Anne Booth a a School of Oriental and African Studies, London. To link to this article:

Transcription

1 This article was downloaded by: [Macquarie University] On: 21 July 2013, At: 05:30 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: Registered office: Mortimer House, Mortimer Street, London W1T 3JH, UK Bulletin of Indonesian Economic Studies Publication details, including instructions for authors and subscription information: Poverty and Inequality in The Soeharto Era: An Assessment Anne Booth a a School of Oriental and African Studies, London Published online: 21 Aug To cite this article: Anne Booth (2000) Poverty and Inequality in The Soeharto Era: An Assessment, Bulletin of Indonesian Economic Studies, 36:1, , DOI: / To link to this article: PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content ) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings,

2 demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at

3 Bulletin of Indonesian Economic Studies Vol 36 No 1, April 2000, pp POVERTY AND INEQUALITY IN THE SOEHARTO ERA: AN ASSESSMENT Anne Booth* School of Oriental and African Studies, London This paper surveys the evidence on trends in poverty and inequality during the years of Soeharto s presidency. That Indonesia saw a decline in the incidence of poverty, and improvements in other welfare indicators such as infant mortality rates and literacy rates over these years, seems indisputable. Comparative studies show that by the latter part of the 1980s, the headcount measure of poverty in Indonesia was below that in the Philippines although above that in Malaysia and Thailand. But relative poverty has declined more slowly, and indeed increased in some urban areas between 1987 and The paper also examines evidence on the determinants of rural poverty in Indonesia in 1993, and suggests that rural development programs targeted to the specific needs of poor people in poor areas will be essential if rural poverty is to be further reduced in future years. Thee Kian Wie is an economic historian by training, and although much of his own research has been contemporary in focus, he has never lost his enthusiasm for the historical study of economic development, both in Indonesia and in other parts of the world. As Soeharto s New Order fades into history, I trust that he will appreciate a paper which explores how one important aspect of the New Order s economic record might be evaluated by future economic historians. Certainly, in the Indonesian context, no one individual since van den Bosch has stamped his personality on the economic policies of an era as has Soeharto and, right up until the crisis hit in late 1997, it was the economic achievements that above all others conferred legitimacy on his regime. My view is that these achievements were substantial, and in spite of the severe economic contraction of they will endure. But they are by no means fully understood, and indeed I would argue that they have often been misrepresented by defenders and critics alike. A full examination of Soeharto s economic legacy is obviously beyond the scope of a journal

4 74 Anne Booth article; my more limited aim here is to review some of the evidence relating to poverty and income distribution, and to discuss the policy implications of this evidence. THE SITUATION IN THE 1960s How poor were Indonesians in March 1966, when the New Order effectively began? The available evidence indicates that they were very poor indeed. Per capita income in Indonesia was well below that of other Southeast Asian economies for which reliable data are available; indeed real per capita GDP is estimated to have been only 535 international dollars (1985 prices) in 1966, compared with 650 dollars in India. 1 The time series compiled by van der Eng (1992: table A-4) shows that real per capita GDP (in 1983 rupiah) was about the same in as in The household survey data collected between November 1964 and February 1965 indicated that in much of Java over half the rural population fell below a very modest poverty line set in terms of the price of cereals (20 kg per capita per month). The situation was little better in urban areas. Outside Java the incidence of poverty was on average lower, although in Bali, Northern Sulawesi and West Nusa Tenggara it was little different from that in Java (Booth 1988: table 4.19). Only in some parts of Sumatra and Kalimantan was the headcount measure markedly lower than in Java. The stabilisation policies carried out by Soeharto and his economics team from 1966 to 1969 were very successful in reducing inflation to single-digit levels, and in returning the economy to sustained economic growth. Their effect on income disparities was more contentious. King and Weldon (1977: 703 6) compared data on household income and expenditures from several sources for Java over the years to They found evidence of growing inequalities in urban areas (especially Jakarta) although in rural areas the trend was far less marked. In , when inflation and economic dislocation were at their height, the Gini coefficient of consumption expenditures in urban Java (excluding Jakarta) was only 0.3; in 1967 it had fallen slightly to But by it had increased to It appears that inflation and economic stagnation had a more severe impact on urban workers, especially those with fixed incomes such as civil servants. In rural areas, the better-off farmers with a surplus of food to sell probably increased their incomes relative to both urban workers and the rural poor. This would explain the quite low urban rural disparities in consumption expenditures that prevailed in the mid 1960s in Indonesia, and also the surprising finding that inequalities were

