MACRO-ECONOMIC ANALYSIS OF OUTPUT, EMPLOYMEN'II I AND MIGRATION IN SIERRA LEONE ’ Dissertation for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY HABIB T. FATOO 1977 III III IIIIIIIIII IIII IIIIIII III III IIIII IIII IIIIIIII 32 328 This is to certify that the thesis entitled MACRO ECONOMIC ANALYSIS OF OUTPUT, EMPLOYMENT AND MIGRATION IN SIERRA LEONE presented by Habib T. Fatoo has been accepted towards fulfillment of the requirements for Ph.D. degree in Agricultural Economics ‘ r (766.64%,K Major professor Date May 20, 1977 0-7639 ABSTRACT MACRO-ECONOMIC ANALYSIS OF OUTPUT, EMPLOYMENT AND MIGRATION IN SIERRA LEONE By Habib T. Fatoo The objectives of this research were to develop an analytic framework to analyze the relationship between output, employment and migration at macro level and to apply this framework to Sierra Leone economy to examine the growth, migration and employment prospects of the economy as well as the implications for output, employment and migration of a number of alternative development strategies. The research was motivated by the lack of an existing framework to analyze comprehensively problems of output, employment and migration facing developing countries-~the Sierra Leone economy was selected because of both the availability of data and the fact that the economy has many features common to other developing countries. However the models are also of general applicability to other developing economies for short- and medium-run policy analysis (5 to lo years). The models are also useful for sector-specific policy analysis. This is because they can run simultaneously with detailed sector models and sector specific policies can be analyzed within a broader macro framework. In this study an improved framework is proposed which has a higher degree of disaggregation than existing models and takes explicitly into Habib T. Fatoo account interactions in both the product and factor markets. In con- trast to conventional classification, the product market is disaggre- gated into a number of interacting sectors on the basis of type of output (agriculture and nonagriculture), scale of operation (small-scale and large-scale) and location (rural and urban). Particular attention also has been given to the modelling of the labor market. Based on disaggregation of the product market into small-and large-scales the labor market is disaggregated into a small—scale sector where wages are competitively determined, and a large-scale sector where they are exogenously fixed. A further refinement is introduced into the labor market by disaggregating the labor force by educational levels to reflect different supply and demand conditions for different educational levels. Migration, specific by educational level between rural and urban areas, occurs in response to the differential between competitively determined rural wage rate and expected urban wage. The expected urban wage is defined as the weighted sum of the wage rates in small- and large-scale sectors in urban areas, the weights being the probabilities of finding an urban job in each sector. This emphasis on intra- and inter-sectoral and regional relationships in both the product and factor markets as they affect output, employment and migration adds strength to the model results. The models were run using aggregated information from comprehen- sive primary data generated by field surveys, unlike most macro models which depend largely on secondary data. The model results indicate that despite the favorable rate of growth of GDP and a slight decline in migration, there is no relief from unemployment if current policies are Habib T. Fatoo continued. This underscores the importance of development strategies which increase employment. In the policy runs the impact of various develOpment strategies were examined with emphasis on how they would affect output, employment and migration. The impact of strategies exa- mined can be broadly classified into three groups. In the first group, the impact of various agricultural development strategies such as agricultural export promotion and an increase in agricultural producti- vity were examined. In the second group, the impact of various strate- gies to promote labor-intensive nonagricultural sectors such as small- scale industry promotion and a switch to labor-intensive techniques of production in large-scale industry were examined. Lastly, the impact of relaxing the foreign exchange constraint by increasing foreign capital inflow was examined. An important finding of this study is that at the macro level there is no trade-off between increased output and employment. This is largely because (a) on the demand side, the consumption demand by the rural population have high income elasticities for labor-intensive products. This consumption demand linkage is important because consump- tion is the largest component of total demand and rural consumers account for a very high proportion of total consumption, and (b) on the supply side, the more labor-intensive sectors are also efficient users of scarce capital and foreign exchange. There is, thus a great potential for designing development strategies which can stimulate both growth and employment. MACRO-ECONOMIC ANALYSIS OF OUTPUT, EMPLOYMENT AND MIGRATION IN SIERRA LEONE By 2‘“ Habib T? I‘Fatoo A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics 1977 ACKNOWLEDGEMENTS I wish to express my sincere appreciation to my thesis supervisor, Dr. Derek Byerlee, for his patience, continued personal interest and helpful suggestions during this research effort. His critical comments and constructive suggestions throughout the period of research were invaluable. I also wish to thank Dr. Carl Eicher, my major professor, for his valuable assistance and direction during my graduate program. Special appreciation is also expressed to other members of my guidance and thesis committee, Dr. Lester Mandersheid for his construc- tive comments and editorial assistance, Dr. Carl Leidholm for his cooper- ation in supplying me with useful information and data related to small- scale industry sector in Sierra Leone,and Dr. Vernon Sorenson and Dr. James Zuiches for their assistance and counselling during my graduate program. I am also indebted to the Department of Agricultural Economics for providing me with an assistantship to finance my graduate program and to the African Rural Economy Project which, through its U.S.lAID Research contract, for financial assistance to carry out the research. Appreciation is also extended to Ms. Linda Buttel, Pamela Marvel and Joanne Berry for providing valuable computer asistance. Thanks are also due to Lucy Wells and Janet Munn for secretarial assistance and their patience in typing the early draft of the thesis. I am deeply grateful to Dr. William Judy and Mr. Frederick Holmes, both of U.S./AID, Dr. Rupert Seals, Dean of the College of Agriculture, ii FAMU and Dr. James Stallings of Auburn University for their initial encouragement I needed to pursue graduate study. Finally, the patience and continual enthusiasm expressed by my parents, sisters and brother provided me with the support I needed to continue my graduate program. Their sacrifices are deeply appreciated. TABLE OF CONTENTS LIST OF TABLES .......................... LIST OF FIGURES ......................... CHAPTER I II III INTRODUCTION ........................ Problem Setting ...................... Objectives of the Research ................. Thesis Outline ....................... METHODOLOGY FOR MACRO ANALYSIS OF OUTPUT, EMPLOYMENT AND MIGRATION ........... , .............. Treatment of Labor and Migration in Theoretical Models . . . Role of Labor and Migration in Economic Growth ..... Integration of Growth and Migration Models ....... Treatment of Labor and Migration in Applied Policy Models Synthesis of Empirical Evidence on Behavior of Rural and Urban Labor Markets and Migration in Africa ........ The Rural Labor Market ................. The Urban Labor Market ................. The Structure of Urban Employment .......... Earnings in the Urban Labor Market .......... Rural-Urban Migration ................ Urban Unemployment .................. Policy Implications ..... ‘ .............. Towards a Framework for Macro Analysis of Output, Employment and Migration ....................... Product Market Disaggregation .............. Factor Market Disaggregation .............. Macro Level Interactions ................ OVERVIEW OF OUTPUT, EMPLOYMENT AND MIGRATION IN SIERRA LEONE National Accounts ..................... Sectoral Performance .................... Population .. ........................ Labor.Force -« ; ...................... Employment ......................... Unemployment ........................ Summary and Policy Issues ................. iv Page vii ix CHAPTER Page IV ECONOMETRIC ANALYSIS OF RURAL-URBAN MIGRATION RATES . . . . 49 Review of Econometric Studies ............... 49 The Migration Function ................... 54 Data ............................ 57 Migration Rates ..................... 58 Rural and Urban Wage Rates ............... 59 Urban Unemployment Rates ................ 59 Estimation Procedures and Empirical Results ........ 61 Implications of the Analysis ................ 67 V MACRO MODELS FOR ANALYSIS OF OUTPUT, EMPLOYMENT AND MIGRAé . TION IN SIERRA LEONE .................... 69 Macro-Economic Model .................... 69 Overview of the Model .................. 70 Model Mechanism ..................... 70 Model Results ...................... 71 Sectoral Disaggregation .................. 72 Structure of the Model ................... 75 Notation ........................ 75 Commodity Balance Equations ............... 78 Inter-sectoral Demand .................. 78 Consumption ....................... 79 Investment ....................... 83 Import Balances ..................... 85 Exports ......................... 87 Foreign Exchange Constraint ............... 89 Labor Force and Employment ............... 89 National and Sectoral Accounting ............ 93 Objective Function ................... 97 Migration Model ...................... 98 Migration Function ................... 98 Expected Urban Wage ................... 99 Probabilities of Finding Employment ........... lOO Wage Rates Determination ................ 101 Model Linkages ....................... 102 Summary and Model Limitations ............... 104 VI MODEL SOLUTIONS ...................... 108 Base Run Projection, 1974-1981 ............... 108 National Account Statistics ............... 109 Sectoral Level Results ................. 112 Value Added by Sectors ................ 112 Employment by Sectors ................ 115 Urban Employment, Migration and Unemployment ...... 119 Employment ...................... ll9 Migration ...................... 123 Unemployment ..................... 125 Policy Runs ........................ 127 CHAPTER Page Results of Agricultural Development Strategies ..... 128 Agricultural Exports Promotion Policy ........ 128 Increased Agricultural Productivity ......... 132 Results of StrategieS- to Promote Labor-Intensive Nonag- ricultural Sectors ........... , ........ 135 Small-Scale Industry Promotion Policy ........ 135 Switch to Labor-Intensive Techniques of Production in Large-Scale Industry ................. 138 Results of Increased Foreign Capital Inflow ....... 139 Discussion of the Results ................. 142 VII SUMMARY AND CONCLUSIONS .................. 145 Summary of the Model Framework ............... 145 Conclusions on the Effects of Various Development Strategies on Output, Employment and Migration ............ 147 Some Policy Guidelines for Increasing Output and Employment 150 Agricultural Development ................ 150 Small-Scale Industry in Rural Areas ........... 151 Large-Scale Sectors ................... 152 Small-Scale Industry in Urban Areas ........... 152 Suggestions for Future Research, .............. 153 BIBLIOGRAPHY ........................... 155 vi LIST OF TABLES TABLE 1 0'1 GOODNOS TO 11 12 13 T4 15 16 17 18 Population Distribution Between Rural and Urban Areas in 1963 and 1974 ........................... Potential Labor Force as a Pr0portion of Population in Each Location, 1963 and 1974 ................... Labor Force Participation Rates by Age and Sex for Urban Popu- lation, 1974 ......................... Labor Force Distribution in 1974 by Location and Sex ..... Employment Increase in Large-Scale Sectors Compared to Total Increase in Employment, 1962-72 ............... Wage Employment in Large-Scale Sectors ............ Unemployment in Sierra Leone ................. Registered Unemployed by Years, 1962-1973 .......... Average Annual Gross Rates of Adult Out-Migration from Rural to Urban Areas by Educational Level Rural Wage Rates by Region .................. Urban Wage Rates by Educational Level ............ Rates of Urban Unemployment by Education ........... Gross Rural to Urban Migration of Adults in Sierra Leone: Ordinary Linear Function ................... Variables and Parameters . . J ................ Input-Output Coefficients of the Sierra Leone Economy, 1974 Average Expenditure Elasticities by Population Groups for Domestic and Imported Commodities, 1974 ........... Capital-Output Ratios by Sector, 1974 ............ Coefficients of Intersectoral Capital Flows-the B Matrix, 1974 vii Page 39 4O 41 42 43 44 45 46 6O 61 62 62 64 76 8O 82 84 86 TABLE 19 20 21 22 23 24 2.5 26 27 28 29 3O '31 32 33 34 Intermediate and Investment Import Coefficients by Sectors, 1974 ............................. Exports in 1974 and Projected Rate of Growth by Sectors Sectoral Employment by Education, 1974 ............ Labor Input Coefficients by Sector, 1974 ........... Geographical Distribution of Sectors and Government Employ- ment, 1974 .......................... Composition of Labor Force by Education Level, 1974 ..... Wage Rates in Large-Scale Sectors, 1974 ........... Types of Variables Transferred Between the Macro Model and Migration Model ....................... National Accounts in 1981 Under Alternative Development Stra- tegies ............................ GDP by Regions in 1981 Under Alternative Development Strate- gies ............................. Value Added by Sectors in 1981 Under Alternative Development Strategies ................. _ ......... Growth Rate of Value Added by Sectors in 1981 Under Alterna- tive Development Strategies ................. Employment by Sectors in 1981 Under Alternative Development Strategies .......................... Growth Rate of Employment by Sectors in 1981 Under Alternative Development Strategies .................... Urban Employment, Migration and Unemployment in 1981 Under Alternative Development Strategies .............. Rural and Urban Wages in 1981 Under Alternative Development Strategies .......................... viii Page 88 9O 91 92 94 95 103 105 110 111 113 116 117 120 121 124 LIST OF FIGURES FIGURE Page 1 Framework for Classifying Treatment of Labor and Migration in Applied Models ........................ 14 2 Product and Factor Market Disaggregation ........... 31 3 Rural Enumeration Areas and Urban Areas of the Migration Survey ............................ 56 4 Sectoral Disaggregation by Scale and Location ........ 74 5 Linkages Between Macro and Migration Model .......... 