Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205277

(1)

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=cbie20

Download by: [Universitas Maritim Raja Ali Haji] Date: 19 January 2016, At: 19:58

Bulletin of Indonesian Economic Studies

ISSN: 0007-4918 (Print) 1472-7234 (Online) Journal homepage: http://www.tandfonline.com/loi/cbie20

The road to pro-poor growth: the Indonesian

experience in regional perspective

C. Peter Timmer

To cite this article: C. Peter Timmer (2004) The road to pro-poor growth: the Indonesian experience in regional perspective, Bulletin of Indonesian Economic Studies, 40:2, 177-207, DOI: 10.1080/0007491042000205277

To link to this article: http://dx.doi.org/10.1080/0007491042000205277

Published online: 19 Oct 2010.

Submit your article to this journal

Article views: 338

View related articles

(2)

THE ROAD TO PRO-POOR GROWTH:

THE INDONESIAN EXPERIENCE IN

REGIONAL PERSPECTIVE

C. Peter Timmer*

Center for Global Development, Washington DC

Indonesia’s long-run ‘pro-poor growth’ record is among the best in Asia. It shows that appropriate policies can free societies from poverty’s worst manifestations in a generation, a crucial message as democracy begins to influence the policy process. This paper places Indonesia’s record in regional perspective, analysing determi-nants of income distribution in Asia and connecting this analysis to Indonesia’s pro-poor growth process and the policy mechanisms that encourage pro-poor growth. Using a data set for eight Asian countries, it examines patterns of change in incomes and distribution across countries and over time. Building on Indonesian experience, the paper presents a pro-poor growth model encompassing three lev-els: improving the ‘capabilities’ of the poor, lowering transactions costs in the econ-omy, especially between rural and urban areas, and increasing demand for goods and services produced by the poor. It finds that rapid pro-poor growth requires simultaneous and balanced interaction between growth and distribution processes.

‘Pro-poor growth’ is the new mantra of the development community. Most donor agencies have active research programs under way to understand the pro-poor process, and the World Bank, with British, French and German bilateral support, is already studying how to operationalise the concept (USAID 2004; World Bank 2004). Definitions vary, but they all revolve around connecting the poor to rapid economic growth so there is a concomitant rapid reduction in poverty. This exploratory essay, commissioned by the Indonesia Project at The Australian National University, places the new interest in pro-poor growth in regional per-spective and attempts to draw historical and policy lessons for Indonesia. The main challenge is to link our relatively robust understanding of the growth process with much more limited understanding of distribution processes. A panel data set of eight Asian countries—selected both for representativeness and for availability of data on income distribution—provides grist for the empirical mill.

POVERTY, GROWTH AND DISTRIBUTION

The sources of economic growth and the distribution of its benefits are among the most important topics that economists seek to understand. From The Wealth of Nations(Smith 1776) to The East Asian Miracle(World Bank 1993), the Asian expe-rience has shaped that understanding in fundamental ways. These range from

(3)

Smith’s concerns about poor economic governance in China as an impediment to growth in the 18th century to the World Bank’s praise for the region’s high sav-ings rates and export orientation as the keys to rapid economic growth in the 20th century. The label ‘miracle’ related in particular to the stability of, or even improvement in, income distribution in these ‘high-performing Asian economies’ during the fastest periods of growth. Rapid, pro-poor growth was invented in Asia, and Indonesia claimed some of the patent rights.

With the onset of the Asian financial crisis in 1997, the sustainability of the East Asian miracle came into question. Indonesia’s success, always on the periphery of the experience in the other East Asian ‘tigers’, has come under special chal-lenge (Booth 2002; Islam 2002). With full recovery of the economy not yet in sight, and escalating political disagreements over income distribution and the fate of the poor, this is an opportune time to bring a regional perspective to the nexus between economic growth and income distribution in Indonesia, and how they affect poverty reduction.

The level of poverty is an outcome, whereas economic growth and change in the distribution of income are processes. In a convenient conceptualisation of the linkages among these three topics, Bourguignon (2004) presents a ‘poverty– growth–inequality’ triangle that illustrates current thinking in the development research community. Figure 1 is adapted from Bourguignon’s original figure, and it highlights the dual routes by which development strategy can reduce poverty and generate ‘pro-poor growth’: through economic growth or through improve-ments in income distribution. The two-way arrow linking these two processes indicates possible causal forces at work in each direction. That is, economic growth might affect income distribution, perhaps widening inequality in the way Kuznets (1955) hypothesised. Also, income distribution might affect economic growth, in a negative direction, as Alesina and Rodrik (1994) and Easterly (2002) demonstrate, or in a positive direction, as Forbes (2000) shows.

If poverty is the outcome measure of interest, both economic growth and income distribution need to be treated simultaneously.1_{For most of the 1990s,} this need was thought to be simplified by the empirical ‘fact’ that income distri-bution seemed to change little over the course of development, so policy atten-tion could concentrate on speeding economic growth, even in settings where poverty reduction was the overriding goal (Dollar and Kraay 2002). Conse-quently, much of the ‘macro’ empirical work in the development profession dur-ing that decade focused on understanddur-ing the determinants of economic growth (Barro 1998; Barro and Sala-i-Martin 2003). That understanding is now reasonably robust, at least at the level of ‘deep’ factors that contribute to long-term increases in real incomes (Maddison 1994; North 1990).

Less is known about country-specific ‘policy’ factors that will speed growth in particular settings (Rodrik 2003). And very little is known about what causes income distribution to change. As evidence has accumulated in the new millen-nium that income distribution can change quite rapidly in a particular country setting, the need to understand the causal factors has also risen.2_{A plausible} hypothesis is that the country-specific determinants of rapid economic growth are linked to the factors that determine the distribution of income. If so, the search for the policy mechanismsthat speed ‘pro-poor growth’ will need to emphasise these connections.

(4)

That is an important goal of this paper. It attempts to understand the determi-nants of income distribution in Asia and to connect this understanding to the process of pro-poor growth in Indonesia.3_{The Asian region in general, and} Indo-nesia in particular, present good examples of contrasting patterns of change in incomes and distribution, both across countries and over time. These contrasts motivate the selection of the eight countries in our panel data set (in addition to the constraints on that selection imposed by data availability).4

REGIONAL PATTERNS OF CHANGE IN INCOMES AND DISTRIBUTION Eight Asian Countries

Pair-wise similarities highlight the sharp contrasts across other country pairs (fig-ure 2); for example, in India and Indonesia, income distribution,as measured by the Gini coefficient, has been remarkably stable for decades.5_{At the same time,} changes in incomes of the rich and poor—especially the growing gapbetween the ends of the income distribution—are challenging political systems to help those left behind as the structural transformation in these societies becomes visible.

In other countries in the region, the Philippines and Malaysia, for example, income distribution has been sharply skewed for a long time, with little apparent political resistance to the inequality. But the economic performance of the two countries has been sharply different, with absolute poverty nearly eliminated in Malaysia following the adoption of pro-Malay policies after communal riots in 1969, while it remains a visible, even growing, problem in the Philippines.

A third category of countries—Thailand and China are key examples—started in the 1960s with quite equitable distributions of income, like those of India and

FIGURE 1 The Poverty–Growth–Distribution Triangle

Development strategy

Aggregate income Distribution and level and growth distributional change

Source: Adapted from Bourguignon (2004).

Absolute poverty and poverty reduction

‘Pro-poor growth’

(5)

FIGURE 2 Changes in Per Capita Incomes and Income Distribution in Eight Countries in Asiaa

100 1,000 10,000

20 30 40 50 60

China and Thailand

China

Thailand

100 1,000 10,000

30 35 40

South Korea and Indonesia

Indonesia South Korea

100 1,000

25 30 35 40

India and Pakistan

India

Pakistan

100 1,000 10,000

40 45 50 55

Philippines and Malaysia

Philippines

Malaysia

a_{The horizontal axis shows the Gini coefficient; the vertical axis shows per capita GDP (log}

scale) for all countries except South Korea and India, for which the log of per capita income is shown; see note 4 for data sources, and appendix 1, table 3, for the period covered for each country. Since economic growth has generally been positive, the movement of points upward on the vertical axis also corresponds to movements over time. Each point repre-sents a year for which data on income distribution are available.

(6)

Indonesia, only to see the growth process induce a sharp skewing of the distribu-tion. Although the political systems (and institutional histories) are quite differ-ent in the two countries, they offer opportunities to see how societies respond to highly visible and rapid changes in the functional distribution of income.

South Korea and Pakistan are the two other countries in the panel of data. They are included for two reasons: the availability of reasonably lengthy time series data on income distribution, and the differences in their economic experience. Averaged over the entire period from 1960 to 1999, South Korea is the fastest growing economy in the region, so it presents an opportunity to study the impact of substantial structural transformation on income distribution, especially the rapidly shrinking relative size of the agricultural sector. Of the countries in this sample, Pakistan started with the lowest per capita income and has not grown rapidly, demonstrating that convergence is not an automatic process.

Figure 2 plots the historical record since the 1960s for real per capita GDP or income (on a log scale) against income distribution (as measured by Gini co-efficients) for the eight countries in the sample. It is clear that not all growth pat-terns are the same: the experience of China and Thailand, for example, stands in sharp contrast to that of India and Pakistan. A rough categorisation of the histor-ical experience in these eight countries is shown in table 1.

The concentration of countries in the lower left ‘triangle’ of the matrix in table 1 is striking. Medium and fast economic growth is associated with low inequality, or with low, rising to high, inequality. Intriguingly, there are no apparent repre-sentatives for the lower right cell, where fast economic growth would exist with

TABLE 1 Countries Categorised by Degree of Income Inequality and Economic Growth, 1960–99a

Economic Income Inequality over Time Growth

Low/Stable Low to High High/Stable

Slow Philippines

Medium India Malaysia

Indonesia Pakistan

Fast South Korea China Thailand

a_{Average annual economic growth from 1960 to 1999 was as follows (percentage per year}

per capita): China, 4.21; India, 2.75; Indonesia, 3.39; South Korea, 5.92; Malaysia, 3.87; Pakistan, 3.38; Philippines, 1.29; Thailand, 4.61. ‘Slow economic growth’ means growth of less than 2.5% per year, ‘medium’ means 2.5–4% per year, and ‘fast’ over 4% per year. Countries with ‘low/stable’ inequality maintained Gini coefficients of less than 35 for the period of observation. ‘High/stable’ countries had Gini coefficients of over 40 for the entire time. The Gini coefficients of the ‘low to high’ countries moved from the 30s to the 40s; see note 4 for data sources.

