E..‘:' S- l”; \‘ " I ' ' n'v . "' ‘ .u .., .' y... ~‘-:~f'3t'«I¢'.-.::tIIf‘JIIIIIii’IieIfiISfi.IIgi-JIi; 1 ' . kWh-I . '~ . I OOMN‘%‘\\‘_‘I&‘ «v.4- _ . .v_ - . . . AN ANALYSIS OF RURAL CONSUMPTION PATTERNS IN SIERRA LEONE AND THEIR EMPLOYMENT AND GROWTH EFFECTS Thesis for the Degree of M. S. MICHIGAN STATE UNIVERSITY ' ROBERT PHILIP KING 1977 ABSTRACT AN ANALYSIS OF RURAL CONSUMPTION PATTERNS IN SIERRA LEONE AND THEIR EMPLOYMENT AND GROWTH EFFECTS BY Robert Philip King The importance of consumer demand in the process of economic growth has gained increasing recognition in the development literature. Several strategies of economic development which rely heavily on consumption based em- ployment effects and intersectoral linkages have been proposed. However, relatively few studies designed to test the validity of the hypothesized consumption effects upon which these strategies depend have been undertaken. This study focuses on growth and employment effects associated with rural consumption patterns in Sierra Leone. In particular, the factor intensity of rural consumer de- mand at different income levels is examined in light of the hypothesis that the labor intensity of consumer de- mand decreases as incomes rise, while the capital inten- sity and foreign exchange requirements increase. Robert Philip King The objectives of the study are: (1) to describe current consumption patterns and to provide a basis for the projection of consumer demand in the rural areas of Sierra Leone; (2) to analyze the impact of consumption patterns at different income levels on employment, capital requirements, and import demand; (3) to determine the nature and strength of consumption based intersectoral and rural-urban market linkages; and (4) to formulate a methodological approach to consumption research in the rural areas of developing countries designed to address specific theoretical and empirical issues. Data for the study were obtained from a survey con- ducted over a twelve month period from May 1974 through April 1975. In addition to information on cash expendi— tures, data on household production and sales were col- lected for the sample of 203 households for the determina- tion of households' subsistence consumption. To permit analysis of the factor intensity of consumer demand and of associated intersectoral linkages, commodity group- ings were kept highly disaggregated during data collection, and the origin of purchased goods was recorded. Average and marginal propensities to consume and total expenditure elasticities by income class for a dis- aggregated set of commodity groupings, as well as average and marginal labor, capital, and foreign exchange require- ments for each of the six income classes, are estimated. Robert Philip King Particular emphasis is placed on the choice of estimation procedures suited to the objectives of the study. The factor intensity results of this study are con- sistent with the hypothesis that capital and foreign ex- change requirements per unit of consumption expenditure increase and labor requirements decrease as incomes rise. Variation in factor intensity is not as pronounced as that reported in Latin American and Asian studies, however. The strength of intersectoral and interregional market linkages is found to be relatively invariant with respect to changes in income. Rural-urban linkages are quite weak, which indicates the impact of rural development pro- grams on urban sectors may be limited. The methodology developed in this study is based on the premise that data collection, data analysis, and the application of research results to theoretical and empir- ical problems are interrelated processes. The survey design and statistical estimation procedures were devel- oped to facilitate the testing of particular hypotheses concerning consumer behavior. In addition to this gen- eral approach, two methodological findings are of parti— cular interest. First, the effects of substituting small positive values for zero observations in statis- tical models with logarithmic dependent variables are investigated. Parameter estimates are found to be quite sensitive to the size of the substituted value which indicates such models should not be used when zero Robert Philip King observations are present. Second, expenditure elastici- ties based on cash and on total expenditure data are compared. Cash expenditure elasticities are found to be reasonable estimates of total expenditure elasticities for commodities not produced by households for their own consumption. AN ANALYSIS OF RURAL CONSUMPTION PATTERNS IN SIERRA LEONE AND THEIR EMPLOYMENT AND GROWTH EFFECTS BY Robert Philip King A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Agricultural Economics 1977 ACKNOWLEDGMENTS For their guidance and assistance, I wish to thank the members of my thesis committee: Derek Byerlee, Carl Liedholm, Lester Manderscheid, and Warren Vincent, my major professor. I also appreciate the help of Carl Eicher, who was largely responsible for my decision to undertake graduate studies in agricultural economics. I am especially grateful to Derek Byerlee, my thesis super— visor, for his interest, patience, and wealth of good ideas. I feel lucky to have had the chance to work with him. Funds for this study were provided by the Agency for International Development under research contracts for the study of rural employment and income distribution. Data for the study were made available by the Njala Rural Employment Project. My appreciation also goes to computer programmers Mike Dege and Linda Buttle for their assistance and to Janet Munn and Lucy Wells, who were responsible for typ- ing this thesis. Finally, I thank my parents and my wife, Jane, for their encouragement and support. ii TABLE OF CONTENTS ACKNOWLEDGMENTS . Chapter 1 INTRODUCTION 1.1. Consumption Research in Develop- ing Countries: The Problem Setting . . 1.2. Relationship of This Study to Previous Research . . 1.3. Objectives of the Study . 1.4. Plan for Remaining Chapters 2 SURVEY METHODOLOGY 2.1. Data for the Study of Rural Consumption Patterns 2.2. Sampling Procedures . . 2.3. Reference Periods for Survey Interviews . . 2.4. The Scheduling of Interviews 2.5. Preparation of the Data for Analysis . 2.6. The Estimation of Subsistence Consumption . 3 A DESCRIPTIVE ANALYSIS OF RURAL CONSUMPTION PATTERNS . 3.1. 3.2. 3.3. The Measurement of Income: Some Conceptual Problems Definition of Income Classes for the Sierra Leone Consumption Study . . The Definition of Expenditure Patterns in Rural Areas . iii Page ii (C(96) H 11 11 13 15 17 18 19 20 20 24 31 Chapter Page 3.4. Description of Expenditure Patterns in Rural Areas . . . 32 3.5. Seasonal Variation in Consump- tion Patterns . . . . . . . . 37 4 METHODOLOGICAL APPROACHES TO THE STATISTICAL ANALYSIS OF RURAL CONSUMPTION PATTERNS . . . . . . . . . . 41 4.1 Engel Curve Analysis . . . . . . 41 4.2 Variables Included in the Analysis . . . . 43 4.3 The Choice of Functional Form . . 45 4.4 Total Versus Cash Expenditure Elasticities . . . . . . 54 4.5 Statistical Problems with Estimation . . . . . . . . . 56 5 THE ESTIMATION OF TOTAL EXPENDITURE ELASTICITIES AND MARGINAL PROPENSITIES T0 CONSUME: EMPIRICAL RESULTS . . . . . 63 5.1. Introduction . . . . 63 5.2. The Effects of Zero Observations . 63 5.3. The Performance of the Two Models . . . . . . 69 5.4. Estimates of Marginal Propen- sities to Consume and Expenditure Elasticities. . . 81 5.5. Total Versus Cash Expenditure Elasticities . . . . . . . . 87 6 EMPLOYMENT AND GROWTH EFFECTS OF RURAL CONSUMPTION PATTERNS . . . . . . . . . . 94 6.1. The Factor Intensity of Rural Consumption Patterns . . . . 94 6.2. Consumption-Based Growth Linkages . . . . . . . . . . 102 6.3. Policy Implications . . . . . . . 