Investors may be losing significant returns by failing to apply factor investing to bonds, says Barbara Petitt 

At a glance

• Harnessing corporate bond factors may help enhance risk-adjusted returns.
• Factor investing in bond markets has not taken off to the same extent as in equity markets.
• The evidence for multi-factor investing in bonds is compelling but there are relatively few options available to implement the approach.

Factor-based investing is a widely used approach in equity markets but far less known and used in bond markets. Yet compelling evidence is emerging of the value of applying factor investing to bonds.

The study ‘Factor Investing in the Corporate Bond Market’ by Patrick Houweling and Jeroen van Zundert published in the CFA Institute Financial Analysts Journal, finds that identifying corporate bond factors and then combining them in multi-factor portfolios may help enhance risk-adjusted returns. This study, by two investment practitioners, should interest not only bond investors, but also all institutional investors seeking to harvest the various premia offered in financial markets.   

It is fair to say traditional allocation by asset class is being superseded by factor investing.

Factors can be described as the persistent return drivers of portfolios. The academic community and the asset management industry have documented more than 300 factors that have been shown to contribute to excess equity returns, and new factors are added regularly.

Numerous studies report that factor investing in equity markets has produced superior risk-adjusted returns both individually and in combination. In particular, strategies based on value, momentum, low-volatility, size and quality factors have been shown to outperform traditional cap-weighted market indexes.

Little wonder, then, that a large number of institutional mandates are now focused on factors and a wide range of financial products has come to market predicated on factor investing. According to investment industry estimates, more than half of institutional investors classify their portfolios by factor, rather than by asset class.

Indeed, an industry has grown up around factor investing. Morningstar estimates that the smart beta industry, a subset of the factor-based universe, has about $400bn (€370bn) in assets under management, which is a four-fold increase since 2010. The total is probably closer to $1trn if separate mandates are included.

However, factor investing in bond markets has not taken off to the same extent as in equity markets. Research into factor-based bond strategies remains limited, probably because of the reduced availability of bond data, which creates obstacles to research. The work of Houweling and van Zundert can help investors better understand the potential of factor investing beyond equities.

The authors studied 1.3m bond-month observations, of which about 900,000 were for investment-grade bonds and about 400,000 for high-yield bonds.

They started their analysis with the construction of factor portfolios that sort corporate bonds according to four specific characteristics – size, low risk, value and momentum. These four factors mirror the descriptions of factors used in equities. But the authors used bond-specific factor definitions that are based on bond characteristics such as credit rating, maturity and credit spread.

The size portfolio is comprised of the bonds of small companies, based on the market value of their outstanding bonds. The low-risk portfolio consists of short-maturity bonds with a high credit rating. The value portfolio is composed of bonds whose credit spread is high relative to a model-implied fair spread. Finally, the momentum portfolio is comprised of bonds with high past returns.

As well as creating portfolios of individual factors, the authors went a step further, constructing a multi-factor portfolio combining these individual factors. It is with this additional step that their work adds most value.

Their study also improves on previous work by using more realistic assumptions. For example, as shorting corporate bonds is difficult and costly, they analyse long-only portfolios. To prevent extreme turnover that would lead to high transaction costs and unprofitable strategies, they also assume a bond-holding period of 12 months instead of one month, as used in a previous study.

The findings of Houweling and van Zundert’s study could affect how investors allocate to bonds.

Both the single-factor and multi-factor portfolios generate economically meaningful and statistically significant alphas over the 21-year period studied (January 1994 to June 2015). That is, the portfolios outperform the benchmark bond indexes – the Barclays US Corporate Investment Grade index and the Barclays US Corporate High Yield index.

For portfolios of investment-grade and high-yield bonds, the authors discovered outperformance above the benchmark for size (1.12% and 5.50%, respectively), low risk (0.41%, and 1.45%), value (1.3% and 4.26%) and momentum (0.3% and 2.04%). The authors note that the outperformance of these factor premiums is substantial, pointing out that investors could have tripled their long-term average excess returns by investing in these factors, compared with passively investing in the corporate bond market indexes.  

The authors also observed that most of the correlations between factors tend to be below 20%, the exception being the correlation between value and size. The lowest correlation is between value and momentum.

Because of these low correlations, the authors investigated the diversification benefits of a multi-factor portfolio with equal allocations to the four factors. They found that the multi-factor portfolio had a lower tracking error and a higher information ratio than each of the single-factor portfolios.

Finally, the study revealed the benefit of applying factor investing not only in the equity market, but also in the corporate bond market. The authors observed that in doing so, investors could increase the alpha of their multi-asset portfolio by more than 1% a year.

The authors say their findings provide investment managers and their clients with a clear opportunity to enhance risk-adjusted returns from their bond allocations.

According to them, a multi-factor strategy is the optimal way to seize this opportunity. The magnitude of the historical premium for any single factor might not be replicated in the future, which argues for a multi-factor strategy. In addition, the tracking errors of individual factors in relation to the market are relatively large, but the tracking error of the multi-factor portfolio is well below the average of the tracking errors of the individual factor portfolios.

But if the evidence for multi-factor investing in the bonds space is compelling, the options for implementing the strategy are less so, the study notes. There are currently few investment vehicles through which investors can harvest factor premiums in the corporate bond market. There is even a dearth of single-factor bond funds which, combined, could serve as the building blocks of a multi-factor portfolio.

This gap may be the result of wide bid-offer spreads on many corporate bonds, or because bond trading execution can be expensive. Or maybe bond managers are deterred from creating single-factor bond portfolios by the large tracking errors of individual factors and their low information ratios, the authors note.

Investors persuaded by the corporate bond factor argument might, therefore, have to take a do-it-yourself (DIY) approach. If investors do decide to create their own strategies, they can control their factor exposures and allocate to factors in the proportions they wish and vary these proportions over time to suit their needs.

However, the DIY approach will probably be available only to asset owners that have the skills to create a suitable benchmark and to measure returns against it. The others may just have to wait until commercial asset managers have developed the expertise to offer off-the-shelf products.

Barbara Petitt is the managing editor of the CFA Institute Financial Analysts Journal