(7)

high and stable income inequality. The political dynamics that make this an ‘empty box’ would seem to be especially troubling for future leaders in China and Thailand. As noted, Malaysia already reflects this reality, coping with high levels of inequality but achieving only medium growth.

The analysis thus far has used the Gini coefficient as the measure of income dis-tribution, but behavioural economists often argue that Gini coefficients offer lim-ited insights into relevant distributive questions. An alternative measure is the

income gapbetween rich and poor, and changes in the size of this gap. Empirically, the income gap can be measured as the absolute difference between the $PPP (dol-lar purchasing power parity) per capita incomes of the richest and poorest quin-tiles at each period for which income distribution data are available. There is no question that income gaps have more political resonance than Gini coefficients, although relatively little economic analysis has compared the two measures.

Several relevant statistics using this income gap are presented in table 2. In both absolute and relative terms, there is substantial variation in the gaps. To standardise the comparisons to some extent, the growth in the income gap over a 20-year period is calculated for all eight countries, and this varies from just $1,109 in India to $9,812 in Malaysia (in constant 1993 $PPP). An alternative rela-tive measure may have more political impact—the size of the income gap com-pared with the per capita income of the bottom quintile of the income distribu-tion. Here too there is great variance, ranging from a low of 3.39 times in Pakistan (and 3.93 in Indonesia) to 10.99 in Malaysia (and 8.76 in the Philippines).

Table 2 quantifies what might as well be called the ‘iron law of distribution’. Despite their political resonance, especially when they are visible in the conspic-uous consumption of the wealthy, income gaps between the rich and poor are a fact of economic life, and the faster the economy grows, the faster the gaps increase. Even with only eight countries in the sample, this is a powerful statisti-cal reality. A simple regression relating the growth rate of per capita incomes and the (log of the) income gap at the start of the period explains more than 95% of the variation in the (log of the) change in the income gap over 20 years.6

It seems likely that growing gaps, because they are so visible (and inevitable), must be managed politically in a way that defuses tensions between the rich and poor; otherwise the growth process itself is threatened. This political ‘manage-ment’ can take various forms, from autocratic repression in Indonesia to open racial preferences in Malaysia, populist rhetoric in Thailand, and extreme agricul-tural protection in South Korea. Democracy is likely to offer different opportuni-ties, and challenges, from those that face autocratic or repressive regimes. The Indonesian experience since 1999 is particularly revealing.

Perhaps the simplest summary of this comparative perspective on economic growth and income distribution in the eight countries examines the experience with ‘pro-poor growth’ directly. Figure 3 plots growth in the per capita incomes of the ‘poor’ (the bottom quintile in the income distribution) against growth in average per capita incomes for the overall economy during the same period.7_The basic picture is well known (Kraay 2004). The long-term performances of all eight countries are in the northeast quadrant and lie close to the 45 degree line, where there is equal growth in the average per capita income and the per capita income of the poor. Of course, as noted, this ‘good’ performance is entirely consistent with rapidly growing gaps between the incomes of the poor and rich.

(8)

Figure 3 demonstrates two key points about pro-poor growth. First, over long periods there is remarkable stability in the close relationship between overall eco-nomic growth and growth in incomes of the bottom quintile. Second, however, there is tremendous variance in this relationship from one time period to another. In China, for example, economic growth was explosively pro-poor in the initial liberalisation period that opened up the rural economy, from 1980 to 1984. Overall economic growth was a remarkable 8% per year per capita, but the incomes of the bottom quintile, nearly all in rural areas, grew by 14.6% per year! Since 1984, the growth process has led to sharply skewed incomes, as rural areas and the interior have failed to match the rapid commercialisation of the coastal

TABLE 2 Changes in Income Gaps between the ‘Rich’ and the ‘Poor’a (constant 1993 $PPP)

Country Time $PPP Per $PPP Changes Period Capita Income Income Gap in Gap

Start End Start End As Multiple Actual ‘20-year’ of Q1 at End

of Period

China 1980–98 Q1 425 960

1,540 6,680 6.96 5,140 6,325 Q5 1,965 7,640

India 1960–97 Q1 352 871

1,382 4,109 4.72 2,727 1,109 Q5 1,734 4,980

Indonesia 1967–2002 Q1 387 1,609

1,525 6,330 3.93 4,805 1,915 Q5 1,913 7,939

South Korea 1965–93 Q1 542 4,373

3,366 18,674 4.27 15,308 8,079 Q5 3,908 23,046

Malaysia 1970–95 Q1 522 1,954

7,656 21,468 10.99 13,813 9,812 Q5 8,178 23,422

Pakistan 1969–96 Q1 415 917

1,406 3,113 3.39 1,707 1,128 Q5 1,820 9,030

Philippines 1961–97 Q1 435 900

5,416 7,886 8.76 2,470 1,257 Q5 5,851 8,785

Thailand 1962–98 Q1 479 1,997

2,503 13,181 6.60 10,679 3,796 Q5 2,982 15,178

a_{The ‘20-year’ gap figure standardises the change in the income gap over a 20-year period,}

assuming the rate of change in the gap is the same per year as for the entire period shown in the table. Q1 and Q5 refer to the per capita incomes of the bottom and top quintiles (Q), respectively. $PPP = dollar purchasing power parity.

(9)

and urban areas. Average per capita incomes still increased by a remarkable 6% per year, but the incomes of the bottom quintile grew only 1.9% per year between 1984 and 1998. From what must have been the most pro-poor growth episode in history, China has reverted to one of the least pro-poor patterns in Asia.8

Indonesia

In some ways, the Indonesian example shown in figure 3 is even more dramatic. Over the entire period, from 1967 to 2002, Indonesia marked up a creditable 4.15% increase in average annual per capita income, with the bottom quintile showing exactly that growth (in percentage terms). But this long-term perspec-tive masks two fundamentally different episodes. From 1967 to 1996, average per capita incomes increased 5.1%, and the incomes of the bottom quintile also increased by 5.1%. From 1996 to 2002, however, average per capita incomes

droppedby 0.4% per year, while the incomes of the bottom quintile fell 0.3% per year, the result of a 13% drop in GDP in 1998, a flat economy in 1999, and slow growth since. This reversal of fortune is one of the most dramatic in recent his-tory. Indeed, as figure 4 shows, there is considerable variance in the ‘pro-poor-ness’ of growth in Indonesia when each available time period is plotted sepa-rately.

F F

-5 0 5 10 15

Annual increase in average per capita income (%)

Annual increase in bottom quintile per capita income (%)

FIGURE 3 Income Growth for the Bottom Quintile Plotted against Growth of Average Per Capita Incomes, for Eight Countries

(constant 1993 $PPP)

Indonesia 1996–2002

India 1960–97 Thailand

1962–98

China 1984–98 China 1980–98 South Korea 1965–93

Indonesia 1967–96 Malaysia 1970–95 Indonesia 1967–2002 Pakistan 1969–96

China 1980–84

Philippines 1961–97

Sources: See note 4.

(10)

An obvious question arises from examination of the growth episodes illus-trated in figure 4. Are the households in the bottom quintile at the start of the period the same households as at the end? If so, poverty is largely structural and reduction will rely primarily on sustained economic growth, plus regionally tar-geted programs (where poverty incidence is especially high and chronic). Alternatively, is there substantial ‘churning’ of households, with a large fraction of poor households moving out of poverty in each period, and others becoming poor?

The question is hard to answer in the Indonesian case because the National Socio-Economic Surveys (Susenas) do not re-sample the same households in each period. However, the fact that a large proportion of Indonesia’s households are clustered near the poverty line, both above and below, suggests that simple meas-urement errors and random processes could lead to substantial churning. The only available empirical data surveying the same set of households at different times are from the Indonesian Family Life Surveys (IFLS), which have been con-ducted in 1993, 1997 and 2000 (with a 25% subsample taken in 1998). These are true panel data and, for the 1997 to 2000 period, show that roughly half the

house-F F F

F F

-5 0 5 10

Annual increase in average per capita income (%)

Annual increase in

bottom quintile per capita income (%)

FIGURE 4 Income Growth for the Bottom Quintile Plotted against Growth of Average Per Capita Incomes, for Various Time Periods in Indonesiaa

a_{Data are from various rounds of the Susenas, Indonesia’s National Socio-Economic}

Survey, conducted by the Central Bureau of Statistics, which provides detailed expendi-ture data on more than 60,000 households every three years. Less detailed data from a ‘core’ questionnaire are available annually for about 200,000 households. Because of differ-ences in approach, it has been difficult to match up results from the two questionnaires.

1996–99

1981–84

1999–2002 1984–87

1990–93 1993–96

1967–76 1976–81 1987–90

(11)

holds that were poor in 1997 had moved out of poverty by 2000, and roughly half the poor in 2000 were new entrants. Strauss et al. (2004), the authors of the major study reporting these results, suggest that 40–50% of this movement is due to measurement errors.

AN ASIAN PERSPECTIVE ON INCOME DISTRIBUTION

In view of the relatively robust knowledge about the determinants of long-run economic growth (Barro and Sala-i-Martin 2003; Rodrik 2003), the more pressing need is to understand the determinants of income distribution. The two together, as illustrated in figure 1, explain the level of poverty and its rate of reduction. An initial effort at this understanding is attempted here for the eight countries in the panel used so far. Income distribution is measured by the Gini coefficient (despite the problems noted earlier), and the dependent variable to be explained in the statistical analysis is the logarithm of the Gini (Loggini). Appendix 1 describes all the variables in more detail and provides their means and standard deviations, by country as well as for the entire sample.

