Should Factor Investors Neutralize the Sector Exposure?

Factor investors face numerous choices that do not end even after picking the set of factors. For instance, should they neutralize the factor exposure? If the investor pursues sector neutralization, does the decision depend on a particular factor? Or are the choices different for the long-only investor compared to the long-short investor? The research paper by Ehsani, Harvey, and Li (2021) answers these questions and provides investors with an interesting insight on this topic.

According to the literature, the return to the standard factor consists of two components, specifically, the within (firm-specific) component and the across (sector) component. The standard factor sorts stocks based on firm characteristics, while the sector-neutralized factor sorts based on the firm characteristics relative to the industry average (within component). The study results have confirmed that the within component of stock characteristics contains more information about the cross-section of expected returns than the across component. However, a mean-variance investor should form the sector-neutral factor only if the ratio of the Sharpe ratios across and within components is less than their correlation. This condition is frequently met in the long-short portfolios, and therefore, the long-short investor likely gains from sector neutralizing. In contrast, the same condition is unlikely to hold in the long-only portfolios, hence the long-only is better off investing in the factor as it stands. The empirical bootstraps of historical data confirm the study results. In particular, keeping the across component produces better long-short factors in only 20% of the trials, while doing so delivers better long-only factors in 78% of the trials.

Authors: Sina Ehsani, Campbell R. Harvey, and Feifei Li

Title: Is Sector-neutrality in Factor Investing a Mistake?

Link: https://ssrn.com/abstract=3959116

Abstract:

Stock characteristics have two sources of predictive power. First, a characteristic might be valuable in identifying high or low expected returns across industries. Second, a characteristic might be useful in identifying individual stock expected returns within an industry. Past studies generally find that the firm-specific component is the strongest predictor, leading many to sector neutralize their factor exposures. We show that this problem is equivalent to the classic two risky-asset problem and derive the condition that determines when the sector component of a characteristic should be omitted. We show both analytically and empirically that the long–short investor is more likely to benefit from hedging out sector bets, whereas the long-only investor is more likely to benefit from investing in the factor as it stands.

As always we present several interesting figures:

Notable quotations from the academic research paper:

“In this paper, we first confirm that the within (firm-specific) component of stock characteristics contains more information about the cross-section of expected returns than the across (sector) component. We then derive a condition that determines when the weaker component of a predictor should be omitted. Using aggregate values for Sharpe ratios and correlation coefficients we predict that this condition—which identifies whether the across-sector component is redundant—will be met frequently in long–short portfolios; therefore, the long–short investor likely gains from sector neutralizing. In contrast, the same condition is unlikely to hold in long-only portfolios; therefore, long-only factor performance is more likely to degrade as a result of sector neutralizing. Empirical bootstraps of historical factor data, constructed using various portfolio construction techniques, show that our analytical results are accurately reflected in the actual data.

Do you want to test these ideas yourself? We offer our readers Historical Trading Data Discounts.

Over the 60-year period of our study, the standard value factor has consistently invested large amounts in the Finance and Utilities sectors and at the same time has shorted large amounts in the Technology and Healthcare sectors. Exposure to any risky asset, such as equity sectors, comes with volatility. The premium associated with this extra risk, and its covariance with the rest of the portfolio, determines its overall contribution to the factor’s risk–return profile. We can construct the standard value factor with the sector exposures of Figure 3 or we can form a value factor using the within characteristics, such that the overall exposure to any one of the sectors is always zero. The sector-neutral value factor is likely less volatile than the standard value factor because it offsets the positive sector exposure of the long leg with an equal amount of negative exposure in the short leg.

Trading BM within the cross-section of every sector is highly profitable, while trading BM in the cross-section of sectors is not. This observation suggests that the entire predictive power of market-wide BM stems from the information in its firm-specific component. A long–short investor that sorts stocks based on a raw BM signal contaminates the useful within-sector information by the noise of the across-sector component. Once again consider the example of tech firms. The original BM shorts tech because most tech companies are considered to be growth companies when compared to the average company. Therefore, the standard value factor does not exploit the variation of BM within the Tech sector because tech stocks are all sent to the short leg of the portfolio.

Our analysis has three caveats. First, our empirical results based on the mean-variance framework are ex post. That is, sector neutralization of long–short (long-only) factors is generally beneficial (detrimental) based on an ex post, historical analysis. Whereas the empirical analysis is ex post, our framework based on the Sharpe ratios of the factor within-industry and across-industry predictability as well as the correlation can be used on an ex ante basis. The second caveat is the mean-variance framework itself. While commonplace in investment management, the choice of whether to neutralize assumes that investors only care about mean and variance. Well known, however, is that investors prefer positively skewed returns and that most factor returns are not normally distributed. For example, suppose the sector component has positive skew. An investor might think twice about expunging sector exposures— even if our mean-variance framework suggests neutralization. Finally, we exclusively focus on sector exposures while other sources such as region or country exposures may also impact factor performance.”


Are you looking for more strategies to read about? Sign up for our newsletter or visit our Blog or Screener.

Do you want to learn more about Quantpedia Premium service? Check how Quantpedia works, our mission and Premium pricing offer.

Do you want to learn more about Quantpedia Pro service? Check its description, watch videos, review reporting capabilities and visit our pricing offer.


Or follow us on:

Facebook Group, Facebook Page, Twitter, Linkedin, Medium or Youtube

Share onRefer to a friend

Subscribe for Newsletter

Be first to know, when we publish new content


    logo
    The Encyclopedia of Quantitative Trading Strategies

    Log in

    MORE INFO
    We boasts a total prize pool of $15,000
    Gain a Share of a Total Prize Pool of $15.000
    MORE INFO
    $15.000
    Gain a Share of a Total Prize Pool
    SUBSCRIBE TO NEWSLETTER AND GET:
    - bi-weekly research insights -
    - tips on new trading strategies -
    - notifications about offers & promos -
    Subscribe