Towards a Healthy Indonesia?

Abstract

During his first presidential term, Joko Widodo increased expenditure on, and the coverage of, several social protection policies, including the conditional cash transfer program. These policies began in the aftermath of the 1997–98 Asian financial crisis and have proliferated in recent years. This Survey will examine these policies, paying particular attention to implementation problems, including effective targeting through the construction of a unified database. It will also examine both food policy and broader health policy issues. It is widely agreed that health problems, such as those relating to early childhood development, must be addressed in Indonesia in a wider context, including through the provision of clean water and sanitation facilities, food security, and social assistance. The Survey will also examine recent discussions of trends in inequality and poverty, several of which claim that inequality has been increasing. Using recent figures published by Statistics Indonesia, it is argued that expenditure inequality has in fact been trending downward in recent years.

Pada periode pemerintahannya yang pertama, Joko Widodo menambah pengeluaran dan cakupan beberapa kebijakan perlindungan sosial, termasuk bantuan langsung bersyarat. Tipe kebijakan seperti ini telah dimulai sejak krisis keuangan Asia di 1997-98 dan semakin berkembang di tahun-tahun belakangan. Survei ini menelaah kebijakan-kebijakan tersebut dan secara khusus memperhatikan berbagai masalah implementasi, termasuk penargetan efektif melalui pembuatan sistem data terpadu. Survei juga menganalisis isu-isu kebijakan pangan dan kesehatan yang lebih luas. Banyak disetujui bahwa masalah-masalah kesehatan, seperti yang mencakup perkembangan dini anak, harus disasar dalam konteks yang lebih luas di Indonesia – termasuk pemenuhan air bersih dan fasilitas sanitasi, ketahanan pangan, dan bantuan sosial. Selain itu, survei ini juga menelaah diskusi terkini soal tren dalam ketimpangan dan kemiskinan – beberapa di antaranya mengklaim bahwa ketimpangan semakin meningkat. Menggunakan angka-angka terbaru yang diterbitkan oleh BPS, penulis berpendapat bahwa ketimpangan pengeluaran justru telah mengalami tren penurunan belakangan ini.

Keywords:

• growth
• health policy
• social protection
• unified database
• food policy
• stunting

JEL classification:

• E20
• E60
• I12
• I15
• I38
• Q18

INTRODUCTION

The parliamentary and presidential elections in April 2019 passed off peacefully, and quick counts conducted after the polls closed suggested that Joko Widodo (Jokowi) had won by a rather larger margin than in 2014. The official results released on 21 May gave Jokowi 55.5% of the vote, and his opponent, Prabowo Subianto, 44.5%. Prabowo claimed that there had been massive cheating and irregularities in the poll, and he will challenge the results in the courts, although it is not expected that the challenge will be successful. Protests by his supporters led to eight deaths and several hundred wounded in Jakarta. The results raised fears in some quarters that Indonesia was becoming more polarised along religious lines; Jokowi polled strongly in Central and East Java, in Bali, and in provinces in Sulawesi and Sumatra with substantial Christian minorities, while Prabowo drew much of his support from West Java, South Sulawesi, and some parts of Sumatra. On the other hand, the results showed that many Muslims in Indonesia continue to support Jokowi’s brand of moderate Islam, and many gave their votes to secular political parties.

Jokowi made several statements in early 2019, promising that, if elected to a second term, he would switch the focus of government spending from infrastructure, on which he claimed that considerable progress had been made, to human resource development (Kompas 2019). Other promises made during the campaign included the ‘Sembako murah’ [Cheap food] pledge. This referred to the prices of the nine staple commodities that have a high weight in the expenditures of low-income households. They are rice, cooking oil, sugar, chicken, beef, eggs, corn, red onions, and soybeans. Although the president appears to want domestic prices of these staples to be ‘cheap’, in several cases, most notably those of rice and sugar, the prices are well above those prevailing in global markets. This is also true of other foods, including fruit and vegetables, the consumption of which is much lower in most Indonesian households than World Health Organization standards recommend (Arifin et al. 2018).

Poor diet, poor sanitation, and inadequate access to clean water are in turn linked to other health problems in Indonesia, which have attracted considerable international attention in recent years. These health problems include child stunting (low height for age) and child wasting (low weight for height). According to a study published by a group of United Nations agencies in 2017, Indonesia has a high percentage of stunting and wasting among children under five, compared with neighbouring countries (table 1).

Table 1. Undernourished Population, and Stunting and Wasting in Children (%)

Source: FAO et al. (2017), table A1.1.

Note: Countries are ranked according to their Human Development Index position in 2015. Stunting refers to the percentage of children who have a low height for their age. Wasting refers to the percentage of children who have a low weight for their height.

During his first term, Jokowi increased expenditure on, and the coverage of, several social protection policies, including the conditional cash transfer program. These policies began in the aftermath of the 1997–98 Asian financial crisis (AFC) and have proliferated in recent years. This Survey will examine these policies from a longer-term perspective than is usual in BIES Surveys, paying particular attention to implementation problems, including the issue of effective targeting. It will also examine both food policy and broader health policy issues. It is widely agreed that health problems, including those relating to early childhood development, must be addressed in Indonesia in a wider context, including through the provision of clean water and sanitation facilities, food security, and social assistance (Rokx, Subandoro, and Gallagher 2018, 5). Improved databases are also essential for compiling lists of beneficiaries. This Survey will look at recent figures on poverty and distribution published by Statistics Indonesia (BPS), and it will review the problems with data from the National Socio-economic Survey (Susenas). It will then assess how useful these surveys are for measuring trends in poverty and inequality, and for framing future social protection policies in Indonesia.

MACROECONOMIC DEVELOPMENTS

Economic growth slowed slightly in the first quarter of 2019, to 5.07% (year on year). In a press statement released on 16 May, Bank Indonesia (BI) cited a decline in global economic growth, combined with uncertainty over the election result, which affected investment expenditures, as a possible cause of this outcome. BI also noted that the impact of the election on consumption expenditures was less than expected. Some commentators thought that in the aftermath of the election, BI might lower interest rates, but in fact it was announced on 16 May that the benchmark rate (the seven-day reverse repo rate) would be kept at 6%. Other rates were also kept on hold. The decision was influenced by a slight weakening of the rupiah in early May, which BI attributed to the trade war between the United States and China; they pointed out that the climate of global uncertainty was leading to a general weakening of emerging market currencies. The policy appears to have been successful at least in the short run; by the end of May, the rupiah had appreciated slightly compared with earlier in the month.

