The Main Points Of The Book "Expected Returns: An Investor’s Guide to Harvesting Market Rewards" By Antti Ilmanen Essay


Antti Ilmanen is a world leading researcher and advisor recognized by the CFA Institute with the 2017 Leadership in Global Investment Award.

Ilmanen began his career as a central bank portfolio manager in Finland after earning a Master of Science in economics and law from the University of Helsinki and a Ph. D in finance from the University of Chicago. Since then, Ilmanen has built a well-respected career through experiences as a senior portfolio manager at Brevan Howard and a number of roles with Salomon Brothers and Citigroup. Antti now manages AQR’s Portfolio Solutions Group where he advises institutional investors and sovereign wealth funds.

Since starting his doctorate program at UofC, Ilmanen has been obsessed with expected returns; a complicated and dynamic concept in “quantitative finance” that he tries to make easily understandable in one place, his book – Expected Returns: An Investor’s Guide to Harvesting Market Rewards.

The book examines expected returns along three different dimensions including: (1) the expected return an investor gets paid for owning any of the major asset classes, (2) the expected return an investor gets paid from pursuing well-known strategies (value, momentum, and carry investing), and (3) the expected return an investor gets from being willing to take the risk of being exposed to economic or other risk.

The term “expected returns” refers to how much an investor makes over time on an investment or strategy which is directly influenced by the risk associated with that investment. Ilmanen introduces the idea of expected returns through a reference to a parable about several blind men and an elephant, as told by American poet John Godfrey Saxe. As the six blind men touch the elephant, they describe the various textures they feel, but each man misses the big picture by only touching one part of the elephant – as will investors who do not consider all four vantage points of expected returns: historical average returns, financial and behavioral theories, forward-looking market indicators, and discretionary views.

Which Asset Classes Deliver Expected Returns

Expected return differentials across assets reflect rational risk premia as well as market inefficiencies and supply-demand effects. Ilmanen attempts to simplify the estimated return drivers by focusing on four major asset classes: equities, government bonds, corporate bonds, and alternative assets.

Equity Risk Premium

The equity premium or equity risk premium refers to the expected return of a broad equity index in excess over some non-equity alternative. Standard economic models and rational expectation theories suggest that such a premium should be negligible or not exist at all, however, academic papers have offered several explanations for the puzzle of stocks much stronger historical performance. Existing theories suggest market frictions, biased samples, and other factors could be the drivers, but there is little consensus. Ilmanen explains the three main factors he believes contribute to such premium include:

  • Rare disaster risk: Investors may have historically assigned a higher probability to rare catastrophic events than materialized during the sample period
  • Structural uncertainty: Investors do not fully know the structure of the economic system, so they must gradually learn about unknown structural parameters
  • Long-run risk: Uncertainty regarding long-run growth risk

Nonetheless, such an equity risk premium appears to exist and the historical annual excess returns of US stocks over government bonds averages 3% to 5% over long windows, a further 1% over short-dated bills, and another 2% higher if arithmetic means are used. Global excess returns of stocks over corporate bonds, on the other hand, are somewhat lower. The variation in average nominal equity returns across very long samples may partly reflect different inflation levels over time, but may also allude to a greater theme that it is crucial to distinguish between realized (ex post) average excess returns and expected (ex ante) risk premia. Because required returns vary over time, past returns may be poor predictors of future returns.

Among forward-looking measures of equity market carry or value, dividend yield was the early leader but fell out of favor after giving a bearish signal through the 1990s equity rally. However, broader payout yields that include share buybacks have replaced dividend yield as the preferred carry measure, while earnings yield and the Gordon Model (i.e. Dividend Discount Model) equity premium have become the preferred valuation measures. Empirical evidence shows that the net buyback-adjusted dividend yield had the highest correlation with next quarter equity market returns (0.27).

From a value measure standpoint, a stock market’s price-earnings (P/E) ratio is the most popular equity market valuation indicator. The spread or, alternatively, the ratio of government bond yield over earnings yield (E/P) is the most popular measure of relative valuation between the two major asset classes – and thus shorthand for the equity-bond premium. Shifts in the relative risk of asset classes are a structural change that undermines the usefulness of valuation signals like yield ratio. This ratio serves well as a mean-reverting signal within one regime, but it typically gives a wrong signal when a structural or regime change occurs. Guiding signals for such secular changes include changes in long-run inflation expectations or in long-run economic growth rates, or in their volatilities.

It is true that high inflation tends to hurt equity markets but so does deflation. Steady, low, but positive inflation appears to be the optimal environment for real growth and asset valuations. Survey based subjective explanations provide direct estimates of changing return expectations, but they may reflect wishful thinking rather than required returns. Retail investors, for example, expect higher equity returns and lower volatility with improving macroeconomic conditions. During recessions, they have an unduly pessimistic view of the macroeconomy and equity markets; their selling drives down equity prices and causes equity expected returns to rise.

Conversely, portfolio managers and CFOs exhibit more cool-headed characteristics and have more countercyclical views when making long-run forecasts. Finally, some investors attempt to implement tactical forecasting methods in order to time the market. Buying stocks when market valuations are cheap, for example, tends to produce much better returns than buying when valuations are rich. Of course, such contrarian market timing takes nerve and sometimes fails, hence why the relation is there.

