5 Rural-Urban Development, Components of Household Income and Inequality in Thailand

Transcription

1 5 Rural-Urban Development, Components of Household Income and Inequality in Thailand Introduction This chapter moves away from the exploration of inequality in the wage sectors and consider the inequality in the distribution of income among all types of households, taking into account the income derived from all sources. In much of the analysis to follow total reported expenditure per household is used as a proxy for income earned. It is well known that in household socio-economic surveys the reported expenditure is a more accurate measure of household welfare that total income reported by the head of the household. Further, since the size of households vary in any region or sector considered, we use the measure of household expenditure per capita as a better measure of household welfare than total expenditure. Globalization has asymmetric impact on different sectors and regions within the same country. This spatial dimension of inequality is an important feature of developing countries specifically those that have witnessed rapid growth such as Thailand. The pioneering work of Michael Lipton (1977) provided important contributions to the literature. Recent studies on spatial inequality that have also explored the rural-urban dimension include Bourguingnon (1998) and Dercon (2002). We therefore begin our analysis of household inequality by concentrating on the ruralurban differences in inequality and their changes over time. Rural and urban sectors in Thailand are, however, not separated very sharply by the sources of income accruing to households. While farm income on doubt accrues predominantly to rural households, the same cannot be said about the two other major components of income nonfarm profit and wages and salaries. Even at the beginning of the growth period of our study, 1988, wages and salaries accounted for 25 per cant of

2 2 household income in the rural areas against 47 per cent in the urban sector. The percentages for non-farm profit were 22 and 11 respectively. Important changes in the relative importance of the factor components of income in the two sectors took place over the various phases of the growth period and the subsequent crisis and recovery. In section II, therefore, our analysis moves to the sources of income and the factors affecting the inequality in their distribution over the period I Rural-Urban Differences The waves of reforms and liberalization undertaken by many developing countries since the mid-1980s are widely believed to have benefited urban areas more than they did rural regions. Both the theory and the empirical studies regarding the effects of reforms and liberalization on the rural-urban gap point to divergent conclusions. The theory provides two differing propositions. First, it is argued that liberalization increases ruralurban inequality with rural areas being worse off. Eastwood and Lipton (2000) argue that the liberalization increases the rural-urban inequality due to several different factors: The better-educated urban population is better suited to exploit new economic opportunities provided by price liberalization. Demographic trends coupled with faster urban fertility transition and the selective urban migration of young and educated may improve the relative earning power of urban population. Since urban and formal activity is inherently more regulated, this sector had most to gain from deregulation advocated by liberalization policies. Successful poverty reduction programs may have left behind a core of low-income, immobile, often regionally and ethnically specific rural groups, which as in China have proved slow to reap the so-called rural spread effects from national growth. Lipton [1977] argues the possibility of a persistent and stable urban bias in developing countries in both allocative outcomes, and political dispositions. This distortions may imply that that rural shares in many sorts of endowment such as investment, health, education, public spending. Such political dispositions are mainly unaffected by adjustment in prices or foreign-trade regimes and in so can offset any pro-rural effects by pro-urban changes in other policies, such as location of public investments and schools.

