Ie cosy SWP364. Income, Consumption and Poverty in Thailand, 1962/63 to 1975/76. November World Bank Staff Working Paper No. 364 r e.

Transcription

1 Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Income, Consumption and Poverty in Thailand, 1962/63 to 1975/76 SWP364 ELUUMSTUCTI'. if'1)m: DEGLPb1ENTL World Bank Staff Working Paper No. 364 r e. November 1979 Prepared by: Oey Astra Meesook Development Economics Departmenl Development Policy Staff Copyright ( 1979 The World Bank 1818 H Street, N.W. Washington, D.C , U.S.A. F The views and interpretations in this document are those of the author and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf. Ie cosy

2 I

3 The views and interpretations in this document are those of the author and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf. WORLD BANK Staff Working Paper No. 364 November 1979 INCOME, CONSUMPTION AND POVERTY IN THAILAND, 1962/3 TO 1975/6 This study uses data from three household sample surveys for Thailand to examine the relationship between regional, as well as urban-rural, disparities in household incomes and the very rapid growth for the economy as a whole over the periodci962/3-to<1975/6;' to determine what has happened to the incidence and distribut-i-onof the-p6or population over this period; and to see whether the government has contributed towards a greater equality in living standards across different population groups through the provision of social services. Data from the national accounts are used to supplement the household survey data. The study concludes that the growth of the national product has been accompanied by a very rapid increase in the incomesf_both urban and..xural households. After pro ccount the changes in definition of urban and rural areas across surveys, the study finds that, contrary to what is generally thought, the disparity in incomes between urban and rural households in the major regions of hafflanid has declined over the period considered. In addition, the gap in rural incomes between the Northeast, which is the poorest region, and the more prosperous Central region has been reduced. The incidence of poverty has declined markedly, although the poor population remains concentrated in rural areas of the North and Northeast. The findings are not in complete agreement with those obtained from national accounts; the study discusses the appropriate use of national accounts data for Thailand in attempting to interpret movements in household incomes over time. It attributes the rapid growth in the agricultural sector to an expansion in cultivated area, improvements in the relative prices of agricultural products, aid switches by farmers into more profitable crops. Thus agricultural growth has directly contributed to both the overall growth of incomes and a reduction in disparities between sectors and across regions. In this limited sense, there has been no trade-off between economic growth and a more equal distribution of income among the major population groups considered. In attempting to relate the issues of growth and equity, it is therefore important to understand the very nature of the growth process itself. Prepared by: Oey Astra Meesook Development Economics Department Development Policy Staff Copyright Q 1979 The World Bank 1818 H Street, N.W. Washington, D.C , U.S.A.

4

5 INCOME, CONSUMPTION AND POVERTY IN THAILAND. 1962/3 TO 1975/6 TABLE OF CONTENTS Page No. INTRODUCTION I. TRENDS IN HOUSEHOLD INCOME AND CONSUMPTION LEVELS OVER TIME II. TRENDS IN URBAN-RURAL AND REGIONAL DISPARITIES III. TRENDS IN THE SIZE AND DISTRIBUTION OF THE "POOR" POPULATION *.....*..*.. 49 IV. CHARACTERISTICS OF POOR AND NONPOOR HOUSEHOLDS CONCLUSION APPENDIX

6

7 LIST OF TABLES Page No. Table 1.1 Table 1.2 Table 1.3 Table 2.1 Real Increases in Per Capita Income and Consumption Expenditure, by Region and Area, Thailand, 1962/3 to 1975/ Real Increases in Consumption Expenditure Per Capita, by Region and Area, Thailand, 1962/3 to 1975/6. 11 Percentages of Households Owning Selected Consumer Durables, by Region and Area, Thailand, 1962/3 to 1975/ Rural Income/Capita and Expenditure/Capita as Percentages of Urban Levels, by Region, Thailand, 1962/3 to 1975/ Table 2.2 Household Total Income, by Region, Thailand, 1962/3. to 1975/ ,..,, 20 Table 2.3 Table 2.4 Table 2.5 Table 2.6 Table 2.7 Table 2.8 Household Total Income for Urban and Rural Areas by Region, Thailand, 1962/3 to 1975/6..., Comparison of Annual Growth Rates of Per Capita Household Income and Per Capita Gross Domestic Product at Current Market Prices, Distribution of Gross Domestic Product by Region, Thailand, 1962, 1968, and Annual Increases in Gross Domestic Product by Region, Thailand, and Wholesale Prices and Areas Planted in Selected Crops by Region, Thailand, 1962 to Proportion of Households in the Agricultural Sector by Region and Area, Thailand, 1975/ Table 2.9 Proportion of Household Total Current Income from. Agricultural Activities, by Region and Area, Thailand, 1975/ Table 2.10 Composition of Household Income: Percentages of Wage Income, Self-Employment Income, Other Money Income and Own Consumption by Region and Area, Thailand, 1962/3 to 1975/ Table 2.11 Average Values of Socioeconomic Indicators, by Region, Thailand

8 Page No. Table 3.1 Table 3.2 Table 4.1 Table 4.2 Table 4.3 Table 4.4 Incidences of Poverty, by Region and Area, Thailand 1962/3, 1968/9 and 1975/ Distribution of the Poor Population by Region and Area, Thailand, 1962/3, 1968/9 and 1975/ Distribution of the Poor by Socioeconomic Class, Region and Area, Thailand, 1975/ Incidence of Poverty by Socioeconomic Class, Region and Area, Thailand, 1975/ Average Household Income Per Capita by Socioeconomic Class, Region and Area, Thailand, 1975/ Some Socioeconomic Characteristics of Households, by Quintiles of Households Ordered by Monthly Per Capita Consumption Expenditures, Villages in Northeast and Center, Thailand, 1975/ Table A.1 Table A.2 Comparison of Sampling Procedures Used in Household Surveys for Thailand Comparison of Definitions Used in Household Surveys for Thailand Table A.3 Household Total Income by Area, Thailand, 1975/6.. 9 Table A.4 Table C.1 Table D.1 Table D.2 Distribution of the Population Across Areas, by Region, Thailand, 1960 to 1975/ Per Capita Household Income, by Region and Area, Thailand, 1975/ Household Total Income and Consumption Expenditures, by Region and Area, Thailand,'1962/3 to 1975/ Household Per Capita Income and Consumption Expenditures, by Region and Area, Thailand, 1962/3 to 1975/

9 INCOME, CONSUMPTION AND POVERTY IN THAILAND, 1962/3 TO 1975/6 INTRODUCTION Over the past two decades, Thailand has unquestionably gone through a period of very rapid economic growth in aggregate terms. At the same time, concern has often been expressed that this growth may not have been equitably distributed, in the sense that the gains from growth are believed not to have reached the poorer population groups to the same extent as the better-off ones. Specifically, it is commonly thought that over this period the disparity in incomes between urban and rural areas has widened and that the poorer regions, with special reference to the Northeast, the poorest major region in Thailand, have lagged behind the better-off regions. These conclusions rely heavily on national accounts estimates of gross domestic product broken down by sector and region. The conventional wisdom would thus seem to be consistent with the view that rapid economic growth cannot be achieved without a deterioration in the distribution of income, at least insofar as the discussion is confined to population groups classified by region and area of residence. In this context it is desirable to take a closer look at the relationship between growth and equity in the recent experience of Thailand. The data base for the country is sufficient for an examination of the following questions. First, over the past two decades, what has happened to income disparities between urban and rural areas and across regions. Second, what has been the nature of the economic growth in Thailand over this period, and in what way can it be linked with the changing disparities which are observed. Third, what has happened to the incidence and size of the poverty population over time and what are the characteristics of

10 -2- this group. Fourth, what role has the government played in the reduction of disparities in living standards across population groups. This paper is essentially concerned with the level of well-being of the Thai population. It accordingly attempts to understand the relationships between the growth process of the economy as a whole and the changes in the level of material welfare of households over time. The objectives are to establish the extent to which the standard of living of the population has risen since the early 1960s, to study urban-rural and regional disparities and the trends in these over time, and to understand the nature and characteristics of poverty among the Thai population. To this end the paper examines what has been happening to household income and consumption of major population groups from the early 1960s to the mid-1970s and, to the extent possible, attempts to characterize households at different income or consumption levels and different socioeconomic classes at a given point in time. The main sources of income and consumption data in Thailand are the various household surveys of incomes and expenditures, which have been conducted by the National Statistical Office since the early 1960s. This paper uses the published results of the Household Expenditure Survey of 1962/3 and the Socioeconomic Surveys of 1968/9 and 1975/6, as well as special additional results obtained from the data tapes of the 1968/9 and 1975/6 Socioeconomic Surveys /1. On the whole, these three surveys are comparable. /1 Household surveys were also undertaken in 1971, 1972 and 1973, with two regions being covered in each year. Results from the surveys for the three years, however, cannot be pooled into a national survey because of large differential movements in producer and consumer prices over the three-year period.

11 -3- Part A of the Appendix to this paper gives a comparison of the sampling procedures and definitions used in the three surveys and discusses the nature of the problems encountered with these surveys which need to be borne in mind in interpreting the results presented in this paper. The first section examines trends in the levels of household income and consumption over time. The observed increases in the overall standards of living of major population groups are supported by information on household consumption of different categories of commodities and household ownership of selected consumer durable goods. In section two, the questions of urban-rural and regional disparities are taken up. Evidence from national accounts data are used both to support and to explain the development over time of these disparities. This leads rather naturally into a discussion of the increasing importance of nonagricultural activities in rural areas. The disparities in the provision of social services across regions and areas are then looked at. The trends in the size and distribution of the poor population across regions and areas over time are examined in section three. Finally, in section four we focus on a comparison of selected socioeconomic characteristics of poor and nonpoor households for 1975/6. The paper has confined itself to an analysis which relies nearly exclusively on official tabulations published by the National Statistical Office. Additional results will be forthcoming from the data tapes of the 1968/9 and, more importantly, the 1975/6 Surveys and promise to bring further insights into other issues which are related to the ones discussed in this paper.

12 - 4 - I. TRENDS IN HOUSEHOLD INCOME AND CONSUMPTION LEVELS OVER TIME There is absolutely no question that the national product of Thailand has been growing at a remarkably rapid rate since the early 1960s. Over the decade 1960 to 1970, gross domestic product at constant 1962 prices grew at an annual rate of 8.0%. Between 1970 and 1976, there was apparently a slowdown, but the annual growth rate of gross domestic product over this period was still 6.5%./l The growth of the agricultural sector was evidently slower; its gross domestic product at constant 1962 prices grew at an annual rate of 5.4% between 1960 and 1970, and 4.1% between 1970 and 1976./2 In any case, since the population grew at a significantly lower rate, an estimated 3.1% per annum between 1960 and 1970, and lower since then, the conclusion must be that per capita income has gone up considerably over these periods, even ln the agricultural sector where the growth rate was lower. Thus, economic growth in the aggregate has not in itself warranted much concern; rather, the issue is that of the sharing of this impressive growth. It is generally agreed that the urban population has experienced a rise in its standard of living; to what extent the rural population has done so is less certain. While incomes have been rising rapidly in the Central plain, the question can be raised as to what has happened to the levels of income in the other regions, in particular the Northeast. /1 Data on gross domestic product are also available as far back as 1970 in constant 1972 prices. They show an annual growth rate of 6.0% between 1970 and /2 In constant 1972 prices, agricultural gross domestic product grew at an annual rate of 4.6% between 1970 and 1976.

13 - 5 - In this section we look at how the increase in aggregate output since the early has translated into changes in the income levels of households and what differences can be found across different regions and areas of the country. The objective at this point is simply to establish whether or not there has been a rise in the standard of living of major population groups, using data from the household sample surveys of 1962/3, 1968/9 and 1975/6. Table D.1 in the Appendix gives average household money income, own consumption, total income, cash expenditures and total expenditures by region and urban-rural area of residence./l Because of the differences in the definitions of urban and rural areas used in the Household Expenditure Survey, 1962/3 and the Socioeconomic Survey, 1968/9, the urban and rural income and consumption figures from these two surveys should not be compared. Sanitary districts were classed with municipal areas as urban in the former survey, while they were classed with villages as rural in the latter survey. Since the 1975/6 Socioeconomic Survey results are given separately for municipal areas, sanitary districts and villages, it has been possible to /1 See Part B of the Appendix for a discussion of the reliability of the household sample surveys used in this report.

14 combine these appropriately for the purposes of comparison with the two earlier surveys. Thus, one set of figures for 1975/6 is to be compared with the 1962/3 figures and the other set with the 1968/9 figures, but any comparison between the 1962/3 and 1968/9 figures is incorrect. A detailed explanation of this problem and the nature and extent of the biases involved in making inappropriate comparisons are given in Part A of the Appendix. Although the household is the entity on which we wish to focus our interest, our ultimate objective is to arrive at some measure of the standard of living of the individual members of households. Thus, for purposes of comparing household incomes across regions and areas and over time, it is preferable to use income per household member figures which take into account the differences in average household size. /1 These are presented in Table D.2 in the Appendix. Concentrating for the moment on changes over time, we see that large increases took place in per capita income and consumption over the period considered, 1962/3 to 1975/6, for all regions. Annual growth rates in total income and total consumption per capita for the periods 1962/3 to 1975/6 and 1968/9 to 1975/6 are given in Table 1.1. Corresponding increases in the consumer price index for urban areas by region are also given, as well as the resulting annual growth rates in per capita income and consumption in real terms. /1 See Appendix C for a discussion of how a comparison of household incomes across regions and areas of residence based on the per capita concept would be affected if a per adult-equivalent concept is used instead.