Another cause for concern was the deteriorating export performance; in April the current account deficit was $25 billion, an historic high. The account deficit for 2019 is projected to be between 2.5% and 3% of GDP. This should be manageable, but if the deficit widens later in the year, there may be further pressure on the rupiah. Some commentators have projected a decline in export receipts in 2019 of between 8% and 9%, which would be higher than in any year since 2015. Inflation accelerated slightly in April compared with the previous month, but at 2.8% (year on year), it remained lower than in 2018 and well within government targets. Recent World Bank forecasts of economic growth in 2019 suggest that it will be about 5.2%, although if both domestic and foreign investor confidence returns after the election, it could be slightly higher (World Bank 2018a, table ES 1).

Where are the clouds on the horizon? Clearly the uncertain global economic climate remains a problem; a growth slowdown in China induced by the trade tensions with the United States could further depress commodity prices and lead to a wider balance-of-payments deficit. Perhaps more worrying are the budget projections for 2019. The budget projections published by the World Bank (2018a, table A4) show an increase of 43.6% in subsidies in the 2019 budget compared with 2018.

This was mainly due to the projected increase in fuel subsidies of more than 100%. These projections were made on the assumption of world oil prices being $70 per barrel; prices of both Brent Crude and West Texas in early June were lower than this. On the other hand, tensions between the United States and Iran could worsen, with consequences for world oil markets that are difficult to predict. But even if the world oil price does fall below $70 per barrel in the second part of 2019, it is likely that budget subsidies for fuel will increase, which would limit fiscal space for other expenditures.

POVERTY AND INCOME DISTRIBUTION DATA

Over the past few years, a growing body of literature has discussed the apparent rise in expenditure inequality, particularly during the first decade of the new millennium (Yusuf, Sumner, and Rum 2014; World Bank 2016; Yusuf and Warr 2018). These estimates use the expenditure data from the Susenas household survey, which has been carried out in Indonesia since the 1960s, although its coverage has been national only since the late 1980s. Problems with the survey have in fact received some attention in recent years, although not all researchers who use Susenas to estimate inequality seem aware of them. The first problem is the large gap between Susenas figures on household consumption and those from the national accounts (Booth 2016, table 8.5). In 1996, the Susenas estimates of household consumption expenditures were about 50% of personal consumption expenditures as measured in the national accounts. After that, there was a steady decline, to about 37% by 2009. There was some improvement after 2010, and from 2011 to 2018, the ratio appears to have fluctuated between 41% and 45%.

The disparity is greater for non-food than food expenditures, which suggests that under-reporting is greater for expenditures such as housing, transport, and leisure activities (table 2). This supports the argument that an important reason for the disparity is that the better-off households in urban areas are difficult to enumerate, either because they are too busy to fill in the form (which takes at least three hours to complete) or because they do not give a full account of their expenditures. In addition, there have been problems with the Susenas sample, which appears to have been skewed towards the middle-and lower-income groups. In 2011, BPS undertook a major change in the sampling methodology, which led to a greater diversity of neighbourhoods being included.¹ As a result of this change, the inequality estimates for 2011 and after are not strictly comparable with those from previous years (World Bank 2016, 41).² But even allowing for these changes, we see that the problem of the disparity between the Susenas estimates and those from the national accounts data persists. While there are valid reasons why estimates in the national accounts data are likely to be higher than those in the household survey data (the national accounts data include an estimate of the rental value of owner-occupied housing and the contribution of the non-profit institutions serving households), a disparity of more than 50% is usually considered to result from understatement in the household survey data.³ In neighbouring countries such as Thailand, Vietnam, and the Philippines, the disparity is less than in Indonesia (Booth 2019, table 7). It appears that in spite of the changes in sampling procedures, more affluent households in urban areas are still under-represented, or, if they are included, that they understate their expenditures.

Table 2. Household Consumption Expenditures: Susenas Estimates as a Percentage of National Accounts Estimates

Source: BPS (Statistics Indonesia). Accessed 1 May 2019. www.bps.go.id.

Note: GDP and Susenas data refer to the March quarter.

A further problem relates to the change in the estimation of the Gini ratio on inequality, from using grouped data to using individual household data; this led to an increase in the Gini and other measures of inequality after 2008 (Asra 2014; Yusuf, Sumner, and Rum 2014). Using individual household data, Yusuf, Sumner, and Rum (2014, figure 3), show that inequality increased from 2003 to 2013. But these estimates ignore the change in sampling after 2010, to which the World Bank (2016) drew attention. At least part of the increase was probably the result of the post-2010 sample including more affluent neighbourhoods, especially in urban areas. If we look at the estimates of the Gini coefficient from 2011 to 2018, which are estimated using individual household data and the same sampling frame, it would appear that there was little change in the Gini after 2011; indeed, between March 2015 and March 2018, there appears to have been a fall, especially in urban areas (figures 1 and 2).

FIGURE 1 Gini Ratio, All Indonesia, 1996–2017

Source: BPS (Statistics Indonesia). Gini index by province, urban and rural areas, 1996–2018. Original source: Susenas.

Note: Until 2005, Susenas was conducted on a three-yearly basis, and from 2007, on an annual basis. Since 2011, the survey has been conducted twice annually, in March and September. From 2011, the data in figure 1 are from the March round of the survey.

FIGURE 2 Gini Ratio, Urban, Rural, and All Indonesia, 2010–18

Source: BPS (Statistics Indonesia). Gini index by province, urban and rural areas, 1996–2018. Original source: Susenas.

*Data labels showing the values in alternate years are for urban and rural areas, starting in 2010.

