Experimental Design
Bono 10,000 was launched in 2010 with loans from the Inter-American Development Bank, the World Bank, and the Central American Bank for Economic Integration, as well as the requirement of an external evaluation. A household received L 10,000 per year if it: (1) resided in a poor village; (2) passed a proxy means test; and (3) enrolled at least one child between 6 and 18 in grades 1 to 9 (Benedetti et al., 2016). The enrollment condition was weaker than PRAF-II, since multi-child households still received the transfer if only one child enrolled. A household received the smaller transfer of L 5,000 if it: (1) included registered children under 6 and pregnant or nursing mothers in a health center, and (2) did not include older, school-aged children. The average household was eligible for annual transfers equal to 18% of median per-capita expenditure (Benedetti et al., 2016).
Of 3,727 villages (aldeas) in Honduras, 816 were eligible for random assignment. In September 2011, 150 of these were randomly selected (without stratification) for the treatment group, and another 150 were randomly selected for the control group. The former received the treatment immediately, and the latter received it immediately after the evaluation’s endline survey was completed (but 5 months before the November 2013 elections). The remaining 516 villages were not monitored by the evaluation team, and received transfers according to standard procedures. We refer to the three arms, respectively, as CCT1, CCT2, and CCT3.
According to program rules, treated households were to receive payments in three installments per year. The first was a small, unconditional payment (1/12 of the total) received at the time of household registration. The second and third were payable upon verification of compliance with the conditions. In practice, administrative data suggest variation in the amount and timing of the payments across the treatment arms. Villages in CCT1 received large payments, on average, just after the study’s baseline surveys were conducted, and just before the endline surveys were to begin. Villages in CCT2 were excluded from the treatment until the final endline survey was conducted on June 23, 2013, although they subsequently received “catch-up” payments.
Finally, villages in CCT3 received transfers at PRAF’s discretion, with a steady increase in the transfers per eligible voter. By election day on November 23, the average CCT1 village had received L 1,773 more per voter than either CCT2 or CCT3. CCT2 and CCT3 received about the same cumulative amount, but CCT2 villages received it closer to the election. Even compared with the upper-bound estimates of PRAF-II payments, the data show that eligible voters in Bono 10,000 villages received substantially larger payments prior to the election.