The basic data on income distribution are from household expenditure surveys compiled by Deininger and Squire (1996, with updates). These data have been the source of a veritable avalanche of empirical work on the impact of income dis-tribution on economic growth, and vice versa (Bourguignon 2004). However, attempts to understand the forces driving differences in income distribution itself, across countries and over time, have been much less successful (Kraay 2004).

The approach here builds on a long literature, beginning with Lewis (1954) and Kuznets (1955), which explores the impact on income distribution of the struc-tural transformation that occurs during economic development. As Kuznets noted, building on the Lewis model, there are strong theoretical reasons for expecting the long-run process of industrialisation and urbanisation, and the accompanying decline in agriculture’s share of both economic output and the labour force, to widen income inequality.

Empirically, however, there have also been a number of countervailing forces, with the result that some studies confirm the Kuznets hypothesis while others reject it. Thus, the relevance of the ‘Kuznets curve’ is in limbo. The simple intu-itive model used here illustrates why this is so, and motivates the selection of variables used in the regression analysis described in this section. Figure 5 pres-ents this simple structural model of income inequality.

Consider a typical poor economy divided into agricultural and non-agricul-tural sectors. For convenience, assume that all workers in the agriculnon-agricul-tural sector have lower incomes than the poorest workers in the non-agricultural sector. Then a standard Gini diagram can be constructed, as in figure 5, with point C at the dividing line between the two sectors. If all workers in each sector have the same income (an important caveatto be relaxed shortly), the Gini coefficient for this society is the area of triangle ABC/2.

This simple model suggests that several elements of economic structure are particularly relevant for explaining income distribution: (1) a ‘synthetic’ Gini coefficient constructed from the share of the agricultural economy in overall eco-nomic activity relative to its share in the labour force, as illustrated by area ABC in figure 5, and which in the regression analysis below is termed (in logarithmic

(12)

form) Logsect; (2) relative labour productivity—usually reflected in sectoral wages—in agriculture and the rest of the economy (Wagegap); (3) the role of non-labour income in these economies, especially income from property ownership (Saving); and (4) the role of the informal sector, especially services—mainly pro-ducing non-tradables in both the rural and urban economies (Gdpserv).

Within each sector separately, a number of variables might affect the distribu-tion of incomes (and expenditures). Poverty and hunger are closely linked, with many poverty lines defined in terms of food intake. If income is held constant, food policy variables are likely to play a significant role in explaining differences in income distribution (because of the linkages illustrated in the ‘Bourguignon tri-angle’ shown in figure 1). In particular, price policy for staple foods, especially rice, has an immediate impact on both food consumption and the profitability of an important rural activity, staple food production. Thus, a simple price variable measuring the marketing margin for rice from farm to retail is likely to have explanatory power (Ricertof), although data for this variable restrict the sample size. As noted, food prices were important in explaining the skewing of urban incomes in China.

FIGURE 5 A Structural Model of Income Inequality

Share in agriculture

(%) 100

A D

E B Cumulative share of total population (%)

0 100

Cumulative shar

e of total income (%)

Shar

e of

income to

agricultur

e (%)

Share in non-agriculture

(%)

Shar

e of

income to

non-agricultur

e (%)

(13)

Food prices more generally are likely to influence household caloric intake in a systematic manner, with poorer households affected more significantly than rich households. Deviations from both the Engel curve (calories against income) and the Bennett curve (starchy staples against income) might be closely linked to income distribution because they are largely explained by differential responses to relative prices (Timmer 1981). Thus residuals from a simple regression relating the logarithm of household incomes to food energy (kilocalorie) intake (Kcalresid) should help explain income distribution (although the causation probably runs in the other direction). Finally, again controlling for income, the quality of the diet might also be an indicator of significant differences in income distribution across countries and within countries across time. Dietary quality is measured here by the starchy staple ratio—the proportion of food energy coming from starchy foods such as cereals and tubers (Ssrresid). Conceptually, these are variables that affect the ‘depth’ of points D and E in figure 5. That is, they help explain the distribution of income within sectors.9

It would be desirable as well to test the impact of differences in educational and health status, for males and females, in urban and rural areas (especially because public investments in these sectors seem to have played an important role in pro-poor growth in Indonesia). Unfortunately, these data are difficult to obtain on the same basis as the other data used in this exercise, without access to the actual household records in the expenditure surveys used to generate the income distribution data. However, these variables are likely to be quite impor-tant, especially over long periods of time, in explaining trends in income distri-bution, and collecting data on them is an ongoing research activity.

Understanding changes in income distribution, which have been quite sub-stantial for several of these countries over the past four decades, requires an understanding of the evolving roles of the structural and policy components of each country’s economy, as illustrated in figure 5. Only rough proxies for each of these structural and policy components are available for all of the countries in the sample; the goal is merely to illustrate that changes in income distribution are not the empirical ‘black box’ commonly thought in the profession. The approach here is simple: it includes the savings rate to proxy for non-labour income (i.e. the total return to capital), the share of services in the economy to measure the informal sector, and the stage in the agricultural transformation to capture the structural variable. However, even these crude variables produce significant contributions to the predicted Gini coefficient for each country over time, as can be seen in table 3.

The columns for equations 1, 2 and 3 in the table provide the key results. In equation 1, all of the structural and food policy variables have the expected sign and are significant, confirming the hypothesis that income distribution does depend in predictable ways on economically plausible variables. Equation 2 tests, roughly, for the ‘Kuznets effect’, by inserting the rate of economic growth into the equation (Gdpgrow). Several other variables drop out, but the main results remain and a positive impact on the Gini coefficient is felt from more rapid growth, although the statistical significance just misses the 5% level. Thus there is a hint that rapid growth increases inequality for the whole sample, a result con-sistent with recent findings summarised by Bourguignon (2004).

(14)

TABLE 3 Explaining the (Log of the) Gini Coefficient, and GDP Growth, across Time and Countries (Indonesia, Malaysia, the Philippines,

Thailand, India, Pakistan, South Korea and China)

Independent Equation Number and Dependent Variable Variablea

1 2 3 4 5 5a 6

Loggini Loggini Loggini Gdpgrow Gdpgrow Gdpgrow Gdpgrow

Constant 1.628 1.635 0.8358 –0.1172 –0.2539 –0.1637 –0.3184 (9.2) (9.2) (2.2) (–1.01) (–1.54) (–1.01) (–2.50) Logsect 0.4322 0.4563 0.4948 0.0149 0.0564 –0.0149 –

(10.1) (10.8) (5.9) (0.58) (2.08) (–0.54) Wagegap 0.0114 0.0131 –0.0360 – – – –

(2.4) (2.7) (7.0)

Gdpserv 0.0020 – 0.0107 0.0025 0.0009 0.0008 – (2.0) (5.8) (3.84) (1.01) (1.06)

Saving 0.0013 – 0.0095 0.0020 0.0014 0.0020 0.0013 (2.0) (8.3) (3.79) (2.01) (3.06) (3.17) Ricertof 0.0432 0.0364 0.1945 – – – –

(2.8) (2.4) (5.5)

Ssrresid –0.4549 –0.7455 0.5258 0.1035 0.1279 0.2759 0.3113 (2.7) (5.4) (3.0) (3.13) (1.05) (2.33) (3.81) Kcalresid –0.00007 –0.00005 –0.00037 0.00002 –0.00004 –0.00003 –0.00003

(2.6) (1.8) (8.2) (1.78) (–1.88) (–1.41) (–2.01) Loggdp – – – –0.0233 0.00006 –0.0201 –

(–1.90) (0.00) (–1.50) Gdpgrow – 0.1949 – – – – –

(1.9)

Logtheil – – – –0.0129 0.0032 – – (–3.26) (0.35)

Loggini – – – 0.0244 – 0.0931 0.0898 (0.98) (2.21) (2.53) Indonesia – – –0.2689 – – – –

(8.1)

Fixed effects? Yb _Y _N _N _Y _Y _Y

R squared 0.611 0.220

Rho 0.945 0.949 0.269 0.394 0.465

a_t_{-statistics are shown in parentheses. For variable means and definitions, see appendix 1. In order for} each country to have roughly equal weight in the panel regression analysis, Loggini is constructed as linearised annual data from the actual Gini observations in the World Bank data set (see note 4). Regressions run with the actual observations themselves obviously have fewer observations and sev-eral statistical results are less robust, but the pattern of results is the same.

b_{When this equation is run with separate country intercepts to extract their values (a specification} equivalent to fixed effects), the country coefficients are as follows (t-statistics in parentheses): China, –0.1209 (3.7); India, –0.1812 (4.7); Indonesia, –0.1058 (3.0); South Korea, 0.1113 (2.6); Pakistan, –0.2177 (4.7); Malaysia, 0.2734 (5.8); the Philippines, 0.1962 (5.7). The intercept for Thailand is part of the over-all intercept.

(15)

Equation 3 replaces the fixed effects specification with a single dummy vari-able to test whether Indonesia is an outlier in this panel of eight countries. Two variables reverse sign when the Indonesia country dummy alone is included— Wagegap and Ssrresid—but these are the least robust variables in other specifica-tions as well, so no deep meaning should be read into the change. Still, the Indonesia country dummy is significantly negative (as it was in the fixed effect specification in equation 4), indicating that income distribution is more equal in Indonesia than in the other countries in the sample, even when controlling for the structural and food policy variables. The country discussion and model of pro-poor growth below attempt to explain why this might be so.10

The ‘Bourguignon triangle’ shown in figure 1 stressed the importance of the connections between economic growth and income distribution in understanding how pro-poor growth leads to reductions in poverty, the ultimate objective of this paper. Equations 4–6 are included to illuminate these connections. In these equa-tions, the dependent variable is the annual growth rate in per capita income (Gdpgrow) and the independent variables are the same as those used to explain income distribution (Loggini). By comparing the coefficients for the different dependent variables, it is possible to identify (at least roughly) where there are complementarities (as with education investments that might help both growth and income distribution) and trade-offs (where a variable might speed growth but worsen income distribution) for pro-poor growth.

The results are quite interesting. Few of the variables that are important for explaining income distribution are significant for explaining growth; this is one possible reason for the empirical impasse over the Kuznets curve referred to above. The exception, however, is important. The savings rate widens income dis-tribution while it speeds economic growth, thus presenting a direct trade-off for pro-poor growth. The role of the starchy staple and calorie residuals in growth (as well as income distribution) is somewhat surprising. Explaining these roles prob-ably requires more of a country focus.

