Wednesday, May 13, 2009


How Much Inflation Is the Market Forecasting?

The strongest argument against those of us who are warning of potentially massive price inflation is that the bond market is apparently disagreeing with us. For example, yields on 5-year Treasurys (as of this writing) are about as low as they have been in the series' history at FRED, which certainly doesn't seem consistent with a decent probability of, say, 10%+ price inflation anytime soon.

Now the worry-warts among us can offer all sorts of explanations for this. The Fed is printing up new money (partly) to purchase Treasurys, and so it's not surprising that the yields might be temporarily held down. It's also possible that the markets are in such disarray that the (virtually) riskless real return on US Treasurys has fallen into negative territory, such that even an expected annual price inflation of 5% over the next few years is showing up as a very low nominal yield.

In theory, the cleanest way to back out the market's actual inflation forecast is to contrast the TIPS yield with nominal yields. The Treasury Inflation Protected Security (TIPS) offers the investor protection against price inflation by adjusting the nominal face value of the bond according to changes in the CPI. So the yield on a TIPS bond (in theory) compensates the investor for the pure deferment of purchasing power, stripping away the inflation risk.

To do it correctly, you would want to correct for the differences in liquidity between standard and TIPS markets, etc., but I think most people use a simple rule of thumb. They take the nominal Treasury yield and subtract the TIPS yield (for the same maturity) and call the result the "market's forecast" of annualized inflation over the life of the bonds.

In April 2009, the difference between the average monthly 5-year Treasury yield and the 5-year TIPS yield was a mere 0.75%, leading people like Scott Sumner to confidently say that inflation is not a threat. In other words, the implication is that "the market" is only forecasting average annual 0.75% CPI increases over the next five years.

But is that right? Unfortunately the TIPS contracts were only introduced in 2003, so we don't have that much time to contrast implied forecasts with actual results. But in the table below, I give the limited data points that we have:

If you ask 85 economists to interpret the above table, you'll probably get 37 different answers. My own guess is that (for whatever reason) the "TIPS forecast" systematically understates expected inflation rates. I am not saying bond traders were dumb in the first half of 2003; rather I am saying that there are probably reasons that the two numbers won't actually be equal, even if traders had perfect foresight.

When I look at that table, I conclude that the "normal" markup should be about 150 basis points from the spread, to get the true market forecast. I think the reason the "underforecast" fell so much was not that the market got smarter in early 2004, but rather that traders failed to anticipate the outright price deflation in late 2008, which ate into the standard markup.

Thus, I think the current spread of 0.75% means that the market "forecast" of 5-year average inflation is above 2%. That's still a lot lower than what I personally am predicting, but my point in this post is that people citing the information from TIPS never justify that tool. With the only results we have available (unless they've done TIPS-type contracts in other countries for longer periods?), it looks like the nominal-TIPS spread isn't a very reliable indicator of inflation at all. Not only is it systematically different from 0, but it swings wildly.

Very cool analysis.
Notice the analysis of Frank Shostak, who argues that excess monetary liquidity has been and will continue to suppress treasury yields.
Post a Comment

Subscribe to Post Comments [Atom]

<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]