摘要：In this paper we use principal components augmented regressions (PCARs), partly in conjunction with model averaging, to determine the variables relevant for economic growth. The use of PCARs allows to effectively tackle two major problems that the empirical growth literature faces: (i) the uncertainty about the relevance of variables and (ii) the availability of data sets with the number of variables of the same order as the number of observations. The use of PCARs furthermore implies that the computational cost is, compared to standard approaches used in the literature, negligible. The proposed methodology is applied to three data sets, including the Sala-i-Martin, Doppelhofer, and Miller (2004) and Fernandez, Ley, and Steel (2001) data as well as an extended version of the former. Key economic variables are found to be significantly related to economic growth, which demonstrates the relevance of the proposed methodology for empirical growth research.
关键词：economic growth;economic convergence;frequentist model averaging;growth regressions;principal components augmented regression