104 7 SUMMARY AND CONCLUSIONS . . . . . . . . . 106 iv APPENDICES 1 BIBLIOGRAPHY Questionnaires Used for Obtaining Cash Expenditure Data for the Sierra Leone Rural Consumption Study Indexing Procedure Used to Fill in Missing Data Page 113 118 122 Table 3.1 LIST OF TABLES Definition of Income Classes and Their Economic and Demographic Characteristics Demographic and Economic Characteris- tics of Sample Households Grouped by Region Budget Shares for Commodities by Income Class Budget Shares for Commodities Grouped by Origin Parameter Estimates for Test Commodities in Zero Observation Experiments Estimated Expenditure Elasticities and Marginal Propensities to Consume for Commodities in Zero Obser- vation Experiment Parameters Estimated with Log-Log Inverse Model . . . . Parameters Estimated with Modified Ratio Semi-Log Inverse Model Household Size Elasticities Derived from the Two Models Estimated Mean Expenditure Elastici- ties and Marginal Propensities to Consume for the Two Models Sums of Marginal Propensities to Consume Derived from the Log-Log Inverse Model vi Page 25 27 33 36 65 68 7O 72 76 77 80 Table 5.8. Page Estimated Total Expenditure Elasti- cities and Marginal Propen— sities to Consume by Income Class . . . . . . . . . . . . . . . . 82 Estimated Parameters for Commodities Grouped by Origin . . . . . . . . . . 86 Estimated Total Expenditure Elasti- cities and Marginal Propen- sities to Consume for , Commodities Grouped by Origin . . . . 88 Cash Expenditure Versus Total Expen- diture Elasticities and Marginal Propensities to Consume . . . . . . . 90 Labor-Output and Capital-Output Ratios for Sectors of the Sierra Leonian Economy . . . . . . . . . . . 97 Average Labor, Capital, and Foreign Exchange Requirements Per Leone of Expenditure by Income Class . . . 99 Marginal Labor, Capital, and Foreign Exchange Requirements Per Addi- tional Leone of Expenditure by Income Class . . . . . . . . . . . . 100 vii LIST OF FIGURES Figure 2.1 Sierra Leone Rural Resource Regions 3.1. Lorenz Curve for Rural Households 3.2. Seasonal Variation in Cash Expen- diture on Rice, Other Food, and Nonfood Commodities viii Page 14 30 38 CHAPTER 1 INTRODUCTION 1.1. Consumption Research in Developing Countries: The Problem Setting Consumer demand has long been recognized as an impor— tant factor in the process of economic growth. Information on consumption patterns is a major input to economic plan- ning and development program design. Only in recent years, however, as interest in questions relating to income dis- tribution has increased, have the effects of variations in consumer demand associated with differences in income been incorporated into growth models and development strategies used in developing countries. As the distinction between growth and development has evolved in the theoretical lit- erature, the nature of the employment and growth effects associated with a movement toward a more equitable distri- bution of incomes--effects which are manifested through consumer demand--has become an important empirical question. Many believe that trade—offs among the multiple goals of equity, employment, and economic growth are minor. Atten- tion has also focused on the role of consumer demand as a medium through which the multiplier effects of growth in the rural areas can be transferred to other sectors of the 2 economy. When this occurs in economies characterized by sharp social and economic divisions between rural and ur- ban sectors, rural consumption patterns can be viewed as a much needed integrative force. Despite widespread interest in consumption patterns and their employment and growth effects, relatively little con- sumption research has been undertaken in developing coun— tries.- Most studies that have been conducted have not addressed the central issues identified in the theoretical literature. Often consumption data are collected only for use in the construction of price indices. While this is an important objective, it is also a rather limited one. A more comprehensive knowledge of consumption can contribute in a number of other ways to planning and program design. At a minimum, consumption research should be designed to include the description of current consumption patterns and estimates of income or expenditure elasticities which can be used for projecting future consumer demand for spec- ific commodities. In many developing countries, projections of consumer demand for even major commodities are based on general estimates of income elasticities, such as those pro- vided by the Food and Agriculture Organization (FAQ). In other cases, elasticities estimated from inaccurate or in- complete time series data are used. In cases where rela- tively reliable estimates of income elasticities are avail- able for major commodities, income elasticities or average budget shares for commodities of lesser importance are often 3 not available, even when such information would be quite valuable. For example, despite increasing interest in appropriate technology and the encouragement of small- scale industry, little is known about the nature of con- sumer demand for the products of small-scale industrial‘ firms. Hymer and Resnick [1969] in a widely recognized theoretical paper have hypothesized that the income elas- ticity for such products is near zero or negative in rural areas. This hypothesis should dampen the interest of gov- ernments in the development of small-scale industry, but it has not been empirically examined. As stated above, consumption patterns have important implications for employment and growth. The hypothesis1 that the labor intensity of goods consumed decreases as in- comes rise while the capital intensity of consumption and demand for imports increases, has been widely accepted in the development literature. The implications for economic planning and development strategy of this intuitively at- tractive idea are great. It implies there need be no trade- off between equity on the one hand and growth and employ- ment on the other, since a redistribution of incomes is ex- pected to result in higher employment and fewer leakages to imported capital and consumer goods. Few empirical studies designed to test the validity of these assertions or the magnitude of the postulated effects have been undertaken. 1First explicitly stated in Towards Full Employment [1970], the I.L.O. study on Colombia. 4 Results that have been reported to date tend to support the hypothesis,2 but no studies addressing this question have been conducted in Africa where the applicability of devel- opment strategies based on Latin American and Asian exper- iences may be questionable. Finally, investigation of the role of consumption pat- terns in the creation of intersectoral linkages as suggest- ed by Mellor [1976] should be another objective of consump— tion research. While earlier studies have emphasized the aggregate effects of current or projected consumption pat- terns, Mellor focuses on the growth and employment impacts of consumption patterns for specific sectors of the develop- ing economy and on the integrative effects of growth real- ized through increased market interaction. He sees new foodgrain technologies as a major impetus for-growth in the agricultural sector and rural consumption expenditures as the primary means by which the multiplier effects of this growth are initiated. In an earlier paper written with Uma Lele [Mellor and Lele, 1970, p. 7], Mellor states: the new foodgrain technologies normally require increased purchase of current inputs and may stimulate greater purchase of fixed capital goods from other sectors. Far more important, however, is the large increase in consumption expenditure which is likely to occur. It is the large aggregate increase in net agricultural income and consequent purchase of con- sumption goods which offer a large potential stimulus to other sectors. 