The first uses a lagged dependent variable approach and the second employs a difference-in-difference framework. Both approaches yield similar estimates, indicating that during the period under examination the subsidy increased support for the incumbent party by 6.2% to 7.5%. While this increase might seem relatively modest, we emphasize that this result is the estimated effect of receiving the subsidy in a single year of a multi-year program and that the individual effects of the subsidy are likely to be attenuated because non-recipients may have also benefitted indirectly, for example through reduced food prices. The main contribution of this paper is to add to the growing empirical literature on the political benefits of poverty-reduction programs . We draw from the new wave of agricultural subsidy programs in Africa to demonstrate that such programs can alter political preferences even in settings like Malawi, where entrenched ethno-regional partisan ties might be expected to limit the political effects of government sponsored programs. In doing so, the paper also contributes to the literature specifically related to the political economy of agricultural subsidies . The findings have important implications for the larger literature on distributive politics and on our understanding of voter behavior in Africa. The conventional wisdom in scholarship on patronage and clientelism, particularly in Africa, is that ruling parties build and maintain support by channeling material favors to core supporters . One of the key debates in the empirical literature on voter behavior in Africa’s emerging democracies is whether government performance affects voters’ political preferences in contexts where social identities, particularly ethnicity, are salient. The analysis presented here shows that distributive programs need not be targeted to core supporters to be politically beneficial to incumbents,greenhouse benches suggesting that leaders in Africa’s emerging democracies may be able to enhance their support by implementing anti-poverty programs that do not discriminate against non-partisans or out groups at the local level.
Should incumbent leaders expect to reap political rewards from implementing targeted antipoverty programs? On the one hand, the answer may seem obvious. The theoretical literature on retrospective voting suggests that voters reward parties that implement desired policies . To the extent that distributive programs lead to real improvements in welfare, voters may well compensate the incumbent at the ballot box. Studies from emerging democracies in the developing world find evidence of such a link. De la O shows that in Mexico a large-scale anti-poverty cash transfer program that provided benefits to lowincome families increased voter turnout and support for the incumbent party. In Brazil, Zucco reports similar effects when examining a cash transfer program aimed at low income families with children. Manacorda et. al show that a short-term poverty relief program in Uruguay had similar effects, increasing support for the incumbent party that launched the program. Pop-Eleches and Pop-Eleches examine a program that distributed coupons to poor families for the purchase of reduced-priced computers and found that beneficiaries were significantly more likely to support parties of the governing coalition. Finally, Harding and Harding and Stasavage present evidence from Ghana and Kenya that voters reward incumbents for improving roads and expanding access to primary education. There are at least two channels through which targeted subsidy programs like the agricultural subsidy we examine in this paper might affect voter preferences. First, retrospective theories of economic voting suggest that when such programs have a positive effect on individual economic welfare, voters will reward the party responsible for implementing the program. Agricultural programs are generally highly visible initiatives that have a direct effect on material well-being for large numbers of citizens. In Malawi, for example, studies have shown a strong positive relationship between expanded fertilizer use resulting from the subsidy program and crop yields . Others have linked the subsidy to dramatic increases in maize output that reduced food insecurity and brought down the price of maize in local markets .
Existing studies suggest that subsidy programs have contributed to improved evaluations of incumbent job performance in Malawi and elsewhere . This is important, because, as Harding and Stasavage argue, voters are more likely to reward incumbent leaders for programs that can be directly attributed to those political actors. The clientelism literature suggests an alternative mechanism through which subsidy programs might affect voter behavior. In contexts where the distribution of valued benefits is controlled by party agents, citizens may trade their vote for material transfers . By this logic, the receipt of subsidy coupons might be expected to strengthen patron-client bonds, solidifying support for the party that controls access to state largess. However, for reasons described below, we expect the clientelist mechanism to be less relevant in the Malawian context that we study because of weak local-level party infrastructure. At the same time, there is reason to be skeptical about anti-poverty programs’ power to influence political preferences, particularly in African contexts where voter preferences are often driven by ethnic and regional identities. Longstanding approaches to political behavior in Africa suggest that voters hold strong preferences for candidates and parties associated with their own ethnic communities and only trust co-ethnics to deliver benefits to their group . Where ethnicity underlies political preferences, voters may be unresponsive to material transfers and may be hesitant to give incumbents credit for distributive programs, even when such programs do not discriminate by ethnicity or partisanship.Moreover, the clientelism literature suggests an additional reason why voters may be unmoved by anti-poverty programs: where local monitoring systems are weak, voters may simply accept government favors but continue to vote according to pre-existing preferences . The existing empirical literature from African cases has so far offered mixed findings on the connection between government performance and voter preferences. Several recent studies provide evidence in favor of retrospective voting theories . Other studies, however, suggest that in some cases ethnicity can trump performance . With the exception of Harding and Harding and Stasavage , these works tend to focus on broad performance measures, rather than specific antipoverty programs. As such, we still know relatively little about the potential effect of particular policy initiatives. There is good reason to be skeptical about the political effects of anti-poverty programs in the Malawian context in particular. In four of five elections after the return to competitive politics in 1994, electoral results exhibited a clear ethno-regional character, with voters in the North, Center, and South lining up en masse behind the party associated with their regions and the ethnic communities in each area . Likewise, the incumbent party at the time of our study – the DPP – was particularly weak. The incumbent president, Mutharika, came to power in the 2004 election as the hand-picked successor of the retiring president, Bakili Muluzi. Mutharika was elected as the head of the United Democratic Front party, over which Muluzi continued to preside after the 2004 election. Subsequent power struggles led Mutharika to abandon the UDF and launch the DPP in 2005. As a new party, the DPP lacked even minimal infrastructure and was poorly suited to monitor clientelist exchanges at the local level. The principal contribution of this article is to show that anti-poverty programs,plant benches particularly in the form of a targeted agricultural subsidy, can affect political preferences even when ethno-regional identities are politically salient and parties lack the ability to monitor distributive exchanges.Intervening in agricultural markets has long been a central political strategy used by African leaders to build and maintain support. In the period following independence African leaders implemented a variety of agricultural policies – tariffs, price controls, subsidies, credit schemes, and so forth – to reward favored constituencies . These interventions favored urban residents and rural notables, not ordinary farmers who posed little threat to incumbent leaders. By the 1990s, governments across the continent had removed or scaled back these programs in response to fiscal constraints and donor pressure to reduce the role of the state in the economy.