To sum up, claims that there was an increase in inequality between 1999 and 2010 are probably correct, but they ignore the very sharp drop in the Gini and other inequality indicators that took place after 1996.⁴ Using 1996 as the base year, we see that there does seem to have been an increase in inequality, especially after 2003,

but that it is hardly as dramatic as some studies have suggested. Yusuf and Warr (2018, 136) report that the Gini (using household data) was .365 in 1996 and .397 in 2016; the 2016 figure is probably overstated relative to the former because of the changes in sampling methodology. So we are hardly looking at dramatic changes in inequality over these two decades. Indeed, since 2015 there appears to have been some decline in the Gini, especially in urban areas. The reasons are not entirely clear, but it is possible that the social protection policies may have contributed to the decline. These policies are examined in more detail below.

The headcount measure of poverty has continued to fall in recent years, although the rate of decline has slowed. Between March 2012 and March 2018, the headcount measure of poverty, using the BPS poverty line, fell from almost 12% to 9.8%. With population growth of about 1.1%–1.2% per annum, the number of poor people has fallen from 28.5 million to 26 million, a decline of only 1.5% per annum, which is much slower than the per capita growth rate of GDP over these six years. It seems that Indonesia is following ‘Gibson’s law’—that as the incidence of poverty declines, the elasticity between poverty decline and GDP growth falls, while the elasticity between poverty and inequality rises (Gibson 2016, 432). Gibson made this observation based on data from Vietnam, but it also seems to apply to Indonesia. To the extent that inequality has decreased only slowly between 2012 and 2018, it is to be expected that poverty decline will also be slow. This implies that the Indonesian government, like other governments in Asia that have experienced a fall in the headcount measure of poverty in recent years, will have to rely on more than just economic growth to further reduce poverty.⁵

EVOLUTION OF SOCIAL PROTECTION POLICIES SINCE 1998

In spite of the fall in poverty as measured by expenditure over the past two decades in Indonesia, both the government and many non-government groups have been concerned about Indonesia’s performance on a range of non-monetary indicators relating to both education and health. Recent studies have shown that, for a middle-income country, Indonesia seems to be doing rather badly on several human development indicators (table 3). These studies feature composite indexes of human capital that have been used to rank countries across the globe. The World Bank’s Human Capital Index, launched in 2018, ranked Indonesia at 87 out of 157 countries; this was worse than the rankings of several other Asian countries with similar or lower per capita GDP, including Sri Lanka, the Philippines, and especially Vietnam. While all these indicators have their problems, the overall message for Indonesia is not encouraging. On a range of health and education/skills indicators, the country should be doing better.

Table 3. Human Development Index Rankings

Source: World Bank (2018b); Lim et al. (2018); World Economic Forum (2017).

Note: Countries are ranked according to 2017 per capita GDP. The World Bank ranked 157 countries; Lim et al. (2018) ranked 195; and the World Economic Forum ranked 130.

The reasons for the rather disappointing performance are complex. Indonesia was bequeathed a poor legacy by the Dutch in terms of both health and education facilities. There was a rapid expansion of access to education after independence, and in the Soeharto era many new schools were built under the Inpres programs. But teachers were often poorly trained and could not teach subjects such as maths, science, or foreign languages. By the 1980s, Indonesia had quite a high teacher-to-student ratio compared with other developing countries, but a considerable number of teachers were surplus to requirements. As many of them were permanent civil servants (pegawai negeri), it was difficult to dismiss them.⁶ The problem has not really improved in the post-Soeharto era. It is well known that Indonesian students have performed badly in international tests such as the Trends in International Mathematics and Science Study (TIMSS) and the Programme for International Student Assessment (PISA).⁷ In spite of initiatives such as increased pay for teachers, the problems of poor student achievement persist and are proving difficult to ameliorate (Kurniawati et al. 2019, 285).

Health facilities were also expanded under Soeharto; the number of clinics (puskesmas) and village health posts (posyandu) grew rapidly. But trained staff were always in short supply, as were medicines and equipment. Newly graduated doctors were obliged to spend up to five years working in puskesmas, but this policy was changed in the 1990s and discontinued in 2007 (Rokx et al. 2010, 33–4). In recent years, it has become clear that puskesmas are unevenly spread across the country, and that there are wide variations in the quality of care they offer. According to Ministry of Health figures, in 2018 only about 40% of all puskesmas had the five categories of staff considered necessary for full preventive health services. Many of the posyandu have become inactive in recent years. While some villages have used their discretionary funds, including their village funds (dana desa), to build health clinics, the problems of shortages of staff and equipment persist. The results of these developments in terms of health outcomes will be discussed further below. Growing worries among Indonesian policymakers in the post-Soeharto era about the country’s poor performance in education and health, together with ongoing concerns about poverty and vulnerability, have been an important factor in the growth of social protection policies.⁸ Certainly funding has increased; one estimate using Ministry of Finance figures shows an increase in budgetary funds for social protection from Rp 19.42 billion in 2005 (.7% of GDP) to Rp 221.22 billion in 2017, or 1.6% of GDP (McCarthy and Sumarto 2019, figure 13.1). What has the money been spent on? An analysis of total expenditures on government programs aimed at assisting those individuals and families considered deprived (tidak mampu) carried out by the National Team for the Acceleration of Poverty Reduction (TNP2 K) found that in 2017 the total amount was Rp 203 trillion, or 1.5% of GDP.⁹ But 58% of this amount went on subsidies for LPG, electricity, and fertiliser (table 4). These benefit large numbers of households (54.9 million in the case of the LPG subsidy), but many of these households are not poor in the sense of being below the official poverty line.

Table 4. Funding of Social Protection Policies and Subsidies, 2017

Source: TNP2K (2018b).

^aThis program comprises four different projects that are intended to help poor households buy a simple house. The most important is the Housing Finance Liquidity Facility (FLPP), which is managed by the Ministry of Public Works and Housing and channelled through the banking system.

^bThere are about 12 other projects in the list compiled by TNP2K, implemented by several ministries. The expenditures for these projects are all less than Rp 1 trillion.

It has been estimated that 40% of the electricity subsidies and 72% of the LPG subsidies benefited non-poor households (TNP2 K 2018a, 118–19).¹⁰ The fertiliser subsidy has been criticised for poor targeting; it has also been claimed that a substantial amount of subsidised fertiliser has leaked to larger farms and plantations, especially in the palm oil sector (Wihardja 2019, 403).

