67308 SOCIAL ASSISTANCE PROGRAM AND PUBLIC EXPENDITURE REVIEW 3 THE WORLD BANK, JAKARTA OFFICE Indonesia Stock Exchange Building Tower II/12th Floor Jl. Jend. Sudirman Kav. 52-53 Jakarta 12910 Tel: (6221) 5299-3000 Fax: (6221) 5299-3111 Website: www.worldbank.org/id THE WORLD BANK 1818 H Street N.W. Washington, D.C. 20433, U.S.A. Tel: (202) 458-1876 Fax: (202) 522-1557/1560 Website: www.worldbank.org Printed in February 2012. Cover photograph and photograph on page 5 provided by Anne Cecile Esteve/Matahati Productions/World Bank. Photograph on page 9 provided by Bimo Nurendro. Copyright protection and all other rights reserved. Designed by Hasbi Akhir (hasbi@aisukenet.com) The Social Assistance Program and Public Expenditure Review policy notes 1 through 8 together comprise Volume 2 of Protecting Poor and Vulnerable Households in Indonesia report. Both the report and the policy notes are products of the World Bank. The findings, interpretations, and conclusions expressed herein do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the Governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. For any questions regarding this report, please contact Vivi Alatas (valatas@worldbank.org) or Jon Jellema (jjellema@worldbank.org). RASKIN SUBSIDIZED RICE DELIVERY SOCIAL ASSISTANCE PROGRAM AND PUBLIC EXPENDITURE REVIEW 3 RASKIN Subsidized Rice Delivery Table of Contents List of Abbreviations, Acronyms and Indonesian Terms 3 Executive Summary 4 1. Background 8 2. Objectives, Program Size, and Benefit Adequacy 11 3. Targeting 15 4. Potential Impacts 20 5. Cost Effectiveness 23 6. Implementation 24 7. Budget Formulation and Measures of Program Cost 25 8. Summary and Recommendations 27 References 28 List of Figures Figure 1: Food Security in Indonesia 10 Figure 2: Raskin Expenditure and Coverage, 2005-2009 11 Figure 3: Raskin Revenue Inflows and Cost Outflows 13 Figure 4. Raskin Coverage and Purchases by subgroup and decile, 2010 17 Figure 5: Food Security vs. Raskin Quota 18 Figure 6: Poor Household Raskin Purchases vs. Raskin Concentrations 19 Figure 7: Raskin Procurement and Raskin Purchases, 2006-2009 20 Figure 8: Raskin Official and Actual Benefits, various years 21 List of Tables Table 1: Raskin at a Glance 12 Table 2: Raskin Expenditure Summary, 2005-2009 14 Table 3: Characteristics within Indonesian Populations, 2009 16 Table 4: Characteristics within Raskin and non-Raskin Populations, 2009 16 Table 5: Spending Efficiency Indicators, 2005-2009 23 List of Boxes Box 1: Raskin budget formulation, revenues, and measures of program cost 25 2 List of Abbreviations, Acronyms and Indonesian Terms AFC Asian Financial Crisis APBN Anggaran Pendapatan dan Belanja Negara (Central Government Budget) BKKBN Badan Koordinasi Keluarga Berencana Nasional (Family Planning Coordination Agency) BLT Bantuan Langsung Tunai (Unconditional cash transfer) bn Billion BPS Badan Pusat Statistik (Indonesian Central Statistics Agency - Statistics Indonesia) BSM Bantuan Siswa Miskin (Cash transfer for poor student) Bulog Badan Urusan Logistik (National Logistics Agency) Desa Village FSVA Atlas Kerentanan dan Ketahanan Pangan (Food Security and Vulnerability Atlas) GDP Gross Domestic Product GOI Government of Indonesia IFLS Indonesian Family Life Survey Kelurahan Urban precinct Kemenkes Kementerian Kesehatan (Ministry of Health) Kemenkeu Kementerian Keuangan (Ministry of Finance) Kemenkokesra Kementrian Koordinator Kesejahteraan Rakyat (Coordinating Ministry for Social Welfare) KemenkomInfo Kementrian Komunikasi dan Informatika (Ministry of Communications and Information Technology) Kota Autonomous municipality LHS Left hand side (of graph) M&E Monitoring and evaluation MDG Millennium Development Goal(s) NGO Non-governmental Organization OPK Operasi Pasar Khusus (special market operation for rice) PKH Program Keluarga Harapan (Conditional cash transfer) Podes Potensi Desa (Village Potential Survey) PPLS Pendataan Program Perlindungan Sosial (Data collection for targeting social protection programs) PSE Pendataan Sosial Ekonomi Penduduk (Socio-economic Population Survey) Raskin Beras untuk Rakyat Miskin (Subsidized rice for the poor program) RHS Right hand side (of graph) Rp Indonesian Rupiah Susenas Survei Sosio-Ekonomi Nasional (National Socio-Economic Survey) UU Undang-Undang (Law) WFP United Nations World Food Program 3 Executive Summary The Beras untuk Rakyat Miskin (Raskin) program was introduced as an emergency food security program in 1998; it delivers rice to be purchased at subsidized prices, prioritized to poor and near-poor households. Raskin is the successor to Operasi Pasar Khusus (OPK), which itself was part of an emergency relief package delivered during the Asian Financial Crisis (AFC). Badan Urusan Logistik (Bulog, the National Logistics Agency) is responsible for purchasing rice from producers and delivering it as far as the over-50,000 distribution points spread throughout Indonesia. Households should be able to make monthly Raskin purchases from these distribution centers; some regions fund logistics operations to bring Raskin rice closer to households. The Coordinating Ministry for Social Welfare (Kementrian Koordinator Kesejahteraan Rakyat, Kemenkokesra) is in charge of planning and determining with Bulog the yearly allocations and prices of Raskin rice. In terms of government expenditure, Raskin remains the largest permanent social assistance transfer targeted to poor households in Indonesia. Though developed as a response to crisis, Raskin has become a permanent program and in real expenditure terms is one of the few social assistance programs with a larger budget in 2010 than in 2005. In 2010 Raskin accounted for nearly 53 percent of all household-targeted social assistance spending carried out by the central government. Over 2000 to 2010, the amount of rice allocated by the Raskin program has averaged over 2 million tons per year; in 2010 the almost 3 million tons allocated could have delivered between 30 and 40 kilograms per month to the approximately 6.2 million households at or below the poverty line. The distribution of Raskin rice does not closely align with the objectives laid out in program manuals and official documentation for at least three reasons. First, not all of the rice procured for the Raskin program makes it to households. In the three most recent years for which there is audited budget data (2007 through 2009), nationally representative household surveys indicate that only half (or less) of the rice procured for Raskin is purchased by households. The readily-available budget and administrative records cannot indicate where the bulk of this “missing” rice exits the delivery chain, and no single agency or authority is in charge of Raskin rice from procurement to household purchase. 4 Second, through Raskin coverage is higher among poorer households, coverage extends to nearly as many non-poor households as poor households. Roughly 50 percent of the entire Indonesian population1 buys Raskin rice at least once a year, making Raskin’s overall coverage 20 percentage points higher than the next largest permanent social assistance program. Raskin allocations to households are decided by local authorities which has resulted in frequent Raskin purchases by poor and non-poor households alike. In some regions a rotating schedule has been established to share Raskin allotments more equally among a greater number of households. These coverage increases (beyond the target population) have been common since Raskin’s inauguration (as OPK) in the late 1990s. Finally, as total Raskin rice is spread more thinly than is specified in Raskin guidelines, few households enjoy full Raskin benefits. In fact, the amount of Raskin rice purchased by a household is roughly constant across the entire consumption distribution, meaning non-poor households buy as much Raskin as poor, near-poor, or vulnerable households. By 2010 the average amount of Raskin rice bought was approximately 3.8 kilograms per month while total rice purchases (of any kind of rice) by poor households amount to nearly 35 kilograms per month while estimated rice needs for an Indonesian household are closer to 40 kilograms per month. As a result, the actual transfer received by households has fallen to about 2 percent of total poor household expenditure, while the kilograms purchased represent less than 10 percent of a household’s rice needs. The implicit value (to households) of Raskin transfers is equal to the difference between the local market price of rice and the official Raskin price times the number of kilos purchased. The price that Raskin buyers pay is often higher than the price stipulated by program manuals and public announcements. These deviations in benefit level and price have reduced the average value of the benefits provided by Raskin 2 to 3 percent of average poor household expenditure in 2010 (from approximately 10 percent in 1999). 1 237.6 million in 2010, as measured by the decadal census produced by Badan Pusat Statistik (BPS, Statistics Indonesia). 