate of gh given by gh ct+v+(v/C)gk, (2) where a is the elasticity of substitution between the capital and labor aggregates. From (2), it can be seen that there is always a high enough saving bias, -$, such that gh - 0, in normal situations with gk