QE 2 and breakevens

alternate title QE 2.MMMCCCLLXXIX as I am a bit obsessed. One of the many bees in my bonnet is the argument over whether the Fed’s QE 2 policy hinted, suggested, teased and announced August 10 2010 through November 3 2010 had much of an effect. My impression remains that it had remarkably little effect on anything for a roughly trillion dollar open market shift. My view is that the problem was the Fed traded money for the most money like asset which wasn’t trading as a perfect substitute for money (7 year Treasury notes — the shortest maturity paying an interest yield noticeably above 0). This would be a way to minimize the effect for a given quantity.

This post is supposed to be about evidence. The thing which struck me long ago is that the price of 7 year notes went down in the period from the first hingt of QE2 through the first actual purchases. This suggests to me that 7 year notes were treated as close substitutes for cash with a relative price not exactly their face value (perfect substitutes don’t have to have equal prices). The reply was that the differential between nominal treasuries and TIPS (bonds indexed to the CPI) went up showing an increase in expected inflation which is the key. To be brief, this pattern corresponds to QE2 working as a signal of future monetary policy and not directly by changing the amount of 7 year notes private agents hold in their portfolios.

The story, as I understand it, is that the key issue is whether the Fed will keep the federal funds rate near zero when unemployment gets down near the non accelerating inflation rate (NAIRU). People with this view tend to argue that the Fed could have convinced people that it would keep rates miniscule causing increased inflation, but chose not to. They tend to argue that QE2 was not very effective, because the Fed was careful to not signal any intention of allowing inflation over 2%. This post does not address that argument, which would be about why QE2 was ineffective. Rather it addresses the straw man who claims QE2 was effective because it shifted expectations of policy in the misty future with non horrible unemployment rates.

OK finally a figure showing breakevens: ordinary nominal treasury yields minus TIPS yields. I’m supposed to warn that it isn’t really investors’ inflation forecast as bond prices depend on other things including liquidity. I have a problem with FRED (not with all available data but with easily available data). FRED doesn’t show yields on specific nominal Treasuries, because there are just too many of them, but does show yields on specific TIPS of which there are quite few. This means that the maturity of the real and npominal interest rate series will not exactly match. I do what I can about this sticking to FRED.

Sorry for the pointless legend. The green line shows the data stressed by say Brad DeLong. It is a 5 year breakeven inflation rate showing the difference between the 5 year constant maturity normal (nominal) Treasury yield and the 5 year constant maturity TIPS yield. The blue line shows the difference between the yield on TIPS maturing July 15 2013 and the 2 year constant maturity (nominal) yield. The red line shows the yield on the TIPS maturing July 15 2012 minus the 1 year constant maturity nominal interest rate. The black line shows the yield on the TIPS maturing July 15 2012 minus the 2 (two) year constant maturity nominal interest rate. The breakeven compared to regular nominal Treasuries maturing July 15 2012 should be somewhere in between.

Well that was a lot of typing for one figure. But the pattern is clear. The increase in the breakeven inflation rate is higher for shorter maturities than for longer maturities. This is the opposite of the pattern strawman would predict. Strawman argues that QE2 caused increased expected inflation in the fairly distant future and this affects inflation and expected inflation in the nearer future through a forward looking Phillips curve. The change in expected inflation should be greater at longer maturities. My guess is that the increase in expected inflation is a coincidence. This post is too long already so more (in progress) after the jump.

I know you guessed, but my belief is that the coincidence is that the price of petroleum increased around the time QE2 was gradually announced. Importantly TIPS are indexed to the CPI not the CPI minus food and energy. Here the black line is, as before, the 2 year constant maturity nominal interest rate minus the yield on the TIPS maturing July 15 2012. The blue line is the monthly percent increase in the price of a barrel of West Texas intermediate crude oil divided by 10 to fit it on the same graph.

Notice that they track although the oil price changes (even when divided by 10) are noisy. They are monthly averages, because that is the easy way to get percent change from a month ago. This isn’t the way things should be, the oil price changes are backward looking and very short term. There is no economic logic behind dividing by 10 (0.1 is larger than the share of petroleum as an input to production of goods and services in the CPI basket). Of course there are very few numbers graphed. This graph may convince no one but me, but it convinces me.

Finally, to discuss an approach which doesn’t convince me at all, some have tried event studies. Here the assumption is that markets are efficient so the full effect of an event on asset prices the moment the event becomes known to the public. In the case of QE2 this is tricky as news dribbled out. For one thing I distinguish between QE2.1 the declaration that the Fed would use proceeds from maturing mortgage backed securities to buy treasuries and not to retire money. This amounted to an estimated (guessed) change in the planned supply of high powered money by about $400 billion and was announced August 10. QE2 is generally used to refer to QE2.2 the purchase of an additional $600 billion in 7 year Treasury Notes with newly issued high powered money. The first public hint of this program was chairman Bernanke’s August 27 speech at Jackson Hole Wyoming. Note that it was just mentioned as a possibility. Announcements that it was actually happening and that the amount would be $600 billion came later on September 21 2010 and November 3 2010.

Michael Bauer at the San Francisco Fed looks at the sum of changes of interest rates on three dates August 10 plus September 21 plus November 3. Oddly he doesn’t look at the relatively famous date of August 27 2010. The sum is – 14 basis points (-0.14 %) for the 10 year constant maturity treasury rate, -13 basis points for BBB corporate bonds and -14 basis points for 30 year MBS. These are very small numbers. The standard deviation of daily changes of the 10 year constant maturity treasury rate was over 6.6 basis points in 2010. The standard deviation of the sum of three changes is root three times that. A change of -14 basis points does not reject the null that nothing special happened those days at standard confidence intervals (the t-statistic with 3 degrees of freedom is about -1.2). Also the 10 year constant maturity rate increased on August 27 2010.

I don’t believe in event studies, but I note that there is almost no evidence of any effect of QE2 in the FRBSF Economic Letter.