Abstract
Since poverty is often believed to be a root cause of clientelism, government policies to reduce poverty should also help to reduce clientelism. However, scholars studying clientelism are more likely to view social policy as a potential resource for clientelist politicians. This article examines this paradox in the Philippine context by offering a general framework to identify when social welfare policies are likely to reduce clientelism, and by applying this framework to the Philippines, focusing on the Pantawid Pamilyang Pilipino conditional cash transfer programme, or Pantawid. I argue that the policies that are most likely to undercut clientelism are universal social protection policies that provide poor families with security, although these are the least acceptable to middle-class taxpayers. This is exemplified by the Philippines, which has tended to introduce social policies that increase the scope for clientelism by making discretionary allocation more likely, rather than policies that offer income security to the poor. The Pantawid programme attempts to overcome these problems by introducing a centralised targeting mechanism to identify beneficiaries and by guaranteeing the benefit to all eligible families, but like all conditional cash transfer programs falls short of guaranteed and universal social protection.
Constituency service, as it is practiced in the United States today, contains echoes of machine politics of old. But many of the functions of the machine have been taken over by governmental bureaucracies, and rarely would an individual constituent be denied access to a social program because he or she has proved to be electorally unresponsive – as is the case for clientelism. Susan C. Stokes, Thad Dunning, Marcelo Nazareno and Valeria Brusco (Stokes et al. 2013: 14)
By and large, political clientelism is more prone to thrive precisely in that area of public policy most neglected in the past and most relevant in the present: social policy, and especially anti-poverty policy.
Diego Abente Brun (Brun 2014: Kindle locations 150–152)
Successful poverty reduction outcomes will lessen dependence of the poor on patronage resources. Careful program design can produce specific good governance outcomes. Joel Rocamora, Philippine National Anti-Poverty Commission (NAPC) (Rocamora 2011: 1)
Introduction
Benigno S. Aquino III was elected president of the Philippines in 2010 on the back of the slogan “If there is no corruption, there will be no poverty” (kung walang corrupt, walang mahirap). His administration presents its anti-poverty initiatives as ways to reduce clientelism and corruption. The most notable of these was the dramatic expansion of the Pantawid Pamilyang Pilipino Programme, otherwise known as Pantawid, a conditional cash transfer (CCT) programme begun by his predecessor. The present article 1 uses the Pantawid to examine whether and when social policy can help to reduce clientelism.
I am grateful to the College of Liberal Arts and Social Science at the University of Guam, and the American Political Science Association for research support, and the College of Liberal Arts at De La Salle University in Manila for affiliation and office space during the summer of 2014. I wish to thank Cecilio Adorno, Eric Batalla, Jackson Cuarteros, Ronald Holmes, Paul Hutchcroft, Cecilia Lero, Zenaida Ligan-Ashburn, Ronald Mendoza, Nestor Pestelos, Julio Teehankee, and the anonymous interviewees listed below for sharing their time and insights. For feedback on earlier versions I thank Kenneth Shadlen, Steve Snow, David Stuligross, participants at the workshop on Electoral Integrity at the 2014 Australian Political Science Association Annual Meeting, and two anonymous reviewers.
The epigrams above summarise the challenge. If poverty is a major cause of clientelism (Stokes 2007b: 617–619), policies that reduce poverty should weaken it; the experience of welfare regimes in the West supports this expectation. Currently, however, social policies in developing countries are often captured for clientelist allocation (e.g., Abente and Diamond 2014). If, social policy can weaken clientelist ties, as some Filipino policymakers believe, we need a theoretical framework that identifies which social policies reduce clientelism and which ones exacerbate it. The first part of this paper develops such a framework. The next applies it to the Philippines.
The framework is developed in three steps. First, like many authors (e.g., Kitschelt and Wilkinson 2006; Clapham 1982b), I distinguish between factors influencing supply and demand for clientelism. I then identify five variables primarily relating to the supply that differentiate types of clientelism, arguing that only two are crucial for understanding the impact of social policy – whether benefit allocation is rule-based or discretionary and whether the voter-politician relationship is of short or long duration. Finally, I identify three aspects of a social policy that would influence its effect on clientelism. First, I note that any policy has both an income effect (which affects demand) and an allocation effect (which affects supply). Second, I differentiate policies by their economic purpose, some having a social investment focus while others are concerned with social protection. Finally, I identify several possible selection mechanisms from universal to sporadic provision. I argue that the social policies that are most likely to reduce clientelism are those that are least likely to find political favour: universal social protection policies that assure immediate income flows or benefits to poor families.
The second half of the paper addresses the Philippines. In the first two subsections I examine historical patterns of clientelism and the place of social policy in the Philippines. I explain that Philippine clientelism is closely associated with local political dynasties using discretionary control over national public policies to nurture clientele for electoral purposes. Therefore, the political discourse of ‘reform’ is strongly associated with reducing the discretionary control that members of Congress have over public resources. Social welfare efforts have been meagre and often, paradoxically, enhance discretionary allocation.
Turning to consider Pantawid in the final subsection, I examine whether the programme breaks the above-mentioned pattern. I find that by guaranteeing a social protection benefit to all eligible households, and using a national targeting system, the Listahanan, to identify beneficiaries for this and other national programs, Pantawid takes an important step towards the kind of policy that could weaken clientelism. However, this potential for reducing clientelism is blunted by Pantawid's relative emphasis on limiting political interference over assuring income security for poor voters and on targeting benefits to households with children over covering all needy households. Consequently, even if Pantawid worked perfectly, it would fall short on the challenge of undercutting clientelism.
As the waves of democratisation and liberalisation in the 1990s receded, academics sought to understand the challenges that followed by turning to some old topics. Concerns about vote-buying revived interest in political clientelism, while fears that neoliberal growth models were bypassing the poor caused even aid agencies to advocate social protection policies. However, both literatures experienced significant shifts in focus that shed light on the conditions that were most conducive to reducing clientelism through social policy. In the first of the two following subsections I identify five sources of debate that I treat as variables differentiating types of clientelism. In the second sub-section, I examine three aspects of anti-poverty policies that mediate their effect on clientelist interaction. Since clientelism is a relationship of exchange, the analysis differentiates how social policy affects the supply and demand for clientelism.
