The Predictability of Australian Listed Infrastructure and Public-Private Partnership Returns Using Asset Pricing Models


31 AUGUST 2014*

Can asset pricing models predict the future returns of publicly listed infrastructure investments in Australia? This study finds that asset pricing models exhibit poor out-of-sample predictive performance when compared to simple, fixed excess return models for the 1997−2012 period. Consistent with recent US evidence, these results suggest that using the long-term historical mean return may be a reasonable starting point for superannuation funds seeking to understand the long-term expected returns of publicly listed infrastructure.

It has long been recognised that well-designed infrastructure investments deliver long-term benefits to investors and the broader economy alike. A number of studies have suggested that infrastructure investments are low risk due to: their regular income; the nature of long-term contracts in the infrastructure industry; their position in lowly competitive markets due to high barriers to entry; and the higher regulatory constraints in which infrastructure firms operate.

Against this backdrop, we examine the predictive (or otherwise) performance of conventional asset pricing models on Australian publicly listed infrastructure investments returns.

There are three motivations for this kind of analysis. First, several US studies in recent years demonstrate that asset pricing models are poor predictors of US stock returns. There is a scarcity of this type of research in the Australian setting and the predictability of infrastructure and PPP returns has never been examined in the literature. Second, the low-risk perception of infrastructure makes it a perfect candidate to evaluate the predictive performance of Australian asset pricing models. If the risks in infrastructure are lower than conventional equity then there is an increased probability that Australian asset pricing models may be able to forecast infrastructure returns. Third, and perhaps most importantly, respective Australian governments have employed the principles of the Capital Asset Pricing Model (CAPM) in their evaluation of public−private partnerships (PPPs) to finance some infrastructure investments. However, there are a number of studies in Australia and globally that have documented poorly performing infrastructure investments. Governments in Australia employ the Infrastructure Australia guidelines for the development of discount rates for PPP infrastructure projects.

This study enables an assessment of the efficacy of using the CAPM in this public policy setting. The findings suggest that Australian asset pricing models are poor predictors of future publicly listed infrastructure returns. Indeed, a simple fixed excess return model provides similar or better forecasts than conventional asset pricing models in many cases. The initial tests before the 2008 global financial crisis (from 1997 to 2007) show that the conventional asset pricing models (including the CAPM and Fama-French three factor model) deliver similar or lower levels of predictive performance than simple fixed excess return models. Testing from 1997 to 2012 indicates that the best predictor of infrastructure returns is a fixed excess return model of 10 per cent per year.

The results indicate that the predictive performance of the CAPM with no intercept is the best of the conventional asset pricing models examined in this study. However, it is important to note that simple, fixed excess return models outperform the CAPM and Fama−French models.

These findings in the Australian setting have important implications for practitioners. An interesting finding is that the predictive performance of the constant returns models tends to gravitate towards its long-term unconditional historical mean return. It suggests that employing the long-term historical mean return is a reasonable starting point for superannuation funds seeking to understand the long-term expected returns of infrastructure. In short, the evidence to date supports employing a simple historical mean return as this seems to outperform conventional asset pricing models.

These findings provide researchers with a number of avenues for future research. First, while this study is limited to the 16 years of empirical data available on Australian infrastructure returns from 1997 to 2012, similar research on longer term US infrastructure data may provide a better understanding of long-run infrastructure returns. Second, while this analysis employs the standard practice of using a rolling 60-month window to capture the inputs for the asset pricing model, researchers may need to experiment with other time frames to evaluate the crudeness of such a time frame.

*This paper is an updated version of Working Paper 2013-05