Image credit: UnsplashAbstract
This paper introduces measurement error to the Revealed Preference test of the Cournot Model building on Carvajal et al. (2013, Econometrica, 2351-2379). In contrast to the standard approach, the Revealed Preference analysis relies only on shape restrictions, i.e., the convexity of the firm’s cost function, but otherwise, no parametric assumptions are needed. In the application for the international market of crude oil, measurement errors might arise in the production quantities data due to the consolidation of information or coordination mistakes. Once measurement error is incorporated, the method requires a centering condition. Here, I assume that the quantity mismeasurement is uncorrelated with prices. In contrast to the deterministic version of the test, the Cournot model hypothesis can no longer be rejected.
