What causes Momentum Returns? Evidence from different Asset Classes

Table of Contents

Table of Contents

List of Abbreviations

1. Introduction

2. Momentum as a Capital Market Anomaly
2.1 Evidence from Equity Markets
2.2 Evidence from other Asset Classes

3. Explanatory Approaches
3.1 Explanatory Approaches for Equity Markets
3.2 Explanatory Approaches for other Asset Classes

4. Cross-Asset Explanatory Approaches
4.1 Approaches in Accordance with the Efficient Market Hypothesis
4.2 Approaches in Accordance with Behavioral Finance

5. Transferability of Explanatory Approaches
5.1 Contrary Observations
5.2 Equal Approaches

6. Conclusion

References

List of Abbreviations

Abbildung in dieser Leseprobe nicht enthalten

1. Introduction

Capital market anomalies are a phenomenon triggering ongoing debates about the trading behavior of investors on financial markets (Asness et al., 2013, p. 929). They are contradicting the core ideas of the efficient market hypothesis (EMH), which considers financial markets efficient and investors rational and fully informed. One of the EMH key hypotheses, especially supported by Fama and by Samuelson, is the principle of random walk (Fama, 1965, pp. 34ff.; Samuelson, 1965, pp. 41ff.). If this principle holds, the prices of assets on financial markets are only influenced by public and firm-specific news, develop apart from that completely random and are not predictable. However, empirical observations question the random walk principle. They tend to in­dicate specific patterns in asset price developments instead of complete ran­domness. Doubts on the EMH and the random walk principle thus cannot be neglected. A common answer to these observations is the existence of addi­tional risk factors which are currently not covered by the applied pricing mod­els. Current asset pricing models mainly rely on Markowitz (1952) and the modern portfolio theory as well as on the Capital Asset Pricing Model (CAPM) from Sharpe (1964), Lintner (1965), and Mossin (1966). These mod­els are rather a benchmark for asset pricing than perfect constructions cover­ing all and any existing risk factors which are relevant for an assets price formation. This essay deals with the potential causes of one of the observed anomalies, the so-called momentum effect. Fama and French (2004) bring attention to the topic as they consider the momentum anomaly “the most se­rious problem” of their three-factor model (p. 40). This paper discusses pos­sible asset-specific and cross-asset explanation approaches for momentum appearance, but it does not provide a final or unique answer. Although under­and overreaction can be seen as key parts of the momentum explaining liter­ature, other possible explanations can by no means be considered less im­portant contributors. The two main threads in literature, stock momentum and momentum of other assets, are discussed separately and subsequently checked for overlaps. The paper also deals with the definition of momentum as an anomaly itself in context of rational and behavioral concepts. It uncovers selected contrary observations and outlines possible conformities. The following structure leads through the paper: Chapter two points out the momentum anomaly and its evidence for stocks and four additional asset clas­ses (commodities, bonds, currencies, and indices). Chapter three provides ex­planatory approaches for momentum in different asset classes, including stocks. Chapter four separates cross-asset explanatory approaches based on their accordance with the EMH. Chapter five examines the generality and transferability of the explanatory approaches for equity momentum. Chapter six constitutes a brief conclusion and an outlook.

2. Momentum as a Capital Market Anomaly

The momentum effect is an often-discussed anomaly observable on financial markets and “[...] probably the most difficult to explain within the context of the traditional risk-based asset pricing paradigm.” (Jegadeesh and Titman, 2011, p. 494). Assets which performed well in the past show significant extra profits in the subsequent period as well. Momentum-based trading strategies exist in retrospect and exploit assumed future return predictability. There are two different approaches to momentum research. A time series observation finds extra profits of the same financial asset over a specific period (Mos­kowitz et al., 2012, pp. 228ff.). The financial assets price is higher than justi­fied by the fundamental data. The alternative approach observes cross-sec­tional momentum. That means a financial asset yields extra profits in relation to comparable financial assets. Time series momentum depicts absolute extra profits from one specific asset, while cross-sectional momentum depicts rel­ative extra profits in relation to similar assets. To proof whether an anomaly really exists, empirical studies have to check their results against potentially existing additional risk factors. Their existence would provide a rational ex­planation and thereby dampen the assumption of an anomaly. Fama and French (1996) define an “anomaly” as a return pattern that is not explained by the CAPM (p. 55). In this essay, an anomaly on financial markets is addi­tionally characterized as an observable price pattern whose past price devel­opment provides hints for the future price development of an asset.

