Commands for testing conditional moment inequalities and equalities

• Donald W. K. Andrews ^[1] ; Wooyoung Kim ^[2] ; Xiaoxia Shi ^[2]
  1. [1] Yale University

     Yale University

     Town of New Haven, Estados Unidos

     
  2. [2] University of Wisconsin–Madison

     University of Wisconsin–Madison

     City of Madison, Estados Unidos

     
• Localización: The Stata journal, ISSN 1536-867X, Vol. 17, Nº. 1, 2017, págs. 56-72
• Idioma: inglés
• DOI: 10.1177/1536867x1701700104
• Texto completo no disponible (Saber más ...)
• Resumen

  • In this article, we present two commands (cmi_test and cmi_interval) to implement the testing and inference methods for conditional moment inequality or equality models proposed in Andrews and Shi (2013, Econometrica 81: 609–666). The cmi test command tests the validity of a finite number of conditional moment equalities or inequalities. This test returns the value of the test statistic, the critical values at significance levels 1%, 5%, and 10%, and the p-value. The cmi_interval command returns the confidence interval for a one-dimensional parameter defined by intersection bounds. We obtain this confidence interval by inverting cmi_test. All procedures implemented are uniformly asymptotically valid under appropriate conditions (specified in Andrews and Shi [2013])