5 Poverty and Inequality in the Soeharto Era: An Assessment 75 TABLE 1 Trends in the Gini Coefficient of Household per Capita Consumption Expenditure Year Urban Rural Urban and Rural Sources: BPS, Statistical Yearbook of Indonesia, various issues; : BPS (1997a). actually lower in urban than in rural areas (Sundrum 1973: tables 5 and 7). These trends were reversed only slowly as inflation fell and economic growth accelerated. By , the Gini coefficient of per capita consumption expenditures in rural Indonesia was 0.34, indicating a moderate degree of inequality. It was slightly lower in urban areas (table 1). International comparisons are difficult, as data for other parts of the developing world often refer to household income rather than expenditure. 2 But there is little reason to think that inequalities in rural Indonesia were much lower than those in the rest of Asia, although they were probably lower than in Latin America or Africa. In urban areas, the cost of living surveys carried out in indicated that the Gini coefficient of household income was close to, or rather greater than, 0.4 in Jakarta, Manado and Yogyakarta, although it was lower in Bandung and Surabaya, and in most of the large towns outside Java (table 2). Bearing in mind that the Gini coefficient of urban household incomes in the Philippines in 1971 was 0.44, and that that country is widely considered to have an unequal distribution of

6 76 Anne Booth TABLE 2 Urban Inequalities in Indonesia: Gini Coefficient of Household Income City a Jakarta Bandung Semarang n.a Yogyakarta Surabaya Medan Padang n.a Palembang Pontianak n.a Banjarmasin Manado Ujung Pandang n.a Denpasar Mataram n.a Kupang n.a Urban Philippines 0.50 (1965) 0.44 (1971) 0.43 (1988) a Data for Banjarmasin and Manado refer to Sources: Sundrum (1974): table 23; BPS (1990): appendix 2. Philippines data from Balisacan (1993): table 3. income at least by Asian standards, it would appear that urban income disparities in Indonesia s larger towns and cities in the latter part of the 1960s were already quite marked. Although the growth in per capita consumption expenditure that occurred after 1968, combined with a slight fall in rural expenditure inequalities, would have led to some decline in poverty in both urban and rural Indonesia between and , there was still widespread concern about the problem of poverty, especially in rural Java. Using the poverty lines proposed by Professor Sajogyo, Sundrum and Booth (1980: table 7) estimated that 61% of the rural population in Java was poor in 1970, and 21% destitute. 3 In urban Java the percentage considered destitute was 26%, while outside Java it was 21% in urban

7 Poverty and Inequality in the Soeharto Era: An Assessment 77 areas and 15% in rural areas. While the Sajogyo poverty line was criticised by some analysts, there could be little doubt that the incidence of poverty was very high in many parts of Indonesia in 1970, whatever the poverty line concept used (Esmara 1986: table 9.7). Ravallion and Datt (1996a: 2,480) point out that in 1970, using a comparable poverty line, India s headcount measure of poverty was very close to that of Indonesia. So there can be little doubt that, in spite of the economic recovery that occurred in the latter part of the 1960s, the great majority of Indonesians were still poor by Asian standards when the first of the Soeharto era fiveyear plans got under way at the end of that decade. GROWTH AND EQUITY DURING THE 1970s The 1970s were, of course, the years when the rapid increase in the world price of oil had important consequences for both economic growth and income distribution in Indonesia. By the mid 1970s there was considerable debate about the distributional consequences of the oil boom, especially in rural areas of the country. It was argued that while the oil boom had led to rapid growth in government expenditure, the benefits of this growth were uneven by sector and region, with urban areas benefiting to a disproportionate extent. In addition, several analysts were pointing to the adverse consequences of the real appreciation of the rupiah (popularly known as the Dutch disease ) for the incomes of producers of non-oil export commodities, especially in the agricultural sector (see, for example, Paauw 1978, for an early statement of this argument). While the rice sector was protected via import controls and producer subsidies, it was argued that the oil boom was harming many millions of smallholder producers of rubber, coffee, tea, pepper and copra. Even before the full effects of the oil boom were felt, the very sharp increase in the price of rice that occurred in was thought to have adversely affected the poorer segments of the population, in whose consumption basket rice and other basic foods had a higher weight. In short, there appeared to be plenty of evidence to support the view that the oil boom was making the rich (especially in urban areas) richer, while the poor might actually have become worse off. With the benefit of hindsight and some careful appraisals of the statistical evidence, it now appears that the rapid inflation induced by both the rice shock of and the onset of the oil boom did lead to faster growth in the cost of the consumption basket consumed by the bottom 40% of the expenditure distribution, especially in rural areas (Asra 1989: table 3). In addition, real per capita expenditure in rural Java on

8 78 Anne Booth TABLE 3 Headcount Measure of Poverty According to the Central Bureau of Statistics Poverty Line, Year Percentage below Numbers in Poverty the Poverty Line (millions) Urban Rural Urban Rural Total 1970 a (53.6) (38.7) a The 1970 estimates are derived from a poverty line based on the 1976 BPS poverty lines, deflated using the price indices for the bottom 40% of the expenditure distribution, as estimated by Asra (1989): table 3. Sources: BPS (1992): table 3.1; BPS (1994): table 3.1; Department of Information (1998): table IV-1. average declined between and 1976 (Booth 1992: table 10.2). Thus the percentage of the population and the numbers falling below the official poverty line actually increased in rural areas between 1970 and 1976 for the country as a whole (table 3). Inequalities, too, increased between and 1976, in both urban and rural areas. As Asra (1989: table 1) has demonstrated, if the 1976 rural expenditure data are corrected for differential changes in prices by decile groups, then rural expenditure inequalities increased in both Java and the Outer Islands, rather than declining as shown in the unadjusted data (table 1). In urban Java inequalities in expenditure also increased; Booth and Sundrum (1981: table 7.22) estimate that expenditure in the top decile group in urban Java grew by 66% between 1970 and 1976, compared with an increase of less than 20% in the bottom decile groups. On average, real per capita expenditure increased more rapidly in Jakarta