The cheap rice program, which was introduced after the AFC, has gone through several name changes, from OPK to Raskin and now Rastra, the Prosperous Rice Program. Between 2001 and 2013, funding grew almost nine-fold, from Rp 2.4 trillion to Rp 21.5 trillion. This increase reflected both higher procurement costs and increased storage and distribution costs (Timmer, Hastuti, and Sumarto 2018, 285–7). The number of households benefiting from the program increased from 20.9 million in 2002 to 32.8 million in 2013.

For two decades, the program has been criticised both for weak targeting and for other failures, including the often poor quality of the rice, and the fact that many beneficiaries received less than the 15 kilograms they were entitled to each month and had to pay a higher price than the stipulated ||Rp 1,600 per kilogram. But in spite of the targeting errors, it has been estimated that 72% of households in the bottom decile did receive Raskin rice in 2014. Although the policy of distributing rice remains popular, the government has announced plans to incorporate Rastra into the non-cash food support program (BPNT), which in 2017 was being implemented in 44 towns. The coverage will be expanded, and about 25% of the population will receive a card that will allow them to buy both rice and eggs in shops equipped with internet facilities (e-warung) up to a limit of Rp 110,000 per month.¹¹

Other programs, such as those designed to help poor families purchase houses, may be better targeted but benefit only a relatively small number of people. This is also true of the scholarships that fund poor students’ university attendance, such as Bidikmisi. The president indicated during the election campaign that this scholarship program, which in 2017 reached only 80,000 students, may be expanded into a much larger program. The Healthy Indonesia Program (PIS) reached 96 million people in 2017.¹² It also paid for health insurance for people judged too poor to pay the premiums for the National Health Insurance fund (JKN). But many of the people receiving this assistance would not have been below the poverty line in their province. The Social Security Management Agency (BPJS), which administers the JKN, has been running a deficit in recent years; some critics argue that much of this deficit is the result of non-poor households that do not have health insurance using health facilities.

The most targeted program has probably been the Hopeful Families Program (PKH), which provided conditional cash grants to 6.2 million poor families in 2017 and 10 million in 2019. The PKH program grew out of previous cash transfer programs that were intended to compensate poor households for fuel price increases. They began on a modest scale in 2007 in only seven provinces and 48 districts and cities (kabupaten and kota). In that year, there were only 387,947 beneficiaries. The expansion in the number of beneficiaries has been accompanied by a rapid growth in the budget. In 2017, the estimated cost of the PKH was Rp 12.7 trillion, rising to Rp 32.6 trillion in 2019. There have also been changes in the amount that each beneficiary, who must be a woman, is entitled to. Households qualify for participation if they fall below a certain income threshold, have a pregnant or nursing mother, at least one child below age six or attending school, or one severely disabled member or a member over age 70. In 2019, the maximum amount a beneficiary can claim has been raised to Rp 10.15 million per year, although the great majority of beneficiaries get less than that.

Given that about 25 million people were estimated to be below the official poverty line in Indonesia in 2017, it is clear that many households who qualified for participation on income grounds were still excluded from the PKH program in that year (TNP2 K 2018a, 74). This inevitably led to resentment and was probably the reason for the substantial increase in both the number of beneficiaries and the budget between 2017 and 2019 (an election year). But the government appears committed to the continued expansion of the PKH to reach the bottom two quintiles of the expenditure distribution, an estimated 28.8 million families. This will involve a considerable increase in the cost of the program.

TNP2 K (2018a, 98–9) has suggested that the Smart Indonesia Program (PIP) for school scholarships and the PKH program be combined. They have a similar target group, and it has been proposed that the government replace both programs with a child benefit of Rp 200,000 per child per month paid to mothers for the first three children. If targeted at families in the bottom three or four income deciles, family benefits of this magnitude could have a significant impact on consumption levels, as well as on school attendance. But the proposed program would involve a substantially greater contribution from the budget than the existing PIP and PKH allocations. Advocates of an expanded PKH program can point to research suggesting that even the much smaller program implemented in earlier years led to increased usage of health professionals for childbirth and halved the numbers of children aged 7–15 who were not in school (Cahyadi et al. 2018). These authors also claimed that children who had grown up in households receiving cash grants showed considerable reductions in stunting.

Other suggested reforms include expanding the coverage of the PIS, although any expansion is likely to run into supply-side constraints, especially in rural areas. The problems facing the supply of health facilities in Indonesia will be examined in more detail below. It is obvious that a reduction, or elimination, of the subsidies listed in table 4 would free up considerable fiscal space for an expanded family benefit scheme, or expanded health benefits. From a political viewpoint, eliminating the LPG subsidy might be easiest, although this would have an adverse effect on many small businesses. A reduction or elimination of both the electricity and the fertiliser subsidy is likely to be more contentious.

In a review of programs intended to support early childhood development, the World Bank (2017, 30) noted that ‘despite the multitude of interventions, programs are neither integrated nor implemented at scale’, which reduces their impact. This criticism would appear to apply to other programs listed in table 4 as well. In spite of the increase, relative to GDP, in budgetary expenditures on social protection policies since 1998, it is still the case that many programs are still too small to have a significant impact on the population they are supposed to reach. Could more accurate targeting improve their effectiveness? This question is addressed in the next section.

TO TARGET OR NOT TO TARGET?

Most of the programs in table 4 involve some form of targeting. Many economists, both Indonesian and foreign, tend to favour targeting on the grounds that universal programs would be very expensive. They also argue that the current policy of targeting programs achieves greater social welfare than universal policies where the available funds would be spread much more thinly, and many of the beneficiaries would not be ‘poor’, however the term is defined. Hanna and Olken (2018) argue that even rather badly targeted programs, which exclude people who should be included, or include those who should be excluded, improve welfare, compared with universal programs. But this argument seems open to dispute on several grounds. The first concerns the cut-off point for receiving benefits. Those below the cut-off point get cash grants or some other benefit, and those above it get nothing. If the cut-off point is determined in a way that most people do not understand, it may appear unfair. How many Indonesians understand how the official poverty lines, or other cut-off points used for programs such as Rastra and PIS, are determined? If they do understand, do they agree with the methodologies, which vary from program to program? It has also been suggested that as families become more familiar with the methods used to establish the threshold level of income, involving, for example, proxy means testing,¹³ these families become aware of the strong incentives for ensuring that they keep their declared income, or ownership of assets, below the threshold level. This is especially the case when the cash and other benefits are likely to be substantial relative to household income.

