Abstract:This paper considers model selection and model averaging in panel data models with a multifactor error structure. We investigate the limiting distribution of the common correlated effects estimator (Pesaran, 2006) in a local asymptotic framework and show that the trade-off between bias and variance remains in the asymptotic theory. We then propose a focused information criterion and a plug-in averaging estimator for large heterogeneous panels and examine their theoretical properties. The novel feature of the p… Show more
“…This way we can easily compare our results with those of Yin et al. (2021). The results for the pooled estimators are presented in Table 7.…”
Section: Empirical Illustrationsupporting
confidence: 62%
“…As the proposed procedure mostly addresses the way factor proxies are constructed (and corresponding sampling uncertainty), and can be used for any model that uses for factor proxies, for example, Focused Information Criterion based model averaging of Yin et al. (2021), the gravity model of Desbordes and Eberhardt (2019), or the discrete choice model of Boneva and Linton (2017).…”
Section: Discussion Of the Main Resultsmentioning
confidence: 99%
“…In what follows we use the empirical strategy of Yin et al. (2021) to investigate the potential causes of the historically increasing wage inequality between high‐skilled and low‐skilled workers in the US manufacturing industries. The empirical specification they consider is a linear regression model of the form: where the variable is the relative wage of low‐skilled workers to high skilled workers.…”
Section: Empirical Illustrationmentioning
confidence: 99%
“…Among other models, Yin et al. (2021) estimated the “full model” in Equation (), as well as the “narrow model” without any controls using the CCE‐MG estimator. Motivated by their setup we estimate these two specifications also using the regularized versions of the CCEP and the CCE‐MG estimators.…”
Section: Empirical Illustrationmentioning
confidence: 99%
“…In our empirical illustration, we re‐evaluate the results from the recent studies in Voigtländer (2014) and Yin et al. (2021), and investigate the causes of the historically increasing wage inequality between high‐skilled and low‐skilled workers in the US manufacturing industries. Our procedure provides strong evidence that, irrespective of the setup considered, the number of the underlying factors is small in comparison with the total number of cross‐section averages.…”
Summary
Cross‐section average‐augmented panel regressions introduced by Pesaran (2006) have been a popular empirical tool to estimate panel data models with common factors. However, the corresponding common correlated effects (CCEs) estimator can be sensitive to the number of cross‐section averages used and/or the static factor representation for observables. In this paper, we show that most of the corresponding problems documented in the literature can be solved once cross‐section averages are appropriately regularized, thus extending the applicability of the CCE setup. As the standard plug‐in variance estimators are not able to account for all sources of estimation uncertainty, we suggest the use of cross‐section bootstrap to construct confidence intervals. The proposed procedure is illustrated both using real and simulated data.
“…This way we can easily compare our results with those of Yin et al. (2021). The results for the pooled estimators are presented in Table 7.…”
Section: Empirical Illustrationsupporting
confidence: 62%
“…As the proposed procedure mostly addresses the way factor proxies are constructed (and corresponding sampling uncertainty), and can be used for any model that uses for factor proxies, for example, Focused Information Criterion based model averaging of Yin et al. (2021), the gravity model of Desbordes and Eberhardt (2019), or the discrete choice model of Boneva and Linton (2017).…”
Section: Discussion Of the Main Resultsmentioning
confidence: 99%
“…In what follows we use the empirical strategy of Yin et al. (2021) to investigate the potential causes of the historically increasing wage inequality between high‐skilled and low‐skilled workers in the US manufacturing industries. The empirical specification they consider is a linear regression model of the form: where the variable is the relative wage of low‐skilled workers to high skilled workers.…”
Section: Empirical Illustrationmentioning
confidence: 99%
“…Among other models, Yin et al. (2021) estimated the “full model” in Equation (), as well as the “narrow model” without any controls using the CCE‐MG estimator. Motivated by their setup we estimate these two specifications also using the regularized versions of the CCEP and the CCE‐MG estimators.…”
Section: Empirical Illustrationmentioning
confidence: 99%
“…In our empirical illustration, we re‐evaluate the results from the recent studies in Voigtländer (2014) and Yin et al. (2021), and investigate the causes of the historically increasing wage inequality between high‐skilled and low‐skilled workers in the US manufacturing industries. Our procedure provides strong evidence that, irrespective of the setup considered, the number of the underlying factors is small in comparison with the total number of cross‐section averages.…”
Summary
Cross‐section average‐augmented panel regressions introduced by Pesaran (2006) have been a popular empirical tool to estimate panel data models with common factors. However, the corresponding common correlated effects (CCEs) estimator can be sensitive to the number of cross‐section averages used and/or the static factor representation for observables. In this paper, we show that most of the corresponding problems documented in the literature can be solved once cross‐section averages are appropriately regularized, thus extending the applicability of the CCE setup. As the standard plug‐in variance estimators are not able to account for all sources of estimation uncertainty, we suggest the use of cross‐section bootstrap to construct confidence intervals. The proposed procedure is illustrated both using real and simulated data.
We analyze Poisson regression when covariates contain measurement errors and when multiple potential instrumental variables are available. Without empirical knowledge to select the most suitable variable as an instrument, we propose a novel model-averaging approach to resolve this issue. We prescribe an implementation and establish its optimality in terms of minimizing prediction risk. We further show that, as long as one model is correctly specified among all potential instrumental variable models, our method will lead to consistent prediction. The performance of our method is illustrated through simulations and a movie sales example.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.