3 Urban-rural inequality might increase due to a persistent global downtrend of farm prices vis-à-vis non-farm prices. Krueger et al. (1995) argue that price distortions against tradables are often harmful to agricultural products; thus when prices are liberalized such distortions are eliminated. This should cause rural-urban inequality to fall and in so leads to improvement in rural welfare. In general, it is argued that overall inequality has been subject to opposing factors. This, known as the offsetting trends in inequality hypothesis (OTI), claims that underlying the overall inequality trend has been a tendency for rising intra-sectoral inequality to be offset by falling rural-urban inequality. Empirical studies also point to diverging conclusions. According to Eastwood and Lipton (2000), China has experienced a substantial increase in rural-urban inequality gap since the mid 1980s. while the rural-urban gap in Brazil and Honduras in the 1990s and in India in has declined. The story for South East Asia and Thailand, in particular, is an open question. Thailand has gone through tremendous growth during the last several decades, a growth period that has resulted in enormous structural adjustment. In the period globalization has led to significant gain by the formal sector, from only 23% in 1988 to more than 37% in The strongest gain in this period has been witnessed during the growth period of but that trend, albeit at slower pace, has continued even in the subsequent recovery period. This move towards formal sector has been a direct result of the move away from farming to the wage sector while self-employment also increased its share. 3 Table 0.1 Percentage Distribution of the Labor Force in Thailand Formal Work Status Wage Earners Self Employed Farmers Area City Sanitary Rural Source: LFS Surveys. For 2002, Farmers and Self Employed, City and Sanitary are reported together. Note: The formal sector is defined as????

4 4 It is also important to note that decades of high growth in Thailand has not led to massive urbanization. By 1988 nearly three quarters of Thailand population lived in rural areas and the initial period of post-liberalization growth ( ) did not alter that composition significantly. The urbanization happened only later in the growth period i.e when the rural share of population declined by more than 4% while the sanitary and urban areas split the population gain. The subsequent crisis and recovery period has led to little urban gain and in so one may safely conclude that is the only period when the population left the rural areas toward sanitary and urban areas in masses. This of course does not mean that rural-urban shifts of labor through internal migration is unimportant. Rather a good deal of the movement of labor is of the temporary or seasonal kind, with a permanent shift of residence being a limited part of the story. In what follows the urban sector includes both the sanitary and the municipal areas.

5 5 Household Income and Expenditure: The Trend The following table demonstrates the trends in per capita income and expenditure of urban and rural households separately in Thailand. In addition to the nominal values, the table also includes the per poverty line values. These values, often reported as welfare units, record the percentage difference between the per-capita income-expenditure of a household and its recorded absolute poverty line with 100 being an income-expenditure equal to the poverty value. As one might expect the urban incomeexpenditure is substantially higher than the rural and has remained so throughout. It is also interesting to note that in 1988 the income and expenditure of the households in particular that of the rural were very close indicating minimal savings. In welfare units, the numbers are more revealing; in 1988 the per poverty unit of households per capita income was 211, only 2.11 times that of the poverty line. For the rural households, this has been even more sever as the welfare income per unit was only 168. The subsequent growth period has seen the income and expenditure grow very rapidly and more so the income. The biggest gain came in the early periods of growth and that gain continued at a slightly lower pace till the crisis. The subsequent crisis and recovery period has seen only mild increase in nominal income-expenditure. In fact, there has been significant deterioration in welfare income in the first several years in the aftermath of the crisis so much that by 2002, 6 years after the crisis, the welfare income has only mildly exceeded that of Table: Trends in Household Income and Expenditure Per capita Income Rural Urban Per Poverty Unit (%) Rural Urban Per capita Expenditure Rural Urban Per Poverty Unit (%) Rural Urban

6 Overall Rural Urban Figure 0.1: Evolution of Welfare Income, Table () reports the growth of income and expenditure for different sub-periods. As indicated earlier, the income and expenditure grew the fastest in the earlier part of the growth period, For this period and in welfare terms, the income shows an astounding annual average growth of 8.7%. The growth has also been very substantial in the subsequent growth period of with the average annual welfare income growing at 7%. However, this growth has been far from being distributed homogenously among urban-rural households. In the initial growth period the welfare income of urban families grew at 12.1% annually far exceeding that of the rural families which increased at only 5.5%. In the subsequent growth period of , the pattern reversed with welfare income of rural families growing much faster (8.5%) than that of the urban families (4.6%). In the post-crisis period of , the welfare income has fallen rather substantially with the urban families suffering from the biggest decline. Household income has risen after 2000 and here the rural income shows the biggest increase. Overall then the relative income growth of urban households has been only confined to the initial period of growth, and the the subsequent cycles have favored the rural sector. Table: Trends in the Growth of Household Income and Expenditure

