Expected Business Conditions and Bond Risk Premia

Abstract

This paper studies the predictability of bond risk premia by means of expectations to future business conditions using survey forecasts from the Survey of Professional Forecasters. We show that expected business conditions consistently affect excess bond returns and that the inclusion of expected business conditions in standard predictive regressions improve forecast performance relative to models using information derived from the current term structure or macroeconomic variables. The results are confirmed in a real-time out-of-sample exercise, where the predictive accuracy of the models is evaluated both statistically and from the perspective of a mean-variance investor that trades in the bond market.Keywords: Bond risk premia, expected business conditions, predictability, economic value, expectations hypothesis, time-varying risk premia JEL Classification: E43, E44, E47, G11, G12. IntroductionThere is mounting evidence that the risk premia required by investors for holding Treasury bonds contain a time-varying and predictable component. One strand of research relates such variations to forward spreads (Fama and Bliss, 1987;Fama, 2006), yield spreads (Keim and Stambaugh, 1986;Campbell and Shiller, 1991), and forward rates (Cochrane and Piazzesi, 2005;Dahlquist and Hasseltoft, 2013;Zhu, 2015), whereas more recent studies link the predictable component to factors whose variations lie outside the span of current yields. Existing research within this strand has uncovered a wide array of factors including jump risk (Wright and Zhou, 2009), option prices (Almeida, Graveline, and Joslin, 2011), and macroeconomic variables (Ilmanen, 1995;Moench, 2008;Cooper and Priestley, 2009;Ludvigson and Ng, 2009;Wright, 2011;Favero, Niu, and Sala, 2012;Joslin, Priebsch, and Singleton, 2014;Zhou and Zhu, 2015).While existing studies mainly rely on information in the current term structure and business environment to explain variations in the predictable component of bond risk premia, our study takes a forward-looking perspective by studying the link between expected business conditions and bond risk premia using survey forecasts from the Survey of Professional Forecasters (SPF). Albeit empirical studies frequently conclude that macroeconomic fundamentals carry information about bond risk premia not already captured by the yield curve, they rarely account for issues with publication lags and data revisions in macroeconomic time series. As recently demonstrated by Ghysels, Horan, and Moench (2014), such features may drive a wedge between the factor spaces spanned by latent common factors extracted from a panel of revised and non-revised data, respectively, implying that a sizable fraction of the predictability documented by Ludvigson and Ng (2009) may largely be driven by revision and publication lag components that are unavailable to an investor in real-time. Survey forecasts, as a result, provide an ideal data source for studying the predictability of bond risk premia as they are model-free, available in real-time, and not...

Predicting Bond Return Predictability

“…We document that bond 1 Early studies include Fama and Bliss (1987), Keim and Stambaugh (1986), Fama and French (1989), and Campbell and Shiller (1991). More recent studies of bond return predictability includes Cochrane and Piazzesi (2005), Cooper and Priestley (2009), Ludvigson and Ng (2009), Cieslak and Povala (2015), Eriksen (2017), Ghysels, Horan, and Moench (2018), Gargano, Pettenuzzo, and Timmermann (2019), Berardi, Markovich, Plazzi, and Tamoni (2020), and Bianchi, Büchner, and Tamoni (2020).…”

“…relative to the EH in Panel A, for EW in Panel B, and for our dynamic forecast combination strategy relative to the EH and the equal-weighted combination strategy in Panels C and D, respectively. In our main results, we set γ = 10 as in Eriksen (2017), but show in the 30 We always use the same variance estimated over the same period as the forecasts for all models so that the optimal portfolio weights only differ because of differences in the excess bond return forecast.…”

“…γ = 5. In order to evaluate the statistical significance of the CER gains, we follow Eriksen (2017) and Gargano et al (2019) and conduct a conventional t-test on the mean of the time series of realized utility differences. Standard errors are obtained using a Newey and West (1987) estimator.…”

“…For example, while in-sample evidence frequently points to predictability by means of variables such as forward spreads (Fama and Bliss, 1987), yield spreads (Campbell and Shiller, 1991), forward rates (Cochrane and Piazzesi, 2005), and macroeconomic variables (Cooper and Priestley, 2009, Ludvigson and Ng, 2009, Cieslak and Povala, 2015, Eriksen, 2017, out-of-sample exercises often fail to deliver consistent evidence of predictability and statistical and economic evaluations often disagree (Thornton andValente, 2012, Sarno, Schneider, andWagner, 2016). Our findings suggest that these puzzling contradictions can be explained by state-dependent bond return predictability.…”

“…We detail the construction of monthly log yields and bond prices in the Internet Appendix. The data are available at https://www.federalreserve.gov/data/nominal-yield-curve.htm.9 The same set of maturities are considered in, e.g.,Fama and Bliss (1987),Cochrane and Piazzesi (2005),Ludvigson and Ng (2009),Thornton and Valente (2012),Eriksen (2017), andGargano et al (2019).…”