9 Poverty and Inequality in the Soeharto Era: An Assessment 79 TABLE 4 BPS Poverty Line as a Percentage of Average per Capita Consumption Expenditure, Year Urban Rural Source: BPS, Survei Sosial Ekonomi Nasional (Susenas), various years; other sources as for table 3. than in other urban areas, and more rapidly in urban Java than in urban areas of the Outer Islands (Booth and Sundrum 1981: table 7.20). A consequence of these trends was a sharp increase in urban rural disparities, especially on Java and especially for non-food items. Some of these trends were reversed in the years from 1976 to Asra s calculations suggest that between 1976 and 1981 the rate of inflation was broadly similar for all decile groups in the expenditure distribution, both in Java and elsewhere (Asra 1989: table 2). In rural areas there was a discernible fall in expenditure inequality. In rural Java on average, real per capita expenditure increased rapidly, although it fell in rural areas outside Java (Booth 1992: table 10.2). The upshot of these trends was a marked decline in the incidence of poverty in both urban and rural areas, although because of faster population growth in urban areas, the numbers under the official poverty line fell more sharply in rural areas (table 3). By 1981 it was estimated that there were 40.6 million Indonesians under the official poverty line, of whom 9.3 million were in urban areas. Over the 1970s, average per capita consumption expenditure had increased more rapidly than the official poverty line used by BPS (the Central Bureau of Statistics, since renamed the Central Statistics Agency). Whereas in 1970 the urban poverty line was over 86% of average per capita expenditure, by 1980 it had fallen to only 56% (table 4). The fall in

10 80 Anne Booth TABLE 5 Estimates of Relative Poverty, 1970 and 1980 Proportion of Population Spending Below 50% of Average per Below 33% of Average per Capita Consumption Expenditure Capita Consumption Expenditure Urban Java Outer Islands Indonesia Rural Java Outer Islands Indonesia Urban + Rural Java Outer Islands Indonesia Sources: 1970: BPS (1973); 1980: BPS (1982a, 1982b). rural areas relative to average per capita expenditure was less rapid but still appreciable. 4 Given these trends it is of some interest to look at trends in relative poverty during the 1970s, where relative poverty is defined in terms of the proportion of the population spending less than one half, or less than one third, of average per capita consumption expenditure. 5 In urban Java, relative poverty defined in both these ways actually increased between 1970 and 1980, although in urban areas of the Outer Islands and in rural areas it fell (table 5). Certainly the decline in the incidence of both absolute and relative poverty in rural areas of Indonesia over the 1970s was a remarkable achievement, especially in comparison with other petroleum economies with substantial agricultural sectors. To what extent was it due to government policy? It is certainly true that the oil boom did see an increase

11 Poverty and Inequality in the Soeharto Era: An Assessment 81 in government expenditure, although little of this increase was explicitly targeted to the poor. Even the regional development grants, or Inpres grants as they came to be known, were not specifically designed to alleviate poverty, but rather to give provincial and subprovincial levels of government more scope to carry out much needed infrastructure rehabilitation and development. 6 After the mid 1970s, the employment generated by the Inpres kabupaten fell continuously (Asher and Booth 1992: table 2.12). The sharp fall in poverty over the latter part of the 1970s was due in part to the successful stabilisation of food prices which, especially in Java, meant that the poor experienced a lower rate of inflation than the rich (Asra 1989: table 3). In addition, the growth in agricultural production that occurred in the 1970s and early 1980s, in considerable part due to the successful dissemination of new production technologies in the foodcrop sector, created new employment opportunities in production, processing and marketing. Indonesian agriculture remained very unmechanised and labour intensive compared with agriculture in most other parts of Asia well into the 1980s, because the small average size of holdings made purchase of agricultural machinery uneconomic for most farmers. In addition, the devaluations of 1978 and 1983 raised the price of farm machinery and discouraged its adoption (Booth 1988: 181). The oil boom also led to rapid growth of sectors producing non-traded goods, such as construction and trade, and this growth provided additional employment opportunities for unskilled workers. More and more people from densely settled rural areas migrated to urban areas on a temporary or permanent basis to seek work. The 1983 Agricultural Census sample survey of farm incomes indicated that, on average, farm households in some provinces in Java were receiving over half their income from non-farm activities, including wage labour (BPS 1987: table 9). In many provinces outside Java, reliance on farm income was greater, reflecting both the larger average size of holdings, and the lower availability of off-farm employment. The Social Accounting Matrices (SAM) constructed by the BPS showed that in 1975, poor farm households (defined as those controlling less than 0.5 hectares) received almost 40% of their income from off-farm sources, and in absolute terms their offfarm income was higher than for medium or large farmers (table 6). This was partly reversed in 1980, when medium farmers (those controlling between 0.5 and one hectare) began to earn more in absolute terms from off-farm activities than small farmers. But the growth in off-farm income for small farmers was significant over these years, and must have played a crucial role in pushing many thousands of households controlling small amounts of land over the poverty threshold.

