Reforming the Social Safety Net1
A. Introduction—The Landscape of Social Safety Nets in Indonesia
1. Over the past decade, Indonesia has made significant progress in a short time toward a comprehensive and adequate social safety net, yet challenges remain, as the country continues to expand the net to reach more of the country’s poor and vulnerable households. The importance of a social safety net to protect the poor and vulnerable is enshrined in the Indonesian constitution and has been recognized continuously by government administrations, in particular the current administration pays great attention to the social safety net in main government planning documents and subsequent budget allocations.2
2. Indonesia’s social safety net is made up of an array of household targeted social assistance programs. These programs have been at the forefront of the central government’s efforts to reduce poverty and inequality while promoting inclusive growth. Yet, the current social safety net is lacking in coverage and adequacy, as a large share of poor and vulnerable Indonesian households are not receiving all the programs they are eligible for. At the same time, the programs may fall short of providing the appropriate benefit for low-income households. Spending on social assistance could be more effective by investing in the most effective targeted programs while reforming existing universal subsidies, which is currently underway.
3. This note will discuss the state of and recommendations for improving Indonesia’s current social safety net as follows. The remainder of this section will discuss the landscape of social protection, noting the larger context in which the programs are set. Section B focuses on the ongoing efforts to improve the social safety net, detailing first the need for greater program integration, coordination and improved targeting then turning to analyze Indonesia’s main household targeted social assistance programs in brief. The programs discussed in this note include a health insurance fee waiver (JKN-PBI), a rice subsidy (Rastra), a scholarship (PIP), a temporary unconditional cash transfer (BLSM), and a conditional cash transfer (PKH). Finally, in Section C, new programs are proposed to address gaps in the net and main recommendations to improve the social safety net are developed.
4. As the Indonesian economy began to recover from the 1998 financial crisis, poverty declined markedly.3 Since then, poverty has fallen at a slowing significantly (Figure 1). However, since 2010 the rate of poverty reduction has slowed down, while in the same year inequality increased markedly, from 38 to 41 Gini points in 2014. Recently, inequality has reportedly fallen to 39 Gini points.4 Most of the reductions in the early 2000s can be attributed to continued pro-poor economic growth following the financial crisis and macro-economic stabilization policies that brought down the price of important staple goods such as rice, which make up a large part of the consumption of the poor. In the later part of that decade and well into this one, the declining poverty (albeit at a slower rate) rate is mainly attributable to initial and sustained nominal investments into social assistance programs. Generally, reducing poverty below 10 percent is difficult because the poor are increasingly further beneath the poverty line. However, further investment in and consolidation of the country’s main social assistance programs is one of the more effective ways to reduce poverty in the short term while promoting better human development in the longer term.
Figure 1.Poverty and Inequality Trends, 2000–16
5. Indonesia’s investment into social assistance programs is relatively new. While national government spending on household targeted Social Assistance programs has increased markedly in nominal terms over the past decade, it has remained within the range of 0.4 percent to 0.8 percent of GDP; low in comparison to other low and middle income countries (Figures 2 and 3).5 Spikes in expenditure are driven by the launching of an unconditional cash transfer used to mitigate the impact of a reduction in the fuel subsidy. Key programs promoting poor and vulnerable household access to education and health services were expanded in 2013 (Scholarship for poor students—PIP) and 2014 (Social health insurance fee waiver, JKN-PBI).
Figure 2.Evolution of Five Main Social Assistance Programs, 2004–16
Sources: 2004-2010 are from 2012 Indonesia Social Assistance Public Expenditure Review: 2011-2016 are revised Budget from Ministry of Finance, Financial Note and BARRENAS. 2016 World Bank estimated nominal GDP.
Figure 3.Social Assistance Spending by Regions, Income Levels, and Selected Countries 1/
Source: World Bank, Aspire database.
1/ For the categories of regions and income levels, the value shown represents a 2008-2014 average. 2016 Indonesia including only planned central government expenditures.