15 Table 1 *1: REAL INCREASES IN PER CAPITA DINCOME AND OWNSUNPTIOI MWENDIURXE. BY REGION AND AREA, THAILANID, 1962/3 ID 1975/6 CE per annum) Urban/a Uurcal/b Urban and Rural Whole Whole Whole Northeast North South Center.jc Kingdom Northeast North South CauterLc Kingdom BangkokLd Northeast North South CenterLe Kingdom Total income per capita /6 Nominal increase Increase in Consumer Price Index Real increase /9-1975/6 Nominal increase Increase in Consumer Price Index Real increase Total expenditures per capita 1962/ Nominal increase Increase in Consumer Price Index Real increase /6 Nominal increase Increase in Consumer Price Index Real increase N unlicipal areas and sanitary districts for 1962/3 and corresponding figures for ; municipal areas for 1968/9 and corresponding figures for 1975/6. Lb Villages for 1962/3 and corresponding figures for 1975/6; sanitary districts and villages for 1968/9 and corresponding figures for 1975/6. /c Excludes changwats Samut Prakan, Nonthaburi. and Pathum Thani in 1975(6. L4 Includes changvats Bangkok and Thonburi in all years and. in addition, changvata Saezut Prakan, Nonthaburi and Pathum Theni in 1975/6. Source: A.PPeadix Table D.2. Increases in Consumer Price Index aer calculated from Bank of Thailand. Monthly Builletin, various 'soluima. Table V.14.

16 - 8 - Two things need to be said about the price deflators. First, the consumer price indices are only available for urban areas of different regions. These have had to be applied to the corresponding rural areas as well for lack of any alternative. Given reasonably efficient markets, it is likely that price movements were similar between urban and rural areas, even though there may be systematic differences in the levels at any given point of time. Second, in the construction of these price indices, the expenditure weights used were obtained from the Household Expenditure Survey, 1962/3. Results from the Socioeconomic Survey, 1975/6 indicate that consumption patterns of households have changed considerably between the two survey periods, as would be expected, and that, if the expenditure weights for the later period are used, then a slower rate of price increase is indicated. At this stage, we do not know the extent of the variation in differences in price increases across regions under the two sets of expenditure weights. Nevertheless, it is likely that real incomes increased more rapidly than indicated here in all regions. Over the thirteen-year period, 1962/3 to 1975/6, real increases in total income per capita were experienced by both urban and rural households in all regions. Two extremely important conclusions emerge. First, with the exception of the rural South, the real increase in per capita income has ranged from 1% to 3% per annum, representing a substantial overall increase considering that it was sustained over so many years. Again with the exception of the South, the rural increases were significantly higher than the urban ones. The reasons behind these differentials in growth rates across regions and areas will be taken up in the next section.

17 - 9 - When we look at real increases in per capita income over the period 1968/9 to 1975/6 we find that, excepting the rural South, they are much lower than for the whole period covering 1962/3 to 1975/6. In fact, they are even negative for all urban areas other than in the Central region, as well as for the rural North. Two explanations can be offered. First, it could well be that real incomes grew less rapidly and even declined in some cases between 1968/9 and 1975/6. The higher rates of increase between 1962/3 and 1975/6 would then represent averages of even more rapid rates of increase over the earlier period, 1962/3 to 1968/9, and much lower rates of increase, or rates of decrease, over the later period. Second, there is the complication of the change in definitions of urban and rural areas, with sanitary districts being classed as urban in the 1962/3 Survey, but rural in the 1968/9 Survey. If increases in real incomes for sanitary districts were larger than those for municipal areas, but smaller than those for villages, then calculations in which sanitary districts fall into rural areas would yield lower rates of growth of income for both urban and rural areas than if they had been classified as urban. This could explain in part the lower growth rates of incomes in the later period. Turning to real changes in per capita consumption levels, these are seen to have been positive for both urban and rural areas for all regions and for both periods considered. With the exceptions of the urban

18 South and Center, the increases in consumption expenditures were smaller than those in incomes for the 1962/3-1975/6 period; the opposite was true for the 1968/9-1975/6 period. In Table 1.2, per capita consumption is broken down into major commodity groups such as food, housing and so on./1 To get consumption /1 Consumption here includes own consumption as well as cash expenditures. increases in real terms, the figures are deflated by the consumer price indices for major commodity groups for urban areas. These apply to urban areas for the whole country and thus can only be taken as a rough measure of the extent of price increases in any particular region. The table gives figures for both the total and annual percentage increases over the 1962/3-1975/6 period. These tend to support the general conclusion that standards of living of Thai households have gone up substantially since the early 1960s. Taking the category of most interest, expenditures on food and beverages ment up in both-urban and rural areas of all regions, the increases being

19 Table 1.2: REAL INCREASES IN CONSUMPTION EXPENDITURE PER CAPITA, BY REGION AND AREA, THAILAND, 1962/3 TO 1975/6 (X per annum) Urban /a Rural /b Urban and Rural Expenditure Whole Whole Whole Group Northeast North South Center/c Kingdom Northeast North South Center/c Kingdom Bangkok/d Northeast North South Center Kingdom Food & 4.07 /e ' Beverages (0.31)/f (2.72) (0.58) (1.49) (1.11) (3.05) (7.59) (0.71) (2.54) (2.51) (-1.03) (2.91) (3.63) (0.91) (2.48) (2.17) Clothing (-0.36) (-1.48) (0.40) (1.96) (0.21) (2.25) (1.44) (0.04) (2.09) (1.52) (1.27) (2.15) (0.99) (0.25) (2.17) (1.47) Housing (0.87) (0.72) (-0.06) (1.27) (4.16) (0.04) (-0.81) (-3.96) (0.56) (3.85) (4.33) (5.38) (4.47) (1.25) (5.68) (4.94) Medical Personal (2.95) (4.48) (1.79) (4.60) (3.44) (6.60) (6.55) (2.74) (4.07) (4.78) (2.58) (6.15) (6.46) (2.78) (4.47) (4.71) Care Transportation (4.73) (6.31) (9.14) (11.55) (7.77) (7.69) (8.50) (8.44) (8.58) (7.97) (3.70) (8.03) (8.38) (8.73) (9.58) (8.10) Recreation, Reading & (-0.68) (5.03) (6.47) (5.07) (3.59) (2.82) (5.23) (3.45) (4.60) (3.78) (2.73) (2.66) (5.92) (5.36) (5.67) (4.91) Education Tobacco & Alcohol (2.66) (5.32) (4.17) (3.10) (3.64) (4.69) (6.35) (4.55) (5.80) (5.26) (2.92) (4.70) (6.26) (4.77) (5.27) (5.17) La Municipal areas and sanitary districts. /b Villages. /c Excludes changwats Samut Prakan, Nonthaburi and Pathum Thani in 1975/6. Ld Includes changvats Bangkok and Thonburi in all years and, in addition, changwats Samut Prakan, Nonthaburi and Pathum Thani in 1975/6. Le Total percentage increase. If Annual percentage increase is given in parentheses. Sources: Household Expenditure Survey, 1962/3, National Statistical Office, Regional Volumes, Tables 1.0 and 1.1. Socioeconomic Survey 1968/9, National Statistical Office, Table 4. Socioeconomic Survey 1975/6, National Statistical Office, Regional Volumes, Table 1. Price deflators are taken from Bank of Thailand Monthly Bulletin, August 1968, Table V.13; January 1973, Table V.14 and October 1978, Table V.13.

20 particularly worthy of note for the rural Northeast and rural North, the poorest areas in the country. Large increases were also recorded for the Central region and urban areas of the North. Clothing expenditures in real terms increased in similar fashion, with the exception of declines in the urban Northeast and urban North. Expenditures on housing showed only small increases or declines. However, expenditures on the remaining four categories, namely (a) medical and personal care, (b) transportation, (c) recreation, reading and education, and (d) tobacco and alcohol, with only one exception, showed large increases in real terms for both urban and rural areas of all regions. On balance the data are thus strongly suggestive of a rapidly rising standard of living in all parts of the country, although of varying extents. Taking the period 1962/3 to 1975/6, the Northeast appears to have done least well out of all urban areas, and the South out of all rural areas. The former case undoubtedly reflects the US presence in the Northeast in the earlier period and the subsequent withdrawal, while the latter case highlights the dependency of the economy of the rural South on rubber and reflects the impact of falling rubber prices over most of this period, a matter which will be taken up in greater detail in the next section.

21 Further corroboration of these results is given by data on household ownership of consumer durable items. Table 1.3 gives the percentages of households owning selected consumer durables in the three survey years, by region and area of residence. As before, the 1962/3 and 1968/9 figures can be compared with the appropriate ones for 1975/6; they should not themselves be compared with each other because of the differences between them in the definitions used for urban and rural areas. The table shows very clearly that there has been a dramatic increase in the ownership of consumer durable goods of households. For example, to take the most commonly owned item, the radio, the percentage of village households owning this in 1962/3 ranged from 5% in the Northeast to 37% in the Central region. The corresponding figures for 1975/6 were 66% and 82% respectively, with the position of the South relative to the Northeast having deteriorated, since its corresponding percentage increase was from 10% to only 64% over this period. The ownership data of other items show similar trends: households in both urban and rural areas are in possession of more and more different items of consumer durable goods. Rural households are beginning to be owners of such items as motor cycles, sewing machines and electric fans, in addition to the now nearly ubiquitous radio, while urban households are also rapidly moving into televisions, refrigerators and automobiles as well.

22 Table 1.3: PERCENTACES OF 8DUSE8OLDS OOIIING SELECTED WDISUKER DURABLES BY RCIOhN MD UIA. TA141SD, 1962/3 TO 1975/6 Urban /a Rural /b Urban and Rural Whole Whole Ibol Hortheat Worth Soutb Center L Kingdom Nortbe t North South Center KiUgdo. ngkok 1A Nortb_et Nortb Sootb Cnter Kir noadio 1962/ / / / Te l evio ion 1962/ M / / * / Automobile 1962/ / / / lotorcycla 1962/ / / / SevinA Mchine 1962/ / / / lef rirerator 1962/ / / / Electric Van 1962/ / / / La Mmicipal &rea. and sanitary districts for 1962/3 and correspondiag figure. for 1975/6; municipal arm for 1968/9 and correaponding figt4e for 1975/6. 1k ViUlase. for 1962/3 and correaponding figures for 1975/6; saaitary districta and villages for 1968/9 and correaoandig figures for L lancludcs changuati Saot Prakan. Nonthaburi and Pathus Than In 1975/6. Lu 7=1advc changwata Bangkok snd Thoaburi in all year and. in addition, cbaagata Sgamt Prakan. Nouthaburi and Patbum Thoai i 1975/6. Boucceot 8ousebold Zzeaditnre Survey. 1962/3, National Statistical Office. gional Volums_ Tabla 12. Socioccnnic Survar 1968/9, National Statistical Office. Tabla 9. 8lowcoc8ede Survey 1975/6. *atieol Statistical Office, Ragional Volumas Uortbe_t - Table 17; lortb Soutb - Tale 161 Citat - tabl 11 gbuk - T1ab 16.

23 Reliable information on the ownership of consumer durable goods is very simple to obtain, when compared with the task of getting accurate data on income and consumption expenditures. placed on the trends shown in Table 1.3. Great reliance can therefore be Ownership of consumer durables adds another dimension to the assessment of trends in living standards of households over time. The picture which emerges quite clearly is one of considerable improvement in the standards of living of the Thai people over the past decade and a half. This is in agreement with our findings based on per capita income and consumption expenditure data. To conclude, data from the household surveys for 1962/3, 1968/9 and 1975/6 on levels of per capita income and per capita consumption expenditures, both in the aggregate and broken down by major categories, supplemented by information on the ownership of consumer durables, provide conclusive evidence that the standards of living of the Thai population havebeen rising quite rapidly when considered in terms of broad population groups involving major regions and areas of residence. It is worth emphasizing that this has been the case for rural areas, and not just urban areas, and for the poorest region in the country, the Northeast, as well as the well-off Central region.

24 II. TRENDS IN URBAN-RURAL AND REGIONAL DISPARITIES One of the most widely debated issues in Thailand has been that involving urban-rural and regional disparities. It is generally believed that the income gap between urban and rural areas is widening and that regional disparities, for example the gap between the worst-off Northeast and the most prosperous Central region, are also increasing. In this section we use the household sample surveys to show that urban-rural disparities in household incomes have been decreasing over time and that, within the rural sector itself, there has been some narrowing of the gap between the Central region and the others. However, we also draw attention to the fact that this type of analysis may mask, and thus distract us from, some of the more significant changes that are taking place with regard to the income positions of households. For example, within the rural sector itself, the income gap between different groups which have been able to take advantage of new opportunities to different degrees has very likely widened. To look first at urban-rural disparities, Table 2.1 gives rural household income per capita and consumption expenditure per capita as percentages of the urban figures for each of the four regions, with Bangkok excluded./l As before, the 1962/3 and 1968/9 figures can in turn be compared with the appropriate 1975/6 figures, but not with each other. /1 Because of the change in definition of Bangkok region to include three additional changwats in 1975/6, it is not possible to provide a meaningful comparison between the incomes of Bangkok and other regions or other urban areas. The gap between Bangkok and the Central region widened between 1962/3 and 1968/9 but would naturally narrow between 1968/9 and 1975/6 because of the inclusion of the three changwats which are poorer and more rural in relation to Bangkok.

25 Table 2.1: RURAL INCOME/CAPITA AND EXPENDITURE/CAPITA AS PERCENTAGES OF URBAN LEVELS, BY REGION, THAILAND, 1962/3 TO 1975/6 (%) /a Northeast North South Center /b Total Income/Capita 1962/ / / / Total Consumption Expenditure/Capita 1962/ / / / /a Rural areas include villages for 1962/3 and corresponding figures for 1975/6; sanitary districts and villages for 1968/9 and corresponding figures for 1975/6. Urban areas include municipal areas and sanitary districts for 1962/3 and corresponding figures for 1975/6; municipal areas for 1968/9 and corresponding figures for 1975/6. /b Excluding Bangkok-Thonburi. Includes changwats Samut Prakan, NDnthaburi and Pathum Thani in 1962/3 and 1968/9; excludes them in 1975/6. Source: Appendix Table D.2.

