Inflation and Inflation Expectations
Inflation expectations play a crucial role in monetary policy making. Not only do they tell policymakers something about the real expected cost of borrowing and hence the viability of investment plans, they also help policymakers gauge the public's perception of the central bank's commitment to maintaining a low and stable rate of inflation. Especially in the current policy environment, where the Fed has been forced by events to take unconventional actions, it is more important than ever to make sure that long-run inflation expectations are well anchored and that the policy message is well understood by the public.
In principle, expectations are not observable. But there are at least two sources that can be used to infer them: surveys and market-based information. With surveys, people are asked directly what they think future price growth will be. There is a variety of surveys that are regularly conducted and that target different types of participants. Here, we will focus on the University of Michigan's Survey of Consumers, theLivingston Survey, and the Survey of Professional Forecasters (SPF). Measures taken from readily available market-based information usually exploit information that is contained in the yield curve of Treasury securities. In particular, some measures rely on the yields of Treasury inflation-protected securities (TIPS), which are traded daily in the secondary market. (See this Economic Commentary for a new method of gauging inflation expectations).
The University of Michigan survey is different from other surveys because participants are actual consumers and not professionals, as they are in the Livingston, the SPF, or the Blue Chip surveys. A look at the survey's measure of one-year inflation expectations (1978-present, monthly frequency), shows that the median forecast quite often lags actual inflation, which suggests that current inflation plays an important role in determining inflation expectations. Most of the time, actual inflation falls inside the 25th and 75the bands, with notable exceptions in the 1980s and most recently: The 2009 dramatic drop in prices was completely unexpected. It is also worth mentioning that the variety of opinions is quite substantial. For example, in the latest available reading of September 2009, only half of the people surveyed reported an inflation forecast that was between 0.1 percent and 4.9 percent; the rest had even more extreme views! In any case, all the percentiles tend to move together and, after increasing the substantially in the spring and summer of 2008, the latest figures give us a description of short-run inflation expectations just below their historic average.
The five-year forecast also has a very similar pattern, with the latest numbers close to their normal levels. However, this series is much less volatile, and, accordingly, the 25th and 75th percentile bands are also narrower. Given the longer horizon, we might be surprised to see that the recent median inflation expectation is quite stable around 3 percent, which is higher than the actual inflation comfort zone of 2 percent - 2.5 percent described often by the Fed. In part, this might reflect a bias due to the fact that, when people think of the CPI, they put less emphasis on the prices of goods they buy less often, like durable goods. At the same time, the prices that have decreased the most in the last few decades have been exactly those for durable goods. Moreover, it is also true that the forecasts vary substantially, which may be in part because each individual consumer perceives inflation in terms of his or her own personal consumption bundle. In any case, the graph below shows that the medium-run inflation expectations of the participants in the University of Michigan survey have not changed.
If the two charts above tell us what common people think about inflation, the next two show what professional forecasters and other business professionals think. In principle, the professionals should be more aware of what they are asked to forecast, and in fact, the levels forecasted are more in line with actual CPI-inflation. At the same time, trends in short-run and medium-run inflation expectations are similar to those of the University of Michigan survey.
One element that is worth mentioning is that uncertainty has recently increased across all forecast horizons in all the surveys discussed here. Uncertainty can be proxied for by the standard deviation of the forecasts at each point in time. The deviation has been going up for all measures, which probably reflects a policy environment that is more uncertain than usual and implies a wider set of views on the effects of the current policy actions on the economy and, hence, prices.
Looking at the charts above, we might argue that surveys are not bad at forecasting actual inflation. What is not shown in the figures, however, is that most measures missed big changes in inflation, like those at the end of the 1960s, the 1970s, and the 1980s (a disinflation). The surge and the drop in inflation of the past three years was also missed. This seems to suggest that big movements in commodity prices and their impact on CPI inflation are always hard to anticipate.
TIPS-based measures of expected inflation are obtained by taking the difference in the yields of conventional Treasuries and TIPS. In principle, the yield differences may provide an accurate measure of market inflation expectations because inflation has very different effects on the returns of the two kinds of securities. The yield on a conventional Treasury must compensate the buyer for any expected erosion in purchasing power due to future inflation. In contrast, the buyer of an inflation-protected Treasury need not worry about future inflation because the principal and interest payments are both indexed to inflation. This spread between the two yields is called the breakeven inflation compensation.
However, the breakeven inflation rate alone is not correctly interpreted as a measure of inflation expectations. Investors also attach premiums for inflation risk and liquidity to their required return. The inflation risk premium means that the breakeven inflation overestimates expected inflation, and the liquidity premium means it underestimates expected inflation.
While the inflation risk premium seems to be quite steady over time, the liquidity premium can move substantially. The liquidity premium arises because the TIPS market is a relatively recent one and it is not as deep as the one for conventional Treasuries. Over time, the value of TIPS outstanding has grown, as has their trading volume. The first indexed Treasury was issued in January 1997, with a maturity of 10 years, and since then, the U.S. Treasury has regularly issued 10-year TIPS and other maturities. However, a substantial difference in liquidity relative to conventional Treasuries persists. Furthermore, changes in the liquidity premium are exacerbated in moments of market turbulence, adding volatility to the inflation breakeven rate that has nothing to do with inflation expectations.
A way to see the effect of liquidity is to plot the spread between an off-the-run (old) and an on-the-run (new) Treasury security. As we can see in the chart below, the premium skyrocketed after the Lehman Brothers collapse.
It is possible to use the liquidity-premium spread to adjust the breakeven inflation compensation for movements in the liquidity premium. This is done in the chart below, where we plot the adjusted inflation compensation that comes out of 10-year TIPS (together with the five-year to five-year forward rate). The adjusted inflation breakeven rate was quite stable up until the end of 2007, between 2 percent and 2.5 percent. After that, its volatility increased, peaking at the moment of highest financial turbulence in the early fall of 2008, and after that, it collapsed, probably signaling deflation fears. However, the most recent readings of the series are back to historically normal values.