Bond Return Predictability Research Articles

Instantaneous volatility of the yield curve, variance risk premium and bond return predictability

This paper proposes a new way of estimating the instantaneous volatility of fixed income securities using derivatives data, which can further be used to construct the corresponding yield curve variance risk premium (VRP). We show that this VRP measure exhibits strong long-horizon predictive power for bond excess returns. After controlling for the shape of the yield curve, the VRP strongly predicts 1-year holding period excess returns for 2-year to 10-year zero coupon bonds. The marginal R2 of VRP is as high as 12.6%. One standard deviation increase in the VRP is associated with 2.224% increase in the bond excess return. This result is robust when we include various other bond return predictors, such as the Cochrane–Piazzesi “tent-shaped” factor. The out-of-sample analysis suggests that this predictability is not only statistically significant, but also can be translated into economic gains. Additional tests suggest that this predictability varies with economic conditions.

Bond return predictability: Macro factors and machine learning methods

AbstractWe investigate the impact of macroeconomic variables on bond risk premia prediction via machine learning techniques. On the basis of Chinese treasury bonds from March 2006 to December 2022, we show that adding macroeconomic factors improves bond return forecasts and generates higher economic benefits to investors. This is achieved when the nonlinear relationship between macroeconomic variables and bond returns is modelled via machine learning methods. Furthermore, the importance of macroeconomic determinants changes along the yield curve. Our study sheds new light on the information contained in macroeconomic variables for treasury bond valuation and highlights the importance of utilizing appropriate machine learning methods.

Moments, shocks and spillovers in Markov-switching VAR models

To investigate how economies, financial markets or institutions can deal with stress, we often analyze the effects of shocks conditional on being in a recession or a bear market. MSVAR models are perfectly suited for such analyses because they combine gradual movements with sudden regime switches. In this paper, we develop a comprehensive methodology to conduct these analyses. We derive first and second moments conditional only on the regime distribution and propose impulse response functions for both moments. By formulating the MSVAR as an extended linear non-Gaussian VAR, all results are available in closed-form. We illustrate our methods with an application to stock and bond return predictability. We show how forecasts of means, volatilities and (auto-)correlations depend on the regimes. The effect of shocks becomes highly nonlinear, and they propagate via different channels. During bear markets, shocks have stronger effects on means and volatilities and die out more slowly.

Are bond returns predictable with real-time macro data?

We investigate the predictability of bond returns using real-time macro variables and consider the possibility of a nonlinear predictive relationship and the presence of weak factors. To address these issues, we propose a scaled sufficient forecasting (sSUFF) method and analyze its asymptotic properties. Using both the existing and the new method, we find empirically that real-time macro variables have significant forecasting power both in-sample and out-of-sample. Moreover, they generate sizable economic values, and their predictability is not spanned by the yield curve. We also observe that the forecasted bond returns are countercyclical, and the magnitude of predictability is stronger during economic recessions, which lends empirical support to well-known macro finance theories.

Predicting Bond Return Predictability

This paper provides empirical evidence on predictable time variations in out-of-sample bond return predictability. Bond return predictability is associated with periods of high (low) economic activity (uncertainty), which implies that violations of the expectations hypothesis are state dependent and linked to features of the business cycle. These state dependencies in predictability, established by introducing a new multivariate test for equal conditional predictive ability, can be used in real time to improve out-of-sample bond risk premia estimates and investors’ economic utility through a novel dynamic forecast combination scheme that uses predicted forecasting performance to identify the best set of methods to include in the combined forecast. Dynamically combined forecasts exhibit strong countercyclical behavior and peak during recessions.This paper was accepted by Lukas Schmid, finance.Funding: This work was supported by the Danish Council of Independent Research [Grants DFF 7024–00020B and DFF 9033–00003B] and the Danish Finance Institute.Supplemental Material: The data files and online appendices are available at https://doi.org/10.1287/mnsc.2023.4713 .

Monetary Policy and Bond Prices with Drifting Equilibrium Rates

AbstractWe study the drift and cyclical components in U.S. Treasury bonds. We find that bond yields are drifting because they reflect the drift in monetary policy rates. Empirically, modeling the monetary policy drift using demographics and productivity trends, plus long-term inflation expectations, leads to cyclical deviations of bond prices from their drift that predict bond returns in- and out-of-sample. These bond cycles can be interpreted as term premia or/and temporary deviations from rational expectations in a behavioral framework. Through the lens of our model, we detect a significant role of the latter in determining the cyclical properties of yields with short maturities.