26 The disparity in household income per capita between municipal areas and sanitary districts, on the one hand, and villages, on the other, has gone down between 1962/3 and 1975/6 in all regions except the South, as seen by the fact that the rural figures have risen as percentages of the urban levels. Between 1968/9 and 1975/6, the disparity between municipal areas and all nonmunicipal areas, which include both sanitary districts and villages, also went down in all regions except the Center. Similar conclusions are reached when per capita consumption expenditures are used, except that, in this case, the gap between municipal areas, with sanitary districts included, and villages went up between 1962/3 and 1975/6, while that between municipal and nonmunicipal areas went down between 1968/9 and 1975/6 for both the Southern and the Central regions. Thus the data strongly support the conclusion that the income and consumption gaps between rural and urban areas have in general been falling over time. It is important to note that, without properly taking into account definitional differences between the surveys, one can easily reach quite a different conclusion, namely that urban-rural disparities increased between 1962/3 and 1968/9, but decreased between 1968/9 and 1975/6. In fact, at least

27 part of the apparent widening in urban-rural per capita income and consumption disparities between 1962/3 and 1968/9 can be explained by differences in the definitions used for urban and rural areas./1 /1 Because sanitary districts were included in rural areas in 1968/9, this wuld bias upward the income estimates for both urban, in this case meaning municipal, and rural, or nonmunicipal, areas in 1968/9 relative to 1962/3. It is not immediately apparent whether this in itself would lead to an upward or downward bias in the level of disparity between urban and rural areas. The figures for 1975/6, however, show that the gap between urban and rural areas is greater when the 1968/9 definition is used than when the 1962/3 definition is used. The only exception to this is when we consider per capita consumption expenditure for the Central region. Turning to the question of regional disparites, Table 2.2 gives household total income by region, for urban and rural areas combined, for the three survey years and, in addition, expresses the levels as percentages of that for the Central region in any given year. From this we see that the poorer regions, especially the Northeast and the North, have been catching up with the Central region. Taking the region as a whole, Northeast household total income has progressed from 61% of Central income in 1962/3 and in 1968/9 and 67% in 1975/6. The North went from 61% in 1962/3 to 66% in 1968/9, but has only reached 67% of incomes in the Center by 1975/6.

28 Table 2.2: HOUSEHOLD TOTAL INCOME, BY REGION, THAILAND, 1962/3 TO 1975/6 Household total income as % of Household total income (in Baht) level for Central Region Region 1962/3 1968/9 1975/6 1962/3 1968/9 1975/6 Northeast 5,915 9,481 16, North 5,987 10,253 16, South 9,411 10,893 19, Center /a 9,731 15,630 24, Bangkok /b 18,690 32,432 37, Whole Kingdom 8,232 13,074 21, /a Excludes changwats Samut Prakan, Nonthaburi and Pathum Thani in 1975/6. /b Includes changwats Bangkok and Thonburi in all years and, in addition, changwats Samut Prakan, Nonthaburi and Pathum Thani in 1975/6. Source: Appendix Table D.1, with urban and rural incomes combined, using appropriate weights as given in Appendix Table A.4.

29 The South apparently suffered a set-back in the 1960s, so that its relative income position declined from 97% to 70% of the level for the Center between 1962/3 and 1968/9. By 1975/6, however, it had partially recovered its original position, with average household income standing at 80% that of the Central region. As will be seen in the next section, the income position of the South is considerably influenced by the level of rubber prices. Table 2.3 also looks at the relative positions of regions, but this time considering urban and rural areas separately. For urban areas, the relative income positions of households in the Northeast and South have worsened when compared with the Center, while that of the North has improved, albeit from a surprisingly low level, between 1962/3 and 1975/6. Considering the period 1968/9 to 1975/6, we see that municipal households in the North caught up slightly with those in the Center, those in the South fell behind a little, while incomes of Northeastern municipal households fell from being 7% above their counterparts in the Central region to 15% below them. Quite a different story is told by the corresponding figures for rural areas. Whether we start in 1962/3, with just villages, or in 1968/9, with sanitary districts and villages together, we find that the gap between household incomes in the Central region and each of the others has narrowed somewhat, with the exception of villages in the South whose position relative to those in the Center appears to have deteriorated markedly.

30 Table 2.3: HOUSEHOLD TOTAL INCOME FOR URBAN AND RURAL AREAS BY REGION, THAILAND, 1962/3 TO 1975/6 Household total income as % of Household total income (in Baht) level for Central Region 1962/3 1975/6 1968/9 1975/6 1962/3 1975/6 1968/9 1975/6 Urban /a Northeast 13,517 28,473 26,471 34, North 10,362 25,955 21,680 35, South 15,164 30,924 24,385 38, Center /c 13,613 32,217 24,863 39, Whole Kingdom 13,090 29,320 24,391 36, , 92.7 Rural /b Northeast 5,259 14,352 8,735 15, North 5,267 14,436 9,498 15, South 8,233 16,548 9,297 17, Center /c 8,957 22,380 14,681 23, Whole Kingdom 6,609 16,296 10,430 17, L /a Municipal areas and sanitary districts for 1962/3 and corresponding figures for 1975/6; municipal areas for 1968/9 and corresponding figures for 1975/6. /b Villages for 1962/3 and corresponding figures for 1975/6; sanitary districts and villages for 1968/9 and corresponding figures for 1975/6. /c Excludes changwats Samut Prakan, Nonthaburi and Pathum Thani in 1975/6. Source: Appendix Table D.1.

31 To conclude, if we measure average household incomes against that in the Central region, then rural households in the remaining regions have made some gains. At the same time, urban households have fallen behind, but not sufficiently to offset the gains made by rural households. Thus, taking the regions as a whole, we still find that the poorer regions have been catching up with the Central region. It may be asked how much of the differences in per capita income across regions and areas of residence is due to the adjustments for household size, since larger households are typically associated with larger numbers of children whose consumption requirements are thought to be lower than those of adults. We have tested the sensitivity of the apparent disparities in per capita incomes by calculating income per adult-equivalent by areas and regions, in which it is assumed that the consumption requirements of children under 11 years of age are only half those of adults. The details are given in Part C of the Appendix. While it is true that the gaps between areas and regions are reduced somewhat after the adjustment, very little real difference is seen in the results. The disparity in income levels between Northeastern villages and municipal areas of the Central region, for example, is simply too big to be appreciably affected by the adjustment which, after all, only involves the difference in the proportions of children in the two areas. This is even after a rather-extreme assumption concerning the consumption requirements of

32 children. We conclude that, even though average household size and, related to this, the proportion of children vary across regions and areas, using per capita income to measure the level of well-being of households, without taking into account possibly lower consumption requirements of children, does not lead to any serious bias in regional and urban-rural comparisons. National accounts data can also be used to investigate the trends in regional disparities over time and to provide some insights into these. Before doing so, however, it will be useful to understand that national accounts and household sample survey estimates of regional incomes are not strictly comparable for many reasons, and that we use the national accounts to supplement our analysis based on household surveys for two purposes. The first is to provide rough corroboration of the survey results. The second is in order to gain a better understanding of the levels and trends in regional incomes through information on the production side. This information at the household level is only available for the 1975/6 Socioeconomic Survey, so that the basis for an analysis over time using the household surveys alone is lacking.

33 There are many serious problems in attempting to compare national accounts and household survey data./l In particular, in the case of Thailand, /1 See Part B of the Appendix for a detailed discussion of the comparability of national accounts and household sample survey data on regional incomes and their growth rates over time. it would be a mistake to consider the national accounts estimates to be accurate and complete and hence a standard against which household survey data should be evaluated. Firstly, there are some differences in concepts and coverage between the two sets of estimates of incomes. Out of total expenditures on gross domestic product, the incomes reported in household surveys would come closest to private consumption; public consumption and gross investment, for example, are not covered. Over time there has been a steady decline in the share of private consumption in total gross domestic product, from 73% in 1962 to 69% in 1968 and 66% in Thus, the growth rate of household incomes over time should be lower than that of gross domestic product. In addition, the household surveys only cover private households and exclude institutional and collective households, whose consumption would be included in private consumption expenditures in the national accounts. The national accounts also include imputed rent to owner-occupied dwellings; we have had to exclude this item since it is not available for all the surveys.

34 Secondly, to the extent that the national accounts themselves are thought to underestimate incomes, either because of incomplete coverage or because the extrapolations of many series over time are based on an assumption of constant per capita consumption and necessarily contain a downward bias, it would be absurd to assess the completeness of coverage of the household surveys by comparing total income reported in the surveys against the national accounts corresponding estimate. Because of the method of estimation of certain series in the national accounts, it is likely that they in fact underestimate consumption to a greater degree than the household surveys. Thirdly, the national accounts estimates use data from the 1962/3 Household Expenditure Survey as benchmark data for numerous series. This dependence on one of the household surveys makes the national accounts inappropriate as a standard of comparison for the purpose of evaluating the coverage of the household surveys. Fourthly, in many cases the method of estimation of a series in the national accounts leaves a great deal to be desired. For example, the estimates are often based on 1962/3 benchmark data extrapolated on the assumption of constant per capita consumption over time. Such crude estimates are unlikely to be as accurate as the direct household survey estimates for later years; they clearly should not be used to evaluate the estimates obtained from the surveys.

35 Finally, the national accounts overstate the share of Bangkok and the Central region in its regional breakdown. This is because of the practice of allocating value added of an enterprise to where its headquarters is located. This affects the regional estimates of incomes at a given point of time, as well as the regional growth rates,if it should be the case that the degree of overstatement in the Bangkok and Central region estimates has changed over time. Thus, it should be borne in mind that estimates of regional incomes from national accounts are used not so much to check the accuracy of the household surveys, but merely to provide a rough confirmation of the levels and trends in household incomes. Table 2.4 gives a comparison of the annual growth rates of per capita income over the period 1962 to 1975 taken from the household surveys and the national accounts. As would be expected from the foregoing discussion, there are many differences between the two sets of estimates, both in the levels of the growth rates and the regional differences in them. Nevertheless, the national accounts figures are in broad agreement with the conclusions obtained from the household surveys that there has been a rapid growth of incomes over the 1962 to 1975 period in all regions, including the Northeast and the North, and that income growth in the South was slow in the period 1962 to 1968.

36 Table 2.4: COMPARISON OF ANNUAL GROWTH RATES OF PER CAPITA HOUSEHOLD INCOME AND PER CAPITA GROSS DOMESTIC PRODUCT AT CURRENT MARKET PRICES, Annual Growth Rate of Per Annual Growth Rate of Per Capita Region Capita Household Income GDP at Current Market Prices (% p.a.) (% p.a.) 1962/3-1968/9-1962/3-1968/9 1975/6 1975/ Northeast North South Center Bangkok Whole Kingdom Source: Appendix Table D.2. National Accounts Division, National Economic and Social Development Board, Bangkok.

37 The national accounts data in Table 2.4 refer to gross domestic product at current market prices, not at constant prices. The distinction is important in a comparison with household survey results, since we are not so much interested in a measure of output as such, but in a measure of the value of output. The constant prices series by definition excludes the impact of changes in prices on income, whereas of course such changes may have a substantial effect on household incomes. Moreover, changes in relative prices affect different population groups differently; the constant prices series do not correctly reflect this. Table 2.5 gives a distribution of the gross domestic product by region for 1962, 1968 and 1975, both in constant 1962 and current market

38 Table 2.5: DISTRIBUTION OF GROSS DOMESTIC PRODUCT BY REGION, THAILAND, 1962, 1968, and 1975 (X) At constant 1962 market prices At current market prices Northeast North South Center Bangkok Whole Kingdom Source: National Accounts Division, National Economic and Social Development Board, Bangkok.

39 prices./i Using the constant 1962 figures, the alarming trend is in the /1 Data on gross domestic product at constant 1972 market prices are also available as far back as 1970, but only for the whole country and not broken down by region. declining shares over time of the Northeast, North and South, with corresponding rising shares for the Center and Bangkok, from 53% of the total in 1962 for these two regions combined to 55% in 1968 and 59% in In fact, the period saw Bangkok gaining at the expense of all the four remaining regions, whereas between 1968 and 1975, the Central region and Bangkok together increased their share, while the Northeast, North and South continued to suffer a fall in theirs. The picture is different when we look at gross domestic product valued at current market prices instead. The shares of the Northeast and South still go down, but that for the North shows no change over the whole period considered. More significantly, the share of the Central region, with Bangkok included, only goes up from 56% to 57% over the period 1968 to 1975 in this case, following a big increase between 1962 and The current prices series would, therefore, indicate that the tendency for the Central region and Bangkok to gain over the remaining regions may be in the process of slowing down in the 1970s.

40 For our purposes, data on gross domestic product are most useful when they are broken down by industrial origin, since they provide a link between household incomes and the production side of the economy. Through them we are able to examine the performances of the agricultural and nonagricultural sectors which help us in the understanding of the movements of household incomes over time. Table 2.6 presents growth rates of gross domestic product by region, with a breakdown for agriculture and, within this, crops. Systematic differences exist between the constant and current prices series which give somewhat different perspectives of the performance of agriculture in relation to the rest of the economy. For example, at current market prices, agriculture in the South grew at more or less the same rate as gross domestic product as a whole between 1962 and 1968, while at constant 1962 prices it grew faster. Between 1968 and 1975, agriculture in the North grew less fast than the total product and crops less fast than agriculture as a whole when valued at constant 1962 prices. The current prices series, on the other hand, indicates that crops grew faster than agriculture,-which in turn grew faster than total product over this period. The reason for these discrepancies is that the relative prices of agricultural products had changed. The current prices figures reflect this fact, while the constant prices series only takes account of output changes. Southern farmers were adversely affected by low rubber prices at the end of the 1960s, while Northern farmers enjoyed more favorable prices for their crops in the 1970s.