Perhaps the most controversial result in equations 5a and 6, which use fixed effects, is the positive coefficient for Loggini when it is included to explain the rate of economic growth. Consistent with Forbes (2000), who also used a fixed effects model, wider income inequalities seem to be associated with faster eco-nomic growth, controlling for the other variables in the model. Again, this would imply a trade-off for proponents of pro-poor growth, but policies to deal with such trade-offs are likely to be highly country specific, as the Indonesia model developed below indicates.

Finally, a measure of income inequality based on dispersion of wages in the manufacturing sector has been proposed by Galbraith and Kum (2003) as supe-rior to Deininger–Squire data, on the basis of both annual coverage and greater reliability. This measure (Logtheil) is included in equations 4 and 5, where it clearly serves as a proxy for a fixed effects model (i.e. Logtheil serves effectively as a separate country-specific variable and loses significance when the fixed effects model is used, unlike Loggini). Appendix 2 discusses the Galbraith–Kum approach in more detail, because it is such a direct challenge to the Deininger– Squire data.

To summarise the regional story, all the structural and policy variables for which there are reasonable proxies in the empirical record are significant and

(16)

have the expected sign, at least in the most basic specifications, as exemplified by equation 1 in table 3. Maintaining greater parity between agricultural and non-agricultural incomes is good for income distribution (and has no impact on eco-nomic growth). Indeed, this is perhaps the most robust statistical result reported.11 _{But a higher savings rate, implying greater total returns to capital,} worsens income distribution, as does a higher share of services in GDP. If the services sector, at the margin, is made up of low-productivity workers fleeing rural poverty, this result makes sense.

The food and agricultural variables that primarily reflect influence on con-sumer behaviour also have surprising power. A wider marketing margin for rice between the farm and retail levels widens income distribution, and it is no won-der that many countries attempt to squeeze the margin through public interven-tions on behalf of both producers and consumers (Timmer, Falcon and Pearson 1983). But this can have devastating consequences for the budget and/or the development of a viable private marketing sector, as China especially learned in the early 1980s, when it devoted 30% of its national budget to subsidising the gap between producer and consumer prices for staple food grains. It is also intrigu-ing to see the role that dietary quantity and quality play in income distribution (and the ways they are affected by it), but further analysis is needed, probably on a country by country basis, to tease out serious policy implications. This work is under way in Indonesia (Molyneaux 2003).

Where does Indonesia fit? At one level, as figure 3 showed, Indonesia can be said to fit at both extremes and in the middle of the Asian experience. Its long-run record on pro-poor growth is not quite as good as that of Malaysia, Thailand or China, but it is better than that of the Philippines, Pakistan and India. When the Soeharto era is split out from the years of the financial crisis, however, then Indonesia’s record is both the worst (1996–2002) and among the best (1967–96). This extraordinary variance begs for explanation. With the regional record for perspective, it will be very instructive to understand Indonesia’s experience in more detail.12

A MODEL OF PRO-POOR GROWTH IN INDONESIA

This section attempts to link the new empirical understanding of the determi-nants of income distribution in Asia, presented in the previous section, with the more developed literature on determinants of economic growth, to develop a model of pro-poor growth. The specifics of the model build on Indonesian expe-rience and thus do not address what in the growth literature are characterised as ‘deep’ (as opposed to ‘policy’) variables. The ‘deep’ variables, such as institu-tions, stability and openness, represent long-run structural issues facing most countries. The ‘policy’ variables take as given these structural factors and focus on changes that affect economic growth and distribution within the time horizon of a sitting government. Obviously, the long run is made up of a sequence of short runs, so the two perspectives are linked.

From this policy perspective, there have been three major sources of economic growth in Indonesia since the mid 1960s: economic recovery and rehabilitation of the existing capital stock and infrastructure (to the mid 1970s); rapid growth in agricultural productivity because of new technology and massive new

(17)

ments in rural infrastructure (mid 1970s to mid 1980s); and eventually the emer-gence of a vibrant manufacturing sector, stimulated by foreign direct investment and exports (mid 1980s to mid 1990s). A rough chronology would have recovery and rehabilitation ending by the mid 1970s, and the dominant contribution of growth in agricultural productivity ending by the mid 1980s, with manufacturing taking the lead role by the late 1980s.

The resulting growth was strongly pro-poor in each episode, because income distribution did not deteriorate to any significant degree from 1967 to 1993, and may even have improved somewhat in rural areas. The regressions in table 3 help explain why. Policy management emphasised maximising productivity for the two scarcest factors of production—land and capital. High productivity for land meant yield-enhancing technologies that were labour intensive; the ‘green revo-lution’ varieties of rice, for example, responded dramatically to greater fertiliser applications, good water control and careful agronomic management. The gap between rural and urban productivity did not widen too rapidly for labour migration to keep wages closely linked.

Working capital in the trade and marketing sector was very expensive, and a small-scale, labour-intensive marketing structure dominated until the rapid emergence of supermarkets after 2001. This part of the service economy was as productive as other opportunities for unskilled labour and was thus not a part of widening income disparities. And despite efforts by the government to build capital-intensive state-owned industries when the budget was flush with oil rev-enues, private sector manufacturing was always highly labour intensive. So the increasing savings rate also helped absorb labour.

The secret to pro-poor growth, not surprisingly, has been rapid growth at the macro level that was simultaneously labour intensive at the micro level. The importance of agriculture and the rural economy in this process is obvious (Timmer 2002).13 _{Where land was the scarce resource relative to labour, as on} Java, labour-intensive cropping patterns using high-yield technologies were poverty reducing. Where land was abundant relative to labour, as in many parts of the Outer Islands, plantation crops yielded better incomes than in labour-sur-plus regions for both labourers and smallholders. If labour-intensive manufactur-ing had not taken off rapidly in the mid 1980s, agriculture on the Outer Islands would probably have contributed more to pro-poor growth by offering migration opportunities from Java. As it turned out, however, more opportunities beckoned on Java than off, and net migration to the ‘overcrowded’ island turned positive in the 1990s.

The Components of the Model

Even a casual reading of the memoirs of the economic technocrats responsible for designing economic policy in the early years of the Soeharto regime reveals their emphasis on economic growth as the only way to reduce poverty (Thee 2003). Their assessment of the economic situation in late 1966 suggested that nearly the entire population was poor by absolute standards—earning half the per capita income of India at the same time, for example. In the short term, as when Park Chung Hee came to power in South Korea in 1961, there was simply no choice but to stress economic growth over poverty reduction—there was nothing to ‘redistribute’.

(18)

In the longer run, of course, strategic choices were available. With the disas-trous experience of ‘politics in command’ of the economy under Sukarno vividly in everyone’s mind, the early strategy focused on stabilising macroeconomic pol-icy through a balanced budget and a realistic exchange rate, stabilising the food economy by controlling rice prices (using market-compatible interventions), and rehabilitating infrastructure using the proceeds from foreign aid (and, later, oil revenues). Trade and investment policy was opened, as in Malaysia and Thailand, with a dramatic liberalisation of the capital account. The external envi-ronment was not particularly hospitable, as global inflation was very high and the United States was deeply engaged in Vietnam. But this permitted the Indonesian government to focus on developing its own strategy for economic growth. In particular, since neither Soeharto nor the technocrats had any experi-ence in policy making, the mechanisms of economic governance needed to be designed almost from scratch to form a workable set of relationships and create a division of responsibilities.

Paradoxical as it seems now, it was Soeharto himself who stressed to the tech-nocrats the importance of connecting the poor to economic growth. Political sci-entists continue to debate why he was so concerned about this connection, but the reality is that much of the emphasis on improving the welfare of the rural popu-lation was initiated by the president (Rock 2002, 2003; MacIntyre 2001). He knew that most of the poor lived in rural areas and that they could be helped through agricultural development, schools, clinics and family planning centres, and rural infrastructure investments. Out of this concern the technocrats evolved a devel-opment strategy that consciouslytried to merge the ingredients of rapid economic growth with powerful connections to the livelihoods of the rural poor.

In retrospect, the pro-poor strategy encompassed three basic levels: improving the ‘capabilities’ of the poor, lowering transactions costs in the economy, espe-cially between rural and urban areas, and increasing demand for goods and serv-ices produced by the poor (or for labour directly). In many respects, this closely mirrored the ‘East Asian’ model practised elsewhere in the region. The main rela-tionships are illustrated in figure 6. Within the framework of Bourguignon’s poverty–growth–inequality triangle, the technocrats were addressing the growth and inequality processes simultaneously, but always with a primary focus on keeping rapid growth under way. Public investments in ‘capabilities’ and con-stant concern for labour intensity connected the poor to that growth. Thus the resulting impacts on poverty from growth and from unchanged inequality were solidly pro-poor.

Macroeconomic policy was directly in the hands of the technocrats, and this was always managed in such a way as to maximise the overall rate of economic growth, subject to controlling inflation through fiscal and monetary discipline. The exchange rate was an instrument of policy, not an objective except in the very short run, and it was managed to maintain profitability of tradable goods produc-tion, especially in agriculture. Such a growth-oriented macro policy should call forth investments from the private sector that become the actual engine of eco-nomic growth, but the institutional foundations for rapid expansion of the pri-vate sector in Indonesia were not in place until the reforms of the 1980s, so a more active public role was necessary to stimulate appropriate investments. Apart from the mid 1970s during the peak of the oil boom, the public role was not

(19)

FIGURE 6 The Road to Pro-poor Growth

Political vision and commitment

Rapid, pro-poor growth Capabilities

Transactions costs

Demand Education and health

Agricultural technology

Technology Infrastructure

Cost of food staples

Exports, exchange rate and trade policy

Rural non-tradables

Corruption

Macroeconomy and rapid growth Empowerment

Regulations

Income distribution

(20)

investment in state enterprises, but rather investment in the soft and hard sup-porting infrastructure for private sector enterprises.