Elements of Clientelism
Clientelism attracted strong interest in political science in the 1960s and 1970s (e.g., Schmidt et al. 1977, and Clapham 1982a), especially in Southeast Asian politics (e.g., Scott 1972; Landé 1973), but waned thereafter, before being revived after 2000 following concerns that vote-buying and electoral corruption were undermining nascent democracies in poor countries. Scholarly focus shifted away from social structures (Scott 1972; Landé 1973) or weak states (Roth 1966; Clapham 1982b) to electoral behaviour (Stokes 2007b: 605; Tomsa and Ufen 2013: 6). What the two periods do share is a preference for public policy founded on an “ethic of public allocation” (Clapham 1982b: 8) not particularistic concerns. Given this shared preference, clientelism is typically defined as a relationship of exchange in which leaders receive political support for narrowly targeted benefits.
Notwithstanding the shared focus, current debates lack a shared definition. The earlier literature converged on a view of clientelism as a network of enduring and dyadic relationships characterised by unequal exchange (Landé 1977: xx). Today, scholars debate the proper use of certain terms 2 – such as patronage, clientelism, and vote-buying – by employing conflicting definitions of them. Five basic fault-lines can be thought of as variables that differentiate among types of “clientelism.” These variables are (a) where patrons’ resources originate; (b) how narrowly benefits can be targeted; (c) how beneficiaries are chosen; (d) whether benefits can be made contingent on recipients providing support; and (e) the duration of the relationship. I argue that only the third and fifth variables are essential for understanding the impact of social policy.
For example, the definitions of patronage, clientelism and vote-buying in Stokes et al. (2013), Schaffer (2007: 5–6), Hutchcroft (2007), and Kitschelt and Wilkinson (2006).
First, it matters whether clientelism is funded by private or public resources. One common view sees pre-modern clientelism as being based on unequal access to private resources, like land, and modern clientelism as based on the exchange of government services for support. This is overstated. Contemporary “vote-buying” can draw on private funds to secure political support, presumably with the expectation of recouping the investment in office. Reform efforts that seek to prevent public resources from being used to supply clientelist benefits without addressing demand may be undermined by the use of private wealth.
Second, the more narrow the manner in which benefits are aimed at particular recipients, the easier it is to demand political support in exchange. This is the dimension that has been most emphasised in the literature and scholars have used different terms to describe it. Narrow benefits are “divisible” (Shefter 1977), “excludable” (Kitschelt and Wilkinson 2006) or “particularistic” (e.g., Hutchcroft 2014) and contrasted to “public” or “collective” benefits. Despite the dichotomous language scholars recognize that this variable really covers a range of options. Many benefits can be limited to one subset of voters but not to particular individuals. These “club goods” (Kitschelt and Wilkinson 2006) or “pork barrel” goods may identify voters by geography, demography or some other criterion. 3 However, “club” goods need not be allocated in a “clientelistic” manner. A pension guaranteed to all seniors is not clientelistic but one that allows political or bureaucratic discretion in selecting beneficiaries is. The degree of divisibility of a public benefit is not crucial.
“Pork barrel” usually refers to collective goods targeted to a geographic area (e.g., Schaffer 2007a: 5). “Club goods” include both pork barrel and what Schaffer (2007a: 5) terms “allocational policies”, which “distribute material rewards to entire, geographically dispersed classes of voters”.
What is central is the third variable, whether beneficiary selection is rule-based or discretionary. Without influencing who receives a benefit, a politician cannot claim credit for it, but virtually any programme can become a resource for clientelism if discretionary allocation is possible. Nominally universal pensions may effectively allow discretionary allocation if individuals need assistance to establish eligibility (e.g., Pellissery 2013). Revenue transfers to local governments may follow a strict revenue-sharing formula, allowing clientelism locally, but not nationally, or may be allocated by political bargaining, ensuring clientelistic relationships permeate national politics. Scholarship programmes for disadvan-taged groups may win support from the group as a whole, but also permit biased beneficiary selection of particular individuals. Not surprisingly, most attempts to reduce clientelism focus on curtailing discretionary allocation.
Fourth, even when a benefit can be provided in a discretionary manner, clientelism requires that its provision be made “contingent” on receiving electoral support (Stokes et al. 2013). This is difficult if the ballot is secret, but not impossible. There are several methods that parties can use to monitor how individuals vote even when the ballot is secret. These include employing brokers with intimate knowledge of their communities, rewarding or punishing regions for aggregate vote performance, and rewarding turnout (e.g., Stokes 2007a; Stokes et al. 2013). However, such methods may be unreliable. Wang and Kurzman (2007: 61–62) found that in the 1993 Taiwanese elections, nearly half of respondents who took money from a candidate reported voting for someone else. Similarly, Tomsa and Ufen (2013: 5) noted that voters often report taking money from multiple candidates. Without reliable methods to ensure voters honour their promise, vote-buying may require establishing norms of reciprocity over time (Schaffer and Schedler 2007: 21–22, 25–27; Kitschelt and Wilkinson 2006: Kindle locations 273–275, 314–315, 328–333).
Therefore, the final variable is the duration of the relationship. Is the exchange of benefit for support a single transaction or a relationship with expectation of future interactions? One-time transactions require active monitoring and suggest a market exchange. Norms of reciprocity develop through extended interactions and imply an ongoing relationship. Some scholars reserve “clientelism” for the latter, preferring to call the former “vote-buying” (Schaffer and Schedler 2007; Tomsa and Ufen 2013: 5). The two may coexist as parties or candidates may supplement public benefits directed to supporters between elections with gifts and cash at election time. Nonetheless, the discretionary allocation of public resources is more likely to be relevant for continuing relationships of political exchange than one-shot transactions. Voters may sell their votes at election time but cultivate patrons on everyday matters and as a hedge against future needs.
Two points stand out from this survey. First, the supply of clientelism depends on discretionary control over benefits. While individually divisible benefits and vote monitoring make clientelism easier, they are not essential; however, without influencing who receives a benefit, clientelist politicians have nothing to offer. Second, the duration of the relationship is the only one of the five variables that directly concerns voters’ motives, or the demand side. Yet this is precisely where social policy is likely to have an impact.
