Predictability Of Excess Returns Research Articles

Solving the Forecast Combination Puzzle Using Double Shrinkages*

AbstractThis study develops a new approach that shrinks the forecast combination weights towards equal weights by using weighted least squares and towards zero weight by using regularization constraints. We reveal the significant predictability of excess returns to the S&P500 index that can be achieved by using this double shrinkage combination (DSC). Furthermore, our DSC approach significantly outperforms the naïve equal‐weighted combination, solving the combination puzzle. The equal‐weight shrinkage has greater effect in economic recessions, whereas the zero‐weight shrinkage dominates in economic expansions. The DSC's superior performance over that of the naïve combination is observed in the application of forecasting macroeconomic indicators.

Commodity futures return predictability and intertemporal asset pricing

We find out-of-sample predictability of commodity futures excess returns using combination forecasts of 28 potential predictors. Such gains in forecast accuracy translate into economically significant improvements in certainty equivalent returns and Sharpe ratios for a mean–variance investor. Commodity return forecasts are closely linked to the real economy. Return predictability is countercyclical, and the combination forecasts of commodity returns have significant predictive power for future economic activity. Two-factor models featuring the market factor and the innovations in each of the combination forecasts explain a substantial proportion of the cross-sectional variation of both commodity and equity returns. The associated positive risk premiums are consistent with Merton’s (1973) intertemporal capital asset pricing model (ICAPM), given how the combination forecasts predict an increase in future economic activity and a decline in stock market volatility in the time-series. Overall, combination forecasts act as state variables within the ICAPM, thus resurrecting a central role for macroeconomic risk in determining expected returns on commodities.

Monetary policy expectation errors

How are financial markets pricing the monetary policy outlook? We use surveys to decompose excess returns on money market instruments into expectation errors and term premia. Excess returns are primarily driven by expectation errors, whereas term premia are negligible. Investors face challenges when learning about the Federal Reserve’s response to large, but infrequent, negative shocks in real-time. Rather than reflecting risk compensation, excess returns stem from investors underestimating how much the central bank eases policy in response to such rare shocks. We show, for the US and internationally, that expectation errors imply excess return predictability from past stock returns.

Infrequent Random Portfolio Decisions in an Open Economy Model

Abstract We introduce a portfolio friction in a two-country DSGE model where investors face a constant probability to make new portfolio decisions. The friction leads to a more gradual portfolio adjustment to shocks and a weaker portfolio response to changes in expected excess returns. We apply the model to monthly data for the US and the rest of the world for equity portfolios. We show that the model is consistent with a broad set of evidence related to portfolios, equity prices, and excess returns for an intermediate level of friction. The evidence includes portfolio inertia, limited sensitivity to expected excess returns, a significant impact of financial shocks, excess return predictability, and asset price momentum and reversal.

Stock return prediction: Stacking a variety of models

We employ an ensemble learning approach, “stacking”, to refine and combine a variety of linear and nonlinear individual stock return prediction models. In an application of forecasting U.S. market excess return, stacking with a simple structure can outperform the traditional historical mean benchmark, Mallows model averaging, simple combination forecast, complete subset regression, combination elastic net forecast, and several other models in terms of both in- and out-of-sample performance measures on a consistent basis. More importantly, we find that the out-of-sample gains of stacking are especially evident during extreme downside market movements. Overall, stacking can generate substantive improvements in market excess return predictability.

Examining the sources of excess return predictability: Stochastic volatility or market inefficiency?

We use a consumption based asset pricing model to show that the predictability of excess returns on risky assets can arise from only two sources: (1) stochastic volatility of fundamental variables, or (2) departures from rational expectations that give rise to predictable investor forecast errors and market inefficiency. While controlling for stochastic volatility, we find that a variable which measures non-fundamental noise in the Treasury yield curve helps to predict 1-month-ahead excess stock returns, but only during sample periods that include the Great Recession. For these sample periods, higher noise predicts lower excess stock returns, implying that a shortage of arbitrage capital in financial markets allowed excess returns to drop below the levels justified by fundamentals. The statistical significance of the predictor variables that control for stochastic volatility are also typically sensitive to the sample period. Measures of implied and realized stock return variance cease to be signficant when the COVID-influenced data from early 2020 onward is included.

Examining the Sources of Excess Return Predictability: Stochastic Volatility or Market Inefficiency?

We use a consumption based asset pricing model to show that the predictability of excess returns on risky assets can arise from only two sources: (1) stochastic volatility of fundamental variables, or (2) departures from rational expectations that give rise to predictable investor forecast errors and market inefficiency. While controlling for stochastic volatility, we find that a variable which measures non-fundamental noise in the Treasury yield curve helps to predict 1-month-ahead excess stock returns, but only during sample periods that include the Great Recession. For these sample periods, higher noise predicts lower excess stock returns, implying that a shortage of arbitrage capital in financial markets caused excess returns to drop below the levels justified by fundamentals. The statistical significance of the predictor variables that control for stochastic volatility are also typically sensitive to the sample period. Measures of implied and realized stock return variance cease to be significant when the COVID-influenced data from early 2020 onward is included.