104 ix I. INTRODUCTION Problem Setting Throughout the developing world, countries are experiencing rapid rates of urban unemployment, rural-urban migration, and urbanization. Urban employment opportunities have been growing too slowly to absorb even the natural growth of urban population. In general, though the rate of output growth has been between 5 to 7 percent per year, the growth rate of nonagricultural employment has been negligible. This slow rate 0f growth of nonagricultural employment accompanied by a high rate of rural ~urban migration creates urban unemployment problems of increasing intens ity. Besides these high rates of unemployment, the rate of growth 01’ unemployment is also of concern. Turnham (1970), in a study of eight countries, concluded that the number of unemployed was growing by 8 per- Cent annually. Rural-urban migration also leads to a high growth rate of urban Ce" 1iers. According to studies carried out by the Population Program Div‘i Sion of the Economic Commission for Africa for the period 1965-70, the growth rate of urban population in Africa was 6.1 percent per annum. A] 1 Owing for the natural growth of urban population at 2.5 percent, a'bout tFA'I'Q‘thirds of urban population growth can be attributed to migration. In “umber of principal cities there has been an urban growth rate averag- ‘i "Q 12 percent per annum. while ina few of these cities the rate has “afiched as high as 15 percent a year (ECA(1971)). Where labor is the dominant input in agricultural production, these high rates of rural outmigration have been a factor contributing to naticanal food deficits and rising food prices. These countries must use scarce foreign exchange to import basic food grains in order to supple- ment domestic prdduction for a growing population. ‘These problems have occurred despite satisfactory output growth rates in a large-scale capital-intensive manufacturing sector, and have given rise to the "growth without development" phenomenon. More specifi- cally/,. there is a growing consensus that a mere increase in output is not enough and that such an increase must be accompanied by other factors, Part‘i cularly income-earning opportunities among the bulk of the popula- tion- Economists have been trying to understand the role of labor in the deve‘l 0pment process. To analyze this, models were proposed based on his- . tor-i Cal experience. Some of the best known models were based on the concept of dual economy which has as its central feature the coexistence 0f. ii 'large agricultural sector--with its traditional technology and low Iabo‘, productivity--with an active and dynamic nonagricultural sector. Lewis (1954) and Fei - Ranis (1964) based models on this concept of clualism and assumed that surplus labor existed in agriculture. The development process was viewed as a shift in center of activity from a9"“iculture towards industry as surplus labor was transferred from agri- (:““‘ture to the nonagricultural sector. I As evidence accumulated that the marginal productivity of labor in aQriculture was positive (Mazumdarv(l965), Hansen (1967)), the assump-.: 1t“ on of surplus labor was abandoned. These neo-classical models, also based on the dual economy concept, focused on the differences in the Marginal productivity and wage rates between the agricultural and indus- trial sectors. The development process again relied on transfer of labor out of agriculture to achieve the optimal resource allocation. Expectations based on these dual economy theories that industrial- ization in developing countries would be associated with an aggregate supp'ly and demand balance in the factor market, proved disappointing. The provision of employment was not automatically ensured by output expan- lsion, and wage rates in the modern sector of urban areas rose substan- tial'ly . Migration, instead of solving the problem of unfulfilled labor demand in urban areas, has created a problem of excessive labor supply and co ntinues despite high unemployment in urban areas. This highlighted the fa ct that the problem of labor absorption and migration was treated inadequately in these models. In view of the high and rising rates of rural-urban migration and urban unemployment, much of the controversy concerning the role of labor I" the process of economic growth has turned to the need for generation ‘Of employment opportunities instead of fulfilling the need for additional workers to the industrial sector. Another problem of great practical concern relates to treatment of Iaer in applied planning/policy models. There is a wide gap between app} ied models and empirical observations. Blitzer (1975) observes that “ht.“ very recently, labor absorption entered only tangentially, if at a‘ 1 , into most applied models. In most cases either labor is not treated at all or is looked at only on the demand side. Where labor is treated, v3 ry rarely is it disaggregated by education. In situations where employ- r“Qnt and migration are education-specific, models that treat labor as thogenous are too simplistic. Also, in most of the cases labor is gr- Constrained at an economy level and in very few cases is it disaggre- gated by location. In cases where labor is disaggregated by location, the Irule has often been to specify exogenously migration from one region to another. Although a fair number of studies now exist on migration that give valuable insights into the process of migration, very little attenvir>t has been made to integrate migration into a policy framework taking into account macro level interactions. I Thus the usefulness of conventional applied models to examine the prob1 ems of employment and migration facing developing countries is limited. There is need for a framework that is more disaggregated and can capture intra- and inter-sectoral and regional interactions in both the product and factor market. Such a framework should also treat migra- tion explicitly in order to meaningfully analyze output, employment, and migra tion. Objectives of the Research The general objective of this.study is to provide a framework to qua“‘lzify the impacts of alternative policies on output, employment and migration using a comprehensive set of micro-level data generated by fiej d surveys.1 Specifically, the objectives are: (1) to develop and improved analytic framework to analyze the relationship between output, employment, and migration at the macro level; \ h IThese field surveys are: Byerlee et a1. (1976) Migration; Lied- le and Chuta (1976) Small-Scale Industry; Spencer and Byerlee (1976) P Q rm Management. a1- (2) to apply the framework to the Sierra Leone economy using pri- mary data to gain insights into growth, labor migration and employment prospects of the economy; and (3) to examine the implication for output, employment and migra- tion of a number of alternative development strategies. The impact of strategies examined can be broadly classified into three groups. In the first group, the impact of various agri- cultural development strategies such as agricultural export promotion and increased agricultural productivity will be examined. In the second group, the impact of various strategies to promote labor-intensive nonagricultural sectors such as small-scale industry promotion and a switch to labor- intensive techniquescfliproduction in large-scale industry will be examined. Lastly, the impact of relaxing foreign exchange constraint by increasing foreign capital inflow will be examined. Thesis Outline In the second chapter, the treatment of labor and migration in theer and practice is critically reviewed in order to identify existing "ea knesses and a more realistic framework for analysis of relationships e":Ween output, employment and migration is proposed. This framework is Used to construct macro economic and migration models which are applied ‘19 the Sierra Leone economy. An overview of Sierra Leone's recent deve- ‘meent with emphasis on output, employment and migration is provided in Cr. apter III. Sierra Leone was selected because of the availability of t:Qmprehensive micro-level data generated by field surveys, and the eco- r‘thmy has features common to many developing countries. In Chapter IV, econometric analysis of migration is presented to quantitatively estimate the magnitude of various factors affecting migration and to test if there is any significant difference between the behavior of educated and unedu- cated migrants. The elasticities of migration derived in this analysis are used int-a migration model. In Chapter V, macro-economic and migration models based on the framework proposed in Chapter III are laid out. The models are applied to the Sierra Leone economy to gain insights into the output, employment and migration potential of the economy and to examine the implications of different policies. Results of both the Projection period and policy runs are presented in Chapter VI. A sunmary of the model framework, and conclusions on the effects of various deve- Torment strategies on output, employment and migration together with the S uggestions for future research are presented in the last chapter. II. METHODOLOGY FOR MACRO ANALYSIS OF OUTPUT, EMPLOYMENT AND MIGRATION The objective of this chapter is to build a framework within which output, employment, and migration can be analyzed. The treatment of labor and migration in both theory and practice is reviewed with empha- sis on their adequacy to analyze output, employment, and migration. Since theoretical models provide the framework within which applied mode1 s are constructed, they are reviewed first. This is followed by a revi ew of the treatment of labor and migration in applied policy models. 1" ‘11l1e third section, empirical evidence on the behavior of labor mar- kets and migration in Africa will be reviewed so that existing conditions i" 1t11e factor market can be incorporated into the proposed framework. Based on identification of weaknesses in present methodological designs, 5" 'irnproved framework is presented in the last section. The proposed framework will be used to construct models of Sierra Leone economy in Order to project the potential of the economy, and to analyze impacts of vahious policies on output, employment and migration at the macro level. Treatment of Labor and Migration in Theoretical Models me of Labor and Migration in Economic Growth Studies of economic growth and development have sought to under- 3 ‘tand the role of labor in the growth process. To analyze the process of Qconomic growth a number of models have been proposed using the concept (If a dual economy. The dual economy models divide the economy into two taroad Sectors: the traditional agricultural and the modern nonagricultural 7 sectors. The first of these models assumed surplus labor in the agricul- tural sector (e.g., the Lewis and Ranis-Fei models). The classical models are based on the assumption that the marginal product of the agricultural labor force is zero or even negative (i.e., surpT us labor) and that wage in agricultural sector is institutionally determined and is above its marginal product. In the early stage of if deve10pment, this redundant agricultural labor is available to the indus- tria1 sector at a constant wage which is equivalent to the institutional agri cultural wage plus a premium to overcome constraints on labor mob-i "I ity (e.g., a higher cost of living in urban area and the psycholo- Slice 1 cost of migrating). Hence the labor supply curve to the industrial sector is perfectly elastic. The development process is then viewed as generating sufficient growth in the industrial sector of the economy to Pennit the transfer of this surplus labor from agriculture to industry. Criticisms of the Ranis-Fei model focused on its level of disaggre- get-i on, its simplicity in the treatment of labor movement, and on the r‘ea‘l ism of its assumptions. Assumptions especially about labor market con ditions are criticized. Labor is assumed, to have zero or even nega- ti Ve marginal productivity in the agricultural sector while implicitly +141 1 employment in the industrial sector is assumed. Empirical evidence from India (Mazmudar, 1965) and Egypt (Hansen, 1966, 1969) refuted the assumption of zero marginal productivity in the ElQricultural sector. Based on this evidence, the assumption of zero marginal product of labor was questioned. , Dual economy models which followed dropped the labor surplus assumption. One of the earliest neoclassical models was Jorgenson's ( 1967). Jorgenson assumes that all factors of production in both sectors have a positive marginal product and that there is a quasi-institutional wage in agriculture. This agricultural wage is variable and is propor- t'i onal to wages in the industrial sector which are determined by marginal Migration is treated as a mechanism which works to equalize The agricul - productivity. the marginal productivity of labor between the two sectors. tu ral labor force moves in response to a wage differential. The absolute P d‘i f‘ferential between urban wage and agricultural wage remains constant l '... :44- I'. over time so that wages grow at the same rate in both sectors and migra- ti on is sufficiently responsive to prevent a widening of the wage gap. Th e development process again relied on the transfer of labor out of ag riculture to achieve optimal resource allocation. The dual economy models provided valuable insights into the impor- ta rice of the agricultural/industrial nexus. However, the expectations, ha sed on these earlier dual economy theories, that industrialization in de veloping countries would be associated with an aggregate supply and demand balance in the labor market, were disappointing. Instead of ff 1 ling the demand for labor in the modern sector, migration has created a problem of excessive labor supply and continues despite high and rising "E'- employment in urban areas. Thus, highlighting the fact that the pro- b‘ an of labor absorption is analyzed inadequately in these models. It bacame evident that the problem of employment could not be studied in is 01 ation from migration. Due to the importance of migration in any ar‘a‘lysis of the role of labor. in economic growth, a methodological fr\alnework within which migration is studied and attempts to integrate m‘ . . . . ‘ Qration within a macro-economic framework are br1ef1y rev1ewed. 1O Integration of Growth and Migration Models Migration is the major link between rural and urban labor markets but treatment Of this process in conventional dual economy models was inadequate. Recently the human capital investment approach, or deriva- ti ve of the approach has become standard framework for economic analysis 0F migration. The human capital investment approach to migration postulates that potential migrants will move if the present value of an expected future income stream in some other region exceeds the present value of expected fu ture income streams in the present region of residence by more than the Human capital investment theory was first extended cos ‘ts of migration. The model can to 'the problem of labor migration by Sjaastad (1962). be expressed as: n v(o) = I [Yu(t) — Yr(t)]e'rt dt - c(0) t=0 where V(O) is the discounted present value, Y'u‘(t), Yr(t) is the income in De riod t in urban and rural regions respectively. C(O) is the cost of "ll Station; 11 is the numberof periods in the migrants' planning horizon; an d r is the discount rate. However, operationalizing this theory P0 3 es problems especially of arbitrary assumption about future wages, th 3 choice of discount rates, and the time horizon. In practice, the 5° 1 ution has been to include variables which approximate the present va Tl ue, e.g., by making migration a function of current income in the or ‘3 Sin and destination areas. Such a procedure implicitly assumes that ”IQ time horizon is unlimited and that both the income and discount rates ar e constant over time. ThQ , , theory provides a cogent explanat1on of the predom1nance of the young This simplifies the computation considerably. 