Gone with the Vol: A Decline in Asset Return Predictability During the Great Moderation

We document and examine a significant shift in the comovement of asset returns and macroeconomic volatility during the Great Moderation. Strong U.S. stock and bond return predictability from several macroeconomic volatility series before 1982 was followed by a significant predictability decline during the Great Moderation (1982–2008). These findings are robust to alternative empirical specifications and out-of-sample tests. In a calibrated equilibrium model with time-varying volatility, the predictability decline requires changes in several model elements. Lower return predictability is consistent with stronger policy responses to inflation and output, a larger slope in the New Keynesian Phillips curve, and reduced sensitivity of both macroeconomic and financial variables to a volatility factor. Our results contribute to the examination of macroeconomic volatility as a driver of expected asset returns and the instability in predictive regressions. We further identify sources of the Great Moderation using asset price dynamics. This paper was accepted by Tomasz Piskorski, finance. Supplemental Material: Data files and the online appendix are available at https://doi.org/10.1287/mnsc.2022.4429 .

Predicting Bond Returns: 70 Years of International Evidence

We use 70 years of international data from the major bond markets to examine bond return predictability through in-sample and out-of-sample tests. Our results reveal economically strong and statistically significant bond return predictability. This finding is robust over markets and time periods, including 30 years of out-of-sample data, prolonged periods of rising or falling rates, and a dataset of nine additional countries. Furthermore, the results are not explained by market or macroeconomic risks, nor can they be easily attributed to transaction costs or other investment frictions. These results reveal predictable dynamics in government bond returns relevant for academics and practitioners.

Real-time Bayesian learning and bond return predictability

The paper examines statistical and economic evidence of out-of-sample bond return predictability for a real-time Bayesian investor who learns about parameters, hidden states, and predictive models over time. We find some statistical evidence using information contained in forward rates. However, such statistical predictability can hardly generate any economic value for investors. Furthermore, we find that strong statistical and economic evidence of bond return predictability from fully-revised macroeconomic data vanishes when real-time macroeconomic information is used. We also show that highly levered investments in bonds can improve short-run bond return predictability.

Bond Return Predictability and its Economic Value: The Case of China

This paper investigates bond return predictability and its economic value. Using regression models, we first examine both the statistical and economic significance of bond return predictability in the Chinese market, and analyze the non-Markov and stochastic volatility properties of bond yields. On the basis of the above analysis, we propose a systematic method for constructing non-Markov dynamic term structure models (DTSMs) under a generalized Heath-Jarrow-Morton (HJM) framework with stochastic volatility. Then, we investigate the roles of the non-Markov property and stochastic volatility in bond return predictability and its economic gains realizing. Finally, we analyze the economic drivers of bond return predictability. Empirical results show that bond return predictability in the Chinese market is statistically significant, which also can be converted into significant economic gains. The non-Markov property and stochastic volatility are of critical importance for this conversion process. Moreover, time-varying risk premia driven by the economic environment are the main source of the bond return predictability in the Chinese market, while unspanned stochastic volatility factors also contain much information for future bond returns.

Forecasting bond returns in a macro model

We construct predicting factors based on the predictive errors of bond yields and macro variables implied by a linearized dynamic stochastic general equilibrium (DSGE) model to study the predictability of excess bond returns. Our predictors consist of two pieces of information. The predictive errors on bond yields capture the time-varying term premiums missing from the linearized DSGE model, and the predictive errors on macro variables reflect the macro risks that are un-spanned by the yield curve. Predictive regressions of one-year excess bond returns indicate that our predictors outperform the CP factor in Cochrane and Piazzesi (2005). We further relate the bond risk premiums to the inflation target. We find that bond risk premiums vary over time with the inflation target, and are earned as compensation for exposure to innovations of the inflation target. Finally, we show that the predictive errors of bond yields implied by a second-order approximation of the DSGE model still have considerable predictive power on excess returns.

The Yield Spread and Bond Return Predictability in Expansions and Recessions

Abstract This paper uncovers that expected excess bond returns display a positive correlation with the slope of the yield curve (i.e., yield spread) in expansions but a negative correlation in recessions. We use a macro-finance term structure model with different market prices of risk in expansions and recessions to show that a very accommodating monetary policy in recessions is a key driver of this switch in return predictability.

Predicting Bond Return Predictability

This paper provides empirical evidence on predictable time variations in out-of- sample bond return predictability. Bond return predictability is associated with periods of high (low) economic activity (uncertainty), which implies that violations of the expectations hypothesis are state-dependent and linked to features of the business cycle. These state-dependencies in predictability, established by introducing a new multivariate test for equal conditional predictive ability, can be used in real-time to improve out-of-sample bond risk premia estimates and investors’ economic utility through a novel dynamic forecast combination scheme that uses predicted forecasting performance to identify the best set of methods to include in the combined forecast. Dynamically combined forecasts exhibit strong countercyclical behavior and peak during recessions.

