Greg Mankiw notes some odd behavior this week in the values reported by the U.S. Treasury for the yields on constant-maturity Treasury Inflation Protected Securities.
In normal times when investors anticipate positive inflation, the yield on nominal Treasuries should exceed that on TIPS, since both coupon and principal on TIPS grow with the CPI. That normal state of things was dramatically reversed over the last month, when the 5-year TIPS came to pay 200 basis points more than the 5-year nominal Treasury. But on Monday, the TIPS yield fell 214 basis points, while the nominal yield was down only 22 basis points, leaving the TIPS yield only slightly above the nominal.
 |
Here is Greg’s interpretation:
That is a huge change over only a few days. What happened? It appears to be, in large measure, a figment of data construction.
Here is how
the data are made:Real yields on Treasury TIPS (Treasury Inflation Protected Securities) at “constant maturity” are interpolated by the U.S. Treasury from Treasury’s daily real yield curve. These real market yields are calculated from composites of secondary market quotations obtained by the Federal Reserve Bank of New York.
And this is what you find in the footnotes:
Starting 12/01/2008, the TIPS yield curve will use on-the-run TIPS as knot points rather than all securities under 20 years.
Why such a large difference between on-the-run (new) vs off-the-run (old) bonds, and why did the issue only arise now? I am not sure, and the Treasury website does not explain, but here is a guess.
TIPS offer asymmetric inflation-protection. If the price level rises, your principal rises as well. But if the price level goes down, you get your initial nominal principal back. (Don’t believe me? Click here.) A new TIPS bond is great when there is risk of deflation. It is a real bond if prices go up, but more like a nominal bond if prices go down. Heads you win, tails you win also.
An older TIPS bond, however, is not as attractive. A lot of price inflation is already built into the adjusted principal. All of that inflation has to be undone by subsequent deflation before the nominal floor on the principal kicks in. As a result, when there is risk of deflation, the older bond has to offer a higher yield to compete with a newer one.
In other words, after a period of inflation, an older TIPS is closer to a true real bond, whereas a new TIPS is an attractive hybrid. This fact could explain the large jump down in the inferred real interest rate when the Treasury changed the raw bond data it uses. And it can explain why the issue became significant only recently, as people have started to seriously worry about deflation, inducing Treasury to change its calculations.
One implication of this hypothesis is that the real interest rate now reported is not a true real interest rate but is infected by the hybrid nature of these bonds. Yields on older off-the-run bonds may be more meaningful.
Of course, my conjecture could be completely wrong, as this is not my specific area of expertise. Another possibility is that the difference between these bonds instead has to do with changing liquidity premia. But one thing I am sure of: It is best to be wary of data from the TIPS market.
In support of Greg’s first conjecture, the 5-year 2011 TIPS and 10-year 2012
TIPS showed the 10-20-basis-point drop of the nominals rather than the 200-basis-point drop of the Treasury’s 5-year constant maturity on Monday.
I note that this interpretation would imply that although the spread cannot itself be viewed as the actual expected inflation or deflation rate, its moves over the last month are attributable to concerns about a fall in the price level. On the other hand, Jeff Hallman speculates that another factor in the recent anomalous behavior of the TIPS yields may have to do with whether they are accepted as collateral for certain transactions.
Any additional insights from our readers would be most welcome.
Technorati Tags: macroeconomics,
,
inflation expectations,
economics,
deflation,
TIPS yields
Perhaps the market is anticipating that this is where the Fed will start with any official quantitative easing.
If I were the Fed and I wanted to conduct quantitative easing, I would want to buy liquid assets that retain their value in the event of high inflation. That way I could reverse the monetary expansion by selling these assets. While buying short-term foreign government bonds might also work, buying TIPS seems a particularly transparent and effective way for the Fed to let everyone know that they can actually reverse the helicopter drop (at which point, we may need a new metaphor).
The main driver remains expected inflation/deflation, though. Should we be surprised to see markets anticipating deflation? I don’t think so. In fact, the most likely scenario is initial deflation of the money supply beacause the expansion of the monetary base will hardly be enough to compensate the credit crunch. But once credit starts expanding again, if central banks are lagging behind, the overexpanded monetary base will cause inflation. How many years will this reverse bubble take is anyone’s guess.
I totally missed Mankiw’s explanation, and it makes a lot of sense to me. kristi and I over at Across the Curve had concluded that this really was just a large illiquidity premium, and not a risk premium, as I’d thought.
Mankiw’s explanation is far more satisfying than both, and indeed, the data is now definitely not useful anymore. Thanks for linking this.
If one values the nominal option in the terminal real return zero-coupon, we have essentially an bond option at the money with the strike at par.