5 RASKIN Subsidized Rice Delivery Raskin’s does not target food insecure regions while extra Raskin rice is often not prioritized for poor households. Some regions in Indonesia are more food insecure than others; Eastern Indonesia in particular faces more difficulty in local production of sufficient calories for its population. Though Raskin could be allocated to enhance food security, Raskin quotas are unrelated (on average) to levels of food insecurity. In areas with more total Raskin available on the ground, the actual amount purchased by poor households is neither greater or nor less, indicating that actual Raskin allocations are not more pro-poor at higher coverage levels. This is a longstanding issue: in 2005, poor households were estimated to have received only 25 percent of total Raskin subsidies while the remaining 75 percent went to nonpoor households. Budgeted operating costs are higher than other social assistance programs but physical management of rice is costly, more so than non-perishable cash, for example. Raskin is delivered from procurement to distribution points by a state-owned company, Bulog, which receives a Raskin budget allocation from the central government. Budget allocations are formulaic and build in a pre-agreed level of expenditure for the administration and operation of Bulog’s portion of the Raskin program. These formulaic operational and management fees are noticeably higher than other national household- targeted social assistance programs in Indonesia, but on top of regular management and administration, Bulog must complete many more processes (related to the production, storage, and distribution of Raskin-ready rice) than other social assistance programs. Tabulations of additional revenue from the eventual sale of Raskin rice are not easy to make, so overall operational or administrative cost margins cannot be calculated with precision. However, when formulaic central government budget allocations for administration and operation are compared with actual benefits delivered, it is clear that Raskin consumes more resources in the production of a smaller de jure benefit. There are few incentives for local-level implementers who are responsible for making Raskin effective for vulnerable households. Kemenkokesra and Bulog together determine yearly allocations and prices and Bulog procures and delivers rice to distribution points; final delivery of Raskin to households is not managed or overseen by Bulog. From the distribution point forward, the implementation of the Raskin program is delegated to local administrations. Some socialization and monitoring and evaluation suggestions are made, but there are few incentives for local administrations to run Raskin efficiently or to protect Raskin quotas for targeted poor households only. Lack of oversight has led to the dilution of benefits and price manipulations mentioned above. Though processes for all support operations are described and suggested in the manual, there is a need for greater oversight to ensure they are done efficiently and a need for attention from policymakers to ensure these subprocesses are receiving the resources they need to effectively inform communities, implementers, and other stakeholders. Currently, even when monitoring is completed according to suggestions in manuals, the monitoring instruments do not provide a suitable foundation for evidence-based program reform. It is crucial that adequate instruments yield a flow of information and that results enter a program-wide monitoring, evaluation, and program improvement cycle. A substantial re-engineering of the Raskin process, together with improvements in oversight and more resources devoted to support operations, could allow Raskin to fulfill its role as regular consumption support for poor households, but without such attention, Raskin may continue to produce meager benefits at relatively high cost. If Raskin benefits as mandated reliably reached poor households, the Raskin program could deliver reasonable last- resort, in-kind income support to vulnerable households. Raskin implementation procedures, however, currently tolerate substantial losses of rice, substantial reductions in benefits transferred, wide variation in targeting procedures, and a lack of central-level oversight of the Raskin program past the distribution points. Changing benefit levels or lists of beneficiary households will not likely improve outcomes under such a regime; rather a complete re-engineering of the Raskin process may be necessary to turn Raskin into an effective social transfer. 6 7 1. Background Raskin has origins in the GOI’s response to the Asian Financial Crisis in the late 1990s, making it the longest- serving of the current household-targeted social assistance transfer programs. Originally called Operasi Pasar Khusus,2 Indonesia’s subsidized rice program first provided (in July 1998) food consumption assistance to households suffering the effects of the Asian Financial Crisis.3 Raskin rice stocks are acquired through the purchase of wholesale quantities from domestic producers at fixed prices4 by the state-owned Bureau of Logistics, Bulog, which then delivers agreed quotas of rice to over 50,000 regional distribution points where eligible households may purchase fixed quantities at below-market prices. Between 2005 and 2010 households could have purchased a maximum of 15 kilograms of Raskin rice per month at 75 to 80 percent lower than the market price.5 2 OPK was given the Raskin name in 2002; program goals and implementation stayed largely the same. 3 See “Social Assistance Progam and Public Expenditure Review 2: BLT” and the references therein for more on the household impacts (including on consumption) of, and household coping strategies during, the Asian Financial Crisis. 4 The price paid to producers is set by the Government of Indonesia. Over 2002 to 2009, Bulog’s procurement price ranged from Rp 2,470 to 4600 per kilogram. It is typically just over or just under the average “farmgate” producer price. 5 These are averages. Kilogram limits for households were at a high in 2005 (20 kilograms) and a low in 2007 (10 kilograms). In 2010, the kilogram limit was 15. In 2007, the Raskin per-kilogram price was Rp 1000 while market prices averaged Rp 5076; in 2010, Raskin prices rose to Rp 1600 and market prices to Rp 5775. 8 Across Indonesia, 54 percent of the population participates in the Raskin program while total Raskin expenditures are equivalent to a quarter of a percent of GDP. A long history combined with a lack of revision to beneficiary registries and redistribution of Raskin buying rights has led to over 50 percent of all Indonesians households buying Raskin rice in any given year. Raskin has become the largest of the permanent household-targeted social assistance programs by expenditure and the second-largest by official targeted coverage. Food availability and food prices remain important issues for poor households in Indonesia… For poor Indonesian households, food expenditures are two-thirds to three-quarters of all expenditures; the majority are net consumers (rather than net producers) of rice; and expensive high-protein items represent only about 10 percent of all household food expenditures. For all these reasons in-kind transfers of important basic commodities may be useful, especially in food insecure areas where regular markets cannot be relied on for a consistent supply of reasonably-priced foodstuffs. Elevated food prices, which still occur fairly regularly in Indonesia, present serious risks for poor households; the 2006 poverty increases were blamed partially on long episodes of elevated food prices. Consequently, a reliable source of below-market-price foodstuffs is valuable for all households. ...while some regions in Indonesia are not able to consistently produce enough calories for their local populations. The United Nations-World Food Programme (WFP) estimates that 13 percent of all districts in Indonesia are food insecure.6 Figure 1 below is a district-level summary showing higher levels of food insecurity in eastern Indonesia (especially Papua), the island districts off the western coast of Sumatra, remote districts in Sulawesi, and the more isolated northern and western districts in Kalimantan. 6 The food security concept and index developed by UN-WFP summarizes three dimensions of food (and calorie) scarcity: food availability, food access, and food utilization. There are 13 indicators that summarize levels and changes in these three broad categories. For example, a ratio of staple food consumption (to all food consumption) is a variable that tracks both food availability and food utilization. The poverty rate and infrastructure access (including electricity, clean water, and health facilities) are proxies for food access. Rainfall deviation, deforestation, and natural disaster frequency all provide proxies for food availability and food access. 9 RASKIN Subsidized Rice Delivery Figure 1: Food Security in Indonesia Source:adapted from A Food Security and Vulnerability Atlas of Indonesia (WFP, 2009) This note assesses the operation and implementation of the Raskin program to determine how well poor households are served by the program and the overall cost of program resources. The note provides quantitative analysis of the coverage, incidence, and average benefit levels of Raskin to determine both the progressivity of the program’s targeting and the adequacy of benefit levels. Qualitative information on program delivery and program operations will also shed light on areas for reform. An evidence-based appraisal of the household-based transfer currently consuming over 50 percent of the entire social assistance budget envelope can provide inputs to the Government of Indonesia (GOI) as it continues to try to achieve both Pro-Poor development for all Indonesians and the Milenium Development Goals (MDGs). 10 2. Objectives, Program Size, and Benefit Adequacy Raskin is an outsize item in the household social assistance budget. Benefits are meaningful on paper, but amounts arriving in poor households are much smaller and not likely to provide real relief. The Raskin program delivers subsidized rice for purchase by eligible households at distribution points throughout Indonesia. The mandated monthly allotments have fluctuated since Raskin’s introduction (as OPK) over 10 years ago, but on average have been between 15 and 20 kilograms per household per month. Raskin has become the largest of the permanent household-targeted social assistance programs by expenditure and the second-largest by official targeted coverage. As food items make up more than two-thirds of overall expenditure on the average consumption bundle for poor households, and rice alone makes up approximately one-quarter of overall expenditure (on the same average consumption bundle), Raskin potentially delivers two significant benefits to beneficiary households. First, Raskin could bring to poor households approximately half of their monthly rice consumption needs (depending on Raskin’s allotment size) at subsidized prices, leaving them additional income to spend on other budget items. Second, the Raskin program could keep rice prices stable for poor households by facilitating purchases of significant allotments of rice at pre- announced prices.7 Raskin consumes the largest share of all regular household-based social assistance expenditure by the central government. Central government expenditures on Raskin totaled nearly Rp 13 trillion in 2009 (approximately US$ 1.4 billion at 2009 nominal exchange rates) meaning Raskin expenditures alone account for 55 percent of all regular household-based social assistance spending.8 Over the five years ending in 2009, nominal Raskin expenditures have doubled while expenditures measured in real terms have also risen (albeit more slowly). In addition, marginal per-beneficiary costs have risen (in both nominal and real terms) as the program has officially expanded to cover more beneficiaries. Recent budget reporting (2011’s Nota Keuangan dan Rencana APBN) indicates that the 2005 Raskin program officially targeted 11.1 million households at a cost of Rp 397,830 per household and by 2009 official targets were 18.5 million households at a cost of Rp 720,000 per household. Figure 2: 15,000 20 Target Number of Beneficiaries Raskin Expenditure (LHS) (Millions of Households) Expenditure 12,000 16 Rp billion Target no. of beneficiary and 9,000 households (RHS) 12 Coverage, 2005-2009 6,000 8 3,000 4 0 0 Sources and Notes: Kemenkeu and World Bank staff calculations. The number of households covered has grown while real benefit levels have fallen. Between 2005 and 2010 7 See Timmer (2004) for more on the role of Bulog and Raskin in rice-price stability in Indonesia; also see Figure 4 and accompanying text below for more on the size of Raskin purchases relative to the size of total domestic rice consumption. 