6. Increases in social assistance spending have not been efficiently allocated. Some of the important programs are in place to address economic and social risks faced by poor and vulnerable households (Figure 2). Yet, main risks remain unaddressed and the increase in nominal spending has not been directed to the programs most effective at reducing poverty and inequality (Figure 4). Based on a recently published study on the distributional impact of fiscal policy in Indonesia,6 direct transfers are most effective at reducing poverty and inequality, yet very little is spent on these (PKH and PIP). Spending on health and education in 2012 was found to be two-thirds and one-third as effective as direct transfers while also receiving a relatively small budget yielding little overall impact as well. While they are being reined in further in 2016–17, energy subsidies comprised 3.7 percent of 2012 GDP and were found to be the least effective in overall terms.7 The proposed reforms discussed in this brief encompass continued coverage expansion of the more effective social assistance programs while further improving inter-ministerial coordination while improving targeting methods and individual program business processes.8
Figure 4.Effectiveness Index and Spending
Sources: World Bank, The Distributional Impact of Fiscal Poticy in Indonesia, 2016.
1/ Delined as how mucha program reduces inequality, and therewith poverty, divided by the total budget spent Direct transfers lnclude PKH and BSM (PIP).
B. Ongoing Efforts and Reforms to Improve Effectiveness
Continued to Improve Targeting, Program Integration and Coordination
7. Targeting accuracy of the main social assistance programs has improved over time, but further reform is needed. The main factor that determines effectiveness in reducing poverty and inequality, is a programs’ targeting accuracy. Accuracy can be measured most simply by estimating beneficiary incidence and targeting errors or how much of the total beneficiaries are fall into different groups of the population and how many of the beneficiaries are correctly targeted. Figure 5 displays the benefit incidence of the five main household targeted social assistance programs. Target levels and their corresponding size vary from the poorest consumption decile (PKH), to the poorest 40 percent (JKN-PBI). Having the lowest target level; PKH manages to allocate just under 80 percent of its benefits to the poor and vulnerable. BLSM and PIP also allocate more than 60 percent of their benefits to these groups. Larger programs such as PBI and Rastra have lower exclusion errors but higher inclusion errors; smaller programs such as PKH will have higher exclusion errors but low inclusion errors as they generally do not go to richer households.
Figure 5.Comparing Program Incidence by Class, 2014-2015
Source: Susenas 2015 (2014 for PKH).
8. Further welfare gains could be achieved through developing a more dynamic targeting system. Currently, Indonesia’s main social assistance programs are targeted using the Unified Database (UDB), most recently updated and expanded in 2015 and based on databases established in 2005, 2008 and 2012.9 The UDB was updated in 2015 to contain 26 million households (just less than 40 percent of all households). The registry of poor and vulnerable households hosts a range of information that is used, together with several other surveys, to rank households by predicted consumption using a Proxy Mean Test (PMT). While mean testing provides more accurate targeting, the implementation of means testing to inform social assistance targeting is a long-term recommendation due to the lack of reliable tax and/or income data. To reduce targeting errors and raise social assistance equity, a dynamic targeting system such as the ones in place in Chile and the Philippines is recommended and a pilot will be implemented in 2017. In such a system, citizens that feel unjustifiably excluded or have never been surveyed can ask to be surveyed and potentially included in a range of social assistance programs targeted at different groups of the population.
9. Thus, developing a more dynamic social registry would help reduce exclusion and inclusion errors. Besides promoting further coverage expansion for programs such as PKH and PBI-JKN, allowing two way updating of the UDB and encouraging continued social assistance program use of it at national and local levels will be key to improving targeting outcomes. This will be crucial to better allocate government spending to reduce poverty and inequality through the development of a more effective social safety net. As noted in previous reviews published by the World Bank, the government should expand the social safety in coverage but also in terms of providing a framework for integration through common standards in delivery processes like targeting and beneficiary identification; socialization, outreach, and enrollment; monitoring and evaluation; and grievance recording, feedback, and remediation.10
10. Households as well as agencies and ministries would be well-served by better coordination between and integration of the existing social assistance programs. Uniting the existing set of independently-operating programs and their implementing agencies via common minimum standards going beyond targeting alone—could provide a boost to consumption expenditure equal to between 14 percent and 21 percent of an average targeted household’s budget and would have an immediate impact on poverty. The government would also experience efficiency gains from integration if the provision of all common program sub-processes mentioned above were rationalized by eliminating the duplication that is currently pervasive.
