Inference in limited dependent variable models robust to weak identification

Abstract

We propose tests for structural parameters in limited dependent variable models with endogenous explanatory variables. These tests are based upon the generalized minimum distance principle. They are of the correct size regardless of whether the structural parameters are identified and are especially appropriate for models whose moment conditions are non-linear in the parameters. Moreover, they are computationally simple, allowing them to be implemented using a large number of statistical software packages. We compare our tests to Wald tests in a simulation experiment and use them to analyse the female labour supply and the demand for cigarettes. Copyright (C) 2010 The Author(s). The Econometrics Journal (C) 2010 Royal Economic Society

Isolating Causality between Gender and Corruption: An IV Approach

“…After simultaneity is addressed, the positive relationship between FLFP and corruption remains powerful, with the respective coefficient settling on a value of 0.003. Magnusson (2010), the weakiv command is applied in Stata.…”

Education and Freedom of Choice: Evidence from Arranged Marriages in Vietnam

“…The estimated marginal effect of education is 13 percentage points, exactly equal to the common estimate from IV Probit, 2SLS, and GMM. Second, we use the recently developed tests for weak IV robust inference in binary choice models with heteroskedasticity developed by Magnusson (2008), and Finlay and Magnusson (2009). Panel B in Table 3 reports the results from the Anderson-Rubin (A-R) tests, the null of no effect of education on arranged marriage is rejected at the 1 percent level (P-value=0).…”

“…Thus weak IV bias is a concern for the male subsample. To address this concern, we report estimates using CUE-GMM estimator which is robust to weak instruments, and also report results from Finaly- Magnusson (2008) weak IV robust inference procedure.…”

Education and Freedom of Choice: Evidence from Arranged Marriages in Vietnam

“…The estimated marginal effect of education is 13 percentage points, the same as the common estimate from IV Probit, 2SLS, and GMM. Second, we use the recently developed tests for weak IV robust inference in binary choice models with heteroskedasticity developed by Magnusson (2008) and Finlay and Magnusson (2009). Panel B in Table 3 The results show that the estimated bounds do not vary significantly with the value of δ.…”