2Studies by Soligo [1973] in Pakistan and Sunman [1974] in Turkey are supportive, while a more recent study by Ballentine and Soligo [1975] using Colombian data indi- cates its long-run validity may be questionable. 5 Consumption linkages between agricultural and other sec- tors of the economy have not been explored in this light in most developing countries. The importance of consumption patterns in the develop- ment process and the relative paucity of research address- ing current theoretical and empirical questions point to the need for an intensification of consumption research efforts. This study, which focuses on rural consumption patterns in Sierra Leone, is directed toward both the es— timation of empirical relationships and the testing of theoretical assertions. In addition, a methodology for consumption research in developing countries which inte— grates data collection, data analysis, and hypotheses test- ing will be formulated. Sierra Leone is a West African country of approximately 2.7 million people which borders on Liberia and Guinea. Its economy is dominated by the agricultural sector, which employs 77 percent of the work force and produces 32 per- cent of the GDP [Central Statistics Office, 1972b]. Mining is also important, contributing 16 percent of GDP but only 5 percent of employment. Approximately 25 percent of the population lives in urban areas, where most employment is in the government, trading, and large-scale industry sec- tors. As is the case in many West African countries, Sierra Leone is plagued with high unemployment in urban centers, balance of payment problems, and stagnation in the agricultural sector. 6 This study was undertaken as one component of an inte- grated nationwide survey conducted in rural and urban areas of Sierra Leone in 1974 and 1975. The survey was designed to provide comprehensive information on output, employment, and income in rural areas for use in evaluating the impli- cations of various policy alternatives on the rural sector of the economy and on the larger national economy and to develop a research methodology applicable to similar stud- ies in other African countries.3 Agricultural production and processing, small-scale industry, migration, and the fisheries industry were other foci of the study. The re- sults from the separate studies, in addition to addressing particular empirical, theoretical, and policy questions, have been incorporated into an aggregate macroeconomic model designed for use in planning and policy evaluation. 1.2. Relationship of This Study to Previous Research Several household budget studies have been conducted in Sierra Leone, though relatively little research has been undertaken in rural areas. A household survey of Freetown and the surrounding Western Area was conducted between 1966 ' and 1968 [Central Statistics Office, 1968].- Similar surveys were conducted in the urban areas of all three provinces of Sierra Leone [Central Statistics Office, 1971a, 1971b, 1971c]. Because these surveys did not contain data on rural 3See African Rural Employment Research Network [1974] for a more detailed statement of the survey's objectives. 7 consumers, a final household survey was conducted in rural areas between 1969 and 1970 [Central Statistics Office, 1972a]. While all of these studies provide average con- sumption data for a highly disaggregated set of commodities and some information on variations in budget shares at dif- fering income levels, no income elasticities or marginal propensities to consume are given. Therefore, findings are of little use for the projection of commodity demands or for analyzing the impacts of rising rural incomes on the economy. Snyder [1971] and Levi [1976], both working with data collected in and around Freetown, do estimate income elas- ticities for a number of goods. There is no a priori rea- son to believe, however, that rural and urban consumers be- have in a similar fashion, especially when subsistence con- sumption is important in rural areas.4 Massell's studies in Kenya [1969] and Uganda [Massell and Parnes, 1969] pro— vide the most comprehensive analysis of rural consumption patterns in African countries. Other rural consumption studies include those by Hay [1966] in Nigeria and Leurquin [1960] in Rwanda-Urundi. These studies, especially those of Massell and Hay, have made considerable contributions to the methodology of estimating income elasticities for rural African consumers. None, however, have analyzed the 4Massell and Parnes [1969] compare estimated elastici- ties for Nairobi with those for rural Kenya and rural Uganda and find both striking similarities and marked dissimilarities. 8 employment and growth effects of variation in rural con- sumption patterns by income group. Elsewhere, outside of Africa, Soligo [1973], Sunman [1974], and Ballentine and Soligo [1974] have investigated the nature of these effects in Pakistan, Turkey, and Colombia respectively. In all of these studies variation in the factor intensity of consump- tion across income classes and its impact on employment and capital requirements were analyzed. Ballentine and Soligo [1974] carry this work the farthest, examining the direct and indirect effects of consumption patterns under differ- ent income distributions. None of these studies, however, develop a unified methodology in which data collection and analysis are designed to test specific hypotheses. The studies referred to above which focus on the factor inten- sity of consumption patterns at different income levels have been based on income elasticities and marginal pro- pensities to consume generated by other researchers, and little is said concerning the estimation of consumption— income relationships. The need for a more integrated meth- odology of consumption research stems from the fact that decisions made when data are being collected or when in- come elasticities and marginal propensities to consume are being estimated often preclude the testing of hypotheses relevant to theoretical or policy questions. 9 1.3. Objectives of the Study The general objectives of this study are fourfold. The first three are synonymous with the major contributions of consumption research discussed above: (1) to describe cur- rent consumption and to provide a basis for the projection of consumer demand in the rural areas of Sierra Leone; (2) to analyze the impact of consumption patterns at dif- ferent income levels on employment, capital requirements, and import demand; and (3) to determine the nature and strength of consumption based intersectoral and rural-urban market linkages. A fourth objective is to formulate a methodological approach to consumption research in the rural areas of developing countries designed to address specific theoretical and empirical issues. 