A beautiful mind

Category: Asia, Hong Kong, Singapore, Global
By David Macfarlane

AAM-CAMRI Prize victor discusses methodology behind winning white paper

In celebration of excellence in regional applied research, Asia Asset Management (AAM) aligned with NUS Business School’s Centre for Asset Management Research and Investments (CAMRI) for the second year running to present the 2016 AAM-CAMRI Prize in Asset Management.

The winning paper of the 2016 AAM-CAMRI Prize (from close to 60 excellent applications from the around the world) is entitled: Market Maturity and Mispricing, which was authored by Heiko Jacobs from the University of Mannheim in Germany. The author presented his research to an audience of academics and senior practitioners and received his prize at the inaugural AAM-CAMRI ESG event in Singapore, which was held on November 17.

In an exclusive interview with AAM, Mr. Jacobs discusses how his research broke new ground in the thinking, practice, policies and issues affecting the Asian asset management industry.

Asia Asset Management: What prompted you to embark on your white paper research into market maturity and mispricing?

Heiko Jacobs: I noticed that, in their white papers and marketing materials, asset management firms often claim that emerging markets are less efficient than developed markets. However, in the academic literature, there is little evidence for superior stock picking skills of fund managers in emerging markets. In short, the empirical evidence seems far from conclusive. That’s why I aimed at revisiting the controversial debate, which appears to exist both in the financial industry and in academia. My findings pose a challenge to the widespread perception of necessarily stronger cross-sectional mispricing in emerging markets.

How did you go about conducting the research and how long did it take?

The most time-consuming part of the project was to gather and clean the large international stock market data set. My baseline sample covers 45 countries, 42,600 firms, and about 115 million firm days between January 1994 and December 2013. Put simply, I then implemented a composite mispricing score for each firm month, constructed long/short mispricing strategies for each country, and then compared the resulting alpha in countries classified as MSCI developed markets with the alpha in countries classified as MSCI emerging markets. In total, the project took about one-and-a-half years.

What techniques did you use to determine and measure mispricing when conducting the research?

In my baseline analysis, I relied on the cross-sectional composite mispricing metric proposed in Stambaugh, Yu, and Yuan (2015, Journal of Finance). For the US stock market, the authors synthesise the information contained in 11 well-established anomalies such as financial distress, share issuance, accruals, or momentum, in a single mispricing score for each stock month. The basic idea is that if a firm is classified as overvalued or undervalued with respect to many of these individual return predictors, then this is quite a strong mispricing signal. Indeed, the approach turns out to be quite powerful.

Nevertheless, it is important to stress some of its limitations. First, its construction is based on public information only, so it does not speak to the strong form of market efficiency, which is concerned with both public and private information. Second, the approach is purely cross-sectional, so I am not able to say anything about market-level under- or overvaluation, such as “bubbles”.

What were the key findings in the white paper and who do you think will benefit most from them?

There are two main findings. First, there is strong evidence for mispricing around the globe. More specifically, for the average country and based on long/short mispricing quintiles, the equally weighted (value weighted) alpha relative to a country-specific Fama and French (1993) three-factor model is about 107 (84) basis points (bp) per month over the 1994 to 2013 period. Equally weighted alphas give much weight to small firms, value weighted alphas are driven by large firms. Second, and perhaps surprisingly, point estimates of mispricing are often higher in developed markets (as classified by MSCI) than in emerging markets.

These insights may be relevant for academics and industry professionals alike. The findings may help to enrich or challenge our understanding of price formation. Coupled with the assumption of higher transaction costs in emerging markets, my results also suggest that popular trading strategies based on public information tend to be more profitable in developed markets.

How does the research compare to previous studies on mispricing – what are the main common themes and contradictions you identified?

Of course, there is already very extensive literature on mispricing, in particular with respect to individual anomalies (as opposed to a composite mispricing score). However, most papers concentrate on the US stock market only. International stock markets are both academically and economically important, but they appear to be under researched. In this respect, my work differs from many existing papers.

Nevertheless, my global asset pricing tests confirm some of the key findings previously documented in the US market. First, composite mispricing approaches, which condense the information contained in many individual anomalies, are very powerful. More specifically, the alphas generated by such approaches are much higher than the alphas generated by the average individual anomaly. This pattern suggests that considering all this information jointly helps to identify mispricing more precisely and to eliminate much of the noise contained in individual anomalies. Second, abnormal long/short returns generated by such a composite mispricing score are concentrated around earnings announcements. This pattern is consistent with the idea that investors have too optimistic (pessimistic) expectations regarding the stocks in the short (long) leg of the mispricing strategy, and then have to partly update their biased beliefs due to the arrival of fundamentally relevant news.

According to your research, what are the key determinants of mispricing and how can these be minimised?

In general, comparing mispricing across countries is difficult as the level and the cost of information production are hard to measure. In addition, countries differ in many dimensions, including unobserved ones. While by no means conclusive, my results indicate that mispricing could be positively related to average firm-specific return variation, to trading activity, and, to a lesser extent, to analyst forecast dispersion. All of these variables may proxy for sentiment-driven noise trading.

It is nevertheless difficult to assess how mispricing could be minimised. Despite decades of research, it is still an open question why exactly mispricing arises in the first place, and why it appears to persist for some time. For instance, we don’t fully understand yet which investor group is primarily responsible for mispricing, and how important limits to arbitrage are in real markets. Indeed, most of the large cross-country variation in return predictability is currently unexplained.

What further research do you think should be conducted on the subject of mispricing?

Given that the academic literature has proposed hundreds of individual return predictors, there is clearly a need for the meta-analysis of market anomalies, in particular in an international context. In other words, I think we could benefit from research that analyses the “big picture” and develops a better understanding of when, where, which, and why anomalies tend to work (or not work). My current and future research aims at progressing on this front.