While poor voters are not alone in seeking clientelistic benefits, clientelism is strongly associated with poverty (Stokes 2007b: 617–619). This is true for both one-shot vote-buying and enduring clientelist ties, and both relative and absolute poverty, but the motives may differ in each instance. When poor voters accept a one-time payment for their vote, they are said to discount future gains at a high rate (e.g., Kitschelt and Wilkinson 2006: Kindle locations 394–401). When they engage in enduring clientelist relationships, however, they may do so precisely because such ties offer long-term benefits, such as informal social insurance or other access to government benefits (e.g., Clapham 1982: 16). Depending on the degree of poverty, the motives may reflect economic uncertainty and a lack of opportunity as well as deprivation.
In theory, either economic growth or social policy could reduce the demand for informal support networks. In practice, policies that affect poverty will have two potentially contradictory effects on clientelism. The policy's income effect – how much it reduces or increases the need of the poor for economic support – will affect demand for clientelism; its allocation effect – whether it reduces or increases discretionary control of officials – will affect supply. This is even evident when comparing strategies for economic growth. A neoliberal economic model might reduce resource allocation by the state, and with it the supply of clientelism, but increase demand by increasing absolute poverty, economic uncertainty or income inequality. A statist growth model might permit greater attention to the poor but expand discretionary allocation.
When it comes to social policy, the income and allocation effect will depend on two factors. One is whether the policy's economic purpose is social investment (enhancing productive capacity) or social protection (placing a floor below consumption). 4 The other is the selection mechanism, which can range from universal provision, through targeted provision (support to a specific needy population) to what might be termed ‘sporadic provision’ (funding inadequate to reach the entire eligible population). Both factors are influenced by politics.
See Bloom et al. (2010a: 13) for definitions of social protection. I use ‘social investment’ for any anti-poverty programme conceived of as an investment, including micro-lending and ‘human capital’.
Social investment policies aim to increase future income rather than current consumption, and were historically preferred by economists and donors (Bloom et al. 2010: 15) but they are not effective against clientelism. They typically offer benefits that can be targeted to favoured districts or clients, such as loans or retraining available to only some households, or local infrastructure (‘pork’). Worse, they often have little immediate effect on the incomes of the poor. Recognising the latter concern, scholars and aid officials have increasingly accepted the need for social protection (Pincus and Cook 2014: 2; Fiszbein, Kanbur, and Yemtsov 2013: 4).
Social protection is “concerned with the protection of basic living standards” by safeguarding current consumption in the face of risks and shocks, and includes “contributory social insurance, tax-financed social assistance and labour-market interventions” (Barrientos 2013: 25). However, as the poor in developing countries are mostly in the informal sector, where contributory social insurance is unfeasible, social protection for them consists mainly of tax-funded social assistance. In the past, this was primarily in-kind, often in the form of subsidies. Subsidies and other in-kind assistance fell into disfavour over various concerns: they primarily benefit the non-poor, involve greater administrative costs than direct assistance, are subject to leakages and corruption, are viewed as paternalistic (by prescribing consumption patterns) or distort market demand (Paqueo 2010: 125–128). The exception is health care, where Thailand's successful Universal Coverage scheme (Ramesh 2009: 158–164) has created interest in universal health care provision.
In other areas, the trend is toward cash transfers. However, cash transfers encounter domestic political resistance from middle-class taxpayers (Duman 2013; Paqueo 2010: 128; Fiszbein and Schady 2009: 9). Two ways to overcome this resistance are by targeting benefits to clearly needy recipients and by requiring socially beneficial behaviour in exchange for it. Both raise problems.
Targeting is challenging for several reasons. First, in countries where the poor work mainly in the informal sector, it is difficult to ascertain applicants’ incomes. A number of more imprecise alternatives are used, including the proxy means test (examining actual living conditions, as well as other household characteristics, such as education and ownership of assets), geographical targeting (focusing on poorer regions), categorical targeting (aimed at particular vulnerable populations like the elderly), and self-targeting (giving assistance in a form that is unattractive to the non-poor) (Pfleiderer 2013; Weiss 2004: 2–3). Several of these are relevant to Pantawid. Second, targeting entails high administrative costs, reducing the funds that are available for recipients, so much so that universal provision might be cheaper (Weber 2010: 49; Weiss 2004: 9). Third, targeting taxes the administrative capacity of poorer countries and is vulnerable to political interference, undermining political support. (This is paradoxical, since the goal of targeting is to reduce political interference and win political support.) Finally, since no targeting method is implemented perfectly, it requires choosing between inclusion errors (benefits going to ineligible recipients) and exclusion errors (intended beneficiaries not receiving benefits). This choice will shape the policy's impact on clientelism profoundly, as we will see.
A second obstacle facing cash transfers is the continuing popularity of normative arguments for funding ‘investment’ over ‘consumption’. This has led to a trend of presenting social assistance programmes as social investment, usually by requiring recipients to perform some behaviour seen as providing a social benefit, such as providing labour on local infrastructure projects or, in the case of CCTs, attending to the health and education needs of children.
Starting in Mexico in the 1980s (Britto 2010; Fiszbein and Schady 2009: 3–8), CCTs have spread widely, as much for political reasons as economic ones. CCTs are seen as investing in future growth, helping the most blameless among the poor (children) and requiring the poor to be more responsible (Bloom et al. 2010: 16–17; Fiszbein and Schady 2009: 10–11). The basic model involves giving cash grants to families below the poverty line with minor children in exchange for school attendance, immunisations, preventive health check-ups, and other activities that are deemed to contribute to long-term human capital development. CCTs are usually accompanied by rigorous targeting and monitoring. CCT programmes have been shown in many studies to be quite successful in improving outcomes on the desired behaviours.
CCTs like Pantawid, however, offer only limited social protection, and experts generally agree they should not become the primary means of providing social protection. First, they work only if the poor actually have access to health and education facilities. Second, they exclude poor individuals and families without children (Bloom et al. 2010: 18; Fiszbein and Schady 2009: 26). Third, as with means-testing, monitoring compliance is expensive, which reduces the amount of money available for the poor, and administratively complex, requiring high administrative capacity without which they may be subject to failures that erode support for the programme (Son 2008: 4–5; Interview 3).
