Where's Waldo? Searching for the Hidden Variable of Corruption

Abstract

The indices of corruption available and used in the literature - government-recorded, perception-based, and experience-based - are all subject to errors in measurement that can lead to biased research results. In this paper, we work to identify the "true" value of corruption hidden by measurement error. We analyze the nature of the measurement errors inherent in frequently used indices of corruption, and we attempt to mitigate the biases that result from these errors to obtain identification and consistent estimators by combining substantive knowledge about corruption with an instrumental variables method for a linear errors-in-variables model.

Keywords: Corruption, Measurement Error