1.4. Plan for Remaining Chapters In Chapter 2, the distinction between cross section and time series data is discussed and the relevance of each form of data to research oriented toward the objectives of this study is examined. The data collection process for this study is then described, particular attention being given to the influence of research objectives on choices relat- ing to the survey methodology. Economic and demographic characteristics of the sample population are described in Chapter 3. Income classes and commodity groups to be used throughout the study are 10 defined and budget shares for each commodity are presented for each income group. Methodological issues relating to the estimation of income elasticities and marginal propensities to consume are examined in Chapter 4. Special attention is given to the statistical and analytical problems particular to data from rural areas in developing countries. Two statistical models, both suited to the needs of this study, are speci- fied in this chapter. The performance of these two models is tested in Chapter 5, and estimated expenditure elasti— cities and marginal propensities to consume are presented. The capital and labor intensity of rural consumption patterns at different levels of income are estimated in Chapter 6, and the results are analyzed to determine the employment, capital, and foreign exchange requirements associated with both current and projected rural consumer demand. Finally, these results are used along with infor- mation on the breakdown by origin of goods consumed in each income class to identify potential growth linkages based on rural consumer demands. The results of the study are summarized in Chapter 7, and additional research re- quirements are outlined. CHAPTER 2 SURVEY METHODOLOGY 2.1. Data for the Study of Rural Consumption Patterns The analysis of consumer demand can be based on either 1 Time series data con- time series or cross section data. sist of periodic observations on aggregate variables taken over a relatively long time frame. It is assumed that dif— ferent time periods are homogenous and that variation in consumption patterns can be explained by variables such as commodity prices, income, and population. Cross section data, gathered in household budget surveys, consist of ob- servations on a number of households over a relatively short period of time. In effect, time, prices, and other market variables are held constant, and the association between consumption levels and variables such as household income, household size, and location is examined. It has been argued by Howe [1966], among others, that the data requirements for any extensive analysis of con- sumption in developing countries can be met only by the initiation of household budget surveys designed to address specific theoretical and empirical questions. Time series data are not appropriate for the analysis of the employ- ment and growth effects of consumption patterns at different 1Much of this discussion is based on a comparison of cross-section and time series data in Klein [1972]. 11 l!” (ilillll l ll‘l' Il‘l“ '.II\ III 12 income levels because they are usually not disaggregated by income subgroups. Even in the projection of consumer demand, time series data may be too highly aggregated to be of use in answering questions relevant to the design of localized development projects. Perhaps the greatest drawback in the use of time series data in developing countries, however, is that they must be collected over a long period of time. Household budget studies can be com- pleted in a relatively short period of time and can pro— vide immediate answers to policy questions. While price effects are not easily analyzed in a cross section frame- work, household budget data can be supplemented by avail- able time series data; and when collected over an extended period of time, they can become a source of time series data. The basis for this study is cross section data collect- ed in a national rural household budget survey of Sierra Leone conducted between March 1974 and May 1975. Expen- diture data were collected for a highly disaggregated set of commodities and information concerning the origin of goods purchased (used in the classification of goods by factor intensity and in the analysis of intersectoral link- ages) was also included in the data set. The survey data were supplemented with household-specific data from an ongoing agricultural production survey and with information on the capital and labor intensity of goods from various sources . 13 2.2. Sampling Procedures The rural consumption survey in Sierra Leone was closely integrated with an agricultural production survey, which was another component of the integrated research project. In the rural household production survey, farm management data were collected in the eight resource re- gions indicated on the map in Figure 2.1. A total of five hundred randomly chosen households2 were surveyed in the rural household production study. One-half of these were chosen at random for the consumption study. For these households, data on labor use, production practices, out— put, and sales, as well as consumption expenditures, were available. This unified sampling approach facilitated the estimation of the value of subsistence production, which has proved difficult in other surveys. The rural household production survey sample was stratified by resource region. For the purposes of the consumption study, stratification by income group to allow the separation of income effects from those attri— butable to regional factors would also have been desirable. Such stratification was not possible, however, since com- prehensive data on household income or even proxy varia- bles, such as farm size, were not available prior to the initiation of the study. In practice, one-half of the sam- ple households assigned to each farm management enumerator _x 2A household is defined as a group of persons who eat from the same pot. l4 I_- . u" m n' v!‘ If u' I I I 0.1—. r.-...‘—.—._.-.—.—u-.\. o _d '- .,..‘ ’ _/. ‘ , ~ ,f' \,‘ if Siltuit '\. c ' \ I" '\. a ‘J. X 0! bole . f I 7 O \ z I ('1 m I - ’5‘ “in“ K J ‘ t. b‘ ‘.\. IA: I I” ' " I; 7 a X} obflamblo f... '1. but . (.u ( - . ,l n. 1‘ n " " ‘ - “-9. ’1. . ~ Ila-(2w '"" "\.. 4.. rfiz 80-“: Gone 'I ' \ I If ’ I'- T“ . L. 0" .0 ° . / o. v — ,0 lament: /. . I /- \ = - Hokcnl .- ! Mon-h l \ oPort loko (“n/‘tvqua "/7 I ‘. m .' o / Hoqb’lroka .' ‘3 k1: .- lun so! :1 ’3 I oKoldu ’- 3‘ 2L. , o c. I I. o. 5 ° ' \ I o I . . 'N \- c.\ .0 J. “- ~.. .’ r 3 w x" -, 1. fix; (’7 —.'.- - Hanna‘s '° °' . ° ' ' . , . u‘ ~o\£3 (Lu/ .\ \"A.., 'dl.~‘\'/ I I " ,_..r" "\\, K ) I new“? ' " laid... ° -’ i \ 3.5 Paulo-ulna ,o’l'”, ’) 8 .9. ‘_ "95“ I, .1 r ’. .Hokanji ./ Segb‘cu’I i G") “"3“?" /. ’IOMIu | N . , ‘ i! l . \ \ \ .- Tam-uh» Tm ' J - a: ‘. . 6 ,I e 0.4 “dork '\ . (" \'. Mulch .’ ' '9 .WJ T OII':\ . .’ Q Mg.) Haul l/ . )' 'I \ ' . I. IuEL .- "o . I " loath .4. \. 0' (it-Thu. “ ' 'J'I II. \, , ' ’Q‘ Hod-m‘ Pom-0n , .l I: . f: ) J" LI $/"'\._\ \ .,\ ’1' known chIcMI location" . .— ..... .0 o. J' “'T'fi‘ Dx humane-ml Boundary ------ - ------ o no to so 40 so so mm bum o, -‘ \ _...___. “:7 hf-Lfi‘.;”'-L-—.-1:11‘:‘:A ‘ 3 /" ("OlfOUM k... - - . . . o - .1 ‘ 3 V. "‘ I | -—- l.-...1 ‘. 7. 0 20 40 so noun-mu RI Run-rec My.“ ......... 4 2 IMO I 1 Wk Atom '3' NOIE: (1) Scarcies, (2) Southern Coast, (3) Northern Plains, (4) Riverain Grasslands, (5) Boliland, (6) Moa Basin, (7) Northern Plateau, (8) Southern Plains FIGURE 2.1 SIERRA LEONE RURAL RESOURCE REGIONS 15 were selected at random. A number of the initial sample of 250 households were dropped from the sample prior to analysis due to inadequacies in data, deaths, or movement of the respondent from the region, leaving a final sample of 203 households. 