8 Regular spending excludes expenditures on the temporary unconditional cash transfer, BLT, which was brought out once in 2005 (for 12 months) and once in 2008 (for 9 months). BLT initiatives were more expensive than Raskin, accounting for 70 percent of all household-based transfer spending in 2005 and 50 percent in 2008. Nonetheless, BLT is run by pre-existing agencies and personnel, and the program does not maintain permanent offices or otherwise have recurring fixed costs or overhead costs, so BLT expenditures are truly “one time” expenditures. 11 RASKIN Subsidized Rice Delivery households have been allowed to purchase up to 15 kilograms of Raskin rice per month at prices approximately Rp 4000 to 4500 lower (per kilogram) than regular market prices, implying a per-kilogram benefit to households of approximately 75 percent of the regular market price.9 That discount, multiplied by the number of kilograms actually purchased is the implicit value of the Raskin transfer to households. As coverage has risen to over 50 percent of all Indonesian households – a rate almost double the rate the program intends10 and twenty percentage points larger than the overall coverage rate of the next-largest household-based social assistance transfer11 – de facto benefit levels have decreased. In 2010, households reported buying approximately 3.8 kilograms of Raskin rice per month. This means that “on-paper “benefit values of almost 8 percent of poor household expenditure12 translate into values “on the ground” of between 2 and 3 percent of poor household expenditure (Figure 8)). Table 1: Official name: Beras untuk Rakyat Miskin (Raskin) Raskin at a Program type: Subsidized Rice delivery Glance Program Type and inaugural year: Permanent, tax-financed, 1998 Coverage: National (100 percent provinces, 100 percent of districts) Official number of target beneficiaries: 18.5 million households Official value of benefit (2009): Rp 720,720 per household per year Benefit value from household records: Rp 125,250 per year Public expenditure (2009) Rp 13 billion (US$ 1.25 billion) Administrative cost per official target beneficiary, Rp 162,703 (US$ 16) budgeted (2009) Percent of poor households covered from 76 household records Key policy and executing agency: Kemenkokesra Key implementation agencies (role): Bulog (1st stage) and local administrations (2nd stage) Support operations partners (role) BPS (targeting and eligibility) Local Government participation Delivery to households, socialization, M&E Sources and Notes: Kemenkeu, Susenas 2010 & World Bank staff calculations. Values are for 2010 unless otherwise noted. Administrative cost is based on budget formulas – see immediately below and Box 1. Kemenkokesra and Bulog are the key policy and executing agency for Raskin (respectively) in stage 1, and responsibility for Raskin delivery to households (stage 2) is delegated to local governments. Kemenkokesra is tasked with determining Raskin policy – monthly allotments, number of months over which Raskin will operate, and coverage levels – and communicating these directives to Bulog. Bulog is then responsible for delivering these agreed quotas to all 50,000-plus distribution points. Bulog receives a budget, which is negotiated with Kemenkokesra, for rice procurement, storage, and logistics. Local governments and village administrative apparatuses are then responsible for notifying eligible beneficiaries and arranging for the transport of rice from the distribution points to households. Regulations imply that local governments should fund their delegated stage 2 Raskin activities from general budget allocations to regions. Total operating revenues and costs are less straightforward to tabulate than other social assistance programs 9 These are averages. Kilogram limits for households were at a high in 2005 (20 kilograms) and a low in 2007 (10 kilograms). In 2010, the kilogram limit was 14. In 2007, the Raskin per-kilogram price was Rp 1000 while market prices averaged Rp 5076; in 2010, Raskin prices had risen to Rp 1600 and market prices to Rp 6397. 10 In 2010, the Raskin program manual stated intended beneficiaries were poor and near-poor households as determined by BPS data (from 2008). Based on the medium-term development plan agreed in 2010, the official target number of Raskin beneficiaries is 17.5 million households. 11 By either expenditure or coverage, the second largest regular program is the health fee waiver program, Jamkesmas; see See “Social Assistance Program and Public Expenditure Review 4: Jamkesmas” in this series. Jamkesmas is also prioritized for poor and near poor households. 12 “On-paper” benefit values are the maximum subsidy that Raskin beneficiaries are allowed according to regulations. This is the product of the stipulated monthly maximum allotments and the difference between the stipulated Raskin price and the market price of rice price. Values “on the ground” are the product of a household’s recorded Raskin purchases multiplied by the difference between average actual Raskin prices and the market price of rice. 12 because of the many stages in the Raskin production line (Figure 3). Bulog is a state-owned, revenue-generating company that undertakes some commercial operations (like buying and selling rice) and receives a regular, yearly allocation from the central government’s budget to operate stage 1 of the Raskin program. Inputs informing the eventual Raskin budget allocation come from both Bulog and Kemenkokesra. Bulog’s inputs consist primarily of a determination of the total cost (including the cost of purchase) of taking a kilogram of procured rice all the way down to Raskin distribution points while Kemenkokesra (in consultation with Bulog) determines the target beneficiaries, the maximum number of kilograms per month each beneficiary can purchase, and the number of months that Raskin distributions will be made. Figure 3: Stage 1 Raskin Kemenkeu Raskin Revenue Kemenkokesra (-) Beneficiaries (APBN) Inflows and Cost Domestic Rice Local (+) (+) Procurement Outflows (-) government (+) (+) Rice 1,575 50,000 Raskin BULOG Procuremeny storage distribution (-) warehouses points (+) (-) (-) International Procurement Rice Administrative and operational costs (+) Stage 2 Sources: Raskin technical manuals and conversations with GOI officials. Based on inputs, the central government allocates a Raskin budget allocation; the allocation is the difference between the total price (as determined by Bulog) and the price households will pay for a kilogram of Raskin rice times the number of kilograms that need to be procured. The allocation is the first source of Raskin revenue in Stage 1, and these allocations are represented by the leftmost green plus sign (+) in Figure 3. Bulog then procures rice from domestic producers and international sources if necessary (procurement is represented by the leftmost red minus sign (-) in Figure 3). Bulog and its regional outposts then expend additional resources in processing, storing, and transporting the procured rice to distribution points (stacked minus signs, middle Figure 3). Finally, regional Bulog facilities together with local governments incur some costs in pushing the Raskin rice the “final mile” from distribution points either directly to households or local markets (the rightmost minus signs in stage 2). Households purchase Raskin rice, representing the last source of revenues and inflow to the Raskin program (rightmost plus signs, stage 2). Revenues from Raskin sales may accumulate at distribution points, with local governments, at regional distribution storage facilities (third rightmost plus sign), or in Bulog itself (second and third leftmost plus signs). Only budget allocations (the first inflow) are observed and summarized in this report. The report also estimates how many Raskin kilograms are purchased and the approximate household price paid, but it does not observe the flow of sales revenues back to Raskin implementation partners. For more detail on what this means for estimating a range of administrative and operational costs, see Box 1 below. 13 RASKIN Subsidized Rice Delivery Table 2: Raskin 2005 2006 2007 2008 2009 Expenditure Total Raskin (Nominal, Rp billion) 6,400 5,300 6,600 12,100 13,000 Summary, 2005-2009 Analytical series: Total Raskin (Constant 2009 prices, 11,227 8,149 9,121 14,153 14,043 Rp bn) Total Raskin (US$, Rp million) 656 580 720 1,240 1,256 Share of central gov. SA spending (%) 49.8 17.9 46.4 36.6 47.3 Share of total central government 1.8 1.2 1.3 1.7 2.1 spending (%) Memo items (all officially announced/budgeted): Target, beneficiary households 11.1 12.7 16.7 19.1 18.5 (Million) Number of months of delivery 12 10 11 12 12 Monthly rice entitlement per 15 13 9 13 15 household (Kg) Value of annual benefit per 397,830 405,632 382,890 517,650 720,720 household (Rp) Sources and Notes: Kemenkeu (Financial Statement and Proposed Planned Budget for 2011), BPS and World Bank staff calculations. 