election.jpg
VoxEU Column Politics and economics

Electoral reciprocity in programmatic redistribution: Experimental Evidence

Conditional cash transfers are a form of programmatic redistribution that can yield electoral benefits for incumbent parties. This column assesses the electoral impact of conditional cash transfers targeting poor areas in Honduras. Voters responded to the net amount of cash transfers and their timing, but the conditional elements of the transfers were not commonly enforced and the distribution of payments did not always conform to schedule. Electoral incentives to improve implementation do not appear to be strong.

In poor countries, vote-buying is a pervasive form of redistribution in which voters receive benefits — cash or in-kind — from party brokers in exchange for their votes (Finan and Schechter 2012, Stokes et al. 2013). Non-programmatic distribution such as vote-buying may introduce distortions in the economy (e.g. Baland and Robinson 2007, Bates 1981, Lizzeri and Persico 2001). In contrast, programmatic redistribution is shaped by transparent and objective rules, and its receipt is not conditioned on political support (Stokes et al. 2013). Beginning in the late 1990s, many countries in Latin America implemented variants of one such policy, known as conditional cash transfers (CCTs). The typical CCT policy objectively identifies poor households using geographic and/or household-level targeting and offers payments in exchange for using school and health services. A largely experimental literature finds that CCTs are successful in increasing the consumption of poor households, increasing the use of school and health services, and reducing child labour on the intensive and extensive margins. Beyond effects on the welfare of poor households, programmatic redistribution via CCTs holds promise for encouraging a shift towards healthier electoral competition that minimises distortions (Diaz-Cayeros et al. 2016). At the same time, a growing literature suggests that CCTs yield electoral benefits for incumbent presidential parties. If this is the case, CCTs might cause distortions of their own, such as the allocation of resources away from policies with potentially larger social returns, but lower electoral ones.

In a recent paper, we contribute to that literature by analysing two CCT experiments in Honduras (Galiani et al. 2016). Before the 2001 presidential elections, the ‘PRAF-II’ experiment randomly assigned households in 40 of 70 municipalities to receive of small conditional cash transfers, only intended to cover the costs of complying with education and health conditions (Galiani and McEwan 2013). Then, in late 2011, the ‘Bono 10,000’ experiment randomly assigned 816 villages to three treatment arms. In a public lottery that took place in September 2011, 150 of these were randomly selected for the treatment group, and another 150 were randomly selected for the control group. The former received the treatment immediately after the baseline was completed, and the latter received it immediately after the evaluation’s end-line survey was completed (but five months before the 2013 elections, a choice that turned out to be useful for our research). The remaining 516 villages were not monitored by the evaluation team, and received transfers according to standard procedures. We refer to the three groups as CCT1, CCT2, and CCT3. CCT1 received the largest cumulative transfers (by design, substantially larger than PRAF-II). CCT2 and CCT3 received the same cumulative amounts, but much less than CCT1. However, CCT2 received its payments closer to the election than CCT3 (Benedetti et al. 2016).

Thus, the two experiments provide variation in both the cumulative amount and the timing of payment sequences. However, both experiments were objectively targeted at poor geographic areas with more than 10% of Honduran voters, and both policies were administered by the same government agency. Hence, our paper is about the electoral implications of a targeted and objectively implemented programmatic social programme, where there is little scope for political or ‘clientelist’ manipulation.

Empirical results

We find that the earlier and smaller PRAF-II transfers did not affect voter turnout or the incumbent party’s vote share, on average. In the later experiment, CCT1’s voter turnout was 3% higher than CCT3, while the incumbent vote share (expressed as a share of registered voters) was 3.4% higher. We interpret this as evidence that transfers mobilised incumbent party supporters, but did little to encourage vote switching.

More puzzlingly, we find that CCT2’s turnout and incumbent vote share also exceeded those of CCT3, suggesting a key role for the timing of payments (see Figure 1). Individuals in CCT2 were more likely to receive ‘catch-up’ payments, closer to the election. The results are consistent with a literature in behavioural economics on the retrospective evaluation of payment sequences. In lab experiments, individuals who receive a sequence of payments tend to overweight the peak and the end payments when assessing cumulative payments. Indeed, they use the peak-end midpoint as a heuristic for evaluating an entire sequence (Fredrickson and Kahneman 1993, Langer et al. 2005, Yu et al. 2008). Our field experiment provides some confirmation of this heuristic in the Honduran context. Despite the same cumulative payments, CCT2’s peak-end midpoints — estimated using administrative payment data — were higher than those of CCT3.

Figure 1. Incumbent voter share increase (percentage points)

We argue that the structure of transfers also mattered in the earlier PRAF-II experiment. Prior research showed that the two poorest strata accounted for all of the substantial increase in child enrolment and reduction in child labour (Galiani and McEwan 2013). There were zero effects in the three less-poor strata. Our paper shows that voter turnout actually declined in the poorer strata, but not in the less-poor strata. A plausible explanation is that households in poorer strata retrospectively remembered net benefits as negative, for which they exhibited punishing behaviour in the voting booth. Three facts support this explanation. First, more households in the poorer strata complied with enrolment conditions, thereby incurring schooling costs as a consequence of accepting the payment. Second, the end of the 2001 school year (just before the November election) coincides with the early part of the coffee harvesting season, in which child labour is especially important. Third, the payments were small and designed to exactly offset costs. Yet, some authors reported sporadic or irregular payments (Fiszbein and Schady 2009, Moore 2008) that could have preceded costs.