41 Table 2.6: ANNUAL INCREASES IN GROSS DOMESTIC PRODUCT BY REGION, THAILAND, and (% per annum) At constant 1962 At current At constant 1962 At current market prices market prices market prices market prices Total Whole Kingdom Northeast North South Center Bangkok Agriculture Whole Kingdom Northeast North South Center Bangkok Crops Whole Kingdom Northeast North South Center Bangkok Source: National Accounts Division, National Economic and Social Development Board, Bangkok

42 Many insights can be gained when we look at data on the prices of agricultural products and the areas planted in different crops by region directly. In Table 2.7, we present data on wholesale prices of, and areas planted in, different crops for which information is available, for the three years roughly corresponding with the years in which the household surveys were conducted. The wholesale prices are not the prices received by farmers for their products; but to the extent that farm-gate and wholesale prices tend to move together, the wholesale prices can be used to indicate price movements over time./l The areas planted are presented by region, since there have been significant differences in the pattern of their increases over time. /1 The advantage of the wholesale series is that they are more complete than the farm-gate ones. In addition, they are probably more reliable. The price data vary from crop to crop. Paddy prices showed little increase between 1962 and 1968, but doubled between 1968 and Rubber prices fell by 27% between 1962 and 1968, and had not recovered even to their original 1962 level by Maize and kenaf prices were similar in 1962 and 1968, but had risen considerably by Cassava prices fell by 24% between 1962 and 1968, but then doubled between 1968 and Other crops such as sugarcane, cotton, groundnuts and beans experienced big price increases between 1968 and 1975.

43 Table 2.7: WHOLESALE PRICES AND AREAS PLANTED IN SELECTED CROPS BY REGION, THAILAND, 1962 to 1975 Wholesale Price Area Planted: Northeast Area Planted: North Area Planted: South Area Planted: Center Crop (baht/metric ton) (000 rai) (000 rai) (000 rai) (000 rai) Paddy 1,096 1,151 2,310 17,820 17,860 24,990 7,673 9,569 11,649 3,150 3,590 3,449 12,525 14,154 13,156 Rubber 8,733 6,387 8, ,174 3,536 5, Maize , , ,157 4, ,015 1,837 2,986 Sorghum n.a. n.a. n.a w Kenaf 2,343 2,526 3, ,429 2, Un Cassava 2,351 1,792 3, , ,367 1 Sugar cane n.a. 136/a ,268 Tobacco n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a Coconut n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. Cotton n.a. 4,090La 7, Groundnuts n.a. 4,200/a 9, Soy beans n.a. 2,410/a 5, , Mung beans n.a. 2,52C/a 7, ,004 1, Castor beans n.a. 2,600/a 4, /a Prices for Source: IBRD, Thailand: Toward a Development Strategy of Full Participation, A Basic Economic Report, Report No TH, September 1978, Tables 7.6, R7.1-R7.4.

44 Comparing the farm-gate and wholesale prices for various crops for which data are available, we see that the farm-gate prices have risen more rapidly for some important crops, notably nonglutinous paddy and maize. For example, between 1962 and 1976, the farm-gate price of nonglutinous paddy went up by 167%, while the wholesale price went up by 106%. For maize, the corresponding figures were 154% and 123% respectively. Thus, the prices received by farmers improved more for these crops than would be indicated by the wholesale prices shown here. However, for many other crops, such as kenaf, cassava, groundnuts, soybeans and mung beans, the wholesale prices went up faster than the farm-gate prices over this period. Table 2.7 also shows that dramatic changes have taken place in the regional pattern of cultivation of different crops, with the exception of the Southern region. Paddy remains very important in all regions, but the area planted in it has remained more or less constant in the Center and South, while increases of 40% and 52% were experienced in the Northeast and North between 1962 and Not only has there been a substantial increase in the area planted in paddy, of the order of 3% per annum, in the Northeast, the areas planted in maize, kenaf and cassava have increased many times, from rather insignificant bases. The situation is similar in the North, where there have been large increases in the areas planted in maize, sorghum and beans. The Central region has seen shifts into maize, cassava, sugarcane and other upland crops.

45 The expansion of paddy cultivation in the Northeast and North, and of other crops such as maize, kenaf and cassava in the Northeast, North and Center, with their accompanying price increases, support the earlier findings that household incomes in rural areas have been growing rapidly. The Northeast and North, in particular, far from giving the impression of a stagnating rural sector, have been able to make shifts into new crops, the prices of which have risen considerably, as well as expand the total areas cultivated. The South gives a totally different picture. Paddy and rubber are the two important crops there. little since the early 1960s. The area planted in paddy has increased very Between 1962 and 1968, rubber prices fell by some 27%, while area planted increased by 11%. The net result was that rubber production over this period fell by 32%. Given the nature of rubber production, in which there is a great deal less flexibility in switching to and from other crops, it is not surprising to find that unfavorable conditions in its export market should have quite perceptible effects on rural household real incomes in the South; their insignificant growth rate between 1962 and 1975 has already been noted in the last section. By 1975, rubber prices were still some 3% below the 1962 figures. Area planted, however, had gone up by 61% from the 1968 level. The apparently better performance of rural incomes in the South between 1968 and 1975 can thus be attributed at least in part to the recovery in rubber prices and the increase in area planted in rubber over the later period. A reasonably consistent story,

46 therefore, is obtained from data on rural household incomes, national accounts data and information on the prices of, and areas planted in, various crops. To conclude, three things should be borne in mind in using national accounts data to help interpret the household survey data. First, the data on real growth rates of regional gross domestic product are more consistent with those of household income when the product is valued at current market prices than at constant prices, because of the changes in relative prices over the period considered. To take crops as an example, between 1968 and 1975, the wholesale price of rubber went up by 33%, while that of cassava doubled and that of maize more than doubled. It makes a big difference, therefore, which prices are used to value the production of these crops. The problem is compounded on the quantity side, since farmers are naturally inclined to shift their production into crops the prices of which are increasing more' rapidly. Thus, for example, over the same period, the area planted in maize quadrupled in the Northeast, doubled in the North and increased by 63% in the Center. Area planted in cassava more than doubled in the Central region and increased by 34 fold in the Northeast. In contrast, area planted in paddy increased by only 22% in the North, 40% in the Northeast and decreased by 7% in the Center over this period. Household incomes are favorably affected not only by an increase in production of the existing crops, but also by shifts into more profitble crops, as well as increases in the prices of their crops. When these things are taken into consideration, the position of the Northeast relative to the Center according to national accounts data improves and is more consistent with the household surveys.

47 Second, the fact that the agricultural sector seems to have a substantial degree of fluidity in it, in the sense that changes in the composition of area planted and of agricultural output can be observed in the aggregate, means that the income position of those who have been able to take advantage of changing circumstances over time has improved and that some distance now exists between them and those who have been unable to respond to the new conditions and opportunities. In such a situation, aggregate figures and trends involving household incomes by region and area conceal a great deal of what is happening to different groups of households. Third, it would be a mistake to associate rural households too closely with the agricultural sector, and urban ones with the nonagricultural sector. Table 2.8 gives the proportion of farm households, defined as those which are headed by farm operators and farm workers, by region and area. We see that, in sanitary districts, generally considered to be quite rural, fewer than half the households are engaged in farming activities in all the regions. And even in villages, the percentage varies from 62% in the South to 80% in the Northeast. Thus a sizable proportion of rural households in each region is engaged in nonfarm activities.

48 Table 2.8: PROPORTION OF HOUSEHOLDS IN THE AGRICULTURAL SECTOR BY REGION AND AREA, THAILAND, 1975/6 (%) Farm Operators Farm Owning land Renting land Farm workers households Municipal Areas: Northeast North South Center Sanitary Districts: Northeast North South Center Villages: Northeast North South Center Bangkok: n.a. 10.3/a All Areas: Northeast North South Center Whole Kingdom: /a 58.5/a /a Excluding farm workers in Bangkok. Source: Socioeconomic Survey 1975/6, National Statistical Office, Regional Volumes, Table 7.

49 A farm household may have income from nonfarm work and a nonfarm household may have some agricultural income. Table 2.9 gives an estimate of the fraction of household total income from agricultural activities by region and area of residence. Income from agriculture is taken to be the sum of profits from farming, the value of home-produced goods and agricultural wages. The value of home production includes some nonagricultural component, but the proportion would be very small and no adjustment has been made for it. Wage income of farm workers has been estimated./1 /1 The official tabulations give a breakdown of total household income into wages and salaries, profits from self-employment, and so on. Wages are not broken down further into wages from farm and nonfarm work. Agricultural wages have been estimated as the wages accruing to all workers wh&ose major occupation is in agriculture. Fortunately, profits from selfemployment are given separately for farming and nonfarming activities.

50 Table 2.9: PROPORTION OF HOUSEHOLD TOTAL CURRENT INCOME FROM AGRICULTURAL ACTIVITIES, BY REGION AND AREA, THAILAND, 1975/6 (%) Estimated Profits Value of income from home Agricultural from farming production wages /a agriculture Municipal Areas Northeast North South Center Sanitary Districts Northeast North South Center Villages Northeast North South Center Greater Bangkok Metropolitan Area All Areas Northeast North South Center Whole Kingdom ja Estimated. Source: Socioeconomic Survey 1975/6, National Statistical Office, Regional Volumes, Table 4; Whole Kingdom Volume, Table 3.

51 The table shows that the proportion of agricultural income in the total. is 6% in municipal areas, 47% in sanitary districts, 72% in villages and 8% in Greater Bangkok. The significance of this is that in the areas which are usually considered rural, a very large part of household incomes originates in nonagricultural activities. Thus 53% of total income of households in sanitary districts are from nonagriculture, and as much as 28%^of the income of village households are. Focusing on villages, we see that the Center and North are the least dependent on nonagriculture; yet even there nonagricultural income makes up 27% and 30% of the total respectively. In the Northeast, 32% of income is from nonagriculture and in the South as much as 36%. Unfortunately, we do not have comparable figures for earlier years. But the combination of a narrowing in the disparities of incomes of urban and rural households, however defined, and similar growth rates in agricultural and nonagricultural gross domestic products by region is strongly suggestive of the increasing importance of nonagricultural activities in the total incomes of rural households. Thus the finding that rural households, in particular those in the Northeast, have experienced high rates of growth in their incomes can be explained by increases in the prices of some agricultural products, switches into more profitable crops, as well as increasing nonagricultural income-earning activities. The shift of households out of farming into other activities can be expected to be accompanied by a corresponding shift in the composition of their incomes from self-employment to wage income. Table 2.10 gives a breakdown of household income into wage inconme, self-employment income, other money income and own consumption, by region anid area for the three survey years.

52 Table 2.10: C0IW0SITOM OF YHDUSEHOLD INCOKE: PERCENTACES OF WACE INCOME. SELF-DIM.limIT DIOU, m mm 1O mm UWvI-II ST IlDIOli AND AREA, TNLIIAR. 1962/3 T0 1975/6 Urbae /l lural /b rbs. mad keral& Whole aol. abol. lortheast North South Center A lingdom Northeast lortb South Center / Ktngdom Iankok ji Northeast Nortb South Ceter Kiogdo= wait itacos 1962/ ; U t U7 31." / S.lf-seplo.ywwt inc s / / U :975/ / " Other wney irnos / t / O.n conoueption Ag / S.IC U n / la MUnIcIpal areas s4d snitsry dletricts for 1962/3 and correspooding figures for 1975/6; umicipal areas for 19U6/9 sad correspodiag figre ft 1975/6. f V.Ilages for 1962/3 and corresponding figures for 1975/6; sanitary district and illeges for 1968/9 and corresponditg ftiurs for 1975/6. /c Eacludes changpts Saout Praken. Nootheburi ad PatIus Thani to 1975/6. Li Includes cung'ats Bangkok and Thonburl id all years and, Ir addition, cbangwate Ssut Praken. ontheburt wad UathumThbi la 1975/6. Le For co.etabiltty across survey years. rental valu of ouned _uss ia scluded. Sources: Household Eapeoditure Sur-ey 1962/3. National Statisticel Office. gona: Volu_. Tbla- 1.0 snd 1.1. S.cios.nooLc Survey 196f/9. alon-l Sttintical Office. Table 10. O. coaueptiou tinte- are fere te or1igiml date tepee of the S -wey. Socloeconcic Survey 1975/6. National Statistical Offica, lonal Volums. Table 4.

53 We notice, first of all, that wage income is clearly more important in urban than in rural areas, while own consumption is a much larger proportion of the total household income of rural households. Secondly, selfemployment income and own consumption combined, representing mainly income from household enterprises, are more important for rural households. Over time there has been a definite trend in rural areas of a rising share of wage income, a falling share of self-employment income and also of total income from household enterprises, even though the proportion of own consumption in the total has risen. Thus, growth in income has been accompanied by an increasing dependence on wage employment. By 1975/6, at least 22% of wage income of village households derived from nonagricultural activities. A more ambiguous trend is seen in urban areas, where no obvious shift pattern can be expected. The evidence on household incomes and production statistics points to a remarkable capacity on the part of households to'adapt to new economic opportunities and to improve their livelihood. Incomes in rural areas, especially in the Northeast and North, are still very low and, as will be seen in the next section, a large proportion of rural households survives under very difficult conditions. Nevertheless, through responding to changing conditions, both in the agricultural sector itself and outside it, rural households as a group have succeeded in maintaining, and even improving, their position vis-a-vis urban households. The narrowing gap between urban and

54 rural households incomes should not, however, be allowed to divert our attention from the finding that there is increasing divergence within the rural sector itself between those households which have benefited from the new opportunities and those which have been unable to respond to them. The record of the Thai government in its supposed attempt to help poor households and to reduce the disparities in the living standards across regions and areas is much less commendable. The provision of different types of social services is still very uneven across regions and areas; there is a definite bias in favor of the better-off regions, and towards urban, as opposed to rural areas. Table 2.11 gives the values of selected socioeconomic indicators by region, covering general government expenditures and the provision of education, health and housing facilities. and require little further elaboration. The numbers speak for themselves Government expenditure per capita increases with the income level of a region; in 1976 it was 573 baht in the Northeast, compared with 945 baht in the Central region. It is true, however, that the growth rate in government expenditure per capita between 1970 and 1976 was slightly lower for the Center, when compared with the other regions. But, far from tending to reduce the existing regional disparities, the allocation of government expenditures would seem to reinforce them.