14 3. Targeting Raskin is intended for poor and near-poor households, but coverage has been extended in many different ways to non-poor households. Raskin’s intended beneficiaries are poor and near-poor households, but Raskin targeting is unmonitored and the allocation of Raskin purchase rights is determined almost entirely at the local level. Like virtually all targeted social assistance in Indonesia, selecting eligible beneficiaries is a two-stage process that includes village-level (or below) consultation and drafting of lists of potentially eligible households. In the first stage, BPS is asked by Kemenkokesra to establish local-level beneficiary quotas based on a national list of poor households.13 In the second stage, consultative village meetings are then meant to be held to verify Raskin beneficiary households. In reality the lists used to determine quotas are imperfect and exclude many poor households.14 Additionally, current program manuals do not mandate that the BPS list of poor households be used in the village meetings and in the Raskin program manual section entitled “Determination of Beneficiaries” it does not state that beneficiaries must in fact be poor. Of equal importance is the variation in the manner and skill with which the consultative meetings are held. In many cases, they are simply not held because of lack of time or funds, or because villages feel that higher-level decisions will suffice. If they take place, the meetings are often not inclusive and include only a few political officials or other authorities, and the meeting activities often do not include a verification of beneficiaries on the centrally-produced quota lists.15 The occurrence of, the content discussed during, and the participants attending or contributing to the consultative meetings are optional in practice and there is no higher-level monitoring or validation of the beneficiaries who receive Raskin rights (see below). As a consequence there are potentially many different ways of allocating Raskin rights. Field studies show district governments in West Sumatra putting pressure on Raskin administrators to use the BPS list with community verification, while in East Java the meetings were held to agree upon equal sharing to avoid community tensions. Village-level decisions on Raskin rights were influenced by the commitment of both the village head (and his officials) and program administrators to a pro-poor program.16 The distribution of Raskin transfers is marginally pro-poor. The results can be seen in Tables 3 and 4 and Figures 3, 4, and 5 below. Overall, Raskin beneficiary households look more like “average” rather than poor Indonesian households. For example, about one-fifth of all Indonesians and one-fifth of Raskin recipients do not have access to safe water; lack of safe water occurs at rates closer to one-third of all poor households. On average, about 9 percent of Indonesian households are illiterate while the same rates among Raskin beneficiaries and poor households are 12 and 15 percent, respectively. Raskin recipient households are generally smaller than poor households as well. It is also clear that Raskin is primarily a rural program that nonetheless has lower prevalence among agricultural workers. 13 A locality might either be a rural village (desa) or an urban neighborhood (kelurahan). Before 2006, Badan Koordinasi Keluarga Berencana Nasional (Family Planning Coordination Agency, BKKBN) classification system for household welfare was used for quota calculation. It is has five indicators (number of meals, housing floor type, possession of multiple clothing types, ability to practice religious principles, and access to modern health care). Since 2006, the number of target households has been determined by a listing of the poor conducted by BPS, which uses a wider range of proxy indicators. 14 The BKKBN indicators in particular are not straightforward proxies for income or consumption; are not easily observable; can be manipulated; and include potentially irrelevant criteria like religious observance – see Sumarto and Suryahadi (2001). However, even the BPS list (the 2005 version is known as PSE05) uses thresholds for determining very poor, poor and near-poor households in order to achieve headcount rates consistent with the nationally representative household survey (Susenas). If the PSE05 scoring system is applied directly to all Susenas households (rather than just the subset of community-designated candidate households that the PSE05 surveyed), more than 50 million of Indonesia’s 57 million households (88 percent) are identified as near-poor or below. See Targeting Poor and Vulnerable Households in Indonesia (World Bank, 2012a) for further detail on targeting methodologies and results for the major social assistance programs including Raskin. 15 See SMERU (2008), page13 and following. 16 The village administration also should have communicated its commitment to the community while program administrators should have communicated their commitment to the village administration and any other officials who would become program operators. However, in practice the poor socialization of the program contributed to a lack of community understanding about the purpose of Raskin and its intended beneficiaries. See SMERU (2008) for a summary of the literature and broader discussion. 15 RASKIN Subsidized Rice Delivery Table 3: % of all % of poor % of Raskin Characteristics Indonesians population recipients within who: who: who: Indonesian Do not have access to bottled, tap, or well water 19 30 23 Populations, 2009 Do not have access to private sanitation 35 58 46 Live in villages without a primary school 1 2 2 Live in villages without a junior secondary school 36 42 42 Live in rural areas 52 63 65 Live with more than 5 household members 27 43 25 Have less than primary education 19 32 32 Are illiterate 9 15 12 Work in agriculture sector 38 59 49 Sources and Notes: Susenas 2009, Podes 2008, World Bank calculations. “Work in …” refers to shares of working individuals, not all Indonesians. Table 4 demonstrates that poor households not receiving Raskin are more likely to be urban, less likely to be working in the agricultural or informal sectors, and have higher levels of education. Non-poor households (by expenditure) receiving Raskin have higher levels of education, are more urban, are less likely to be in agriculture, and are more likely to be formal employees than poor beneficiaries. They are also noticeably smaller households. Relative to poor households not receiving Raskin, non-poor households with Raskin are slightly less educated and less often in the formal sector, but more often urban and with smaller households. In other words, non-poor households receiving Raskin are in some ways better off and some ways worse off than poor households not receiving Raskin. Table 4: Poor households Non-poor households Characteristics Raskin non-Raskin Raskin non-Raskin within Household or Household head is/has: (% of households) Raskin and non-Raskin Primary school or less 82 71 76 43 Populations, Agricultural Sector 65 56 57 30 2009 Formal Employee 18 27 23 45 Urban 21 26 31 55 (average among households) Household members 4.5 4.7 3.2 3.7 Child dependency ratio 58 61 39 39 Sources and Notes: Susenas 2009 and World Bank calculations. Figure 4 below demonstrates that while Raskin coverage among the targeted group of households is 70 to 80 percent, 52 percent of the next four deciles also participate, and even 23 percent of the second richest decile, leading to nearly 70 percent of all beneficiaries being non-poor and non-poor households receiving well over half of all program benefits. It also demonstrates that conditional Raskin purchases are approximately equal across all levels of household consumption. In a comparison of targeting performance among the three major social assistance programs, where outcomes are compared to a random distribution of benefits, Raskin shows targeting gains of 13 percent while BLT and Jamkesmas had gains of 24 percent and 16 percent (respectively).17 As for most Indonesian social assistance programs female-headed households and rural households are more likely to be beneficiaries of the program regardless of economic status (Figure 4). 17 That is, targeting outcomes under Raskin (BLT, Jamkesmas) are 13 (24, 16) percent better than if the same number of benefits had been distributed randomly. See World Bank (2012a) for more detail. 16 If households below 1.4 times the poverty line or below 1.6 times the poverty line are allowed to count as targeted households, the targeting gains for Raskin increase to 20 and 27 percent (respectively), indicating that some Raskin households are close (by expenditure level) to official targets. However, while Raskin coverage of the poorest 40 percent is much higher than other international examples of in-kind transfer programs, so too is Raskin coverage of the most well- off 40 percent much higher than these same international examples; see directly below for more detail on high coverage in both poor and nonpoor households.18 Figure 4. 100 15 Kilograms per household per month Raskin Percentage of Household Covered Coverage and 13 80 Purchases by subgroup and 10 decile, 2010 60 8 40 5 20 3 0 0 Source: Susenas 2010 Raskin allocations show no relationship with measures of food insecurity. Potentially, more Raskin resources could be allocated to areas that are more food insecure and where regular markets are unable to consistently provide an affordable source of nutrients.19 Figure 5 below demonstrates that Raskin allocations are not provided in larger amounts to food insecure areas. For as many food insecure areas with a relatively larger share of Raskin benefits (areas that are shaded green with stripes), there is an equal or greater number of food secure areas with the same relatively large share of Raskin benefits (red with stripes).20 Likewise, for as many food secure areas with relatively smaller shares of Raskin benefits (green with dots), there are about as many food insecure areas with the same relatively small shares of Raskin benefits (red with dots). In other words, district-level Raskin benefit shares have no correlation with district-level food security measures. 