Summary

Voters responded to the cumulative, net amount of transfers as well as their timing, insofar as timing influenced remembrance of net amounts. This leaves open the question of why voters changed their behaviour in response to a programmatic sequence of payments and costs. A plausible explanation is that some voters in programmatically targeted households were intrinsically reciprocal (Finan and Schechter 2012, Lawson and Greene 2014, Manacorda et al. 2011, Sobel 2005). Voters reciprocated because they derived utility from aiding political parties that helped them, and from punishing parties that hurt them. On the other hand, voters might have been instrumentally reciprocal. Their putative generosity in the polling booth was a rational response in a repeated game between voters and political parties, in which voters maximised the present value of future payoffs from CCTs or similarly-targeted programmes. However, instrumental reciprocity is not consistent with our findings of punishing behaviour, unless voters somehow anticipated that punishment would lead to higher payoffs in the future (Sobel 2005). Even if there is some role for instrumental reciprocity, intrinsic reciprocity makes it easier to maintain a repeated game between voters and political parties (Finan and Schechter 2012, Sobel 2005).

Finally, our results explain two stylised facts of CCT implementation in poor countries. First, the use of conditional rather than unconditional transfers is common, perhaps to assuage taxpayers (Fiszbein and Schady 2009). However, the conditions themselves have often been imperfectly enforced (Baird et al. 2014, Ozler 2013). Weaker conditions imply larger net benefits for a subset of households who would otherwise incur costs of complying with conditions. Second, due largely to operational complexities and fiscal constraints, the distribution of payments in CCT policies does not always conform to scheduled amounts and timing, in ways that our results suggest could influence electoral outcomes. In the early implementation of Mexico’s Progresa, for example, Skoufias (2005) shows that payments were sometimes delayed and larger-than-expected due to ‘catch-up’ payments. Even if politicians do not actively undermine CCT implementation, our evidence suggests that electoral incentives do not prod them to fix it.

References

Baird, S, F H GFerreira, B Özler, and M Woolcock (2014), “Conditional, unconditional and everything in between: A systematic review of the effects of cash transfer programmes on school outcomes”, Journal of Development Effectiveness, 6, 1-43.

Baland, J, and J Robinson (2007), “How does vote buying shape the economy?”, In F Schaffer (ed.), Elections for sale: The causes and consequences of vote buying, Lynne Rienner Publishers.

Benedetti, F, P Ibarrarán, and P J McEwan (2016), “Do education and health conditions matter in a large cash transfer? Evidence from a Honduran experiment”, Economic Development and Cultural Change, 64, 759-793.

Diaz-Cayeros, A, F Estévez, and B Magaloni (2016), The political logic of poverty relief: Electoral strategies and social policy in Mexico, Cambridge University Press.

Finan, F, and L Schechter (2012), “Vote-buying and reciprocity”, Econometrica, 80, 863-881.

Fiszbein, A, and N Schady (2009), Conditional cash transfers: Reducing present and future poverty, Washington, DC: World Bank .

Fredrickson, B L, and D Kahneman (1993), “Duration neglect in retrospective evaluations of affective episodes”, Journal of Personality and Social Psychology, 65, 45-55.

Galiani, S, and P J McEwan (2013), “The heterogeneous impact of conditional cash transfers”, Journal of Public Economics, 103, 85-96.

Galiani, S, N Hajj, P Ibarraran, N Krishnaswamy, and P McEwan (2016), Electoral reciprocity in programmatic redistribution: Experimental Evidence, NBER.

Kahneman, D, P P Wakker, and R Sarin (1997), “Back to Bentham? Explorations of experienced utility”, Quarterly Journal of Economics, 375-405.

Langer, T, R Sarin, and M Weber (2005), “The retrospective evaluation of payment sequences: Duration neglect and peak-and-end effects”, Journal of Economic Behavior and Organization, 58, 157-175.

Lizzeri, A, and N Persico (2001), “The provision of public goods under alternative electoral incentives”, American Economic Review, 91, 225-245.

Manacorda, M, E Miguel, E, and A Vigorito (2011), “Government transfers and political support”, American Economic Journal: Applied Economics, 3, 1-28.

Moore, C (2008), “Assessing Honduras’ CCT programme PRAF, Programa de Asignación Familiar: Expected and unexpected realities”, Country Study No 15, International Poverty Center.

Özler, B (2013), “Defining Conditional Cash Transfer Programs: An Unconditional Mess Development Impact”, World Bank.

Skoufias, E (2005), PROGRESA and its impacts on the welfare of rural households in Mexico, Research Report 139 Washington, DC: International Food Policy Research Institute.

Sobel, J (2005), “Interdependent preferences and reciprocity”, Journal of Economic Literature, 43, 392-436.

Stokes, S C, T Dunning, M Nazareno, and V Brusco (2013), Brokers, voters, and clientelism: The puzzle of distributive politics, New York City: Cambridge University Press.

Yu, E C, D A Lagnado, and N Chater (2008), “Retrospective evaluations of gambling wins: Evidence for a ‘peak-end’ rule”, In B C Love, K McRae, and V M Sloutsky (eds.), Proceedings of the 30th Annual Conference of the Cognitive Science Society, p.64-70, Austin, TX: Cognitive Science Society.

735 Reads