Can common risk factors explain infrastructure equity returns? Evidence from European capital markets

Abstract

This is the first paper to test the ability of conventional asset pricing models to explain the excess returns of European infrastructure stocks. Specifically, we firstly run the well-known Fama and French three-factor model, including three common stock market factors (market risk, size risk and value risk), and subsequently augment the model with two common bond risk factors (term and default risk), as infrastructure firms should be closely related to bond markets. The times-series regressions span the period from July 1992 to June 2014 and are conducted using an individually created infrastructure equity data-set. With the help of an intensive screening process, we only include those infrastructure stocks that in fact own and/or operate physical infrastructure. The results reveal that the three-factor model is unable to capture most of the variation in infrastructure returns. Therefore, bond risk factors should be included in asset pricing models in order increase the goodness of fit, as infrastructure stocks prove to be sensitive to interest rate changes. Nevertheless, even the augmented asset pricing model leaves a substantial part of the variance unexplained, thus indicating that infrastructure firms exhibit a high level of idiosyncratic risk. In addition, the results suggest that there may be further risk factors which should be investigated in future studies.

Keywords:

• Infrastructure investments
• asset pricing
• systematic risk
• Fama and French
• common risk factors

Notes

1. Source: www.privatizationbarometer.net; retrieved on July 5, 2015.

2. SIC codes: 4812, 4813, 4822, 4832, 4833, 4841, 4899, 4011, 4013, 4111, 4013, 4111, 4119, 4173, 4231, 4491, 4493, 4581, 4612, 4613, 4619, 4785, 5171, 4911, 4922, 4923, 4924, 4925, 4931, 4932, 4924, 4925, 4931, 4932, 4939, 4941, 4952, 4961, 4971.

GICS codes: 50101010, 50101020, 50102010, 25401025, 20304010, 20305010, 20305020, 20305030, 10102040, 55101010, 55102010, 55103010, 55104010, 55105010.

3. We deem this approximation as sufficient for our research purposes. Firstly, other asset pricing studies with a European context have used US data in order to derive the default premium (e.g. Bauer, Cosemans, and Schotman, 2010). Secondly, the two bond risk indices yield comparable descriptive statistics and high correlation.

4. For reasons of brevity, these results are not reported.