These concerns have led experts to increasingly embrace unconditional cash transfers (UCTs) to the poor (The Economist 2013; Bajaj 2013). UCTs range from non-contributory “social pensions”, through unconditional grants to poor families (Haushofer and Shapiro 2013) to ambitious “basic income” proposals that provide an income floor to all families. With the exception of social pensions, political support for most of these UCTs is weak.
This overview suggests there is an inverse relationship between making poverty reduction politically acceptable and using it to reduce clientelism. First, while social investment programmes are politically attractive, they have a limited near-term income effect and are vulnerable to clientelist capture. Second, high administrative costs associated with targeting or conditionality reduce the income effect for any given level of funding. Third, any deficiencies in the monitoring mechanisms make political interference more likely and undermine political support. Finally, while political support requires that inclusion errors (minimising ‘undeserving’ recipients) be limited, curtailing clientelism is better served by minimising exclusion errors (ensuring the needy receive support). In short, the more universal and generous the social assistance programme, the more likely it is to weaken clientelism and the less likely it is to win support from middle-class taxpayers. This dilemma shapes the relationship between clientelism and social policy in the Philippines.
Clientelism has long been central to the study of Philippine politics. In the early 20th century, wealthy families converted local, economic dominance into national political power when the United States replaced Spain as the colonial power and introduced elected legislatures on a narrow suffrage (Cullinane 2003; Hutchcroft 2000). After 1935, when suffrage was extended to literates and an elected national executive created, holding office in Manila became an avenue to wealth and was sought for that purpose (Machado 1974). The system of patronage based on local dominance appeared broken when President Ferdinand Marcos declared martial law in 1972 and centralised patronage in his own hands. Marcos’ overthrow in 1986 returned politics to a pattern that was rooted in local domains but nourished by resources flowing from national politics.
The centrality of families as units of political organisation is the most visible manifestation of clientelism (De Dios 2007: 164-165). Locally, families rather than parties acquire and nurture a reputation for delivering goods and services. Members of political dynasties secure election to the national Congress or as provincial governor, from which posts they divert resources to their own use in a variety of ways. Dispensing cash at election time (‘vote-buying’) and discretionary allocation of government services between elections (‘clientelism’) help assure the election of these individuals. Family ties among office-holders have helped to maintain the continuity of clientelist ties. The 1987 constitution limited how many consecutive terms an individual could occupy a single office but did not set a lifetime limit, which allowed politicians from one family to rotate back and forth between offices (Eaton 2002, 2003; Hutchcroft 2012). As Mendoza (2013a) illustrated, the ability of a family to occupy multiple offices means the family as a whole controls far more public resources than is evident from considering individuals.
The congressperson's vote is the crucial commodity traded to secure resources, both in the form of budgetary expenditures and through other decisions that allocate economic rents. The provincial governor's vote is almost as important. Well below these two, and somewhat dependent upon them, are the mayor and barangay captain, chief executives of the municipality and barangay (village or neighbourhood), respectively. The governors, mayors and barangay captains acquired a measure of revenue independence with the Local Government Act of 1991 which, among other things, guaranteed a transfer of revenues following statutory guidelines. However, the simultaneous devolution of many unfunded administrative responsibilities to local government units (LGUs) ensured that their solvency would remain dependent on discretionary flows secured by members of Congress (Hutchcroft 2012).
Until recently, two important tools for delivering particularistic benefits to supporters were the Priority Development Assistance Fund (PDAF) and Congressional Insertion Allocations (CIA). The PDAF was a discretionary fund given to all members of Congress in the post-1987 period. It sought to curb Congressional lobbying for pork barrel funds by giving all members an equal amount and insulating it from presidential vetoes. The CIA, by contrast, was an allocation for a specific project or expenditure inserted after the formal budget had passed. It required a peculiar budgetary trick. Congress would appropriate funds to service the external debt in excess of what was likely to be required by assuming an unfavourable exchange rate, then allocate the unused funds to particular projects sought by members of Congress. Unlike the PDAF, the CIA was dependent on the individual members’ bargaining leverage (Interview 1).
Both tools came under attack during the Aquino administration. The CIA was curbed by removing debt servicing from the budget so that Congress was no longer required to estimate how much was required. The PDAF was struck down by the Supreme Court in 2013. Shortly thereafter, a scandal came to light involving the use of PDAF allocations by some legislators to enrich themselves by funding fraudulent civil society organisations. Subsequently, some very prominent politicians were prosecuted, including Senate President Juan Ponce Enrile and former president Joseph Estrada's son, ‘Jinggoy’ (Arquiza 2013; Kirk 2014).
Largesse from these and other sources fund a wide variety of divisible benefits from pork barrel or club goods to individual assistance. The term ‘pork’ is often used in the Philippines to refer to discretionary allocation of public resources generally (e.g., Rocamora 2011), without distinguishing between collective (‘club’) goods and individual benefits. Nonetheless, familiar instances of ‘pork barrel’ politics like infrastructure projects that serve private interests are seen as paradigmatic of patronage politics. In addition, political intervention in the selection of beneficiaries for divisible public benefits can range from using PDAF funds to provide individuals with health benefits directly to influencing access to other public benefits administered by national agencies. One popular practice involved congresspersons depositing money with a hospital to cover the medical costs of any person the legislator nominates turning health care into a discretionary benefit. 5
I heard of this from several sources, including Paul Hutchcroft (personal communication).
Efforts to control clientelism have been a recurring feature of Philippine politics since the 1930s. They come from above and from below. Presidents Manuel Quezon in the 1930s and Ferdinand Marcos in the 1960s sought to curtail local dynasties by centralising clientelism in their own hands (McCoy 1988), while self-styled reformers like Ramon Magsaysay in the 1950s and Fidel Ramos in the 1990s presented themselves as technocrats, seeking more rule-based resource allocation. On the civil society side, the National Movement for Free Elections (NAMFREL), which was founded with covert backing from the United States in 1953, successfully prevented vote-rigging by President Quirino, ensuring Magsaysay's election, and has periodically played a similar role since (Wurfel 1988: 104; Calimbahin 2010). Middle-class reform groups, from NGOs to ‘good government’ parties like Akbayan focus much of their public agendas on curbing clientelism.