11. The existing array of main social assistance programs target the same range of poor income households, but integration at the household level is lacking. For instance, in 2015 only 10 percent of the poorest households are estimated to be receiving both PBI-PBI and PIP. Similarly, only 29 percent of the poorest decile receive Rastra, and JKN-PBI assistance fell from 43 percent in 2014 to 29 percent of the total population in early 2015, likely driven by a decrease in Rastra coverage. Although the majority of poorest households are eligible to receive all three of the aforementioned programs, only about 8 percent do.11 The “overnight” reduction in the headcount poverty rate that would result from social assistance program integration (estimated for PKH, PIP and Rastra) is expected to be 2 percentage points to 4 percentage points.12 For instance, extending a social assistance package that combines the three current direct cash or near-cash transfers in one to 10 million households would create a benefit with a magnitude similar to that in countries where direct transfers reduce poverty without distorting labor market decisions.13
Expand Coverage, Adequacy and Efficiency of Key Social Assistance Programs
12. This section briefly describes the main five social assistance programs with references on targeting performance (see Figure 5) and specific reforms proposed by program. Table 1 provides an overview of total central government expenditure and targeted beneficiaries since 2004.
|Total Central Government Expenditure||2004||2006||2008||2010||2012||2014||2016|
(In millions of households and individuals 1/)
|Targeted Beneficiaries (In millions)||2004||2006||2008||2010||2012||2014||2016|
HH stands for households; Ind for individuals.
HH stands for households; Ind for individuals.
Penerima Bantuan Luran—Jaminan Kesehatan Nasional—PBI-JKN
13. PBI, previously known as Jamkemas is a health insurance fee waiver and has accomplished major coverage increases while it has successfully been absorbed by JKN. Managed by The Ministry of Health and BPJS Health, PBI-JKN is the largest single source of health insurance coverage in Indonesia, covering over 92 million individuals (approximately one-third of the population) in 2016 at the cost of Rp 25.5 trillion. PBI’s value to households is potentially significant—it promises a nearly unlimited-in-value health benefit to poor and vulnerable households—so, on paper PBI has become the largest social assistance transfer.14
14. PBI-JKN targeting outcomes for the poor and vulnerable populations are relatively good but appear to have slightly worsened.15 This trend is most likely due to the coverage expansion as well as due to the fact that although PBIs’ initial quota was generated by querying a household list containing some socio-economic and demographic information, those households given Jamkesmas cards in 2012 were identified by locally-based Ministry of Health staff, service provider staff, and local government with varying eligibility criteria. Furthermore, there were issues with beneficiary lists as Jamkesmas merged into JKN-PBI in early 2014, and beneficiaries were reportedly often not aware of the extent of benefits they would receive in either Jamkesmas or Jamkesda (local government variant of Jamkesmas).
15. Several reforms are needed to improve the adequacy of the PBI component of JKN. While utilization is not significantly different from other forms of health insurance in Indonesia, it could be increased by improving outreach, facilitation and beneficiary support. In particular, promoting greater dissemination of information to beneficiaries about the scope of PBI-JKN health care coverage and the standards of practice they should be expecting would improve utilization. In tandem, the Government should merge PBI monitoring of the healthcare service sector with that of the JKN system.
16. Rastra, the second largest social assistance initiative in terms of coverage—has positive potential but in its operation fails to achieve fundamental social assistance goals. Rastra, previously known as Raskin, is a subsidized rice delivery program implemented by the Ministry of Social Affairs and Bulog, the State Logistics agency. In 2016, Rastra targeted 15.5 households and program expenditures amounted to Rp 22.5 trillion.16 The consistent provision of a basic food package could protect poor households from food-price volatility, calorie scarcity and malnutrition. However, Rastra suffers from dilution of benefits and inclusion errors, missing rice, and hidden financing burdens, all of which reduce the transfer values provided to target households.17 Poor targeting, dilution of benefits, and missing rice are long-standing and well-known Rastra issues.
