Our revisits the idiosyncratic return momentum effect of Gutierrez and Pirinsky (2007) and Blitz et al. (2011). Idiosyncratic momentum is calculated based on the stock -specific (idiosyncratic) returns, e.g. the returns orthogonal to the three factors that explain a major part of the variation in average returns – the market, size, and value factors.

We conclude that idiosyncratic momentum presents an even bigger challenge to the asset pricing literature, and that the underreaction explanation for the premium seems more likely than the various risk-based and behavioral explanations that have been proposed for conventional momentum.

Our contribution to the literature is fourfold:

  • we show that idiosyncratic momentum is priced in the cross-section of stock returns, and that it cannot be subsumed by any of the established asset pricing factors, including conventional momentum
  • we review some of the most prominent explanations for the sources of the momentum premium, such as non-linear crash risk, investor overconfidence, and overreaction, and show that none of them explain the superiority of idiosyncratic over conventional momentum
  • we bolster the link between idiosyncratic momentum profits and underreaction by showing that idiosyncratic momentum forecasts high long-term excess returns, and that it can be used to differentiate between high momentum stocks whose future returns reverse, and those whose do not
  • we provide evidence for the anomaly in markets outside of the U.S. where the effect was originally discovered

Abstract

Sorting stocks into portfolios on their idiosyncratic, as opposed to total past returns generates returns that are anomalous from the standpoint of any commonly employed factor asset pricing model. Our contribution to the literature is fourfold: (i) we show that idiosyncratic momentum is priced in the cross-section of stock returns, and that it cannot be subsumed by any of the established asset pricing factors, including conventional momentum; (ii) we review some of the most prominent explanations for the sources of the momentum premium, such as non-linear crash risk, investor overconfidence, and overreaction, and show that none of them explain the superiority of idiosyncratic over conventional momentum; (iii) we bolster the link between idiosyncratic momentum profits and underreaction by showing that idiosyncratic momentum forecasts high long-term excess returns, and that it can be used to differentiate between high momentum stocks whose future returns reverse, and those whose do not; (iv) we provide evidence for the anomaly in markets outside of the U.S. where the effect was originally discovered.

Introduction

This paper revisits the idiosyncratic return momentum effect of Gutierrez and Pirinsky (2007) and Blitz et al. (2011)1. We contribute to the literature by showing that (i) idiosyncratic momentum is a distinct phenomenon that cannot be explained by the commonly used asset pricing factors, including total return momentum; (ii) overconfidence, overreaction, and risk-based explanations that arguably explain conventional momentum cannot explain idiosyncratic momentum; (iii) the existence of idiosyncratic momentum profits is consistent with the underreaction to news (slow diffusion of information) hypothesis; (iv) idiosyncratic momentum shows robust performance in international equity markets, including the one market where conventional momentum is known to be ineffective – Japan.

Idiosyncratic Momentum

Idiosyncratic Momentum

The momentum effect is one of the most pervasive asset pricing anomalies documented in the financial literature: stocks with highest returns over the past six to twelve months continue to deliver above-average returns in the subsequent period (see Jegadeesh and Titman, 1993, 2001). Momentum strategies are known to exhibit significant dynamic exposures to systematic risk factors (styles). For instance, in bull markets high beta stocks tend to, on average, outperform low beta stocks, and a zero-investment momentum factor has a net positive exposure to the market factor. The opposite happens in bear markets. Such exposures can be particularly hurtful during style reversals: the Fama-French momentum factor returned -83% in 2009, when stocks that had suffered the largest losses during the financial crisis made a strong recovery.