2.3. Reference Periods for Survey Interviews The accuracy of consumption expenditure data is de- pendent on the length of the reference period used in the survey questionnaires. The reference period is the length of time over which interview respondents are required to recall events from memory. The ability to remember events, such as consumption expenditures, diminishes as the length of the reference period increases. This problem of reduced recall capacity is most severe for events that occur fre- quently, such as the purchase of food, tobacco, beverages, and regularly consumed household goods. Another source of bias is what Prais and Houthakker [1971] call the end period effect. This occurs most often for durables and other less frequently purchased commodi- ties. Respondents tend to include expenditures from earl- ier time periods in their reporting of consumption, espec- ially for items for which there has been no expenditure during the time period under inquiry. Therefore, short reference periods can lead to some overestimation of expen- ditures for goods of this sort. 16 In order to reduce biases caused by these effects, two questionnaires with different reference periods were used.3 A questionnaire with a reference period of four days (RER/ Cl) was used to record all consumption expenditures made by a household within the recall period. This question— naire was intended as a source of data on expenditures for food, beverages, tobacco, and other commonly purchased items. The second questionnaire (RER/C2) had a reference period of one month and was used to record only expendi- tures on durables and less frequently purchased goods. Expenditures on food, tobacco, beverages, and other non- durable personal items were not recorded on the survey forms for this questionnaire. In both the weekly and monthly questionnaires, infor- mation on commodity purchased, its origin, the place of purchase, quantity, and cost were collected. Both survey forms were partially pre-coded by commodity to remind enu- merators to ask about certain commonly purchased items. Origins were grouped into five categories: rural, large urban, smaller urban, imported, and undetermined4 loca- tion. This information was gathered for the analysis of the locational impacts of rural consumption patterns. All quantities were measured in local units. 3See Appendix 1 for copies of both questionnaires. 4This category included expenditures on school fees, medical services, transportation, etc., which could not be attributed to a particular location. l7 Enumerators were instructed to be as specific as possible concerning the nature of commodities. In this way a minimum of information relating to consumption of rather specific commodity groupings and to the factor in- tensity of total consumption was lost. 2.4. The Scheduling of Interviews Interviewing for the consumption study was conducted over an entire cropping year in conjunction with the farm survey, using the same enumerators. Enumerators inter- viewed in each household on a twice weekly basis in connec- tion with the farm survey, but it was felt that such repe- titive collection of consumption data might quickly lead to fatigue on the part of both enumerators and respondents, which could have resulted in standardization of responses. To avoid this problem, the short reference period ques- tionnaire was administered only twice each month for successive four—day reference periods. It was assumed that consumption of commonly purchased goods is relatively constant during a month. The scheduling of consumption interviews was estab- lished by grouping the sample for each enumeration area into four groups, each corresponding to a week of the month. In general, each group consisted of three house— holds. For a given week of a month the three households in the associated group were administered the short refer- ence period questionnaire. The long reference period 18 questionnaire was administered to each household in the sample during the last week of the month.5 In this way, the enumerator's work load was distributed evenly through- out the month and continuous data within each enumeration area were obtained. 2.5. Preparation of the Data for Analysis Because the purchases of commonly consumed goods were recorded for only one week in four, it was necessary to "puff up" the data. This was done under the assumption that consumption of these goods is relatively consistent from day to day. Therefore, if data were available for seven days out of thirty in a given month, recorded expen- ditures for a particular good were multiplied by 30/7 to estimate expenditure for that good for that month. Missing data were also a problem in some cases. When the amount of data present for a household met certain minimum standards,6 months for which no data were avail- able were filled in using the indexing procedure described in Appendix 2. 5Observations that were obviously recorded on both questionnaires were screened at the time of coding to avoid duplication. 6At least three months of consumption data, a valid month being defined as having at least three days of data from the short reference period questionnaire and the pre- sence of the long reference period questionnaire. 19 2.6. The Estimation of Subsistence Consumption The data collected through the administration of the two survey questionnaires provide an accurate representa- tion of cash expenditures in consumption goods, but they do not measure the value of subsistence consumption, i.e., the value of goods produced and consumed by a household without entering the market. Data on output and sales from the farm management survey were used to estimate households' consumption of home produced goods. Subsistence consump- tion was defined simply as the difference in the value of what a household produced and what it sold. Both output and sales were valued at farm gate prices. This method of estimation caused some difficulties since sales data were apparently underestimated in a number of cases.7 In gen- eral, though, this approach seems to have yielded satis- factory results. Total consumption for a given commodity, then, was defined as the sum of cash expenditures and the value of subsistence consumption for that good. 7Subsistence consumption of coffee and cocoa, which are not generally consumed in rural households in any quan- tity, for example, was estimated to be quite high by this method. Similar problems were encountered with small- scale industry products. Because of these obvious diffi- culties, subsistence consumption for these goods was set at zero. The accurate measurement of sales has been a difficult problem in many studies. CHAPTER 3 A DESCRIPTIVE ANALYSIS OF RURAL CONSUMPTION PATTERNS 3.1. The Measurement of Income: Some Conceptual Problems In the theory of consumer behavior, levels of consump- tion for distinct goods or sets of goods are determined to a large extent by their respective prices and by the income of the consumption unit. Other factors such as tastes and preferences, household size and composition, and environ- mental constraints also affect consumption decisions. Since prices are assumed constant in this study, income be- comes an even more important variable for exploring differ— ences in consumption. Net household income can be considered to be a rea— sonable measure of current income for the households in the survey sample. It is defined by the following functional relationship: I = S + M — F + W (3.1) where I is net household income, S is the value of subsis- tence consumption, M is the value of total farm and non- farm sales, F is the value of purchased and nonpurchased factors of productivity excluding unpaid family labor, and W is the value of wages received from off-farm employment. A measure of current income such as net household income is often used as an explanatory variable in the analysis of consumption behavior. It can be argued, however, that 20 21 another measure of income, total consumption expenditure, may, if the time period over which data extend is suffi- ciently long, be a better indicator of permanent income, which Modigliani and Bromberg [1954] and Friedman [1957] hypothesize to be the true determinant of consumption be- havior.1 Total consumption expenditure, Y, is the measure of household income used in this study. It can be defined as the sum of the value of subsistence consumption, S, and cash expenditures: Y = S + C . (3.2) Cash expenditures and the quantity (M — F + W) cannot be