Regular readers of this blog know that I track the break-even TIPS spread to follow changes in inflation expectations. Doing so also provides an implicit (and imperfect) estimate of changes in the real interest rate. (For an explanation of why the break-even TIPS spread is an imperfect estimate of inflation expectations and the real interest rate, see the Cleveland Federal Reserve Bank website.) Since early in May, the data show a fairly striking anomaly in real interest rates: real interest rates over a 5-year time horizon have been rising (though still negative) while real interest rates over a 10-year horizon have been falling.
Why is this anomalous? Because real interest rates at the 5-year and 10-year time horizons are generally closely correlated. The chart below shows fluctuations in real interest rates at constant 5- and 10-year maturities since the beginning of 2012. The two lines track each other closely until the beginning of May when the 5-year real interest rate begins to rise while the 10-year real interest rate continues to fall. The coefficient between the 5-year and 10-year real interest rates from January 3 to May 24 is slightly over .8. From January 3 to May 3, the correlation coefficient is almost .86; the correlation coefficient since May 3 is -.72.
I have no explanation for this anomaly. Anybody out there like to take a crack at it?
UPDATE: It just occurred to me that the increase in short term real rates is reflecting a liquidity premium associated with an increasing perceived likelihood of a financial crisis associated with a breakdown of the euro. Not a very happy thought as I prepare to call it a night.
It could be that the federal government is changing its duration portfolio.
Interest rates on government bonds have a supply / demand aspect that is not caught in your “inflation expectations” model. Obviously, an increase in short term issuance will tend to push up short term interest rates while a decrease in long term issuance will tend to push long term interest rates down – all else being equal.
LikeLike
I also recall that the TIPS spread has a risk term built into it when you work out the exact difference. Relative risk over these time horizons could be changing due to god-knows-what expectations about the availability of Euro debt or whatever, although it’s a bit late at night for me to check maths and dates.
LikeLike
David,
I’d need to see the underlying data being used for calculating the inflation expectations. In absence of data, the two usual suspects would be:
1) A zero bound problem. If rates are expected to be near zero for five years, but not ten, falling inflation expectations manifest as rising real interest rates for the next five years.
2) Oil prices. If prices are expected to fall sharply in the near-term due to economic contagion effects, near-term inflations expectations could fall sharply while longer-term inflation expectations remain unchanged.
LikeLike
A third possibility would be the market is anticipating a premature Fed tightening. But I’d lean strongly toward (1) above.
With 5-year nominals already below 1%, are real interest rates at rising toward -1%, I’d guess it’s a zero bound problem. What would Keynes have to say about liquidity preference?
LikeLike
make that: *and* real interest rates rising
Yes, I’m thinking it’s definitely a zero bound falling inflation expectations liquidity preference problem. My fingers are faster than my brain.
LikeLike
I wouldn’t know the answer. Reading some of the comments the zero bound answer sounds plausible to me.
LikeLike
FWIW nominal rates have been doing the same thing – just a bull flattening.
LikeLike
It could be, as you say, due to liquidity concerns. Investors might be fleeing from illiquid 5-year TIPS to more liquid 10-year TIPS and thus 5-year TIPS yields have had to rise so as to compensate investors with an adequate return.
But are TIPS at the 5-year end of the spectrum actually less liquid than those at the 10-year? This depends on the relative size of the various TIPS issues, the depth of each market, and the dispersion of holdings. I’m not so sure that near TIPS are less liquid than more distant ones.
If not, then people could be moving from 5-year to 10-year TIPS because they expect short term deflation. The embedded option in TIPS loses value when inflation is expected to be negative, although for longer terms the option will still have plenty of time value remaining.
LikeLike
David: I think your liquidity premium/Euro crisis hypothesis makes sense. Here’s on small bit of supporting data: since early May the Canadian dollar has been falling relative to the USD: http://ca.finance.yahoo.com/echarts?s=CADUSD%3DX#symbol=;range=3m;compare=;indicator=volume;charttype=area;crosshair=on;ohlcvalues=0;logscale=off;source=undefined;
This (a falling CAD/USD exchange rate) normally seems to happen whenever there’s increased fear of liquidity problems. My reasoning is that the US dollar is “the moniest of all monies”, because it is the medium of exchange for all other media of exchange. (If you want to swap Canadian for Australian dollars, for example, you have to first swap Canadian for US dollars, then US dollars for Australian.)
Must get my mind back in gear, and respond to your Yeager/Tobin post!
LikeLike
Frank, Is there any evidence of a change in the duration of the government’s duration portfolio? Could the portfolio change enough in the course of 3 weeks to produce a substantial change in yields? I am more inclined to look at the demand side to explain short term fluctuations and the supply side to explain long-term trends.
Ben, The risk component is what the Cleveland Fed estimates try to isolate. When their estimates of real interest rates and inflation expectations come out next month for May, we will be able to observe whether the riskiness of inflation expectations has also increased as your comment suggests.