A second problem concerns what is often called income churning. Evidence on this has been available for some years from panel data derived from rounds of the Indonesian Family Life Surveys. Between 2007 and 2014, it is estimated that one-third of all households climbed to a higher consumption quintile, while another third fell back. Only one-third of households stayed in the same economic class over this seven-year period (TNP2 K 2018a, 43). There are obviously many reasons for these movements; some households with low incomes may benefit from remittances from members working in another province or abroad, while others may be affected by the illness or death of the main earner. It is essential if government assistance is to be seen as fair by non-beneficiaries that records of household income be kept as up to date as possible.

In Indonesia, much effort over the past two decades has gone into producing and maintaining a ‘unified database’ (UDB), which by 2019 contained detailed socioeconomic information on almost 100 million people (see box). Although there has been some progress in gaining the trust and cooperation of many local governments across the country, it appears that a significant number of districts do not regularly revise or update their data. Thus, the targeting of social protection programs in these districts is likely to be flawed. Over time, there is a danger that the entire targeting system may appear to many people to be arbitrary and unjust. This is especially true where household income undergoes considerable change from year to year. It is almost certain that public doubts about the targeting system will increase if households that are considered poor when the UDB is compiled stay in the system and continue to get grants, even if their income improves, while others who drop down the income scale receive nothing.

BOX Indonesian Unified Database (UDB)

When Indonesia began to implement targeted social assistance programs after the AFC, it used locally validated data from the National Family Planning Coordinating Board (BKKBN). As the economy recovered, several of these programs were continued and new ones were introduced. The need for better information on individual households emerged when the government cut the fuel subsidy in 2005 and 2008, and needed to launch compensation programs. This led to special surveys being implemented; in 2005, a socioeconomic population survey (PSE 2005) took place—and later was updated in 2008 through the Data Collection for Social Protection Program (PPLS)—to establish a database that could be used to compile a beneficiary list for some compensation programs, including the unconditional cash transfer (BLT) program, Rice for the Poor (Raskin) program, and Health Insurance for the Poor (Askeskin) program. For both the PSE 2005 and the PPLS 2008, the main source of the household pre-lists was input from village administrations.

In the next PPLS round in 2011, the government mandated TNP2 K and BPS to improve the data enumeration and the method by which the household pre-list was compiled. The 2011 PPLS used a combination of poverty mapping and community suggestions to compose lists of poor households. Field surveys were carried out by BPS using proxy means testing, but there were problems. Implementation guidelines were not always followed, and in many districts, households deemed non-poor by enumerators or by community leaders were removed from the lists. Despite the problems, the 2011 PPLS covered 45%–50% of the population. Results from Susenas were used to determine the number and percentage of households in each decile and district. Households included in the UDB were divided into the four poorest deciles (Bah, Nazara, and Satriawan 2015). The PPLS 2011 laid the foundation for the UDB, which was intended to be a national registry of households identified as poor and needing government assistance. After the 2011 PPLS was carried out, a permanent national targeting unit in TNP2 K was established to manage the UDB and to facilitate its use by line ministries and local governments. Since 2012, the main social assistance programs—PKH, Raskin, PIP, and the national community health insurance program, Jamkesmas—have used the UDB to identify beneficiaries. Local governments have also used the UDB to plan their own social assistance programs and to target the beneficiaries of the program. The most recent nation-wide updating of the UDB was carried out in 2015. In composing the household pre-list for updating, TNP2 K and BPS conducted Development Research Forums (FKP) at the local level to scrutinise the lists, using input from both local governments and representatives from the community.

The cost of establishing the UDB has been estimated at Rp 600 billion for data collection (PPLS 2011), with an annual average operating cost of Rp 16.3 billion between 2012 and 2014 (Bah, Nazara, and Satriawan 2015). This worked out at Rp 8,700 per registered household per year. This was less than the cost per registered household for similar programs in parts of Latin America. Also, new social assistance programs can share the database rather than having to fund and design their own targeting systems. Between 2012 and 2014, the costs of the UDB were about .5% of the four main programs, PKH, Raskin, PIP, and Jamkesmas.

Since 2017, more regular updating of the UDB has been introduced, managed by the Ministry of Social Affairs. By 2018, the UDB contained detailed socioeconomic information on 28.8 million families. Responsibility for managing the UDB has gradually been transferred to the Ministry of Social Affairs and its regional offices (dinas sosial). Updating the database is now done twice yearly with support from local governments, although not all districts comply. In September 2018, it was estimated that 388 districts had updated their data, and 126 had not. Officials in the Ministry of Social Affairs have pointed to several reasons why districts are reluctant to update. These include the difficulties involved in visiting all households (especially in remote areas), high budget costs, and the lack of skilled enumerators in some districts. In addition, it appears that not all local governments understand the UDB and its potential benefits.

Some studies have claimed that targeting is likely to be more successful in rural areas than in towns and cities, because village government is quite strong and local leaders usually know who and where the poor are, and they will ensure that the available cash and other assistance get to the people who need it. But that may not be the case everywhere in the country. One study in Aceh found that the zakat was better targeted than government programs such as the PKH, because local religious leaders really did understand where the needs were greatest (McCarthy and Sumarto 2019, 376). The authors argued that half the people benefiting from the PKH cash grants were not poor based on the community’s own criteria. When poor villagers saw that they were excluded from receiving government grants while their better-off neighbours were not, they became resentful and blamed the village leaders.

