Tuesday, August 29, 2006

Unbiased but wrong...

I'm picking up on this debate a bit late, so I apologize in advance.

On Thursday, the Wall Street Journal ran this op-ed piece by Arthur Laffer. In it he claimed:

"Fortunately, the Treasury has over the recent past begun to issue an instrument called the Treasury Inflation-Protected Security (TIPS) in a wide range of maturities including a 10-year note. The yield on this TIPS note is for all practical purposes the expected real yield over the horizon of the maturity of the note."

To which Bret Swanson replied:

"Dr. Laffer says expected inflation gleaned from TIPS bonds is the best predictor of inflation, but in fact TIPS have not been very good at all at predicting inflation."

Other blogs are writing about this debate, including a couple of my favorites: Capital Spectator, and Macroblog (here and here).

You should read all these pieces, because they both make various points, all of which are interesting to econgeeks like myself. Anyway, I'll weigh in on the debate as to whether TIPS can be used as an estimation of the markets inflation prediction or whether the TIPS spread rarely predicts CPI correctly. They are both right.

As I've written before (here for example), the forward yield curve is an unbiased predictor of future rates. Because the rates market is impacted by innumerable variables, no predictor could possibly be particularly accurate. In fact, efficient markets theory tells us that as soon as a predictor became known, it would become part of the current price and therefore cease to be an effective predictor. That being said, the slope of the yield curve tells us something about where the market thinks rates (and the economy) is headed.

TIPS spread is similar. The wider the spread, the more inflation the market expects and vice versa. But inflation is a complex thing, and there are many possible paths inflation might take. To take an example from statistics, the mean of a sample is an unbiased predictor of the mean of the population. But odds of the mean of the sample being exactly the mean of the population? Close to zero.

Take another example. Every year, Las Vegas puts out odds on who will win the Super Bowl. Every year there is a favorite. Without having any data on the subject, I'll wager that the pre-season favorite wins the Super Bowl less than 10% of the time. Why? Because any random injury, or wind-blown kick, or tipped interception can change the course of a game and therefore a whole season. Not to mention how athletes will progress or age is extremely difficult to predict. Does that mean that the Vegas odds are meaningless numbers? Of course not. The preseason favorite probably does have the best chance of winning, its just that there are so many possibilities even the most likely scenario is still fairly unlikely.

So the TIPS spread is a valuable indicator of what the market expects for future inflation. Intermediate-termed TIPS spreads are a particularly good indicator of Fed credibility. So long as 5-10 year TIPS spreads are hovering in the 2-3% range, the market is telling you that it expects inflation to generally remain under control.

If anything, the TIPS spread overstates the market's expectations for future inflation. This is because TIPS are like insurance against inflation. For any investor where inflation is part of their liability stream, like a pension fund with a COLA increase in their benefit payments, TIPS are a way of controlling that liability. Therefore, the TIPS market probably has a premium priced into it, particularly in long-dated issues.

So will we have an inflation spike? Maybe. But the market is certainly not betting on it.

Tags:
, , ,

No comments:

Post a Comment

Comment rules:
All comments must contribute to the conversation
All comments should be civil
No comment should include any personal attacks, however minor, on the author or other commenter.
Do not hawk your own website unless its a specific reference to the article
If you post anonymously, please give some identifyer
I will delete any comment which doesn't fit this criterea