Steve, Real interest rates at 5-year and 10-year maturities have been negative for some time, so I don’t see the relevance of the zero lower bound issue. As for the effect of lower oil prices, my question is why those aren’t showing up in more rapidly falling inflation expectations in the shorter maturities rather than rising real rates. I think your point about an expected Fed tightening is somewhat along the lines of my conjecture in the UPDATE I added. The difference is that the tightening I am envisaging is not a deliberate policy, but a failure to respond to the deflationary effect (as manifested for example in the recent appreciation of the dollar relative to the euro) of the Eurozone crisis.
Tas, Again, I don’t get the zero-bound argument in this context.
Tubulus, You’re right, the nominal term structure is showing a similar divergence, but it is less pronounced than the real term structure.
JP, It may not be so much that they are fleeing TIPS because of liquidity concerns, but that they are expecting a liquidity effect to cause an increase in inflation-adjusted yields on shorter durations.
Nick, Yes, the timing fits. The shift to US dollars from Canadian dollars starts at the end of April. The euro/dollar exchange rate was at about $1.323 on April 30 and May 1 close to the euro highs for 2012. On May 2, the euro fell to $1.315, and the euro has fallen steadily for the rest of the month hitting a two-year low yesterday just over $1.25.
I look forward to seeing your comments on Yeager and Tobin.
LikeLike
David, your use of the term “liquidity premium” confuses me. I use it as Keynes used it… a liquidity premium is a “convenience yield” specific to a given asset and makes up part of that asset’s return. The premium is derived from that asset’s ability to be sold quickly in secondary markets.
You seem to be using the term to refer to a general liquidity event. Given your definition, I don’t disagree with you. The only thing that could push down 5 year TIPS prices but increase 10 year TIPS is some sort of near term flight to liquidity. Same with Nick’s C-dollar falling. Same with stock markets falling. Same with oil falling. Same with copper falling.
To really see this effect, here is a 5yr TIPS that is close to maturity, due April 15, 2013. The rise in its yield is especially pronounced:
http://research.stlouisfed.org/fred2/series/DTP5A13?cid=82
Here is the July 15, 2012 10 year TIPS:
http://research.stlouisfed.org/fred2/series/DTP10L12?cid=82
LikeLike
JP, You are right that I was not following Keynes in my usage of “liquidity premium, but you seem to have figured out what I meant, nevertheless. Can you recall if there was any specific event or news on or about May 3 that might have triggered the expectation of a crisis?
LikeLike
Maybe the Greek elections on Sunday, May 6? Dunno. I think France had its elections around then too.
LikeLike
David,
First thing I would check is whether you are looking at an “on the run” 5yr TIPS — the 4/15/17 maturity was recently opened, and this should be the most liquid 5yr. An “off the run” bond (such as the short-maturity ones JP Koning links to) can be much less liquid.
If the 5yr you are using is on the run, then I would wonder why a shorter term maturity would have a more financial crisis-sensitive liquidity premium. Theory predicts the opposite. (BTW, the on the run 5yr is comparable in size to the 10yr, so the size of the issue should not be determining liquidity.)
Let’s say the 5yr is on the run. What would explain real rates rising from deeply negative levels as inflation spreads fall? The only alternative explanation I can come up with: short-term U.S. sovereign credit risk (rather than liquidity risk) is rising. This explanation is unlikely for a number of reasons. The main one is that the U.S. issues in its own currency, so U.S. “credit risk” is seen by market participants as “inflation risk”. Expectations of meaningful U.S. ratings downgrades would arguably drive higher, not lower, inflation premia.
LikeLike
David G and David P, in the interests of trying to understand this strange phenomenon, I made some charts of my own, come by and see:
http://jpkoning.blogspot.ca/2012/05/tipsy-tips-spreads.html
I made long-term (back to 2006) charts of actual TIPS vintages rather than using constant-maturity data, which is a constructed number. The current period is not the only period to have shown rising near-term yields and falling distant yields (in 2010 and 2011 this also happened). Near-term TIPS are the most sensitive on the TIPS yield curve and also the most correlated with changes in stock markets (using a rough visual approach).
LikeLike
JP Koning,
I think those near-term TIPS are old, off-the-run bonds. I believe TIPS issuance is concentrated in 5yr and 10yr instruments, so a 1 yr would be the remnant of a 5yr opened four years ago. These would be illiquid and most sensitive to rises in liquidity premia.
LikeLike
You’ll notice that at the end I compared the on-the-run 5-year 2017 issue to an old 10-year issue maturing at the same time. If there was a rush to invest in on-the-run issues in order to enjoy their superior liquidity, one would have expected the price of the 5-year 2017 issue to rise relative to the 10-year 2017 issue, but this does not seem to be the case.
LikeLike
Missed that one. Barring the “on the run” theory, I don’t know what explains a shorter maturity having a higher liquidity premium than a longer one.
LikeLike
Has it occurred to anyone that the CPI component of the inflation protection is not updated continuously? It is updated semi-annually (May and October).
Here is the latest press release from the U. S. Treasury:
http://www.treasurydirect.gov/news/pressroom/currenteebondratespr.htm
If markets are supposed to be forward looking then why did the selloff occur after the May release?
LikeLike