Tuesday, March 05, 2013

A bit more on lagged inflation and TIPS Breakevens

I'm afraid that exploring the high correlation of inflation over the past year and the breakeven inflation rate which would make the return on Treasury Inflation Protected Securities (TIPS) equal to the nominal treasury rate threatens to move from hobby pass time to work.  It is very easy and even fun to make some graphs using FRED and show a surprisingly very high correlation between inflation in the past year and the 5 year constant maturity breakeven.

This is OK for a blog post, but won't do for anything more for several reasons.  First, for some reason, the FRED series only go back to 2003 for some reason (even though TIPS were first auctioned in January 1997).  Second the matched series of real and nominal interest rates are not simple calculations from the prices of specific assets but "constant maturity" series based on estimation of a yield curve and interpolation.  I don't know exactly how this is done and I should find out.

This little post (here because it is not up to www.angrybearblog.com standards) is a pathetic attempt to deal with the first problem by Googling.   I was looking for older data on TIPS breakevens.  Pu Shen at the Kansas City Fed wrote a working paper on the topic (warning pdf).  Shen's point is that the breakeven inflation rate was low and variable and interprets the average excess returns on TIPS as compensation for their low liquidity.  A constant difference would not be a problem, but the TIPS breakeven varied a lot while median Blue Chip survey expected inflation and actual achieved iflation changed very little.

Anyway I stole the graph (I don't have the numbers I just took a screen shot).  The part from 2003 through 2006 is familiar from the FRED based post to which I link above.  For comparison have a corresponding FRED graph of the average of CPI inflation and CPI minus food and energy inflation over the past year.  The correlation in the period 1999-2003 was impressive if not perfect.

The correlation isn't perfect by any means, but the breakeven series looks much more like the past year inflation series than like the survey forecast of inflation to put it very mildly.

1 comment:

Ken Houghton said...

Constant Maturity adjusts for (among other things) declining liquidity.

Quarterly issuance is Feb, May, Aug, Nov. Monthly is monthly. Weekly (bills) is weekly.

But people pay more for new securities (higher price, lower yield, more liquidity). So if I buy the 5-year note issued in February, I lose liquidity (demand; price) as newer five-years are issued. Also, my security is no longer a five-year security. After the first coupon payment, it's a 4.5 year security. Two years in, you could re-open that "five-year" security to sell as the "new" three-year.

Shorter the above: you can't stick with the same Treasury and call its yield "the five-year rate."

The "answer" is to roll the "basket" of securities looked at so you have a Constant Maturity Treasury. (You would expect this of the people who came up with "Cheapest to Deliver.")