expected to be equal, though they can be expected to be highly correlated. To the extent that they differ, so will total consumption expenditure and current net household in- come. To facilitate the description of consumption patterns and the discussion of analytical results, sample households were grouped into income classes on the basis of household consumption expenditure per person. As Kuznets [1976] points out in the following passage from his recent essay on the demographic aspects of income and distribution, this is the only valid form of income measurement for the 1Prais and Houthakker [1971] note that in surveys con- ducted over a very short period, a single expenditure on a major durable may cause total expenditures to grossly overrepresent permanent income. Since households in this survey were observed over an entire year, this should not be a problem here. 22 analysis of income distribution among households of vary- ing sizes [Kuznets, 1976, p. 87]: It makes little sense to talk about inequality in the distribution of income among families or house- holds by income per family or household when the under- lying units differ so much in size. A large income for a large family may turn out to be small on a per person or per consumer equivalent basis, and a small income for a small family may turn out to be large with the allowance for size of the family. Size dis- tributions of income among families or households by income per family or household, reflecting as they do differences in size, are unrevealing--unless the per family or household income differences are so large as to overshadow any reasonably assumed differences in size of units, or unless the latter differences are minor. Neither of these conditions is realistic. It follows that, before any analysis can be undertaken, size distributions of families or households by income per family or household must be converted to distri- butions of persons (or consumer equivalents) by size of family or household income per person (or per con-- sumer). Classification of households by consumption expenditure per person, while clearly superior to a grouping based on consumption expenditures unadjusted for household size, fails to take the composition of a household into account. This factor can also affect consumption decisions. Given two households, each with the same income and the same num- ber of members, for example, a household composed entirely of adults may be expected to meet minimum caloric require- ments less easily than a household made up of only two adults and several small children. Ideally some consumer equivalent scale should be used to adjust household size in order to compensate for differing percentages of child- ren and for other compositional factors such as the ratio 23 of males to females and the proportion of elderly persons in the household. Several consumer equivalent scales have been used in African consumption studies. Massell [1969] treats all adults, male and female, as equal consumer units, and weighs children at one-half a consumer unit. Howe [1966], in a Nairobi consumption study, uses the following weights: males, sixteen and older, 1.0; females, sixteen and older, 0.8; and children under sixteen, 0.6. The difficulty with such consumer unit scales is, as Prais and Houthakker [1971] note in their excellent discussion of this topic, that in actuality consumer unit scales should be commodity specific. They present a formulation in which, concep- tually, "household size” or the number of unit consumers is different for each commodity. The measure of household size to be used in the determination of total per capita consumption expenditure is based on the sum of commodity specific "household sizes" weighted by appropriate aver- age propensities to consume. While the approach outlined by Prais and Houthakker is theoretically attractive, it can prove to be difficult and expensive to implement. Also, it can lead to a defi- nition of the average consumer that may not be in accord- ance with that of policymakers and planners. It was de- cided, therefore, to use unadjusted household size in de- scribing consumption patterns and to make appropriate ad- justments for household consumption in the regression 24 equation used to estimate expenditure elasticities and marginal propensities to consume. 3.2. Definition of Income Classes for the Sierra Leone Consumption Study Using unadjusted annual per capita consumption expen- ditures as a criterion for grouping, six income classes were established. The first and sixth comprise, respec- tively, the lower and upper 10 percent of households ranked by per capita consumption expenditures. Classes two through five are made up, respectively, of households in the second and third, fourth and fifth, sixth and seventh, and eighth and ninth deciles of the ranked sample population. This classification accentuates the difference between the high- est and lowest income class-and so should facilitate the analysis of the effect of income on consumption. Lower and upper bounds, as well as mean expenditure levels for the six income classes, average household size, the percentage of the members in a household who are less than sixteen, and the percentage of total value of goods consumed attri- butable to subsistence consumption are given in Table 3.1. Examination of the figures given in Table 3.1 reveals a consistent trend in household size and in the percentage of children in a household across the range of incomes. Both decline steadily as per capita consumption expendi- tures rise.2 A simple economic explanation for this 2Kuznet's [1976] results indicate a similar pattern for households in the United States, Germany, Israel, Taiwan, and the Philippines. 25 .oaonomso: on» :H noesmqoo cad oceanouq tooom Op manmusnauupd ennufincodxm aofipgesmcoo Hap0u «o mwdpaooummn .msms\vsmfl as oH.Ha u mgomo Ha .MHHo ho>Hsm ”mumpow we. em. m.® mm.®HH am.mmv ®¢.om mom magawm wnfipcm mm. HH. m.m mm.vow Hm.mmv oo.on as mfifiomu to a: .m wv mm. m.n mu.HbH mm.mom oo.me av mewooo npcfic cam npsuwm .m uv. mm. m.m mo.mma mm.mvH oo.voH ow mmHHoon nuco>vm nus 2pxflm .v mm. hm. m.» mw.ww mm.moH 00.09 HV mmHHooo mummy new nausea .m mv. vv. >.m Hm.mm om.w® oo.mv Nv mmflqooc unfinu can ncooom .N am. on. v.0H mm.mm mm.mv m¢.om. om moHHoon hosed .H Amocomuv Amocooav «Amocomuv mNHm endpfivsooxm venom venom nofiudm convaflno uHozmmnom dpfiado pom waspflocwdxm .onaufincodxm mUHonwmsom monoumfimpmm mwmucmonom mwmpo>< Hmscc< smog some: umsog mo umnfisz mmaao meoocm mUHBmemeodmde UHzmmn_cmfixxrnunéd swan; 0.H 0.H 0.0 0.0 0.« 0.0 0.H ooomnou can mowaumpon Hausa 0.H0 0.50 0.05 0.05 0.05 «.00 0.05 0000 HH< «.0 «.0 0.0 «.0 0.0 0.0 0.0 0000 cwmmooonm 0.5 0.0 5.5 0.0a 0.5 0.5 0.0 nmflm 0.« o.m 0.H H.H m.H 0.H 0.H mposuonq xUOpmo>flH can new: 5.0 «.H «.H 0.H m.H 0.H 0.H mucosaucoo and pflam umppoqu 0.0 «.0 0.0 0.0 0.0 0.0 0.0 H00 umsuo and “Ham Hahnm 0.0 0.5 0.0 «.5 «.0 0.0 0.5 H00 sang 5.« 0.« 0.0 0.0 «.« «.0 0.« moanmummm> was mpfisam 0.0 5.0 0.0 5.0 0.5 0.0 «.0 mqouo 0000 0:0 wfiwoumo 0.«0 0.00 0.«0 0.00 0.00 0.00 0.00 0000 mHHoma wmawomn npqaz moHHowQ nuaw>om moawomn £0000 moafiooa chase mafiomn 9003000 and 002000 can 30x00 new saunas and 0coowm 000300 magsam oufluam 0030 0883 you 5m: onsufiocmaxm H0009 mo madpcmoumn 0:090 mpfivoaunu mm¢40 MEOUZH rm mmHBHQOEEOO mom mmmpmm HH< H.H 0.H 0.0 0.0 5.0 0.0 5.0 mwofi>pow uwngu 0.0 0.0 0A H20 04 5A 0.0 unmannfldfimusm 00d 33080.80 0.0 0.« 0.H 0.H 0.H 5.« «.« PfiXfifidflu 0.0a 0.0H 0.«H 0.«H «.0fi 0.«fl 0.0a mposnoaa,vmpnanH 98 030005 mHaowlmuudH HQ 0.0 0.0 0.0 H.« 0.« 0.« H.0 nnquV mcoom 3:00.80 and 3023.8: .850 0A0 H.H 5.0 0.0 «.H H.H 040 mmoEw 0.0 0.0 «.« H.0 0.« 0.H 0.0 apoHU 0.H 0.H 0.H 0.H 0.« 5.H 0.H mcanpoao . «.H. 0.