18 See also World Bank (2012a), especially Figure 2.9 and the surrounding text for international comparisons. 19 If a household cannot secure enough calories for all its members over an extended period of time, the household will disappear, so food insecurity cannot necessarily be measured in a static cross-section of households. For that very reason, it would be conceptually difficult to target food insecurity (as distinct from household expenditure) using household-level indicators. 20 In Figure 4, areas colored red with dots or green with stripes are food insecure areas, or the priority areas 1 and 2 from Figure 1 above, while areas that are green with dots or red with stripes are food secure, or the priority areas 5 and 6 from Figure 1 above. A district is over quota if it has more beneficiaries than target population (determined from Susenas near-poverty rates), and under quota if it has fewer beneficiaries. 17 RASKIN Subsidized Rice Delivery Figure 5: Food Security vs. Raskin Quota Legend: Priority and Over/Under quota (%) High Food Insecurity & Raskin Under-quota High Food Insecurity & Raskin Over-quota Medium Food Insecurity Low Food Insecurity & Raskin Under-quota Low Food Insecurity & Raskin Over-quota Source: Food Security and Vulnerability Atlas (WFP, 2010) and Susenas 2010 Total Raskin available on the ground and the amount purchased by poor households show no relationship with each other either. When more Raskin resources are available, the extra allocations could be prioritized for purchase by poor households. However, Figure 6 below demonstrates that this is not likely in practice: when more Raskin is available, this sometimes leads to larger purchases by poor households (green shading) but also frequently leads to smaller purchases by poor households (pink shading).21 For districts receiving smaller per-capita Raskin allocations, there is at least one with relatively high poor-household Raskin purchases (yellow shading), but many more with relatively low poor- household Raskin purchases (red shading). When more Raskin is available, poor households do not (on average) appear to have greater priority; poor households are not prioritized either when there is relatively little Raskin rice to go around.22 Nationally, the conditional amount of Raskin rice purchased is approximately equal across households of all wealth levels (Figure 4): on average, households with lower consumption levels are not buying more Raskin rice than higher consumption levels. 21 Concentration is measured as total Raskin available divided by total district population (as measured by Susenas). Total Raskin available is measured as the cumulative sum (within a district) of all Raskin purchases (by all households) recorded in Susenas 2010. Across districts in Indonesia, this ratio averages 6.2 kilos (per 3-month period, so just over 2 kilos per month) while the 25th percentile is 3.3 kilos and the 75th percentile is 7.6 kilos. “High Concentration” districts are those with a ratio above 7.25; “Low Concentration” districts are those with a ratio below 3.5; and “Medium” districts are all those with a ratio in between 3.5 and 7.25 (inclusive). Raskin purchases by poor households average 9.1 kilos (per 3-month period) while the 25th percentile is 5.3 kilos and the 75th percentile is 11.8 kilos.; “Low poor hh purchases” are districts where average poor household purchases are below 10 kilos (inclusive) and “High poor hh purchases” are districts where average poor household purchases are above 10 kilos. Outliers are compressed to a theoretical minimum of 3 kilograms (per 3-month period, or one kilogram per month) or a theoretical maximum of 42 kilograms (per 3-month period, or 14 kilograms per month), or the maximum allowable purchase as defined by Raskin program regulations. 22 This is also true in urban areas only; see RAND (2011) pp. 72 and following. 18 Figure 6: Poor Household Raskin Purchases vs. Raskin Concentrations Indonesia_kab_shp Low Raskin Concentration & Low Purchases Low Raskin Concentration & High Purchases Medium Raskin Concentration High Raskin Concentration & Low Purchases High Raskin Concentration & High Purchases Source: Susenas 2010 19 RASKIN Subsidized Rice Delivery 4. Potential Impacts Raskin transfers are on average 2 to 3 percent of the poverty line. Amounts of Raskin rice purchased are small and the implicit value of Raskin transfers is low because of noticeable discrepancies between total Raskin rice procured and total Raskin rice purchased; between total benefit promised and total benefit received; and between total number of beneficiaries targeted and total number of actual Raskin beneficiaries (see Section 3 above). Ministry of Finance budget reporting and Susenas household records show that of the Raskin rice procured23 to deliver promised benefits, only about half of the procured kilograms (in recent years) are actually purchased by eligible households (Figure 7). It is not clear at which stage – procurement, storage, transport, or storage and sale (Figure 3) – in the Raskin delivery process the bulk of the procured Raskin rice goes missing, but the situation on the ground indicates that distribution was not as planned. Earlier estimates (between November 2003 and January 2004) found that up to 30 percent of Raskin allocations went “missing” in between the distribution points and Raskin-buying households; in some areas the estimated rate was as high as 75 percent.24 Figure 7: s Raskin 20,000 l c o a Procurement 10,000 g l e and Raskin Purchases, 2006-2009 4,000 3,330 Million kilogram 3,000 2,674 2,000 1,736 1,625 1,339 1,000 1,029 1,023 783 0 Sources: Kemenkeu and Susenas (various years). 2007 estimates of the poor and near-poor population is potentially unreliable, so the estimate of 2007 total rice purchases for poor and near-poor households is omitted. 23 In Figure 7, “Total Raskin rice procured” represents the total amount of rice procured by Bulog (according to Bulog financial reporting documents confirmed by the Ministry of Finance) while “Total Raskin household purchases” is the cumulative sum of all Raskin purchases (by all eligible and ineligible households) recorded in Susenas in a given year. Total rice consumption by either poor and near-poor households or by all households is the cumulative sum of all rice purchases by that set of households, as recorded in Susenas in various years. Poor and near-poor status is determined by household expenditure as recorded in Susenas. 24 See World Bank (2006), especially pages 215-217 and references therein, for more details. 20 In 2010, based on public announcements, Raskin should have delivered to poor and near-poor households at least 13 to 16 kilograms of rice per month at a subsidized price of Rp 1,600 per kilogram.25 These amounts would have translated into approximately 11 percent of poverty line expenditure and between 30 to 40 percent of estimates of an Indonesian household’s rice needs (between 35 and 45 kilograms per month, according to GOI press reports). Actual purchases as reported by households,26 however, were far less (Figure 8, Figure 4). Figure 8: Raskin Official 20 8% and Actual Benefits, Kilogram of Rice per month various years 15 8% 10 6% 5 2% 2% 2% - 2004 2007 2010 Promised benefit as % of poor household % expenditure Budgeted benefit (kg/hh/mo) Actual benefit (kg/hh/mo) Average actual benefit as % of poor % household expenditure Sources: Kemenkeu and Susenas (various years). Finally, the Raskin price for households is often higher than the stipulated price. In 2004 and 2006 Raskin buyers paid approximately Rp 1,160 and 1,253 per kilogram respectively; in 2005 prices were as high as Rp 2,900 in Jakarta while the official price was Rp 1,000.27 Between November 2006 and January 2007, households reported paying Rp 1,689 per kilogram of Raskin rice while the official price was again Rp 1,000. The Susenas household survey shows average Raskin prices paid by households rising from Rp 1,283 in 2007 to 1,994 in 2010. Higher prices charged are often justified by local-level Raskin implementers as operational or transport costs, and in some cases Raskin rice is brought closer to household clusters (from the distribution points). However, markup amounts are often incommensurate with actual transportation costs: for example in a province where market costs for taking goods the distances in between Raskin distribution points to household clusters were approximately Rp 44 to 125 per kilogram, the Raskin program charged Rp 200 to 300 per kilogram.28 25 In Figure 7, “Promised benefits” are taken from press releases, Raskin regulations, and Raskin program manuals, which contain information on target households, maximum monthly purchase amounts, and the prices at which Raskin rice should be sold “Actual benefit” is the average amount that any Raskin-buying household claims to have bought in the last 3 months, divided by 3, also taken from Susenas. “Promised benefit as % of poor household expenditure” takes “Promised benefits”, multiplies them by the difference between the market price of rice and the Raskin price announced by Bulog (e.g., Rp 1500 per kilogram in 2010), and expresses the product as a percent of average poor household expenditure. Finally, “Actual benefit as % of poor household expenditure” takes actual average Raskin purchases, multiplies by the difference between the market price and the actual average price charged for Raskin rice (recorded in Susenas), and expresses the product as a percent of average poor household expenditure. 26 Among poor households only, Raskin purchases averaged 3.9 kilograms; see also Figures 4 and 6 and accompanying text. In the Susenas questionnaire, the questions pertaining to Raskin purchase amounts are potentially ambiguous; however, there is independent confirmation of average Raskin purchase levels in both IFLS (where Raskin purchases averaged 3.4 kilograms per month in 2007) and SMERU (2008). IFLS questions to households regarding Raskin purchases are unambiguous and demonstrated that over 90 percent of those with Raskin rights buy the full amount available if it becomes available; IFLS is representative of 80 to 85 percent of the Indonesian population. SMERU questions regarding Raskin are also unambiguous, but SMERU answers are based on visits (by analysts) to a limited number of districts. Average Raskin purchases (as calculated by recorded by SMERU (2008) using Susenas questionnaires) have been falling since at least 2002 and in 2006 were measured at just over 5 kilograms. 