Do average higher moments predict aggregate returns in emerging stock markets?

PurposeIt has been demonstrated in the US market that expected market excess returns can be predicted using the average higher-order moments of all firms. This study aims to empirically test this theory in emerging markets.Design/methodology/approachTwo measures of average higher moments have been used (equal-weighted and value-weighted) along with the market moments to predict subsequent aggregate excess returns using the linear as well as the quantile regression model.FindingsThe authors report that both equal-weighted skewness and kurtosis significantly predict subsequent market returns in two countries, while value-weighted average skewness and kurtosis are significant in predicting returns in four out of nine sample markets. The results for quantile regression show that the relationship between the risk variable and aggregate returns varies along the spectrum of conditional quantiles.Originality/valueThis is the first study that investigates the impact of third and fourth higher-order average realized moments on the predictability of subsequent aggregate excess returns in the MSCI Asian emerging stock markets. This study is also the first to analyze the sensitivity of future market returns over various quantiles.

Commodity Futures Return Predictability and Intertemporal Asset Pricing

We find out-of-sample predictability of commodity futures excess returns using forecast combinations of 28 potential predictors. Such gains in forecast accuracy translate into economically significant improvements in certainty equivalent returns and Sharpe ratios for a mean-variance investor. Commodity return forecasts are closely linked to the real economy. Return predictability is countercyclical, and the combination forecasts of commodity returns have significantly positive predictive power for future economic activity. Two-factor models featuring innovations in each of the combination forecasts and the market factor explain a substantial proportion of the cross-sectional variation of commodity and equity returns. The associated positive risk prices are consistent with the Intertemporal Capital Asset Pricing Model (ICAPM) of Merton (1973), given how the predictors forecast an increase in future economic activity in the time-series. Overall, combination fore- casts act as state variables within the ICAPM, thus resurrecting a central role for macroeconomic risk in determining expected returns.

Predictability of Stock Market Excess Returns With Household’s Obligation Ratio

In this study, I test the predictability of stock market excess returns with households’ obligations ratio. Using U.S stock market data, I show that household’s debt service ratio can predict stock market returns at short horizon and over business cycle frequencies. I show that between 1980 and 2016, mean deviations from debt service ratio is a better forecaster of future returns both in-sample and out-of-sample than dividend-price ratio, dividend yield, earnings-price ratio, investment-capital ratio, and several other popular forecasting variables. The results remain significant using quarterly data and annual data.

주택가격의 초과수익률 예측가능성과 초과변동성 검정

본 논문은 동태적 고든 모형을 서울과 6개 광역시 아파트 매매시장에 적용하여 다음 주요 결과를 도출했다. 첫 번째로 모든 지역에서 아파트 초과수익률의 예측가능성이 높게 나타나 아파트 매매시장에는 투기적 수요가 존재하는 것으로 나타났다. 두 번째로 초과변동성 검정은 적정 시차에 따라 대도시 지역에 적용가능 여부는 달라지는 것으로 나타났다. 세 번째로 로그 임대료-가격 비율이 단위근을 가지는 경우에는 동태적 고든 모형을 결과 해석에 유의할 필요가 있다. 네 번째로 동태적 고든 모형을 이용한 지역별 아파트 근본가치평가를 실제 주택가격과 비교하여 과다 또는 과소 평가 여부를 파악할 수 있다.

Decomposed Higher-Moment Risk Premiums and Market Return Predictability

In an empirical study of Standard & Poor's 500 index options, this paper analyses the predictability of future market excess returns by means of decomposed higher-moment risk premiums. The study proposes a new measure of kurtosis risk premium and suggests a decomposition of higher-moment risk premiums up to the fourth moment into downside and upside premiums. Thereby, the paper enhances the understanding of higher-moment risk premiums. The decomposition uncovers valuable information for return forecasts, as decomposed higher-moment risk premiums deliver improved in-sample predictions. In an out-of-sample study, the predictive power of decomposed higher-moment risk premiums is shown to be particularly driven by downside higher-moment risk premiums.

Bond return predictability: Evidence from 25 OECD countries

In this paper, we test for bond excess return predictability in 25 Organisation for Economic Co-operation and Development countries, using a set of 12 predictor variables, including macroeconomic, financial, and commodity-based indicators. The novelty of our approach is that we employ both time-series and panel data models that accommodate key data features, namely, persistency, endogeneity, heteroscedasticity, and cross-sectional dependence. We note two key findings. The first is that commodity variables, both oil spot and futures prices, and the world commodity price index predict bond excess returns. Our second finding confirms that macroeconomic variables (the term spread and T-bill yield) are successful predictors of bond excess returns.

Corrigendum: Bond Risk Premiums with Machine Learning

Abstract In this note we revisit the empirical results in Bianchi, Büchner, and Tamoni (2020) after correcting for using information not available at the time the forecast was made. Although we note a decrease in out-of-sample $R^2$, the revised analysis confirms that bond excess return predictability from neural networks remains statistically and economically significant.