17. Rastra remains the least well targeted of any Indonesia’s SA programs and the average benefit package is significantly diluted when the right to buy Rastra rice is re-allocated at the village level to include many non-poor households. Rastra actual coverage is estimated to far exceed the targeted level of 15.5 million households (23 percent of the population in 2015): according to survey data, 28.6 million households, just over 40 percent of the total population, are receiving Rastra, with only 57 percent of beneficiaries being in the poorest 40 percent. According to the latest available Susenas data, households are receiving only 4 kilograms per household per month, 11 kilograms less than households should be receiving.18 In addition, a lack of clarity concerning responsibilities and financing at the “last mile” of Rastra distribution means that Rastra-purchasing households—especially those in remote areas—receive a lower per-kilogram benefit than promised.
18. To begin addressing over coverage and an overall inadequate benefit package, the Rastra program should base allocations exclusively on the UDB and monitor allocations and ‘last-mile’ rice delivery with a stronger oversight system. Furthermore, the program should continue experimenting with different methods of allocating Rastra, in-kind, in cash or in a monetized transfer to be spent only on Rastra rice or other staple goods. The Government is planning to implement an e-voucher reform initiative in 44 cities in early 2017. The new delivery mechanism called E-Warong aims to leverage existing market players to deliver Rastra benefits but need close monitoring and evaluation as it scales up.
Program Indonesia Pintar—PIP
19. PIP, Indonesia’s’ scholarship program has matured and has begun to demonstrate its full potential in recent years but can still deliver more to those most in need. PIP, previously referred to as BSM, is a scholarship program for poor students implemented by the Ministry of Education and The Ministry of Religious affairs. PIP targeted 19 million students in 2016 at a budget of Rp 14 trillion. With recent increases in coverage and reforms to implementation, PIP is now making significant positive contributions to welfare in poor and near-poor households (with students) and to the Government’s drive to provide universal basic education.
20. 2015 PIP targeting outcomes have shown improvement over 2013 BSM levels. Beneficiary incidence change shows that while incidence decreased slightly for the poorest decile (from 22 percent to 20 percent in 2015) increased for the vulnerable (42 percent to 47 percent) and decreased for the non-poor (18 percent to 14 percent).19 This change is also reflected in the decline in exclusion and inclusion errors at all levels. That said, exclusion errors are still quite high and likely due to low program uptake rates as students living in households in the poorest 25 percent of the population are technically eligible and may receive PIP if they present a KPS or KKS card (social protection card sent to 15.5 million households).
21. To improve the program further, PIP should focus on continuous and coordinated monitoring, evaluation, and improvements in delivery. A key obstacle is the institutional fragmentation within the two Ministries that oversee the programs’ implementation. A reform that is well within reach and long overdue is, benefit-level updating which should occur more frequently in order for the transfer to remain relevant. PIP should be at the forefront of positive outreach to poor students, especially those approaching the senior secondary or university levels and facing the highest out of pocket and opportunity costs. To reach that goal, outreach facilitation could be developed further with the creation of beneficiary support modules for senior secondary and university level drop-outs.20
Bantuan Langsung Sementara Masyarakat—BLSM
22. BLSM, Indonesia’s Unconditional Cash Transfer (UCT), has a clear objective: temporarily supplement consumption for poor households facing anticipated, policy-based price increases. BLSM, previously called BLT, is a program implemented by the Ministry of Social Affairs. The payment targeted the poorest 25 percent of households (15.8 million in 2015) and provided monthly payments of Rp 150,000 for seven and six months respectively for each tranche. The total budget disbursed in 2015 amounted to Rp 14.6 trillion. Over the years, the government has reduced existing fuel subsidies and compensated poor and near-poor households for the subsequent rise in fuel, food, and transport prices with a temporary unconditional cash transfer. Research has shown that the BLT UCT was effective in preventing an increase in poverty due to increased fuel prices; instead of rising the poverty rate fell by about 1 percentage point. Furthermore, there is no evidence that UCT or conditional cash transfers (CCT) affect labor supply or generate dependency. Heads of households were not likely to leave work due to receiving the BLT transfer, which comprised just about 15 percent of household monthly expenditure.21
23. BLSM has better targeting outcomes when comparing to other programs of similar size (Rastra targeting 15.5 million households, just under 25 percent of the population and PBI targeting 88.2 million people or approximately 23 percent of the population). A reason for this may be that BLSM is a more direct and simple transfer program than the other Social Assistance programs and so suffers less from implementation issues in using the UDB lists. In 2013, incidence to the target group, the poorest 25 percent of households, is around 45 percent and just about the same as with the 2014 BLSM round (Figure 5).
