Reducing the factor zoo for bonds

Back in the day when I was working on multi-asset portfolios, I always struggled with bond selection. Unlike in equities, where different factors like value or momentum are a good guide to selecting the best stocks, hardly anything seems to work in bonds. Ironically, new research indicates that I was on to something.

The factor zoo in equities has by now grown to several hundred, but only a small number of these factors are important. The ground-breaking study of Campbell Harvey established that only value and momentum are factors that are statistically above suspicion. Recently, I wrote about research by Alexander Swade and his collaborators that showed that with five factors you can cover the entire space of all equity factors: value, momentum, quality, and two growth-related factors.

A new paper tried to do something similar for bonds. It looked at a total of 563 trillion possible combinations of 49 different bond and equity factors identified in the literature to check which one of these factors is likely to be a unique risk factor priced in bond markets.

Looking at these different factors and the probability of these factors ending up in the final selection of explanatory factors they conclude that only three factors are likely to contribute to bond pricing.

The first one is the market factor, meaning bond prices are driven by their beta to the overall bond market. Nothing too surprising there.

The second factor is the steepness of the yield curve and its reflection of expected economic conditions. A steeper yield curve means that riskier bonds will outperform safer ones and vice versa.

Finally, the third factor has only been described in the literature a couple of years ago and it is the bond equivalent of post-earnings announcement drift. I will talk about this effect in some detail next week, so for now, I will leave it at this cliffhanger.

However, looking through the paper of Alexander Swade, I cannot help but notice that it all boils down to just some simple rules for bonds. First, look at the beta of bond price to the market overall and this will drive most of your performance. Second, if the steepness of the yield curve is negative or falling, move into safer bonds, if it is steepening or positive, move into riskier bonds. And that’s about it. No fancy factors there, just common sense.