Local demand shocks, excess comovement and return predictability

Abstract I investigate the importance of local demand shocks on excess comovements and return predictability for 4560 twin-pairs of Exchange-Traded Funds (ETFs) from 15 country-pairs. The returns on ETFs traded in the same country comove excessively with one another. These comovements are stronger for funds with greater liquidity and more competitors in the local market. In contrast, comovements are not materially different among ETFs that are attractive to fundamental (factor) investors. A local measure of mispricing, based on price-deviations between ETFs and their foreign peers, strongly predicts ETF return reversals. Betting against local mispricing yields significant abnormal returns of up to 20 percent per year after trading costs.

Predictability and pricing efficiency in forward and spot, developed and emerging currency markets

We study the predictability of forward and spot exchange rates of currencies of emerging and developed economies from 1994 to 2016. Our purpose is to shed light on the efficiency of currency markets and how and why it has evolved over this time. For the currencies of emerging economies, our analysis of rates of return on forward contracts finds some evidence of excess-predictability, especially in the earlier parts of the sample period, consistent with the view that this portion of the foreign exchange market has only become efficient in recent times. When we turn our attention to excess-returns computed from spot exchange rates and spot interest rates, however, we find much less predictability. In particular, over our full sample period, we find no evidence of excess-predictability, in contrast with the results reported by Hsu et al. (2016) but in agreement with Kuang et al. (2014). The different predictability of spot excess-returns and rates of return on forward contracts is a manifestation of the widespread violation of covered interest parity which emerged with the onset of the 2008 financial crisis.

Commodity Futures Return Predictability and Intertemporal Asset Pricing

We find out-of-sample predictability of commodity futures excess returns using forecast combinations of 28 potential predictors. Such gains in forecast accuracy translate into economically significant improvements in certainty equivalent returns and Sharpe ratios for a mean-variance investor. Commodity return forecasts are closely linked to the real economy. Return predictability is countercyclical, and the combination forecasts of commodity returns have significantly positive predictive power for future economic activity. Two-factor models featuring innovations in each of the combination forecasts and the market factor explain a substantial proportion of the cross-sectional variation of commodity and equity returns. The associated positive risk prices are consistent with the Intertemporal Capital Asset Pricing Model (ICAPM) of Merton (1973), given how the predictors forecast an increase in future economic activity in the time-series. Overall, combination forecasts act as state variables within the ICAPM, thus resurrecting a central role for macroeconomic risk in determining expected returns.

Predictability of Treasury Bond Returns: Risk Premia or Overreaction?

We examine whether the predictability and business-cycle dependence of excess returns in US Treasuries can be more naturally explained in terms of state-dependent risk premia or a specific cognitive bias (representativeness). We show that the extremely parsimonious cognitive-bias model in Shleifer and Gennaioli (2018) accounts very well for a large number of stylized facts about the predictability of excess returns, and their business-cycle dependence. We also test the risk-premium explanation by looking at the correlation between the payoff of the carry strategy and (several proxies for) consumption. When we do so we find that this correlation either has no explanatory power for returns, or has the wrong sign. We conclude that undue extrapolation in the future of recent information provides a parsimonious, simple and arguably more compelling account of excess returns than a risk-premium explanation.

Investigating the expectation hypothesis and the risk premium dynamics: new evidence for Brazil

We re-examine the validity of the expectation hypothesis (EH) of the term structure for the Brazilian fixed income market, using data from January 2000 to June 2017. Furthermore, we investigated the out-of-sample predictability of bond excess returns by means of common factors extracted from a cross-section of Brazilian macro-variables and zero-coupon interest rates. The EH is rejected throughout the term structure examined on the basis of the statistical tests across the entire maturity spectrum considered. Our results confirm previous findings, mostly obtained for developed markets, that a linear combination of forward rates and macroeconomic factors can explain a substantial portion of movements in bonds excess returns, contributing novel and up-to-date evidence from a large and dynamic emerging bond market, such as Brazil. Furthermore, we find that the factor extracted from a large panel of macroeconomic variables generates significant gains in forecasting bond excess returns relative to yield curve information.

Short-Rate Expectations and Unexpected Returns in Treasury Bonds

I document large and persistent errors in investors’ expectations about the short-term interest rate over the business cycle. The largest errors arise in economic downturns and during Fed easings when investors overestimate future short rates and, thus, underestimate future bond returns. At a one-year horizon, errors about the path of the real rate (as opposed to inflation) account for 80% of short-rate forecast error variance, with more than half of that number attributed to the Fed easing more aggressively than the public expected. Short-rate forecast errors induce ex post predictability of excess returns on Treasury bonds that is not due to time-varying risk premium. Received June 10, 2016; editorial decision February 1, 2018 by Editor Robin Greenwood. The author has furnished an Internet Appendix, which is available on the Oxford University Press Web site next to the link to the final published paper online.