Prior studies on stock return predictability generally use a time-invariant prediction model to predict stock returns based on lagged predictor variables. This approach has played an important role in the development of asset allocation strategies. However, recent studies indicate that predictions allowing for model instability can provide evidence suggesting more practical investment opportunities for investors (e.g., Henkel, Martin & Nardari, 2011; Kinnunen, 2013; Marfatia, 2014; Boamah et al, 2017a). Structural breaks in the parameters that relate stock returns to predictor variables can occur for a number of reasons, including major changes in market sentiment as well as monetary policies and institutions. Macroeconomic shocks (e.g., unexpected inflation) that affect economic growth or risk premia may also contribute to these breaks. Additionally, if return predictability is partly a result of market inefficiency rather than mere time-varying risk premia, such a predictive relationship should disappear provided that sufficient capital is available to exploit the opportunity. 1 The presence of breaks is important because it fundamentally affects the extent of return predictability and introduces new sources of investment risk. As evidenced by Pettenuzzo and Timmermann (2011), model instability has a large effect on a long-run investor’s optimal asset allocation. Failure to consider structural breaks in modeling stock returns could yield misleading results that incur significant losses for investors. Despite its evident importance, model instability within a stock return forecasting context has received limited attention in the existing literature (Pettenuzzo & Timmermann, 2011). Instead of formal tests, instability is typically addressed by examining stock return prediction models across various subsamples, with the results showing time-varying predictable patterns. For example, Mohanty, Nandh and Bota (2010) uncover instability in return models, based on changes in oil prices, when data from the recent global financial crisis is added to the sample. Similarly, Fayyad and Daly (2011) detect the increased forecasting ability of oil prices for stock returns during the global financial crisis when considering structural breaks. Although econometric tests allowing for breaks (e.g., Chow, 1960; Brown et al., 1979; Andrews, 1993; Andrews et al., 1996; Bai & Perron, 1998, 2003; Elliott & Mueller, 2003) have been widely used in the existing literature, they have been rarely considered in examining stock return predictability.