Low-risk investing in equities has been in the spotlight in recent years probably due in particular to the disappointing performance of equity markets since the start of the new millennium and until the 2008 crisis. The main focus of low-risk investing is to reduce portfolio risk, defending the portfolio in equity market downturns, while capturing the positive alpha from low-risk stocks to improve risk-adjusted returns. Indeed, the positive alpha found in low-risk stocks explains why the Sharpe ratio of strategies invested in these stocks has been larger than that for the market capitalization index. Low-risk investing also naturally excludes the riskier stocks which have been delivering the poorest risk-adjusted returns and have had significant negative alpha.
Low-risk investing dates back to the seminal paper of Haugen and Heins (1972) with empirical evidence that between 1926 and 1969 portfolios systematically investing in U.S. low-volatility stocks would have delivered much larger returns than expected from their low level of beta, while portfolios invested in high-volatility stocks would have delivered returns much below what should have been expected from their high level of beta. Brennen (1971) and Black (1972) showed that the violation of one of the assumptions behind the Capital Asset Pricing Model (CAPM) – that investors have no constraints, e.g. on leverage or borrowing – is sufficient to reduce the slope of the relationship between returns and beta. Blitz (2014) has recently reviewed the academic literature and summarized the different effects that have been proposed by academics to explain the low-risk anomaly.
The low-risk anomaly appears almost universally. Haugen and Baker (2012) demonstrated empirically that it can be found in the cross-section of stock returns of almost all developed and emerging market countries in the world. The comprehensive empirical analysis of De Carvalho, Dugnolle, Lu and Moulin (2014) strongly suggests that the low-risk anomaly goes beyond equity markets and can also be found in the cross-section of bond returns of all major segments of fixed-income markets and regions. Their results show that portfolios invested in low-risk bonds with the lowest beta generated the largest positive alpha, while portfolios invested in the riskier bonds with the highest beta generated the most negative alpha. This result was found for government bonds, quasi & foreign government bonds, securitized & collateralized bonds, corporate investment-grade bonds, corporate high-yield bonds, emerging market bonds and aggregations of some of these universes, and for bonds in USD, EUR, GBP and JPY. Frazzini and Pedersen (2014) suggest that the low-risk anomaly is also observed in commodities, currencies and at top-down level in fixed income and equities, i.e. in the cross-section of the returns of currency forwards, index futures, equity and Treasury country indices, portfolios aggregated by ratings, and in the cross-section of all these put together. Baker, Brendan and Taliaferro (2014) have recently looked at the decomposition of the low-risk anomaly into top-down country and industry contributions and bottom-up contributions. They found a risk anomaly in the cross-section of country returns and, to a lesser extent, in the cross-sectional of industry returns. Asness, Frazzini and Pedersen (2014) gave stronger evidence of a low-risk anomaly in the cross-section of industry returns by using more granular industry definitions.
The low-risk anomaly is not only found in the cross-section of asset classes but also in the time series of asset class premiums and in the time series of factor premiums. Perchet, De Carvalho, Heckel and Moulin (2014) showed that the time series of asset class returns shows volatility clustering, i.e. the volatility forms two distinct volatility regimes, one with low volatility and high average returns and on with high volatility and low average returns, or even negative, for most asset classes. In turn, Perchet, De Carvalho and Moulin (2014) showed that the time series of value and momentum factor returns in equity, government bonds and currency markets also shows volatility clustering, with two distinct volatility regimes: higher returns for the low volatility regime and lower returns for the high volatility regime.Download to read more