55 Table 2.11: AVERAGE VALUES OF SOCIOECONOMIC INDICATORS, BY REGION, THAILAND Northeast North South Center Bangkok-Thonburi Government expenditure per capita, 1970 (baht/year) Government expenditure per capita, 1976 (baht/year) Annual growth rate in government expenditure per capita, (% per year) CAO expenditure per nonmunicipal population, 1974/5 (baht/year) Population per government employee, Percentage of population 6-29 in school, CAO upper primary school enrollment rate, 1975 (%) Secondary school enrollment rate, 1975 (%) Current expenditure on primary education per capita, 1975 (CAO) (baht/year) Capital expenditure on primary education per capita, 1975 (CAO) (baht/year) Total expenditure on primary education per capita, 1975 (CAO) (baht/year) Population per Ministry of Public Health hospital, , , , , ,010 Population per hospital, ,340 20,819 14,546 11,413 2,272 Population per hospital bed, ,188 1, Population per provincial hospital bed, ,565 1,626 1,296 1, Population per doctor, ,805 18,234 20,131 14,481 1,863 Population per government doctor, ,439 19,768 23,737 15,897 2,033 Population per nurse, ,720 7,628 6,456 5, Rural population per first class health center, , ,528 94, ,215 - Rural population per health center, ,440 7,882 5,297 6,366 - Expenditure on health per capita, 1977 (baht/year) Percentage of houses with piped water, Percentage of houses with electricity, Percentage of villages with electricity, Source: Meesook, Oey Astra, "A Study of Disparities in Income and Social Services across Provinces in Thailand", Thammasat University, Faculty of Economics, Research Report Series No. 7, September 1978, Table

56 In the area of education, we see that school enrollment rates are lower, teachers are less qualified and government expenditures are lower in the Northeast and North than in the South and Center. The distribution of health personnel and facilities show a similar bias. In the poorer regions, a larger number of people have to share doctors, nurses, health centers and hospital beds. They are also allocated smaller amounts of expenditures on health. We conclude that the record of the govertiment in its contribution to the reduction of disparities across regions is quite unimpressive.

57 IIIo TRENDS IN THE SIZE AND DISTRIBUTION OF THE "POOR" POPULATION In this section we consider trends in the incidence of poverty and the distribution of the poor population over time, by region and area of residence. Our interest is focused specifically on the question of what has been happening to the poorest groups in the population, given that Thailand has experienced a very rapid rate of growth in aggregate income. Thus we wish to find out what has happened to the size and distribution of the poverty group, where poverty is defined in absolute terms so that the same standard is used in all years and the only adjustments made are to take into account price increases. The poverty line for 1975/6 has been set at 1,981 baht/person/year in rural areas and 2,961 baht/person/year in urban areas /1. Consumer price indices for urban areas by regions have been used to calculate the corres7 -ponding poverty lines in 1962/3 and 1968/9 prices.\ From these we have estimated the incidences of poverty, the proportions of the population below the poverty line, for 1962/3, 1968/9 and 1975/6 for each region and area of residence. Several estimates of the incidences of poverty in Thailand have already previously been made./2 However, comparisons over time need to be corrected since proper account has never been taken of the inconsistency in /1 See Annex I of IBRD, Income Growth and Poverty Alleviation in Thailand: Some Special Studies, Report No TH, June 1979, for details. /2 For example, Oey Astra Meesook, Income Distribution in Thailand, Thammasat University, Faculty of Economics, Discussion Paper Series No. 50, May 1976; IBRD, Thailand: Toward a Development Strategy of Full Participation, A Basic Economic Report, Report No TH, September 1978, Tables 3.16 and 3.17.

58 -50 - the definition of urban and rural areas between the 1962/3 and 1968/9 Surveys./l As already pointed out in this paper, sanitary districts were included in urban areas in the 1962/3 Survey but in rural areas in the 1968/9 Survey. This means that comparisons of the incidences of poverty between these two Survey years are invalid for two reasons. First, the coverages of the urban and rural populations are different, so that incomparable entities are being compared. Second, there is also an implicit inconsistency in the definition of the poverty line for sanitary districts; in the 1962/3 Survey, the urban definition is used for them, whereas the rural definition is used in the 1968/9 Survey. Thus a direct comparison between the incidences of poverty between 1962/3 and 1968/9, which necessarily contains these two discrepancies, contains a bias in that it exaggerates the improvement made between the two dates. In relation to 1968/9, the 1962/3 estimates contain an upward bias in the incidences of poverty. First, the inclusion of sanitary districts in urban areas implies a higher incidence of urban poverty. At the same time, excluding them from rural areas and retaining only villages also lead to a higher incidence of rural poverty. Second, using the urban definition of poverty on sanitary districts leads to a higher incidence of poverty overall than if they were treated as rural and hence subjected to a lower poverty line, as was the case in 1968/9. /1 For a detailed explanation, see Appendix A.

59 Sanitary districts are not as urban as municipal areas but at the same time not as rural as villages. But, given that the gap in average household incomes between municipal and nonmunicipal areas is greater than that between municipal areas together with sanitary districts and villages for all regions, one would be more inclined to classify sanitary districts as rural along with villages, if a two-way urban-rural breakdown is required. However, the essential point in comparisons over time is to ensure comparability in the definitions of urban and rural areas in each case, and this has been done in this report. Table 3.1 presents estimates of incidences of poverty for 1962/3, 1968/9 and 1975/6. Two sets of estimates are given for 1975/6, one for comparison with 1962/3 and the other with 1968/9. The first set includes sanitary districts with urban areas; the second includes them with rural areas. Thus in one case the urban definition of poverty is applied to sanitary districts, while in the other the rural definition is used. The first set shows consistently higher incidences of poverty for both urban and rural areas, but especially for urban areas, for reasons already given. The overall incidence of poverty is 33% instead of 31%, which is an indication of the degree of bias in applying the urban definition of poverty to sanitary districts instead of the rural one. Each set-of estimates, however, can be compared, since the figures are based on identical definitions of urban and rural areas, as well as poverty lines. The table shows that the incidence of poverty in all three areas of residence has been declining over time. For municipal areas and sanitary districts combined, the poverty incidence has fallen from 38% of the population in 1962/3 to 22% In 1975/6. Over the same period, the poverty incidence

60 Table 3.1: INCIDENCES OF POVERTY, BY REGION AND AREA THAILAND, 1962/3, 1968/9 AND 1975/6 (% of population group) Region and area 1962/3 le 1975/6 /e 1968/9 /f 1975/6 /f Northeast Urban /a Rural /b Total North Urban Rural Total South Urban Rural Total Center /c Urban Rural Total Bangkok /d Whole Kingdom Urban Rural Total /a Municipal areas and sanitary districts for 1962/3 and corresponding figures for 1975/6; municipal areas for 1968/9 and corresponding figures for 1975/6. lb Villages for 1962/3 and corresponding figures for 1975/6; sanitary districts and villages for 1968/9 and corresponding figures for 1975/6. /c Excludes changwats Samut Prakan, Nonthaburi and Pathum Thani in 1975/6. /d Includes changwats Bangkok and Thonburi in all years and, in addition, changwats Samut Prakan, Nonthaburi and Pathum Thani in 1975/6. /e The poverty line is defined in terms of household total income to be B 1,981/person/year in villages and B 2,961/person/year in municipal areas and sanitary districts. /f The poverty line is defined in terms of household total income to be B 1,981/person/year in sanitary districts and villages and B 2,961/person/ year in municipal areas. Sources: Household Expenditure Survey 1962/3, National Statistical Office, Bangkok. Socioeconomic Survey 1968/9, National Statistical Office, Bangkok. Socioeconomic Survey 1975/6, National Stat-istical Office, Bangkok.

61 in villages fell from 61% to 37%. The proportion of people below the poverty line in the whole country accordingly declined from 57% to 33%. Over the 1968/9 to 1975/6 period, the averall incidence of poverty (under a less stringent poverty line for sanitary districts) was reduced from 39% to 31%. For municipal areas alone, the reduction was small, from 16% to 14%, while in sanitary districts and villages together there was a decline from 43% to 35%. The corrections in the calculations performed here, compared with previously reported results, are quite interesting, especially with respect to urban poverty. The data now suggest that the incidence of poverty has been falling in sanitary districts and villages, but that the improvement in municipal areas has been quite slow. In other words, the incidence of urban poverty did not drop dramatically between 1962/3 and 1968/9, from 38% to 16%, and then level off for the later period, as reported previously. Making the appropriate comparisons shows that the 38% incidence in 1962/3 had only fallen to 22% by 1975/6. The incidence of poverty in Bangkok for 1975/6 is high relative to the earlier years because of the inclusion of changwats Samut Prakan, Nonthaburi and Pathum Thani in the Greater Bangkok Metropolitan Area in 1975/6. Table 3.2 gives the distribution of the poor population by region and area of residence. We see that, under either set of definitions of urban and rural areas, the urban population has come to constitute a somewhat larger

62 Table 3.2: DISTRIBUTION OF THE POOR POPULATION BY REGION AND AREA, THAILAND, 1962/3, 1968/9 AND 1975/6 (Z of total poor population) Region and area 1962/3 /e 1975/6 /e 1968/9 /f 1975/6 /f Northeast Urban /a Rural /b Total North Urban Rural Total South Urban Rural Total Center /c Urban Rural Total Bangkok /d Whole Kingdom Urban Rural Total /a Municipal areas and sanitary districts for 1962/3 and corresponding figures for 1975/6; municipal areas for 1968/9 and corresponding figures for 1975/6. /b Villages for 1962/3 and corresponding figures for 1975/6; sanitary districts and villages for 1968/9 and corresponding figures for 1975/6. /c Excludes changwats Samut Prakan, Nonthaburi and Pathum Thani in 1975/6. /d Includes changwats Bangkok and Thonburi in all years and, in addition, changwats Samut Prakan, Nonthaburi and Pathum Thani in 1975/6. /e The poverty line is defined in terms of household total income to be B 1,981/person/year in villages and B 2,961/person/year in municipal areas and sanitary districts. /f The poverty line is defined in terms of household total income to be B 1,981/person/year in sanitary districts and villages and B 2,961/person/ year in municipal areas. Sources: Household Expenditure Survey 1962/3, National Statistical Office, Bangkok. Socioeconomic Survey 1968/9, National Statistical Office, Bangkok. Socioeconomic Survey 1975/6, National Statistical Office, Bangkok.

63 proportion of the total poor population. Under one set of poverty lines, the population in sanitary districts and municipal areas made up 20% of the poor in 1975/6, compared with 12% in 1962/3; under the other set of definitions, 10% of the poor were in municipal areas in 1975/6, compared with 6% in 1968/9. Nonetheless, the overwhelming majority of the poor are to be found in rural areas, especially in the Northeast and the North. In terms of the total number of people living in poverty, this has decreased under one set of poverty lines by 14% from 1962/3 to 1975/6, or 1.1% per annum, from 15.8 million to 13.6 million people. Alternatively, under the other set of poverty lines, the number has gone down by 4% between 1968/9 and 1975/6, or 0.5% per annum, from 13.4 million in 1968/9 to 12.9 million in 1975/6. Thus under a situation of relatively rapid population growth, the reduction in the incidence of poverty has nevertheless left the country with close to the same numbers of poor people as it had before. The size of the poor population in the Northeast has remained more or less constant at 7 million over the period considered. In conclusion, the revisions in the estimates of the incidence and distribution of poverty presented here do not require a corresponding revision in our understanding of where the poor live or whether substantial improvements have taken place. The data suggest, however, that the decline in the incidence of poverty has been more gradual than was originally supposed; part of the apparent sharp fall in the poverty incidence previously observed between 1962/3 and 1968/9 can be accounted for by the differences in the definitions of urban and rural areas and, associated with these, the poverty line implicitly used for sanitary districts.

64 IV. CHARACTERISTICS OF POOR AND NONPOOR HOUSEHOLDS In the last section we looked at the trends in the incidence and distribution of the poor population across regions and areas over the period 1962/3 to 1975/6. We found that the proportion of the population below a poverty line defined in constant prices has been declining over time, so that rapid growth has indeed been accompanied by an improvement in the standard of material well-being of at least some of the poor. We turn now to look at households at a single point in time, in an attempt to get a clearer picture of who the poor are; in what ways they are different from the rest of the population; and whether they have the same access to social services as the better-off groups. The information is taken from the latest Socioeconomic Survey, conducted in 1975/6, for which greater details are available. In accordance with the preferred definitions of urban and rural areas, in which sanitary districts are classified as rural, we use poverty lines of 2,961 baht/person/year in municipal areas and 1,981 baht/person/year in sanitary districts and villages, in 1975/6 prices. Table 4.1 presents a

65 Table 4.1: DISTRIBUTION OF THE POOR /a BY SOCIOECONOMIC CLASS, REGION AND AREA, THAILAND, 1975/6 (% of poverty population) Socioeconomic class Professional, Clerical, sales, Nonfarm technical & Farm & services & Economically Farm own-account administrative general production inactive All operators workers workers workers workers households households Municipal areas: Northeast North South Center /b Sanitary districts: Northeast North South Center /b Villages: Northeast North South Center /b Greater Bangkok /c Whole Kingdom /a The poor are defined to be those who belong to households which have household total incomes under B 1,981/person/year in sanitary districts and villages and B 2,961/person/year in municipal areas, in 1975/6 prices. /b Excluding changwats Samut Prakan, Nonthaburi and Pathum Thani. /c Including changwats Samut Prakan, Nonthaburi and Pathum Thani. - Sample too small. Source: Special tabulations from the Socioeconomic Survey 1975/6, National Statistical Office, Bangkok.