Grundy and Martin (2001) show that dynamic hedging of momentum strategy’s market and size exposures substantially reduces the volatility of the strategy without a loss in return, but Daniel and Moskowitz (2016) show that the superior performance of their strategy is crucially dependent on the fact that they use ex-post factor betas to hedge these exposures. A hedging strategy based on ex-ante betas does not generate the same improvement. Gutierrez and Pirinsky (2007) propose an alternative method to reduce these systematic style tilts by making individual stock returns in
the ranking period orthogonal to the three factors that explain a major part of the variation in average returns – the market, size, and value factors. Using this approach, the authors document that, after a similar performance in the first year after formation, this idiosyncratic momentum strategy continues to generate abnormal returns for years, while the total return momentum strategy reverses strongly. Although their results suggest that the performance difference in the first year after formation is negligible, Blitz et al. (2011) observe that the idiosyncratic momentum strategy exhibits only half of the volatility of the conventional momentum strategy without a significant reduction in return, thus doubling the Sharpe ratio of the strategy. However, neither Gutierrez and Pirinsky (2007) nor Blitz et al. (2011) address one of the fundamental asset pricing questions, namely, if idiosyncratic momentum is a distinct factor that expands the efficient frontier comprised of already documented factors, particularly if conventional momentum is included.

This paper provides strong evidence that idiosyncratic momentum is a distinct phenomenon. Using a set of time-series, cross-section, and factor-spanning test, we show that idiosyncratic momentum cannot be explained by any of the established asset pricing factors, such as market, size, value, operating profitability, and investment, even if the total return momentum factor is included. In fact, idiosyncratic momentum subsumes total return momentum in some tests, while the converse is never the case. We conjecture that idiosyncratic momentum presents an even greater challenge to the asset pricing literature than conventional momentum.

When examining the importance of factors on which stock excess returns are orthogonalized in order to obtain idiosyncratic momentum scores, we find that the market is by far the most important one. This should not come as a surprise as it is the factor with the highest risk premium, volatility, and power in explaining the variation in returns in the time series. Adding size (SMB) and value (HML) factors further enhances risk-adjusted returns of the strategy, but RMW and CMA add value only marginally.

Idiosyncratic Momentum

We also provide a fresh perspective on the various explanations for the momentum phenomenon that have been put forwarded in the literature. These include investor overconfidence, investor over-and under-reaction, as well as risk-based explanations. Gutierrez and Pirinsky (2007) argue that idiosyncratic momentum is an underreaction phenomenon, caused by gradual diffusion of information, given their finding that abnormal returns do not reverse over multi-year holding periods. Prior research has established links between conventional momentum profits and investors’ overconfidence, overreaction, and risk-based explanations. If idiosyncratic momentum is a distinct phenomenon that is driven by something else, such as investor underreaction, one would expect such links to be absent, or at least much less pronounced. If, on the other hand, idiosyncratic momentum and total return momentum are driven by the same underlying market mechanisms, the superiority of idiosyncratic might simply be due to more extreme exposures to these sources. We empirically test this and reject the latter hypothesis, i.e. we find that the strong link between conventional momentum and investors’ overconfidence, overreaction, as well as risk-based explanations, is much weaker for idiosyncratic momentum.

Our results support the underreaction hypothesis. First, we find that, controlling for other known predictors of stock returns in the cross-section, idiosyncratic momentum forecasts high short and long-term excess returns, while conventional momentum forecasts high short-term, and negative long-term excess returns. Second, as conventional and idiosyncratic momentum strategies are positively correlated, we argue that one can use idiosyncratic momentum as a signal to distinguish between momentum stocks with high future returns, that are more likely to be caused by underreaction, and those whose returns reverse, consistent with initial overreaction and long-term reversal. Congruous with this, we show that a portfolio that is long idiosyncratic momentum winners and short losers within past conventional momentum winners generates high returns continually over the next five years. On the other hand, a portfolio that is long conventional momentum winners and short losers within past idiosyncratic momentum winners generates negative long-term returns.

The final contribution of this paper is to document that idiosyncratic momentum shows robust out-of-sample performance in international markets, including Japan, where conventional momentum does not work, at least unconditionally, therefore causing data mining concerns. Chaves (2016) builds on the work of Blitz et al. (2011) using a simplified definition of idiosyncratic momentum2 and finds evidence for the effect in 21 developed countries, in addition to the U.S. We, on the other hand, use the original definition and apply it uniformly in all considered regions, including emerging markets. The work of Chaves (2016) also shows that the effect is robust to the methodological choices that we make.

By David Blitz, Matthias X. Hanauer & Milan Vidojevic, read the full article here.