These infrastructure investments lowered the costs of market connections that generated jobs and raised the productivity of the poor. Indeed, public sector investments and regulatory improvements to lower transactions costs as an approach to market development are arguably the crucial link between growth-oriented macroeconomic policy and widespread participation by poor house-holds in the market economy.14 _{In Indonesia, as in China in more recent times,} these investments were in roads, communications networks, market infrastruc-ture and ports, and irrigation and water systems. Many of them were built as labour-intensive public works, making millions of jobs available to unskilled labour willing to work at local market wages.

Lower transactions costs mean more market opportunities and faster economic growth. They also mean easier access for the poor to markets and better connec-tions to economic growth; the positive coefficient on the rice marketing margin variable in equation 1, table 3, shows that these costs are also important for income distribution. For access to translate into participation, however, the capac-ity of poor households to enter the market economy needs to be enhanced. Investments in human capital—education, public health clinics and family plan-ning centres—improve the ‘capabilities’ of the poor to connect to rapid economic growth. Of course, other barriers can also impede participation of rural house-holds in market-led growth, hence the crucial importance of improved local gov-ernance to lower transactions costs with respect, for example, to property rights, market access, permits and education.

The three-tiered strategy for pro-poor growth shown in figure 6 links sound macroeconomic policy to market decisions that are facilitated by progressively lower transactions costs, which in turn are linked to household decisions about labour supply, agricultural production, and investment in the non-tradable econ-omy. As had been the experience elsewhere in the region, the rate of poverty reduction driven by this strategy depended on the array of assets controlled by the poor—their labour, human capital, social capital and other forms of capital, including access to credit.

The most important way Indonesia attempted to influence returns to the port-folio of assets held by the poor was through human development expenditure, especially on education and public health. At least during the parts of the Soe-harto era when the pro-poor strategy was most effectively implemented, efforts to influence wage rates directly were generally avoided—another pillar of the East Asian rapid growth pattern—and organised labour was actively suppressed. The technocrats closely monitored Indonesia’s wages relative to those of competi-tors such as Malaysia and Thailand in the early years, and China, Vietnam and India in the later years. The concern was always for job creation and the prof-itability of labour-intensive activities, especially for export. The wage gap between formal and informal labour markets was not allowed to get too large, and this helped stabilise income distribution, while also attracting investments to labour-intensive manufacturing.

An active price policy for rice also attempted to stabilise the returns to small-holder producers. At least until the mid 1990s, there was no long-run effort to raise these returns above trends in the world market, converted at the open-market

(21)

exchange rate. This price stabilisation policy, as implemented by the market-oriented operations of the logistics agency, Bulog, had a highly positive impact on farm productivity, consumer welfare and national food security.15_{Both farmers} and consumers gain if the average prices they receive and pay are stabilised at their long-run mean (or around their long-run trend). Reduced variance for the same mean improves the performance of a diversified asset portfolio, and reduced ‘noise’ from price signals raises investment efficiency by improving sig-nal extraction. It has been argued that until the 1990s, the costs of this price pol-icy were modest (Pearson 1990).

How well does this pro-poor strategy work? The answer depends on the effi-ciency of transmission mechanisms that connect the poor, through factor and product markets, to the overall growth process. The efficiency of these mecha-nisms depends on demand and supply pressures in the markets for unskilled labour and on how well integrated these markets are across skill classes and regions. Initial conditions for income and asset inequalities seem to play an important role in the connection process, possibly because of failures in credit markets that make it hard for the poor to invest in their own human capital. Thus public investments in education and rural public health are likely to be necessary for the transmission mechanisms to work effectively for the poor (Gugerty and Timmer 1999; Bourguignon 2004).

Further, migration, job mobility and flexibility in the face of shocks all help maintain upward mobility during the growth process, and cushion the irre-versibility, seen in so many countries, of a sudden fall into poverty. In Indonesia a resilient rural economy turned out to be crucial as a social safety net during the crisis, when there was a sharp reversal of the structural transformation, and agri-culture absorbed millions of displaced industrial and service workers (a process seen in Thailand as well). Again, keeping a reasonable balance between growth in rural and urban economies during the structural transformation has a positive impact on income distribution (as seen in table 3) and contributes to pro-poor growth if the resulting balance does not sacrifice economic growth to a significant degree. The evidence from equations 4 and 5 in table 3 suggests that it does not.

The pro-poor growth strategy in Indonesia emphasised rapid increases in the demand for unskilled labour (Manning 1998). Macroeconomic policy stressed (a) stability (to lower risks to investors); (b) a competitive exchange rate (to keep tradable goods production profitable); and (c) a monetary and fiscal policy that did not subsidise the use of capital. This was the ‘umbrella’ over the market econ-omy. Markets were the arena for participation by the poor in economic activities that improved their productivity and household incomes. If the household econ-omy is the ‘foundation’ of the pro-poor strategy, with public investments used to improve human capital and capabilities, the market economy is the bridge to the macro policy. For the most part, this market economy is accessible to the poor only if the transactions costs of engagement are manageable and the risks are low.

Here too public investments are the key to making the process pro-poor. There are no doubt important trade-offs in how the public sector manages the array of invest-ments needed, from human capital in rural areas to infrastructure that links rural households to market opportunities. Eventually all of these investments need to be made for pro-poor growth to succeed, and there is only limited evidence on sequencing when resources are critically short. Indonesia was able to make these

(22)

investments faster because of large oil revenues in the 1970s and extensive for-eign assistance from the late 1960s on. But most countries in similar circum-stances squandered the largesse. Why Soeharto carried out the pro-poor strategy as aggressively as he did remains a mystery of political economy. And how the country will find the resources to continue these pro-poor investments in the face of massive debt and dwindling petroleum exports is an even murkier question.

A final question facing Indonesia is how to reconcile sharply conflicting meas-ures of poverty. The definition of a poor household is often based on the real wages household members earn, as the poor usually have little to sell but their own labour. But the link between real wages for unskilled labour and the extent of poverty is more complicated than expected, and this raises quite basic issues about the definition and measurement of poverty. Especially during the Asian financial crisis, from which Indonesia has not yet recovered fully in macro-economic terms, the Susenas household expenditure surveys (the basis for the data used in the empirical analysis in this paper) are telling a different story from that told by data on real wages. According to Papanek (2004), in 2003 real wages for agricultural workers were still 20% below their previous peak in 1997, but the headcount index of poverty had returned to its previous low.

Part of the problem is the difficulty of choosing a reliable deflator for nominal wages during a period of rapid relative price changes. Part of the problem is the changing structure of employment between formal and informal, and the possi-bility of a short-run break in the strong integration of labour markets seen histor-ically. And part of the problem may be a growing importance of self-employment and remittances in stabilising household expenditures. This is a major research topic for the future, because reconciling these two views on the current status of the economy and its impact on the poor will be crucial to designing pro-poor growth strategies that work as well in the future as they did during the three decades of rapid economic growth under the Soeharto regime.

SUMMING UP AND LOOKING FORWARD

This paper has explored Indonesia’s growth experience and prospects, and how they are illuminated by the recent experience of her regional neighbours. Economic growth in Indonesia has always benefited the poor. There are episodes when income inequality increased and episodes when it decreased, so Indonesia has experienced both ‘weak’ and ‘strong’ pro-poor growth. But over the long run Indonesia’s record on ‘pro-poor growth’ is nearly as good as the best in Asia, and is better than most. As the analytical and empirical discussions in this paper have stressed, rapid pro-poor growth requires simultaneous and balanced interaction between growth and distribution processes.

The balanced interaction between growth and distribution—the two sides of figure 1—that generated rapid pro-poor growth in Indonesia was based on a con-scious strategy of integrating the macro economy with the household economy. Transactions costs were lowered in both factor and product markets, principally through investment in infrastructure and human capital. Luck also played a role, with powerful new agricultural technology becoming available in the late 1960s, just as the country was putting in place the economic strategy to make it effective. In the 1980s, foreign direct investment arrived, mainly from Northeast Asia, just

(23)

as Indonesia needed to restructure its manufacturing sector to be more labour intensive and export oriented.

After more than three decades of autocratic rule under President Soeharto, Indonesia successfully elected a democratic legislature in 1999, which in turn selected a new president. With representative democracy came a new political economy of economic policy, especially from populist voices ostensibly speaking on behalf of the poor. An important test is under way to determine if Indonesia’s pro-poor growth experience under a highly centralised and politically dominant regime has put down sustainable, even irreversible, roots, or whether the very foundations of the strategy will come undone under political challenge. Put bluntly, will what worked then work now?

Economic history has many examples of reversals of fortune, from the collapse of early civilisations to more modern experiences in Myanmar, Argentina and Zimbabwe. In the short run, politics is always the master of economics, but in the long run good economic governance is essential for growth. Indonesia has expe-rienced its own reversals of fortune over the centuries, but the current challenge is unprecedented in the memories of most voters. It is already clear that the tran-sition from the autocratic rule of Soeharto, with economic policy designed and administered by an insulated group of skilled technocrats, to a politically respon-sive system with few public institutions in place to protect economic policy from polemicists, is going to be difficult for botheconomic growth and its connection to the poor. It is entirely possible that Indonesia will follow a path that looks more like Africa than East Asia.16

But that path is not inevitable. The Asian experience with pro-poor growth in general, and the Indonesian experience in particular, provide hope that desper-ately poor societies can escape from the worst manifestations of their poverty in a generation, provided appropriate policies are followed. This is an important message for the Indonesia of the future, unsure as it is over what path to follow during its democratic transition. Although built on Indonesian experience, the three-tiered strategy of growth-oriented macroeconomic policy, linked through progressively lower transactions costs to product and factor markets, which in turn are linked to poor households whose capabilities are being increased by public (and private) investments in human capital, is a general model accessible to all countries, including the future Indonesia.

NOTES

* Inevitably, many intellectual debts are incurred in writing a paper such as this. I can-not hope to thank all my colleagues in Indonesia and elsewhere who over the years have helped me understand the processes of economic growth, income distribution and poverty reduction. The intellectual framework used here goes back at least to Food Policy Analysis (1983), and my co-authors for that volume, Wally Falcon and Scott Pearson, have remained close colleagues and sounding boards over the years. The Food Policy Support Activity (FPSA), funded by USAID in Indonesia, has provided continuing support and colleagues for the development of my ideas. My daughter and co-author on a related manuscript, Ashley Timmer, has provided wonderful technical assistance on the data analysis for this paper. Finally, several reviewers for BIESoffered helpful comments. Many thanks to all, and blame to none for continuing problems and errors.

