On the effect of prior assumptions in Bayesian model averaging with applications to growth regression

• Published in: Journal of applied econometrics (Chichester, England) Vol. 24; no. 4; pp. 651 - 674
• Main Authors: Ley, Eduardo, Steel, Mark F.J.
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.06.2009; John Wiley & Sons; Wiley Periodicals Inc
• Subjects: Aggregate Productivity; Bayesian analysis; Bayesian method; Bayesian networks; Cross-Country Output Convergence O470; Datasets; Economic growth; Economic growth models; Economic models; Economic theory; Forecasting Methods; Forecasts; Inference; Linear models; Linear regression; Measurement of Economic Growth; Model Construction and Estimation C510; Modeling; Observation; Predictive modeling; Probabilities; Probability; Regression analysis; Simulation Methods C530; Single Equation Models; Single Variables: General C200; Steels; Studies; Variance
• ISSN: 0883-7252; 1099-1255
• DOI: 10.1002/jae.1057

We consider the problem of variable selection in linear regression models. Bayesian model averaging has become an important tool in empirical settings with large numbers of potential regressors and relatively limited numbers of observations. We examine the effect of a variety of prior assumptions on the inference concerning model size, posterior inclusion probabilities of regressors and on predictive performance. We illustrate these issues in the context of cross-country growth regressions using three datasets with 41-67 potential drivers of growth and 72-93 observations. Finally, we recommend priors for use in this and related contexts.