Presidential efforts, however, often centre on asserting presidential control over the allocation of public resources rather than shifting to a rule-based process. This partly reflects the nature of the constitutional regime itself. Because of the need to secure legislative majorities in a Congress in which party loyalty is notoriously weak, even ‘reformist’ presidents used their power to withhold funding for particular projects to obtain support from individual members of Congress for presidential legislative priorities, or developed sources of funding outside the Congressional appropriations process that allow them to personally direct pork barrel spending. I discuss a recent example below.
Proposals for institutional changes to reduce the supply of clientelism are a recurring feature of Philippine politics. A constitutional convention called in the early 1970s to propose reforms became consumed with preventing Marcos from retaining office and eventually led to his declaration of martial law (Abueva 1979: 37). The 1987 post-Marcos constitution incorporated four major reforms: term limits for members of Congress and the president; setting aside 20 per cent of the seats of the House of Representatives for party list proportional representation; calling for a local government code; and calling for an anti-dynasty bill (Eaton 2003: 475). However, as all but the first of these reforms were left to Congress to implement, they were subsequently undermined. The party list reforms were enacted in such a way that only very small parties could be elected on it and most seats were left unfilled (Eaton 2003: 477–480). The 1991 Local Government Code devolved more responsibilities to local governments than revenue, ensuring that local governments remained dependent on discretionary transfers; additionally, provisions to ensure the participation of NGOs in a watchdog role remained largely inoperative (Eaton 2003: 484–486). The term limits applied only to consecutive terms, allowing the politicians to rotate between national and local office. Furthermore, Congress refused to pass an anti-dynasty bill (Eaton 2003: 480–482), and has since failed to act when one is proposed, as in the current (Benigno Aquino) administration (Diaz 2015).
Another thrust has been fiscal reforms, which have again been a supply-side measure. The enduring challenge of the Philippine political economy, after the concentration of land, has been the fragmentation of fiscal and regulatory policy by rent-seeking politicians, often in concert with bureaucrats appointed by patronage, which has eroded both the revenue base of the country and the coherence of economic policy (Hutchcroft 1991). The Ramos administration (1992–1998), as part of an overall package of economic reforms, sought to broaden the tax base with a value-added tax and rationalise tax collection by limiting the Bureau of Internal Revenue's discretion to offer tax concessions to particular enterprises (Eaton 2002: 106–123, 174–181). The Aquino administration introduced an initiative termed ‘Bottom-Up Budgeting’ (BUB), which sought to involve civil society organisations in the allocation of social welfare funds.
However, like Quezon and Marcos, both Ramos and Aquino have been accused of using pork barrel to reward individual politicians. In the Aquino administration, the Disbursement Acceleration Programme (DAP), set up ostensibly to accelerate the implementation of necessary projects without awaiting the cumbersome process of Congressional appropriation, effectively allowed the president to personally direct local infrastructure projects where he wished. Inevitably, the president was accused of favouring political allies. Portions of DAP were ruled unconstitutional at the same time as PDAF, and the threat of prosecution still hangs over Aquino (PCIJ 2014; Businessworld 2016).
Unlike efforts to reduce the supply of clientelism, attempts to limit demand fall largely under the heading of moral suasion. Schaffer (2007b) described efforts at “voter education” aimed at poorer voters, while Thompson (2010) described the emergence of a “reformist” narrative that appeals to middle-class voters on a classic “good government” platform as a staple of presidential politics. Poverty reduction, while supported in principle, has not generally been discussed as a means of achieving this. Indeed, Thompson (2010) observed that the “reformist” narrative is contrasted to a “populist” narrative, which rhetorically champions poverty reduction.
Social Welfare Policy
Given the association of local infrastructure spending with clientelism, it is striking that many poverty alleviation efforts in the Philippines focus on infrastructure. Ramos’ Social Reform Agenda invested in health care and education, but emphasised building physical facilities in poorer areas (Haggard and Kaufman 2008: 239–240). KALAHI-CIDSS – Macapagal-Arroyo's flagship programme, which was designed with considerable input from international agencies – built on this, focusing on local infrastructure projects related to poverty alleviation, but sought to reduce the influence of clientelist politicians by including civil society organisations in project selection (World Bank 2013).
Social protection programmes were historically small and patchy. At the start of the Benigno Aquino administration, the country was viewed as a laggard in terms of social protection for the poor (ADB 2013) 6 and in poverty reduction generally (NAPC 2011a: Figure 7). Social assistance programmes and other measures show that the poor represent a small fraction of total social protection. In 2009, total social protection expenditure was 2.5 per cent of GDP, but 80 per cent of that was for social insurance programmes, with pensions alone taking up 43 per cent of all social protection. Social assistance programmes amounted to 13 per cent of the total, or 0.3 per cent of GDP spread out among 19 programmes. 7
According to ADB's Social Protection Index, in 2009 the Philippines was slightly above the median of lower-middle-income countries in Asia-Pacific overall, but well below it when only social assistance is considered (ADB 2013: Table A.3.5). Over the course of the Aquino administration this improved considerably and Pantawid alone now accounts for 0.5 per cent of GDP (Interview 3).
Campos-Hamper (2012), Table 11. Approximately 5 per cent of social protection expenditure was accounted for by the sponsored Health Insurance Programme (Table 12), discussed below, which pays health insurance premiums for poor households but is classified under social insurance. Transferring that to social assistance gives 75 per cent for social insurance and 18 per cent for social assistance. The largest outlay in social insurance was for government pensions. In social assistance it was for a Department of Education programme that combined poverty relief with adult literacy. 4Ps was second at 25 per cent of all social assistance spending (Table 14) but had the highest number of beneficiaries at nearly 4 million. Again this is for figures in 2009, before Benigno Aquino's election.
This spread means that “only a few [programmes] reach significant numbers of people and offer sufficient benefits to make a noticeable contribution to poverty alleviation” (Ramesh 2014: 47). Moreover, even programmes that do reach large numbers are underfunded and fall well short of their target populations. This is illustrated by two broad-based social protection programmes established in recent years: the National Health Insurance Programme, or PhilHealth, established in 1995, and the Social Pension, established in 2010.
