Inflation and inflation expectations

15 Nov 2010 by Jim Fickett.

The possibility of future inflation is currently much under discussion, and inflation expectations, as measured in the bond markets, play a significant role in that discussion. However, although measures of inflation expectations may be useful to policy makers, they probably have little predictive power for investors.

Many investors, commentators, and economists use bond yields as a measure of inflation expectations. The simplest measure is just to look at straight Treasury yields – if yields are very low, as currently, most investors must not be very worried about inflation. More common is a slightly more sophisticated measure, the so-called break-even inflation rate, which is the difference in yield between nominal bonds and Treasury Inflation Protected Securities (TIPS) of the same maturity. It is called the “break-even” rate because if that yield difference turns out to be the actual inflation rate then it makes no difference whether you buy the nominal bond or the TIPS.

Most investors realize that the break-even rate should be taken with a grain of salt, because many other factors enter into bond prices besides inflation expectations, including liquidity issues, risk premia and, in the case of TIPS, deflation as well as inflation protection. Measures of inflation expectations based on bond prices can become quite complicated.

But there is a fundamental question behind this whole area that is rarely addressed, and that is, do inflation expectations actually tell you anything about the future? Or, to be more specific, do inflation expectations in the present correlate with actual inflation in the future?

It would be nice to do a careful quantitative comparison of the break-even rate, or one of its more sophisticated cousins, to actual forward inflation, over a long time period. However the Federal Reserve only supplies historical data on TIPS yields back to 2003. That is too short a history to be useful. The UK has a longer history with index-linked gilts, which are similar to TIPS, but even there the historical data only go back to 1981. The study would not be very interesting without including at least one period of higher inflation.

So we are limited to nominal Treasury yields. But we can make use of even this basic information. The baseline is current inflation – bond yields tend to be pretty closely related to current inflation plus a small real yield. So, if the bond markets have any ability at all to predict inflation, one would at least expect that the excess of current yields over current inflation would correlate with the excess of actual future inflation over current inflation.

The markets are probably not very good at predicting anything many years in the future but since, for example, growth in the money supply does correlate with inflation 2-3 years later, one might hope that the bond market could help predict inflation over the next few years. For each point in the following scatter plot the x coordinate is the 5-year Treasury nominal yield in a particular month, less the year-over-year percent change in the CPI as of that same month, and the y coordinate is the CAGR in the CPI over the following five years, again minus the concurrent YOY CPI change. Data are for the period 1954-2004.

There is no useful correlation. The vertical line shows, for example, the current situation. The 5-year Treasury is yielding 1.5% and the most recent figure for the CPI is 1.1%, so the difference is 0.4%. At that yield, the predicted inflation increase could be anywhere from about -8 to +3 percentage points.

Note that I am not saying inflation expectations have no value. Central bankers attempt to manage both actual and expected inflation. If they are going to manage inflation expectations they need to measure them. So inflation expectations, as measured in the bond markets, are certainly useful as policy tools. But they are not useful, in any pragmatic predictive sense, to investors.