11 araci educated in the migration stream. Younger people migrate more often because they have a longer time horizon over which to capitalize earnings d‘i fferentials. According to the theory, one would expect the rural- urban income differential to be eliminated eventually as a result of mi gration. It does not explain adequately why migration continues in sp‘i te of the high and rising unemployment in urban areas in deve10ping co untri es. An extension of the human capital investment approach has been deve- lo Dad by Todaro (1969) who explicitly relates rural urban migration to urban wages and unemployment in developing countries. The model can be ex pressed as: n V(O) = I [P(t)Yu(t) - Yr(t)]e'rt dt - C(O) t=O wl'n ere P(t) is the probability of being employed in urban destination as of: period t. V(O), Yu’ Yr’ C, Y and n are as defined previously. The decision to migrate from rural to the urban areas is related no ‘1: only the urban-rural wage differential but also the probability of fi nding an urban job. This probability of finding urban employment is de‘Fined as equal to the fraction of the urban labor force actually employed I" the manufacturing sector. Expected urban income is defined as the "r‘ban wage weighted by the probability of finding an urban job. Migra- ti On is then functionally related to expected wage differential which is u. e difference between the expected urban wage and the rural wage. The Todaro model is a contribution towards theory of migration by e): '31 icitly noting the interrelationship between wages and unemployment an d explains why migration can take place despite high and rising um enlployment. 12 Since the treatment of migration in the conventional dual economy models was too simplistic, a logical extension was to integrate the dual economy and'migration models. To analyze the interrelationship between growth, employment and migration, Harris and Todaro (1970) incorporated the expected wage differential model of migration proposed by Todaro wi th the conventional dual economy model. The Harris-Todaro (1970) model divides an economy into two sectors: ma nufacturing and agricultural with manufacturing located exclusively in urban area and the agricultural sector in rural area. The sectors' pro- du ction functions are neoclassical. Labor and capital are inputs in th e production of manufacturing and land and capital are inputs into ag ricultural production. Supply of land and capital is fixed while 1a bor is the variable factor and is allocated endogenously between the two sectors. Furthermore, they assume an institutionally determined wa ge rate in urban areas and a wage determined by labor supply and demand in rural areas. The behavioral function of rural-urban migration is a modified version of the Todaro (1969) model of migration. Though the treatment of the labor market interaction is more rigor- °US and migration is more explicitly treated than in the dual economy models of Ranis-Fei and Jorgenson, there are several weaknesses in the ”3 Y‘ris-Todaro model. As a dual economy model, it is highly aggregated. AS V‘iculture is the only activity allowed in the rural region, and manu- faCturing is the only activity allowed in the urban region. The model jg"'Iores the urban traditional sector which affects both urban income and employment probability and, consequently, expected urban income. The "IQ del does not treat adequately important interactions indie product ma "ket between the agricultural and nonagricultural sectors. 13 These theoretical models which analyze interactions in the factor Iar1 analyze employment and migration within a macro-economic framework. Treatment of Labor and Migration in Applied Policy Models The theoretical models together with empirical observations provide a Iaase for construction of applied policy models. There, however, is a w' 1‘ de gap between theory and practice. Until very recently, labor absorp- t “i on and the labor transfer process entered only tangentially, if at a.Tl ‘1, into most applied models. In this section, applied policy models will be reviewed briefly w'-i ‘th emphasis on the extent to which labor and migration has been treated. For this purpose, three criteria are used to evaluate the ade- . q t_aiacy of the applied policy models to analyze employment and migration. I) Is there a treatment of labor? If so, does the model look I only at demand or at both demand and supply? 11) Is labor assumed homogenous or is there disaggregation by education or skill? III) Is labor constrained at the economy level or is it disaggre- gated by location into rural and urban areas? Does treatment include rural-urban migration? I Using the first criteria, there are models that exclude labor a j together (see Figure 1). For example, four of the seven planning deels in a well-known volume on methodology of planning, edited by 14 memooz oqumc< 2H zomhmHmm<4u mom gmo3m25 mmcox .am\~Fw¥m QSH »a + macwaaw_wca , u a<> - mean: an oupxmz e.g.,“? 92%.... s x w mmmgmwmmre I mmpgmam oczgm once: a cosmmcmm a u a + .o P an awaamwz ea .maLmH an even“ mmomwwwm- .mmmmpo mgmmm Xe ownsmN . .um\Fchm ocsam 9% e . .. .......M..__e.__.._uw umum we < . . a axuancogp wLaNaz Ammmpv gucwm e maocwmoe°= an seas an arucu Ammmpv oponam . o 985 m 395. <_ 295 c .InIIIIIIIIIIIIIIILrIIIIIIIIIIIIIIIIIIuIIIIIIIaIIIIaIaIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIILHN 15 Ade'l man and Thorbecke (1966), have no discussion of labor. These models area of Pakistan by Chenery and MacEwan, Zambia by Seers, India by Bergs- mar1 and Manne, and Mexico by Manne. Some models look only at demand for labor and employment projection is (jone either within a partial equilibrium framework or within a general equ ‘ilibrium framework. The partial equilibrium model focuses on the Ni ncipal variables involved in the particular problem studied, and treats the rest of the economy as exogenous. The general equilibrium approach takes simultaneity into account and captures both the direct and ind firect effects of the changes a given variable makes ,in the system. In partial equilibrium framework, employment is a function of inde- pendent variables and depending on the values of independent variables, einp 'loyment is projected (e.g., Sabolo (1967) and Singh (1969)). Sabolo (l 9 69) regressed the logarithm of sectoral employment as a proportion of The to tal population on the logarithm of per capita GDP for six sectors. reg ression equations provided estimates of sectoral employment elastici- ‘ ti es with respect to per capita income. These elasticities were used to project employment trends assuming a given GDP growth rate. In a general equilibrium approach, employment is put within an I" put-output orllinear programing framework (e.g. Harzari (1970)). In tI‘ ‘3 s format it is assumed the country modelled has surplus labor and does "I. t regard labor as a constraint. The only constraints on the produc- ti Ve capacity of the economy are those related to capital, foreign 9" Change and other intermediate inputs. Projected employment can be a9 Qregated or disaggregated at various levels by skill or by location °" by both skill and location. An equation might look'like: 16 where: x1. is gross output in sector 1 lik is the labor input coefficient for sector i specific for labor Sk111 k Eik is employment in sector 1 of type k labor Models that exclude supply or look only at the demand side impli- ci tly assume surplus labor. In situations where labor is not surplus or where a category of labor (e.g. skilled) is in short supply, models that do not check for consistency on the supply side are of limited use. A more realistic group of models are those that examine labor in bo th demand and supply side and treat it in the same way as capital, fc> reign exchange, and other intermediate inputs. Labor allocation is th us a constraint in this group. AcCordingly, the supply of labor is as sumed fixed exogenously at some upper level and the equation may take th e following form: lIik Xi —<-I'k WI"- ere: Iik and X1 are same as before and tk is an exogenous projection "M: 1' °F the total supply of labor in category k. This ensures that the output Ia\nt'el is consistent with availability of labor besides other intermediate I“ puts: Within this supply and demand approach, the model can either b3 static (e.g. model of Peru by Thorbecke (1970), Israel by Bruno (1966) a“ (1 India by Sandee (1957)) or dynamic (e.g. Nigeria by Byerlee (1971), Bhazn by Yap (1976)). Using the second criteria, labor can be treated either as a homoge- "QUs input (e.g. Peru byp'Thorbecke (1970), India by Sandee (1969)) or d ii =5iiiggregated by education or skill (e.g. Israel by Bruno (1966) and 17 Philippines by 1L0). Where employment and migration are education- or ski ll-specific, models that treat labor as homogenous are simplistic. Using the third criteria, labor can be aggregated either at an economy level (e.g. Peru by Thorbecke (1970) and Israel by Bruno (1966)) or disaggregated by location into rural and urban areas (e.g. Nigeria by Byerlee (1971) and Brazil by Yap (1976)). Where labor is constrained at the economy level, i.e., where total demand for labor in all sectors is less than or equal to total labor supply, the assumption that labor is pe rfectly mobile between rural and urban areas is implicit and the labor su pply constraints by geographical locations are assumed away. Hence, these models cannot analyze interrelationships between employment and mi gration. Models that do disaggregate labor by location capture interactions in the factor market between rural and urban areas and can examine employ- me fit and migration more comprehensively. The model allows transfer of labor between regions and there is explicit migration behavior. In most 0": these models (e.g. Nigeria by Byerlee (1971) and Brazil by Yap (1 976)), the migration function used are extensions of the Todaro-Harris ”Odel of migration based on the human capital investment theory discussed ea r1 ier. Synthesis of Empirical Evidence on Behavior of Rural andFUrban Labor Markets and Migration in Africa In this section, labor markets and rural-urban migration will be ré‘l‘iewed so that existing conditions in the factor market can be incor- pQ i"taed into the proposed framework. 18 The Rural Labor Market The great majority of pe0ple in African countries live in rural areas. The estimates range from over three quarters of the population to 99 percent, depending on the country and definition used for rural areas. It is estimated that the natural rate of population growth in these areas is about 2,3 to 2.6 percent (ILO (1971)). Underemployment rather than unemployment is a major problem for rural people in Africa. There is evidence that labor is a binding con- straint at the peak agricultural seasons (Spencer and Byerlee (1976), Norman (1973)). The demand for hired labor is at its peak just at the time when the need for potential wage earners to work their own holdings is also at its highest, and hired labor constitutes only a small pro- portion of the total labor input. Rural nonagricultural activities are an important source of employ- ment. Employment in nonagricultural activities can be a primary or part-time occupation during the slack labor demand period. Studies in Nigeria (ILO (1970)) and Sierra Leone (Byerlee et a1. (1976)) estimate that about one;fjfth_of the males in rural areas engage in nonfarm acti- vities as primary occupations. In Sierra Leone, Spencer and Byerlee (1976) estimate limpercent of the male papulation in rural areas worked part-time in rural nonfarm activities. Evidence is accumulating indicating that the rural labor market is cgmpetitiye. In Sierra Leone, Spencer and Byerlee (1976) analysis showed wage rates varied by labor type and season reflecting differences inthe opportunity cost of labor of different types and at different sea- sons of the year. Findings from Sierra Leone (Spencer and Byerlee (1976) and Nigeria (Norman (1973)) indicate a good correspondence between the MVP of labor and the wage rate. 19 The Urban Labor Market The urban labor market is reviewed with respect to its various related dimensions. These are a) the structure of employment, b) earn- ings, c) rural urban migration, and d) unemployment. The Structure of Urban Employment Following the familiar notion of urban duality, urban labor mar- kets in Africa can be divided broadly into two categories: large-scale (or organized) and small-scale (or unorganized) sectors. Factor inten- sity and characteristics of labor employed between the two sectors differ. Large scale sectors are usually characterized as capital_jntensive in contrast to small-scale labor-intensive sectors. Educated persons com- prise a higher proportion of employment in the large-scale compared to small-scale sectors. Within the large-scale sectors a further subdivision can be made between public sector employment (or government) and employment in the private sector. Ggygrnment employment, in fact, accounts for half or more of the total employment in the large-scale sectors. In Kenya, ILO (1972), estimates government employment to account for 40 percent of total employment in the large-scale sectors. Small-scale sectors include small-scale trade, transport, services, and small-scale manufacturing, and employment in this sector is impor- tant. This sector is relatively easy to enter in the sense that very little capital--human or physica1--is required and there is a high job turnover. In Sierra Leone, Chuta and Leidholm (1976) observed that in 1974 small-scale industrial establishments employed 88.6 thousand (or 95.6 percent) of the total industrial employment of 92.7 thousand. In Nigeria, Kilby (1969) estimated a total employment of 28.7 thousand in 20 small-scale sectors. This sector is also assumed to be the major employer of uneducated people. In the case of Sierra Leone, Byerlee et a1. (1976) observed that one-half of the employed migrants without educa- tion were located in small-scale sectors. It is also important to examine the trend in employment in the large-scale and small-scale sectors. Despite high rates of growth of output in the large-scale, modern nonagricultural sectors, the growth of employment in these sectors was disappointing. Though the growth rate of output in the modern sectors was about 5 to 8 percent per annum, the growth rate of employment was less than 2 percent. In Kenya, Ghai (1971) observes that while output in real terms increased at an annual rate of 6.3 percent between 1964 and 1969, employment grew at a rate of only 3.5 percent. And even then most of the expansion in employment was due to an increase in governmental jobs, where employment increased at an average annual rate of 5.5 percent. A number of reasons have been advanced for this higher growth rate of output relative to rate of growth of employment (or productivity increase). A major reason has been due to institution of policies which have distorted factor markets--making capital cheaper and labor expen- sive--encouraging capital/labor substitution. Policies which have dis- torted factor markets in favor of capital include overvalued exchange rates, subsidized credit arrangements, tax policies such as accelerated depreciation and investment allowances and duty-free imports of capital goods. Policies relating to minimum wage rates have made labor expen- sive. A number of other factors also have contributed to this productivity 21 increase, including more and better training of labor and technological progress which has generally been capital-intensive.1 Data on employment expansion in the small-scale sectors are vir- 'tually nonexistent. Estimates can be made by subtracting large-scale employment and unemployment from the total labor force in urban areas. Byerlee and Tommy (1970) used this method to data from Freetown, Abidjan and Nairobi, and they estimate employment in small-scale sectors to be increasing at least 10 percent per year in these urban areas. Earnings in the Urban Labor Market Earnings inthe urban labor market differ between large-scale and small-scale sectors. It is observed that the wage rates in the large- scale sectors are not determined by market conditions but by such factors as government minimum wage legislation and relative bargaining strengths of trade unions. Observations in Kenya (Ghai (1968)) and in Nigeria (Diejomah and Ormalide (1971)) suggest that wage rates in the modern large-scale sector are higher than dictated by market forces. Within the large-scale sectors there is some evidence that the private large-scale sector pays a higher wage than government. There is also a large difference in wages between the large- and small-scale sectors. Wages in the small-scale sector are determined by supply and demand of labor. Evidence from Kenya (ILO (1972)) and Sierra Leone (Byerlee et a1. (1976)) show that earnings inithe small-scale sector are below the minimum wage established for government employment. (197 )lsee Frank (1967), Berg (1970), Eicher et a1. (1970) and lie 2 . . 