| 讲座简介： | In this paper, we propose statistical tests for weighted functional inequalities with estimated weights. Such null hypotheses have proven highly relevant in settings where subjects' response times may convey information about important variables that are typically unobservable (e.g., utilities, happiness, or political attitudes). Our tests, implemented through an easy-to-use multiplier bootstrap procedure, have reasonable power and are consistent. Applying our method to revealed preference analysis, we show that it pinpoints the subject's preference relation in 68.83% of cases with choice reversals. Moreover, the out-of-sample predictions based on our test results have a remarkable accuracy rate of 96%. |