It also needs to be recognised that Indonesia is now urbanising rapidly and over time fewer people will be living in traditional village communities. Increasingly the demand for health and education services will come from urban and peri-urban households. Quite a high proportion of the urban and peri-urban populations will be recent migrants. Even if the local officials are reasonably honest and competent, it will not always be easy for them to assess household income in order to target programs effectively. It may be easier, especially in urban areas, to screen out the better-off households, using proxy means tests, than to determine which households are in the bottom three or four deciles of the income distribution at any point in time. But even this may prove to be unpopular, and local officials may be reluctant to screen out potential beneficiaries.

FOOD POLICY ISSUES

Food security is a crucial component of social protection policy, especially in Indonesia, where there is still widespread evidence of poor nutrition, particularly among children. Stunting and wasting in children appear widespread; stunting affected at least 30% of the under-five age group in 2018 (table 5). There is much medical evidence that poor nutrition in early childhood affects cognitive development (Rokx, Subandoro, and Gallagher 2018, 6–7). The food security problems in Indonesia centre around rice policy and broader food availability issues, which in turn lead to the vexed question of greater reliance on food imports. Rice availability and the rice price are important in Indonesia because the poor still spend about 26% of their expenditure on rice and 65% on all foods, tobacco, and beverages. Researchers have shown that increases in rice prices, other things being equal, lead to an increase in the headcount measure of poverty. But over time, economic conditions do change; from 2010 to 2018, the headcount measure of poverty fell in spite of an increase of over 70% in the wholesale price of rice (table 6). Of course, it is likely that poverty would have fallen faster if the price of rice had been stable, or had risen less rapidly.

Table 5. Health Indicators for Children under Five: 2007, 2013, and 2018 (%)

Source: Ministry of Health (2019).

Table 6. Rice Prices: Ex-Vietnam and Indonesia Wholesale

Source: Vietnam export prices in dollars: International Rice Research Institute. https://.ricestat.irri.org. Indonesia wholesale price of rice and poverty incidence: BPS (Statistics Indonesia). www.bps.go.id. Exchange rate: Bank Indonesia. www.bi.go.id/en/statistik.

The reasons for the rapid increase in the rice price after 2010 are complex and cannot be attributed simply to rising protection, as the domestic price of rice was already close to twice that of the ex-Vietnam export price in 2010, and the differential has not changed much since then. The explanation seems to lie, atleast partly, with the depreciation of the rupiah, which also affects the price of other important food staples in Indonesia, including corn, wheat, soybean, and meat. Supply factors would also affect the trends shown in table 6. How much rice is in fact being produced in Indonesia? Until 2015, BPS used figures from the Ministry of Agriculture, although over the years many observers have raised doubts about the reliability of these figures, especially those relating to harvested area. In 2018, BPS published the results of a major survey of rice production in 15 provinces. It demonstrated what the critics have long suspected—that the planted and harvested area of rice is much less than the official statistics, supplied by the Ministry of Agriculture, have claimed.¹⁴ This in turn means that rice production figures are overstated. The 2018 estimates put production of paddy (gabah kering giling) at 56.54 million tonnes, which converts to 32.42 million tonnes of milled rice. Consumption availability of rice is estimated at 29.57 million tonnes or about 112 kilograms per capita (BPS 2018, 12).

The difference between production and consumption of about 2.85 million tonnes in 2018 was presumably added to stocks, together with imports, which have fluctuated in recent years but reached 2.25 million tonnes in 2018. These stocks were used for the Rastra subsidised rice program and also to stabilise domestic prices in the run-up to the election in April 2019. In fact, domestic price rises between 2016 and 2018 were modest, at under 5% over two years. This compares with rises of more than 8% per annum from 2010 to 2016. Although this reversal in policy seems to have received little publicity, the government now appears to be using rice imports to stabilise domestic prices, albeit with domestic prices pegged at a much higher level than the international price.

If the second Jokowi administration is to honour its pledge of keeping the prices of staples stable over the next five years, it seems inevitable that imports not only of rice but also of other foods will rise. In fact, imports of both wheat and corn have been increasing over the past two decades; according to figures from the United States Department of Agriculture, Indonesia is now among the largest importers of wheat in the world. Especially in urban areas, consumption of wheat-based foods (bread and noodles) is growing rapidly. But rates of effective protection on all food crops, as well as on milled rice and wheat flour, remain high in Indonesia, and indeed they appear to have increased between 2007 and 2015. Estimates of effective rates of protection on both food crops and food-processing industries show that they more than doubled over these eight years (Marks 2017, table 4), and in 2015 effective protection on food crops was five times the overall average on all traded goods. The increases appear to be mainly due to increased quantitative restrictions. Effective rates of protection were highest on field rice and sugar cane, followed by fruit, vegetables, and corn.

As will be seen below, health experts have expressed concern about the low levels of consumption of fruit and vegetables in Indonesia, especially among children. A more liberal import policy would reduce domestic prices and increase consumption. Indeed, economists have pointed out that import barriers on food affect the poor more than the non-poor, as food comprises a greater share of their expenditure, and most poor households are net consumers rather than producers of food. Should the Indonesian government focus more on removing barriers to trade in food rather than allocating budgetary resources to conditional cash transfers and other social protection policies? This question merits further research.¹⁵ Producer welfare would be adversely affected if food crop protection were reduced or eliminated, but poor consumers would benefit, and scarce budgetary resources could be re-allocated to other sectors to achieve improvements such as greater access to clean water and better sanitation, especially in poor areas.

THE HEALTH SECTOR

Since the AFC and the departure of Soeharto from office, the health sector has been through a number of changes; probably the most important have been a decline in the role of the government in providing and staffing public clinics, and a rise in the number of health professionals in private practice (Rokx et al. 2010, 50–54). For much of the Soeharto era, most doctors were civil servants, although many had private practices as well as public employment. Immediately after the AFC, the number of physicians relative to population declined, although it is unclear whether this was due to retirement or emigration. Between 1996 and 2006, there was an increase in the total number of physicians relative to population, although because of rapid urbanisation over these years, there was a decline in the number of doctors per 100,000 people in urban areas. The number of midwives increased relative to population in both urban and rural areas. The number of physicians per puskesmas also increased in the decade up to 2007, but this growth was accompanied by an increase in the number of puskesmas without a doctor, because doctors tended to cluster in urban areas where the opportunities for private practice were greater (Rokx et al. 2010, 42–8).