(24)

1 Bourguignon (2004), drawing on Datt and Ravallion (1992), explains the simple mathe-matics of this identity. ‘A change in the distribution of income can be decomposed into two effects. First, there is the effect of a proportional change in all incomes that leaves the distribution of relative income unchanged, i.e. a growtheffect. Second, there is the effect of a change in the distribution of relative incomes which, by definition, is inde-pendent of the mean, i.e. a distributionaleffect … A change in poverty can then be shown to be a function of growth, distribution and the change in distribution’ (pp. 3–4). 2 Three recent volumes from the World Institute for Development Economics Research

of the United Nations University (UNU/WIDER) address these issues directly (Cornia 2004; Shorrocks and Van der Hoeven 2004; and Van der Hoeven and Shorrocks 2003). 3 These issues can also be approached from a political perspective. In the face of

resent-ment over rapidly changing income distribution, political challenges can spill directly into economic policy, which affects the speed and direction of economic growth. A paper pursuing these political economy dynamics in the Asian context is under way (see Timmer and Timmer forthcoming).

4 The eight countries selected are the only countries in Asia with consistent and reliable data on income distribution over at least two decades—each country has at least eight observations on the distribution of household income or expenditures. The basic data are from World Bank Indicators (www.worldbank.org/data/wdi2003/) and a special data set developed for the project ‘Operationalizing Pro-poor Growth’ (World Bank 2004). Appendix 1, table 3, shows the period covered by the data for each country. 5 The last recorded Gini value for India in figure 2 suggests a dramatic increase in

inequality during the period of more rapid growth, but there is some doubt as to whether this reflects a trend.

6 The actual regression is as follows (with t-statistics in parentheses): Ldgap20 = 0.798 + 0.426 * Pcigrow + 0.692 * Lgapstart,

(0.9) (9.2) (5.4) (R squared = 0.956) where

Ldgap20 = logarithm of the change in the income gap over 20 years (calculated from the rate for the entire period);

Pcigrow = growth rate in per capita income over the entire time period (% per year); and

Lgapstart = logarithm of the income gap at the start of the period (in constant 1993 $PPP).

7 Most researchers have found that relationships that hold for the bottom quintile hold closely for other definitions of the poor as well; see Cord, Lopez and Page (2003) for a review.

8 Income distribution has widened within sectors as well, although roughly two-thirds of the rise in the overall Gini in China can be attributed to the growing wage gap between urban and rural workers. But it has been argued that in urban areas income distribution has also deteriorated and poverty increased, largely because of increases in food prices and the need to spend more on services previously supplied by the state (Meng, Gregory and Wang 2004).

9 A similar attempt to disaggregate changes in income distribution into urban and rural differentials is described in Eastwood and Lipton (2004).

10 Regressions similar to those presented in table 3 for Loggini as the dependent variable were run with the log of the income gapas dependent variable, as well as the relative gap. In general, the results were similar to those for Loggini;they are not presented here in the interests of space.

11 There are two points here. The first is the economicsignificance of the variable, in con-trast to its statistical significance as reflected by the t-statistic. To judge this, each esti-mated coefficient is multiplied by one standard error in the variable itself, to judge the

(25)

size of the impact such a change would have. The following ‘importance’ coefficients result:

Logsect 0.075 Wagegap 0.030 Gdpserv 0.014 Saving 0.013 Ricertof 0.015 Ssrresid 0.038 Kcalresid 0.015

If all of these variables had a one standard deviation move in the ‘bad’ direction at the same time, the Gini coefficient would rise from 38.0 to 46.3, which would represent almost exactly a one standard deviation movement in the Gini for the whole sample.

The second point is whether sectoral ‘parity’ means policies that protect domestic agricultural producers during the structural transformation. Although that is one option, followed aggressively by Japan and South Korea, the alternative is to invest in more flexible connections between the rural and urban economies, including migration and financial flows. This path is still open to Indonesia (Timmer 2003).

12 For an excellent short summary of the Soeharto years, see Hofman, Rodrick-Jones and Thee (2004). For greater historical detail, see Hill (1996) and the ‘Survey of Recent Developments’ in each issue of the Bulletin of Indonesian Economic Studies.

13 A recent working paper from the SMERU Research Institute attributes two-thirds of the reduction in the overall poverty headcount index achieved at the provincial level between 1984 and 1996 to growth in agriculture. Industrial growth was only margin-ally significant in reducing urban poverty (Sumarto and Suryahadi 2004).

14 Sharp differences between transactions costs in Holland and in colonial Netherlands East Indies (Indonesia) have been offered as a major reason for their different develop-ment paths from the 18th through the early 20th centuries (see Van Zanden 2002, 2004; Van der Eng 2002).

15 ABIESreviewer suggested that this was a very controversial interpretation of Bulog’s activities from 1970 to 1996. The analytical arguments and empirical defence are sum-marised in Timmer (1996), with more institutional detail available in Timmer (1991). 16 See Artadi and Sala-i-Martin (2003) for an explanation of the economic policies that led

to Africa’s decline since 1960. Every variable in their list of contributing factors has direct parallels to issues now facing Indonesia.

REFERENCES

Alesina, Alberto, and Dani Rodrik (1994), ‘Distributive Policies and Economic Growth’,

Quarterly Journal of Economics109 (2): 465–90.

Artadi, Elsa, and Xavier Sala-i-Martin (2003), ‘The Economic Tragedy of the XXth Century: Growth in Africa’, NBER (National Bureau of Economic Research) Working Paper No. 9865.

Barro, Robert J. (1998), Determinants of Economic Growth: A Cross-Country Empirical Study, MIT Press, Cambridge MA.

Barro, Robert J., and Xavier Sala-i-Martin (2003), Economic Growth, 2nd edition, The MIT Press, Cambridge MA.

Booth, Anne (2002), ‘Rethinking the Role of Agriculture in the “East Asian” Model: Why Is Southeast Asia Different from Northeast Asia?’, ASEAN Economic Bulletin19 (1): 40–51. Bourguignon, François (2004), The Poverty–Growth–Inequality Triangle, World Bank,

Washington DC, February (mimeo).

Cord, Louise, J. Lopez, and John Page (2003), ‘When I Use a Word … Pro-Poor Growth and Poverty Reduction’, in M. Krakowski (ed.), Attacking Poverty: What Makes Growth Pro-Poor? Nomos Verlagsgesellschaft, Baden-Baden.

(26)

Cornia, Giovanni A. (ed.) (2004), Inequality, Growth, and Poverty in an Era of Liberalization and Globalization, Oxford University Press for UNU/WIDER (United Nations Uni-versity/World Institute for Development Economic Research).

Datt, Guarav, and Martin Ravallion (1992), ‘Growth and Redistribution Components of Changes in Poverty Measures: A Decomposition with Application to Brazil and India in the 1980s’, Journal of Development Economics38 (3): 275–95.

Deininger, Klaus, and Lyn Squire (1996, with updates), ‘A New Data Set Measuring Income Inequality’, The World Bank Economic Review10 (3) (September): 565–91. Data available at www.worldbank.org/research/growth/dddeisqu.htm.

Dollar, David, and Aart Kraay (2002), ‘Growth Is Good for the Poor’, Journal of Economic Growth7: 195–225.

Easterly, William (2002), ‘Inequality Does Cause Underdevelopment’, Working Paper No. 1, Center for Global Development, Washington DC, January.

Eastwood, Robert, and Michael Lipton (2004), ‘Rural and Urban Income and Poverty: Does Convergence between Sectors Offset Divergence within Them?’, in G.A. Cornia (ed.),

Inequality, Growth, and Poverty in an Era of Liberalization and Globalization, Oxford University Press, Oxford: 112–43.

Forbes, Kristen (2000), ‘A Reassessment of the Relationship between Inequality and Growth’, American Economic Review90 (4): 869–87.

Galbraith, James K., and Hyunsub Kum (2003), ‘Estimating the Inequality of Household Incomes: Filling Gaps and Fixing Problems in Deininger & Squire‘, UTIP Working Paper No. 22 (revised), 29 October, University of Texas, Austin.

Gugerty, Mary Kay, and C. Peter Timmer (1999), Growth, Inequality and Poverty Alleviation: Implications for Development Assistance, Prepared for CAER II, USAID, Washington DC, November.

Heston, Alan, Robert Summers and Bettina Aten (2002), Penn World Table Version 6.1, Center for International Comparisons at the University of Pennsylvania (CICUP), Philadelphia, October.

Hill, Hal (1996), The Indonesian Economy since 1966, Cambridge University Press, Cambridge.

Hofman, Bert, Ella Rodrick-Jones and Thee Kian Wie (2004), Indonesia: Rapid Growth; Weak Institutions, Paper prepared for the Shanghai Conference ‘Scaling Up Poverty Reduction: Lessons and Challenges from China, Indonesia, Korea and Malaysia’, 25–27 May 2004, preliminary draft (May), The World Bank, Jakarta (mimeo).

Islam, Iyanatul (2002), Poverty, Employment and Wages: An Indonesian Perspective, ILO– JMHLW–Government of Indonesia Seminar on Strengthening Employment and Labour Market Policies for Poverty Alleviation and Economic Recovery in East and Southeast Asia, 29 April – 1 May, ILO (International Labour Organization), Jakarta (mimeo). Kraay, Aart (2004), ‘When Is Growth Pro-Poor? Cross-Country Evidence’, World Bank

Policy Research Working Paper 3225, Washington DC, March.

Kuznets, Simon (1955), ‘Economic Growth and Income Inequality’, American Economic Review45 (1): 1–28.

Lewis, W. Arthur (1954), Economic Development with Unlimited Supplies of Labour, The Manchester School of Economics and Social Studies, Manchester.

