Cash transfer programs, as a means of providing financial support to poorer households, are becoming increasingly popular among policymakers in developing countries. These countries, however, often lack necessary infrastructure to accurately target beneficiaries of social assistance programs. Previous studies have shown that poor targeting results in under-coverage of eligible recipients and leakage of benefits to non-eligible households. The consequences can extend beyond financial losses to the destruction of trust and social capital and feelings of inequality among constituents., This study seeks to explain how mistargeting affects crime rates and social capital within communities.
Indonesia’s Bantuan Langsung Tunai (BLT) program, with a total budget of approximately one billion USD, was one of the largest cash transfer programs implemented to date. The program aimed to compensate approximately 18.6 million poor households for the 185.7 percent increase in kerosene costs due to the removal of fuel subsidies in 2005. BLT provided households with a monthly per capita expenditure of less than Rp 175,000 (17 USD) with a stipend of Rp 100,000 (9.71 USD) per month for six months.
On a tight schedule and in the absence of a national database of household incomes and expenditures, the government used a statistical test to target poor households. The data collection process, however, was marred by sloppiness and favoritism. BLT covered approximately 21 percent of ineligible households and missed about 35 percent of the eligible population. The resulting social unrest was widely reported in the media and ranged from protests across the nation to stoning or burning of the offices of village leaders.
The researchers tested the effects of poor BLT targeting and implementation on criminal behavior using two datasets: the National Indonesian Economic Census (Susenas) of 2006 and the Indonesian Village Census (PODES) of 2005. The Susenas collected basic demographic and economic variables (including crime rate and BLT use) from 277,202 households and over 1.1 million individuals, and the PODES collected information on types of crimes experienced at the village level throughout all of Indonesia (approximately 65 million households or 230 million individuals).
Researchers first used multivariate regression to identify the probability that a household would be a victim of a crime if they were in a village with BLT recipients. To identify a causal link, researchers used a difference-in-difference model constructing the baseline level of crime with data from PODES and comparing it to self-reported crime from the 2006 Susenas data. They then compared the relative increase in crime between villages that had BLT recipients and those that did not. Because household level crime data was only available after the implementation of BLT, only village-level differences in crime were tested. After identifying changes in crime levels, the authors used a third data set, from the Indonesian Family Life Survey (IFLS), to approximate differences in social capital, measured by participation in community groups. Social capital ratings were then used to help explain the mechanism by which mistargeting may have influenced crime.
Results and Policy Implications
Results from levels
Multivariate regression results indicated that poor targeting was associated with an increase in village-level crime. In the household-level analysis, researchers found that living in a village with BLT increased the probability of being a victim of crime by 23.6 percent. For every additional 10 percent of ineligible households that received the BLT payment, the probability of being a victim of crime in that village increased by 4.4 percent.
Results from DID analysis also showed a relationship between crime and BLT participation. The analysis shows that BLT did account for a 19 percent increase in the proportion of villages that experienced crime. In both analyses, crime was more prevalent in richer communities, indicating that the correlation between BLT – intended to target poor households – and crime was not due to poverty. Rather, the data indicates the increase in crime was likely due to leakage, or giving cash to ineligible households.
The analysis of IFLS data showed that leakage of funds to the non-poor is associated with decreases in community participation and a decrease in the perception of safety in the villages. Interestingly these findings are driven by women’s participation in village activities—not men’s. This is consistent with findings in the experimental economics literature that women are more concerned with the welfare of others and fairness than men (Eckel and Grossman, 1998). This paper suggests that poorly executed social programs do more than waste money; they also adversely affect the cohesion of a community and make it more susceptible to crime. Moreover, accurate and transparent targeting practices for cash transfer programs are integral to ensure efficacy and promote social cohesion.
 Coady et al., 2004
 Putnam, 2000