7 Per capita Income Rural Urban Per Poverty Unit Rural Urban Per capita Expenditure Rural Urban Per Poverty Unit Rural Urban The rapid growth of was fueled by the rapid increase in the income of both the entrepreneurs-technical class as well as that of laborers and employees and was hindered by the slower growth of the farming class. This to a large extent explains the substantial underperformance of the rural sector in this sub-period. In the subsequent period of , farming income rebounded and grew faster than the average of 7% at 9.7%. But the rural families also expanded their income faster than the urban families from the other two classes, namely, entrepreneur and technical, and laborer and other employees as well. In the crisis and recovery period of the farming sector has fared worse than the average but has significantly rebounded in period. Table: Growth of Welfare Income-Expenditure by Household Class Overall Farming Entrepreneur, Technical Laborers, other Employees Rural Farming Entrepreneur, Technical Laborers, other Employees

8 Famers Laborer, Other Employees Entrepeneurs, Technical Figure: The Evolution of Welfare income by Household Class,

9 9 Per Capital Income by Deciles: Rural-Urban Trends Table() provides the distribution of the per capita income by deciles. In the initial periods of growth , the urban income grew faster than the rural at every decile but progressively so at higher deciles. This implies that while the lower income urban households benefited marginally more than the urban households, the growth of the income for the rich urban families was almost double that of their rural counterparts. In the later stage of the growth period this phenomenon reversed-- the rural families grew their income faster than the urban at every decile and more so at higher deciles. Interestingly in the aftermath of the crisis ( ) the urban families have regained their advantage except for the very high deciles where the rural families have grown their income faster. In recent years ( ) where the growth has restarted, albeit at a much slower pace, the rural families have enjoyed a stronger income growth in particular at the mid deciles. This rural-urban pattern of growth to a large extent has been reflected in the between sector inequality pattern. We will discuss this later in this section (this to be added soon).

10 10 Table: Distribution of per capita income by Deciles (Nominal) D=1 D=2 D=3 D=5 D=8 D=9 D= Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Growth Rates Rural Urban Rural Urban Rural Urban Rural Urban Trend in Inequality: Rural and Urban Areas As one can see from the following graph, the rural-urban inequality differential has changed rather substantially throughout. While, as one might expect, both rural and urban Gini coefficient remained substantially lower than the overall index, the initial growth period demonstrate a substantial increase in urban inequality. Between , both rural and urban Gini coefficient rose. But while rural Gini rose only marginally from

11 0.421 to 0.436, the urban inequality rose substantially greater from to This divergence between inequality rates can also be observed if one considers the kernel density of the urban versus rural households as represented in Figure(). The share of the upper income group in the urban households shows a dramatic increase between while rural households show only a moderate increase towards the median accompanied by flattening of the curve. This pattern resulted in a substantial increase in between group inequality from only 9.17% of total inequality to 13.37%. The dramatic rise in the urban inequality in this period was accompanied by even greater rise in inequality in Bangkok metropolitan area. The subsequent period ( ) shows a mild reduction in the overall Gini coefficient but this reduction has more to do with the drop in urban inequality, which has resulted in reduction of urbanrural inequality. The drop in the urban inequality in this period has been a direct result of the substantial convergence between the upper decile and lower decile as can be observed from the normalization of the density curve. The crisis and the post crisis years (1996-=2000) reversed the trend of the later growth period and inequality increased in the country as whole. As can be seen from the figures below this was due entirely to the growing inequality in the rural sector largely because of the substantial increase of households with low earnings. Urban inequality trended slightly downwards. The post-crisis recovery period again shows reverting to the earlier growth experience of increasing urban inequlity. 11