Additionally, some investors utilize monetary policy indicators that have exhibited quite mild predictive correlations thereby adhering to the old adage “Follow the Fed”. Other common slogans have also shown to have validity in the equity markets such as “Sell in May and Go Away”. The best-known seasonal regularities are the so-called January and Halloween effects largely fueled by behavioral science factors such as a greater risk appetite at the beginning of the year.

Bond Risk Premium

The bond risk premium (BRP) or term premium is the expected return advantage of long-duration government bonds over short-term bonds. In the case of bonds, rearview-mirror evidence is obviously less important than with regards to equities because market yields provide relatively transparent information on expected nominal returns. A key focus regarding the bond risk premium is how to extract information about it from the yield curve. The yield curve reflects both the BRP and the market’s interest rate expectations. However, the YC may not able to predict multi-year excess bond returns and it may be a poor BRP proxy because mean-reverting rate expectations dominate curve high or low.

Since rate expectations taint information about the BRP in the yield curves, a natural solution is to estimate rate expectations and subtract them from bond yields. steepness when short term rates are exceptionallyEmpirical evidence suggests that the Treasury market’s risk-reward relation is positive, but the more interesting pattern is that the relation is quite nonlinear. The long-run risk-reward relation in the Treasury market looks like a hockey stick: very steep up to two years and flat thereafter.

Nonetheless, yield curve steepness is the simplest and most popular proxy for the ex ante BRP. The steepness of the yield curve (YC) performs fairly well as a forecaster of bonds near term expected return but is quite a poor proxy for the long-term BRP. The best explanation for this puzzle appears to be mean-reverting rate expectations. Because curve steepness reflects both rate expectations and required risk premia, the short rates triggers investors to expect increases from record-low rate levels and declines from record-high levels. The rate expectation and risk premium components in the YC are related in opposite directions to the rate level thereby weakening the overall link between the YC and BRP.

Therefore, it is pivotal to break down the YC into unobservable rate expectation and risk premium components, or which the author favors using survey data to prevent the values from reflecting model specification or estimation error. Ilmanen further recommends breaking down the four major drivers of ex ante bond premia and real yields into: level-dependent inflation uncertainty, equity and/or recessions-hedging ability, supply-demand factors, and cyclical effects.

The second and fourth factors, in particular, can be motivated by covariance with bad times, while the first and third factors are related to bonds’ standalone risks. The inflation risk premium (IRP) is the most important secular driver of required expected real bond yields and BRPs; it contributed 3%-4% to nominal bond yields when they peaked in the early 1980s and subsequently fell close to zero. The long-term inflation and bond market volatility have high predictive correlations over the 5 year horizon (0.31 and 0.64), but show low correlation over short horizons.

The story is simple and rings true with investment practitioners: higher inflation levels are associated with greater inflation uncertainty, which in turn warrants higher required premia for holding nominal bonds. The various supply-demand factors that influence the pricing and required returns on bonds can also be viewed as contributing to the time-varying nature of risk premia and can be classified as one of three: fiscal effects, regulatory effects and pension fund demand, and foreign flows. Empirical studies with US and global data suggest that a 1% rise in the public-debt-to-GDP ratio raises bond yields by 2 to 6 basis points. Yield curve shape is also closely related to business cycles, credit cycles, and monetary cycles.

A steep YC coincides with a high unemployment rate and predicts fast economic growth. YC countercyclicality may explain its ability to predict near-term bond and stock returns: high required premia near business cycle troughs result in a steep YC, while low required premia near business cycle peaks result in an inverted YC.Finally, bond risk premium is also influenced by safe haven influences and other cyclical trends. For example, negative-equity returns, and high equity market volatility are bullish for bonds both contemporaneously and predictively. From a cyclical standpoint, a strong economy empirically predicts low bond returns with surprisingly large correlations for the longest horizon.

Credit Risk Premium Bonds exposed to credit risk have outperformed Treasuries only marginally over long histories, arguably giving poor compensation for their low liquidity and poor timing of losses. Long-dated corporate bonds have performed especially poorly, while barely speculative-grade bonds (BB-rate) have performed best. In this generally bleak picture, short-dated top-rated credit bonds have given an attractive reward-volatility ratio that arbitrageurs are unable to remove because of the financing rates they face. These corporate bonds are also generally very risky with the largest losses occurring during financial crises.

Besides credit considerations such as the break-even spread due to expected default or downgrading losses, yield spreads over Treasuries likely reflect a premium for the lower liquidity of corporate bonds. Disentangling these unobservable components is an even greater challenge than disentangling the market’s rate expectation and required bond risk premium components in Treasury yields.

Nonetheless, the realized average outperformance of credits should be an accurate measure, not suffering from the overstatement of prospective return advantage that is an inherent characteristic of yield spreads. In any case, historical average returns are not very informative about the future if expected returns vary over time, but we study them in the hope of gleaming something useful about long-run reward.Historical returns tell us that lower rated investment grade (IG) classes have earned higher mean returns, but excess returns are small and inconsistent. The BB-rate sector, just below the IG threshold, provides the best long-run performance of any bond-rating category.

This success likely reflects partial market segmentation caused by the constraints under which many portfolio managers operate. Fallen angels (bonds downgraded from IG to HY) appear to outperform bonds originally issued as high yield (HY). The riskier corporate bonds, CCC, have clearly underperformed Treasuries since 1985. Even with limited outperformance, a simple argument could be made that corporates are superior investments to treasuries: they offer slightly higher long-term returns and slightly lower volatilities. Critics, on the other hand, argue that the additional risks associated.

How to cite this essay: