What drives long-term bond yields?
Yesterday, Ben Carlson of “A Wealth of Common Sense” asked how it could be that more than $13 trillion of government debt has a negative yield. As someone who lived in Switzerland for more than two decades I can tell you that this situation can last for many years, yet, one has to wonder who buys these bonds? In the case of Switzerland it is almost exclusively pension funds and insurance companies who have to own a lot of government bonds for regulatory reasons. The effect these regulations have on the yield curve can be quite substantial. So here is a partial answer to Ben’s questions:
Long-term government bond yields are amongst the most widely used benchmarks for long-term investors. Most of the time, 10-year government bond yields are used as benchmarks to assess the relative performance of equities but often these yields are also used for regulatory purposes, for example to calculate minimum return requirements for institutional investors like pension funds and insurance companies.
If you go beyond maturities of 10 years, pension funds and insurance companies become increasingly dominant as a demand factor for these bonds. Because these institutions have extremely long-dated liabilities, they are often forced to hold long-dated assets to reduce the interest rate risks in their portfolios. And of course, the local regulator demands that these long-dated assets should be held in the local currency rather than in foreign currencies.
But what happens if there are more pension fund and insurance assets than long-dated bonds in the market? One would assume that if there is more demand for long-dated bonds the yields on these bonds decline. And this is what a recent study by Robin Greenwood of Harvard and Annette Vissing-Jorgensen of UC Berkeley found. They looked at the yield spread between 30-year and 10-year government bonds and compared it to the size of the pension and insurance system in a country.
The chart below, taken from their paper, shows the net demand for long-dated assets (calculated as the assets under management of pension funds and insurance companies minus the amount of government debt relative to GDP) and the yield spread. Countries like Denmark or Switzerland with their mandatory private pension system have a very large pension sector and hence a large demand for long-dated bonds. Unfortunately, these countries have very low debt/GDP ratios so the net demand for long-dated government bonds is large.
In comparison, the pension system in Germany and Italy, for example. is largely based on a pay-as-you-go system so the pension funds tend to be smaller. The impact these institutional differences have on the yield spread are clearly visible. The larger the pension and insurance system in a country, the lower the yield spread between 30-year and 10-year bonds, i.e. the flatter the yield curve at the long end. While their paper does not investigate the steepness of the yield curve at shorter maturities, I would not be surprised to see a similar result there.
While these results are not too surprising in and of themselves, there is a hidden risk there. If the regulator in a country changes the discount rates used by pension funds and insurance companies or makes other regulatory changes that impact the asset allocation of these institutions, it may have a direct effect on the yields of long bonds independent of the macroeconomic environment. As the authors show in their paper based on six case studies from Denmark, Sweden and the Netherlands, the yield curve reacted strongly to changes in discount rate rules for pension funds and insurance companies. As a result, yields for long bonds increased significantly as demand declined and the financing costs for the government and corporates increased. What a wonderful case of unintended consequences.
Steepness of yield curve and size of pension system
Source: Greenwood and Vissing-Jorgensen (2018).