Showing papers by "Serena Ng published in 2010"
TL;DR: In this article, a regression model with endogenous regressors and common factors is considered, and the authors show that the estimated common factors can be used as instrumental variables and they are more efficient than the observed variables in their framework.
Abstract: We consider estimation of parameters in a regression model with endogenous regressors. The endogenous regressors along with a large number of other endogenous variables are driven by a small number of unobservable exogenous common factors. We show that the estimated common factors can be used as instrumental variables and they are more efficient than the observed variables in our framework. Whereas standard optimal generalized method of moments estimator using a large number of instruments is biased and can be inconsistent, the factor instrumental variable estimator (FIV) is shown to be consistent and asymptotically normal, even if the number of instruments exceeds the sample size. Furthermore, FIV remains consistent even if the observed variables are invalid instruments as long as the unobserved common components are valid instruments. We also consider estimating panel data models in which all regressors are endogenous but share exogenous common factors. We show that valid instruments can be constructed from the endogenous regressors. Although single equation FIV requires no bias correction, the faster convergence rate of the panel estimator is such that a bias correction is necessary to obtain a zero-centered normal distribution.
176 citations
TL;DR: This article used PANIC residuals to estimate the pooled autoregressive coefficient and a sample moment, and established their large-sample properties using a joint limit theory and showed that PANIC-based pooled tests have nontrivial power.
Abstract: An effective way to control for cross-section correlation when conducting a panel unit root test is to remove the common factors from the data However, there remain many ways to use the defactored residuals to construct a test In this paper, we use the panel analysis of nonstationarity in idiosyncratic and common components (PANIC) residuals to form two new tests One estimates the pooled autoregressive coefficient, and one simply uses a sample moment We establish their large-sample properties using a joint limit theory We find that when the pooled autoregressive root is estimated using data detrended by least squares, the tests have no power This result holds regardless of how the data are defactored All PANIC-based pooled tests have nontrivial power because of the way the linear trend is removed
162 citations
TL;DR: In this paper, the same filter is applied to both the data and the model variables and the filtered variables are stationary when evaluated at the true parameter vector, and the estimators are approximately normally distributed not only when the shocks are mildly persistent, but also when they have near or exact unit roots.
49 citations
01 Jan 2010
TL;DR: In this paper, an indirect inference approach that exploits the biases in an auxiliary model to identify the parameters of interests has been proposed, which is non-standard because the covariates cannot be held xed in simulations, and the auxiliary parameters must be chosen to vary with the nuisance parameters causing inconsistency.
Abstract: This paper considers an indirect inference approach that exploits the biases in an auxiliary model to identify the parameters of interests. The proposed augmented indirect inference estimator (IDEA) is non-standard because (i) the covariates cannot be held xed in simulations, and (ii) the auxiliary parameters must be chosen to vary with the nuisance parameters causing inconsistency. We provide a Panel-ME algorithm for mismeasured dynamic panel models that does not require fully specifying the joint distribution of the data. A simple modication leads to a Panel-IV algorithm for models with endogenous variables. A Panel-CS algorithm is also proposed for dynamic panel models with cross-section dependence which, like measurement error models, also have time varying latent components. A Monte Carlo study shows that all three algorithms have impressive nite sample properties.
1 citations