Bond markets are shrugging off inflation fears, but what do they know that we don’t?
As the debate continues over the potential inflationary impact of President Joseph R. Biden Jr.’s $1.9 trillion “American Rescue Plan,” one key indicator suggests that markets are no more worried about excessive inflation than the Federal Reserve is. “Inflation compensation” in US 10-year Treasury securities has jumped 0.25 percentage point since January 14, when Biden first proposed the package. (Inflation compensation, also known as “breakeven inflation,” refers to the yield on a conventional bond minus the yield on an inflation-indexed bond of the same maturity.) The current value of inflation compensation, however, is fully consistent with the Federal Open Market Committee's goal for inflation of 2 percent, after having been somewhat below that level last year.
So, despite all the hand-wringing in some quarters, the bond market is apparently not worried about excessive inflation. On the other hand, to give the hand-wringers their due, history suggests we should not take much comfort from the bond market, because it has not managed to predict any notable change in inflation since World War II in the major advanced economies.
A previous post showed that bond yields in the United States, United Kingdom, Japan, and France since the 1950s have not been good predictors of future inflation. An important qualification of those results is that they are not able to distinguish between movements in bond yields owing to expected future inflation and movements owing to the expected real rate of return.
The development of inflation-indexed bond markets in advanced economies since the 1980s creates a security whose yield is fixed in real, or inflation-adjusted, terms. Inflation compensation can thus be measured directly as the (nominal) yield on a conventional bond minus the (real) yield on an inflation-indexed bond of the same maturity. The main limitation of inflation compensation as a guide to the predictive power of bond markets for inflation is that it is not available during the key decades of the 1960s and 1970s, when inflation rose dramatically, or the early 1980s, when inflation began to fall.
The figure below shows 10-year inflation compensation from government bond yields for the United States, the United Kingdom, and Australia, the advanced economies with the longest records of inflation compensation. In all three countries, inflation compensation is less volatile than actual inflation. There is no apparent tendency for inflation compensation to move in advance of major moves in inflation. Instead, movements tend to be simultaneous. In particular, the sustained declines in inflation in the United Kingdom and Australia in the early 1990s were not predicted in advance by inflation compensation.
The table below presents regressions of 10-year inflation compensation on the current 12-month consumer price index (CPI) inflation rate, an average inflation rate over some period in the future or the more distant past, and the current 12-month change in Brent oil prices in local currency. The 10-year backward inflation average is the average rate of inflation over the 10 years before the yield is observed, and the 10-year forward inflation average is the average rate of inflation over the following 10 years. The 10-year forward inflation average is what 10-year inflation compensation is supposed to predict.
In all three countries, backward averages of inflation provide better models of inflation compensation than forward averages, with uniformly higher R2s. A one percentage point increase in the 10-year average of past inflation raises inflation compensation 0.26 percentage point in the United States, 0.52 percentage point in the United Kingdom, and 0.60 percentage point in Australia. When a backward average of inflation is included in the regression, the current inflation rate has only a small effect in the United States and the United Kingdom and a moderate effect in Australia.
Inflation compensation has essentially no correlation with future inflation in the United States and only a modest correlation with future inflation in Australia, limited to the 1-year and 2-year horizon. Inflation compensation is highly correlated with long-term future inflation in the United Kingdom, as shown by the moderately high coefficients on 5-year and 10-year forward averages of inflation. Nevertheless, even in the United Kingdom, inflation compensation is better explained by a long backward average of inflation than any measure of forward inflation as evidenced by the R2 statistics.
Overall, the evidence that 10-year inflation compensation provides extra predictive power (on top of that of current inflation) for future inflation is nonexistent in the United States and very weak in Australia. Even in the United Kingdom, where regressions suggest that inflation compensation has some extra predictive power for inflation, inflation compensation completely failed to predict the one large and sustained change in inflation (in the early 1990s) that has been observed since inflation compensation data became available.
|Regressions of 10-year inflation compensation on inflation and change in Brent crude oil prices|
January 1999–February 2021
January 1985–January 2021
Q4 1985–Q4 2020
|Measure of average inflation||Average inflation||Current inflation||R2||Average inflation||Current inflation||R2||Average inflation||Current inflation||R2|
|n.a. = not applicable|
|Note: This table presents results from monthly regressions (quarterly regressions for Australia) of 10-year inflation compensation on the current 12-month inflation rate (which is also the 1-year backward average), an average inflation rate over the indicated period in the past or future, and the current 12-month change in Brent oil prices in local currency. Regressions for Australia include a dummy that takes a value of 1 in Q3 2000–Q2 2001 to control for the introduction of the Goods and Services Tax in July 2000. Inflation averages are compound annual growth rates of respective price indexes over periods ranging from 12 to 120 months (4 to 40 quarters in the case of Australia). Consumer price indexes are used for Australia and the United States, and the retail price index is used for the United Kingdom. Inflation compensation is expressed in terms of zero-coupon yields for the United States (for more details, see this webpage). Data ranges displayed refer to availability of the inflation compensation series. Regressions with forward averages of inflation have shorter sample periods based on the leads needed to calculate future inflation averages.|
|Sources: Authors’ calculations based on data from the Australian Bureau of Statistics, Bank of England, Federal Reserve Board, International Monetary Fund, Reserve Bank of Australia, UK Office for National Statistics, and US Bureau of Labor Statistics via Macrobond.|
1. 10-year inflation compensation is 2.33 percent in terms of the consumer price index (CPI). The Fed aims for inflation of 2 percent in terms of the personal consumption expenditures price index, a measure of inflation that tends to run a bit lower than CPI inflation. Data are from Bloomberg.
2. An inflation-indexed bond, such as a Treasury Inflation Protected Security (TIPS), pays a yield that is fixed in real terms because the principal value used to calculate coupon and redemption payments grows by the rate of inflation each year.
3. The current 12-month inflation rate is effectively a 1-year backward average of inflation. In the United Kingdom, retail price index inflation is used because inflation-indexed bonds are linked to that measure of prices. For Australia, the regression includes a dummy variable to control for the imposition of the Goods and Services Tax in July 2000. The coefficients on Brent oil price changes are statistically significant, but the results for the predictive power of future inflation are not sensitive to dropping oil prices.
4. Regressions of future inflation averages on past inflation and past inflation compensation are consistent with the results in the table, with no (or even negative) predictive power of inflation compensation in the United States, a moderate amount for 1-year and 2-year forward inflation in Australia, and significant predictive power for up to 5-year inflation in the United Kingdom (but compensation had a negative coefficient with 10-year inflation). These results are based on Newey-West standard errors with lags commensurate with the time period over which the dependent variable is constructed.