2.1 Evidence from Equity Markets

Momentum in equity markets is a quite robust observation, already stated by Levy in his dissertation from 1967. He outlined that stocks, whose prices are significantly higher than their average of the past 27 weeks, gain significant extra profits (pp. 595ff.). This hypothesis faced critics from Jensen and Ben­nington in 1970, who accused him of a selection bias since he examined 68 different trading strategies in his paper. Both examined the momentum effect within a longer period than Levy did and did not find any profits significantly higher than the buy-and-hold benchmark portfolio (pp. 469ff.). In their paper from 1993, Jegadeesh and Titman (J&T) provide empirical evidence for the existence of momentum in the US stock market while conducting a so called relative-strength-strategy. Within the years from 1965 to 1989, stocks listed at the New York Stock Exchange (NYSE) and American Stock Exchange (AMEX) show significant extra returns over a holding period from three to twelve months after portfolio formation. The compiled portfolio contains re­cent winner stocks which outperformed in the last three to twelve months ahead of the portfolio formation as long positions. Additionally, it holds loser stocks which underperformed within the same period as short positions. If the first week after portfolio formation was left out to avoid noises, such as bid ask spread or short-term reversal, returns even got higher. To underline the existence of an anomaly, alternative explanations such as systematic risk or delayed reactions to non-firm specific factors had been eliminated (J&T, 1993, pp. 65ff.). Cooper et al. (2004) confirm this existence of momentum in the US stock market and preclude macroeconomic driving factors (pp. 1345ff.). In their follow-up study from 2001, J&T confirm the results from 1993 within the consecutive years in the 1990s and thus reduce the potential influence of data mining, thus increasing the robustness of the findings (pp. 699ff.). A more international picture was drawn from Asness et al. in 2013. Return premia have been revealed in stock markets of the USA, United King­dom, Europe, and Japan. All results except those of Japan show statistically significant results. The first month after the portfolio formation was left out, to minimize the influence of short-term reversal (pp. 929ff.). These results are in line with the outcome of Fama and French in their paper from 2012, which examines momentum return patterns besides size and value effects in the same stock markets (pp. 457ff.). Momentum effects are negatively correlated to firm size and not significantly observed for Japan either (Fama and French, 2012, pp. 457ff.; Hong et al., 2000, p. 265; Chui et al., 2010, pp. 361ff.). Rouwenhorst also examines the observation of momentum return develop­ments in an international context in his paper from 1998. Twelve European countries, such as France, Germany, the United Kingdom, and Italy, had been examined. Extra returns in all countries during a period from 1980 to 1995 are found. One internationally diversified portfolio consists of past winners and outperforms another portfolio that consists of past losers by about one percent-point per month. The abnormal returns remain for about one year af­ter portfolio formation. The effect is negatively related to firm size and overall matches with the results of J&T from 1993. The paper outlines a correlation between European and US relative strength strategies, hinting at a common factor which influences international momentum patterns (pp. 267ff.).

2.2 Evidence from other Asset Classes

Evidence for momentum of the four asset classes indices, currencies, com­modities, and bonds traces back to the 19th century. Hurst et al. (2017) outline momentum return patterns over a period from 1880 to 2016 (pp. 15f.). With slightly different asset combinations, the results are confirmed by Asness et al. (index futures replace indices) across markets, Moskowitz et al. (bond fu­tures replace bonds) for time-series momentum, and Jo and Kim (equity re­place indices) for recent time-series momentum from 1984 to 2017 (Asness et al., 2013, pp. 929ff.; Moskowitz et al., 2012, p. 228; Jo and Kim, 2019, pp. 767ff.). Moskowitz et al. additionally state that the “[...] correlation among time series momentum returns is stronger than the correlation of passive long positions across the same asset classes [.]” (2012, p. 249). For commodities in particular, Erb and Harvey (2006) point out excessive returns of the GSCI commodity futures index during the 1982 to 2004 period due to momentum effects (pp. 1ff.). Novy-Marx reveals commodity short-term momentum regarding a one-month horizon (2012, p. 446). Regarding currencies, Okunev and White (1990) provide findings indicating profitable momentum trading and rule out long-term drift, cross-correlation, and different risk levels as po­tential sources of the return pattern (pp. 425ff.). Menkhoff et al. (2012) con­duct a buy winner and sell loser strategy over a period from 1976 to 2010, for 48 countries. They find extra returns of up to ten percent per year, while about 50 % of the extra returns derive from “minor currencies” (pp. 682f.). Novy- Marx finds currency momentum for a short one-month horizon (2012; p. 447). Jostova et al. (2013) reveal “strong evidence” for momentum in the US corporate bond market in a period from 1991 to 2011 (p. 1690). The winner bonds over the last 6 months prior to portfolio formation yield 59 basis points (BPS) more than the loser bonds over a holding period of 6 months (p. 1649). Muller and Ward (2010) find significant momentum extra returns for 70 country equity indices depicting the MSCI world index. They compiled a portfolio of the four best performing indices over the last eleven months with a holding period of one month. This portfolio yielded extra profits of about ten percent annually over the period from 1970 to 2009 (p. 111). Novy-Marx confirms country indices momentum for one-month horizons (2012; p. 445). The evidence for momentum in other asset classes than stocks is not as clear as it seems to be while considering the literature. For example, Okunev and White (1990) tend to refute market efficiency regarding their currency trading strategy with the technique of J&T (pp. 425ff.). In contrast, the findings of Pukthuanthong-Le et al. (2007) indicate a drift of currency momentum strat­egies from “major” to “exotic” currencies (pp. 114ff.). The latter might indeed imply some form of market correcting behavior, since “exotic” currencies seem to be less liquid and therefore seem to enable lower correcting arbitrage opportunities than the highly intensive traded “major” currencies. However, even these “exotic” currencies show return declines, possibly pointing to an arising weak form of market efficiency (Pukthuanthong-Le et al., 2007, pp. 114ff.). In line with this reasoning are the findings from Jostova et al. (2013), pointing out momentum extra returns only with non-investment grade (NIG) bonds after checking for interest rate risk, credit risk, trading frictions, and microstructure noise (p. 1650, pp. 1666, p. 1690).

[...]