Impacts of Conditional Cash Transfer Programs on Voting in Honduras
Previous research has shown that conditional cash transfers (CCT) can have a significant effect on areas of health and education. CCTs could also affect election outcomes, if recipients "reward” political leaders with votes. However, little research examines whether national CCT programs affect elections. In this randomized evaluation in Honduras, researchers assessed how the national Programa de Asignación Familiar (Family Allowance Program) CCT program, which provided a combination of transfers to households and transfers to support public goods, influenced voting in local and national elections. In municipalities where households received direct transfers, incumbent mayors faced a higher likelihood of re-election (an increase of 39 percent), but there was no impact on presidential election outcomes for any combination of transfers. Overall, cash transfers were more effective at boosting political support than interventions providing public goods to citizens in low-income areas.
Policy issue
Conditional cash transfer (CCT) programs offer cash assistance to households under certain conditions, including sending children to primary school or visiting health clinics. Previous research has shown that conditional cash transfers (CCT) can have a significant effect on areas of health and education. CCTs could also affect election outcomes, if recipients "reward” political leaders with votes, even if the transfer is a targeted social program and unaffiliated with a political party. However, little research examines whether national CCT programs affect local elections. How do national social programs, such as conditional cash transfer programs, affect voting in local and presidential elections?
Context of the evaluation
In Honduras, from 2000 to 2005, the national Programa de Asignación Familiar (PRAF), or Family Allowance Program, provided two types of cash transfers to rural households and communities. The first offered direct cash transfers to households, including educational and nutritional vouchers, in return for households enrolling their children to school and regularly taking their infants to local health centers. The second type offered public good transfers, in the form of cash to parent associations and health care teams, in order to strengthen local schools and health clinics. On average, the value of the transfers was US$18 per capita per year, which represents roughly 3.6 percent of participants’ total spending.
In Honduran presidential elections, two out of five registered political parties have historically received over 95 percent of votes: the Liberal Party (PLH) and the National Party (PNH). These parties are the oldest in Honduras and have retained support through patronage networks since the 19th century. Studies show that party preferences in Honduras are largely determined by client networks and party loyalty, rather than ideology. For this reason, citizens rarely change party loyalties, even when they are dissatisfied with the regime or low government performance.
Details of the intervention
Researchers conducted a randomized evaluation of the PRAF program in Honduras to examine how conditional cash transfers affect local voting, and whether targeting households or communities has a stronger electoral effect.
In October 1999, prior to the start of PRAF, researchers randomly assigned the 70 targeted municipalities into one of four groups (three intervention groups and one comparison). To assign the groups, researchers first divided the 70 municipalities into groups of 14, based on their rank in height-for-age, a commonly used indicator of child nutrition that is correlated with poverty. Then, municipalities in each group were assigned to one of the following four groups by a lottery:
- Intervention group 1 (G1): Municipalities in this group received household-level conditional cash transfers or vouchers only.
- Intervention group 2 (G2): Municipalities in this group received vouchers for mothers (household-level transfers) and transfers for service providers (public good transfers).
- Intervention group 3 (G3): Municipalities in this group received public good transfers only.
- Comparison group (G4): Municipalities in this group did not receive any transfers or vouchers, and did not participate in the PRAF program.
Researchers analyzed data on presidential elections from 1993 through 2005, and municipal elections from 1997 and 2001, from the Honduran National Electoral Tribunal, including the vote share for the five Honduran political parties, the number of registered voters, and voter abstentions.
Researchers also examined data from the International Food Policy Research Institute (IFPRI), which measures the impact of PRAF on socioeconomic outcomes and contains survey responses from members in 80 randomly selected houses in each of the 70 municipalities, from 2000 and 2002. This data included information on household socioeconomic levels including average daily wage, literacy, and availability for remittance.
Results and policy lessons
The national PRAF program increased the likelihood of an incumbent mayor’s re-election, but did not affect voting behavior in presidential elections. The results suggest that Honduran voters may have mistakenly believed that mayors played a role in their receipt of transfers and may be less loyal to their parties in municipal than presidential elections.
Impact on national elections: The PRAF program did not impact voting behavior in national elections for any treatment group. Researchers suggest that Hondurans may tend towards party loyalists in presidential elections, as well as lack information regarding the source of the cash transfer (for example, incorrectly assuming that mayors played a role in the transfer, rather than the national government).
Impact on municipal elections: Municipalities that only received household-level, direct cash transfers (Group 1) exhibited different voting behavior in municipal elections than other treatment groups. In particular, the PRAF program led to an increase in the incumbent party’s vote share by 6.6 percentage points in Group 1 municipalities. This corresponds to an increased likelihood of re-election by nearly 39 percent. Meanwhile, neither the municipalities that received both household-level and public goods transfers (Group 2), nor those that received only public goods transfers (Group 3), changed their municipal voting behavior. Researchers hypothesize that households in Group 1 may have attributed the cash transfer to the mayor and thus rewarded the incumbent mayor’s party in municipal elections.
As the PRAF program did not benefit the party that initiated it in national elections, this suggests that Honduran voters may neither be perfectly informed, nor loyal to one specific party, in their voting behavior. Overall, results show that while the national PRAF program increased incumbent re-election in municipal elections, it had no noticeable effect on vote shares in presidential elections.
Linos, Elizabeth. 2013. "Do conditional cash transfer programs shift votes? Evidence from the Honduran PRAF." Electoral Studies 32, no. 4: 864-874.