Investor demand and government bond pricing

Summary:

• Central banks have started to reverse a decade of government bond purchases while sovereign debt is currently projected to remain high across most advanced economies. This raises the question of how different investor groups will respond to an increase in debt supply, and what the implications are for long-term yields. To answer this, we use data on investor holdings of sovereign debt to investigate how investor demand and composition affect government bond yields in advanced economies.
• We start by documenting differences in investor composition across government bond markets. We then study how debt supply is typically distributed across investors. We identify who the marginal buyers of government debt are by estimating the marginal responses of investor holdings to changes in government debt. We find that non-banks (such as investment funds) increase their holdings at a faster rate than banks. Investor marginal responses also show marked differences across countries.
• Using a demand system for government bonds that directly relates investor holdings to bond prices, we estimate the price elasticity of demand for each investor group. We find that foreign non-banks are by far the most elastic investor while domestic banks tend to be quite inelastic.
• We combine our estimates of marginal response and demand elasticity by investor group to quantify the yield impact of an increase in government debt. All else equal, a 10 percent increase in government debt increases ten-year bond yields by approximately 60 basis points on average. Using differences in investor composition, we also quantify the yield impact across countries. For the US, we estimate a yield impact of approximately 100 basis points.
• We use our demand system approach to evaluate the role of quantitative easing in explaining the gap between long-term real yields and long-term growth expectations over the past decade. We also investigate the implications of its reversal, known as quantitative tightening, for long-term yields. We find that the effects of quantitative tightening on yields can be meaningful, especially with the prevalence of price-inelastic investors. For instance, we estimate that if the US Federal Reserve were to sell an amount of bonds equal to 50 percent of the size of its purchases during the Covid-19 pandemic, yields could increase by about 80 basis points.