27 SMERU (2008), pages 26 to 27. 28 SMERU (2008) (Executive Summary pg. x); the excess over the market cost of transport may have been divided among program implementers and village officials. 21 RASKIN Subsidized Rice Delivery The distribution of Raskin rights to a larger set of beneficiaries also dilutes benefit levels. As mentioned above, local-level decisions on Raskin rights have produced many different outcomes. In some cases, community norms regarding sharing, conflict avoidance between neighbors, and social envy, combined with visible demands from the non-poor for Raskin rights all led to a distribution of Raskin rights that tended toward universal. In those areas, Raskin rice is shared among a greater number of individuals and most individuals participating purchase a smaller amount than the maximum allowed. In other cases, Raskin rights were rotated monthly among a larger set of individuals; again the dilution of benefits and greater coverage is the result. In isolated cases, due dates for village payments to Bulog for Raskin rice necessitated the quick sale of large quantities of remaining Raskin rice to those who could afford to take large deliveries.29 Actual benefits from the purchase of Raskin rice are lower than in other social assistance transfers. The higher prices charged to households combined with the small amounts purchased mean that actual Raskin benefit values from the 3.8 kilograms of Raskin purchased are between 2 and 3 percent of the household poverty line,30 5 percent of per- capita food expenditure, and only 6 to 11 percent of a household’s monthly rice needs in 2010 and earlier years (Figures 7, 8).31 The next least-generous social assistance program targeted to poor households has transfer values that are nearly 5 times as large as average Raskin benefits while benefits in the only program targeted to very poor households have a cash component that is 10 times as large as average Raskin benefits.32 Households are not exiting the Raskin program (on net), meaning actual benefit levels will continue to fall. Raskin quotas are based on the poor and near-poor populations: in 2000, 2004, and 2010 this meant 16 million, 13 million, and 12 million households respectively. However in between 2004 and 2010, the total number of households buying Raskin rice increased by over 40 percent (from 20 to 28 million households), implying that the number of Raskin buyers is growing while total poor and near poor households are decreasing.33 During those same years, net entry into the Raskin program from among the set of households who were never poor was 10 percent.34 From the group who was poor in 2000 but by 2007 were no longer poor, net entry was approximately 20 percent. For the group of households that became poor over the period and the group of households that became non-poor over the period, rates of “always bought Raskin” were equal (at approximately 40 to 50 percent). As quotas and total amounts procured are based on headcounts of poor and near-poor populations, overall net entry into the Raskin program will mean continuing declines in average benefit levels. 29 SMERU, 2008 (Executive Summary pg. viii-ix). 30 The March 2010 individual poverty line is IDR 211,726. Poor households in Indonesia typically have 4 to 5 members. 31 Additional comparisons (see Figure 7): total Raskin purchases by all households are approximately one-fifth the size of total rice consumption by poor and near-poor households; total Raskin purchases by poor and near-poor households are approximately one-tenth the size of total rice consumption by poor and near-poor households; total Raskin purchases by all households are approximately one-twentieth the size of total rice consumption by all households (source: Susenas and World Bank staff calculations). 32 The next least generous program is Scholarships for the Poor (BSM) – see “Social Assistance Program and Public Expenditure Review 5: BSM” in this collection; the program targeted to very poor households is Program Keluarga Harapan (PKH), the pilot conditional cash transfer – see “Social Assistance Program and Public Expenditure Review 6: PKH” in this collection. In addition to the cash component, PKH sends facilitators to assist beneficiary households in accessing other social services (health and education); this facilitation is not easy to value but does mean that PKH’s implicit transfers are larger than the cash component only. 33 Susenas various years. 34 IFLS 2000, 2007. Here, net entry is defined as share of households who buy Raskin only in 2007 (entry) minus the share of households who buy Raskin only in 2000 (exit). 22 5. Cost Effectiveness Assessing the efficiency of Raskin spending is challenging. The Raskin production and delivery process is unique and managed in two stages by two different sets of actors. Government transfers are meant to comprise a large portion of the budget for the first-stage implementing agency, Bulog, but tabulations of additional inflows and outflows (of revenues from sales and procurement, management, staff and capital costs) in both the first and second stages are not readily available. Figure 3 above demonstrates that total expenditures on administration and operation cannot necessarily be determined from budget allocations and formulas alone, making it difficult to precisely quantify operational or administrative cost and therefore total real Raskin spending. See Box 1 below for alternative estimates of a potential range of administrative and operational expenditures. Budget estimates indicate that Raskin administrative are increasing as the official number of beneficiaries has grown. According to the formulaic budget allocation procedures described above (Section 2), allocated administrative and operational costs in Raskin have risen as the official target number of beneficiaries has expanded. For example, in 2009 (the last year with realized budget data) about Rp 163,000 was allocated (per official beneficiary) for administrative and operational expenses, which is a nearly 80 percent increase from 2006 when the official number of target beneficiaries was approximately 30 percent lower at 13 million households. Table 5: Spending 2005 2006 2007 2008 2009 Efficiency Unit Cost (Total spending/no. Indicators, 576,577 417,323 395,210 633,508 702,703 beneficiaries, Rp) 2005-2009 Administrative costs per beneficiariy 252,793 91,226 84,107 161,008 162,703 (Non-benefits/No. beneficiaries, Rp) in US$ 26 10 9 17 16 Administrative overhead ratio (Non- 44% 22% 21% 25% 23% benefits/Total spending) Cost of delivering benefits ratio (Non- 78% 28% 27% 34% 30% benefits/Benefits) Memo items: Official Number of target beneficiary 11.1 12.7 16.7 19.1 18.5 households (Million) Official Value of annual benefits (Rp) 397,830 405,632 382,890 517,650 720,720 Total spending (Rp bn)* 6,400 5,300 6,600 12,100 13,000 o/w benefits 3,594 4,141 5,195 9,025 9,990 o/w Admin 2,806 1,159 1,405 3,075 3,010 Sources and Note: Kemenkeu and World Bank staff calculations 23 RASKIN Subsidized Rice Delivery 6. Implementation There are no direct incentives encouraging Raskin operatives to make the program more effective for vulnerable households. Program support operations are underprovided, of low quality, and do not feed into a program improvement cycle. Suggested Raskin socialization procedures contain few details on either instruments or processes and fewer positive incentives or negative sanctions for effective socialization. This leads to variability in content, method, and effectiveness of socialization activities. Few Raskin-specific socialization activities were provided to program implementers at any level; socialization that did occur usually took place as part of village or neighborhood-level general purpose meetings (called musyawarah) where general introduction and development of the Raskin program, as well as general discussion of all other initiatives and general activity planning, take place. To beneficiaries and communities socialization was informal, sporadic, and often lacking in substantive content. Field research has found that program details like amounts, prices, frequency of Raskin distributions, eligibility, and more generally program goals are not widespread.35 Such information, as well as information regarding previous distributions, upcoming distribution schedules, changes to prices or beneficiary lists is not generally publicly available. Without frequently-updated information on Raskin objectives, allocations, and outcomes, beneficiaries and wider communities will not be able to effectively regulate the program. Likewise, implementers at any level are not given strong direct or indirect incentives to improve Raskin efficiency or effectiveness for targeted poor and near-poor households. Though Bulog hired a consortium of universities to provide assistance to local implementers to improve outcomes in targeting, socialization, and distribution, the number of villages actually reached by the teams was far less than 1 percent of all Indonesian villages.36 There is no regular socialization refresher or capacity building schedule for Raskin implementers at any level or for communities directly. Monitoring and evaluation activities need serious strengthening and redesign to understand village-level implementation constraints and improve outcomes for households. Monitoring and evaluation and audit activities are scheduled on a rotating basis for the central- through subdistrict-level implementing units while village-level and below monitoring activities are not explicitly scheduled by either Bulog or Kemenkokesra. Analysts have described M&E carried out by Bulog as a formality that in content simply records the success or failures of implementers in meeting targeted allocations, payment timeliness, and rice quality and weight rather than effectiveness or outcome indicators.37 Likewise, the complaint handling system needs more considerate socialization before it will gather information relevant to real-time improvements or long-term reform. According to program manuals, complaints, feedback, and suggestions can be lodged at the village level and will be transmitted up to district- and province-level complaint handling units. However, field research notes that communities are not generally aware of this procedure, the complaint handling units, or the toll-free telephone line for communicating complaints that Bulog maintains. To report complaints38 or give feedback, beneficiaries turn to whoever is most obviously their interface with the program: operatives at distribution points, community leaders, or village administrative officials. Field research indicates that virtually no complaints regarding Raskin are addressed or resolved above the village level. For a more effective Raskin program, incentives in both stages of the program need thoughtful revision. Program implementers are not evaluated according to the effectiveness of the program for communities or beneficiaries (see above). A Brawijaya University study concluded that M&E activities are ineffective because in all levels of administration no one feels specifically responsible for outcomes and no additional funds are allocated for program monitoring.39 Beneficiaries and communities are not being given enough information regarding program roles and guidelines, nor is current operational information being made available. There is very little potential for effective grass- roots program oversight from the beneficiary, household, or community level. 