ABSTRACT
We demonstrate a strong relationship between short-term small-firm premium and future low-beta anomaly performance. Rises (declines) in small-firm prices temporarily improve (deteriorate) funding conditions, benefiting (impairing) the short-run returns on the low-beta strategy. To investigate this phenomenon, we examine returns on betting-against-beta (BAB) and small-minus-big (SMB) factor portfolios in 24 developed markets for the years 1989–2018. A zero-investment strategy of going long (short) in BAB factors in the quintile of countries with the highest (lowest) three-month SMB return produces a mean return of 1.46% per month. The effect is robust when controlling for major risk factors in equity markets, alternative portfolio construction methods, and subperiod analysis. The predictability of BAB performance by SMB returns is also present in the time series of individual country returns, forming the grounds for effective timing in the low-beta strategies.
Acknowledgements
The author thanks Lasse Heje Pedersen, Turan Bali, Pim van Vliet, Soosug Hwang, David Sraer, and Bader S. Alhashel for helpful comments and discussions that benefited the paper. All errors are my own. Also, the author thanks the participants of the Citation2018 Paris Financial Management Conference, 2019 Citi Quant Conference in Valencia, and the research seminar at University of Dubai. This paper is a part of the Project No. 2019/33/B/HS4/01021 of the National Science Centre of Poland.
Author’s note
Correspondence concerning this article should be addressed to Adam Zaremba, 1) Montpellier Business School, 2300 Avenue des Moulins, 34185 Montpellier cedex 4, France, [email protected], or 2) Poznan University of Economics and Business, al. Niepodleglosci 10, 61-875 Poznan, Poland, e-mail: [email protected].
Notes
1 A recent study of Scott (Citation2020) extends this phenomenon for other “non-market” betas.
2 Banz (Citation1981) argues that low-capitalization firms produce higher returns than the high-capitalization firms. The effect is usually described as the “small firm anomaly.” Some recent studies have cast doubt on the validity of the size premium (van Dijk, Citation2011); nonetheless, it is still considered an essential factor for many well-established asset-pricing models. Furthermore, Asness et al. (Citation2018) demonstrate that once the role of firm quality is controlled, the size premium remains powerful and robust across many countries and industries.
3 The formulas in this section follow the notation of Frazzini and Pedersen (Citation2014).
4 As explained on the AQR webstite, “with each data update, the entire historical series is updated and may reflect changes as a result of the various data vendors subsequently fixing data errors” (https://www.aqr.com/Insights/Datasets/About-the-AQR-Data-Library). Therefore, any dataset obtained at a later date from the same source may not necessarily yield perfectly identical results. These minor discrepancies should not qualitatively change our findings.
5 The implementation by Asness and Frazzini (Citation2013) leads to slightly better profitability of the value strategy than the original framework. Nonetheless, when we use factor returns computed in the manner of Fama and French (Citation1992, Citation1993) there is virtually no impact on our results.
6 The results for the extended months from l = 13 to l = 60 are available upon request.
7 For robustness, Tables A4 and A5 in the Internet Appendix report the analogous results for portfolios from one-way sorts on the SMB returns in the past one and two months.
8 To assure the robustness of our outcomes, we also examine the BAB predictability according to SMB returns for selected markets, sourced from French (Citation2018), as well as the long-short quintile portfolio returns from sorts on market capitalization based on the same source. We find no qualitative difference in the results. Importantly, our conclusions hold also when we employ long-only small firm portfolios, which is still consistent with our explanation based on time-varying funding conditions.
9 Tables A9 and A10 in the Internet Appendix uncover similar results for the two- and three-month sorting periods applied to SMB returns.
10 For brevity, we demonstrate only the results for regional and global portfolios.
11 For robustness, we also consider the average returns through the past two and three months. The results reveal no major qualitative differences and are presented in Tables A15 and A16 of the Internet Appendix.
12 The detailed statistics are available upon request.