22 Data available do not indicate any major changes in the level of wages in large-scale sectors in recent years. In a survey of African economies, the highest increase was recorded in Kenya, where average earnings increased by 5.8 percent in 1972. In other countries of East Africa, wages have stagnated. In several West African countries, minimum ‘wage rates have remained at the same level for many years (ECA (1973)). One of the reasons advanced is that with the recognition of the inequa- lity between rural and urban areas and between large- and small-scale sectors in urban areas, governments have been reluctant to raise minimum wages. Data on trend in earnings in the small-scale sectors are practically nonexistent. It is very likely that earnings in this sector have also stagnated or even decreased as a result of increased absorption of labor, forcing labor productivity down (Byerlee et al., (1976)). Rural-Urban Migration The growth rate of urban population in Africa for the period 1965- 1970 is estimated at 6.1 percent (ECA (1971)). There is a tendency for migration to be directed mainly towards one or two of the largest cities in each country so that the growth rates of cities with populations of 100,000 or more are often higher than the rates for the urban areas as a whole. Thus, in a number of principal cities, there has been an urban growth rate averaging 12 percent per annum, while in a few capital cities the rate has reached as high as 15 percent a year (ECA (1971)). If allowance is made for the natural growth rate of urban population at 2.5 percent, about two-thirds of urban population growth can be attributed to migration. The significance of rural-urban migration for the problem 23 of urban unemployment is that it makes the already serious problem of unemployment far worse . Typically, migrants are younger and better educated. In Tanzania, Barnum and Sabot (1975) observed strong, positive relationships between rates of migration and educational level, and that educational selecti- vity has increased over time with secondary school leavers forming a higher proportion of total rural-urban migrants. In Ghana, Caldwell (1969) reported that 65 percent of respondents with no education had never migrated or did not intend to migrate, compared to 17 percent for those respondents who had some, secondary schooling. In Kenya, ILO (1972) Observed that the probability of migrating for persons with nine years Or more of schooling is about five times greater than for persons with 1 ess education and over twenty times greater than for those without S chool ing. Studies of rural-urban migration consistently show the importance Of economic factors in migration. The basic economic consideration being rural and urban income (either absolute or expected) or their difference. I n Ghana (Rourke and Sakyi-Gyinae (1972)), Nigeria (Diejomaoh and Ormi- 1 ade (1971)), Kenya (Todaro (1971)), and Uganda (Knight (1968)) have Observed that there is a significant rural-urban income disparity. In a S urvey of economic conditions in Africa by ECA (1973), it was found that Wages paid to urban employees are generally higher than incomes in the agricultural sector. In Kenya, ILO (1972) observed that statutory mini- mum wages inurban areas are well above the incomes of all groups in the Y‘Ural areas except for the more prosperous small-holder and the average DWner of a nonagricultural rural enterprise. A number of reasons have been advanced for this gap such as minimum wage rates in urban areas and 24 low rural incomes. This has been augmented by government bias in the provision of social services to urban areas. A number of authors have also observed thatrural-urban income dis- parity is higher for educated than for uneducated persons. In Tanzania, Sabot (1975) observed a strong positive relationship between wage incomes and levels of education. For urban wage earners, average income rose from Sh. 251 for those with no education to Sh. 861 for those with some secondary education. In Kenya, ILO (1972) observed that gains from mi gration are usually much greater for the more educated than for those With less education. ‘Despite the remarkable similarity in the response of migrants to V‘Ural and urban incomes, the effects of education on migration have differed markedly. Beals (1967) found that in Ghana migration decreased with higher levels of education at both the origin and destination, while Greenwood (1971) found that in Egypt migration increased with higher 1 evels of education at both the origin and destination regions. This ambiguity is due partly to the inclusion of aggregate educational attain- r"lent variables or educational enrollment variables in econometric models. Very few studies (e.g., Barnum and Sabot (1976)) have disaggregated the population by education. l-l Irban Unemployment The evidence available does provide ample empirical confirmation that urban unemployment is of a high magnitude and that it is growing In most of the African countries, the aggregate rates are typi- V‘apidly. The ILO (1972) Cally between 10 and 15 percent of the labor force. mission to Kenya estimated that the level of urban unemployment was around 15 percent. Nonofficial statistics relating to unemployment 25 estimate that for all of Africa in 1970, the level was 10.84 million or 7.9 percent of the economically active population (Sabolo (1969)). The available evidence shows that the rates are high among youths and the educated. An ILO (1972) mission to Kenya observed that the majority of the unemployed were between 15 and 24 years of age. While the rate of growth of educated workers due to rapid expansion of school systems is impressive, the rate of growth of wage jobs has been negligible, giving rise to phenomenon called "educated" unemployment. In most of these countries, only a small portion of students completing primary school enters secondary school; the group which leaves school after approximately seven years of education forms the bulk of job seekers ‘i In the urban areas. In Kenya, Elkan (1973) estimates that of 150 thou- Sand leaving primary school each year no more than 30 thousand go to Secondary school. This accelerated educational system combined with educational selectivity of migrants results in higher urban unemployment r‘ates for educated. In Sierra Leone, based on Household Survey (1971) data, the‘unemployment rates for the educated in urban areas were con- 8 istently higher than for the uneducated. The survey also shows that unemployment is worse in large urban areas than in small urban areas (e.g., lL‘reetown had 15.5 percent, Bo 15.1 percent compared to 9.5 percent in 0 ‘ther small urban areas). Data from individual countries tend to confirm the increasing inci- dence of unemployment over time (ILO (1970)). There is very little ‘3 Information on trends in unemployment at a disaggregated level. Barnum and Sabot (1975) observed in Tanzania the differential in rates of Qrowth of unemployment by educational level. 26 gglicy Implications The above review of labor markets in Africa shows that urban unem- pl oyment is high and probably increasing. Both supply and demand factors have contributed to the emergence of the urban unemployment problem. While the supply of labor in urban areas has increased rapidly, partly due to high rate of natural growth of population and largely due to rural- urban migration, the demand for labor in the modern large-scale sector has stagnated or increased very slowly, primarily due to adoption of capital intensive technologies . Various strategies have been suggested for approaching the employ- ment problem at the macro level. These policies can be grouped into those which attempt to increase the demand for nonagricultural employment and those which seek to decrease the supply of labor in urban areas. In the former group, for example, are policies that encourage small-scale 1 abor-intensive sectors. In the latter group, supply of labor can be r‘educed in the long run through reduction in the natural rate of popu- 1 ation growth. The labor supply also can be decreased in the short run by reducing the rate of rural-urban migration. Among the policies sug- gested for decreasing rural-urban migration are reducing the rural-urban Ti Income differential by increasing rural incomes through agricultural Cl evelopment programs . Urban unemployment, however, cannot be studied at the macro level W ithout reference to the total economy. In particular, the impact of Various development strategies must consider the relationship between Qrowth, employment and migration. For example, the interactions in the product market between agricultural sectors and nonagricultural sectors are important. Likewise, interactions in the factor market between rural 27 and urban areas are important in determining the supply of labor to each region. As observed earlier in this chapter, the existing analytical framework toanalyze these interactions in the product and factor‘markets at macro level are inadequate. Towards a Framework for Macro Analysis of Output, Employment and Migration In reviewing applied policy models, it was observed that they do not gi' ve adequate attention to labor and migration. Models that do not disagg regate labor by education and rural urban locations are of limited 1158.1" analyzing employment and migration. In this section an improved framework is suggested. This framework will be used to construct macro economi c and migration models which are applied to Sierra Leone economy tO‘ aha ‘Iyze output, employment and migration. I n order to analyze output, employment, and migration, a more rea- HSHC disaggregation of both product market and labor market and expli Ci t treatment of labor migration is needed. ESQ-w Market Di saggregation Characterizing rural areas with agricultural production and urban areas with modern manufacturing seems unrealistic. Evidence reviewed earl 3 er in this chapter showed the importance of the nonagricultural acti- VII-i es, especially small-scale manufacturing and trade, as a source of income and employment for rural populations. The evidence also showed . the importance of traditional small-scale industries in urban areas, Which are operating under differenttechnological frontiers and produc- thn functions “than large-scale sectors. A number of authors have proposed a higher level of disaggregated framework. Oshima (1962) argues that an adequate model should distinguiéh 28 tIlr‘ee sectors--agriculture, industry, and traditional trade services. Reynolds (1969) distinguished four sectors--two traditional (agriculture and urban trade services) and two modern (industry and government). Byerl ee and Eicher (1972) have put forward a case for dividing the eco- mm on the basis of three criteria--type of output, firm size, and Iocati on. Dividing the economy on the basis of these three criteria, they then subdivide the economy into at least four sectors--small-scale agricu‘l ture, small-scale rural nonfarm, small-scale urban, and large- scale urban. I n this study, output will be disaggregated on the basis of three criteri a- following Byerlee and Eicher (1972). are: These three criteria type of output, scale of operation, and location. The first criter-i on, type of output, divides the economy into agriculture and DOMQY‘i culture. This divislion is needed to capture interactions between these two sectors and their relative distribution as development Pmcest or due to impact of different policies. A factor which changes the relative distribution of these two sectors is the difference in income elasticities of demand for their output. The second criterion, scale of operation, divides the economy into 1a"ge'~scale and small-scale sectors. These two sectors differ markedly "I eclonomic characteristics as was reviewed earlier. Firms in small- scal e sectors are usually family Owned, depend largely on indigenous rest)urces, and are labor-intensive, in contrast to the capital-intensive I“‘Qe-scale sector. The small-scale sector also employs a relatively “lgher proportion of uneducated labor in contrast to large-scale sectors WI‘EY‘e educated labor dominates. Besides these there are other differences such as demand patterns which may lead to different employment implica- tions. 29 The third criterion for disaggregating output is on the basis of location. This divides the economy into rural and urban sectors. This rural-urban distinction is important because rural and urban production and employment problems differ greatly. This also facilitates linking the economy with labor markets, where rural-urban migration is affected by income and labor market conditions in both the rural and urban areas- This will also enable analysis of impacts of various policies on rural and urban income distribution. Factor Market Disaggregation Where there is a separation of labor market in the urban areas by educati onal level and where there is a differing response by educational level t:<3 factors affecting migration, the models that regard labor as homogen 0115 are unsatisfactory. ‘ I '1 this study, labor markets will be disaggregated on the basis of two cri teriaulocation and education. The first criterion, location, divides the labor market into rural and urban. This is necessary in order to Properly analyze rural and urban labor markets and to give explicit treatme nt to the process of rural-urban migration as a link betvieen ““91 and urban labor markets. An added advantage of this disaggrega- tion i S that it allows the structure of consumption demand to vary by NV“ and urban population. Given the significance of consumption, and t0 the extent that the demand patterns differ between the two groups, thls (11‘ saggregation adds more realism. Consumption will be computed eOd°Qenously for each comnodity for rural and urban groups separately, (151119 population-specific elasticities. The second criterion for disaggregating labor market is by educa- tional level. The labor force cannot be regarded as homogenous if 3O employment and migration are to be analyzed adequately. For this study, the labor force will be disaggregated by education into two groups: uneducated and educated. This dichotomy is of value in understanding employment and migration which are highly education-specific. Migration will specifically be disaggregated by educational level into two streams, uneducated (those with less than four years of educa- tion) and educated (those with four or more years of education). This disaggregation is valuable in an analysis of migration which is education- speci Fic. The small-scale sectors in urban areas are explicitly incor- porated - These sectors affect urban employment probability and wages and, hence, expected urban wages. Figure 2 shows the division of output and labor on the basis of the cri teria discussed. Only sectors that are of practical importance based on empirical evidence and observation in Sierra Leone are used for thi 5 study though conceptually some sectors can exist, e.g., large- scale sectors (be they agricultural or nonagricultural) in rural areas are not of practical significance. er1 Interactions The framework captures important linkages within rural and urban 53m"?! and between rural and urban sectors in product and factor markets . At the macro level, several types of intersectoral linkages are i"“l:>ortant. Linkages in the labor market include labor allocation between agricultural land nonagricultural activities in rural areas, and mom ‘Force distribution, through migration, between rural regions and 1 urban areas. Linkages within product (markets include backward and forward interindustry demand linkages between agriculture and other Sectors of the economy and demand linkages through consumption 31 ‘ ' .. l ' .