12 Total Rural Urban Figure: Trends in Inequality, kdensity WIncome x Rural Urban Graphs by year

13 13 Rural-Urban and Regional Inequality In Thailand rural-urban differences in inequality is related to regional inequality. Rates of urbanization vary enormously between the North and the North-East on the one hand and the Central and Southern regions on the other. At the same time the differences in rural-urban per capita income difference as well as the degree of inequality in the distribution in income varies substantially between the regions. Table() provides the decomposition of Theil index of inequality as well as the Gini coefficient by area and region. The decomposition of the Gini index is more problematic because the Gini coefficient in general is not decomposable. Shorrocks (1999), however proposed a decomposition based on the so-called Shapely value that can be used.. This decomposition involves two general steps. In the first step, the decomposition of total inequality into inter-group and intra-group contributions is calculated. In the second step, the total intra-group contribution is decomposed into the sum of the contributions of each of the groups. To calculate the within and between group inequality using Shapely decomposition, the rule of thumb is to initially eliminate intra-group inequality and to calculate inter-group inequality. To do so a vector of incomes is used such that each observation has the average income of its group. Then to calculate intra-group inequality we have to eliminate inter-group inequality. To do so a vector of incomes is used where each observation has its income multiplied by the ratio. The results of this decomposition exercise is provided in Table(). In the upper part of this table, the overall Theil index as well as by area and sector are reported along with the share of the between groups inequality. The overall Theil index shows the same trend as before while the rural and urban Theil index imply considerably lower within group inequality reflecting the contribution of the between group inequality on overall inequality. As reported here the share of between group inequality rose significantly in the initial stages of growth ( ), then showed considerable decline in the later period, a trend that continued in a moderate way in the immediate aftermath of the crisis. It declined during the post-crisis recovery period. It is interesting to note that if regional groupings are considered rather than the rural-urban, then the crisis period shows far greater increase in between group inequality.

14 14 Table: Decomposition of Inequality Indices for Thailand Theil Index Total By Area Rural Urban Between as a % of Total By Region Bangkok Central South North North-East Between as a % of Total Gini Index Total By Area Rural Urban Between as a % of Total By Region Bangkok Central South North North-East Between as a % of Total

15 15 The reported trends in regional inequality demonstrate a rather interesting and revealing picture. In the North and the North-East, where the households are predominantly rural, inequality has risen throughout the growth period with only a moderate decline witnessed in the final years. This is also true for other regions with the exception of Bangkok metropolitan, where the decline in inequality started earlier. Hence, the Gini coefficient draws a picture where the inequality in Bangkok metropolitan has moved rather aggressively in both directions and in so doing has dictated the trend in overall inequality. A major finding here reveals a significant contrast with other East Asian development models. The initial stages of growth period have witnessed a rapid increase in between urban-rural inequality, a rise that has only been partially offset in the subsequent period. The increase in between sector inequality has also been accompanied by a substantial increase in within sector inequality in particular a rapid increase in urban inequality in the initial stages of growth. We will use the next session to further assess the components of this rise in inequality, in particular the m\impact of the rise in share of formal sector. In other East Asian countries such as Taiwan- Korea the poor households move away from the farm sector to other sectors, resulted in substantial reduction in within-group inequality. The evidence provided so far shows a striking difference. II Components of Income Thailand s economy has witnessed rapid change since The process of structural adjustment has seen the share of farm ownership decline rapidly from close to 33% of the households (40% of the population) to close to 19% of the households (21.2% of the population) with the bulk of the decline occurred in the latter part of the growth period i.e This decline has also been witnessed for farm operators. The decline of the farm sector has been accompanied by increases across other non-farm sectors. The biggest increase has been witnessed in the other employees category which include clerical, sales and service workers and production and construction workers. Economically inactive group, which primarily includes those receiving assistance or pensions also shows a rapid increase in particular after This increase has slowed down in the immediate period after the crisis and has since intensified further.