MacIntyre, Andrew (2001), ‘The Politics of Agricultural Policy-making: The Importance of Institutions’, in Ammar Siamwalla (ed.), The Evolving Roles of State, Private, and Local Actors in Asian Rural Development, Oxford University Press, Hong Kong: 243–70. Maddison, Angus (1994), ‘Explaining the Economic Performance of Nations, 1820–1989’,

in William J. Baumol, Richard R. Nelson and Edward N. Wolff (eds), International Convergence of Productivity, Oxford University Press, New York.

Manning, Chris (1998), Indonesian Labour in Transition: An East Asian Success Story?

Cambridge University Press, Cambridge.

(27)

Meng, Xin, Robert Gregory and Youjuan Wang (2004), Poverty, Inequality, and Growth in Urban China, 1986–2000, 5 April, Australian National University, Canberra (mimeo). Molyneaux, Jack (2003), Starchy Staple Consumption and Household Nutrition: A Fresh

Look at Indonesian Welfare, Indonesian Food Policy Program (www.macrofoodpolicy. com), Jakarta, February (mimeo).

North, Douglas (1990), Institutions, Institutional Change and Economic Performance, Cambridge University Press, Cambridge.

Papanek, Gustav (2004), The Poor during Economic Decline, Rapid Growth and Crisis: The Case of Indonesia, Prepared for the USAID Project on Pro-Poor Growth conducted by DAI (Development Alternatives, Inc.) and BIDE (Boston Institute for Developing Economies), Bethesda MD (mimeo).

Pearson, Scott R. (1990), ‘Financing Rice Price Stabilization in Indonesia’, Indonesian Food Journal 4 (7): 63–83.

Rock, Michael T. (2002), ‘Exploring the Impact of Selective Interventions in Agriculture on the Growth of Manufactures in Indonesia, Malaysia, and Thailand’, Journal of International Development14 (4): 485–510.

Rock, Michael T. (2003), ‘The Politics of Development Policy and Development Policy Reform in New Order Indonesia’, William Davidson Institute Working Paper 632, University of Michigan Business School, Ann Arbor MI, November.

Rodrik, Dani (2003), ‘Growth Strategies’, ksghome.harvard.edu/~.drodrik.academic.ksg/ growthstrat10.pdf (forthcoming in Handbook of Economic Growth, to be published by Elsevier/North-Holland, Amsterdam).

Shorrocks, Anthony, and Rolph van der Hoeven (eds) (2004), Growth, Inequality, and Poverty: Prospects for Pro-Poor Economic Development, Oxford University Press for UNU/WIDER, Oxford.

Smith, Adam (1776), An Inquiry into the Nature and Causes of the Wealth of Nations; see the edition of Smith’s work based on Edwin Cannan’s careful 1904 compilation of Smith’s fifth edition of the book (1789): Cannan, Edwin (ed.) (1904), An Inquiry into the Nature and Causes of the Wealth of Nations. An Introduction and Notes, Methuen and Co. Ltd, London.

Strauss, John, Kathleen Beegle, Agus Dwiyanto, Daan Pattinasarany, Elan Satriawan, Bondan Sikoki, Sukamdi and Firman Witoelar (2004), Indonesian Living Standards: Before and After the Financial Crisis, Institute of Southeast Asian Studies, Singapore; RAND Corporation, Santa Monica CA; and Gadjah Mada University, Yogyakarta.

Sumarto, Sudarno, and Asep Suryahadi (2004), The Role of Agricultural Growth in Poverty Reduction in Indonesia, The SMERU Research Institute, Jakarta, February (mimeo).

Thee Kian Wie (ed.) (2003), Recollections: The Indonesian Economy, 1950s–1990s,Institute of Southeast Asian Studies, Singapore.

Timmer, C. Peter (1981), ‘Is There “Curvature” in the Slutsky Matrix?’, Review of Economics and Statistics63 (3): 395–402.

Timmer, C. Peter (1991), ‘Food Price Stabilization: Rationale, Design, and Implementation’, in Dwight H. Perkins and Michael Roemer (eds), Reforming Economic Systems, Harvard Institute for International Development, Harvard University, Cambridge MA, distrib-uted by Harvard University Press: 219–48 (chapter) and 456–59 (notes and references). Timmer, C. Peter (1996), ‘Does Bulog Stabilise Rice Prices in Indonesia? Should It Try?’,

Bulletin of Indonesian Economic Studies32 (2): 45–74.

Timmer, C. Peter (2002), ‘Agriculture and Economic Growth’, in Bruce Gardner and Gordon Rausser (eds), The Handbook of Agricultural Economics, Vol. II, North-Holland, Amsterdam: 1,487–546.

Timmer, C. Peter (2003), ‘Food Security and Rice Price Policy in Indonesia: The Economics and Politics of the Food Price Dilemma’, in T.W. Mew, D.S. Brar, S. Peng, D. Dawe and

(1)

Look at Indonesian Welfare, Indonesian Food Policy Program (www.macrofoodpolicy. com), Jakarta, February (mimeo).

North, Douglas (1990), Institutions, Institutional Change and Economic Performance, Cambridge University Press, Cambridge.

Pearson, Scott R. (1990), ‘Financing Rice Price Stabilization in Indonesia’, Indonesian Food Journal 4 (7): 63–83.

Shorrocks, Anthony, and Rolph van der Hoeven (eds) (2004), Growth, Inequality, and Poverty: Prospects for Pro-Poor Economic Development, Oxford University Press for UNU/WIDER, Oxford.

Sumarto, Sudarno, and Asep Suryahadi (2004), The Role of Agricultural Growth in Poverty Reduction in Indonesia, The SMERU Research Institute, Jakarta, February (mimeo).

Thee Kian Wie (ed.) (2003), Recollections: The Indonesian Economy, 1950s–1990s,Institute of Southeast Asian Studies, Singapore.

Timmer, C. Peter (1981), ‘Is There “Curvature” in the Slutsky Matrix?’, Review of Economics and Statistics63 (3): 395–402.

Bulletin of Indonesian Economic Studies32 (2): 45–74.

Timmer, C. Peter (2002), ‘Agriculture and Economic Growth’, in Bruce Gardner and Gordon Rausser (eds), The Handbook of Agricultural Economics, Vol. II, North-Holland, Amsterdam: 1,487–546.

Timmer, C. Peter (2003), ‘Food Security and Rice Price Policy in Indonesia: The Economics and Politics of the Food Price Dilemma’, in T.W. Mew, D.S. Brar, S. Peng, D. Dawe and

202 C. Peter Timmer

(2)

B. Hardy (eds), Rice Science: Innovations and Impact for Livelihood, Proceedings of the International Rice Research Conference, 16–19 September 2002, Beijing, International Rice Research Institute, Chinese Academy of Engineering, and Chinese Academy of Agricultural Sciences: 777–88.

Timmer, C. Peter, Walter P. Falcon and Scott R. Pearson (1983),Food Policy Analysis, Johns Hopkins University Press for the World Bank, Baltimore MD.

Timmer, Ashley, and C. Peter Timmer (forthcoming), ‘The Distributive Foundations of Economic Growth: An Essay on Behavioral Political Economy’, Social Science Research Council and Center for Global Development, Washington DC.

USAID (2004), Pro-Poor Growth: A Guide to Policies and Programs, Prepared by DAI–BIDE for the USAID Office of Poverty Reduction, Washington DC.

Van der Eng, Pierre (2002), ‘Indonesia’s Growth Performance in the Twentieth Century’, in Angus Maddison, D.S. Prasada Rao and William F. Shepherd (eds), The Asian Economies in the Twentieth Century, Edward Elgar, Cheltenham.

Van der Hoeven, Rolph, and Anthony Shorrocks (eds) (2003), Perspectives on Growth and Poverty, United Nations University Press for WIDER, Tokyo.

Van Zanden, Jan Luiten (2002), On the Efficiency of Markets for Agricultural Products: Rice Prices and Capital Markets in 19th Century Java, Paper for the 62nd Annual Meeting of the Economic History Association, 11–13 October, International Institute for Social History, St Louis MO (mimeo).

Van Zanden, Jan Luiten (2004), ‘On Global Economic History: A Personal View on an Agenda for Future Research’, accessed at www.iisg.nl/staff/jvz.html.

World Bank (1993), The East Asian Miracle: Economic Growth and Public Policy, Oxford University Press for the World Bank, London and New York.

World Bank (2004), ‘Operationalizing Pro-Poor Growth’, A Research Project Sponsored by AFD (Agence Française de Développement), DFID (the UK Department for Interna-tional Development), GTZ (Deutsche Gesellschaft für Technische Zusammenarbeit, Germany’s government-owned development cooperation enterprise), KfW Banking Group and PREM (Poverty Reduction and Economic Management Department of the World Bank), Washington DC.

(3)

APPENDIX 1: DATA

TABLE 1

Variable List and Definitions

Gdpgap Average per capita gross domestic product of the bottom income quintile sub-tracted from average per capita GDP of the top income quintile. Calculated using quintile income-share data from an unpublished database (‘Pro-Poor Growth Database’) compiled by Emily Sinnott and Humberto Lopez of the World Bank; World Bank (2004); and GDP data as defined below (under Loggdp) for GDPPC.

Gdpgrow Annual growth (%) in per capita GDP, calculated using GDP data as defined below, from Heston, Summers and Aten (2002), pwt.econ.upens.edu/. Loggdp Logarithm of gross domestic product per capita (GDPPC), in

purchasing-power-parity adjusted dollars, from Heston, Summers and Aten (2002), pwt.econ.upens.edu/.

Gdpserv Percentage of gross domestic product value added from services, from World Bank, World Development Indicators Online, publications.worldbank. org/WDI/.

Loggini Logarithm of the whole-country Gini index of inequality, using the ‘high-quality’ subset of data from Klaus Deininger and Lyn Squire, ‘Measuring Income Inequality: A New Database’, www.worldbank.org/research/ growth/dddeisqu.htm.

Logsect Logarithm of the computed Gini using only agricultural and non-agricultural shares of GDP and agricultural employment and non-agricultural employ-ment shares. Data are from World Bank, World Developemploy-ment Indicators Online, publications.worldbank.org/WDI/.