— I Rural , I , Urban ’ Supply Uneducated vMigratiIm‘r Uneducated Uneducated Educational 0f """ Stream's ''''' ‘- Level Labor , a , + D1sagg. Rural T” .‘T' Urban Educated Educated ' Educated + \ _. \L 1) Y| 41» I _ Rural Rural Urban Urban Smgll Small-Scale Small-Scale Small-Scale Small-Scale Scale Demaricj Agric.' Nonagric. Agric. Nonagric. + Scale or: " - Labor. urban La: 9 D‘sagg. Large-Scale ScaIe Nonagric. + 4— Agric. +I + Nonag. +| 4- Agric. +| 4- Nonag. + I +————— Rural 4 4‘ Urban —_—* -— ‘ Locational Disaggregation . I' X: 01’ No Pratical Significance FIGURE 2 PRODUCT AND FACTOR MARKET DISAGGREGATION expend itures by rural and urban. populations. These linkages should be captur ed in order to meaningfully analyze output, employment and 11119131: ion. III. OVERVIEW OF OUTPUT, EMPLOYMENT AND MIGRATION IN SIERRA LEONE In this chapter the Sierra Leone economy is briefly described, with emphasis on output, employment and migration. Many of the features of the Sierra Leone economy common to other developing countries are highl i ghted. The first section summarizes the national account statistics. The Performance of the most important sectors of the economy is described in the second section. The third section discusses the population and labor force with emphasis on its distribution between the rural, small urban and large urban regions. Employment, unemployment and migration are di scussed in the last section. National Accounts (Brass Demestic Product (GDP) at factor cost in constant 1963/1964 prices grew at an average rate of 4.2 percent per annum, from Le 214.8 mlIIIOn in 1963/64 to Le 287.9 million in 1970/71. The population is esimatedl to have grown at 2.3 percent per annum over the same period, "0‘“ 2 -18 million in 1963/64 to 2.71 million in 1970/71, so that GDP per capita - at constant 1963/64 prices grew at 1.9 percent per annum, from Le 93 in 1963/64 to Le 106 in 1970/71. Over the period 1963/64 - 1970/71, hOWEVer, the economy grew at an uneven rate. Whereas GDP grew by 4.2 percent per annum for the years 1963/64 - 1965/66, for the recession A I" 1See section under population. 32 33 years 1965/66 - 1967/68 it fell by 1.1 percent per annum. The economy recovered in 1968/69 and grew by 7.5 percent per annum during 1968/69 - 1969/70 period, but the growth rate again fell in 1970/71 to 1.6-percent. GDP can be disaggregated by expenditure into consumption, invest- ment, savings and imports and exports. Investment grew at an average annual rate of 18.3 percent over the period 1963/64 - 1970/71 with sub- stantial year-to-year variations. During the recession years 1965/66 - 1967/ 68, it dropped by 3.4 percent per annum. The share of investment in GDP 112:5; been increasing and almost doubled from 10.1 percent to 18.1 percent during the period 1963/64 - 1970/71. (Zonsumption grew at an average rate of 1.6 percent per annum during the period 1963/64 - 1970/71. Consumption, like GDP shows a substantial variaitz-ion and the trend closely follows GDP.‘ It grew steadily during the 1 963/64 - 1966/67 period, dropped sharply during the recession, and again catching up during the recovery period 1968/69 - 1970/71. The share of consumption in GDP has been falling steadily and declined from 97-2 percent in 1963/64 to 80.1 percent in 1970/71. ‘The other accounting activity of GDP expenditure is foreign trade. Tota] imports have consistently been greater than total exports, except duri"9 1968/69. The value of exports grew at an average rate of about 5 per‘C‘ent per year between 1964/65 and 1972/73. Imports rose at an ““31 average rate of 3.7 percent during the same period. Exports 3“ predominantly minerals and agricultural commodities and the mix of the tnnmo has changed very little. Mineral exports represented more than 75 Pevcent of the total exports, while agricultural exports represented 17 Percent. Total exports averaged about 25 percent of GDP at market prices and imports about 30 percent during the period 1963/64 - 1970/71. 34 Sectoral Performance Gross Domestic Product can also be disaggregated by industrial origin. Agriculture constitutes. the largest single sector in the economy and it is the dominant source of employment. According to the 1963 population census almost three quarters of the labor force was in agri- culture. Agricultures' contribution to GDP has been slowly declining from 38.6 percent in 1963/64, to less than 30 percent in 1973/74. Agri- cultu re grew only at about 1.5 percent per annum during the period 1963/64 to 1970/71. This growth is less than the rate of population growth which i ncreased from 1.5 percent a year in the 1960's to 2.2 percent a .Year 1 n the 1970's. In fact, the real GDP per capita in agriculture actual 'l y declined. Compared to the growth rates in other sectors, aQPlCu'I ture had the lowest growth rate. The impacts of these low rates 01: QPOWth in agriculture were felt both in terms of foreign exchange foregone because of food imports, especially rice, and in terms of pro- Viding food and employment for the growing population. One of the reasons for the poor performance of the agriculture ““0" is the limited investment allocated to the sector. Although agricuj tural investment expenditures have increased from 4 percent of development expenditures in 1963/64 to about 25 percent, they still account for less than 1 percent of GDP. Another reason for this poor PerfOY‘mance of the agricultural sector is the pricing policy which New“)! taxes agricultural output. In 1971/72, the farmers' share of the export price (f.o.b.) was about 70 percent for palm kernels and betWEEn 40 to 50 percent for coffee and cocoa. Over the period 1968/69 to ‘972/73, the total taxes raised directly from the agricultural sector amounted to about Le 19 million. In contrast, total government 35 expenditures, both current and develOpment, directly allocated to agri- cul ture were only about Le 14 million (World Bank, 1974). Rice is the main staple food in Sierra Leone and accounts onthe average for about 40 percent of the total value of crop production. In 1970/71, 808,000 acres were devoted to rice cultivation or over 50 per- cent of all land under cultivation. It is grown by about 81 percent of all fa rmers. Failure: to produce enough domestic rice for self-sufficiency has continually troubled policy makers. Prior to the early 1950's, Sierra Leone was self-sufficient in rice production, but rice became a major component of food imports in the early 1960's. Average annual imports of rice for 1970/71 - 1972/73 of 26,000 tons, were more than double 1950/61 - 1962/63 levels and were close to 10 percent of total rice con- sumpt-i on in 1974/75. ' Production of other food crops (excluding export crops) has shown an a"H'Tual increase of 2 to 3 percent during the period 1963/64 to 1970/71. Some Of the most important crops in this category are cassava, millet, groundnuts and citrus fruits. EEprort crops such as palm kernels, coffee and cocoa form the bulk 0f 39"? cultural export earnings. They represented 17 percent of total exports , with palm kernels accounting-for 8.7 percent, coffee 3.7 percent and cocoa 2.8 percent of the total exports during the period 1963/64 to 1970/71. ’ I‘lining is the second largest sector in the economy following agri- cU‘tUY‘Ea. Mining increased at an average annual rate of 3.1 percent during the period 1963/64 to 1970/71. Mining contribution towards GDP averaged around 17 percent during the period 1963/64 to 1970/71 with fittle year-to-year variation. The importance of the sector to the 36 economy is brought out more clearly by its contribution to export earn- ings and public revenues. Export of minerals has contributed more than 76 percent of the total value of domestic exports during the period 1963/64 to 1970/71. Exports of diamonds alone contributed 60 percent of export earnings during the period. The next important mineral is iron ore which contributed about 16.8 percent of export earnings during the period. The sector also contributes to revenues of the government through taxes on the mining companies, export duty on diamonds, royalties and license fees and profits of the joint enterprise, the National Diamond Mining Company of Sierra Leone. In 1970/71, the contribution of mining to current government revenue amounted to 16.6 percent. Much of the mining activity is of an enclave type, i.e., capital-intensive, foreign owned and with relatively few links with the rest of the economy. There is also the feeling that wealth provided by the diamonds has been respon- sible for the lack of urgency regarding reforms in agriculture. Manufacturing and handicrafts contributed on the average slightly more than 5 percent of the Gross Domestic Product with little year-to- year variation during the period 1963/64 to 1970/71. The average growth rate of the sector during the 1963/64 to 1970/71 period was only 2.9 percent per annum. A distinction should be made between the large-scale factory type industry and small-scale industry because economic charac- teristics of the two differ. Liedholm and Chuta (1976), in their analysis of data from a small-scale industry survey in Sierra Leone, found that small-scale industries make extensive use of labor and are parsimonious in their use of capital. The labor-capital ratio for small- scale industry is substantially higher than for large-scale industry and small-scale industries possess higher output-capital ratios. Liedholm 37 and Chuta (1967) estimated that small-scale industry in 1974/75 accounted for approximately 2.9 percent of Sierra Leone's GDP or approximately 43 percent of the entire manufacturing sector's GDP, empahsizing that small-scale establishments are indeed a significant component of Sierra Leone's industrial sector. Transport and communication has been expanding steadily at an average annual growth rate of 11.6 percent during the period 1963/64 to 1970/71. Its contribution to GDP increased from 6.8 percent in 1963/64 to almost 10 percent in 1970/71. Wholesale and retail trade is the third largest sector in the economy following agriculture and mining. The sector's average annual growth rate was 6.5 percent during the period 1963/64 to 1970/71. Wholesale and retail trade sectors' contributions towards GDP averaged between 13 and 14 percent. Construction grew at an average annual rate of 12.8 percent during the period 1963/64 to 1970/71. This growth rate was almost three times the growth rate of GDP and faster than all other sectors except utili- ties. Construction sectors' contributions to GDP increased steadily from 3.3 percent in 1963/64 to 5.0 percent in 1970/71. Utilities were the fastest growing sector, growing at 14.8 percent per annum during the period 1963/64 to 1970/71. However, the sector is the smallest of all the sectors and contributed less than 1 percent to GDP in 1970/71. Population Data for population in 1963 by age-sex and location are derived from the population census of Sierra Leone. For 1974 estimates were available of the total population and its distribution by location. The ,38 age-sex composition within each location for 1974 was assumed to be the same as 1963. The component method whereby the population can be projected by age-sex from 1963 was not feasible as the available data on birth and death rate by location are extremely fragmentary and con- tradictory. It is worth examining the changes in the pattern of distribution of population between rural and urban areas in 1963 and 1974. Table 1 shows the population distribution between rural and urban areas in 1963 and 1974. In 1963, 77.3 percent of the p0pu1ation was rural. Within the urban areas, the small urban areas had a higher percentage of the population than the large urban areas. In 1974, 73.0 percent of the population was rural, showing the relative decline of the population in rural areas. The remaining 27.0 percent was more or less distributed evenly between the small and large urban areas, indicating the importance of large urban areas in 1974 compared to 1963. Although the population as a whole increased by 2.3 percent per annum during the period 1963-1974, the rate of growth of the rural popu- lation is only 1.8 percent, reflecting the out-migration of population from rural areas. This contrasts with the rate of growth of the urban population where the small urban areas grew at 3.3 percent and the large urban areas at 6.9 percent. Allowing for the natural rate of growth of 2.2 percent, this yields a growth rate due to migration of 1.1 and 4.7 percent respectively for small urban and large urban areas. Table 2 shows the potential labor force (defined as popu1ation aged 10 to 64) as a proportion of total population in each location in 1963 and 1974. \This proportion is higher in urban areas, approximating 70 percent, but only 64 percent for rural areas, reflecting the greater 39 TABLE 1 POPULATION DISTRIBUTION BETWEEN RURAL AND URBAN AREAS IN 1963 AND 1974 1963 1974 Average Annual Po ITgiion Rate Of p Total Percent Total Percent Growth, 1963-1974 ('000) ('000) (%) Sierra Leone 2,180.3 100.0 2,733.1 100.0 2.3 Rural 1,685.6 77.3 1,994.9 73.0 1.8 Small Urban 272.0 12.5 361.8 13.3 3.3 Large Urban 222.7 10.2 376.4 13.7 6.9 Sources: 1963 Population census of Sierra Leone Central Statistics 0ffice,Estimates for 1974 burden of dependency in rural areas. It also reflects out-migration of younger people from rural areas. Labor Force The size of a population, its age-sex composition and locational distribution combined with the participation rates specific for each age- sex and location group are the primary determinants of the size of labor force available to the economic sectors and to each location. 'The labor force participation rates used for urban areas were obtained from 1L0 (1971). These are based to a large extent on compara- tive analysis of labor force structure in different countries at differ- ent stages of economic development. These labor force participation rates are shown in Table 3. The rates for males 20-64 years are on the average 90 percent, while for females they are about half of that for 40 TABLE 2 POTENTIAL LABOR FORCE1 AS A PROPORTION OF POPULATION IN EACH LOCATION, 1963 AND 1974 Location 1963 1974 . (percentage) —(percentage) Sierra Leone 64.8 66.4 Rural 63.6 64.7 Small Urban . 68.6 ' 69.6 Large Urban 70.9 71.8 1Population between the ages of 10 thru 64. Sources: 1963 Population census of Sierra Leone Central Statistics 0ffice,Estimates for 1974 males. Since the concept of labor force participation rates as it usually is defined does not have much meaning in rural areas, it is simply assumed that all males are 20 to 65 and 90 percent of the females in that age group participate in the rural labor force. It is instructive to compare these labor force participation rates with labor force participation rates for migrants in urban areas observed by Byerlee, Tommy and Fatoo (1976). Overall, the labor force participa- tion rates for migrants were consistent with the labor force participa- tion rates of the urban population as a whole. Male migrants aged 25+ had on the average 90 percent participation rates; this is identical with the rates of males in that age group in urban population as a whole. For female migrants aged 25+, uneducated had lower (28 percent), while educated had higher (52 percent) participation rates, compared to an average of 45 percent for urban female population in that age group. 41 TABLE 3 LABOR FORCE PARTICIPATION RATES BY AGE AND SEX FOR URBAN POPULATION, 1974 Age Males Females (percent) ‘Tpercentl' 10-14 21.3 16.0 15—19 56.4 33.5 20-24 85.6 43.5 25-44 96.9 47.6 45-54 . 