Thursday, November 20, 2014

How well do TIPS predict inflation?

Yesterday's post focused on the market's current inflation expectations for the next 5 and 10 years, as embedded in the pricing of TIPS and Treasuries. It concluded that inflation expectations are not unreasonable, and they are in line with past inflation and what is reasonable to expect going forward. This post focuses on how well the bond market's inflation-forecasting skills have proven to be.

Conclusion: the bond market has more often than not tended to underestimate future inflation. For the first 12 years of the existence of TIPS, 5-yr breakeven inflation rates were lower than inflation turned out to be about two thirds of the time, and there have been some sizable errors at times.

Introduction for neophytes: "TIPS" is an acronym for Treasury Inflation Protected Securities. TIPS were introduced in early 1997. They are a unique type of bond, because they pay a fixed, "real" rate of interest plus a rate that is equal to whatever inflation happens to be. The effective interest rate a TIPS investor receives thus has a real component (known in advance) and an inflation component, and both are guaranteed by the U.S. government. In practice, TIPS work like this (in simplified fashion): they are issued with a par value of $100, and that value is subsequently increased by the rate of consumer price inflation. They are also issued with a fixed coupon that is paid on the inflation-adjusted principal. Say you buy some 5-yr TIPS on January 1st with a fixed coupon of 2% and a par value of $100. At the end of the year it turns out that the CPI has increased by 3%. You would then end up with a total return of a little over 5%: the principal value of your TIPS would be 102 at the end of the year, and you would be paid a coupon of 3% on that value: 1.02 * 1.03 = 1.0506, or 5.06%. The market price (not the par value) of TIPS once they are issued varies inversely with their real yield: if real yields increase, the market price declines, and vice versa.

If you compare 5-yr TIPS to 5-yr Treasuries, you can easily determine the bond market's expected, or "breakeven" inflation rate. (Breakeven is the technical term, since that is the inflation rate that would make the return on holding 5-yr TIPS equal to the return on holding 5-yr Treasuries to maturity.) You simply subtract the real yield on 5-yr TIPS from the nominal yield on 5-yr Treasuries.

The chart above compares the trailing 5-yr annualized rate of CPI inflation (red line) with the 5-yr breakeven inflation rate (i.e., the expected annualized rate of inflation over the subsequent 5 years) as predicted by 5-yr TIPS and Treasuries (blue line). Think of the blue line today as the market's inflation forecast made 5 years ago.

For example: 5-yr TIPS first started trading in July 1997, when the breakeven or expected inflation rate for the next 5 years was 2.3%. The starting value for the blue line is thus 2.3% in July 2002, five years after their first issuance. In their first month of trading, the 5-yr breakeven inflation rate embedded in TIPS prices turned out to be exactly right, since the CPI rose at an annualized rate of 2.3% for the 5 years ending July 2002.

As you can see from the chart, the blue line was almost always lower than the red line from 2002 through 2009. That means that the breakeven or expected inflation rate embedded in the pricing of TIPS and Treasuries was almost always lower than actual inflation turned out to be, and by a considerable amount, during the 7-yr period from 1997 through 2004. But from late 2004 through late 2007 (2009 through 2012 on the chart), the expected inflation rate tended to be somewhat higher than actual inflation, and the errors were of much smaller magnitude.

Since mid-2008 (mid-2013 on the chart), the bond market's expected rate of inflation proved again to be much higher than actual rate of inflation. In general terms, TIPS have under-estimated inflation for 8 of their first 12 years of trading, and over-estimated inflation for four of the first 12 years.

I note also that there have been times in the past where Treasuries and TIPS have egregiously under-estimated inflation. In late 1998, for example, (late 2003 on the chart) the expected rate of inflation over the next 5 years was about 0.6%, but actual inflation turned out to be about 2.5%. In late 2001, the expected rate of inflation was about 0.8%, but actual inflation turned out to be about 2.6%. In late 2009, the expected rate of inflation was about -0.4%, but actual inflation over the subsequent 5 years turned out to be 2%.

I think there's a reasonable explanation for why the bond market has at times underestimated future inflation by a considerable amount. 1998, 2001, and 2009 were all times of great distress in the markets: the Russian default and LT Capital debacle in 1998, the recession of 2001, and the financial crisis of 2009. During times of great uncertainty, Treasury prices have been bid up to unrealistically high levels due to the market's craving for certainty; this pushed Treasury yields down relative to TIPS yields.

Today, 5-yr TIPS and Treasuries are predicting that inflation over the next 5 years is going to be about 1.6% per year (the last datapoint for the blue line). Since 1966, inflation over any rolling 5-yr period has only been that low or lower for seven months—from April 2013 through October 2013—and that period was dominated by the negative rates of inflation we saw in 2009.

In short, the U.S. bond market currently is expecting inflation to be historically low for the next five years. Since TIPS were first issued in 1997, the bond market has had a strong tendency to underestimate future inflation, especially during periods of rising inflation and during periods of financial market stress. Although the current expected rate of inflation is consistent with what we have seen in recent years and consistent with the significant decline in energy prices over the past six months, the market is arguably still afflicted with a substantial amount of risk aversion, as I've noted repeatedly. And it would appear that the bond market does not see hardly any chance that the Fed commits an inflationary error as it unwinds QE and moves away from zero-bound interest rates.

The bond market could be right about the future of inflation, but it's making some pretty heroic assumptions, and past experience suggests it's probably underestimating future inflation today.


Benjamin Cole said...

Excellent blogging.

Well, maybe the TIPS market is underestimating inflation.

But, we see a Federal Reserve that is consistently undershooting its 2% PCE inflation target with out any concern.

Today a Fed dove is someone that wants the Federal Reserve to hit its 2% target which is obviously become a ceiling.

Maybe the TIPS market is right.

The price for this microscopic rate of inflation will be very slow growth. said...

The Federal Reserve Bank of Cleveland reports that its latest estimate of 10-year expected inflation is 1.83 percent. In other words, the public currently expects the inflation rate to be less than 2 percent on average over the next decade.

Scott Grannis said...

sgt: That's not surprising, since it is exactly what 10-yr TIPS and 10-yr Treasuries are saying; see the second chart in this post.

NormanB said...

Thanks, This is great.