« H.H «.H «.H 0.0 0.H naHmug xuoz H0002 5.« 5.« 0.0 0.0 5.0 0.0 H.0 pnwfla can Hmsm 0:000 00380 £52 03300 50900 0.03000 "300.0 00:03 and? 03000 000me 0:0 £2000 0:0 0830 05 5.30.0 05 000000 00033 3930 2330 00.30 0885 how 882 aaomvhpflfiahbo mm eesspssmssm msHe> eessefiemssm esse> sesseesmnsm messesv 0.30.8 000345 5: 5 5 Isa/m Ihomno I I 18.05 955 .5 05.5 00 .5 0565.508 0098.5. 0352 53005560 mezmsHmmaxm on5<>mmmmo omms zH mmaeHaoszoo.emms mos m05<255mm 0050:0000 H . m @1229 66 .hhpmncnw oddomldaasm wouMOfiudH Hmmo 0000050000 00005000 00 00300.3: 05 05.000008 000003093 0.50000 5 0003 63030.00 0 93 0503980 5 000: 0a ado 0005. 0000003008 0008.300 0:0 00 0000 0.0000000 00.0 0000500000 5 0000.052a .8000000 09 you :03 and 0800 05 5.0800 «00.0.5000 0000:3000 5000000 0.00 0003030090 0000 o: 00:30 .300 0950 "8080 Awm.mvv Amw.mhv Ah®.hwv Avm.mmav Av®.v A¢O.Hv AmN.HV Aam.Hv MCH>00 mnH. mvm. QQN. mom. www.mHI mwm.HwI mnm.moHI Hmm.bmal mmo.l NM®.I Nww.l 0H0.HI m®.® mm HdcofiuSufipmcH 005.000 000.000 A50.000 000.050 000.0 000.0 005.0 000.00 mom. NmN. CNN. HmH. hon.mHI mwfi.ml me. wov.HN H00. NOH.H mvm.H ONQ.H mm.om m mafinpoHo AON.mmv Amm.mmv Awm.mhv AVN.ONHV Avm.v Ammw.v ANO.HV Am®.Hv oAHmm moH. NmO. mOH. voa. N©H.OH Nvm.@v Hmh.H® mam.mHH NNH. mwm. mmH.H mON.N bN.H bmH x003 H0003 mm. Ho. mmoo. mmoooo. mm. Ho. mmoo. mmoooo. mm. Ho. mmoo. mmoooo. A000» 00m 0305 0003000950 051$ 000330230 03g 000330920 00:00av . 002000 000003 >\H a :H Iamaxm 100090 I I H308 003 00 95.5 mm .00 0000003000 0w000>< 00952 3009080 mezmszmmNm ZOHH<>mmmmO OMMN zH mmHBHQOEEOU Bmmh mom mm8<2namm mm9m23. «3. a: o3. 82. n2. o8.” Sag sag ~24 25. «a: 83.89.— So. So. one. o8 as. as. Ev. 8v. E”. as. 88. Ba. 88.5 8.45 n8. 58. 58 n8. «.8. 88. as; 08. Sad 384 854 m8.“ :38 3.5 HP. «2.. E. «2. 8s. 8m. 86.“ 884 So: m8. 88.” Bag d5: 8:68 8:88 8:58 8:68 8:68 8:68 56:68 8:68 £52 5888 5:5 28:. 552 5858 3:5 .55. 288 Ba Ea .98 28 288 288 98 98 Ba Ba 6:68 6938 883m in $85... 8868 898 «8:3: 581.". 55m $88 888 883 g mag—H 868 3H 8: 8:33 52.5 lawnmo hm nflnbomo mananooaioo mOh Mlbmzoo OF mNHBHmZHQOHA AdZuOfldfl n24 mNHBHOHBm< N. o "H.542. 100 .38 555 ”25m omo. mb.b oH. moo. ob.b maaomo pmmaMfim omo. mH.w Va. ooo. oH.w mmafiomo spud: oqw_nnawam Hmo. oo..m.m E . moo. ovo mwflomo 585mm . can 5me omo. obo 3. moo. Saw wmfiooo 5.3m onwonphgpm woo. oo.m vH. Hoo. No.m mmafiomo cafinp can 288m omo. mm.m ma. ooo. om.m mawomo pmmsca Amongv , Amazomlaomnmmv H338 ~85 Amfiocomdv 8“ 953% mmfiaoé . Ammaoflv Agfiugomv was p.895 no...” 398.300 awfimnom H333 Hons 8503 mmfio NEOUZH rm WMDBHQZHQNH ho NZOHA A8... unoomm... _.. .3 a. .ma ... .NH ... .3 * .oH * .m m i .w J a. .h I a. .o 1 198 .m m monumhdma .v . 8088. .m mmzoumZ .m "5808..“ .588th A occwoycm A .m.. :2 .0H J 3mm .m . So 53 .5 m H H \\\\\\ 820%?»me p \ \WWNN €958 ..u. m :mE .a. m. $281625 £2. .m s . 82 .4 828.3 8E A 350 3 woman 8.80 3 . @320 33 13m 33 LE . Boo H809 fig: :53 .895 53.6 «moo H38. ”:35 -55 89E 53.5 633 .HO 3 \ \ >683. E. \ \ $8883 830.288 53 amnesz 33o: >5 P8009. . b ... .vH _ _ 3. .MH 1 ... .NH 1. .HH a. .0." ..u .m m 4 ... .m m * .n I * ..c 1 93w ..u. w monumhmuwo .v 8038. .m $93.32 .N 2338.5 Eéumofim \ \ \ \ \ \ ocmmeom .H 5 .NH 1 .HnHI l .2 J .0 m. “Ham .m H6 Sam .5 m H H \ NR \ 898 Ear: .m p \ \ x N \ flanged .m m . :me .v 1. «Sam 829 2.0.2 .m m. 8E .m 8388a 8E .H 85o 3 83.0 3:8 3 888 |..|.I|., .33 13a .33 LE 88 H38. :5 -550 283 53.5 $8 H38. 55 -53 895 53.6 28H H0 3. \ \ >856 E \ \ .2882 833.889 8: 116 NJALA UNIVERSITY COLLEGE (University of Sierra Leone) RURAL EMPLOYMENT RESEARCH PROJECT Consumgtion Study Form RER/C—Z SECTION I: I D E N T I F I C A T I O N 1. Enumeration Area Number 2. Chiefdom 3. District 4. Name of Holder 5. Locality 6. Holder Number 7. Interviewed By 8. Date 197 MONTH OF: 11L7 SECHWII-WOFWWI’EBMM mar: (Emluding Food, Drinks, and Tobacco) Descript ion of Quantity Total Value Item Purchased (Give Unit) (Be Specific) (a) (d) Household Items 1 . APPENDIX 2 INDEXING PROCEDURE USED TO FILL IN MISSING DATA 118 119 APPENDIX 2 INDEXING PROCEDURE USED TO FILL IN MISSING DATA Where observations on a household's consumption ex- penditure patterns for a particular month were absent or insufficient,1 the indexing procedure given below was used to estimate expenditure levels for various commodi- ties. Households that failed to meet the minimum data requirements of three months' data from RER/Cl and three months' data from BER/C2 were dropped from the sample be- fore indexing was performed. At the time of indexing, the data file contained observations on expenditures by each household on each of 112 commodity-origin categories and the number of days in each month for which data from the short reference period questionnaire (BER/Cl) were present. Data from RER/Cl and RER/CZ were separate. To lessen com- ,putational expense, the 112 commodity-origin categories were grouped into three large categories: rice, other food, and nonfood; and within each of these, the seasonal pattern of expenditure was assumed to be the same for all commodities. Sets of monthly indices for the relevant year2 were calculated for each large commodity category for each research region for both RER/Cl and BER/C2 data. 1Data for a month were considered insufficient if data from RER/Cl covered less than three days. 2May 1974 - April 1975. 120 The first step in the indexing procedure was to "puff up" the data from BER/Cl so that it represented monthly expenditure levels. If, for example, seven days' expen- ditures were recorded in September for a household, the sum of observed expenditures for that month for each commodity was multipled by 30/7, while the quantity 30/5 would be used for a household with only five days of ob- servation. By puffing up the data, expenditure totals were given in monthly terms and were comparable for all households. This process was not necessary for BER/CZ data since it was already in monthly form. Next, the indices for each commodity group were cal- culated in the following manner. First, the average ex- penditure on a commodity group for the 3th month in the 1th region, 513, was determined using the formula: _ N13 eij = E ehij /Nij (A.2.l) h-l where ehij = expenditure on the commodity roup in question by the hth household in the i h region during the Jth month, household h being one for which valid data are present Nij = is the number of households in region i for which valid data is present for month 3. Monthly average expenditures were then summed over the year as in equation A.2.2, 12 E = 26 , (A.2.2) 1 1_1 13 121 to obtain the average annual expenditure on the commodity group in region 1, E1. Monthly indices for the 1th region, Iij’ were then calculated using equation A.2.3.: Iij = eij/Ei (A.2.3) The procedure used constrains the sum of the monthly in- dices to equal unity. The determination of the adjusted total expenditure on a particular commodity-origin category by the hth household in region i, Tfii’ for households with missing data was the next step in the process. Th1 was calculated using equation A.2.4., k * = — Thi [:1/(1 jgllij):] Thi , (A.2.4) where Thi = unadjusted total expenditure on the commodity by household h in region 1 k 2 I = the sum of the indices for the appropriate j=1 13 large commodity group for months with miss- ing or inadequate data. Finally the estimated value cash expenditure on the com- modity in question for missing month m by household h in region 1, tfiim’ was calculated using equation A.2.5.: tit him = Tfiilim . (A.2.5) BIBLIOGRAPHY BIBLIOGRAPHY "African Rural Employment Study: Progress Report and Plan of Work, 1972-1976." 1974. African Rural Employ- ment Working Paper No. 1, Department of Agricultural Economics, Michigan State University, East Lansing, Michigan. Allen, R.G.D. and Bowley, A.L. 1935. Family Expenditure. London: Staples. Ballentine, J. Gregory and Soligo, Ronald. 