96.0 49.7 55-64 86.5 39.5 Source: ILO (1971, p. 117). Based on these analyses of total population, its age-sex structure, its diStribution by location and the activity rates specific for age-sex and location, the labor force available in each location in 1974 is esti- mated and shown in Table 4. About three-quarters of the labor force is in rural areas, the remaining quarter divided more or less equally between the two urban locations. Females comprise 40 percent of the rural labor force while in urban areas they form slightly less than one- third of the labor force. This is partly the reflection of the activity rates assumed in the computation. Employment According to estimates prepared by the Central Planning Unit, the total labor force increased from 927,000 in 1962 to 1,094,000 in 1972 or at an average annual rate of growth of 1.7 percent during the period. 42 TABLE 4 LABOR FORCE DISTRIBUTION IN 1974 BY LOCATION AND SEX Sierra Small Large Leone Rural Urban Urban (in thousands) Total 1,287.46 1,021.46 126.00 140.00 Male 729.81 539.15 86.53 100.21 Female 557.65 482.31 39.47 ' 39.78 Source: Estimates based on applying 1L0 (1971) participation rates to Sierra Leone population data. Only about 149,000 or 89.2 percent of the 167,000 additional workers were able to find employment. From the sectoral distribution of the employment shown in Table 5, it is evident that agriculture absorbed the largest share (56.0 percent) of the new entrants to the labor force. Construction, commerce, trans- port and public administration provide most of the remaining employment. Most of these increases in employment were in the small-scale sec- tors. With the exception of utilities and mining, contribution of employment creation by large-scale sectors was slight. 0n the average, large-scale sectors accounted for only 7.9 percent of the total increase in employment during the period 1962-1972. Time series data available for wage employment in large-scale sec- tors are shown in Table 6. The average annual rate of increase in employment in large-scale sectors was 2.2 percent during the period 1962- 1972 and is largely a reflection of accelerated growth during the period 1962-1965. Employment in large-scale sectors since 1966 has actually declined from 67,692 (in 1968) to 65,728 (in 1972). This decline in 43 TABLE 5 EMPLOYMENT INCREASE IN LARGE-SCALE SECTORS COMPARED TO TOTAL INCREASE IN EMPLOYMENT, 1962-1972 Total Increase Emngsmzfitjgn in Employment, Large-Scale Sectors, (In Thousands) (In Thousands)————- Agriculture, forestry 83 1.1 hunting and fishing Mining and quarrying -5 1.2 Manufacturing 10 2.6 Construction 9 -2.7 Electricity, water and l 0.6 sanitary services Commerce 24 1.2 Transport, storage and 10 2.0 communications Services 17 5.8 All sectors 149 11.8 1Establishments with six or more workers. Sources: Bank of Sierra Leone, Economic Review . Ministry of Development and’Economic Planning employment in large-scale sectors, in spite of increase in output, is due to a productivity increase in the large-scale sectors. The increase in productivity can be attributed to a number of factors. New investment can be capital-intensive in response to various market imperfections which encourage capital-labor substitution. Productivity increases can also be due to on-the-job training of both labor and management. 44 TABLE 6 1 WAGE EMPLOYMENT IN LARGE-SCALE SECTORS Year Wage Employment Annual Rate of Increase (At the End of the Year) In Percentage 1962 53,628 1963 58,146 8.4 1964 61,699 6.1 1965 67,692 9.7 1966 67.388 -O.4 1967 63,643 -5.6 1968 63,070 -0.9 1969 64,513 2.3 1970 64,315 -0.3 1971 65,318 1.6 1972 65,728 0.5 1Establishments with six or more workers, excluding government employment. Source: Bank of Sierra Leone, Economic Review (1972). Unemployment In assessing the magnitude of urban unemployment it should be kept in mind that the discussion in this section is on the visibly unemployed. These rates of unemployment do not include underemployment in the tradi- tional sectors of the urban areas. According to the survey of the Central Statistics Office (1967-1969) there is substantial unemployment in urban areas. These rates of unemployment shown in Table 7 indicate 45 TABLE 7 UNEMPLOYMENT IN SIERRA LEONE Percentage of Labor Force Location Visibly Unemployed Western Area Freetown 15.5 Other Urban 13.5 Southern Province Urban (1968) 10.1 Bo (1968) 15.1 Northern Province Urban (1968) 11.0 Eastern Province Urban 9.5 Source: Central Statistics Office (1967-1969). variation in unemployment rates ranging from 9.5 percent in urban areas of the Eastern Province to 15.5 percent in Freetown. There are no comprehensive statistics which show the trend of unemployment over the last decade. The time-series which exist cover only job-seekers registered at employment exchanges. These job seekers constitute only a fraction of the total number. There is sometimes a relationship between the unemployed who register at the exchange with the unemployment rate. The higher the unemployment rate, the fewer the chances of finding work, therefore, fewer persons register. Hence the number of job seekers registered is not a safe indicator of the total 46 TABLE 8 REGISTERED UNEMPLOYED BY YEARS, 1962-1973 Year Registered Unemployed 1962 9,006 1963 8,509 1964 11,604 1965 12,315 1966 . 13,632 1967 14,704 1968 14,603 1969 15,502 1970 14,156 1971 13,483 1972 12,839 1973 12,122 Source: National Accounts of Sierra Leone number of unemployed and should be interpreted with caution. The number of registered unemployed shown in Table 8 indicates that unemployment has increased over time. Both demand and supply conditions have contributed to the emergence of the urban unemployment problem. Demand for labor in the modern large-scale sector has either stagnated or increased very slowly, pri- marily due to adoption of capital-intensive technologies. However the supply of labor in urban areas has.increased rapidly, partly due to high natural rates of population growth and largely because of rural-urban migration. 47 Summary and Policy Issues In this chapter some of the features of the Sierra Leone economy relevant to this studywere described. The review has shown that the Sierra Leone economy has much in common with other developing countries. Agriculture is the dominant sector of the economy. However, the growth rate in the agricultural sector has lagged far behind that of the rest of the economy. This disparity in growth between the agricultural and nonagricultural sectors is reflected in a disparity between development in rural and urban areas. Like most developing countries, urban unemployment rates are high and increasing, while high rates of rural urban migration continue to aggravate the problem. Greater awareness of the economic and social problems created by rural-urban migration and unemployment has been shown by the Sierra Leone government. The general objectives of the employment policy of the Sierra Leone National DevelOpment Plan, 1974/75- 1978/79 are (l) to accelerate the growth of productive employment, and (2) to reduce unemployment. The development strategy of the plan con- tains several elements stimulating labor-intensive production and encour- aging fuller utilization of human resources. Many policies and programs to achieve these objectives are contained in the development plan. Among these are to increase the overall rate of growth of the agricul- tural sector from 1.7 to 5.4 percent per annum. Agricultural develop- ment is expected to serve employment objectives in several ways. First, agriculture is the most labor-intensive of all the sectors and has a potential for labor absoption. Secondly, the increase in rural income is expected to slow rural-urban migration and consequently decrease urban unemployment. It is also possible that the increase in rural income 48 will increase demand for labor-intensive products, thereby stimulating total employment. Within the agricultural sector, the goal is to increase rice production. Export crap production is also expected to be increased to provide another source of export. Small-scale indus- tries will be encouraged in order to increase employment. There is, thus, a need to analyze the impact of these policies on output, employ- ment and migration at a macro level. IV. ECONOMETRIC ANALYSIS OF RURAL-URBAN MIGRATION RATES1 The objectives of this chapter are two-fold. The first is to esti- mate quantitatively the magnitude of various factors affecting migration. These elasticities of migration will be used in a migration model to forecast the distribution of the labor force between rural and urban areas. The second objective is to test for significant differences between the behavior of educated and uneducated migrants. If the response of these two groups is found to be significantly different, this will reinforce the argument for disaggregating the labor market by educational level. In this chapter, previous econometric studies of migration are cri- tically reviewed first. Based on the review, a migration function which avoids earlier deficiencies is presented in the second section. Discussion of the data used and estimation procedures are presented in the third section. In section four empirical results are discussed and these are used as a basis for policy implications in the final section. Review of Ecbnometric Studies Econometric analysis of migration rates is now standard part of research on migration by economists. Most of the studies are concerned with the response of migration to economic variables, and the framework 1This chapter is based on a paper by Byerlee, Tommy and Fatoo (1976) "Rural Urban Migration in Sierra Leone: Determinants and Policy Implica- tions," African Rural Economy Paper No. 13, Dept. of Agricultural Econo- mics, Michigan State University. For details about the characteristics of migrants and migration process, consu1t this paper. 49 50 of the model is based on human capital investment approach or its derivative. However, several problems are inherent in past analysis of this type in developing countries. Most studies on migration have had to rely on census data, restricting the specification of the model by the use of birth place data instead of place-to—place migration flows. In these studies (e.g., Beals, Levy and Moses (1967), Sahota (1968), Adams (1969) and Greenwood (1969)), migration data employed refers to cumulative lifetime migration from one region within a country to another, i.e., the number of persons born in region i and enumerated in region j. The use of such data may result in simultaneity bias in the estimates of the coefficients, since migration which has occurred over a long period of time is likely to have influenced the independent varia- bles such as wage rates employed in the regression models. Moreover, it is questionable to relate migration which has occurred over a longer period of time to variables defined at present time. Second, most analyses of migration have focused on interregional migration. Interregional migration includes besides rural to urban migration, rural to rural, urban to rural and urban to urban. As such, these studies (e.g., Beals, Levy and Moses (1967), Mabogunje (1970)) do not give reliable estimates of response of rural-urban migration to various factors. Third, although numerous studies of migration in Africa have iden- tified economic motives as dominant hithe decision to migrate, they have suffered'h1the measurement of income. Most of them have used secondary data or proxies for income such as regional per capita income (e.g., Sabot (1975)) or even per capita food production (e.g., Levi (1972)). Sabot (1976), Essang and Mabawonku (1974) and Rempel (1971) have 51 carefully measured urban incomes, though none has measured incomes of rural households from which migrants originate. In this study, rural wages are obtained from a sample of 16,000 rural wage observations obtained in a farm management survey by Spencer and Byerlee (1976). Finally, for rural-urban migration, various studies in developing countries indicate that education has a significant effect on migration, but it has not been possible to provide consistent interpretation of the observed relationships. Beals, Levy and Moses (1967), Greenwood (1969, 1971), Sahota (1968) and Schultz (1971) used regression analysis to estimate labor force migration in Ghana, Egypt, India, Brazil and Colombia respectively. The education level of the migrants could not be determined in these studies, but the education levels of the origin and destination regions were included as explanatory variables in order to examine the relationship between education and migration. One of the problems with this procedure is that it constrains the level of precision at which we can analyze the determinants of migration. The estimated regression constrains the coefficients of the independent variables to- be the same for each education subgroup. As pointed out by Barnum and Sabot (1975) and Levy and Wadycki (1974), a significant association between regional average educational levels and migration rates is not sufficient to confirm that the educated have a higher propensity to move than the uneducated. The estimated coefficients of the education variables may have captured a number of effects, including the effect of education on an individual's willingness to migrate as well as the attraction of educational opportunities for potential migrants. Even if it is established that the educated have a higher migration propen- sity there is no way to determine whether this is due predominantly to 52 higher level of responsiveness to a given rural-urban income differen- tial or to a wider income differential for the educated than the unedu- cated. It is not surprising that the estimated effects of education on migration have differed markedly among studies of migration in differ- ent countries, despite the remarkable similarity in the estimated responses of migrants to such factors as regional income and urbanization levels. Greenwood (1969, 1971) and Sahota (1968) found that migration decreased with higher levels of education at the origin and increased with higher levels of education at the destination. Beals (1967) found that migration decreased with higher levels of education at both the origin and destination, while Greenwood (1971) found that migration increased with higher levels of education at both the origin and destin- ation regions. Very few studies besides Levy and Wadycki (1974) and 4 Barnum and Sabot (1975) have disaggregated the population by education and tested whether or not these structural differences are statistically significant. Levy and Wadycki (1974) found significant difference in the urban income elasticity between migrants who have had a secondary education and those who did not have any primary education. The income elasticity for educated group was higher. Barnum and Sabot (1975) did not find any significant difference in the expected rural urban wage differential elasticity for educational categories. However, Barnum and Sabots' results should be interpreted with caution as they did not exclude those 1 who had migrated to attend school or were apprentices. Given the edu- cational system in Tanzania, students from rural areas are very likely 1Barnum and Sabot (1976) used as their dependent variable men born in the country who came to town after the age of 13. 