In recent years, however, large-scale subsidy programs have reemerged in several countries as initiatives to combat stagnant agricultural productivity and chronic food insecurity . One estimate suggests that seven leading African countries presently spend over US$ 2 billion per year on subsidy programs . Malawi’s Agricultural Input Subsidy Programme was launched in response to a sharp decline in rainfall in the 2004/05 growing season3 that that left an estimated 4.9 million Malawians vulnerable to hunger and food insecurity . The AISP targeted poor subsistence farmers starting with the 2005/06 season. The program represented a massive expansion of efforts to alleviate food insecurity that had been initiated earlier in the decade . The program is often heralded as a success: one report, for example, credits the AISP with taking Malawi from having a 43% food deficit in 2005 to achieving a 53% food surplus in 2007—becoming a net food exporter in just two years . These claims, however, have been challenged by those questioning the validity of such statistics and by those who have studied household-level data to measure the program’s enduring effects and its ability to reduce food insecurity and poverty . The AISP’s core objective was to increase resource-poor smallholder farmers’ access to improved agricultural inputs to achieve food self-sufficiency and to increase smallholder farmers’ incomes through increased food and cash crop production . Beneficiaries received coupons to be redeemed at government-designated retailers for either seed or subsidized fertilizer. The AISP distributed three types of coupons: a fertilizer coupon, a seed coupon, and what was referred to as a flexible coupon, which could be exchanged for a variety of seed options. The program grew in subsequent years, reaching an estimated 65% of farm households at its peak in 2008/09 at a cost of approximately USD $285 million in that year, equivalent to 16% of the government’s annual budget or about $22 per citizen in a country were GDP per capita is less than $300 . Allocation decisions4 are made at the national level by officials from the Ministry of Agriculture. District allocations are based on the amount of land under cultivation and the number of farm families per district, information that is provided by village-level officials and cross-checked by agricultural extension agents. At the village level, allocation was jointly determined by the Ministry of Agriculture, District Development Committees, Area Development Committees, and Traditional Authorities. Individual beneficiaries were supposed to be identified through use of the farm household register and open meetings held by Ministry of Agriculture staff. These committees submitted to Ministry of Agriculture staff a list of names of those in need, from which the Ministry of Agriculture selected beneficiaries. District officials then transferred coupons to representatives of village committees , and these committees distributed coupons to beneficiaries. Given the multiple actors involved in the identification of recipients and the delivery of coupons, there were a number of opportunities for the government to engage in targeting.The literature on distributive politics offers contrasting views on whether incumbents should be expected to target rewards to core supporters , swing voters or some mix of the two. Empirical studies from other contexts find evidence consistent with both models . With specific regard to targeting in agricultural input subsidy programs in African countries, relevant studies found evidence of political targeting toward opposition strongholds in Ghana , core support areas in Zambia , and an absence of targeting with respect to patterns of prior electoral support in Malawi .6 More work is needed to identify the conditions under which leaders adopt alternative allocation strategies. Regardless of whether the Malawian government may have sought to reward core supporters or to court potential swing voters through the allocation of subsidy coupons across districts, we expect that the ability to target individuals at the local level was constrained in Malawi by the relatively high degree of ethnic homogeneity at the village level.The prerequisite for individual targeting is that party agents must be able to identify core supporters, opposition supporters, and possible swing voters. In most election years, ethnicity serves as a useful predictor of electoral preferences , making it possible to identify core and swing districts based on ethnic demography. However, because rural villages have little ethnic diversity, ethnicity is much less useful for identifying core and swing individuals within villages.Two additional features of the Malawian context limit the potential for village-level targeting. First, during the period under study partisan orientations were abnormally fluid, making ethnicity even less useful as a marker of party allegiances.