Decentralisation reforms created further problems for the health sector. It was one of the sectors where responsibility was devolved to the districts, so almost 250,000 health workers were transferred to local governments. But in reality, the centre retained considerable financial control (Rokx et al. 2010, 32–3). Overall expenditure on health care remained low in Indonesia; the United Nations Development Programme (UNDP 2010, 198–9) estimated that in 2007, per capita spending on health care in dollars adjusted for purchasing power parity was lower than in any other country in the medium development category except Pakistan. Since 2011, government spending on health has risen as a proportion of the national budget, from 3% to about 5%, but it is still less than 1% of GDP. Of the Rp 111 trillion budget allocation to health in 2018, Rp 21 trillion went to the Healthy Indonesia Card (KIS) program, whose direct benefit to the poor has been questioned. Other government expenditures on health provision are also not well targeted to poor people. Johar et al. (2018) find that only access to outpatient care in puskesmas was pro-poor in the years from 2011 to 2016. Access to most other facilities, both public and private, tends to benefit the non-poor, although in urban areas in-patient care at public hospitals is pro-poor. The serious supply-side problems in the health sector, which were already obvious before 1998, persist and in some areas have worsened. They affect access both to facilities (clinics and hospitals) and to skilled health workers in these facilities.

In 2018, Ministry of Health figures showed that of the 283,370 posyandu in the country, only 61% were active in the sense that they offered services to the public. But there was enormous variation across provinces, from 99% of posyandu being active in North Sulawesi to only 8% in Maluku. The posyandu were set up in the 1980s to be the village-level units offering midwifery services, baby measuring and weighing programs, and advice on nutrition to expectant and nursing mothers. The staff were unpaid, although they did receive some travel money. They were trained only in delivering basic disease prevention and primary care (Rokx, Subandoro, and Gallagher 2018, 53–9). They were linked to the PKK, a Soeharto-era family welfare institution designed to improve health and welfare in villages. The PKK groups were often headed by the wives of village officials, who did not always have much knowledge of primary health care. Nonetheless, the posyandu have been credited with some success in the reduction of some health problems, including stunting. After the AFC, there was a steep drop in the number of children attending the posyandu.

In 2018, there were on average 656 active posyandu per million people in Indonesia, but there was considerable variation across regions (table 7). There were slightly more than 17 posyandu per puskesmas in Indonesia, but again there was considerable variation across regions. Puskesmas staff are supposed to support and supervise posyandu in their activities, but in many cases they lack both the staff and the equipment to carry out their tasks. In Java–Bali, each puskesmas on average supervises over 30 active posyandu, which imposes a considerable burden on staff. In more remote rural areas, the ratio appears better, but distances are considerable and puskesmas staff often have neither the time nor the budget to visit posyandu in more remote villages on a regular basis. Between 2014 and 2017, an estimated 6,504 villages built new health clinics (polindes), which offer the services of midwives to local women in fixed premises, although it is unclear whether the advent of a new building improves the quality of service, or improves supervision by the puskesmas. But in spite of the ongoing supply-side problems in health provision, there have been some signs of progress, especially from 2013 to 2018. A key source of information has been the three rounds of the Basic Health Research survey (Riskesdas), conducted by the Ministry of Health in 2007, 2013, and 2018. These data show a modest improvement in the nutritional status of children under five, and a decline in stunting and wasting, although the proportion of stunted children was still 30% in 2018 (table 5). But other indicators were less encouraging. The figures on rates of immunisation for infants aged 12–23 months showed a slight fall in the percentage receiving all the recommended vaccinations between 2013 and 2018 (from 59% to 58%). The numbers of those not being vaccinated have stayed quite stable at a little over 9% between 2007 and 2018. A worrying trend concerns the prevalence of anaemia among pregnant women, which increased from 37% in 2013 to almost 49% in 2018. Although the government has tried to increase the supply of iron tablets available in puskesmas, the medication does not appear to be getting to the pregnant women who need it. Many aspects of the Indonesian diet are inadequate; for example, in 2018, 95.5% of the population were not getting the recommended five portions of fruit and vegetables per day. The high cost of fruit and vegetables relative to income is no doubt one reason for this.

Table 7. Number of Puskesmas and Posyandu per Million People by Region, 2018

Source: Kurniawan et al. (2018).

A striking feature of the Riskesdas data was the very considerable regional variations in many health indicators. The proportion of children under five who were stunted varied from 17.7% in Jakarta to 42.6% in East Nusa Tenggara. The percentage of those considered wasted varied from 4.6% in North Kalimantan to 14.4% in West Nusa Tenggara. The proportion of babies who had received all the recommended vaccinations varied from only 20% in Aceh to 90% in Bali. While there is a general tendency for health indicators to be worse in Eastern Indonesia than in Java–Bali and Sumatra, this is not always the case. A possible explanation for the low, and in some provinces falling, percentage of vaccinated children is an active campaign on social media promoting the view that vaccinations contain fluids derived from pigs. It is often very difficult for the government to counter such views, without offending Islamic opinion.

There are sound economic reasons why Indonesia should devote more resources to improving nutrition among the general population, especially among pregnant and nursing mothers and children under five. Hoddinott et al. (2013) construct what they term credible estimates of benefit–cost ratios for a plausible set of nutritional interventions aimed especially at the reduction of stunting in children. They obtained very high estimates for several countries in South and Southeast Asia; in Indonesia the ratio was 48, and in the Philippines it was 44. Although these estimates can be criticised, they do suggest that there has been significant underinvestment in improving nutrition in many developing countries, with Indonesia being a striking example.¹⁶ The key problem for Indonesian policymakers will be to design a package of nutritional and other interventions that will achieve improvements in the health of young children, at a reasonable cost.

