One of the best finance books I've read in some time (and I read a lot of them)
Reviewed in the United States on July 7, 2022
One of the best finance books I have read in some time. The author uses loads of data to provide a robust investing framework. Although low expected returns (hint: low treasury yields) are featured, this book is more broad than just that issue. I highly recommend it.
Part I: Setting the Stage
Figure 1.2 identifies 9 asset class premia that have, since 1926, delivered persistent, pervasive, and robust rewards that are statistically and economically significant: 4 asset class premia (equity, term, credit and commodities) and 5 style premia (value, momentum, carry, defensive, and trend).
Expected returns in all major asset classes have fallen to near historic lows because everything is discounted by the low yielding treasury bonds. It’s not just bond prices that move inversely to yield.
Part II: Building Blocks of Long-Run Returns
A diversified portfolio of commodity futures has historically earned 3-4% over T-bills.
US equities outperformed foreign equities, but mostly due to much faster real growth in dividends (1.9% in the US vs. near zero elsewhere). Sharpe ratios for 10-year treasuries and global government bonds were comparable to equities.
Antti previously questioned whether or not corporate bonds added value over a blend of equities and treasuries, but he says that newer research clearly answers in the affirmative. He concludes that despite a positive (0.25) correlation with equities, the credit premium has been a useful contributor to investor portfolios. High yield corporate bonds have higher returns, higher volatilities, and higher equity correlations than investment grade bonds. Based on historical experience the average breakeven spread to offset expected default losses is modest (15-25 bps) for the broad investment grade market, but much higher (200-250 bps) for the high yield market.
Commodities, with their growth and inflation exposures, are opposite to bonds. Commodities also have quite mild correlations with either stock (+) or bond (-) markets, so they are excellent diversifiers. Long-run commodity indices use front contracts and roll these to the next contract before expiry. Long-run returns to these indices can naturally be split into spot returns and roll returns. Commodity futures have a long history, although until the late 1940’s are mostly on grain products. Most of the long-run return came from the spot return, with the roll return being consistently negative. Since 1877 commodities had a long run geometric mean of 2-4% and a sharpe ratios near of 0.3.
Part III: Putting it all Together
Long/short strategies enable much more aggressive use of diversification, through shorting and leverage than long-only tilts. Antti contends that multi-metric and industry-neutral value strategies have outperformed.
Value strategies are inherently short a structural change and major value drawdowns have coincided with technological revolutions. Value and momentum work well at different horizons as many assets exhibit trending tendencies up to one-year, but mean-reverting tendencies at multiyear horizons. The negative correlation (often near -0.5) between value and momentum strategies make them great complements and near 50/50 is the implied optimal blend.
Trend-following was profitable every decade for the composite and almost without exception for each asset class. The risks for momentum and trend following come from sharp market turns and whipsawing trendless markets, respectively. Cost-effective trading execution is especially important for these high turnover strategies. Trend has performed very well in the worst equity market drawdowns, especially if they are protracted. Stock selection momentum also tends to perform well during these crashes, but there have been momentum crashes after the market has turned.
A broad definition of carry is an assets return in unchanged market conditions (yield or spread over funding rate). The best known carry trade favors high yielding short term interest rate currencies over low yielding ones. In practice this involved buying emerging market currencies and shorting developed market currencies. This strategy is significantly exposed to equity market risk and thus Antti has characterized this as “picking up pennies in front of a steam roller”.
Quality can be proxied by the bet against beta strategy, where the long side of low beta stocks are levered up and the short side of high beta stocks are levered down to produce a beta neutral strategy. If this strategy is not levered (dollar-neutral) it results in negative beta, but even so offers outperformance through diversification (risk-reduction) even if the raw returns are zero. Betting against beta has offered a higher sharpe ratio and better out of sample performance than Value and Momentum.
Risk parity investing involves taking equal risk in three or four nearly uncorrelated asset classes with similar sharpe ratios and thereby boosting the portolio sharpe ratio to 1.5 – 2x the typical single asset class sharpe ratio. Long short premia can do even better, especially if four of these styles can be applied in four or five asset classes in lowly correlated ways. Thus, portfolio sharpe ratios could plausibly be doubled by style diversification and doubled again by multi-asset applications. The math is basically square root of n, where n is number of uncorrelated investments with similar sharpe ratios.
This math only works by reducing volatility, if investors want to convert the reduced volatility to higher returns, they need to use leverage. Most practical asset allocation has shorting or leverage constraints. Relaxing these constraints is the key. Portfolio volatility declines when more assets are added but the decline is much steeper when lowly correlated investments are combined. For a single style in one asset class, the average SR was 0.4, after combining three to four styles per asset class, the average multi-style SR was 0.8. Then after diversifying across the asset classes, the all-in composite SR was 1.5, an almost four-fold increase. Importantly, this does not include trading costs or fees. Long/short style pairs have near-zero correlations to each other and multi-style composites across asset classes are also very low. These low correlations are not available when using a long-only framework.
Antti describes unlimited liabilities such as selling short a stock as particularly dangerous. Investors need to put survival first. Antti seems to support trend following as downside protection during long slow bear market, but that it is vulnerable to sudden market falls.
AQR published a study on trading/market impact costs using data from 1998-2016 and found that trades cost 9-19bps per dollar traded on average for large/small caps.
Antti makes a compelling case for applying long/short strategies across multiple asset classes, seeming to imply a quadrupling of the Sharpe ratio. My main question is what kind of leverage this would realistically take and what borrowing costs would be for all the short positions.
2 people found this helpful