12 82 Anne Booth TABLE 6 Sources of Income for Agricultural Households, Year Poor a Medium a Rich a Percentage of total household income accruing from farm operation Non-agricultural income per farm household (Rp 000) a Poor farm households are those cultivating less than 0.5 hectares, medium, 0.5 to one hectare, and rich, over one hectare. Source: BPS (1996): table 4.37; BPS (1999): table The SAM tables also indicated that, over the decade , per capita incomes of farm labourer households (the poorest of the socioeconomic groups in the 1975 SAM) grew slightly faster than the national average (table 7), while those of poor farmers (those operating less than 0.5 hectares) grew more slowly, so that by 1985 farm labourer households were earning slightly more on average than poor farm households. But middle and especially larger farmers had increased their incomes even faster over these years, relative both to the national average and to incomes of farm labourers. The income data collected in the 1976 Labour Force Survey (Sakernas) indicated that, in both Java and the rest of the country in 1976, the Gini coefficient of household income distribution was in fact higher in rural than in urban areas, and higher for agricultural households than for any other category of household (Sigit 1985: tables 2 and 3). Clearly we cannot assume that the distribution of incomes within

13 Poverty and Inequality in the Soeharto Era: An Assessment 83 TABLE 7 Agricultural Household Incomes as a Percentage of the National Average, Year Labourer Poor a Medium a Rich a a See table 6 for definitions of poor, medium and rich. Source: BPS (1999): table the agricultural sector was especially egalitarian at the time the green revolution was gathering pace in Indonesia in the mid 1970s. Neither can we assume that agricultural household incomes became more equally distributed as a result of the dissemination of the new agricultural technologies, even if the years from 1976 to 1987 were characterised by some decline in inequality in the distribution of per capita consumption expenditures in rural areas (table 1). 7 STRUCTURAL ADJUSTMENT POLICIES AND POVERTY DECLINE, One of the most striking aspects of the decline in poverty in Indonesia since the mid 1970s, as depicted in the official statistics, is that it continued more or less regardless of changing macroeconomic policies. Between 1976 and 1981, the years of the oil boom, the annual average decline in the numbers below the BPS poverty line was 5.6%. After 1981, Indonesia s revenues from oil exports began to fall, and the government adopted a series of measures designed to increase non-oil exports, diversify the domestic tax base, attract more foreign investment, deregulate the financial sector, and improve the efficiency of public sector enterprises,

14 84 Anne Booth TABLE 8 Estimates of Relative Poverty, Year Urban Rural Urban + Rural Percentage of the population spending: Below 50% of average per capita consumption expenditure Below 33% of average per capita consumption expenditure Sources: BPS (1973); BPS, Statistical Yearbook of Indonesia, various issues; BPS (1997a). measures which have been much discussed (e.g. Pangestu 1991; Azis 1994). Over the years from 1981 to 1987, GDP growth was lower than in the oil boom years, and government expenditures were cut back. But in spite of fiscal austerity and slower GDP growth, the annual percentage decline in the numbers under the official poverty line was only slightly slower than in the oil boom years. In addition, the incidence of relative poverty declined, in both urban and rural areas (table 8). The Indonesian experience of has attracted considerable attention as one of the few examples of a sustained program of macroeconomic policy reform, along the lines advocated by the World Bank and the IMF, that has not had an adverse effect on poverty (Stewart 1995: 206 8; Thorbecke et al. 1992). 8 Thorbecke et al. (1992: 135 6) developed a structural adjustment model for Indonesia that simulated

15 Poverty and Inequality in the Soeharto Era: An Assessment 85 the impact of several different retrenchment policies in the period. They found that the policy package actually adopted came quite close to the most favourable simulation in terms of the impact on poverty. They argued that a more drastic budget retrenchment would have resulted in overkill. Economic stagnation and a worsening income distribution would have been a very high price to pay for maintaining the level of public debt prevailing at the outset of the crisis. The budget cuts implemented after 1981 were most severe in the more capital-intensive sectors such as energy, transmigration and grants to public enterprises, which did not have much direct impact on employment (Asher and Booth 1992: table 2.7). Government expenditure as a share of GDP (estimated in international dollars; see note 1) fell from 16.1% in 1982 to 14.5% in 1984, and remained at around this level for the rest of the 1980s. 9 Indonesia in fact managed to maintain a reasonably expansionary fiscal stance between 1982 and 1988 while at the same time keeping inflation in check in the wake of two substantial devaluations, thus engineering a substantial real devaluation over these years (Intal 1992: 113). One reason was that domestic monetary growth was held down, although not at the expense of the credit needs of the export sector. In addition, the government relied on foreign borrowing to balance the budget, which reduced the necessity for borrowing from the domestic monetary authorities with potentially inflationary consequences. Well judged macroeconomic policy thus created an environment in which producers could take advantage of the opportunities created by the large real devaluation to expand traded goods production. At the same time, improved infrastructure and flexible labour markets made it easy for people to move to take advantage of new employment opportunities, or indeed to diversify their sources of income while staying in rural areas. In their analysis of the Susenas for 1984 and 1987, Huppi and Ravallion (1991: 1,672 3) stressed the importance of increasing wage incomes for poor self-employed farmers in Central and East Java. These two provinces accounted for over half the total income gains made by the rural farming poor in the country as a whole over these three years. Outside Java, where wage earning opportunities were less abundant, there was less income diversification, and wage earnings were of little importance in poverty alleviation (Huppi and Ravallion 1991: 1,673). THE RETURN TO RAPID GROWTH, Over the period , Indonesia managed to achieve a substantial real devaluation without a severe domestic recession. What Intal (1992) termed the policy of aggressively adjusting the exchange rate as the

16 86 Anne Booth TABLE 9 Regional Breakdown of GDP, Population and Poverty Incidence, 1996 (%) Region GDP a Population Poor Poverty Population Incidence b Jakarta Java/Bali Mining four c Sumatra Kalimantan Sulawesi Eastern islands d Indonesia a GDP data are provisional. b Poverty incidence refers to the percentage of the population under the BPS poverty line in c Provinces of Aceh, Riau, East Kalimantan and Irian Jaya. d Provinces of West and East Nusa Tenggara, East Timor and Maluku. Sources: Department of Information (1998): table IV-4; BPS (1997b). terms of trade declined thus paid off. Non-oil export industries grew rapidly, in response to both the real depreciation of the rupiah and the other reforms aimed at improving incentives for exporters, including the introduction of a duty drawback scheme. The government was also able to push through major tax reforms and, in 1988, a substantial deregulation of the financial system (Nasution 1993; Cole and Slade 1996: ). But although these reforms ushered in a period of faster GDP growth after 1987, which was sustained right up until 1997, the years from 1987 to 1996 witnessed a slower decline in the numbers under the BPS poverty line. This was especially true in rural areas: whereas from 1976 to 1987 the number of people below the official poverty line in rural areas declined at almost 7% per annum, between 1987 and 1996 the annual rate of decline slowed to just over 3% per annum. Drawing attention to this slowdown in a speech in 1994, President Soeharto suggested that increasingly the poor were isolated in pockets of poverty which government development programs were not reaching. By the mid 1990s it was

17 Poverty and Inequality in the Soeharto Era: An Assessment 87 TABLE 10 Estimates of Relative Poverty, 1987 and 1996 Proportion of Population Spending below 50% of Average per Capita Consumption Expenditure Urban areas Jakarta West Java Central Java East Java North Sumatra South Sumatra South Sulawesi Indonesia Rural areas Indonesia Whole country Sources: BPS, Statistical Yearbook of Indonesia (1989, 1996); BPS (1997a). obvious that these pockets of poverty were located in many parts of the archipelago and not just in Java: in 1996 the incidence of poverty was higher in Kalimantan and Eastern Indonesia than in Java and Bali (table 9). The unadjusted household survey data show a rise in inequalities in urban per capita consumption expenditure between 1990 and 1996 (table 1). The household income data collected in the urban cost of living surveys (not yet available for the 1990s) show more or less constant inequality in Jakarta from the late 1960s until the late 1980s, but an increase in the Gini coefficient in Bandung and Surabaya (table 2). Although the proportion of the population under the official poverty line continued to fall in both urban and rural areas, there was an increase in the incidence of relative poverty (defined as the proportion of the population spending less than half the monthly average) in most parts of the country. It was especially pronounced in Jakarta, and in urban areas of West Java (table 10). The rapid population growth of greater Jakarta, and of other large cities in Java, due mainly to a huge influx of migrants, has obviously

18 88 Anne Booth led to sharper disparities in consumption expenditure and living standards. In their study of trends in household expenditure inequalities in Indonesia between 1987 and 1993, Akita, Lukman and Yamada (1999: table III) demonstrate that both Theil and Gini indices increased in urban areas over these years, although most of the increase occurred between 1987 and There was a slight decline in rural inequalities. These authors decomposed the Theil indices by region and by educational achievement. They found, perhaps surprisingly, that interprovincial inequalities are not a major factor in overall national inequalities, confirming earlier results which showed that interprovincial disparities in household expenditures are far smaller than interprovincial disparities in GDP (Akita and Lukman 1995). They argued that rural urban inequalities accounted for a much larger share of overall inequalities and thus that policies aimed at reducing them should be given high priority. They also argued that educational attainment was a significant determinant of expenditure inequality, accounting for around 33% of total inequalities by This finding was confirmed by Cameron (1999: 26), who concluded from her analysis of the household expenditure data for 1984 and 1990 that increased educational attainment was the largest single determinant of the inequality increase over these years. The policy implications of these findings are examined in the next section. Between 1993 and 1996, the Gini coefficient of per capita expenditure in urban Indonesia increased from 0.33 to 0.36; in rural Indonesia it increased only slightly, and remained much lower than that for urban areas (table 1). The increase in relative poverty in rural areas was also much more modest; by 1996 the proportion of the rural population spending less than half of average expenditure was about half that in urban areas (table 10). Some students of rural change in Indonesia find the evidence of quite low expenditure inequalities hard to accept. It must be stressed again that these data refer to expenditure only, not to income or wealth, and attention has already been drawn to the evidence that income inequalities in rural areas of Indonesia in the mid 1970s were much higher than inequalities in per capita consumption expenditure (Sigit 1985). 10 There is also evidence that since the mid 1980s the mechanisms promoting an egalitarian distribution of income in rural areas may have been working less effectively than in the decade For example the SAM data show that, between 1985 and 1998, incomes from off-farm employment rose faster for richer farmers than for poorer ones. By 1993, farmers owning more than one hectare of land were earning about the same proportion of their total income from off-farm activities as were farmers owning less than half a hectare, but the absolute amount

19 Poverty and Inequality in the Soeharto Era: An Assessment 89 was much larger (table 6). Indeed household incomes for all types of farm households were lower relative to the national average in 1993 than they were in 1990, but only for farm labouring households were they much lower than in 1975 (table 7). To the extent that off-farm employment was having an equalising effect on rural incomes in the late 1970s and early 1980s, this was apparently no longer the case by the early 1990s. The 1990s also saw a steep decline in all agricultural incomes relative to the national average, but the decline was sharpest for agricultural labourer households (table 7). By 1998, per capita disposable incomes in farms operating more than one hectare of land were over three times those of farm labourer households, compared with less than twice in 1980 (BPS 1999: table 3.14). But the SAM data also show that disparities widened between rural and urban areas and especially between the rural poor and the urban well-to-do. In 1985 the upper stratum of urban households had an average disposable income per capita of less than four times that of agricultural labour households; by 1998 it was 9.5 times the level of agricultural labour households. 11 THE CHANGING TRADE-OFF BETWEEN GROWTH AND EQUITY IN INDONESIA The lesson of the years from 1987 to 1996 would seem to be that the elasticity of poverty decline with respect to economic growth fell in Indonesia. In other words, the rapid growth of these years was accompanied by increasing inequality, especially in urban areas, and this increase in inequality reduced the impact of the growth on poverty decline. 12 Much more analysis of the available data is needed before we can arrive at convincing explanations for this, but several hypotheses suggest themselves. Since 1987, the manufacturing and modern service sectors have been the principal engines of growth in the Indonesian economy; the agricultural sector has been relegated to a secondary role in policy debates, and receives a falling share of budgetary resources. In other parts of Asia, it has been argued that the sectoral composition of economic growth has an important effect on poverty reduction. Ravallion and Datt (1996b: 19) argue that in India both primary and tertiary growth have reduced poverty nationally and within urban and rural areas, whereas secondary sector growth has had little discernible impact on poverty reduction. Thus at least part of the reason why growth has been

20 90 Anne Booth less pro-poor in Indonesia since 1987 could be that it has relied less on agriculture and more on industry. Another reason put forward in some official documents, and espoused by former President Soeharto, is that increasingly since the mid 1980s, the poor in Indonesia have been located in remote regions outside Java, and unable to benefit from Java s rapid manufacturing-based growth. To the extent that these regions are characterised by poor soils, erratic rainfall and poor market access, they have not benefited much from agricultural modernisation either. According to this explanation, there is nothing inherently anti-poor about manufacturing-led growth, but if people are caught in isolated poverty traps which are not connected to the more dynamic parts of the economy, and from which they cannot escape, then they cannot benefit from such growth. A third reason why growth has been less egalitarian in Indonesia since the latter part of the 1980s concerns access to education, and especially what may be termed international quality higher education. People with degrees from the most prestigious domestic universities or from overseas, especially in disciplines such as medicine, accountancy and engineering, have been able to command considerable rents in the booming private sector, and their incomes have increased relative to those in other parts of the urban labour force. In spite of the rapid growth in educational enrolments at all levels in Indonesia since the 1970s, access to international quality higher education, or even good quality secondary education, was still severely limited in the late 1980s and early 1990s, and these rents were not dissipated through increased supply. This explains the findings of several analysts that educational disparities contributed significantly to overall expenditure inequalities in these years. As far as the first explanation is concerned, Ravallion and Datt (1996b: 19) stress that the Indian result, that capital-intensive industrialisation behind extremely high protectionist barriers had negligible benefits for the poor, is hardly surprising. In other economies, such as Taiwan, where the early phase of industrialisation was more oriented to export markets, and based on more labour-intensive production technologies, the sectoral shift from agriculture to industry did not lead to a perceptible increase in income inequalities. In fact they declined between 1953 and 1972 (Kuo 1975). Indonesia after 1987 can be viewed as an intermediate case between India and Taiwan: although exports of labour-intensive manufactures grew rapidly, a substantial part of manufacturing value added was still

21 Poverty and Inequality in the Soeharto Era: An Assessment 91 produced in sectors enjoying high levels of protection and quite capitalintensive production technologies. Thus although there was a significant increase in the manufacturing labour force over these years, and indeed over the entire period from 1971 to 1995, the rate of growth was not as fast as in Taiwan or South Korea (Manning 1998: 93 4). 13 The second explanation, based on the notion of pockets of poverty, also has considerable empirical support. By 1996, 43% of the poor population were outside Java and Bali; over 20% were in Kalimantan, Sulawesi and the eastern islands (East and West Nusa Tenggara, East Timor and Maluku). There seems to be little doubt that relatively backward agriculture, combined with small holding sizes, was one explanation for the high incidence of rural poverty in these regions. The results of an OLS (ordinary least squares) regression analysis, shown in table 11, indicate that a large part of the interprovincial variation in rural poverty in Indonesia in 1993 can be explained by just three variables: average holding size, value added per hectare in smallholder agriculture, and the proportion of total value added in smallholder agriculture accruing from foodcrop agriculture. In those provinces where per hectare agricultural productivity is low, and this is not compensated by larger holding sizes, and where a higher proportion of agricultural value added accrues from the foodcrop sector than from treecrops, the poverty problem is likely to be worse than the national average. By contrast, in the densely settled provinces of Java/Bali, low holding sizes are more than compensated for by high per hectare productivity, so that agricultural value added per farm household is higher and, in addition, off-farm sources of income are more plentiful. Thus rural poverty levels are lower. 14 The policy implication of this is that agricultural and rural development policies which raise per hectare productivity or holding sizes should be given high priority in those parts of the Outer Islands where the incidence of poverty is high. This is not just because agricultural growth in itself helps to alleviate poverty, but also because it fuels the growth of off-farm employment. Indeed, the Indonesian government has in recent years given priority to agricultural development outside Java, especially in Eastern Indonesia. But the problems associated with both extensive and intensive agricultural development outside Java remain formidable. The biophysical environment in Eastern Indonesia is much less favourable to intensive agricultural development, and the seedfertiliser-water technologies that worked well in Java and the well irrigated parts of Sumatra and Sulawesi are not easily transferable to the more arid climate that prevails in many of the eastern islands, nor to the soils of Kalimantan. In addition, the distribution of land is more skewed in many parts of the Outer Islands than in Java, and the benefits of

22 92 Anne Booth TABLE 11 OLS Regression Results: Dependent Variable Rural Poverty, 1993 (headcount measure; all provinces) Explanatory Variables Equations a AHS ( 2.015) ( 3.283) ( 2.105) ( 3.256) PLF ( 0.104) ( ) AHH ( ) (0.1336) VA ( 4.513) ( ) ( ) ( 4.884) GLF (0.5698) (1.0233) FVA (2.139) (2.549) R 2 (adjusted) F-statistic F (6,19) F (5,20) F (3,22) F (2,23) SE of regression a All equations estimated using logs of variables; t ratios given in brackets. AHS: average holding size. PLF: percentage of the rural labour force employed in agriculture. AHH: percentage of agricultural households with more than 70% of household income derived from on-farm activities. VA: value added per hectare in smallholder agriculture. GLF: growth of the agricultural labour force, FVA: percentage of total value added in smallholder agriculture from foodcrops. Sources: Poverty data taken from BPS (1994); data on value added in agriculture from BPS (1997b); other agricultural data from BPS (1995a, 1995b). Labour force data from Sakernas, 1986 and 1994.

23 Poverty and Inequality in the Soeharto Era: An Assessment 93 agricultural development policies are more likely to accrue in the first instance to the non-poor population. 15 The third hypothesis I have suggested to explain the fall in the elasticity of poverty decline with respect to GDP growth relates to access to education. Since the late 1960s, Indonesia has witnessed an enormous expansion of educational opportunities at all levels. This expansion has been driven by greatly increased provision of facilities by both the government and the private sector; in addition, parental demand for education has expanded, although this increase has not been uniform by region or by social class. As in most other Asian countries, public subsidies per student rise sharply by level of education. In the mid 1980s, the cumulative public subsidy per student in higher education was more than four times the subsidy to students in primary education (Tan and Mingat 1992: 80). 16 Several analyses of household survey data have shown that enrolments at the primary level vary little by income decile, but at the secondary and tertiary levels the disparity is very marked (Gertler and Rahman 1994: table 4.10; King 1997: 169; Oey-Gardiner 1997: 152 3). King s analysis shows that government subsidies for primary education are quite strongly pro-poor, because primary-aged children represent twice as large a fraction of the population in the poorest deciles as in the richest, but at the secondary and tertiary levels, especially the latter, the benefits of public subsidies are skewed very heavily towards the richest decile. But public subsidies do not cover the full cost of education even at the primary level. Parental top-up contributions have long been required by government schools, and these fees are a higher proportion of household expenditure in poor households (Gertler and Rahman 1994: 180). In Indonesia, as in most other parts of the developing world, education is viewed by parents as a crucial channel of social mobility, and the level of education is clearly an important determinant of lifetime earnings. In 1996, an employee with university qualifications was earning, on average, more than three times as much as an employee who had only completed primary education (table 12). Thus, as far as an individual is concerned, the higher the level of educational attainment that he or she can reach, the higher the expected earnings stream over a lifetime. But at the national level this simple relationship must be modified. The earnings of any particular type of worker are determined, ultimately, by supply and demand, and if the supply of workers with, for example, post-secondary qualifications increases relative to demand, their earnings can be expected to fall. It is clear from table 12 that the very high dispersion of earnings by educational attainment that was found in Indonesia in the 1970s was in fact compressed during the 1980s, as the supply of workers with at

24 94 Anne Booth TABLE 12 Monthly Employee Remuneration by Educational Level (Rp 000) No schooling 7.5 a 35.4 a 92.1 Not completed primary Completed primary Lower secondary (general) Lower secondary (vocational) Upper secondary (general) Upper secondary (vocational) Diploma I/II Diploma III/Academy 57.6 b b University Average a Average for no schooling and incomplete primary. b Average for academy and university. Source: BPS (1988, 1997c). least primary qualifications increased. In addition, the level of education required to gain entry to particular occupations has increased, while at the same time new jobs have become available, especially in the factory sector, which demand only relatively modest educational qualifications. Given these changes in the labour market, and the fees associated with post-primary education, which are high relative to average per capita consumption expenditure, it appears that, by the late 1980s, many parents decided there was little point in keeping children in school beyond the primary level, especially when paid employment opportunities for teenagers were increasing. Over the Repelita V period ( ), there was an absolute decline in numbers enrolled at both the lower and the upper secondary levels (Booth 1994: table 14). Thus, by the 1990s differences in the level of educational attainment in the labour force, combined with the still limited access to high quality tertiary education, were an important factor contributing to expenditure inequalities. Akita, Lukman and Yamada (1999: 213 4) point out that if

25 Poverty and Inequality in the Soeharto Era: An Assessment 95 there had been no disparities in mean expenditures by educational attainment, total expenditure inequalities would have been much lower. They argue that raising general educational levels in Indonesia should have a significant impact on inequalities over time, other things being equal. But matters may not be as simple as this. Many parents have apparently decided that the costs of keeping children in secondary education (including forgone earnings) outweigh the perceived benefits, and all the evidence suggests that, beyond the primary level, participation in education is very dependent on family income. Although the most recent evidence indicates that the continuation rate from primary to lower secondary education has increased in Indonesia over the 1990s (and has not been much affected by the crisis of ), it is still the case that many millions of children are not completing the nine-year cycle of primary and lower secondary education (Booth 1999c). An important reason for this is the not inconsiderable fees associated with secondary schooling in Indonesia relative to household incomes, and there is little reason to expect that these will fall in coming years. Thus the findings of King and others that government subsidies at the post-primary levels are skewed to the better off are unlikely to alter in the immediate future. Given the evidence on the incidence of educational expenditures, it could be argued that a better outcome for the poor would be achieved if fees were actually increased at the upper secondary and tertiary levels, and public subsidies reduced. 17 The savings could then be used to achieve the goal of universal nine-year schooling. This would entail building new schools in remote areas, and providing better incentives for teachers to work in such areas. In addition, more funds could be directed to improving existing primary and lower secondary facilities, and increasing teacher remuneration. Only when all children in the 7 15 year age groups are in school should government spending on upper secondary and tertiary facilities be increased, preferably through the provision of meanstested assisted places. The problem with this strategy is that many bright children from poorer households will be deprived of access to upper secondary and tertiary education, as the fees would be prohibitive. Of course it could be argued that this is already the case, and that raising fees at these levels simply reduces the subsidy paid to better-off households. But to the extent that enrolments at the upper secondary and tertiary levels stagnate or even decline, the transition to skill-based industrialisation in Indonesia will be retarded. The experience of Thailand over the past decade offers some cautionary lessons for Indonesia. Educational enrolments at the secondary level stagnated in Thailand for much of the 1980s, and by the early 1990s severe skill bottlenecks were emerging in many manufacturing industries. Only when the government embarked on a massive program