No Need To Run Millions Of Regressions
Jakob De Haan
University of Groningen - Faculty of Economics and Business; De Nederlandsche Bank; CESifo (Center for Economic Studies and Ifo Institute for Economic Research)
KOF Swiss Economic Institute; CESifo (Center for Economic Studies and Ifo Institute for Economic Research); CESifo (Center for Economic Studies and Ifo Institute for Economic Research) - Ifo Institute for Economic Research
CESifo Working Paper Series No. 288
We argue that in modelling cross-country growth models one should first identify so-called outlying observations. For the data set of Sala-i-Martin, we use the least median of squares (LMS) estimator to identify outliers. As LMS is not suited for inference, we then use reweighted least squares (RLS) for our cross-country growth models. We identify 27 variables that are significantly related to economic growth. Subsequently, applying Sala-i- Martin?s approach for the data set without outliers hardly reveals any additional information. Variables that are insignificant according to the RLS method are generally not significantly related to economic growth under the Sala-i-Martin approach.
Number of Pages in PDF File: 17
Keywords: Sensitivity analysis, outliers, economic growth
JEL Classification: O4, C21, C52working papers series
Date posted: November 5, 2000
© 2013 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo3 in 0.313 seconds