Model Averaging and Its Use in Economics

Abstract

The method of model averaging has become an important tool to deal with model uncertainty, for example in situations where a large amount of different theories exist, as are common in economics. Model averaging is a natural and formal response to model uncertainty in a Bayesian framework, and most of the paper deals with Bayesian model averaging. The important role of the prior assumptions in these Bayesian procedures is highlighted. In addition, frequentist model averaging methods are also discussed. Numerical techniques to implement these methods are explained, and I point the reader to some freely available computational resources. The main focus is on uncertainty regarding the choice of covariates in normal linear regression models, but the paper also covers other, more challenging, settings, with particular emphasis on sampling models commonly used in economics. Applications of model averaging in economics are reviewed and discussed in a wide range of areas including growth economics, production modeling, finance and forecasting macroeconomic quantities. (JEL C11, C15, C20, C52, O47).

A Conceptual Introduction to Bayesian Model Averaging

“…Despite the intuitive application of BMA in examples from daily life, it is severely underutilized in the social-sciences literature (for noteworthy exceptions, see Gronau et al, 2017; Kaplan & Lee, 2018; for recommendations and guidelines, see Appelbaum et al, 2018), even though it is regularly employed in other disciplines (see Fragoso, Bertoli, & Louzada, 2018, for a review; see Steel, in press, for a survey on BMA in economics). This mismatch suggests that researchers in the social sciences remain unaware of the advantages of adopting Bayesian statistics (Vandekerckhove, Rouder, & Kruschke, 2018) or do not have access to easy-to-use software that implements BMA and other Bayesian methods.…”

Does Education Really Improve Health? A Meta‐analysis

“…Having the wrong variables in the equation leads to misspecification bias and invalid inference. Model selection and model averaging are two popular strategies employed in the literature to address model uncertainty (Steel, 2020). The most frequently used model selection is stepwise regression.…”

Is there a link between temperatures and COVID-19 contagions? Evidence from Italy

“…In FMA, the parameters are treated as fixed, yet unknown, and are not assigned any probabilistic interpretation associated with prior knowledge or learning from data. To date, the empirical literature of FMA has mainly focused on forecasts combinations ( Steel, 2017). However, as we show below, this approach can also be employed to derive metrics of great interest to analyze the importance of the potential drivers of COVID-19 differentials.…”