PhilHealth is a universal, mandatory and contributory health insurance programme. The Sponsored Programme, which pays premiums for the poor, was added in 1997 with money to come equally from the national, provincial and local governments. Although one-quarter of the population qualify for the programme (Haggard and Kaufman 2008: 141), budget shortfalls, particularly from local governments, and a “lack of accurate poverty data and political meddling” kept between one-third and one-half of the eligible population off the rolls as recently as 2009 (Ramesh 2014: 47). This turned a universal programme that might have reduced clientelism into one that was ripe for clientelist intervention. 8 Moreover, the actual medical costs covered by the programme are quite limited (Interview 2). Nonetheless, PhilHealth is popular and politicians cultivate support by providing access to it. Just as Congressmen used PDAF funds to pay for medical costs of constituents, local governments sometimes offer medical benefits for their own employees. Cebu City, for example, offers medical benefits to supplement PhilHealth – but only if they have voter IDs in Cebu – and was proposing to pay PhilHealth premiums for barangay workers (Interview 8). As noted below, one of the more popular benefits of Pantawid is the fact that poor families are automatically enrolled in PhilHealth. In 2014, a bill was passed to pay PhilHealth premiums for all seniors over 60 (Rappler 2014a, 2014b). Time will tell if funding levels meet the requirement.
One reason for the low level of enrolment may be a requirement that beneficiaries must have two types of valid, government-issued ID (Zenaida Ligan-Ashburn, personal communication). While there are provisions to allow the poor to acquire the requisite documents, it is possible that these require clientelist intervention.
The Social Pension is less ambitious but has faced similar constraints. By law, it provides 500 PHP a month to all indigent citizens over 60, but budget shortfalls forced the Department of Social Welfare and Development (DSWD) to limit benefits to those over 77, an astonishingly high age. Wealthier provinces sometimes supplement the Social Pension by offering one with a lower eligibility age. Cebu City, for example, pays 1,000 PHP to seniors aged 60 and above but, tellingly, again only to those with voter ID cards in the province (Interview 8). In 2014 a bill was passed to fund expanded implementation and in 2015 DSWD announced it would lower eligibility to 65. During the debate over this measure, one senator pointed out that the expansion could be funded by taking a small portion of the generous funding for Pantawid (Torregoza 2014).
This background provides the context for the Aquino administration's efforts to tie poverty alleviation to political reform. It is noteworthy that members of the administration make the explicit case that poverty reduction will lead to improved governance. Thus, National Anti-Poverty Commission (NAPC) Secretary Joel Rocamora wrote: “Successful poverty reduction outcomes will lessen dependence of the poor on patronage resources. Careful program design can produce specific good governance outcomes” (Rocamora 2011: 1). Similar views are found in the administration's National Anti-Poverty Programme (NAPC 2011b: 6). This document also notes that the Aquino administration reduced the budgetary allocation to infrastructure and economic services (associated with “pork”) and increased funding for poverty alleviation over 20 per cent (NAPC 2011b: 1–2). These documents also identify specific links between the major anti-poverty programmes and governance outcomes (NAPC 2011b: 3–7). The flagship of these efforts is the Pantawid programme, which had its budget doubled in the administration's first year (NAPC 2011b: 1) and set an ambitious agenda for its first three years of increasing coverage by Pantawid from one million households to four million over three years and, separately, “also targeting 5.2 million families for subsidized health coverage” (NAPC 2011a: 1).
The Pantawid Pamilyang Pilipino Programme was initiated on a pilot basis in 2007 by the Arroyo administration, with funding from international agencies, in response to concerns that the Philippines would fall short of its Millennium Development Goal (MDG) targets (Interview 3). Like most CCTs the programme gives cash grants to the poorest families with young children, contingent upon the families’ meeting certain conditions relating to school attendance and child-related healthcare. The Aquino administration expanded the programme, which was focused on the poorest barangays, into a nationwide programme, making it the largest social welfare programme in the country. Starting with 4000 households in 2008, it grew to cover over four million families in 2014 (Interview 10). In addition, under the Aquino administration 4Ps beneficiaries are automatically enrolled for free in PhilHealth. The programme identified beneficiaries with a combination of geographical targeting and a proxy means test, conducting a census in the twenty poorest provinces, and in the poorest barangays of other provinces, to identify all poor households, and then selecting those with small children to receive the Pantawid benefit (Interview 10). Today the targeting system is nationwide and Pantawid is implemented in all but one province, and a second nationwide proxy means test enumeration was carried out in 2015 covering fifteen million households (Interview 3).
Preventing local politicians from interfering with beneficiary selection was a priority from the start. A World Bank note on governance issues in 4Ps focused entirely on preventing resource diversion (Arulpra-gasam et al. 2011). Eligible households are identified through a nationwide assessment, the National Household Targeting System (NHTS), also known as Listahan, which was established during the Arroyo administration for this purpose and designed to minimise input from elected officials. Rocamora (2011: 1) argued strongly that the programme has reduced the supply of clientelism:
Pantawid Pamilya and PhilHealth have removed substantial patronage resources from trapo (traditional politicians). Through the simple expedient of a nationally run targeting system, the National Household Targeting System for Poverty Reduction (NHTS-PR), the two programs have significantly lessened leakage […].
Consequently, Rocamora says, “It is no accident that the most vehement opponents of Pantawid Pamilya have been local politicians.” As we will see below, however, observers vary widely in their perceptions of how effective the insulation from political interference has been.
Conceptually, in addition to attacking the supply of clientelism by attempting to insulate beneficiary selection, the programme attacks demand through both social protection (cash and health benefits) and social investment (promoting behaviour that might improve future income). Supporters of 4Ps mention both economic goals, but the social investment argument is made more frequently (e.g., Mendoza 2013b); this reflects the fact that, as a conditional cash transfer aimed at families with school age children, 4Ps is more a social investment than a social protection programme. The goal is to prevent intergenerational transmission of poverty (Interview 3), not to lift those presently poor out of poverty. Indeed, the effects on income poverty have been very limited. Notably, both World Bank and ADB officials said that the choice of conditional over unconditional cash transfers was made by the government, rather than donors, out of deference to popular (middle-class) conceptions of the poor as “lazy” (Interview 3, 10), a view the officials themselves rejected.