16 16 Distribution of Households (population) Farm Owners Farm Operator Entrepreneurs, Trade Industry Professionals Laborers Farm Workers General Workers Other Employees Economically Inactive Concentrating on the rural households as in Table(), the rapid decrease in the farm-owners class has not been restricted to a specific group, rather it has been felt across all farm-owners, small and large. This decline has also been felt among the farm operators and to a lesser extent the farm workers. Here again the decline of the farm sector has caused a rapid extension of other sectors in particular Other Employees and Economically Inactive. This increase, however, has been even stronger than the overall picture indicating the dominance of the rural households in determining the pace of the overall structural adjustment. Distribution of Rural Households (1) Total Population Average HH Size Farm Owners < 10 rai

17 >=10 and <40 rai >= 40 rai Farm Operators Entrepreneurs, Trade Industry Professionals Laborers Farm Workers General Workers Other Employees Economically Inactive The share of wage and non-farm profit in total household income has risen rapidly in the growth period of This rise has been in particular strong during earlier periods of growth i.e when the share of wage has risen from 35.4% to 39.1% while the share of non-farm profit shows a 2.2% gain from 16.3% to 18.5%. The rapid rise in the share of wage and non-farm profit has been accompanied by a rapid decline in the farm-profit from 15% in 1988 to 12.4% in 1992 and then 11.9% in Furthermore, the rapid monetization of the economy has led to a significant decline in the income in-kind from 24.3% in 1988 to 21.4% in 1988 and then to 18% in The increase in the share of wage income has continued after 1996 while the share of non-farm profit has declined, this has accompanied some initial rise in the share of farm profit and income in-kind. The urban/rural distribution of factor components also confirms these results. However, for the rural families the increase of the wage share has been noticeably stronger. The share of wage factor in rural households income has increased from 24.9% in 1988 to 29.8% in 1992 to further 33.0% in It has then decreased just before the crisis and has only recovered to 1994 pick at The share of non-farm profit for both urban and rural families follow the same pattern of the overall with the noticeable exception of the share of non-farm profit income which does not show a decline after the onset of the crisis. It is also worth noting that share of the current transfers in rural families income has increased almost throughout whereas for the urban families it has declined in the growth period but has increased substantially following the onset of the crisis. The current transfers include components such as assistance and remittance, pensions ad annuities, and terminal pay.

18 18 How this ongoing and somewhat rapid structural adjustment has transformed the rural families is still open for debate. Whether farm families have succeeded in diversifying their income source during this period or that they have altogether abandoned their farming practice is an important question with profound ramifications. This question is even more important considering the fact that the unique pattern structural adjustment in Thailand has not resulted in urbanization rather in families moving from their traditional farm practice to wage bearing occupations or small household business. To assist in understanding this question better Table () provide a different summary of the distribution of the household families by their income source. Here, the decline in the households whose only source of income has been farming has been almost perfectly offset by the increase in rural households whose income source has been restricted to wage or non-farm business. More interestingly, the share of the joint income rural families has trended down. That includes those families reporting both farm income and wage income or business income. A general conclusion that can be drawn from all these is the unique nature of the Thai development process. In this process, the rural families have in general stayed away from permanent migration to the urban areas. However, that has not meant that the farming sector has not suffered, rather the rural families have abandoned their farms but yet stayed in rural areas. The biggest beneficiary of this so-called occupation change has been the so-called wage sector; by mid 1990 the percentage of rural families reporting only wage income has exceeded those reporting only farm income the rural families. The immediate question here concerns the occupational distribution of these so-called rural-wage jobs. This has been addressed in Table () Distribution of Total Household Income by Factors Total Population [????] Average HH size Wages and Salaries Non-Farm Profit Farm Profit Property Income Current Transfers Total Money Income Income In-kind Other Money Income

19 Urban Only % of Total Households Wages and Salaries Non-Farm Profit Farm Profit Property Income Current Transfers Total Money Income Income In-kind Other Money Income Rural Only % of Total Households Wages and Salaries Non-Farm Profit Farm Profit Property Income Current Transfers Total Money Income Income In-kind Other Money Income Percentage Distribution of Rural Households All Rural families Farming Only Wage Only Business Only In-kind Income Only Farming & Wage Farming &Business Farming & Wage & Business Farm Owners & Operators (% of all Rural households) Farming Only In-kind Income Only

