For at least the past five years we have told investors that we expect yield curves to flatten and that long dated bonds would outperform. For those of us who have managed bond funds over multiple decades, such a conclusion is easy to grasp. However, as the ‘Taper Tantrum’ of 2013 clearly demonstrated, not all investors share that intuition. Back then, at least 99.9% of investors seemed to think that short dated bonds would outperform, despite the fact that short dated bonds had lost money in each of the previous three tightening cycles.
If we look at total returns for US Treasuries since that now infamous speech by Ben Bernanke, 30-year bonds did indeed outperform, with the 30-year gaining just over 4% in total return terms, and 5-years losing 4.1% based on returns for constant maturity indices. Recent experience is no different with the past 12 months showing gains for 30-years (up 2.1%) and losses for 5-years (down 1.9%) using the same indices.
Looking ahead, we continue to see flattening as short rates rise, with 3 rate hikes from the Fed in 2018 being our best guess. So, when asked about the shape of the US Treasury yield curve our answer is straightforward. Unless inflation picks up sharply such that Fed gets ‘behind the curve’, that will likely remain our view until we see the tightening cycle reaching a peak. Our analysis suggests that is some way in the future.
However, as we focus mainly on sovereigns, quasi-sovereigns and corporate bonds of creditor nations, rather than Treasuries, the shape of the yield curve, whilst important, has relatively little influence on our returns. So instead of the shape of the yield curve, we often get asked a related question about the ‘duration’ of our portfolios. Underlying this question is the assumption that knowing the ‘duration’ of the portfolio will provide some useful information.
Unfortunately, knowing the ‘duration’ of a portfolio, or even the ‘duration’ of a particular bond provides no more information than knowing that the temperature outside is ‘32 degrees’. If you are based in London you would probably guess that the temperature being referred to was 32 degrees Fahrenheit, but without such a qualification ‘32 degrees’ has an ambiguous meaning. Similarly with ‘duration’. To know the implications of a duration of x, you need to know what type of duration is being referred to. Had you begun 2008 with portfolio of high yield bonds with a 10-year duration then your year would have been rather unpleasant. Had your portfolio consisted of US Treasuries with a 10-year duration going into 2008 though, then you would have been delighted.
To bring this to life with a more recent example, let’s compare the returns of Southern Copper 2035’s, a very undervalued bond with a high expected return 12 months ago, with the US Treasury 5.25% Nov 2028, which has a similar duration. Over the past 12 months the Southern Copper issue has returned 15.41% whereas the Treasury returned just 0.31% over the same period.
Rather than focus purely on duration, a better measure of risk would be to compare duration times spread. A high yield bond with a five year duration and a spread of 400 over is roughly twice as risky as a bond with a 10 year duration with a spread of 100. It’s not perfect, and provides only an illustration, but it’s a vastly superior measure than duration in isolation.