Corruption in the Lab and in the Field: Evidence from a Market for Credence Goods

Topic & Objectives

Large and persistent variation in cheating behavior, even under asymmetry of information, poor enforcement, and weak fines, raises the question concerning motivations underlying such behavior. We investigate the role of intrinsic motivation for honesty in explaining this variation. Our study takes place on informal milk markets in Delhi (India), which provides a unique setting for our study. Milkmen in these markets sell fresh milk directly to their customers without any intermediaries. These markets, however, are prone to asymmetric information as both ex-ante and ex-post detection of milk quality is challenging. This asymmetry of information together with the near absence of monitoring of these markets by the government and NGOs provides milkmen with strong incentives to cheat by adulterating milk with water. We invited milkmen from such markets to take part in a novel experiment to measure their motivation for honesty. In the experiment, milkmen have to roll a die 40 times and self-report the outcome of each roll. The die we use is Bluetooth-enabled, which allows us to contrast actual with self-reported outcomes and construct measures of motivation for honesty at both the extensive and intensive margin. We then combine these experimental measures with field outcomes on cheating, as measured by added water in milk, by the same milkmen. We find a strong association between motivation for honesty and cheating in the field: the more dishonest milkmen are in the experiment, the more water they add to milk. Finally, we contrast our results with those obtained from using conventional measures that rely only on self-reported outcomes and find that the latter suffer from sizable measurement errors, leading to underestimation of the association between motivation for honesty and cheating in the field.

Key Findings

The project generates three main findings of interest:

1. Our refined experimental design reveals that milkmen differ not just in their willingness but also in their degree of dishonesty.
2. We find a strong association between the motivation for honesty and cheating in the field: the more dishonest milkmen are in the experiment, the more water they add to milk.
3. Conventional measures based on the binary classification of individuals are prone to sizable measurement errors. The downward bias resulting from this error is large, up to 26 percent.