The Jokowi government has initiated a national strategy to combat child stunting (StraNas Stunting), which has prioritised 100 districts across the country where the problem appears most severe.¹⁷ A major government concern is that multiple government programs to expand access to basic services on maternal and child health, parenting and nutrition counselling, water, and sanitation, as well as to provide social protection, are still not reaching many households across the country (Rokx, Subandoro, and Gallagher 2018, 59). Simply increasing the health budget, or the social protection budget, may not in itself be sufficient to solve the problem of coordination across several central government departments and many district-and village-level administrations. Indeed, it could make the problem worse.

Starting in 2018, more than 3,000 human development workers will operate in five pilot districts to coordinate programs at the village level, and to support posyandu in implementing StraNas Stunting programs. These initiatives are likely to be scaled up in the next five years.

CONCLUSIONS

Over the past two decades, a number of social protection policies have been implemented in Indonesia, including the distribution of subsidised rice, cash transfers (unconditional and conditional), health cards, and grants to assist children to stay in school. In addition, subsidies have been used to lower the prices of LPG, electricity, and fertiliser. The total cost of these policies has risen rapidly both in absolute terms and relative to GDP. But there are concerns about targeting. The government has devoted considerable resources to the construction of the UDB, which provides comprehensive data on the income and assets of close to 100 million Indonesians, in order to target social assistance programs more accurately. But it appears that a minority of local governments have been slow in updating the information (see box).

Some research has shown that the beneficiaries of the various social protection programs have been able to access health care and keep their children in school, although it could be argued that other policies, including the reduction or elimination of quantitative controls on food imports, could have a greater impact on the incomes of the poor. The high cost of many basic foods in Indonesia, relative to import prices, is one reason for poor nutrition among children under five and many adults, including pregnant and nursing mothers. But there are other reasons for Indonesia’s rather weak performance on a number of health indicators. The supply of health workers (doctors, nurses, and midwives) has grown over the past two decades, but many have opted for private practice, and government facilities are often understaffed and unable to offer a full range of treatments. Government spending on health has increased as a share of the government budget over the past decade but is still low relative to GDP. The decentralisation of health services to the districts does not appear to have improved the range of services available to many Indonesians, especially in Eastern Indonesia. There are often marked disparities between provinces in both maternal and child health, which government policies at both central and local levels have yet to address.

Notes

1 The sample of 300,000 households is now drawn from the 800,000 census blocks in Indonesia, which are divided into three groups according to a wealth index compiled by the BPS methodology division. But it is probable that the upper-income group contains many households that are not in the top expenditure decile.

2 This change was probably the reason for the jump in inequality between 2010 and 2011 (World Bank 2016, 41).

3 A discussion of the disparity between national accounts data and those from household surveys is given in Deaton (2010, 190–220). He examines the disparity in the context of India and China but not Indonesia.

4 It is important to stress that this drop was in nominal terms; the data were not corrected for inflation, which may have adversely impacted poorer groups more than the rest. One study that attempted to correct the data for inflation between 1997 and 1999 using the 100 Village Survey found that various measures of inequality increased by up to 23% over these two years, once the impact of inflation was considered (Skoufias, Suryahadi, and Sumarto 1999, table 4).

5 There are other issues with the poverty line in Indonesia that we cannot examine in detail in this Survey. Priebe (2014) has documented the many changes in the BPS methodology for estimating poverty since 1984; he argues that the methodology has been consistent only since 2007. Booth (2019, table 4) shows that in 2011 the BPS poverty line was rather low in comparison with other countries in Southeast Asia, including the Philippines and Vietnam.

6 By the end of the Soeharto era, around 40% of civil servants were teachers or workers in the education sector (Booth 2016, table 9.5).

7 In both mathematics and reading, Indonesian students’ PISA results fall below the global 25^th percentile average, although they improved between 2003 and 2015. But it should be noted that as more low-and middle-income countries join the PISA tests, the average score of the lowest 25% drops (Kurniawati et al. 2019, figures 10.2–3).

8 A useful summary of the social protection policies introduced between 1998 and 2014 is given by Tohari, Rammohan, and Parsons (2017, 1–7).

9 This is slightly lower than the figure given by McCarthy and Sumarto (2019). The programs in table 4 were selected from the following budget categories: social assistance, government assistance, and subsidies. Programs such as Pamsimas, the community-based water supply and sanitation program partly funded by the World Bank and implemented by the Ministry of Public Works, were excluded on the grounds that they targeted areas rather than people. Between 2008 and 2015, Pansimas provided around eight million people with access to improved water facilities and 7.7 million with improved sanitation in 10,287 villages (World Bank 2017, 25).

10 Estimates indicate that the cost of the LPG subsidy will increase to Rp 69–73 trillion in 2019. Discussions are ongoing within government on how this can be reduced.

11 In fact, the termination of Rastra may be delayed, as the national logistics agency Bulog still holds large stocks of rice that it argues can be more quickly run down through Rastra than through the BPNT program.

12 The Healthy Indonesia Program targets not only people living below the official poverty line but also those considered vulnerable. It covered about 35% of the population in 2017.

13 Proxy means testing involves allocating transfers, or other government grants, on the basis of household scores based on easily observed household characteristics such as quality of housing or possession of consumer durables. These are assumed to be adequate proxies for the actual income or expenditure of households. Community-based targeting is based on local knowledge, but this is often provided by local officials who may have their own agendas when it comes to the distribution of government grants (Ravallion 2016, 560–65).

14 A joint study by BPS and the Japan International Cooperation Agency in 1998 found that the harvested area was over-reported by 17.1%. The problem seems to have worsened since then. BPS estimated that the harvested area of rice in 2018 was 10.9 million hectares. In 2017, the Ministry of Agriculture figures gave the total harvested area of rice as 15.7 million hectares, or 44% higher. It has been suggested that one reason for the over-statement of harvested area is that farmers think they will receive more subsidised fertiliser if they claim they are planting more rice than is actually the case.

15 Using a CGE model, Yusuf and Warr (2018, 149) find that increasing protection in the food crop sector does increase poverty, but the effect is small. The effect is greater in urban than in rural areas.

16 A detailed assessment of the problems in estimating the benefits of nutritional interventions can be found in Alderman, Behrman, and Puett (2017).

17 The number of districts covered by the program rose to 160 in 2019 and is projected to increase to 260 in 2020. By 2023, there will be full national coverage of the program.