66 distribution of the poor by socioeconomic class, region and area of residence. This gives, in each instance, the proportion out of the total poor population belonging to a particular group and thus takes into account both the incidence of poverty for the group and its weight in the total population. The poverty problem in Thailand is, first of all, a rural problem. Fully 83% of the poor reside in villages, and 6% more in sanitary districts. We can narrow this down further since 66% of all the poor can be found in villages of the Northeast and North, 46% in Northeastern and 20% in Northern villages. The other dimension of the problem concerns the socioeconomic class of households. The overwhelming majority of poor households, 75%, are those of farm operators, while 9% more are farm and general workers. The concentration of the farm operators making up 75% of the poor population is clear enough, since 69% out of the total of 75% are in villages, with 41% in villages in the Northeast and 15% in villages of the North. In Table 4.2, we have estimated the incidences of poverty for households belonging to different socioeconomic classes, as well as region and areas of residence. In municipal areas, the highest incidences of poverty are found among farm operators, farm workers and general workers. These groups have incidences of poverty which are well above that for municipal areas as a whole in all four regions. Nonfarm own-account workers and clerical, sales, services

67 Table 4.2: INCIDENCE OF POVERTY /a BY SOCIOECONOMIC CLASS, REGION AND AREA, THAILAND, 1975/6 (% of population group) Socioeconomic class Professional, Clerical, sales, Nonfarm technical & Farm & services & Economically Farm own-account administrative general production inactive All operators workers workers workers workers households households Municipal areas: Northeast North South Center /b Sanitary districts: Northeast North South Center /b Villages: Northeast North South Center /b Greater Bangkok /c Whole Kingdom /a The poverty line is defined in terms of household total income to be B 1,981/person/year in sanitary districts and villages and B 2,961/person/year in municipal areas, in 1975/6 prices. lb EKcluding changwats Samut Prakan, Nonthaburi and Pathum Thani. /c Including changwats Samut Prakan, Nonthaburi and Pathum Thani. - Sample too small. Source: Special tabulations from the Socioeconomic Survey 1975/6, National Statistical Office, Bangkok.

68 and production workers have incidences of poverty in the neighborhood of the overall municipal figure. By far the lowest incidences are found among professional, technical and administrative workers. In villages, it is clearly farm operators in the Northeast, North and South, and farm and general workers in the Northeast and North who face the highest incidences of poverty. Village households in the Central region are considerably less likely to be poor than those in the remaining regions. As many as 55% of households of farm operators and 42% of those of general workers in Northeastern villages are below the poverty line as defined. Nearly 40% of the corresponding classes of village households in the Central region are. All of the other socioeconomic classes in villages are considerably better-off, judging from their much lower incidences of poverty. Households in sanitary districts have been classified here as poor according to the same definition used for villages. We should expect them to exhibit characteristics which are somewhere between those of municipal areas and villages, including their incidences of poverty. For all classes together the poverty incidence for sanitary districts is higher than for municipal

69 areas and lower than for villages in the Northeast and North, but sanitary districts have a lower poverty incidence than municipal areas in the South and Center. And there are many more exceptions when we look within each socioeconomic class. For example, in the South 41% of farm operators in municipal areas are below the poverty line, while only 34% of those in sanitary districts are. This suggests that it would probably be appropriate to define the poverty line for sanitary districts to be somewhere between those for municipal areas and villages, in other words higher than the one being used at present. The class of economically inactive households contains a great variety of situations: people who are retired and well taken care of financially, as well as those who are unable to seek employment. This is reflected by an absence of any pattern in the incidences of poverty for such households across regions and areas. Table 4.3 presents average household incomes per capita for different socioeconomic classes by region and area of residence. Although broadly speaking these are In agreement with the incidences of poverty in Table 4.2, for example the income levels for households of farm operators, farm workers and general workers are much lower than for the other groups, there is by no means a one-to-one correspondence between the level of per capita income and the incidence of poverty. This helps to emphasize the fact that the distribution of income within a group, as well as the average level, is important in the determination of the incidence of poverty for the group.

70 Table 4.3: AVERAGE HOUSEHOLD INCOME PER CAPITA BY SOCIOECONOMIC CLASS, REGION AND AREA, THAILAND, 1975/6 (baht/month) Socioeconomic class Professional, Clerical, sales, Nonfarm technical & Farm & services & Economically Farm own-account administrative general production inactive All operators workers workers workers workers households households Municipal areas: Northeast North South Center /a Sanitary districts: Northeast North South Center /a Villages: Northeast North South Center /a Greater Bangkok /b , /a Excluding changwats Samut Prakan, Nonthaburi and Pathum Thani. /b Including changwats Samut Prakan, Nonthaburi and Pathum Thani. - Sample too small. Source: Special tabulations from the Socioeconomic Survey 1975/6, National Statistical Office, Bangkok.

71 The incidences of poverty and average household incomes per capita give us a good description of the economic position of different socioeconomic groups, but the group under consideration may be small and hence relatively insignificant from the point of view of policy planning. The distribution of the poor population in Table 4.1 points more directly to those population groups which demand government attention.

72 -64 - The policy implication of our findings for poverty alleviation is clear. Poverty programs must be oriented to rural, farming households; 83% of the poor are to be found in villages, 75% of them belong to households of farm operators and 9% to households of farm and general workers. Any governmental effort which is concentrated on farm households in villages of the Northeast and North has the potential to reach nearly 60% of the poor. Nearly every other household in Northeastern villages is poor, while one out of every three in Northern villages is. In order to better understand the differentiating characteristics of poor and nonpoor households, Table 4.4 gives values of selected socioeconomic characteristics for quintiles of households ordered by per capita consumption. The results would be similar if per capita income were used instead. We restrict ourselves to village households and use only the Northeast and the Center,.as representing the poorest and least poor regions respectively. Thus we are in a position to examine differences across households at different economic levels, as well as regions. at different stages of development. Definite patterns emerge from Table 4.4. The poor households, best represented by the bottom quintile, and the better-off ones, represented by the top quintile, exhibit very different tendencies. The differences can be

73 Table 4.4: SOME SOCIOECONOMIC CHARACTERISTICS OF HOUSEHOLDS, BY QUINTILES OF HOUSEHOLDS ORDERED BY MONTHLY PER CAPITA CONSUMPTION EXPENDITURES, VILLAGES IN NORTHEAST AND CENTER, THAILAND, 1975/6 Northeast villages Center villages Quintile Quintile I II III IV V I II III IV V Per capita consumption/month (baht) Average age of head Average household size % with eight or more members Number of earners Number of members per earner Persons per sleeping room % having electricity % lived in amphoe less than 10 years % farm operators % farm workers % nonfarm entrepreneurs % members attending school male female % members with higher than elementary grade education male female Source: Socioeconomic Survey 1975/6, National Statistical Office, Northeast and Center Volumes, Table 9.3.

74 traced progressively across successive quintiles within each region. Moreover, comparing corresponding quintiles of the Northeast and the Center yields the expected result that middle-income Central region households exhibit the characteristics of a better-off quintile in the Northeast. With respect to some demographic characteristics, we find that households with lower levels of per capita consumption have larger average household sizes and younger heads; they are more likely to be of very large sizes, with eight or more members. Poorer households are more likely to have to share each of their sleeping rooms between more people and are less likely to have the use of electricity. They tend to be less mobile, in the sense of being-less likely to have moved recently. The average number of income earners in poor households is larger than in rich households, but each earner also has to take care of the needs of more people, since the dependency ratio is higher for the poor. Poor households contain a larger proportion of farm operators and farm workers, and a smaller proportion of nonfarm entrepreneurs. Finally, in terms of educational opportunities, children in poor households are much less likely to attend school than children from richer households, and female children less than male when economic class is controlled for.

75 The data on access to electricity and education are especially interesting. First of all, great differences exist across households at different levels of income. For example, 0.5% of households in the bottom quintile in Northeastern villages have electricity, compared with 16.7% in the top quintile. For villages in the Central region, the comparable figures are 3.7% and 24.5% respectively. And whereas 1.2% of males of ages in households in the bottom quintile attend school, 12.8% of those in the top quintile do so. In the Central region the range is from 3.6% in the bottom quintile to 21.4% in the top quintile. Secondly, whereas we already know that the provision of public services is uneven across regions and, in particular, community types, the results here already control for region and area of residence. So it is not simply that all households in villages are equally less well provided with public services when compared to urban households. Rather, there is differential access to services, in this case electricity and schools, even among village households of a region. Data from the Socioeconomic Survey 1975/6 show differential access to electricity across socioeconomic classes as well, even after controlling for region and area of residence. For example, in Northeastern villages, only 3.4% of households of farm operators, 4.1% of farm workers and 2.4% of general laborers have electricity, compared with 12.8% of production workers, 16.9% of nonfarm own-account workers, 20.5% of professional, technical and administrative workers, and 32.8% of clerical,

76 - 68- sales and services workers. Similarly, in Central villages, 6.7% of households of farm operators, 4.0% of general laborers and 11.4% of farm workers have electricity, compared with 26.7% of production workers, 24.7% of nonfarm own-account workers, 44.3% of professional, technical and administrative workers, and 40.0% of clerical, sales and services workers. To conclude, the evidence is that the poor population consists mainly of farm households, especially in Northeastern and Northern villages. Poor and nonpoor village households have quite different characteristics and the poor have less access to social services. CONCLUSION The main conclusions of this paper are as follows. Between the early 1960s and mid-1970s, households in all areas of residence in all regions of Thailand have experienced a substantial rise in their standards of living, as measured by real increases in per capita income and consumption expenditure and in the ownership of consumer durable goods. The disparity between urban and rural areas in household incomes has been declining over this period in all regions. Moreover, the gap in rural incomes between the Central and remaining regions has also narrowed. The improvement in the position of rural households can be attributed to an expansion in the area cultivated, increases in the prices of agricultural products and, directly related to this, switches

77 into more profitable crops. In other words, agricultural prices have played a crucial role in improving the standard of living of the rural population. There is also evidence that rural households are relying more and more on income from nonagricultural activities; this has accounted in no small part for their continuing relatively favorable position over time. The increase in average incomes has been accompanied by a reduction in the proportion of the population in absolute poverty. This improvement in the position of the poor, however, has not much altered the composition of the poverty population, which still consists largely of households of farm operators and farm workers in Northeastern and Northern villages. It would appear that some of the households that used to be poor have shown themselves to be highly adaptable to changing economic conditions. They have been able to increase their total cultivated area, shift their production into more profitable crops and take advantage of off-farm incomeearning opportunities. Thus on average their relative position has improved over time. Nonetheless, there are also households which have been unable to benefit from the new economic environment. When we examine the performance of the Government with respect to its provision of public services, we find that great disparities also exist across regions and areas of residence. There is a bias in the availability of public services in favor of the urban population, the better-off regions and the better-off households within each area and region. To the extent that the total level of material well-being of the population is composed of'social services as well as personal consumption, then the Government's contribution is to increase, rather than reduce, whatever disparities exist in the personal component. The poor have managed to survive, and even to better their livelihood, quite apart from what their Government has done for them.

78 In considering the related issues of growth and equity with respect to Thailand, we thus conclude that economic growth has been accompanied by a reduction in regional and urban-rural disparities in household incomes and in the proportion of the poor population in the total. Moreover, to the extent that the reduction in disparities originates from direct improvements in the agricultural sector, then we have not witnessed any trade-off between economic growth and equity; agriculture-based growth has itself improved agricultural incomes, and hence also aggregate income, and at the same time led to a reduction in disparities between agriculture and nonagriculture, as well as across regions. One would, however, have to know what has happened to the distribution of income within the agricultural and nonagricultural sectors themselves before drawing final conclusions on this issue. As far as the role of the government is concerned, however, the resources available to it have not been used in such a way as to contribute to a reduction in the disparities in the living standards across major population groups.

79 APPENDIX A. Comparability of the Household Sample-Surveys for Thailand This appendix compares the three household sample surveys used in the study in terms of their sampling procedure and definitions. These have bearings on the interpretation of some of the findings of the study. The Household Expenditure Survey of 1962/3 and the Socioeconomic Surveys of 1968/9 and 1975/6 were all conducted by the National Statistical Office. They are, in a great many ways, similar to each other and comparable. However, partly in the attempt to improve successive surveys, there are some differences between them which are worth pointing out here. We first consider the surveys in terms of their sampling procedures and then of the comparability of their definitions. Sampling Procedures The sampling procedures for the three household surveys were similar, but not identical. Table A.1 gives a summary of the sampling techniques and sample sizes. The exact wordings in the published reports are used. Many minor differences can be seen, but the most significant is that the first-stage sampling was done by amphoe in the 1962/3 and 1975/6 Surveys, and by changwat in the 1968/9 Survey. Sampling by amphoe means that more

80 Table A.1: COMPARISION OF SAMPLING PROCEDURES USED IN HOUSEHOLD SURVEYS FOR THAILAND Household Expenditure Survey 1962/3 Socioeconomic Survey 1968/9 Socioeconomic Survey 1975/6 Sampling technique Three-stage stratified sampling Stratified multistage probability Self-weighting, stratified, three-stage sampling sampling First-stage sampling unit Amphoo: with probability proportional Changwat: In each of nine subregions, Amphoe: with probability proportional to size two sample changwats were to population size chosen with probability proportional to size. Bangkok-Thonburi was treated as a self-representative sample. Second-stage sampling unit Census Fnumeration District (ED) city MA & large SD: sample block MA: sample block block or village Small SD & rural areas: sample village SD & V: sample village Third-stage sampling unit Household: listed and classified by Household: Preliminary survey to divide Household: listed and classified by number of persons in house- households into six occupa- size and occupation of hold. In towns, further tional groups in sample household head classified by owner-renter blocks, two occupational 12 addresses in sample block & 15 status. groups in sample villages. addresses in sample villages were At least two households were selected systematically selected. from each occupational group. Sample size MAs & SDs Vs MAs SDs Vs MAs SDs Vs Northeast 330 1,470 Other regions 1, ,400 Northeast ,115 North Bangkok-Thonburi 1,500 North ,410 South South Center 440 1,070 3, Center ,095 Bangkok-Thonburi ,325 2J310 4,120 Total sample size - 6,000 Bangkok Metropolis: Total sample size - 6,430 (Carried out in three rounds, using Core Suburbs FrinRe areas 4,500 sample households in each round; about 13,500 filled-in questionnaires 1, were obtained.) Total sample size - 12,189 MA - municipal area SD - sanitary district V e village tli 31

81 APPENDIX changwats are included in the sample, but yet the data should not be used by changwat. Sampling by changwat probably means that changwat estimates of variables are better, for those changwats which are included; yet there is the serious problem that in the 1968/9 Survey, estimates for each of nine subregions are based on only two changwats, so that the outcome of the firststage sampling will necessarily have a big impact on the regional estimates, quite regardless of how the subsequent stages of the sampling are done. For purposes of obtaining regional and national estimates, sampling on the basis of amphoes would appear to be superior. Further differences in the sampling can be found in the second and third stages, but these are minor in relation to the first-stage differences. Excluding Bangkok, which is treated as a self-representative sample in the surveys, the sampling fraction for villages was relatively larger in the 1962/3 Survey compared with the later ones. The total number of households sampled was roughly twice as large in the 1975/6 Survey as in the two earlier ones. To conclude, as far as the sampling procedures go, the most serious difference between the three surveys is the first-stage sampling unit. The amphoe was used in the 1962/3 and 1975/6 Surveys, while the changwat was used in the 1968/9 Survey.

