Survey-Based Measures of Inflation Expectations
Expectations play a central role in economics. When faced with a pricing or an investment decision, people typically consider what they expect the future to look like and base their decisions accordingly. Their expectations for inflation not only reflect their perceptions about the future, they also directly affect actual levels of current and future inflation. Because of their importance in shaping economic outcomes, the expectations about the future price level are one of the major areas on which central banks focus.
There are two sources of data on inflation expectations. One is derived from market prices of various financial securities; the other is surveys of the general public and professional forecasters and economists. A well-known market-based measure uses the spread between the yields of nominal Treasuries and TIPS. Another is inflation swap rates. (For a new, model-based measure of inflation expectations that uses these market prices, see this Economic Commentary). Popular surveys of expectations include the University of Michigan’s Survey of Consumer Attitudes and Behavior (U of M survey), the Survey of Professional Forecasters (SPF), and the Blue Chip Survey.
Here we look at recent trends in these survey-based measures of inflation expectations. Survey-based measures provide some information that most market measures don’t, including shorter-term expectations and distributions among different survey participants.
The respondents of the U of M survey are consumers. Among other things, they are asked how much they expect prices to change over next 12 months, but in general terms, not relative to any price statistic or consumption basket. In contrast, the SPF and the Blue Chip Survey ask participants—professionals and economists—for their inflation expectations with respect to particular measures including the CPI. The SPF and Blue Chip Survey ask respondents about their inflation expectations for specific quarters, and the annual inflation expectations are computed from the quarterly figures. In addition, the SPF is conducted quarterly, whereas the other two are conducted monthly. We interpolated monthly figures for the SPF from the quarterly ones.
When the inflation expectations calculated from these three surveys (specifically, the median expectation for the next year) are plotted alongside CPI inflation that has been shifted 12 months forward (to line up expectations with the inflation they predict), we see that the survey-based estimates are imperfect predictors of actual inflation. In relatively stable periods, they are better in forecasting, but they generally fail to predict big movements in inflation. Another observation is that early in recessions, expectations generally exceed actual inflation. This may reflect the difficulty in predicting recessions since recessions are generally associated with less inflationary pressure.
Median short-term inflation expectations increased at the onset of the recession, but they dropped sharply during the financial crisis. They have picked up to moderate levels since March 2009. The swings following the recession were substantial; for example, U of M expectations declined to 1.7 percent in December 2008 from 5.2 percent in May 2008. For the other two surveys, the declines were not as large, but they were still significant: 1.1 percent for the SPF from the second quarter of 2008 to the first quarter of 2009, and 1.5 percent for the Blue Chip Survey from May 2008 to March 2009. Since then, U of M inflation expectations have displayed a rising trend, reaching 2.7 percent in February 2010, while both of the other measures have hovered around 1.7 percent to 1.8 percent.
What about longer-term inflation expectations? Median 5-year inflation expectations from the U of M survey have shown a declining pattern since mid-2008. The 5-year inflation expectation from the SPF declined until recently, while the 10-year expectation hardly moved except for the last quarter of 2009. These two measures bounced up in the first quarter of 2010.
On the other hand, a market-price-based measure of inflation expectations (from 5- and 10-year TIPS and nominal Treasury securities), which we plotted for comparison, has been quite volatile in this recession. It rapidly declined to 0.7 percent in December 2008 from its level of around 2.6 percent in the earlier months of 2008. Since then, it has gradually been increasing, and as of February 2010, it is 2.6 percent. Though we have used this measure to alleviate problems arising from liquidity and inflation risk premiums, its volatile behavior shows that it may be substantially contaminated by these effects. Overall, recent longer-term inflation expectations are below or around their historical averages, showing no substantial pressure for future inflation.
The median values from the surveys show only a general tendency for inflation expectations. When we look at the distribution of expectations among the survey respondents, we see substantial disagreement. While such disagreement used to be associated with the volatility of the economic environment, it may also arise from different perspectives, experiences, or information sources among the participants. Whatever the underlying source, the divergence of expectations for short-term inflation picked up substantially following the recession.
During the volatile period of 2008, the range between the 25th and 75th percentiles of 1-year inflation expectations from the U of M survey was at its highest level since the early 1980s disinflationary period. SPF 1-year inflation expectations also became more dispersed during this time. Some participants of the surveys expected a disinflationary or deflationary period, while others expected an increase in inflation. (The deflationary expectations can be seen in the 25th percentile of the 1-year U of M inflation expectations, which were negative between November 2008 and April 2009. Although we can see a disinflationary expectation for some SPF respondents, we don’t see a negative figure for the average of the bottom 10 of the SPF.) This, in itself, shows how volatile the environment was after the financial crisis. The dispersion of expectations is still quite high, although it has declined substantially recently.
Disagreement about longer-term expectations also increased in this recession according to the U of M Survey and the SPF. However, it did not increase quite as substantially as it did for short-term expectations.
In sum, we see that the short-term inflation expectations were very volatile following the recession and that, recently, median expectations have hovered around 2 percent and 2.5 percent (respectively, for professional forecasters and consumers). However, there was substantial disagreement about future inflation expectations early in the recession, reflecting opposite concerns among survey participants: some fear deflation or disinflation and others fear higher inflation. The longer-term inflation expectations are currently around their historical levels, and the dispersion for these expectations also reflects a better anchoring of inflation rates over the longer term.