20 20 Farming & Wage Farming &Business Farming & Wage & Business Table () summarizes the distribution of the rural wage workers. The share of rural wage workers has trended slightly up in the growth period with the bulk of the increase occurring in the early part of the growth period, i.e reflecting the fact that the growth of the wage sector has been stronger in the rural areas. Consistent with the overall contraction of the agriculture sector, here again the agriculture industry shows the biggest decline in the wage sector while all other sectors show gain to differing degrees. Manufacturing shows tremendous gain, in particular in the period. This increase has slowed down substantially in the subsequent growth period but has continued unabated even in the aftermath of the crisis. Construction has shown tremendous gain rising from only 8.4% of the rural wage sector to 14.7% by 1994 and to more than 20% in Interestingly but perhaps not surprisingly this share is much higher than the overall average indicating a disproportionate concentration of the rural wage workers in the construction sector. Owing to the nature of this sector, the crisis has lead to almost 40% reduction n the share of this sector from 20.2 in 1996 to 12.7 in 1998 and the subsequent recovery period has yet to register a sizable recovery. Service sector is the main sector for which the rural wage share does not show a specific trend. This is an indicator of the nature of the structural adjustment in Thailand where the overall growth of the rural wage sector has been primarily fuelled by exogenous sources. The long term factors that have fuelled the manufacturing boom has lead to tremendous growth in the wage sector in the rural areas. This coupled with cyclical boom in the construction sector has lead to disproportionate strengthening of the wage sector in the rural areas. Distribution of Rural Wage Earners % of total Wage Earners By Industry Agriculture Manufacturing & Mining Construction Commerce Transport & Utility

21 Services By Occupation 1 Professional & Technical Managers Clerical & Sale Workers Services - Transport Agriculture and Fishers Production workers Note: LFS Surveys 21 The wage sector in the rural area has advanced faster than the national average so much that the percentage of rural manufacturing wage earners has increased from only 35.6% in 1988 to 51.5 by The crisis has dampened the growth of the manufacturing and in so has hit the rural areas harder. By 1998 the percentage of rural manufacturing wage earners has fallen to 45.9 and by 2002 has yet to recover to the heydays of the growth era. Reflecting the general state of the manufacturing in Thailand, the resource-based manufacturing has dominated in the early years. However, the growth of the labor-intensive manufacturing dominated the growth of the rural manufacturing wage rising from 17.0 to 22.1 in 1992 and continued rising even after the crisis. Interestingly enough, the capitalintensive manufacturing also shows some healthy gains in the growth period. The rural manufacturing sector in Thailand has been mainly concentrated in the central region around Bangkok. There has been slight share gain for the central region at the expense of the south region in particularly in the growth period. However, all the rural regions have expanded their manufacturing sector substantially during this period. It is also interesting to note that at the beginning our period (1988) a sizable portion nearly 37% of the wage workers in rural manufacturing was concentrated in small-scale establishments with employment size of less than 10 workers. The growth period saw a massive shift towards largescale manufacturing with the proportion of employment in >100 establishments exceeding 50% of total rural manufacturing employment in The subsequent crisis has negatively affected rural manufacturing in general. However, the effect has been primarily felt in small scale and particularly medium scale manufacturing. In fact, the overall employment of large-scale establishment in this period grew rather substantially: the employment share of large establishments reached more than 60% which exceeds that of urban manufacturing. Interestingly enough several years