35 SMERU, 2008 (pg. 9) 36 SMERU, 2008 (Executive Summary pg. xi) 37 SMERU, 2008 (Executive Summary pg. xi-xii). Meanwhile, from the external monitoring and evaluation activities requested by Bulog (and carried out by independent universities, NGOs, and international organizations), very few recommendations have been incorporated into subsequent operating manuals or decrees governing Raskin implementation. 38 SMERU, 2008 (Executive Summary pg. xii-xiii). The few complaints reported are most often related to quality issues, volume and quota issues, or targeting issues. 39 SMERU, 2008 (pg.35) 24 7. Budget Formulation and Measures of Program Cost Public financial management in the Raskin program is not straightforward given the division of responsibilities in different program stages. Bulog, as the state-owned logistics company, can undertake commercial activities in addition to performing its public-service obligations: maintenance of national security food stocks and exclusive import of non-domestic rice, public procurement in support of farmgate prices, emergency food response, and implementing Raskin. It also receives a regular, yearly allocation from the central government’s budget to operate the first stage of the Raskin program. However, neither Bulog nor the central government administrators in charge of oversight make final delivery of benefits to households.40 Figure 3 above demonstrates that total expenditures on administration and operation cannot necessarily be determined from budget allocations and formulas alone, so a holistic perspective of financial management across the entire Raskin production process is unavailable (given data constraints). Box 1 below presents alternative estimates of a potential range of administrative and operational expenditures given what is observable. Box 1: Raskin The first stage in the production and delivery of Raskin involves primarily Bulog. Bulog is a state- budget owned, revenue-generating company that undertakes some commercial operations (like buying and selling rice) and receives a regular, yearly allocation from the central government’s budget to operate formulation, the Raskin program. revenues, and measures of Inputs informing this Raskin budget allocation come from both Bulog and Kemenkokesra. Bulog’s program cost inputs consist primarily of a determination of the total cost (including the cost of purchase) of taking a kilogram of rice all the way down to Raskin distribution points. Processing costs in the first stage of procurement (including surveying, weighing, milling, transport from procurement place to warehouse) plus overhead costs in later stages (including storage and maintenance costs, rebagging and packaging, and distribution) plus general administrative costs are included in this total cost. Kemenkokesra (in consultation with Bulog) determines the target beneficiaries, the maximum number of kilograms per month each beneficiary can purchase, and the number of months that Raskin distributions will be made. These inputs determine the total kilograms of rice that will be procured and distributed. Bulog, Kemenkokesra and Kemenkeu then determine three prices: the first is a Raskin price for beneficiaries (Ph), the second sets the price that Bulog will pay to producers to acquire rice (Pf) for use in the Raskin program (or to add to stocks), and the third is equivalent to the total cost described above (Pt), which is the sum of the agreed procurement price plus additional administrative and management costs in the Raskin production process (see also Figure 3 above). Central government budget allocations for Raskin are based on a formula: Agreed budget allocations for Raskin-stage 1= (Pt - Ph)*QRaskin, where Pt is the Bulog-calculated total price, Ph is the price households eventually pay for Raskin rice and QRaskin is amount of rice to be procured and distributed. For example, in 2009 Bulog was allocated Rp 12.99 trillion from the central government’s budget (APBN) to implement the Raskin program. This amount was based on a Bulog total price (Pt) of Rp 5500 per kilogram, a household Raskin price (Ph) of Rp 1600 per kilogram, and targets of 18.5 million households, 15 kilograms per month (per household), and a 12-month program (or a total Raskin procurement of 3.3 million tons). This allocation formula embeds a markup amount (Pt - Pf, where Pf is the agreed price Bulog pays to farmers) over the price of procurement that could be considered as an amount given for administrative and operational expenses. This markup has averaged about 23 percent over Pf from 2002 to 2009; in 2005 (when Pf was Rp 2805 per kilogram) Pt was 50 percent higher than Pf and in 2009 (when Pf was set at Rp 4600 per kilogram) Pt price was 20 percent higher. This markup (over Pf) is one measure of administrative cost, as it represents the amount of money allocated to Bulog that is over and above Bulog’s cost of procurement. 40 In addition, neither Bulog nor any other agency has been delegated the task of monitoring or improving the post-distribution point outcomes 25 RASKIN Subsidized Rice Delivery Box 1: continued However, budget inflows are not the only Raskin revenues nor are Bulog’s stage 1 costs the only outflows in the production of Raskin (see Figure 3 above). In fact, historically, Pf > (Pt - Ph), meaning that the price Bulog agrees to pay to producers has been higher than the per-kilogram budget allocation. From 2005 to 2010, (Pt - Ph) has been smaller than Pf by an average of about Rp 460 per kilogram (or about 11 percent of Pf). So budget allocations clearly plan on additional revenue inflows into Raskin. In Figure 3, these inflows are represented by payments for Raskin rice from households. In a scenario where each Raskin kilogram is sold at the stipulated price, and all such revenues are absorbed back into the Raskin balance sheet, this revenue (from sales) plus total Raskin budgets received from central government are total inflows in the Raskin program. Such hypothetical inflows can be compared to either total budgeted Raskin expenditure or total official benefits received: these hypothetical inflows have averaged 26 percent (of either total central government expenditure or total official benefits distributed) between 2005 and 2010. Comparing such hypothetical inflows to total Raskin operating losses – the amount Bulog loses in procurement and sale, or (Pf-Ph)* QRaskin – they have averaged 35 percent between 2005 and 2010. These ratios (total inflows to total expenditure, total benefits, or total operating losses) are measures of hypothetical administrative cost under a non- problematic and complete Raskin disbursement at stipulated prices. However, summaries of actual Raskin purchases made and prices paid for each kilogram of Raskin rice (as recorded at distribution points) and revenues from those sales are not readily available. It is also unknown how much of Raskin sales revenues reenter the budgets or balance sheets of Raskin implementation partners above the local level. As mentioned, Raskin sales are not overseen or managed by Bulog. Local governments are delegated responsibility for managing the sales process (and any additional transportation they choose to provide) and are meant to use general regional allocations to complete those tasks. However, anecdotal evidence suggests not all local governments fund stage 2 Raskin operations this way, which means revenues from Raskin sales may accrue to local-level implementers. This in turn makes it is difficult to calculate total inflows and total outflows in either stage 1 or stage 2 of the Raskin production process. A third indication of administrative costs for the Raskin program compares total budget allocations with actual benefit values received. Household surveys indicate that half (or less) of Raskin rice procured makes it to households; that average household Raskin purchases are small; that average prices paid for Raskin rice are higher than the stipulated price; and that many more households (than officially targeted numbers) buy Raskin rice, so one can compare actual total budget allocations with actual benefits provided through Raskin rice purchases. For example, in 2010 Raskin budget allocations were equivalent to Rp 442,177 per actual Raskin household. Likewise in 2010, average actual purchases of Raskin rice were 3.8 kilograms per month, or about 45 kilograms per year, meaning per-beneficiary-benefit values were actually equivalent to about Rp 292,660 (given Pm-Ph, or the difference between the market price of rice and the official Raskin price). Therefore, about 66 percent of the total allocated budget became benefits for households and the remaining 34 percent was not benefits. This ratio accounts for problematic implementation and potentially incomplete absorption of Raskin sales revenue, the size of which, as well as where sales revenue is accruing, is unobserved. It is also unknown at which point officially procured rice exits the Raskin delivery chain and total revenues from this leaked rice. In other words, the 34 percent calculated is an estimate that builds in rice loss and incomplete sales revenue absorption. Three different estimates, all imperfect, indicate that a noticeable portion of budgeted amounts for Raskin are not being turned into benefits for households. This issue needs further clarification and generating solutions would require a more detailed look at all the inflows and outflows in the Raskin process as well as a better accounting of all procured Raskin rice at each point in the delivery chain. 