Kcalresid Residuals from OLS (ordinary least squares) regression: caloric intake = α+ β*log(GDPPC). Average daily caloric intake data are from FAO, FAOSTAT Agriculture Data, apps.fao.org/page/collections?subset=agriculture. Ricertof Ratio of annual average retail price of rice to farm price, adjusted for milling

using a 0.67 conversion factor. Price data from International Rice Research Institute, World Rice Statistics, tables 24 and 26, www.irri.org/science/ ricestat/index.asp.

Saving Gross domestic savings rate, from World Bank, World Development Indicators Online, publications.worldbank.org/WDI/.

Ssrresid Residuals from OLS regression: starchy staple ratio = α+ β*log(GDPPC). The starchy staple ratio was computed using total average calorie intake and starchy staple average caloric intake data from the Food and Agriculture Organization, FAOSTAT Agriculture Data, apps.fao.org/page/collections? subset=agriculture.

Logtheil Logarithm of the Theil index of manufacturing wage dispersion, from James Galbraith and Hyunsub Kum, ‘EHII: An Estimated Household Income Inequality Data Set for the Global Economy’, utip.gov.utexas.edu/ web/Data/EHIIUTIP22/introducingehii.htm.

Wagegap Non-agricultural value added per worker divided by agricultural value added per worker. Calculated using value added data and employment shares from World Bank, World Development Indicators Online, publications. worldbank.org/WDI/.

204 C. Peter Timmer

(4)

The Road to Pr

o-Poor Gr

owth: Indonesian Experience in Regional Perspective

205

APPENDIX 1 TABLE 2

Variable Means by Country (standard deviations in parentheses)

China Indonesia India South Korea Malaysia Pakistan Philippines Thailand Full Sample

Gini coefficient 33.3 33.0 32.4 34.8 50.2 31.7 48.0 44.5 36.9

(4.61) (2.13) (2.25) (2.60) (1.60) (0.82) (2.24) (3.35) (7.14) Theil index 0.0025 0.0826 0.0764 0.0284 0.0339 0.0460 0.0616 0.0941 0.555 (0.001) (0.020) (0.018) (0.007) (0.009) (0.016) (0.015) (0.032) (0.030)

Intersectoral Gini 67.4 64.5 64.6 50.9 61.3 66.3 64.5 72.1 63.8

(0.92) (4.96) (2.72) (14.95) (5.49) (1.33) (2.94) (3.46) (8.69)

Non-agricultural/ 5.8 3.0 3.0 2.4 1.5 3.0 2.8 9.6 3.9

agricultural wage ratio (1.73) (0.70) (0.32) (0.47) (0.15) (0.39) (0.38) (1.15) (2.61) Share of GDP from 28.4% 37.3% 38.8% 45.5% 41.9% 45.7% 42.9% 48.2% 41.6% services (4.78) (3.09) (5.09) (4.80) (3.02) (4.45) (5.73) (2.21) (6.86)

Growth in 5.8% 3.4% 2.8% 6.0% 4.0% 2.9% 1.4% 4.7% 3.8%

per capita income (4.2) (3.9) (3.2) (4.1) (2.9) (2.9) (3.0) (3.6) (3.7) Per capita income 1,904 2,134 1,366 6,519 5,480 1,285 2,789 3,346 3,116 ($PPP) (878) (1,051) (459) (4,509) (2,294) (457) (413) (1,947) (2,706)

Gap between top and 707 794 414 1,935 2,632 446 1,355 1,560 1,210

bottom quintile income (363) (322) (134) (973) (721) (122) (148) (919) (901)

Saving rate 37.0% 23.9% 18.5% 24.6% 32.2% 10.8% 20.1% 26.0% 23.7%

(4.2) (10.1) (3.5) (10.9) (8.2) (3.7) (3.9) (6.6) (10.0)

Ratio of retail 0.70 1.46 1.28 1.01 1.22 1.25 1.44 1.70 1.30

to farm rice prices (0.27) (0.32) (0.18) (0.07) (0.20) (0.17) (0.11) (0.25) (0.34)

Starchy staple ratio 0.06 0.08 0.03 0.11 0.08 0.11 –0.04 0.02 0.00

(residual from GDP regression) (0.04) (0.02) (0.01) (0.07) (0.04) (0.02) (0.02) (0.04) (0.08)

Caloric intake –64.9 59.3 85.9 207.6 38.9 141.6 314.6 174.1 0.0

(5)

APPENDIX 1 TABLE 3 Summary of Data Availability

Gini Data Macro/Sectoral Data Rice Price Data

China 1980–95 1975–2000 1975–95

Indonesia 1967–2002 1960–2002 1966–95

India 1951–97 1960–2000 1966–96

South Korea 1965–98 1960–2000 1966–93

Malaysia 1970–97 1960–2000 1966–89

Pakistan 1969–96 1960–2000 1966–94

Philippines 1957–97 1960–2000 1961–99

Thailand 1962–98 1960–2000 1961–97

APPENDIX 2: GALBRAITH–KUM INEQUALITY ESTIMATES

In a recent paper, Galbraith and Kum (2003) argue that Asian income inequality,

as measured by household expenditure surveys ‘… is much higher than a casual

reading of the D & S (Deininger and Squire) data would suggest’. Galbraith and

Kum argue that one reason is the poor quality of the income distribution data

themselves.

Galbraith and Kum propose an alternative—to use data on dispersion of

man-ufacturing wages, along with two structural variables, to construct a substitute

for the D & S data. The advantages, they claim, are annual availability of the wage

data and much less ‘noise’ than in the D & S data. Further, they claim, their

con-structed substitute has more plausible levels of income distribution than are

pre-sented in the D & S data, especially for Asia.

Unfortunately, their approach is basically irrelevant in Asia. The empirical

results reported in appendix 2, table 1, show that the key structural dimensions

of each economy, along with differences in the food and agricultural sectors,

con-tribute significantly to understanding differences in income distribution across

the countries in the sample explored in this paper and over time. The

Galbraith–Kum specification fails to contribute at all.

When these variables are included in the predictive equation, the Theil

coeffi-cient on manufacturing wage dispersion (Logtheil) drops out entirely! As best the

analysis can indicate, the Theil distribution data on wage dispersion basically

substitute for country-specific effects. Indeed, a country’s data on GDP per capita

(Loggdp) dominate statistically the log of the Theil coefficient in all the results.

206 C. Peter Timmer

(6)

APPENDIX 2 TABLE 1 Explaining the (Log of the) Gini Coefficient across Time and

Countries (for Indonesia, Malaysia, Philippines, Thailand, India, Pakistan, South Korea

and China) using the Galbraith–Kum Measure of Inequality (LogTheil)

Independent Equation Number

Variablea

A-1 A-2 A-3

Constant 1.853 –2.068 0.4686

(7.9) (5.9) (0.9)

Logsect 0.4160 0.7498 0.4828

(9.5) (11.3) (4.6)

Wagegap –0.0094 0.0003 –0.0314

(0.7) (0.1) (4.6)

Gdpserv 0.0022 –0.0103 0.0161

(1.8) (4.7) (6.3)

Saving 0 0025 –0.0126 0.0082

(2.2) (8.1) (5.2)

Ricertof 0.0039 0.0812 0.1690

(0.2) (2.2) (2.8)

Ssrresid –0.2100 –0.1607 –0.1306

(1.1) (1.4) (0.7)

Kcalresid –0.00008 –0.00031 –0.00039

(2.6) (8.8) (6.8)

Loggdp –0.0181 0.4107 –

(0.8) (17.1)

Logtheil –0.0194 –0.0022 –0.0668

(1.6) (0.2) (3.1)

Fixed effects? Y N N

R squared 0.820 0.517

Rho 0.966

a_{See table 3 for details of the regression specifications.}

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205277

Bulletin of Indonesian Economic Studies

The road to pro-poor growth: the Indonesian

experience in regional perspective

C. Peter Timmer

THE ROAD TO PRO-POOR GROWTH:

THE INDONESIAN EXPERIENCE IN

REGIONAL PERSPECTIVE

APPENDIX 1: DATA

TABLE 1

Variable List and Definitions

APPENDIX 1 TABLE 2

Variable Means by Country (standard deviations in parentheses)

APPENDIX 1 TABLE 3 Summary of Data Availability

APPENDIX 2: GALBRAITH–KUM INEQUALITY ESTIMATES

In a recent paper, Galbraith and Kum (2003) argue that Asian income inequality,

as measured by household expenditure surveys ‘… is much higher than a casual

reading of the D & S (Deininger and Squire) data would suggest’. Galbraith and

Kum argue that one reason is the poor quality of the income distribution data

themselves.

Galbraith and Kum propose an alternative—to use data on dispersion of

man-ufacturing wages, along with two structural variables, to construct a substitute

for the D & S data. The advantages, they claim, are annual availability of the wage

data and much less ‘noise’ than in the D & S data. Further, they claim, their

con-structed substitute has more plausible levels of income distribution than are

pre-sented in the D & S data, especially for Asia.

Unfortunately, their approach is basically irrelevant in Asia. The empirical

results reported in appendix 2, table 1, show that the key structural dimensions

of each economy, along with differences in the food and agricultural sectors,

con-tribute significantly to understanding differences in income distribution across

the countries in the sample explored in this paper and over time. The

Galbraith–Kum specification fails to contribute at all.

When these variables are included in the predictive equation, the Theil

coeffi-cient on manufacturing wage dispersion (Logtheil) drops out entirely! As best the

analysis can indicate, the Theil distribution data on wage dispersion basically

substitute for country-specific effects. Indeed, a country’s data on GDP per capita

(Loggdp) dominate statistically the log of the Theil coefficient in all the results.

APPENDIX 2 TABLE 1 Explaining the (Log of the) Gini Coefficient across Time and

Countries (for Indonesia, Malaysia, Philippines, Thailand, India, Pakistan, South Korea

and China) using the Galbraith–Kum Measure of Inequality (LogTheil)

Dokumen yang terkait

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205187

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205196

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205204

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205213

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205222

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205231

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205240

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205259

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205268

Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205286

Dokumen yang Anda mencari sudah siap untuk unduhkan