Program Keluarga Harapan—PKH
24. PKH is Indonesia’s CCT program that is implemented by the Ministry of Social Affairs. In 2016, the program is slated to scale up to reach 6 million families in all districts of Indonesia at a budget of Rp 9 trillion. PKH has generated significant and positive impacts in welfare, health, education and nutrition seeking behavior.22 Proven impacts in behavioral change translate into long run development impacts as educational and health attainment through improved nutrition of young children.
25. PKH is a relatively well targeted program. Based on the latest available data in 2014, the poorest 10 percent of families received over one third of the benefits available; the bottom 20 percent received over half of the benefits available; and the bottom 30 percent received over two-thirds of the benefits available. There are likely many factors that together lead to better targeting results in PKH including the CCT’s early adoption of the UDB-based beneficiary identification and verification system that includes two-way dynamic updating of program participants and eligibility status. It must be kept in mind however, that PKH is a small program targeted at the poorest 10 percent of the population and so it is expected that leakage to the non-poor is minimal: inclusion errors are very low but exclusion errors are very high as a result.
26. To strengthen the program and prepare for further scale up, the program should continue strengthening institutional capacity, its IT and HR systems and the capacity of affiliated service providers (a task requiring higher level coordination). Continuous enhancement of core program functions is essential for efficient delivery of benefits and effective access for households: timely verification of beneficiaries’ status and conditionality fulfillment; regular MIS updating, adjustment of benefit levels and timely disbursements; determination of local level capacity for distributing benefits; as well as suggestions for remediation of local supply inadequacies in health, education and program socialization are some of the aspects that need strengthening.
C. Looking Ahead: Building a More Comprehensive Social Safety
27. At the moment, not all important household risks to well-being are covered. For instance, programs to address old age and disability exist but are low in coverage (ASLUT and ASODKB).23 In addition, there is no social assistance program to foster poor household low-cost access to early child hood education and development services (ECED). Such a program would stimulate learning and social interaction at an early age while allowing mothers to seek employment at the same time. In addition, an adequate social safety net should provide active support to individuals and households moving from a state of vulnerability and dependence to one of independence and resilience through livelihood and labor market activation initiatives. While Indonesia has some of these programs (some under the heading of the Sustainable Livelihoods initiative P2B, launched in 2015), they are scattered over many Ministries and it is unclear which of these are effective and able to be scaled up. An actionable reform would be to evaluate, consolidate where appropriate and promote scale up existing programs.
28. Finally, Indonesia’s nascent Crisis Monitoring and Response System (CMRS) should be fully operational and automatically activated as part of the social safety net when needed. Households in Indonesia are vulnerable to stresses that the international and national economies inevitably produce, and there is as yet no automatic response mechanism providing social and economic support during times of uncertain outlook. A functioning monitoring system is already in place, but it depends upon timely, high quality data inputs from across several government agencies. Response protocols are needed so that programs can be automatically funded, activated, and implemented when needed and so that budgetary and parliamentary procedure does not prevent timely assistance from being released.
29. Indonesia has made significant progress towards the creation of a holistic social safety net. Key programs are in place to protect against economic and social risks faced by poor and vulnerable populations. To improve the adequate, accurate and timely allocation of social assistance to eligible households, some of these programs will need a higher budget allocation and reforms in the sense of increased coverage (PKH), increased benefit levels (PKH and PIP), revised delivery system (Rastra) and standardized systems to conduct monitoring and evaluation, grievance redress, socialization and feedback (all). Tying all programs together to reach the same poor and vulnerable population will require significantly more cross ministerial coordination and the standardized use of the UDB as well as consolidated progress towards a more dynamic and open targeting system, allowing for greater social inclusion.
30. While some reforms are already underway, others are left at the way side. Indeed, some important risks such as disability, old age, inadequate skills to enter the labor force, stunting, and lack of access to early child hood education remain unaddressed in a systematic way. To spend better, social assistance spending should continue move away from universal subsidies toward targeted programs. To begin protecting better, Indonesia should consider the expansion or integration of existing smaller programs as well as creation of new programs. If gaps are left unaddressed or the social net incomplete, these risks hinder the potential of Indonesia’s population to become a more prosperous and equal society.