26 8. Summary and Recommendations Poor households in Indonesia benefit from stable prices and tools which help safeguard food expenditure… Food is the largest single category of expenditure for poor households and instability in food prices, especially rice prices, threatens progress in poverty reduction. A well-timed, well-targeted, and effectively managed in-kind transfer could supplement poor household income as well as keep price increases in check. …but Raskin will not effectively deliver either service without a major overhaul. Raskin does not effectively accomplish either of the above goals for poor households because of overlapping weaknesses in design, operation, and oversight. In order to be an effective expenditure of public revenues that contributes to Indonesia’s social safety net, Raskin needs at a minimum to accomplish reforms in (1) targeting, allocation, and local-level distribution so that vulnerable households benefit from low-cost access to a significant benefit; (2) oversight and monitoring of Raskin policies and outcomes below the regional distribution points; and (3) socialization and dissemination of program goals, outcomes, financial performance, and delivery schedules. Raskin quotas and household allocations should be based on the 2011 registry of poor households (PPLS11) and the unified targeting system when it becomes available. Bulog has made progress in securing the level of rice necessary for supplying all poor and near-poor households with the mandated monthly household quota. When the new registry of poor households becomes available, Bulog should base procurement on this improved list. Raskin allocations below distribution points should be re-engineered and monitored to ensure that a full allocation reaches all eligible households identified in the PPLS11 registry of poor households. Local “equal distribution” practices are part of the reason why households receive such a small fraction of the Raskin rice to which they are entitled. Operation manuals do not address this issue. Raskin implementation partners must revise agreements with local authorities so that there are clear incentives for providing Raskin in a timely manner to poor households and clear penalties for leakage, missed deliveries, or receipt of benefits by non-poor households. The same authorities should develop a monitoring architecture and explicitly assign monitoring of local-level outcomes to an agency or external partner. Raskin socialization should be redesigned with performance incentives so local communities can develop grass-roots monitoring and provide feedback to implementers. Current socialization activities are sporadic at best; the information content is not appropriate; and results, including financial performance, of Raskin operations are not made public. If eligible beneficiaries are not informed of their rights and do not expect efficient service, they will not demand it from local authorities. Likewise, community groups do not have enough information to effectively speak on behalf of eligible beneficiaries; in fact, given the tendency to share Raskin widely, community groups may be actively opposed to enforcing Raskin standards, as it will mean the removal of a benefit most members currently enjoy. Raskin implementing costs are too high for the level of service provided. Though it is more expensive to deliver in-kind benefits than cash, central government budget allocations tolerate rice losses and incomplete absorption of Raskin sales revenue. Within household-based social assistance programs, Raskin has noticeably higher non-benefit spending and produces an actual benefit that is much lower (on average) with much lower provision of safeguarding activities. Two strategies – delivering Raskin to food-insecure areas and developing a Raskin “trigger and response” mechanism for food price crises – could pay immediate dividends for poverty reduction in Indonesia. Raskin allocations are currently uncorrelated with levels of food insecurity. Increased Raskin allocations are often suggested as a solution when Indonesian food price indices spike, but such increases are often negotiated as budget add-ons and valuable time is spent in satisfying various parliamentary procedures. If Raskin were instead automatically disbursed in either scenario – in regions with increased risk of food insecurity, or when food price crises hit – poor and vulnerable households could take advantage of the insurance provided by such a scheme. Instead of depending on usual coping strategies – reducing expenditures, putting more members to work – they will instead be confident that enough food will be available low cost when a shock occurs. Without a comprehensive overhaul, Raskin should consider focusing on price stabilization only. The current Raskin does not deliver a substantial benefit to poor and vulnerable households and has not attempted to substantially improve its service for those households. Raskin may instead be more effective as a market-targeted, supply-side intervention that keeps local markets for rice “thick”, with as many buyers and sellers as possible. If Raskin were reliably delivering rice to existing markets, the current lack of local-level oversight of outcomes could be tolerated and all households with market access would still derive benefits from low and stable prices and a functioning market in rice. 27 RASKIN Subsidized Rice Delivery References Grosh, M., del Ninno, C., Tesliuc, E., and Ouerghi, A. 2008. For Protection and Promotion: The Design and Implementation of Effective Safety Nets. Washington, DC: World Bank. RAND. 2011. “Indonesia Urban Poverty Analysis and Program Review.” Draft report for upcoming World Bank study : Urban Poverty in Indonesia. SMERU. 2006. “A Rapid Appraisal of the Implementation of the 2005 Direct Cash Transfer Program in Indonesia: A Case Study in Five Kabupaten/Kota.” Research Report. Jakarta, Indonesia: The Smeru Institute. SMERU. 2008. “The Effectiveness of the Raskin Program.” Jakarta, Indonesia: The Smeru Institute. SMERU. 2009. “A Rapid Appraisal of the Implementation of the 2008 Direct Cash Transfer Program and Beneficiary Assessment of the 2005 Direct Cash Transfer Program in Indonesia.” Research Report. Jakarta, Indonesia: The Smeru Institute. Sumarto, Sudarno and Asep Suryahadi. 2001. “Principles and Approaches to Targeting: With Reference to the Indonesian Social Safety Net Program.” SMERU working paper. Jakarta, Indonesia: The SMERU Research Institute Timmer, C.P. 2004. “Food Security in Indonesia: Current Challenges and the Long-Run Outlook.” Center for Global Development Working Paper 48. World Bank. 2006. Making the New Indonesia Work for the Poor. Jakarta, Indonesia: World Bank. World Bank. 2010. Indonesia Jobs Report: Towards Better Jobs and Security for All. Jakarta, Indonesia: World Bank. World Bank. 2012a. Targeting Poor and Vulnerable Households in Indonesia. Jakarta, Indonesia: World Bank. World Bank. 2012b. Protecting Poor and Vulnerable Households in Indonesia. Jakarta, Indonesia: World Bank. World Bank. 2012c. “Social Assistance Program and Public Expenditure Review 1: Public Expenditure Review Summary.” Jakarta, Indonesia: World Bank. World Bank. 2012d. “Social Assistance Program and Public Expenditure Review 2: BLT.” Jakarta, Indonesia: World Bank. World Bank. 2012e. “Social Assistance Program and Public Expenditure Review 3: Raskin.” Jakarta, Indonesia: World Bank. World Bank. 2012f. “Social Assistance Program and Public Expenditure Review 4: Jamkesmas.” Jakarta, Indonesia: World Bank. World Bank. 2012g. “Social Assistance Program and Public Expenditure Review 5: BSM.” Jakarta, Indonesia: World Bank World Bank. 2012h. “Social Assistance Program and Public Expenditure Review 6: PKH.” Jakarta, Indonesia: World Bank. World Bank. 2012i. “Social Assistance Program and Public Expenditure Review 7: JSLU, JSPACA, and PKSA.” Jakarta, Indonesia: World Bank. World Bank. 2012j. “Social Assistance Program and Public Expenditure Review 8: History of Social Assistance in Indonesia.” Jakarta, Indonesia: World Bank. 28 Indonesia Social Assistance Program and Public Expenditure Reviews are policy notes summarizing current issues in the Government of Indonesia’s major household-targeted social assistance programs. Design, implementation, coverage, impacts, budgetary demands, cost effectiveness, stakeholder perceptions, and program history will be examined through qualitative and quantitative analysis for each program: conditional and unconditional cash transfers, subsidized rice distribution, a health fee waiver, cash scholarships, and cash transfers for the neglected elderly, disabled, and at-risk youth. Evidence-based appraisal can assist policymakers in evaluating programs and deciding whether and how they are achieving Indonesia’s social protection goals. Reviews 1 through 8 together comprise a companion volume to the “Protecting Poor and Vulnerable Households in Indonesia” report. “Protecting Poor and Vulnerable Households in Indonesia” provides a public expenditure review of the wider social assistance sector, summarizes results from Reviews 1 through 8, explores gaps in both population coverage and risk coverage of current social assistance programming, and offers recommendations and suggested reforms for achieving an integrated and coordinated household-based social assistance system. THE WORLD BANK, JAKARTA OFFICE Indonesia Stock Exchange Building Tower II/12th Floor Jl. Jend. Sudirman Kav. 52-53 Jakarta 12910 Tel: (6221) 5299-3000 Fax: (6221) 5299-3111 Website: www.worldbank.org/id