Predicting Bond Returns: 70 Years of International Evidence

We examine the predictability of government bond returns using a deep sample spanning 70 years of international data across the major bond markets. Using an economic, trading-based testing framework we find strong economic and statistical evidence of bond return predictability with a Sharpe ratio of 0.87 since 1950. This finding is robust over markets and time periods, including 30 years of out-of-sample data on international bond markets and a set of nine additional countries. Furthermore, the results are consistent over economic environments, including prolonged periods of rising or falling rates, and is exploitable after transaction costs. The predictability relates to predictability in inflation and economic growth. Overall, government bond premia display predictable dynamics and the timing of international bond market returns offers exploitable opportunities to investors.

Bond Risk Premiums with Machine Learning

Abstract We show that machine learning methods, in particular, extreme trees and neural networks (NNs), provide strong statistical evidence in favor of bond return predictability. NN forecasts based on macroeconomic and yield information translate into economic gains that are larger than those obtained using yields alone. Interestingly, the nature of unspanned factors changes along the yield curve: stock- and labor-market-related variables are more relevant for short-term maturities, whereas output and income variables matter more for longer maturities. Finally, NN forecasts correlate with proxies for time-varying risk aversion and uncertainty, lending support to models featuring both channels.

Underlying Bond Return Predictability by ETF Returns

This paper investigates whether ETF returns lead the returns of underlying bonds and similar style bond funds. Bond prices are often stale due to their lack of liquidity, and price discovery may occur in ETFs and then in underlying bonds. As predicted, we find that ETF returns predict its own NAV returns and aggregated ETF returns for each bond also predict the underlying bond returns on a monthly basis. We show that bond liquidity is the determining factor of the predictability and the role of authorized participants is critical to the dissemination of information from ETFs to underlying bonds. Bond returns are more predictable for illiquid bonds and bonds with less ETF ownership and less absolute ETF flows. Next, we examine whether bond fund returns are also predictable. We find the predictability of bond fund returns by same style ETF returns and a bond fund portfolio strategy based on this predictability, which indicates that the bond fund market is potentially vulnerable. Our research adds to the discussion of the impact of ETFs on market efficiency.

Real-Time Macro Information and Bond Return Predictability: Does Deep Learning Help?

This paper examines whether deep/machine learning can help find any statistical and/or economic evidence of out-of-sample bond return predictability when real-time, instead of fully-revised, macro variables are taken as predictors. First, when using pure real-time macro information alone, we find that deep learning cannot help find any statistical evidence for forecasting both non-overlapping and overlapping excess bond returns. In contrast, some machine learning models can help find some statistical evidence for forecasting overlapping excess bond returns. Second, when using both pure real-time macro information and yield curve information, we find that deep learning performs well for forecasting medium- and long-maturity overlapping excess bond returns, but such predictability is dominantly driven by yield curve information. Third, all statistical evidence of predictability is much weaker than that found from using fully-revised macro data and generates minimal economic gains for a mean-variance investor, regardless of her level of risk aversion and whether she can take short positions.

Monetary Policy and Bond Prices with Drifting Equilibrium Rates

We study drift and cyclical components in U.S. Treasury bonds. We find that bond yields are drifting because they reflect the drift in monetary policy rates. Empirically, modeling the monetary policy drift using demographics and productivity trends, plus long-term inflation expectations, leads to cyclical deviations of bond prices from their drift that predict bond returns in- and out-of-sample. These bond cycles can originate from term premia or temporary deviations from rational expectations in a behavioral framework. Through the lens of our model, we detect a significant role of the latter in determining the cyclical properties of yields with short maturities.

Sovereign bond return prediction with realized higher moments

This paper analyzes whether realized higher moments are able to predict out-of-sample sovereign bond returns using high-frequency data from the European bond market. We study bond return predictability over tranquil and crisis periods and across core and periphery markets at the index and country level. We provide fresh evidence that realized kurtosis is the dominant predictor of subsequent returns among higher moments and other predictors such as CDS spreads, short-term interest rates and implied stock market volatility. Our findings further underline that sovereign bond return predictability is stronger during crisis periods and more pronounced for bonds of lower credit ratings.

Gone With the Vol: A Decline in Asset Return Predictability During the Great Moderation

We document a significant shift in the comovement of asset returns and macroeconomic volatility during the Great Moderation. Strong U.S. stock and bond return predictability from several macroeconomic volatility series before 1982 was followed by a significant predictability decline during the Great Moderation (1982-2008). These findings are robust to alternative empirical specifications and out-of-sample tests. In a calibrated time-varying-volatility equilibrium model, the predictability decline is consistent with changes in monetary policy and shock dynamics. A stronger policy response to inflation and lower cost-push shock variance reduce the sensitivity of macroeconomic variables and asset returns to a persistent volatility factor, explaining the lower predictability. The results contribute to examine macroeconomic volatility as a driver of expected asset returns, and identify sources of the Great Moderation using asset price dynamics.