53 to go to regional capital or regional urban areas rather than to urban areas with higher income per se. This is because students have little choice as to which urban area school they can attend and are directed by the ministry 0f education. Even if students were free to choose urban destination, variables such as the location and quality of schools proba- bly are more important. Levy and Wadycki (1974) included education-specific wage rates but the study suffered from the use of nonspecific unemployment rates. If there is a separation of the labor market in the urban areas, a single unemployment rate is inadequate. In the case of Sierra Leone, the urban destination unemployment rate for the educated is higher than for the uneducated, in which case a single average unemployment rate in urban destinations would understate the unemployment for the educated and overstate for the uneducated. These varying urban unemployment rates by education were observed in Tanzania by Barnum and Sabot (1975). In this study some of these deficiencies in earlier analyses are overcome through data collected specifically for the purpose of analyzing migration rates. The survey data were used to compute education-specific rates of migration for the last five years. Furthermore, in analyzing migration rates students are specifically excluded for reasons explained earlier. The function is disaggregated by two educational subgroups using education-specific urban wage and unemployment rates. To test for the significance of the difference between corresponding parameter estimates in regressions, observations for the two groups will be pooled. 54 The Migration Function The migration function is based on the theory of investment in human capital discussed in Chapter II. Rural-urban migration is viewed within a framework of costs and returns of investment in human capital. Costs are comprised of money costs and psychic costs. Money costs include costs of transportation, increased expenditures on food and lodging during the period spent on traveling and in searching for a new job. Psychic costs are costs such as homesickness, acclaimatization, strain and so on. Since these costs are likely to vary with miles tra- veled, distance is used as a proxy. Also, distance is likely to be a factor in determining available information. The opportunity cost of migration is the income foregone in the origin. The economic return is the income the rural resident expects to receive in the urban area. These economic costs and returns should be discounted. Since precise informa- tion on time horizons, discount rates and changes in income are not available, migration rate is related to the current income in origin and destination areas. The expected economic returns to migration cannot be estimated on the basis of the income of those employed in urban destinations in situations where there are high levels of unemployment. In such a situa- tion a potential migrant cannot be sure of finding a job, and unemploy- ment has to be taken explicitly into the migration decision (Todaro, 1969). The size of the urban area is included to represent a number of factors such as a larger labor market and urban amenities (i.e., "bright lights") which influence economic components of the costs and returns. 55 Education and migration appear to be complementary human capital invest- ments. One of the objectives of this chapter is to empirically analyze education-migration relationships and therefore the function is disaggre- gated by educational level. The rural-urban migration function is then given by: Mijk = f (Ni, ij, U P. D.., e) J'k’ J’ 13 average annual gross rate of adult migration for the kth educational cohort from rural origin i to urban destination 3 W. = average monthly income of adult males in rural region i where Mijk ij, U.k = average monthly income and percentage unemployed respec- J tively for the kth educational cohort of male migrants in the j h urban center Pj = population size of the jth urban area Di' = the road distance in miles between the main center of rural 3 region i to urban center j e = random error and i = 1, 2, . . .8, corresponding to the eight rural resource regions of figure 3 j = 1, 2, . . .5, corresponding to the five urban centers above 20,000 population--Freetown, Kono, Bo, Kenema and Makeni x ll 1, 2, representing two educational cohorts--less than four years education and four or more years education. Some comments on the specification of the function are in order. The measure of rural income used here is wage rate rather than household income. This measure of rural income was chosen because (a) it was shown that an active and competitive rural labor market exists (Byerlee and Spencer, 1976), and (b) given this competitive market and dominance of household rather than individual decision making, this wage rate should be a close approximation of the value of marginal product (VMP) of labor 56 1 '1'. ”- "' j | 1 I I 1 ' I 1 . r.‘- I 1 . : ‘,......................\I O _. ; 1 1 .. -' "\.~.’."’ 1. I z I I f, Soll‘o'mq ‘1‘" C 1 I 1 1 _I '\_ I I 3 I I \ ’ 1' l i 1‘ "1 ' 1 Chum , I 1 1 [I " 7 \. 1 I ' 1 ~. .. k: 1 I I ) “it.“ \n.’.. ‘3 l-.\ f“ 1 rk?’ l ‘“““* '3 I i I\ \...\ S I" .1 7 I I l .—- \‘( ole-5n ’ ., (_ l ' .~_/ Jon-50‘: './ F.5WI 7 -. (I. I I ; ‘0'.__ 2"“, —. \' ‘3 -. rézuowi 50-. .'/ v." ['r’.‘ 1 .1 (II-161w K0. by". 1 0’ \""" 1 ._: .1 ' . ' mum'- ./ 1 1 1 ”m l \ um Lou / '41:}. {/7 ’ -‘ I 1 '. /‘ M 'o . I ' I Lou-:0! '. “WT-n : I I 1 3‘ 1.\ ’3’ I ololdw ,-/) I I ' \ I JM‘T‘H ’. 1 \ Cf 3 ‘8 f” 1'1 1' r '1 ’1’. _1 1 l V" .- u .‘o o. . ' 00 J I 1 ~°'\ 5 ’ "/ .\ \“A”, -"i\ / 1'. ‘ 1 w . ‘ I. \ '. d I 3 ( ”Hunt? \'. \o. ‘ lab ,' ) I ' I. \. 5". Fourth I.’/-'". 1 I l 8 1 9' ' !. 2 'm 1 "' : 0m“. ' E 1 .1 "A 1 ./ Supt-u“ | —I , ' ‘3 ‘I ”‘3'" °lnoll l 1 \ ‘ u‘ loan-Na 1w ‘ \ J- \ *1 I v'- * . : .’ 9 $455.: M, r ~ w- 6 .1 c . Q Not "/"- ).' [-I Q I on. /' ASL,I~, . _/ \ \0‘ W ‘1 ' ' " ‘I w. 'Ijo.~un J I. [I \' 1 u ,- e" 1 r—fi I ‘Jfiu‘ \ -\ I learnt lulu-o1 huge”... 9......”1 ' Sufi-$3.1 ava" >————-J 1 I -o to )o u , \ u r . .. , ._.__ ' I -——§.- 3; '“--OL1'O_ to... . M. '. ‘ \ but “loan: Ma y r—__— I 1 y >— ; ' -;1 .=.. ’1 (“ouch- Ann... . .. .I A: I ; I 0 u u . ' . 1 . . . . .0 ”h- . . ‘ ‘5ch liq-u. ........ .4 a 1 I 1 “3 I I when Ann I 0|- I 1 I .10 u' 1. 9' Sierra Leone FIGURE 3 RURAL ENUMERATION AREAS AND URBAN AREAS OF THE MIGRATION SURVEY 57 (Knight, 1972).1 Furthermore, since females have a low participation rate in the urban labor market, male wage rates were used. However, the same rural wage rate was used for both educational cohorts on the assump- tion that educated persons receive the same wage rate in traditional farming activities as those without education. Though the model is formulated in terms of variables whibh are more relevant to male migrants, who comprise most of the labor force, the migration rates include both males and females. The most important reason for female migration is marriage to a male migrant usually from the same rural area, female migration is correlated with male migration. To determine the relationship between male and female migration, a correlation coefficient was computed. The coefficient between male and female migration from specific rural origin to specific urban destina- tion was 0.78 for uneducated migrants and 0.87 for educated migrants. For these reasons, the model is formulated in terms of variables which are more relevant to male migrants. Since persons in the labor force provide an economic base for other nonworking migrants, particularly housewives from the same rural area, the model is used to explain both male and female migration. Qata_ All data with the exception of urban unemployment and urban size were obtained from a migration survey. Although urban unemployment data were available, the sample was too small to estimate education-specific unemployment rates for the medium size towns of 80, Makeni and Kenema. 1In the case of individual decision making, the relevant income is the value of the average product if income is shared among household members. 58 Unemployment data were derived from the urban household survey of the Central Office of Statistics (1967-1971) which were shown by Byerlee, Tommy and Fatoo (1976) to be highly consistent with unemployment data from the migration survey. Also, the sample size prevented reliable income estimates for the small towns (less than 20,000 persons) and these towns were exluded from the analysis. Migration Rates Migration rates can be expressed as gross migration or net migra- tion. Net migration is the difference between out-migration and in- migration. Net migration rates are indicators of rural out-migration or urban in-migration. Where the characteristics of in-migrants differ from out-migrants, net migration rates are less meaningful. In a situation where the rural out-migrants are dominated by young and educated while the rural in-migrants are older persons, gross rural out—migration rates are better indicators of those entering the urban labor force. A corre-’ lation coefficient was computed to determine the extent to which varia- tions in net migration are the results of variations in gross rural out-migration or variations in gross rural in-migration. The correlation coefficient between net migration and gross rural out-migration is 0.89 compared to -O.l4 between net migration and gross rural in-migration, indicating that the large prbportion of variation in net migration is due to variations in gross rural out-migration. Hence, gross rates of rural out-migration are used.1 1For a discussion of the information loss involved in models of net migration as compared to models of gross migration see Sjaastad (1962) and VanderKamp (1972). 59 An added advantage of using gross rural out-migration rates is that they are more reliable than net migration rates. In computing net migra- tion rates, residual error is compounded due to errors in estimating rural-to-urban migration and urban-to-rural migration rates. Gross rates of adult out-migration from rural region i to urban destination j, specific for education group, are computed using the following equation: 111.. M.. =41! x 1,000 13k Nik where mijk is the number of adults in the kth education cohort migrating from origin i to destination j and Nik is the number of people in the kth education cohort in the origin i population. These rates are shown in Table 9. The table shows that the educated persons have consistently higher propensity to migrate than those without education. Rural and Urban Wage Rates Rural wage rates used are from the wage rates reported in a farm management survey by Spencer and Byerlee (1976). The hourly wage rates were multiplied by the average number of hours worked per month by an adult male. These wage rates per month are shown in Table 10. Urban wage rates were computed by destination, specific for each education group and are shown in Table 11. Comparison of these wages between the education groups shows that educated migrants in urban areas consistently earn higher wages than uneducated migrants, except in Kono. Urban Unemployment Rates Urban unemployment rates are shown in Table 12. The unemployment rate for the educated in urban areas is consistently higher than for the 60 TABLE 9 AVERAGE ANNUAL GROSS RATES OF ADULT OUT-MIGRATION FROM RURAL T0 URBAN AREAS BY EDUCATIONAL LEVEL1 . Urban Destinations Rural Origin Region EdggszIon Freetown Kono Makeni Kenema Scarcies Uneducatgd2 4.5 .8 .2 0 Educated 20.0 0 O 0 Southern Coast Uneducated .9 1.2 0 .4 Educated 19.2 2.7 0 8.2 Northern Plains Uneducated (3.3 5.2 1.2 .7 Educated 51.3 51.3 20.5 0 Riverain Grasslands Uneducated .6 .5 0 .6 .5 Educated 11.3 11.3 0 2.8 .3 Bolilands Uneducated 16.2 2.9 1.6 .6 Educated 37.8 0 5.4 0 Moa Basin Uneducated .4 2.2 .2 3.0 Educated 15.8 17.1 1.3 17.1 Northern Plateau Uneducated 1.7 7.9 0 3 Educated 12.9 12.9 6.5 0 Southern Plains Uneducated .8 4.6 O 1.3 Educated 34.7 29.2 2.8 19.4 1Rates per thousand of population. 2Uneducated are those with less than four years of education. 3Educated are those with four or more years of education. Source: Migration survey 61 TABLE 10 RURAL WAGE RATES BY REGION Rural Region (Leonewgg: Month) Scarcies 14.03 Southern Coast 9.82 Northern Plains 9.60 Riverain 7.52 Bolilands 7.61 Moa Basin 7.32 Northern Plateau 10.53 Southern Plains 12.82 Source: Spencer and Byerlee (1976). TABLE 11 URBAN WAGE RATES BY EDUCATIONAL LEVEL . Wage Urban Destination (Leone per Month) Uneducated Educated Freetown 43.27 73.83 Kono 80.28 68.35 Makeni 52.00 62.50. Kenema 44.22 54.38 80 41.27 50.26 Average Urban Wage 48.17 65.74 Source: Migration survey 62 TABLE 12 RATES OF URBAN UNEMPLOYMENT BY EDUCATION Unemployment Rates Urban Destination Uneducated Educated Freetown 14. 4 l7 . 8 Kono 12.5 16.9 Makeni 7.7 17.7 Kenema 6.3 17.1 Bo _ 20.6 20.6 Source: Central Statistics Office, Household Survey(197l). uneducated. The unemployment rate for the educated does not vary as much with destination as it does for the uneducated. Estimation Procedures and Empirical Results The estimation procedure employedeas ordinary least squares regres- sion. .10 test if any significant difference exists between the behavior of educated and uneducated migrants, data for both were pooled and the following linear relationship was fitted: M = b + b E + D W. + b EW. + b W. ijk o 1 2 3 4 jk + bSijk + b6Ujk + b7EUjk + bBPj + ngPj + b1001j+ b11EDij + where all variables except E are as defined previously. E is a dummy variable for education such that E = 0 for an observation on uneducated 63 migration and E = l for educated migration. Consider now the coefficient of "i and.EWi. The coefficient b2 of ”i indicates the influence of wage in rural area on the uneducated migrant, while the sum of the coeffi- cients of W]. and EWi (i.e. b2 + b3) indicates the:inf1uence of rural wage on the educated migrant. The coefficient of EWi (i.e. b3) indicates whether b2 and (b2 + b3) differ significantly. In other words, b3 indi— cates whether or not the influence of rural wage differs significantly for the uneducated migrant as compared to educated migrants. Table 13 contains the estimated relationships for rural-urban migra- tion by educational subgroups. The first figure below each coefficient is the "t" statistic, while the second figure is the elasticity cal- culated at the mean value of the variables. Up to four equations are reported for each group. First, there is the standard linear form on all variables in the model. In the case of educated migrant, however, strong multicollinearity exists between urban size, Pj, and urban wages, ij. Therefore, a second run was made in which urban size was dropped. A more relevant measure of urban wages is the expected wage which takes into account the probability that the migrant will be unemployed in the urban destination. That is, the expected wage is computed as ng = (l - Ujk) ij where Ujk and ij are the unemployment rate and aver- age wage rate respectively for kth education group in destination j. Accordingly, the unemployed variable and wage variable were incorporated into an expected wage variable--W§k. Finally, the expected wage differ- ential (ng - Hi), which is the difference between the expected urban wage I and the rural wage was used. The expected wage differential takes into account not only the difference between rural and urban income but also the probability of finding an urban job. 64 .cc_o A_oe.ov Ae_m.ov A_em.ov emu: w~_oo.- ammoo. ~me.. eeeeeaxw can: empoc.- empoo. m_hoo. _o~_.- «5389.- eeeeeaxw tonnage eNem o~_oo.- “Napo. cameo. ame~_.- mmmp_.- cent: ee.oo.- ¢_Foo. ammoo. omo~_.- _oeeo. uceeeeem empooa _.o._- mNP.m .cc_o Amoa.~v “www.me Aoom.pv can: eo__o.- .m_o_. some..- ueeeeaxm mwo._- eme.o P_~.m moo.o- Ammp.mv Aeoe.Pv Ammo._v Amo_.ov Aemm.ov cue: .eoo_o.- Npmoo. .Pomeo. moem_.- memm._- caeeeaxm mam.o- oeo.~ ~me.¢ aeo.o- emanate Amom.mv Aemm.ov Aoe~.ev Awop.ov ANom.PV e~_m .memoo.- cemmp. rapmmo. Nemep.- emmo.m- cent: ooo.F- wmm.o emw..- moo.o- Aepo.m3 Aoom.ev Aeee.ov Ame..ov Am-.,v remmoo.- .Pmaoo. NmooP.- mmmmp.- mmo~.m ecceecem caboose“ m_m._- ane.o me~.P .cceo Aemm.mv Aoe~.ev Ammm.~v Aepo.ov one: .mepoo.- .ep_oo. .Nmmoo. memo. neeeeaxu omm.F- mom.o ~m~.~ cam.o- “mew.mv Aeee.ev Aooe.mv A_mm.ov Amem.ov can: .ee_oo.- re~_oo. .w_eoo. meeo_.- .memmo.- eeeeeaxu ~mm..- aom.o wep.o- mem.~ mam.o- Ammo.me Amom.mv Am-.ov Aoe~.ee Aopo.ov Aeo~.ov .oo_oo.- .a_Poo. mm_oo.- .eoooo. ooeo_.- opmm.