82 APPENDI1 Survey Definitions Table A.2 gives a comparison of some of the definitions used in the three household surveys. First are some definitions concerning the basic sampling unit, the household, and its membership. The definitions are given exactly as they appear in the official published reports. The definitions of the household itself have superficial differences, but are in fact quite consistent with each other. The definition of the head of household in all three surveys involves recognition as head by all the other members. However, in the 1968/9 Survey, he was also supposed to be "responsible for the welfare of other members." The upshot of this was that in this Survey all household heads were found to be economically active, presumably because the question asked itself suggested that an economically active individual should be named. The published reports of the two earlier surveys did not spell out how household membership was to be determined; more detailed instructions which may have been given to interviewers are not available. For the 1975/6 Survey, detailed instructions were given on whether or not absent household members, boarders and lodgers, and servants should be included in the main household or treated as a separate household. It is not known whether there were any departures from corresponding instructions for the earlier surveys.

83 Table A.2: COMPARISION OF DEFINITIONS USED IN HOUSEHOLD SURVEYS FOR THAILAND Definition Household Expenditure Survey 1962/3 Socioeconomic Survey 19E8/9 Socioeconomic Survey 1975/6 Household A group of persons who lived together. A private household consisting of a A group of two or more persons who make Some households contain two or more person or a group of related or unrelated common provision for food and other living "economic families" who share the persons who lived, ate and consumed other essentials. Members of a household may dwelling and sometimes ate together living essentials together but some pool their income and have a common budbut kept their finances separately. persons might keep their finances get to a greater or lesser extent. They separately. may be related or unrelated or a combination of both. A household could also be one person who makes provision for his own food and other living essentials without having common housekeeping or financial arrangements with other persons. Household head The person generally recognized as the The person generally recognized as head Person recognized as head by all other head by other members of the family. by other members of the household. members. Usually he was responsible for the welfare of the other members. Household membership Not explained Not explained Absent members; boarders & lodgers; servants (criteria for membership given). Earner A household member receiving money A household member receiving money A household member 11 years of age or _n income; same as an income receiver income; same as an income receiver in older who is economically active (selfin 1975/6 Survey. 1975/6 Survey. employed, employees and unpaid family workers) A household member receiving money income is called an income receiver. Urban/rural Urban - municipal areas + sanitary Urban - municipal areas Municipal areas, sanitary districts and districts Rural - sanitary districts + villages villages reported separately. But for Bang- Rural - villages kok, use core, suburbs and fringe areas. Central region Includes Samut Prakan, Nonthaburi Includes Samut Prakan, Nonthaburi and Excludes Samut Prakan, Nonthaburi and and. Pathum Thani. Must add "Eastern" Pathum Thani. Pathum Thani. region which is given separately. Bangkok Only municipal areas of Bangkok- Bangkok, Thonburi Bangkok, Thonburi, Samut Prakan, Thonburi were included. Nonthaburi and Pathum Thani t X

84 -76- APPENDIX The most serious discrepancy in definition across surveys is the breakdown between urban and rural areas. The 1962/3 Survey classified sanitary districts as urban, in addition to municipal areas, leaving villages as rural./1 In the 1968/9 Survey, only municipal areas were treated as urban, 1± According to the 1970 Population and Housing Census, the definitions of municipal area and sanitary district are as follows: Municipal Area A municipal area is a legal unit established by the Royal Decree of the 1953 Municipality Act. There are three categories of municipal areas, Nakhon (city), Muang (town) and Tambon (commune). A tambon municipality is established wherever it is deemed appropriate. A muang municipality is established in each area where the administrative seat of the provincial government is located or where the population is at least 10,000 persons, with an average density of not less than 3,000 persons per square kilometer. The sources of tax revenue must also be sufficient for the execution of municipal affairs as stipulated in the 1953 Municipality Act. A nakhon municipality is established in areas where the population is at least 50,000 persons, with an average density of not less than 3,000 persons per square kilometer. Tax revenues must also be sufficient for the execution of municipal affairs as stipulated in the 1953 Municipality Act. Sanitary District A sanitary district is established by the Ministry of Interior under the provisions of the Sanitary District Act of Under the provisions of the Municipality Act, any sanitary district may be established as a municipal area.

85 APPENDIX while sanitary districts, together with villages, were considered rural. The 1968/9 definitions imply higher income averages for both urban and rural areas, when compared with the 1962/3 definitions, since average incomes of sanitary districts fall between municipal and village incomes. not clear from this how urban-rural disparities are affected. However, it is Fortunately, the 1975/6 Survey tabulations are done separately for municipal areas, sanitary districts and villages, so that we can use 1975/6 data to see how the definitional differences affect various estimates of income disparities.

86 APPENDIX To use household total incomes as an example, Table A.3 presents these by region for 1975/6, for municipal areas, sanitary districts and villages separately, as well as the weighted averages for municipal areas and sanitary districts, and for sanitary districts and villages. We note, first of all, that the gaps between the three different community types are different across regions. The distances between sanitary districts and villages are quite similar in all the regions except the Northeast, where incomes in sanitary districts are 78% higher than those in villages; in the North, South and Center they are only 38%, 32% and 28% above village incomes. However, the Northeast has a smaller gap between municipal and sanitary district incomes than the North and South; the former is 33% above the latter in the Northeast, compared with 80% in the North and 76% in the South. In the Centtal region, municipal incomes are only 40% above those of sanitary districts, which themselves are already closer to village incomes than in any other region. The last two columns in Table A.3 give urban incomes as percentages of rural incomes, using the two sets of definitions of urban and rural. Three things can be seen from these figures. First, the ranking of regions by the degree of urban-rural income disparity depends on the choice of definition. The Northeast has the largest disparity, followed by the South, if the 1962/3 Survey definitions are used, in which sanitary districts come under urban areas. When the 1968/9 Survey definitions are used, so that

87 - 79 APPENDIX Table A.3: HOUSEHOLD TOTAL INCOME BY AREA, THAILAND, 1975/6 Household total income (B) Relative income (M) MAs SDs MAs + SDs MAs Region MAs SDs Vs SDs Vs Vs SDs + Vs Northeast 34,032 25,524 14, ,473 15,538 North 35,772 19,872 14, ,955 15,128 South 38,280 21,768 16, ,924 17,176 Center /a 39,852 28,548 22, ,217 23,532 /a Excluding changwats Samut Prakan, Nonthaburi and Pathum Thani in 1975/6. Note: Source: MA = Municipal area SD = Sanitary district V = Village Socioeconomic Survey 1975/6, National Statistical Office, Regional Volumes, Table 4.

88 APPENDIX sanitary districts are classified as rural, the North has the greatest disparity, followed by the South. Second, according to either definition, the urban-rural disparity is considerably lower in the Central region than in any of the others. Third, for all regions, the 1968/9 Survey definitions consistently show a higher disparity between urban and rural incomes than the 1962/3 definitions. Thus there are serious problems in comparing the published results from the 1962/3 and 1968/9 Surveys. In particular, comparisons of urban and rural household incomes will be seriously biased toward showing increasing disparities over time. Fortunately, the results of the 1975/6 Survey are given separately by community type; these can be combined according to either of the two sets of definitions to make the suitable comparison over time. Another problem associated with the urban-rural breakdown is that communities get reclassified from time to time, so that in effect better-off villages become sanitary districts and better-off sanitary districts become municipal areas. The income and other estimates for these community types can be affected, and hence also the estimates of disparities between urban and rural areas, when reclassification takes place. Table A.4 gives the distribution of the population across community types for the Census years, 1960 and 1970, and for 1968/9 and 1975/6 as obtained from the Socioeconomic

89 APPENDIX Table A.4: DISTRIBUTION OF THE POPULATION ACROSS AREAS, BY REGION, THAILAND, 1960 TO 1975/6 Sanitary Municipal areas districts Villages 1960 Whole Kingdom Northeast North South Center Bangkok-Thonburi /b (100.00) (sanitary districts + villages) 1968/9 Whole Kingdom Northeast North South Center /a Bangkok-Thonburi /b Whole Kingdom Northeast North South Center la Bangkok-Thonburi /b /6 Northeast North South Center /a Greater Bangkok Metropolitan Area /b /c Core 51% Suburbs 36% Fringe areas 13% la Changwats Samut Prakan, Nonthaburi and Pathum Thani excluded in 1975/6 but included in all other years. /b Changwats Samut Prakan, Nonthaburi. and Pathum Thani included in Greater Bangkok Metropolitan Area in 1975/6 but excluded from Bangkok-Thonburi in all other years. /e For Greater Bangkok Metropolitan Area, from Socioeconomic Survey 1975/6, National Statistical Office, Bangkok Volume, Area Coverage, p. 1, the proportions of the population in the Core, Suburbs and Fringe Areas living in municipal areas, sanitary districts and villages were obtained. These were used in conjunction with the distribution of households in the Core, Suburbs and Fringe Areas, given in the Socioeconomic Survey 1975/6, to calculate the distribution of the population across areas. Sources: Population Census, taken from Household Expenditure Survey 1962/3, National Statistical Office, Whole Kingdom Volume, Table L, p /9 - Socioeconomic Survey 1968/9, National Statistical Office, Table A, p Population and Housing Census, National Statistical Office, Regional Volumes and Changwat Volumes for Bangkok and Thonburi, Table /6 - Calculated from "Socioeconomic Survey 1975/6, Weight Factors for Combining Household Averages and Percentages," National Statistical Office, and average household sizes by region and area taken from Socioeconomic Survey 1975/6, Regional Volumes, Table 1.

90 APPENDIX Surveys. We see that considerable reclassification has taken place over time. For example, in 1975/6 there are larger proportions of people in K municipal areas in the South, and in sanitary districts in all regions, when compared with Reclassification means that we are no longer referring to the same population group, even when we compare the same community type over time. One last definitional problem concerns the recent inclusion of changwats Samut Prakan, Nonthaburi and Pathum Thani in the Greater Bangkok Metropolitan Area. This means that in the 1975/6 Survey, Bangkok considered as a region refers to these three changwats, in addition to Bangkok and Thonburi of the earlier surveys. Correspondingly, the Central region in the 1975/6 Survey excludes these changwats, which had previously been included. The extent to which this change affects the estimates of incomes and other particulars for these two regions is not entirely clear. The three changwats are somewhat more urban than the rest of the Central region; 11.3% of their population lived in municipal areas in 1970, compared with 9.5% for the region as a whole. Thus the change means a smaller proportion of municipal population in both the Central region and Greater Bangkok. Household income estimates are not available for these three changwats alone but 1976 per capita gross domestic product is higher for them than for either Bangkok-Thonburi or the Central region. Compared with the Central region, the figure is lower for Pathum Thani, but higher for Nonthaburi and much higher for Samut Prakan, which had by far the highest per capita product in There is no way of adjusting the published tabulations of the 1975/6 Survey to take this change of definition into account.

91 APPENDIX B. Reliability of the Household Sample Surveys for Thailand and their Comparability to National Accounts Data The Household Expenditure Survey, 1962/3 and the Socioeconomic Surveys, 1968/9 and 1975/6 are on the whole well designed and theoretically capable of producing reliable estimates of incomes, expenditures and so on. In practice, they suffer from many shortcomings which are usually encountered in sample surveys of this type. These problems affect the various survey estimates in a given year, as well as their growth rates between different survey periods. It is difficult to determine the extent of the errors involved in the estimates and, in many cases, even the directions of the biases produced. The sample surveys for Thailand are national surveys based on a stratified random sampling technique which is designed to give unbiased regional and national estimates of incomes, expenditures and so on. The survey questionnaire is reasonably well-designed and very detailed, which should help increase the accuracy of the estimates. The quality of the estimates of different variables may, however, vary from survey to survey, depending on the emphasis given in each particular case. To give some examples on how the more important estimates are obtained, we consider household income and consumption expenditures. The income questions are asked of each individual household member, not of the household as a whole, and is broken down into different categories, such as

92 APPENDIX wages and salaries, profits from self-employment and so on. For the 1975/6 Survey, for which we have the instructions for interviewers, the interviewers are asked to check that the answers, both in terms of the type of income and the amount, are reasonable, given the individual's occupation, education and so on. Information on consumption expenditures was collected for the household at a very detailed level. Both cash expenditures and own consumption are covered, so that it is possible to derive total household income and total household consumption. The interviewers are asked to check for reasonable consistency between the income and expenditure figures, and to reinterview if necessary where the discrepancy is too great. Based on their sample design, the sampling technique used, the questionnaire and the attention to detail reflected in the instructions to the interviewers, the quality of the household sample surveys for Thailand appears to be good. Nonetheless, we point out some potential problem areas in actually conducting the surveys. These problems fall into two types, sampling and nonsampling errors. Sampling errors mean that the sample is not necessarily representative of the underlying population. They occur because the sampling frame is inaccurate or out of date, such as when it is based on the last Population Census preceding the household survey in quesion, or when households which should be included in the sample are left out. For example, many slum households in Bangkok may be outside the sampling frame altogether, so that they are systematically excluded from the final estimates. On the other hand, some households may be in the sampling frame and still be systematically excluded. For example, a remote village may be selected to be in

93 APPENDIX the sample, but another village may have to be used as a substitute because of security problems in the village originally selected. Since villages with security problems are likely to have special characteristics, their systematic exclusion from the sample would bias the estimates to some extent. Another population group which is likely to be underrepresented in the sample is wealthy households in Bangkok. These households are in the sampling frame but the interviewers may be too intimidated to conduct the interview, especially if the household does not show immediate cooperation. With regard to nonsampling errors, the most serious involve the underreporting of incomes and expenditures. Low-income households are generally thought to underreport both their incomes and consumption because of the difficulty of recalling all of own consumption which is a significant proportion of total consumption. However, extremely poor households may deliberately overstate their incomes and consumption out of embarrassment or shame, telling the interviewer what they think they ought to be eating rather than what they are actually eating, for example. Rich households, on the other hand, would be inclined to underreport their incomes out of the fear of having to pay more taxes, because they have acquired their incomes through illegal means, or simply out of a desire to hide things from government officials. luxury items. They are also likely to underreport their expenditures on All households are prone to the underreporting of expenditures of individual members on food, transportation, personal care and so on which are made outside the home, since the housewife, who is usually interviewed for the household survey, is probably unaware of them.

