The momentum effect: omitted risk factors or investor behaviour? Evidence from the Spanish stock market

Abstract

In this paper we use generally applicable non-parametric methods in an attempt to sort out the possible sources of momentum in stock markets (behavioural theories or omitted risk factors). Specifically, we present the results of bootstrap analysis and stochastic dominance tests for the Spanish stock market. Our results from the bootstrap analysis are found to depend on the resampling method used (with or without replacement). Nevertheless, the various stochastic dominance techniques applied lead us to the same conclusion, namely that the winner portfolio stochastically dominates the loser portfolio, which is not consistent with the general asset-pricing models developed for risk-averse investors. This promotes interest in analysing theories that relax the unbounded rationality assumptions that support many of the classical asset-pricing models.

Keywords:

• Market efficiency
• Behavioural finance
• Empirical asset pricing
• Anomalies in prices

Acknowledgements

The authors would like to thank the helpful comments made by the anonymous referees. This paper received financial support from the ERDF and the Spanish Ministry of Science and Technology (SEC2003-07808-C03) and from the Spanish Ministry of Education and Science (SEJ2006-14809). Financial support from FUNCAS is also gratefully acknowledged (‘Programa de Estímulo a la Investigación’ WP No. 252). This article received the BME Spanish Exchanges, prize for the best article of the year 2006 on ‘Variable Income’ at the Spanish Finance Association's Fourteenth Finance Forum.

Notes

†Most of the past evidence regarding the existence of the momentum effect in the Spanish stock market is to be found in the research of Forner and Marhuenda (2003, 2006) and Muga and Santamaría (2007a, c). There is also evidence, consistent with that obtained for other markets, to suggest that abnormal returns to momentum strategies cannot be explained by the traditional risk assessment models (Forner and Marhuenda 2006, Muga and Santamaría 2007a). The explanation is still not entirely convincing, even when asymmetric risk factors are added (Muga and Santamaría 2007c). Clear findings also fail to emerge from tests of the implications drawn from the different behavioural models (Forner and Marhuenda 2004, Muga and Santamaría 2007a). This is due both to the type of portfolio analysis that is required and to the relatively small number of stocks listed on the Spanish stock market.

†Because of the peculiarities of the bootstrap analysis selected for this study, it was performed only for the time period January 1991 to December 2000 and not for the entire sample, as will be explained later.

‡Data for the instrumental variables, DY and TERM, were supplied by Belén Nieto of the University of Alicante.

§This is in line with the procedure used by Forner and Marhuenda (2006).

†Forner and Marhuenda (2006) find monthly returns ranging between 0.5% and 1.3% for January 1965 to December 2000.

‡HML and SMB factors are only available from 1982. For this reason the adjusted returns using the Fama–French model have been computed for the 1982–2004 period. The momentum raw returns for this period are very similar to those obtained for the whole period (1973–2004). These results are available from the authors upon request.

§Jegadeesh and Titman (2001) obtain a momentum return of 1.39% per month for the US in the 1990s and Rouwenhorst (1998) 1.16% per month for a sample of developed markets.

†In Karolyi and Kho (2004), moreover, the above models are extended with a GARCH-type structure. This was not done in our case, however, because the parameters for most of the selected assets were non-significant using monthly data.

†The reason for this choice of time period was that data are required for a minimum number of stocks, for portfolio diversification reasons, and a long enough time horizon, to avoid as far as possible the small sample bias to which bootstrap analysis is susceptible.

‡This table presents the returns obtained through the various bootstrap simulations for each of the momentum strategies, a first p value, p-value(1), showing the percentage of simulations that have yielded a negative return, and a second p value, p-value(2), showing the percentage that have outperformed the original sample.

§These results are consistent with those reported by Forner and Marhuenda (2004) using a similar procedure.

†Fong et al. (2005) were the first to apply stochastic dominance techniques to analyse the momentum effect and to test whether the loser portfolio was stochastically dominated by the winner portfolio.

†The test is robust to the presence of heteroskedasticity in the data series (Chow 2001).

†Although, in this case, the significance level for the J = 3, K = 3 strategy is 10%.

†The critical value for the case at hand is 2.81 for the 5% level of significance.