WHAT TO DO (AND NOT TO DO) WITH TIME-SERIES CROSS-SECTION DATA

WHAT TO DO (AND NOT TO DO) WITH TIME-SERIES CROSS-SECTION DATA

Vol. 89, No. 3 September 1995 | NATHANIEL BECK University of California, San Diego JONATHAN N. KATZ California Institute of Technology
Nathaniel Beck and Jonathan N. Katz examine the estimation of time-series cross-section (TSCS) models, questioning the conclusions of many published studies, particularly in comparative political economy. They critique the generalized least squares (GLS) approach of Parks, which they argue produces standard errors that lead to extreme overconfidence, often underestimating variability by 50% or more. They propose an alternative estimator of standard errors that is more accurate when the error structures are complex, as found in TSCS models. Monte Carlo analysis shows that their "panel-corrected standard errors" perform well. The authors demonstrate the utility of their approach by reanalyzing a "social democratic corporatist" model by Hicks and Swank, which they find to be inconsistent with their findings when using corrected standard errors. They conclude that researchers should avoid the Parks method and instead use ordinary least squares (OLS) with panel-corrected standard errors, which are more accurate and reliable.Nathaniel Beck and Jonathan N. Katz examine the estimation of time-series cross-section (TSCS) models, questioning the conclusions of many published studies, particularly in comparative political economy. They critique the generalized least squares (GLS) approach of Parks, which they argue produces standard errors that lead to extreme overconfidence, often underestimating variability by 50% or more. They propose an alternative estimator of standard errors that is more accurate when the error structures are complex, as found in TSCS models. Monte Carlo analysis shows that their "panel-corrected standard errors" perform well. The authors demonstrate the utility of their approach by reanalyzing a "social democratic corporatist" model by Hicks and Swank, which they find to be inconsistent with their findings when using corrected standard errors. They conclude that researchers should avoid the Parks method and instead use ordinary least squares (OLS) with panel-corrected standard errors, which are more accurate and reliable.
Reach us at info@study.space