Pantawid provides two grants, an education grant per child, up to a maximum of three children per family, and a health grant of 500 PHP per family. Initially, the education grant was 300 PHP and stopped when the child reached the age of 14. Therefore, the maximum benefit a family could receive was 1,400 PHP, which is approximately 30 USD. The condition for receiving the education grant is 85 per cent school attendance, as registered by teachers and reported to the local DSWD representative (Chaudhury and Okamura 2012). The conditions for the health grant include such practices as ensuring children take deworming pills and get immunised, or pregnant women get pre-natal check-ups (Fernandez and Olfindo 2011). In addition, recipients are required to attend monthly family development sessions, the topics of which vary greatly but can include civic education intended to discourage vote-buying (Interview 3).
The separation of the grant into two means a family can fail to meet one obligation without losing the entire amount, but all Pantawid families qualify for both grants. Until 2013, the education grant was limited to children under 14, although families with more than three children might continue to receive the benefit as older children are replaced by younger ones. In 2013, the government, on its own initiative and without funding from the ADB or the World Bank, increased the age limit to 18, and increased the grant for children aged 14 to 18 to 500 PHP a month taking the maximum monthly benefit to 2,000 PHP per month (Interviews 1, 2, 3, 10).
The programme's design limits its income effect. As with all CCTs, it excludes the elderly, single adults or families without school-age children. Moreover, the five-year time limit – since removed – meant the programme did not provide a permanent income floor. This ensures that parents with a long time horizon would maintain clientelist ties. While meeting the conditions may generate income in the future, this will not insulate parents from economic insecurity. Additionally, it appears even the income provided by the programme is inadequate to undermine clientelism. A recent World Bank assessment (Chaudhury, Friedman, and Onishi 2013) suggested that the programme is boosting school attendance and immunisation rates but not meeting its social protection aims. Specifically, although it was designed to provide 23 per cent of household expenditure – more generous than any other CCT programme except Mexico – actual disbursements averaged only 11 per cent in the households surveyed, well below comparable levels in Latin America (Chaudhury, Friedman, and Onishi 2013: 11). The explanations related to the conditionalities – non-compliance by households, failure to report compliance in a timely manner by schools and clinics, and failure by programme staff to update the database – all seemed to play a role (Chaudhury, Friedman, and Onishi 2013: 26–27, 33). The evaluation attributed to these failures the fact that it found only a marginal difference in poverty reduction between 2008 and 2011 in areas that had Pantawid compared to those that did not. 9 Additionally, grants are not indexed to inflation so that their real value declines over time (Interview 3).
Poverty levels dropped by 17.5 per cent in the 4Ps areas compared to 14.8 per cent in the control areas (Chaudhury, Friedman, and Onishi 2013: Table 3). These were areas from which the programme “was purposefully withheld for the purpose of this evaluation” (Chaudhury, Friedman, and Onishi 2013: 21) and which started to receive it immediately after the study. It should be noted that the period of the study (2008–2011) mostly falls before the Aquino administration.
Nonetheless, the programme does seem to provide an important safety net. When I interviewed Pantawid recipients in one barangay in Bohol (Interview 5), the gratitude for the programme was clear. Respondents detailed how the extra money had made a difference in their household budgets. All expressed great appreciation for their automatic enrolment in PhilHealth, even though only one had had occasion to use it. Most importantly, when asked to whom they gave credit for the programme, all respondents identified the president (referring to him by his nickname, ‘PNoy’) rather than local officials.
However, respondents also knew that the programme did not confer long-term security. While some could count on receiving the benefit for younger children as older ones lost their eligibility, one woman receiving the benefit for her grandchildren expressed concerns for her future once the money ran out. Moreover, while enrolment in Pantawid is often a gateway to other anti-poverty programmes, ranging from livelihood programmes to scholarships, access to these benefits is neither guaranteed nor governed by universalistic rules (Interview 4). Being identified as poor by NHTS made families eligible for these other programmes but did not rule out discretionary selection of some eligible families to receive their benefits. Moreover, since these benefits are divisible, the potential for clientelism is high.
Turning to the allocation effect of Pantawid, the picture is mixed. The NHTS, which was set up with World Bank and ADB assistance, used census data to identify the poorest areas and carried out household surveys to identify poor families. A proxy means test to determine whether a household was poor enough was based on an assessment form filled out by surveyors who worked for DSWD rather than by local governments, the information for which was then encoded by DSWD. Subsequently, the lists of poor households derived from the survey were validated at the local level before being finalised. Of the poor households identified by the NHTS, only those eligible children were enrolled in Pantawid, but other poor households were eligible to be enrolled in many other national programs including in PhilHealth.
Despite these precautions, assessments of the NHTS's selection process vary widely. NAPC, World Bank and ADB officials expressed confidence that political interference had been kept to a minimum (Interviews 1, 2, 3, 10). Considerably more scepticism was expressed in Cebu City. One reform-minded city councillor reported anecdotal evidence that barangay captains insert their own people into the lists (Interview 7). However, one barangay captain in Bohol stated she had forwarded fifteen names for inclusion in the list during the validation process but heard nothing back (Interview 6). It is generally believed that there are both inclusion and exclusion errors in the final list. Statistical models of the proxy means tests suggest inclusion errors were greater in the provinces, while exclusion errors were greater in Manila (Interview 3). Of course, this does not tell us whether exclusion stems from administrative oversight or political interference, but one local social welfare official in Cebu City suggested exclusion errors were common among migrants without an address (Interview 8). The same concern was expressed by one critical analysis that questioned whether the very poor were being reached (Reyes and Tabuga 2012: 9).
One area where political interference could have occurred is the initial survey. While the NHTS was a door-to-door household survey, it initially conducted a full household census only in the 20 poorest of the country's 81 provinces. In the others it surveyed only the poorest barangays (Interview 2). These surveys are conducted by surveyors hired by DSWD on contract, not career staff, and the integrity of the survey depends greatly on the care the surveyor takes to ascertain the occupants of a house. 10 One mayor sketched a hypothetical instance of interference: if a house is empty because the residents have migrated, a local official, typically a barangay captain, might give the name of a supporter, a “retired teacher or police officer,” as the residents of the house and they would then be able to collect money without being eligible (Interview 9). Others have questioned whether this scenario is realistic (Interview 3). Nonetheless the suggestion reflects the scepticism I often encountered.
This is, of course, generally true of censuses as it would be too expensive to keep a large enough staff permanently on salary. In the 2015 NHTS enumeration, a full census in all but one province DSWD “hired 40,000 field staff to enumerate 15.6 million households, about % of all households in the Philippines” (Interview 3).
