Hjalmarsson (2010) considers an OLS-based estimator of predictive panel regressions that is argued to be mixed normal under very general conditions. In a recent paper, Westerlund et al. (2016) show that while consistent, the estimator is generally not mixed normal, which invalidates standard normal and chi-squared inference. The purpose of the present paper is to study the consequences of this theoretical result in small samples, which is done using both simulated and real data.
|Number of pages||11|
|Journal||Journal of International Financial Markets, Institutions and Money|
|Publication status||Published - Nov 2016|
- Common factors
- Mixed normality
- Panel data
- Predictive regression