22 22 after the crisis by year 2002, the total employment of small and medium scale <100 rural manufacturing is still below that of Distribution of Manufacturing Rural Wage Earners % of all Manufacturing By Manufacturing Category Resource based Labor Intensive Capital Intensive Differentiated and Science Based By Region Central South North & North-East By Size > Note: LFS Surveys, Labor intensive, Textile-Garment, Resource Intensive such as Food Capital Intensive such as machinery, Differentiated such as appliances, electronics. Table() and Table() present by deciles the distribution of the factors of income together with the household characteristics. The lower deciles are pre-dominantly rural families and that has been so for all sub-periods. In fact the ratio of urban families in the first deciles has been between 6 and 7 percent throughout and the same type of analysis is also applicable to the second and third deciles. The reverse can be said about the higher deciles where the predominant families are urban. Interestingly the percentage of female-run families appear to be trending up and while the data does not show much a of a positive relation between gender and income at the lower deciles, there is a definite gender gap between the lower and the higher deciles. For instance, while in 1996 only 20.3% of the second deciles families were headed by a female, 25.6% of the families in the ninth deciles category had a female head. Among the rural families, the landowners are by far the most likely to be poor. Not surprisingly, the land-owners rely on the so-called income inkind as their primary source of income. Interestingly the richer rural

23 families include those with wage or non-farm profit as well as those that report high level of farm profit. (more ) 23

24 24 Table: Distribution of Components of per capita income D=1 D=2 D=3 D=5 D=8 D=9 D= Income Components (%) Wages and Salaries Non-Farm Profit Farm Profit Income In-kind HH Characteristics HH Size Urban (%) Age (Years) Sex (% Female) Income Components (%) Wages and Salaries Non-Farm Profit Farm Profit Income In-kind HH Characteristics HH Size Urban (%) Age (Years) Sex (% Female) Education (Years) Income Components Wages and Salaries Non-Farm Profit Farm Profit Income In-kind HH Characteristics HH Size Urban (%) Age (Years) Sex (% Female) Education (Years) Income Components Wages and Salaries Non-Farm Profit Farm Profit Income In-kind HH Characteristics HH Size Urban (%) Age (Years) Sex (% Female) Education (Years)

25 25 Table: Distribution of Components of Per capita income (Rural) D=1 D=2 D=3 D=5 D=8 D=9 D= Income Components (%) Wages and Salaries Non-Farm Profit Farm Profit Income In-kind HH Characteristics HH Size Land owners (%) Age (Years) Sex (% Female) Education (Years) Income Components (%) Wages and Salaries Non-Farm Profit Farm Profit Income In-kind HH Characteristics HH Size Land owners (%) Age (Years) Sex (% Female) Education (Years) Income Components Wages and Salaries Non-Farm Profit Farm Profit Income In-kind HH Characteristics HH Size Land owners (%) Age (Years) Sex (% Female) Education (Years) Income Components Wages and Salaries Non-Farm Profit Farm Profit Income In-kind 58, HH Characteristics HH Size Land owners (%) Age (Years) Sex (% Female) Education (Years)

26 26 Trends in Between and Within Sector Inequality Differential The following Table presents trends in inequality indices for per capita as well as welfare income. The Gini coefficient rises in the earlier segment of the growth period and then falls. This initial increase especially for the period has been rather substantial from to for per capita income and from to for the welfare income. The gradual fall in the inequality in the subsequent period was not sufficient to reverse the initial rise in inequality fully and in so by 1996 the inequality remains still higher than 1988 and only marginally lower than that of The crisis has reversed this downward trend to some extent and only after 2000 a noticeable improvement in the post-inequality index has been witnessed. Looking at the quantile ratios, one can see that the P90/P10 demonstrate similar pattern as the Gini coefficient while this cannot be said about the P90/P50. The share of income for the 9 th decile has remained almost constant throughout the initial growth period and shows a rather significant drop in the growth period of This implies that while the medium income earners have reduced their income gap with the high earners, the poor have lost ground. Hence the overall increase in the inequality index can be attributed mostly to the fact that the poor has not been able to grow their income as fast as the rich. A kernel representation of the distribution function for the farm, nonfarm self-employed and the wage sectors in Thailand is also presented in Figure(). While the distribution of the non-farm self-employed and wage sectors re closely matched, the distribution of the farm sector lies way to the right of both distributions. This shows the potential existence of substantial earnings gap for new earners, those moving away from the farm sector. Interestingly, this gap widened rather dramatically in the period, with the farm households distribution showing little change while both non-farm self-employed and wage sectors moved significantly to the right. Hence the increase in the relative income of the 9 th decile with respect to the 1 st decile as reported before has more to do with the increase in the earning gap between the farm sector and the rest of the economy.

