Has expected inflation really risen?

Several economics and finance bloggers, such as Greg Mankiw and Felix Salmon, have been pointing to an apparent increase in long-term inflation expectations in the TIPS market that Greg Ip wrote about Thursday in the Wall Street Journal’s economics blog. It would all be very worrisome to me, except that when I look at the actual data, they don’t seem consistent with a reasonable story about rising inflation expectations.

Greg Ip gives a clear description of the indicator being used and why it is used:

The Fed has long looked at the difference between yields on nominal Treasurys and inflation-protected Treasurys (TIPS) for a sense of what investors expect inflation to be. The difference is the so-called “breakeven” rate — the inflation rate that equates returns on the two. The Fed also tries to strip out near-term inflation disturbances related to fluctuating energy and food prices by looking at what the market expects inflation to be starting five years from now and running for the next five years (i.e. from 2013 to 2108). This is the “five-year, five-year forward” breakeven rate.

I’m not sure exactly how the Fed, or Barclays Capital (the source for Greg Ip’s numbers), does the calculation, but I’m going to use what I think is a reasonable approximation. (I emphaize that it is an approximation, most likely not the calculation that the Fed does, but it should give a reasonable idea of what the general picture looks like, and it has the advantage that I can do it in my head just by looking at four easily available yields.) Here are the four relevant pieces of data for my approximation, taken from the Fed’s constant maturity data for two points in time (the day before Ben Bernanke’s Jan. 10 speech, and the day before Greg Ip’s Jan. 31 blog post):

                                Jan 9   Jan 30
5-year TIPS yield               0.94%    0.84%
5-year nominal treasury yield   3.10%    2.96%
10-year TIPS yield              1.56%    1.45%
10-year nominal treasury yield  3.82%    3.78%

Using brute force subtraction, these data imply the following approximate current breakeven inflation (BEI) rates:

                                Jan 9   Jan 30
Current 5-year BEI              2.16%    2.12%
Current 10-year BEI             2.26%    2.33%

The approximate “five-year, five-year forward” breakeven inflation rate is two times the 10-year BEI minus the 5-year BEI, as follows:

                                Jan 9   Jan 30
5-year, 5-year forward BEI      2.36%    2.54%

This change is similar to what Barclays found, so I trust that my approximation isn’t too far off. But it’s important to think about the current BEI rates and what they mean. The first thing you should notice is that the current 5-year BEI rates are below the 5-year forward BEI rates. That should make you a little suspicious already. Think about those “near-term inflation disturbances related to fluctuating energy and food prices” that the Fed is trying to filter out. There has been a huge run-up in commodity prices over the past 6 months, and over the past 5 years. Presumably this should have more effect on the inflation rate over the next 5 years than it will on the inflation rate over the subsequent 5 years. If these BEI rates were unbiased indicators of expected inflation, you would probably expect the current BEI to be higher than the 5-year forward BEI. Not that anyone really thinks they are unbiased indicators, but this observation underscores the point that risk premiums and liquidity premiums are more important than inflation expectations when comparing the behavior of different 5-year and 10-year securities.

Take a look specifically at the change between Jan. 9 and Jan. 30. The current 5-year BEI rate actually went down by 4 basis points. That observation, it seems to me, is rather hard to reconcile with a story that says investors were reacting to a dovish shift in Fed policy. Is there any way that a dovish shift could reduce the inflation rate over the next 5 years? Of course, the (inflationary) dovish shift might have been outweighed by (disinflationary) weak economic news. But in that case should we really be worrying about inflation expectations? In particular Greg Mankiw might need to reconsider his conclusion:

A rise in expected inflation is not consistent with the conventional wisdom that the economy is on the verge of a serious slump driven by inadequate aggregate demand. It is, however, consistent with the hypothesis that policymakers are overreacting to some bad economic news with excessive monetary and fiscal stimulus.

These data suggest to me that the chance of a serious slump has risen and that the monetary and fiscal stimulus has not caught up with that rising probability.

It’s still possible that the market’s perception of the Fed’s preferences has shifted in a dovish direction, and that would be one way to explain the behavior of the 10-year BEI. But in that case the market must also think the Fed will be less able to implement those dovish preferences in the immediate future.

An alternative explanation, which doesn’t require such a subtle analysis of the Fed’s preferences and abilities, is that inflation expectations in the near future have fallen (as the weak economic news would suggest) but that the liquidity premium on TIPS has also fallen (as have other liquidity preference indicators, such as the TED spread). If the same liquidity premium applies to 5-year and 10-year TIPS, then a drop in that liquidity premium, without any change in expected inflation rate, would result in higher current BEI rates at both the 5-year and 10-year horizon. That shift by itself would increase the 5-year, 5-year forward BEI rate. If, in addition to the drop in the liquidity premium, the current 5-year expected inflation rate fell but the 5-year forward expected inflation rate remained the same, that would result in exactly the pattern that we observe.