- uceeecem eeeeeeeees Ha a ..am~o seesaw m wu<3 eeem_a aaoa “mm: qazmzn mo<3 euozmo chowumz .mcowo mggmwm we ucmscgm>oo mmugoom xumocoowm ocm Aouopv mmpgmao can gmucmam .Aosopv muago ocm spagomwo mmogsom xgmswgo moopucp mmmgh .mmogoom maowgm> seem owuuogumcoo ”mmugoom .n sumooesou mo «we: use mo comuuooogo com convocmg p zumooseou mo ucoosm on» mzozm open» as» cm .Wmm .ucmwowommou sumo . H o. mop. Nuo. o. omo. mop. moo. ooo. ooo. ooo. o. mop. o. o. mm0w>cmm o momgh mpmum-mosmo .ep coo. o. opo. moo. eoo. coo. moo. Npo. Npo. o. o. o. o. o. mmmuwpwpo .m_ omo. omo. o. opo. ooo. woo. NNo. oeo. omo. o. o. o. o. o. ocwgzuummocaz mpmumumogmo .Np o. o. o. o. o. o. o. o. o. o. o. o. o. o. ocwcwz .—~ Fpo. ono. omo. m_o. o. o. o. o. o. o. o. o. o. o. cowuuzgumcoo .op o. o. muo. o. o. o. NNF. omm. omm. omm. mmo. ono. moo. omo. awash mpmomupmem .m “No. moo. Poo. omo. moo. oop. o. omp. ooF. omp. Poo. moo. omo. oeo. coopmpgoomcmgp .o o. o. Poo. o. woo. woo. o. o. o. o. o. o. o. o. cmngo mocmo ..o:H upmum-_FmEm .m o. o. moo. o. «oo. Foo. o. o. o. o. Poo. poo. poo. Poo. cone: __mEm ..o:H m—mumu_Fmsm .o o. o. mpo. o. opo. moo. o. o. o. o. o_o. opo. oFo. opo. Pagom ..o:H mpmomuppmsm .m o. o. moo. o. o. o. o. ooo. Koo. Npo. o. o. o. o. .owgo< .moowmmm .e o. o. o. o. o. o. o. o. o. o. o. o. o. o. moogo ugooxu .m o. o. mmo. o. o. o. o. ooo. omo. ooo. o. o. o. o. moogo coon Lmspo .N o. o. moo. o. o. o. o. ooo. omo. ooo. o. o. o. o. muwo .F P mgopumm «F mp ~_ FF op o o n o m e m N P cowpuaoogo n mgouumm .quomcoz upcomlmogmo. .uwgomooz mpmumuppuEm. mg=u~30wso< cowumcwummo Houop .>zozoou.uzom4 <¢mmHm mzh do mhzmoooaamou hoahno-»=azH mp u4mwQ Pmcowpmz .mcomg mLwam $0 HemECLm>ow ocm AcNo—v .m .cOmmmmm Eocm omuoogumcoo "mmugoom .ummscmz oom. mop. “mm. omm. ocm. omc. Pop. coc. occ. omm. mum. onm. mum. ohm. w—mumumogmo ooc. cum. mom. ch. oop. cmc. ocm. omc. ch. com. cpc. cpc. cpc. mpc. compuogumcoo cp .. mp N— PF o_ o o N o m c m N F Paupaau msoupaoweomcoz «Foam mogmo mgoppoumgomcoz mpmum-~_msm mgoppouwgo< mo cwopgo mo gopumm mgouumm cowumcwummo cnop .xumh Investment, Consumption, Etc. Labor Force Migration 1 Between Rural and Urban Areas Population Distribution Between Rural and 1 Urban Areas FIGURE 5 LINKAGES BETWEEN MACRO AND MIGRATION MODEL Summary and Model Limitations In constrast to conventional macro models, the models have a higher degree of disaggregation in both the product and factor markets and take explicitly into account interactions in both the product and factor markets. This emphasis on intra-sectoral and inter-sectoral relation- ships as they affect output, employment and migration adds strength to the results. The macro and migration models are useful in analyzing output, employment and migration at macro level. They are also useful for sector- specific policy analysis, because they can run simultaneously with detailed sector models and sector specific policies can be analyzed within 105 TABLE 26 TYPES OF VARIABLES TRANSFERRED BETWEEN THE MACRO MODEL AND MIGRATION MODEL Variables Transferred Variables Transferred From Migration Model ' From Macro Model to To Macro Model Migration Model 1. Number of migrants from rural l. Wages in rural areas. region to urban areas by size of urban areas and education 2. Wages in small-scale sectors level. by urban size and education level. 3. Employment by urban size, education and scale of operation. a broader macro framework. However the models running independently of sector models have considerable value in analyzing output, employment and migration at the macro level. The macro economic model is also useful in identifying bottlenecks in the economy. Although the models are applied to Sierra Leone they are of general applicability to other developing economies. The degree of diSaggrega- tion determines the data requirements of the models. In case of the macro model corresponding to each component is a key set of parameters e.g. input-output coefficients of the production component and income elasticities of demand of the consumption component. An issue frequently raised in connection with the use of models is whether the information available in developing countries is sufficient in quantity and quality to justify the use of models. In the case of Sierra Leone a comprehensive 106 set of aggregated data generated from field surveys were available for the small scale sectors. Data for the large scale sectors were available from secondary sources and were fairly reliable. In other developing countries input-output tables are becoming increasingly available in developing countries. Similarly, data for other components will not usually be a limiting factor, although disaggregation by rural and urban regions may not always be possible. Such a modelling does demonstrate the need for this type of information and may lead to increased efforts to collect data that might otherwise be ignored. Some of the most important limitations of the models should be kept in mind. First, there are no prices in the model. The macro model essen- tially allocates real resources optimally with respect to a given objec- tive function. The optimizing process under linear programming system generates shadow prices which are resource costs in terms of the objective function. There are no market prices. Shadow prices usually differ from real prices not only because of imperfections in the market that are not included in the model, such as monopolies and government regulations but also because shadow prices reflect the models' structure. However some sort of pricing is still needed to give relative weights to sectors and for aggregation. Hence though the market prices have no role, all goods and services are expressed in value terms instead of real quantities. Secondly, the production functions are homogenous of the first degree i.e. they show constant returns to scale and have fixed input coefficients. For short-run projection where the technical coefficients are not expected to change, this may not be a severe limitation. The problem of changing technical coefficients can be handled within the eXisting model by means of statistical revision of the input-output table. 107 In addition, all production relationships are accounted for directly in inputs and outputs, so there are no external economies or diseconomies. These limitations should be kept in mind when assessing the empirical results. VI. MODEL SOLUTIONS The models are deisgned to provide a framework within which to examine, quantitatively, the potential of the economy and impact of dif- ferent policies on output, employment and migration at the macro level. First the results of basic runs which are made for two time periods, 1974 and 1981 are presented. This is followed by policy runs. The objectives of the basic runs are: (I) to analyze the results of optimization and examine the output, employment and migration potential of the economy, and (II) to use the base run 1981 as a bench-mark for analyzing impli- cations of different policies on output, employment and migration. Base Run Projection, 1974-1981 The basic projection was run under a set of assumptions about the values of exogenous variables discussed in Chapter V and reflect recent historical trends. Government consumption and employment is exogenous. Exports and foreign exchange available from sources other than exports also are assumed exogenous. Total population is exogenous but the labor force distribution between rural and urban areas is determined endogenously in the migration model. The rate of migration in response to expected rural-urban wage differential was estimated for each educational subgroup in Chapter IV. The sectoral composition of employment by education is assumed constant. However, labor productivity changes are assumed in the 108 109 large-scale sectors. Wages in the large-scale sectors are exogenously determined and are assumed to increase over time. While wages in the small-scale nonagricultural sectors in urban areas and rural wage rate are related to productivity growth. These assumptions, together with the limitations, particularly the problem of changes in parameter values over time discussed in Chapter V, should be kept in mind. National Account Statistics Table 27 shows the level and growth rate of national accounts. GDP is projected to grow at an average annual rate of 4.7 percent, from Le 446.2 million in 1974 to Le 593.1 million in 1981. This projected rate of GDP is higher than the historical rate of 4.2 percent for the period 1964-1971. Assuming that the population grows at 2.6 percent per annum, the GDP per capita will grow from Le 163.2 in 1974 to Le 183.6 in 1981 or at an average annual rate of 1.8 percent. Table 28 shows the rate of growth of GDP by regions. Though the GDP for the country as a whole is projected to grow at 4.7 percent per annum, the rural area will be grow- ing at less than this average, while the urban areas will be growing at the same rate or higher than the national average. The growth rate of GDP in rural areas is projected at 4.4 percent compared to 5.4 percent for small-urban and 4.7 percent for large-urban. The projected distribution of GDP between the regions will change very little during the period. Investment is projected to grow at an average annual rate of 5.2 percent from Le 64.6 million in 1974 to Le 88.3 million in 1981. The share of investment in GDP is projected to increase from 14.5 percent to 14.9 percent during the projection period. Consumption is projected to grow at an average annual rate of 3.9 percent from Le 353.0 million in 1974 to Le 449.1 million, in 1981. 110 Ncm.ocp Ncm.oc~ Ncm.ocp Ncm.ocp mnommcp F.m Ncm.ocp ooc.m- mugooxm Ncm.oN_ Ncm.oN_ Ncm.onp Ncm.oNp NNo.mNP o.N Ncm.oNP mNN.pc~ masons“ mco.mo Fco.oo Ncm.oo Foo.oo NmN.Fo N.m FNm.oo www.co ucwsumm>=~ mmm.omc ocm.omc cco.occ NNm.mmc moo.ch o.m cmF.occ Npo.mmm coppoe=mcoo NNc.Noo ooN.mom moo.mom oc_.ooo Nmm.Nom N.c om_.mom mop.occ moo 6.8.55 E: 23.28 “a 8? 8.7. EB sopwco xgumocco cowposogo poopucNoF Powwomo mpmumumogmo xgpmooco xpw>wpooooco cowuoEoLo N or poop cNo— pcmcoosoo cowmgou cw mocwcgump upmum .owco< mugooxm m m are; pczouu< ommomguco m>wmcmucfi Loom; uppmsm ommwmgoco .uwgo< wows; .>mo m>HHwuu=oogo cowpoEogo . op o cowmgom cw mocwcnuwh . mpmom .uwgo< mucooxm m m N are; . Poop, cm P ommmmguca m>wmcmucm Loam; uppmsm ommmmgucH .Pouwgo< ammo“ .>o mesa ammo m com c com o cam N com P com F < c mmHomHmo m>Hhm moo 112 Assuming growth of population at 2.6 percent per annum, consumption per capita will increase from Le 129.2 in 1974 to Le 139.0 in 1981, repre- senting an average annual growth in per capita consumption at 1.0 percent. Since the growth rate of consumption is less than the rate of growth 6f GDP, the share of consumption in GDP is projected to decline from 79.1 percent to 75.7 percent during the period. The average annual growth rate of imports is projected at 2.9 per- cent from Le 141.2 million in 1974 to Le 170.3 million in 1981. The model projects the share of consumer goods imports in total imports to decline from 54.6 percent to 51.6 percent; while the share of intermediate goods inputs is projected to increase from 32.0 percent to 35.1 percent during the projection period. The share of investment goods imports is projected to remain stable at around 13.0 percent. . There are no competitive imports of rice and the model prefers to produce rice domestically rather than import due to foreign exchange constraint. All available foreign exchange is required for noncompeti- tive imports. Sectoral Level Results Value-Added. Value added by different sectors is shown in Table 29. The average annual rate of growth of the agricultural sector as a whole, in terms of value added, is projected at 3.9 percent, from Le 122.0 million in 1974 to Le 155.1 million in 1981. It is the slowest growing sector compared to other sectors. Due to this slow growth rate, the sectorls contribution towards total GDP is projected to decline from 27.3 percent to 26.1 percent. .owoom-m=~m> newscgm>oo mmoopucnp 113 NNo.Po— coo.oo— mNN.oo mo_.oop cco.oop o.m Noo.oo moo.oN moup>gmm ocm moose mmN.c cc~.c oop.c Pop.c oop.c o.m oop.c NmN.m mmwuwpwgo oco.mm moo.cm FNN.mm moc.mm oN_.mm o.c moo.mm cmo.mN ocwgapummacmz mNm.co mNm.co MNm.co MNm.co MNm.co P.m MNm.co NoN.oo ocwcmz ooo.Nm moo.Fm _c_.~m Npo.om Foo.Nm N.m coF.Fm NoN.NN cowuoacumcow oNP.ooN occ.cmN opm.NmN cmo.NmN coc.omN o.c mmo.NmN ooo.oo— «mmooz mpmumum on; moo.cc oom.mc ooo.mc ooo.mc mmo.mc o.c NoN.mc mom.Nm munch MNP.om ch.mm mnm.mm coo.Nm coo.mm P.m ocN.mm moo.oc ugoamcmgh ooo.o o_N.o omo.o an.o ooN.o F.m oNo.o oNo.c :mngoumogmo NNN.c cco.c ooN.c Noo.c ooo.c o.c oNo.c omc.m concoupmem mmm.m~ oom.mp NFP.oF moo.mp NNo.mF N.m on.mp oom.FF ngzm Poc.NN oco.oN Nmo.NN moN.NN ooo.NN ~.m moo.oN mNN.o_ .macm: m Geog—Em Nmo.NNp —oo.mNp Noo.NN_ ooo.MNF mnm.oN_ o.m on.mN— PNo.No «mace: mqum-F_mam o~m.oN opm.oN omN.oN ooc.oN on.oN N.c oNN.oN Nop.m_ .upgo< .owmmm NmN.FN NNN._N NmN.PN NmN.PN om—.mN m.N NmN.pN omo.o_ moogo ugooxm ooo.om ocN.oo occ.oo cco.~o on.om o.m mpm.oo N~_.mc nook Lafio Pop.mm Noo.Nm oNN.Nm cco.om NNN.Nm c.N NoN.mm onm.mc mumm ooo.cmp oNP.mmp oon.cmp ooN.moF Poo.cmp o.m omp.mm~ nmo.pNF mcoupoowgo< NNc.Noo ooN.mom .ooo.mom ocp.ooo Nmm.Nom N.c om~.mom mop.occ powuuaoogo comuosogo .n cw Poop cNmF cowmgom cw mscwcsuw» mpmom .umgo< mugooxm . mung mgovuwm ommmmguco m>mmcwuco Lonmo uppmsm ommmmgucH .umgo< guzogo mcom wmmo m cam c com m com N com p com Poocc< .>< mmmomhmo m>~h<2mm~o< mmgz: ~mm~ zH muchomm >m ouoo< m=4<> om m4m pcoscgm>oo woos—ucH H N.m o.c o.c o.c o.c o.c .u_com:oz mpmumumogmo 3 3 3 M: Na 9m 32352 m_momumesm 3 9m 3 m4 3 mm 3.22 o.m o.c N.c o.c o.c N.c Npmuok - ”cocoa emu ucmucwmv zgumooco zopmco mpmum1mogmo Popcomo cw mzcwcsumh cowposoga Nuw>muuaoogo cowpoeogo _oo~ :owmgou m>mmcwpco NgumoocH .owgo< mugooxm , tomowgucH Lonoo mpoumuppoEm ummmmxucm .owgm< com omom m com c cam m com N com _ com mmuwmhmo m>HHm omoo< w=4<> no mhoo mononucm a 117 c.8N N.m~ P.0N N.@N c.0N c.m o.mN o._N mmumsemm a munch c.c m.c m.c N.c m.c _.~ m.c c.c mm_umea= m.c c.c c.c m.c N.c N.N m.c m.m mcweapuacaeaz m.cm a.cm m.cm c.cm m.cm m.o m.cm c.Nm mepcwz c.Nm m.om N.om m.om o._m o.m N.om c.mN gawuuzepmcou N.mop o.oop c.oo_ m.oo_ m..op N.N c.cop P.5m .maeoz a_aum-amea3 N.NN “.0N c.N“ ..NN c.N“ m.c c.c“ _.Nm munch m.mm m.Nm m.Nm _.mm o.mm _.m N.Nm m.NN utoamcaeh F.m_ N.N_ _.m_ w.NP m.N_ c.N, m.m =ane:-mmeag N.m o.m m.m o.m m.w m.m c.c eanL=-_paEm m._op o.oop m.mo_ m.FoP m.oop N.am P.MN _aeam _.mmp m._N_ c.mN_ m.mNF N._NP _.m m.om_ m.mm .zcaz m_aom-Fpa2m p.mMN m.mmN m.ocN m.mm~ “.mmm o.m c.cmm “.mm_ .macoz m_aum-__asm c.mNp m.mNF c.0NP m._N_ _.NNF N.c N.©NP m.cm .8wem< .cwmmm N.oo~ N.oo_ N.oo_ N.Nmp o.mNP m.N N.omP m.mmp macaw “coax“ w.mmm c.cam _.Nmm m.Pmm o.msm m.m m.~mm N.mNN coca assoc c.Nmm m.mmm m.cmm o.me N.cmm c.N F.mmm c.com mu_m m.mmo.~ c.Nmo._ c.cmo._ m.mmo._ c.cmo.F m.m m.mmo.F F.Npm mesa—auwemc o._cc.. m.mmc.P F.Pcc.F o.occ.F c.Pcc.P m.m N.mmc.P m.mN..p Hpauoh massages ooo.,- pwcoo Ngumoocn zopceo m.aum-wmcas _mm_-cNmF .& c. _mm. cum, Pmuwamo cw macwcsum» compoeoco ouw>wpuooogo cowpoEoLo corona; w>wmcmucn Ngumooco .umgo< masooxm mama . . owmmmsucm Loom; upmumupmem cummmguco .umgo< suzoco mcom ammo mesa c :3. m :3. N :3. P :3. 2553 .>< mmmomhmo m>HHm hzmz>ooa2m Fm m4moo mmoopucH a c.N N.N N.N N.N c.N N.N drama: mpmumum no c.m _.m m.m N.m N.m o.m .UPemacoz w—mumupmem o.c o.m o.c o.m o.c o.m "pouch [basses Ema ucmugmav Ngumooco zepwcm upoumumogmo qupamo cw moowccumh :o_uoeoga xuv>wuu=oocq copuosogo Poop zowmgom m>wmcmaca acumoocu .Uwgo< mpgooxm on own ommmmgocH goon; mpmumnppmsm ommmwgucH .uwgo< m o m com c com m com N cam P com mmnomhmo m>~hm hzmz>ooo2m mo mhwuu:ooca :o_uosoca cowwcou :_ m:o_:;um» acumzocm .uwcoc mucooxo ommwwcuc~ m>_m:mu:o Loooo mpmum-__mem owmomcoco .u_coo. mcsm mmoo m com c com o cam N cam F cam (1 mm_wmhuo m>~hooo2m22 oz< zo~H04m:m zwpuooogo cowpoEogo Poop cNoP cowmgou m>wmcmucH agpmoocu .upco< mpgoaxm we: own ommmmgucu Loam; mpmum1PFmsm ommmmguco .upgo< m o m ::m c com o com N com P com mumwmhmo m>HH