27 Table: Thailand- Household inequality Static Inequality Decomposition Results (Household Data) Per Capita Income Gini Coefficient General Entropy E(0) E(1) E(2) Quantile Ratios P90/P P90/P P10/P Welfare Income Gini Coefficient General Entropy E(0) E(1) E(2) Quantile Ratios P90/P P90/P P10/P kdensity WIncome x Farm Wage Earner Self Employed Graphs by year

28 28

29 III 29

30 30 The Decomposition of Inequality by Factor Components Methodology This section discusses and measures quantitatively the contribution of different sources of income to the over-all inequality measure for different groups and periods documented above. It might seem at first sight that the overall inequality index, like the Gini, would simply be the weighted average of the Ginis of the individual components. But this is not so since the income from particular source might be distributed among the recipients of this type of income in a different way than the total income is distributed among the members of the group we are considering. The widely used technique to deal with this problem is that of the pseudo-gini. The pseudo-gini coefficient for a particular income component is similar to the Gini coefficient for that component, except for the modification that the income recipients are ranked not in terms of the income from this source, but in terms of their total income from all sources. The over-all Gini is then the weighted sum of the pseudo- Gini s of the components, where the weights are given by the share of the component in total income (Shorrocks 1982). The impact of each source on total inequality has been traditionally measured by the relative share of each source on total inequality index. Based on this, each factor will have an equalizing effect only when its corresponding share is negative, or when the corresponding pseudo-gini is negative. This interpretation, however, has been criticized by Podder (1993). He suggests that those factors that have higher pseudo-gini than the overall Gini will contribute positively to the inequality index while those with lower pseudo-gini will contribute negatively and in so doing will be considered to be equalizing. The pseudo-gini for a particular component divided by the true Gini for that component can be shown to be equal to the rank correlation co-efficient between the income from that source and total income (called the Gini correlation). A higher value of the Gini correlation means that the particular income source is contributing more to the income of the richer in the group and hence ceteris paribus enhancing inequality. The contribution of a particular component to the overall inequality (Gini) can then be found in two alternative ways: (i) the product of the pseudo-gini of the component and its share in the total income; or (ii) the products of three terms: the true Gini of the component, its share in total income, and the Gini correlation.

31 A formal algebraic account of these relationships, following Silber (1993) and Fluckiger and Silber (1995) is given in the following section. 31 Assume X it denote the value of income source i for individual j and let X i and X j are defined as: n X.i = j=i I X j. = i=1 X X ji ji Here I is the total number of income sources while n stands for the number of individuals. Furthermore, define S ji, S i, and S j as: S ji = X S.i = X.i j. = X ji S j. / X / X / X in which X represents the total income of the population, S j stands for the share of income source i in total income and S j denotes the share of individual j in total income. Borrowing from Silber's (1989) analysis of the decomposition of income inequality by income factors, it is possible to write the Gini Index (I G ) of overall income inequality as: I G = [ e ] G [ S ] where e is a (1 by n) vector of population shares, each equal to (1/n), while [S] represents a [n by 1] vector of the income shares S j. Here, G is a (n by n) matrix with elements g hk equal to 0 if h=k, equal to 1 if h<k and equal to +1 if h>k. Here, income shares S j are ranked by decreasing value of the total income of the various individuals. The share of S J individual j can also be represented by: S j. = S I i=1 ji