A second route to insert clientelist influence is the validation process undertaken at the municipal level. The initiative rests with the com-
munity: households can request to be included, while members of the community are invited to identify households that should be excluded. Mayors validate and forward such complaints and have the option of aggregating the complaints. Approximately 300,000 households were removed from the rolls (Interview 2). The same respondent, however, pointed out that mayors were likely to resist forwarding names for removal, quoting one mayor as saying “politics is addition, not subtraction.” This would be consistent with the view that inclusion errors were more common in the provinces. The issue of families being added to the Pantawid rolls during the validation process became a national controversy when the Commission of Audit's (COA) 2013 report found that over 360,000 Pantawid recipients could not be found in the NHTS database. The implication of the critics was that these households had been inserted due to political connections. DSWD responded that COA was wrong: approximately 100,000 households were in the database, and the remainder were covered “under the Modified CCT” which was intended to extend coverage to “streetdwellers and indigenouse [sic] people” who were not captured by the NHTS (DSWD 2014). The issue has since been resolved in favour of the DSWD. 11
Based on email follow up with Interview 3, who also observes that Pantawid has an independent grievance redress system (GRS) that is accessible to all via text, phone hotline, internet or by filing in a form and does not go through mayors or other politicians.
In one barangay on Bohol, anecdotal evidence and figures were both consistent with a picture of few exclusion errors but some inclusion errors. Of the 127 households receiving Pantawid benefits in the barangay, 121 were actually included in the NHTS list of poor households. 12 Six were added afterwards in a community assembly in 2009. Conversely, twenty households were on the NHTS list of poor households but not included in Pantawid because they did not include children. Beneficiaries I interviewed largely agreed that there were no families that should be receiving the benefit but were not. However, several interviewees acknowledged, with some amusement, that “there might be” families getting the benefit that should not (Interview 5). 13
The data was provided to me by local DSWD officials themselves, and I have no reason to suspect malfeasance.
Each beneficiary was interviewed in English in the presence of others and local DSWD officials. Each answered for herself in English. I recorded verbal responses, non-verbal cues, and biographical data but not names.
What are we to make of these divergent views on the selection process? Some of the scepticism regarding the NHTS probably reflects justifiable cynicism stemming from long experience with past clientelist practices rather than evidence of current malfeasance, but some other scepticism is clearly merited. Ascertaining how much would require a much larger-scale research project. What we can say is this: for social protection to reduce demand for clientelism by providing subsistence guarantees, what matters more is reducing exclusion errors. While including non-poor recipients due to political interference indicates the continued relevance of clientelism and undermines political support for the programme, a social protection programme that guaranteed the poor were covered would weaken the structural foundations of clientelist demand. From that point of view, it is significant that the main sources of exclusion error seem to be structural – migrants without a physical address and families without children – rather than political.
To sum up, then, while Pantawid introduced the Philippines to the kind of guaranteed social protection programme that might undercut clientelism, it is still well short of that goal. While the selection of beneficiaries has been found to be very successful in reducing political interference it has obviously not eliminated the suspicion that clientelism plays an important role. At the same time, the exclusion of families without children and the relatively low average benefit found in provinces surveyed by the World Bank undercut the potential effect on clientelism, which was one of the main political selling points of the programme.
The idea that social protection policies can weaken clientelist networks is intuitively appealing, theoretically well-grounded and has historical support. Centralised welfare policies that deliver goods directly to the poor can create a direct link between central authorities (states) and poor voters, bypassing traditional clientelist structures. However, as I argued earlier, the circumstances under which social policy could undermine clientelism are quite restrictive and exclude many popular types of poverty reduction policy. This method depends on designing welfare policies to simultaneously reduce bureaucratic discretion in identifying beneficiaries (addressing the supply of clientelism) and providing adequately for the security needs of poor voters (addressing the demand). Anti-poverty policy is seldom designed to achieve these goals. The traditional bias toward social investment policies, and political and fiscal considerations that push policymakers to implement highly targeted and conditional policies, are major constraints.
The Aquino administration's efforts to tie a nationwide CCT programme to the goal of curtailing clientelism exemplifies these challenges. In the Philippines, where clientelist practices have long dominated electoral politics and rest on the discretionary provision of divisible benefits, effective social protection programmes are scarce. While Pantawid has made a commendable effort to insulate beneficiary selection from political interference through the NHTS, the suspicion that political interference continues erodes political support for the program. Even as these concerns undermine political support, the securing of which was one motive for adopting a CCT model, it is evident that the programme's exclusion of households without children and apparently meagre actual benefit, limit its income effect.
Yet there seems to be real potential for using social protection to reorient voters away from the clientelist networks that sustain political dynasties, if we do not equate this with inculcating in poor voters a middle-class ideal of civic voting. One indication of this is the fact that Pantawid benefits are associated in recipients’ minds with the president, rather than with local power brokers. Another is the popularity of the PhilHealth component of Pantawid. Unlike the CCT benefit, enrolment in PhilHealth is available to all poor households identified by NHTS and has no time limit. While the PhilHealth benefit is relatively meagre, it does have the potential to be expanded to a scale where there is a significant impact on the dependency of the poor on clientelist politicians. Another area where social protection could help would be in further strengthening the social pension.
The main obstacle to this expansion is the absence of any real political support for such a strategy among the middle class. One aspect of this is the strong middle-class bias against expansive social benefits unless they are justified in terms of a developmental rationale. A second is the emphasis in reform discourse of moralistic strictures against vote buying. In this context, the Aquino administration's efforts to tie poverty explicitly to clientelism are a significant step forward. Future efforts to tie reducing corruption to reducing poverty may wish to put more emphasis on permanent and broad-based social protection policies, like the social pension and PhilHealth.
List of Interviews
Interview 1 Senior official, National Anti-Poverty Commission
Interview 2 Senior official, National Anti-Poverty Commission
Interview 3 Senior Specialist, Asian Development Bank
Interview 4
Interview 5
Interview 6 Barangay captain, Tagbilaran, Bohol
Interview 7 City Councillor, Cebu City
Interview 8 Senior Official, Provincial Social Welfare Services, Cebu
Interview 9 Mayor of a small town in Siquijor
Interview 10 Senior Specialist, World Bank