Prepared by Juul Pinxten, Pablo Ariel Acosta and Changqing Sun (all World Bank, Jakarta).
See Indonesia, National Medium-Term Development Plan 2010–2014 and 2015–2019.
Prior to the financial crises in 1998, Indonesia managed to halve poverty from 40 percent to 18 percent in the period between 1976 and 1996; one of the most pro-poor growth periods in the economic history of any country. World Bank, 2006, Making the New Indonesia Work for the Poor.
Statistics Indonesia, multiple years.
This note refers to the five main social assistance programs and that are centrally executed and household targeted (as referred to in Figures 2 and 3). While ASLUT, JSPACA/ASODKB and PKSA calculate into the total central government expenditure on household targeted SA as a share of GDP, they are very small programs and thus not included in the discussion.
See World Bank, 2016a, The Distributional Impact of Fiscal Policy in Indonesia for more detail. Effectiveness here is defined as how much a program reduces inequality, and therewith poverty, divided by the total budget spent. Direct transfers include PKH and PIP (previously called BSM).
This note is based on a forthcoming Indonesia Social Assistance Public Expenditure Review Update (2017) and several other publications by the World Bank and other institutions.
See TNP2K, 2015, Indonesia’s Unified Database for Social Protection Programs for more detail.
World Bank, 2012, Indonesia Social Assistance Public Expenditure Review and 2016 update.
For reference, if actual headcount poverty continues to fall at the rate experienced between 2013 and 2014 (approximately one-tenth to one-fifth of a percentage point per year), it would take approximately 10 years to achieve the “overnight” reduction that the least expensive integration scenario achieves immediately, see World Bank (2016b) for details.
In Philippines, the transfer of the conditional cash transfer program represents 21 percent of the average income of the poor. In Mexico and Colombia, the transfers of conditional cash transfer programs range between 21 percent to 25 percent of average consumption of target groups.
Meanwhile, in the new National Health Insurance scheme (Jaminan Kesehatan Nasional, or JKN), the premiums that the government contributes on behalf of Jamkesmas card holders account for a large share of all JKN premiums currently collected. In other words, without Jamkesmas beneficiaries, JKN would cover a far smaller share of the population and collect much less in revenue.
Statistics Indonesia Susenas 2010, 2013, and 2015; and World Bank, 2016c, Benefit Incidence of Main Social Assistance Programs in Indonesia. Unpublished note provided to Ministry of Finance.
Revised central government budget from The Ministry of Finance, Financial Note, 2015.
World Bank, 2016b, Indonesia Social Assistance Public Expenditure Review (forthcoming 2017).
A main reason for low program adequacy is that large proportions of rice procured for Rastra do not reach localities and no extra effort is made to put Rastra rice in targeted households when total supplies are low. This is due in part to a lack of clarity concerning responsibilities and financing at the “last mile” of Rastra distribution means that Rastra-purchasing households—especially those in remote areas—receive a lower per-kilogram benefit than promised.
Statistics Indonesia Susenas 2013 and 2015; and World Bank (2016c) Benefit Incidence of Main Social Assistance Programs in Indonesia. Unpublished note provided to Ministry of Finance.
World Bank, 2016b, Indonesia Social Assistance Public Expenditure Review (forthcoming).
World Bank, 2013, Indonesia Economic Quarterly Adjusting to Pressures; and JPAL, 2016, The Impact of Cash Transfers on Labor Supply, presentation at JPAL SEA Policy Conference.
TNP2K, 2016, Evaluating Longer-Term Impact of Indonesia’s CCT Program: Evidence from a Randomized Control Trial, forthcoming; and World Bank, 2011, Main Findings from the Impact Evaluation of Indonesia’s Pilot Household Conditional Cash Transfer Program. The midline evaluation demonstrated that PKH was responsible for statistically significant increases in pre-natal care. The likelihood of attending at least four prenatal visits: increased by 9 percentage points while newborn delivery at a facility or attended by a professional increased by 5 percentage points. Post-natal care improved by almost 10 percentage points while, immunizations, and growth monitoring check-ups increased by 3 percentage points and 22 percentage points respectively.
These programs are being merged into PKH as of late 2016 and may be scaled up within that program.