Three Essays on Strategic Misreporting

Chun Song, Purdue University

Abstract

This dissertation studies the economics of strategic misreporting and the effect of different anti-misreporting approaches based on theoretical, experimental, and quasi-experimental evidence. In Essay 1, I propose a theoretical model to study the efficacy of absolute and relative inspection standards in reducing misreporting when agents are heterogeneous in their reporting cost. I extend from previous theoretical studies by examining explicitly the performance of competitive endogenous audit rule (i.e., tournament audit) compared to the random audit as a function of agent’s heterogeneity parameter. I find that a tournament audit reduces average misreporting and the dispersion of misreporting relative to a random audit, and that the magnitude of the reduction is independent of the degree of heterogeneity among agents. A larger number of audits (presumably delivered by a softer budget constraint), a higher degree of imperfect monitoring, and larger risk aversion among agents reduce the effectiveness of the tournament audit in lowering misreporting. However, the magnitude of the reduction remains independent of heterogeneity in those cases. Theoretical predictions from the first essay are built on a strategic equilibrium concept that relies on rather sophisticated assumptions. Testing these predictions in a controlled environment is thus of empirical importance. In Essay 2, I study misreporting decisions in laboratory experiments, and I test predictions from the first essay. The game played by subjects carefully recreates the environment used to generate theoretical predictions. The experiments have two sources of exogenous variation. The first varies the audit scheme, while the second varies heterogeneity in the cost of reporting. This allows me to test the key predictions from Essay 1 by comparing outcomes across different combinations of treatments. The experimental results largely support the theoretical predictions that a tournament audit reduces misreporting, both with homogeneous and heterogeneous agents. It also supports the prediction that the magnitude of the reduction in misreporting under a tournament audit relative to the random audit is largely independent of the degree of heterogeneity. However, the misreporting reduction is smaller than predicted, as subjects in the experiment tend to misreport less in the random audit baseline. This result is consistent with subjects being risk averse as characterized in Essay 1. Similarly, efficiency gains associated with lower misreporting are smaller than predicted. In the third essay, I study a reform that conferred Chinese provincial authorities more monitoring power over air pollution performance by cities in those provinces. I use quasiexperimental methods to quantify the effects of this reform on misreporting by local authorities. Implemented in 2016, the reform gave the provincial authorities direct access to local (municipal) pollution monitoring stations, thereby making it harder for local authorities to misreport after the reform. The reform was introduced only in some provinces, many treated and untreated provinces have similar pollution trends before the reform and significant overlap on observable characteristics. These features aid me in establishing a causal effect of the reform on misreporting. The estimation involves two steps. First, I quantify different types of misreporting following recently proposed methodologies. Second, I regress estimated misreporting on the reform indicator using a difference in difference estimator. I found that the reform reduces hard misreporting, which takes place when local authorities interfere with the pollution monitoring facility, both during regular days and during heavily polluted days. The reform does not appear to reduce soft misreporting, which takes place when local authorities tamper with the pollution data. The results are robust to a number of robustness tests, and suggest that through proper institutional reform, the upper-level government can prevent certain types of misreporting at the local level. This dissertation delivers a characterization of strategic misreporting by heterogeneous agents and studies the impact of different anti-misreporting schemes based on theoretical, experimental, and observational evidence. Results from this dissertation provide evidence that regulators can use mechanisms that: 1) curb misreporting without enhancing monitoring (tournament audits), or 2) that enhance monitoring to ultimately curb misreporting (adoption of monitoring technologies), or 3) a combination of both. This is important given the pervasiveness of misreporting among regulated agents, and substantial heterogeneity among those agents.

Degree

Ph.D.

Advisors

Sesmero, Purdue University.

Subject Area

Accounting|Finance|Political science

COinS