From Project Syndicate:
The prices of long-term government bonds have been running very high in recent years (that is, their yields have been very low). In the United States, the 30-year Treasury bond yield reached a record low (since the Federal Reserve series began in 1972) of 2.25% on January 30. The yield on the United Kingdom's 30-year government bond fell to 2.04% on the same day. The Japanese 20-year government bond yielded just 0.87% on January 20.
All of these yields have since moved slightly higher, but they remain exceptionally low. It seems puzzling – and unsustainable – that people would tie up their money for 20 or 30 years to earn little or nothing more than these central banks' 2% target rate for annual inflation. So, with the bond market appearing ripe for a dramatic correction, many are wondering whether a crash could drag down markets for other long-term assets, such as housing and equities.
It is a question that I am repeatedly asked at seminars and conferences. After all, participants in the housing and equity markets set prices with a view to prices in the bond market, so contagion from one long-term market to another seems like a real possibility.
I have been thinking about the bond market for a long time. In fact, the long-term bond market was the subject of my 1972 PhD dissertation and my first-ever academic publication the following year, co-authored with my academic adviser, Franco Modigliani. Our work with data for the years 1952-1971 showed that the long-term bond market back then was pretty easy to describe. Long-term interest rates on any given date could be explained quite well as a certain weighted average of the last 18 quarters of inflation and the last 18 quarters of short-term real interest rates. When either inflation or short-term real interest rates went up, long-term rates rose. When either fell, so did long-term rates.
We now have more than 40 years of additional data, so I took a look to see if our theory still predicts well. It turns out that our estimates then, if applied to subsequent data, predicted long-term rates extremely well for the 20 years after we published; but then, in the mid-1990s, our theory started to overpredict. According to our model, long-term rates in the US should be even lower than they are now, because both inflation and short-term real interest rates are practically zero or negative. Even taking into account the impact of quantitative easing since 2008, long-term rates are higher than expected.
But the explanation that we developed so long ago still fits well enough to encourage the belief that we will not see a crash in the bond market unless central banks tighten monetary policy very sharply (by hiking short-term interest rates) or there is a major spike in inflation.
Bond-market crashes have actually been relatively rare and mild. In the US, the biggest one-year drop in the Global Financial Data extension of Moody's monthly total return index for 30-year corporate bonds (going back to 1857) was 12.5% in the 12 months ending in February 1980. Compare that to the stock market: According to the GFD monthly S&P 500 total return index, an annual loss of 67.8% occurred in the year ending in May 1932, during the Great Depression, and one-year losses have exceeded 12.5% in 23 separate episodes since 1900.