Factor Investing in Sovereign Bond Markets: A Time-Series Perspective

printer-friendly version


Factor Investing in Sovereign Bond Markets: A Time-Series Perspective



The abundance of theoretical and empirical research on factor investing in the equity universe (see for example Amenc and Goltz (2016) for an overview) stands in sharp contrast to the relative scarcity of research about how to efficiently harvest risk premia in bond markets. That relatively little is known about the out-of-sample performance of factor-based bond portfolio optimisation models is perhaps surprising given the importance of fixed-income investments in institutional and private investors’ portfolios. From the investment practice standpoint, a similar contrast actually exists between factor investing in the equity space, which is a relatively mature subject, and factor investing in bond markets which still is in its infancy.

The paper "Factor Investing in Sovereign Bond Markets: A Time-Series Perspective" is the first of three papers that provide a detailed analysis of the theoretical, empirical and practical challenges related to factor investing in sovereign bond markets. This first paper has a focus on factors that are relevant from a time-series perspective (i.e. factors such as the “level” or “slope” of the yield curve that explain for any maturity a large fraction of differences over time in bond returns). In two companion papers (Maeso et al. (2019a,b)), authors present a related analysis of the factors that are relevant from a cross-sectional perspective (i.e. factors such as “value” or “momentum” that explain at any point in time a large fraction of differences in the cross-section of bond returns).[1] It should be noted that both of these papers address the question of factor investing from the perspective of a single credit-risk-free issuer, which is a priori the purest and most difficult problem since neither time-series nor cross-sectional differences in risk and performance can be explained by differences in creditworthiness, as in the case of a multiissuer universe. In other words, authors' ambition is to explore whether it is possible to identify economically justifiable strategies which, after accounting for transaction costs and other forms of trading frictions, generate excess returns when investing in a relatively homogenous set of highly correlated securities.

In this first paper, authors provide a positive answer to this question with respect to time-series factors such as the level of interest rates and, albeit to a lesser extent, the slope of the yield curve. More precisely, they first confirm (using both yield curve and CUSIP-level data in the US) the well-known finding that long-term bonds do appear to offer a higher unconditional excess return over shortterm rates, an excess return which is also known as the bond risk premium. They also confirm that the magnitude of the excess return generated by such an unconditional carry strategy is relatively small (much smaller than the equity risk premium) and has been negative for extended periods. As a matter of fact, the risk premium associated with an unconditional exposure to the slope factor (via flattener or steepener strategies) is even smaller, if not zero).

On the other hand, they show that one can efficiently exploit the presence of state-dependencies in excess returns from the level factor to implement a conditional version of a carry strategy based upon a variety of return-predicting factors. Interestingly, this strategy shows a much better resilience to increasing interest rates environments with respect to the benchmark or to an unconditional version of the carry strategy. It also has the highest outperformance potential in bear equity markets, thus suggesting very attractive diversification benefits.

They additionally find, using yield curve data, that the performance of a conditional version of a flattener strategy based upon a time-varying exposure to the slope factor also generates economically significant additional performance, but such excess performance is limited by the presence of leverage constraints, and we therefore remain cautious with respect to its implementability with individual bonds.

A natural question arises with respect to the explanation of such time-series predictability in risk-free government bonds. In particular, one may wonder whether this predictability admits a rational explanation, or whether it is due to such behavioural features as investor overreaction. From a rational investor perspective, the well-documented predictability of time series excess returns in Treasuries is most naturally explained in terms of state-dependent risk premia. However, if the excess returns were truly due to risk premia, their asset-pricing origin should leave a detectable signature in the good and bad the pay-offs of the securities occurring in good or bad states of the world. Can this signature be detected? And, if not, can alternative explanations be offered for the return predictability? Irrespective of whether the predictability stems from a statedependent risk premium or from other sources, how can one best harvest the corresponding premium, both from an unconditional perspective, where the exposure is maintained over time irrespective of changes in market conditions? And how can the same corresponding premium be harvested from a conditional perspective, where the exposure is adjusted over time as a function of changes in market conditions? These are some of the key questions that authors would address in this paper.


The research from which this study is drawn was produced as part of the Amundi research chair at EDHEC-Risk Institute on “ETF, Indexing and Smart Beta Investment Strategies.”


A copy of the mentionned paper can be downloaded via the following link: Factor Investing in Sovereign Bond Markets - A Time-Series Perspective

You can access the editorial – EDHEC-Risk on Factor Investing in Sovereign Bond Universes – published in the April edition of EDHEC-Risk Institute quarterly newsletter.



[1] There exists of course a strong relationship between the time-series and cross-section perspective. In particular, the intertemporal capital asset pricing model of Merton (1973) suggests that factors that proxy for time-varying investment opportunities should be priced in the cross-section.