94 APPENDIX In terms of aggregate measures, such as average household income and income inequality, the underreporting of rich households has the greater effect. Even though the total number of households involved is small, the incomes are relatively large. The underreporting of incomes, especially of rich households, means that in the aggregate both average household income and the degree of income inequality are biased downward. Moreover, it cannot be assumed that the extent of the bias is uniform across different population groups at a point in time or across surveys at different points in time. A standard method for evaluating income and consumption estimates from household sample surveys is to compare them with the corresponding national accounts estimates. In this way the degree to which total income or consumption is underreported in a sample survey can be estimated. In addition, for two or more surveys it is possible to check on the rates of growth of income and consumption over time, if some assumption is made concerning the relative extents of underreporting in the different surveys. In most countries, the collection of data for national accounts and for household surveys are independent activities. The national accounts are supposed to be more comprehensive in their coverage, with their own checks for consistency, so that estimates of income and consumption based on them are considered to be reliable in terms of their levels at a point in time, as well as their trends over time. Because national accounts estimates are presumed to be complete and accurate, they can be used in the assessment of household survey estimates with respect to their completeness and accuracy.

95 APPENDIX However, in the case of Thailand it would be a mistake to evaluate the reliability of the household surveys by comparing their estimates with those obtained from national accounts. There are at least five reasons why national accounts estimates do not provide the appropriate standard against which survey data can be evaluated. The first is that there are differences in coverage and concepts between the two sets of estimates. The second is that the national accounts themselves are not complete in their coverage and seriously understate national income./l The third is that the national /1 Levy, Emmanuel, National Accounts and Basic Statistics of Thailand, mimeo., March The discussion of national accounts which follows relies heavily on this report. accounts estimates are not independent of the household sample survey estimates; many series which go into national accounts are based on bench-mark values obtained from the 1962/3 Household Expenditure Survey. It becomes absurd then to check the validity of the household surveys using the national accounts estimates. The fourth is that the methodology used in the estimation of many series which make up national income are highly questionable or even improper. The fifth is that the allocation of the total product across regions is biased in favor of Bangkok and the Central region. In our judgment, therefore, it is far from clear that the income and consumption estimates from the household surveys would be inferior to the national accounts estimates, both as to levels and trends over time. In many instances, it would be more appropriate to evaluate a national accounts series by comparing it with the household survey estimate. When

96 APPENDIX we use national accounts data in conjunction with the household sample surveys, it is with a view of providing rough checks on the levels and trends of both sets of estimates, not of checking one set against the other which is somehow assumed to be more accurate. Even for rough comparisons, it is useful to be aware of some of the major sources of differences between the national accounts and household sample survey estimates. The item in national accounts which is closest to household income is private consumption expenditure valued at current market prices. The constant market prices series cannot be compared with the income estimates of household surveys which incorporate changes in producer prices over time, whereas the current prices series would be comparable. Even private consumption is not directly comparable to household income estimates because of differences in coverage. The household sample surveys include only private households and excludes all institutional households, such as monasteries, penal institutions, welfare homes and hospitals; collective households such as hotels, boarding houses and living quarters for workers with at least six members who have meals provided by the management; and the military. The national accounts estimate of private consumption is supposed to include imputed rent of owner-occupied dwellings. A comparable estimate for imputed rent was obtained in the 1975/6 Socioeconomic Survey. However, since imputed rents were not estimated in the 1962/3 and 1968/9 Surveys, we have left them out of the 1975/6 Survey estimates as well, in order to have comparability across surveys. This means that our income and consumption estimates are less complete in terms of coverage than national accounts.

97 APPENDIX The national accounts underestimate many items of food production and consumption because of the method of estimation. For example, production of fresh-water fish is based on the total number of boats and the average catch per boat, while that of meat is based on the number of animals slaughtered, with a correction for own consumption based on consumption data from the 1962/3 Household Expenditure Survey. Many series in the national accounts are based on the assumption of a constant per capita consumption over time. Since it is likely that per capita consumption of many of these items has in fact gone up over time, the national accounts estimates for later years are too low. The problems of underestimation of consumption, in particular own consumption, in household surveys would seem to apply with even greater force in the corresponding national accounts estimates. Many series in the national accounts are based on estimates of consumption from the 1962/3 Household Expenditure Survey extrapolated for later years under an assumption of constant per capita consumption, for example consumption of chickens, ducks and other poultry; eggs; fruits and vegetables; laundries, barber shops; domestic services; coffee, tea and cocoa; footwear and other personal effects; reading and education. Clearly there is nothing to be gained by checking the validity of the 1962/3 Household Expenditure Survey against the national accounts data for that year, or of the 1968/9 and 1975/6 Socioeconomic Surveys against the extrapolations of the 1962/3 bench-mark figures.

98 APPENDIX In general, the national accounts estimating methods are to be regarded with suspicion. The 1962/3 bench-mark figures are by now too out of date, and the assumption of constant per capita consumption over the years both perpetuates and compounds the original errors. Other obvious sources of errors abound. Housing rent is based on 1962/3 benchmark figures which purportedly included both actual and imputed rents. not clear even now that imputed rent was in fact included. However, it is Originally estimates were obtained for average rent per square meter of accommodation for municipal areas and per housing unit for nonmunicipal areas. The municipal area figure was then converted to a per housing unit basis and used in that form since. The resulting estimates of rents cannot be very reliable. Many of the national accounts estimates are based on assumptions. the basis and justification for which are mysterious. Retail prices are based on import data at COIoF. values converted by using a standard mark-up 0 Estimates for production of fat assume that there are 18 kilograms of fat per swine. Gasoline consumption is based on an assumed 5-liter consumption per car and 1-liter consumption per motorcycle per day. It is also assumed that 65% of cars and 55% of motorcycles are privately owned. Lastly, the practice of allocating value-added of enterprises to that region in which the enterprise headquarters is located means that some of the product which is allocated to Bangkok and the Central region should properly belong to the remaining regions.

99 APPENDIX Given the very crude and often questionable methods of estimation used in the national accounts, we conclude that they cannot be used to evaluate the income and consumption estimates of the household sample surveys. The latter have the advantage of being based on actual observations and not extrapolations or assumptions. Thus, if anything, one would be more inclined to check the validity of the national accounts estimating methods by comparing their estimates against the household survey estimates for recent years. For indications of trends in income and consumption over time, the household surveys and national accounts will yield different results if the share of private consumption in the total has changed; if the share of private households in private consumption has changed; and if the assumptions used in national accounts, especially those involving constant per capita consumption, lead to different degrees of bias in the resulting estimates over time. Although total consumption of households can be most closely approximated by private consumption in expenditures on gross domestic product at current market prices, this item is only useful as a possible check on the level and trend of consumption. Gross domestic product by industrial origin at current market prices is more useful if we are trying to explain the changes in household income and consumption over time, since the contributions of different sectors to total gross domestic product are given. Moreover, until recently, the figures were available at the regional level and thus could be used to give insights into differences in regional growth rates. The problem is obviously that it is not possible to separate the total product into that accruing to private households. The national accounts data are not, therefore, necessarily expected to give results

100 APPENDIX which are all that consistent with the household survey data, but only to give indications of what has been happening on the production side of the household sector, in order to see if these are broadly in agreement with the observed changes in household income and consumption over time. C. The Sensitivity of MeasurinR Household Income on a Per Capita Basis as Compared with a Per Adult-Equivalent Basis Table C.1 gives per capita household income for 1975/6 for municipal areas, sanitary districts and villages separately. The figures are also expressed in terms of the municipal Central level, and take into account differences in average household sizes across areas and regions. Significant differences are seen to exist between them and it may well be asked how much of these is due to the adjustment for household size. Typically, larger household sizes are associated with larger numbers of children. Thus, granting that some adjustment must be made on total household income before an acceptable measure of household economic well-being can be obtained, the question remains whether there is a downward bias for large households, if it should be the case that consumption requirements of children are lower than those of adults./l /I For example, is the disparity between villages in the Northeast and municipal areas in the Center exaggerated, given that the former has an average household size of 6.0 and the latter 5.1. How much in fact of the difference in per capita household income can be explained by the difference in the proportions of children in the two areas.

101 APPENDIX Table C.1: PER CAPITA HOUSEHOLD INCOME, BY REGION AND AREA, THAILAND, 1975/6 Per capita % of Adjusted Adjusted household household household per capita Per capita income members size income household relative to under 11 relative relative income municipal years of to actual to munici- (baht/year) Center age size pal Center (%) (Z) (x) Municipal Areas Northeast 6, / North 7, South 7, Center 7, Sanitary Districts Northeast 4, North 4, South 4, Center 5, Villages Northeast 2, North 2, South 3, Center 4, Greater BanRkok Metropolitan Area Core 7, Suburbs 6, Fringe areas 4, Source: Socioeconomic Survey 1975/6, National Statistical Office, Regional Volumes, Tables 4 and 10.

102 APPENDIX Column 3 of Table C.1 gives the percentage of household members who are under 11 years of age; this varies across region and area, from 21% in the city core of Greater Bangkok to 35% in Northeastern villages. The low income areas also tend to have larger proportions of young people; thus per capita income disparities would be exaggerated if their consumption requirements are indeed lower. Without making a judgement on what the consumption requirements of children relative to adults should be, we test the sensitivity of the apparent disparities in per capita income by making the assumption that, on average, the requirements of children under 11 years of age are only half those of adults. Column 4 of Table C.1 then shows the adjusted household size, the number of adult-equivalents, as a percentage of the actual size. Finally, column 5 gives the per adult-equivalent income relative to the municipal Center level. While it is true that the gaps between areas and regions are reduced somewhat after the adjustment, very little real difference is seen in the results. The disparity in income levels between Northeastern villages and municipal areas of the Central region is simply too big to be appreciably affected by the adjustment which, after all, only involves the

103 APPENDIX difference in the proportions of children in the two areas. This is even after a rather extreme assumption concerning the consumption requirements of children. We conclude that, even though average household size and, related to this, the proportion of children vary across regions and areas, using per capita income to measure the level of well-being of households, without taking into account possibly lower consumption requirements of children, does not lead to any serious bias in regional and urban-rural comparisons.

104 Table D.1: HOUSEHOLD TOTAL INCOME AND CONSUMPTION EXPENDITURES, 8T REC1ON AND AREA, THAILAND, 1962/3 T0 1975/6 (Bhct) Urban /a Rur-l /b Urban and Ru-ral Whole Whole Whole0 Northeast North South Center / Kingdom Northeat worth South Center A Kingdom Bangkokjd Northeaat North South Center /c Kingdom X Money Incone 1962/3 13, , , , , , , , , , , , , , , /9 24, , , , , , , , , , , , , , , ,$ /6 25, , , , , , , , , , , , , , /6 32, , , , , , , , , , Ovn Cons-notion /a 19C2/ , , , , , , , , , , /9 1, , , , , , , , , , , , , , , /6 2, , , , , , , , , , , , , , , , /6 1, , , , , , , , , , Total Iscoce 1962/3 13, ,362,24 15, , , , , , , , , , , , , , /9 26, , , , , , , , , , , , , , , , / , , , , , , , , , , , , , , , /6 34, , , , , , , , , Cas, Exe.ndlturea 1962/3 12, , , , , , , , , , , , , , , , /9 18, , , , , , , , , , , , , , , , /6 24, , , , , , , , , , , , , , , , /6 30, , , , , , , , , , Total Consnnption Ex,end i t.res 1962/3 13, , , , , , , , , , , , , , , , /9 20, , , , , , , , , , , , , , , , /6 27, , , , , , , , , , , , , , , , / , , , , , , , , , D Municipal reas and sanitary districts for 1962/3 and corresponding figures for 1975/6; muncipal area" for 1968/9 and corresponding figures for 1975/6. /b Villageo for 1962/3 and corresponding ftgures for 1975/6; sanitory districts and villages for 1968/9 atod correepoodijg figures for 1975/6. /r Zxcludea changuat Samt Prakan. Nonthaburt and Pathum Thani in 1975/6. /Ld Includes changwato Bangkok and Thonburi in all years and, in addition, changwats Smut Praken, Nonthaburi end PatwA Thani in 1975/6. /e For comparability across survey years. rental value of ovned homs is secludad. Sources: Household Expenditure Survey 1962/3, National Statistical Office, Regional Volume, Tables 1.0, 1.1 end 6. Socioeconnolc Survey 1968/9, National Statistical Office, Tables 10, 4 and 2. own c0 su ptloa stiutes are from the originl date taes of the Survey. Socioeconomic Survey 1975/6, National Statistical Office, Regional Voluas, Tables 4 and 1.