Subscribe to E-Bulletin Donate to CEGA

Corruption Dynamics: The Golden Goose Effect

Development Challenge

Corruption imposes enormous social and economic costs on developing countries by undermining progressive policies and market efficiencies, limiting governments’ ability to provide efficacious programs, discouraging private investment, and generating higher costs for a variety of transactions in the public and private sectors. Transparency International, a global organization dedicated to promoting government accountability, estimates that the costs of achieving the U.N. Millenium Development Goals for water and sanitation alone will be US$48 billion higher as a result of corruption.[1]   Recognizing these concerns, researchers are eager to identify effective strategies to combat corruption. While previous research has focused on the design of incentives for corrupt officials – including auditing and other mechanisms for increasing transparency – the pervasiveness of corruption in many developing countries suggests that a more dynamic approach to addressing the issue is necessary.

Context

India’s National Rural Employment Guarantee Act (NREGA) guarantees every household in rural India 100 days of paid employment per year. Workers are paid by the day (for daily wage projects) or by output (for piece-rate projects). Employment and payment history are documented through jobcards, which are obtained by households and filled out by a job site manager. Theft in NREGA takes several forms: managers can over-report days worked on a project and pocket the difference; managers can over-report output and pocket the payment for extra production; or managers can simply keep a portion of a worker’s daily wage.

This study explores the impact of an increase in the minimum daily wage in the Indian state of Orissa (effective May 1, 2007) to measure corruption under NREGA. The researchers use the wage shock to test their hypotheses about a “golden goose effect”— in which an increased wage raises the value of future theft by government officials, leading them to steal less now to preserve a future stream of benefits.

Evaluation Strategy

The evaluation compared data from the official records from NREG (including the names and addresses of participating households) to independent survey data reporting the duration of each period of NREG employment and amount of compensation paid to a sample of program beneficiaries.  To ensure work histories were accurate in the survey, researchers tied dates to major holidays to jog survey respondents’ memories. Subjects' recall was also enabled by the fact that the NREGS was a new and salient program, and periods of NREG employment were likely to be memorable and distinct compared to other forms of work.

Any discrepancies between official records and survey data were likely the result of over-reported days of work and the under-payment of wages. To evaluate the impact of a golden goose effect, researchers looked at panel data for 215 villages, and compared official jobcard data over several months in 2007 to work histories collected from the household survey. Importantly, they also compared households in the state of Orissa – where a policy change enacted a higher statutory wage and provided a more lucrative corruption opportunity for officials – to households in neighboring Andra Pradesh, which experienced no wage shock.

Researchers used the panel data to study differences between official and actual employment before and after the wage increase. They controlled for district effects, time trends on the periodicity of projects, total days worked on a project, and geographical regional effects (using data from Andhra Pradesh). Using a difference-in-differences approach, they further identified differential effects on villages that do more wage rate (vs. piece-rate) projects, as well as differential effects in Orissa using neighboring AP as a control.

Results and Policy Implications

While the official wage rate increased from 55 Rs. to 70 Rs. in May 2007, the researchers found no change in the actual wage rate. Overall, they found that daily wages are driven by the prevailing market wage, not the statutory wage. The direct effect of the wage increase on the number of over-reported days was positive but not statistically significant. However, the interaction of the wage shock and the forward-looking fraction of daily wage projects showed a strong and statistically significant negative effect. This suggests that officials’ future opportunities depressed theft from the program in the short term.

Although the wage increase did not apply to piece-rate payments, researchers found a significant decrease in theft on these projects. Officially reported payments fell immediately after the wage shock while actual payments rose. Following the daily wage shock, officially reported payments for piece-rate projects fell by about 70 Rs. a day (or about 26.5 percent of the daily theft rate before the shock). This change is attributed to “the golden goose effect,” as increased expected future benefits caused officials to be more cautious to avoid losing their jobs. Researchers estimated that without the wage increase theft would have been 73% higher than it would have in NREGA.

The results of this study suggest that program design may be a crucial component in determining possible corruption levels. Programs with a longer timeline raise the value of expected future benefits, which may lead to lower levels of short-term theft. Lax enforcement and monitoring also raises the value of future benefits from theft, and the resulting “golden goose effect” explains why efforts to limit corruption may be less effective than expected. While this study’s findings are significant, it is unclear if “the golden goose effect” would result in less corruption in the long-run, suggesting clear avenues for follow-up research in this area.

Timeline

2007-2009

[1] Transparency International (2010), “Corruption and Public Procurement,” Millennium Development Goals Report, Working paper.