1975. "Con- sumption and Earnings Patterns and Income Redistri- bution." Program of Development Studies Paper No. 67, Rice University, Houston, Texas. Central Statistics Office. 1968. Household Survey of the Western Area: Final Report. Freetown: Government of Sierra Leone. 1971a. Household Survey of the Eastern Pro- vinceep Urban Areas: Final Report. *Freetown: Govern- ment of Sierra Leone. 1971b. Household Survey of the Northern Pro- vincesL Urban Areas: Final Report. Freetown: Gov- ernment of Sierra Leone. 1971c. Househgld Survey of the Southern Pro- vince, Urban Areas: Final Report. Freetown: Gov- ernment of Sierra Leone. 1972a. Household Survey_pf the Rural Areas of theiProvinces: Final Report. Freetown: Government of Sierra Leone. _ . 1972b. National Accounts of Sierra Leone, 1964/65 to 1970771. Freetown: Government of Sierra Leone. Cline, William R. 1972. Potential Effects of Ingome Re- distribution on Economic Growth. New York: Praeger Publishers. 122 123 Desai, B.M. 1972. "Analysis of Consumption Expenditure Patterns in India." Occasional Paper No. 54, Depart- ment of Agricultural Economics, Cornell University, Ithaca, New York. Dutta-Roy, D.K. 1969. "The Eastern Region Household Budget Survey." Technical Publication Series No. 6, Institute of Statistical, Social and Economic Re- search, University of Ghana, Legon, Ghana. and Mabey, S.J. 1968. "Household Budget Sur- vey in Ghana." Technical Publication Series No. 2, Institute of Statistics, University of Ghana, Legon, Ghana. Fatoo, Habib. 1977. "Macro-Economic Analysis of Output EmploymentzuuiMigration in Sierra Leone." Ph.D. Dissertation, Michigan State University (forthcoming). Food and Agriculture Organization. 1972. "The Impact in Demand of Changes in Income Distribution." Monthly Bulletin of Agricultural Economics and Statistics 21: 1—11. Friedman, Milton. 1957. A Theory of the Consumption Function. Princeton: Princeton University Press. Hay, George. 1966. "Computing Income Elasticities in Nigeria." EDI Working Paper No. 8, Economic Develop- ment Institute, University of Nigeria, Enugu, Nigeria. Holmberg, Johan. 1973. Surveyoftgonsumption Patterns in Etheya Extension Area (CADU Publication No. 90). Addis Ababa: Government of Ethiopia. Houthakker, H.S. 1957. "An International Comparison of Household Expenditure Patterns, Commemorating the Centenary of Engel's Law." Econometrica 25: 532-551. Houyoux, Joseph. 1973. Budgets Menagers, Nutrition et Mode de Vie a Kinshasa. Kinshasa: Presses Univer- sitaires du Zaire. Howe, Charles W. 1966. "The Use of Household Budget Surveys in Economic Planning in East Africa." Bulletin of the Oxford Institute of Economics and Statistics 28: 199-209. Hymer, Stephen and Resnick, Stephen. 1969. "A Model of an Agrarian Economy with Nonagricultural Activities." The American Economic Review 59: 493-506. 124 International Labour Office. 1970. Towegds Full Employ- ment. Geneva: International Labour Office. Johnston, J. 1963. Econometric Methods. New York: McGraw—Hill. ' Klein, Lawrence Robert. 1962. An Introduction to Econo- metrics. Englewood Cliffs: Prentice-Hall. Kmenta, Jan. 1971. Elements of Econometrics. New York: Macmillan Publishing Co., Inc. Kuznets, Simon. 1976. "Demographic Aspects of Size Distribution of Income: An Exploratory Essay." Economic Develgpment and Cultural Change 25: 1-94. Land, James W. and Soligo, Ronald. 1971. "Income Dis- tribution and Employment in Labor Redundant Economies." Program of Development Studies Paper No. 9, Rice University, Houston, Texas. Lele, Uma J. and Mellor, John W. 1971. "The Political Economy of Employment Oriented Development." Occa- sional Paper No. 42, Department of Agricultural Economics, Cornell University, Ithaca, New York. Leser, C.E.V. 1963. "Forms of Engel Functions." Econometrica 31: 694-703. Leurquin, Phillippe. 1960. Le Niveau de Viedes Popula- tipps Ruralesgu Ruanda-Urungi. Louvain: Institut de Recherches Economique et Sociales (IRES), Univer- site de Leopoldville. Levi, John. 1976. "African Agriculture: Economic Action and Reaction in Sierra Leone." Oxford Institute of Agricultural Economics, Oxford, England. Liedholm, Carl and Chuta, Enyinna. 1976. "The Economics of Rural and Urban Small-Scale Industries in Sierra Leone." African Rural Economy Paper No. 14, Depart- ment of Agricultural Economics, Michigan State Uni- versity, East Lansing, Michigan. Liviatan, Nissan. 1961. "Errors in Variables and Engel Curve Analysis." Econometrica 29: 336—362. Massell, Benton F. 1969. "Consistent Estimation of Ex- penditure Elasticities from Cross-Section Data on Households Producing Partly for Subsistence." Review of Economics and Statistics 51: 136-142. 125 and Heyer, Judith. 1969. "Household Expenditure in Nairobi: A Statistical Analysis of Consumer Be- havior." Economic Development and Cultural Change 17: 212-234. Massell, Benton F. and Parnes, Andrew. 1969. "Estima- tion of Expenditure Elasticities from a Sample of Rural Households in Uganda." Bulletin of the Oxford Institute of Economics and Statistics 31: 313-329. Mellor, John W. 1976. The New Economics of Growth. Ithaca: Cornell University Press. and Lele, Uma J. 1971. "A Labor Supply Theory of Economic Development." Occasional Paper No. 43, Department of Agricultural Economics, Cornell Uni- versity, Ithaca, New York. 1972. "Growth Linkages of New Foodgrain Tech- nologies." Occasional Paper No. 50, Department of Agricultural Economics, Cornell University, Ithaca, New York. Modigliani, Franco and Brumberg, Richard. 1954. "Utility Analysis and the Consumption Function: An Interpre- tation of Cross—Section Data," Post Keynesian Eco- nomics. Edited by Kenneth K. Kurihara. NewBruns- wick: Rutgers University Press._ Mukhrjee, M. and Rao, D.S. Prasada. 1972. "Non—monetized Consumption in the Analysis of Consumer Behavior." Economic Affairs 17: 49-64. Ostby, Ivar and Gulilat, Taye. 1969. "A Statistical Study of Household Expenditure in Addis Ababa." East African Economic Review 1: 63-74. Paukert, Felix. 1973. "Income Distribution at Different Levels of Development: A Survey of Evidence." International Labor Review 108: 97-125. Prais, S.J. and Houthakker, H.S. 1971. The Analysis of Family Budgets. Cambridge: Cambridge University Press. Riemenschneider, Charles H. 1976. "The Use of the Gini Ratio in Measuring Distributional Impacts." M.S. Thesis, Michigan State University. Sinha, R.P. 1966. "An Analysis of Food Expenditures in India." Journal of Farm Economics 48: 113-123. 126 Snyder, Donald W. 1971. "An Econometric Analysis of Household Consumption and Savings in Sierra Leone." Ph.D. Dissertation, Pennsylvania State University. Soligo, Ronald. 1973. "Factor Intensity of Consumption Patterns, Income Distribution and Employment Growth in Pakistan." Program of Development Studies Paper No. 44, Rice University, Houston, Texas. Spencer, Dunstan S.C. and Byerlee, Derek. 1977. "Small Farmers in West Africa: A Descriptive Analysis of Employment, Incomes and Productivity in Sierra Leone." African Rural Economy Working Paper No. 19, Depart— ment of Agricultural Economics, Michigan State Uni- versity, East Lansing, Michigan. Stevens, Robertl). 1965. "Elasticity of Food Consumption Associated with Changes in Income in Developing Countries." Foreign Agricultural Economic Report No. 23, Economic Research Service, U.S. Department of Agriculture, Washington, D.C. Summers, Robert. 1959. "A Note on Least Squares Bias in Household Expenditure Analysis." Econometrica 27: 121-126. Sunman, Tuncay Mustata. 1974. "Short Run Effects of Income Distribution on Some Macroeconomic Variables: The Case of Turkey." Ph.D. Dissertation, Rice University. IIHIIMIHIllllilllllllllillllllIIHIIINIHIIlllllllllllllll 2931 190429 9619