To get such a different in yields between of the run and on the run (200bps p.a. during 5 years)requires a huge inflation rate volatility (400 to 500 bps per annum). It is of the order of magnitude of what we had recently on the off the run short maturity TIPS, but it sounds at first sight a bit stretched to see such a volatility staying so high for 5 years.
A possible explanation is that, in the current environment, a gaussian random walk of the instantaneous inflation rate is not a valid representation any more. With the economy teetering on the verge of deflation on one side, and prominent economist like Ken Rogoff advocating a 6% inflation target on the other side, the “market-implied” distribution of the 5Y maturity CPI may look bi-modal, with one peak at 80% of current (I.e. Great Depression redux) and another 130/140% of current (I.e. Bernanke’s Apocalypse now – slightly adapted – :
“I love the smell of the Fed’s trading room at the end of the week . You know, one time we had an quantitative easing for 1 trillion dollars. When it was all over, we didn’t find one of ’em, not one remaining defaulted mortgage. The smell, you know that arm pit smell, the whole bond market. Smelled like… victory. Someday this stagflation’s gonna end…” :-))
On-the-run TIPS are a composite bond: a risk-free bond and a call option on CPI. Seasoned TIPs bonds are closer to a pure inflation bond (CPI plus real return). In addition to collateral issues, there is complexity risk and tax inefficiency that impede broad distribution of TIPs bonds. That’s why the liquidity is less (bid-ask spread are wider).
By changing the time-series, the Treasury has created a data-disjunction that will be exploited by partisans to claim that deficit spending doesn’t increase inflation expectations. I can’t wait to read (for years to come) how the market proves that monetized deficits do not increase inflation expectations.
Prof. Hamilton, I am a fan of the blog and I am quite thankful for the knowledge shared here. I’ll attempt to comment on the issue at hand. The answer will begin to appear if you look at the yield difference between the apr/13 and jul/13 TIPS. The april issue is the on-the-run 5 year yielding aproximately 1.75% versus 3.75% for the july issue. This difference in yield is what is causing confusion here (given similar maturity, liquidity being relevant, but not that relevant) and it is mostly due to the pricing by the markets of the “par-put” on the principal of the april issue, like Prof. Mankiw pointed out. Since the april 2013 TIPS was recently issued, its accrued inflation index means that a little deflation will make it hit its deflation floor, thus that option is “in-the-money”. That is not the case for the july issue which is much older and has much more accrued inflation to its principal (therefore the deflation put is still way out-of-the-money). My opinion is that a truer measure of real yields and implied inflation/deflation should be extracted from issues which have accrued significant inflation and are not exhibiting “optionality” in their pricing. This can be observed to varying degrees across the TIPS curve, but is most relevant for the apr/13 issue. Hope this helps.
SD
A couple of weeks ago, Mankiw posted another hypothesis to explain the negative TIPS spread.
If I understand it correctly…
Assume that 1) investors seek a real return; 2) investors prefer certainty to uncertainty; and 3) future inflation is now, for some reason, positively correlated with the broader market’s random walk.
Then nominal Treasuries are now anti-correlated with the market, and can therefore be worth “something extra” when added to a diversified portfolio, because they help cancel out the randomness. Thus Treasuries might be more attractive than TIPS simply because they are more useful for providing a less uncertain real return in a diverse portfolio.
For this explanation to work, you have to assume not only the three statements above, but also that inflation is not normally so correlated with future market returns; otherwise, the TIPS spread would have been negative all along.
I am not saying I believe this explanation. (For one thing, it presupposes an awfully academic “random walk” model of markets.) But I must admit it is a possible alternative to the market expecting negative 2% annualized CPI over the next five years.
Charles and SD have described exactly what’s happening in the TIPS market.
Nemo also brings up a good point. Since the beginning of the year, a lot of the volatility in the spread between TIPS and nominals has had nothing to do with expected inflaton. For example, in the 5yr sector, TIPS swap spreads (adjusted for inflation) have narrowed from 20 to -100, while nominal swap spreads have widened from 70 to 85. This disparity would make it look like expected inflation fell by 135 bps more than it really did.
I’m confused. If I buy an off-the-run TIP (old) bond that has a positive adjusted principal due to inflation at the time of purchase, and I hold that bond to maturity during a period of deflation, can I lose money? In other words, can the principal adjust downward during the deflationary period such that the returned principal at maturity is less than the price I paid for the bond?
Where are you getting your data? The 5yr TIPS yield didn’t fall by 200+ bps yesterday. However you are measuring the 5yr TIPS yield is incorrect.
I’m guessing that you may have had an index that was referencing the July 2012s or July 2013s and then switched to referencing the April 2013s.
The April 2013s have a yield approximately 2% lower than the other TIPS because of the deflation floor. The inflation premium in TIPS accrues in a factor. TIPS principal pays out with a factor of at least 1, which is called the deflation floor. The factor for the April 2013s is 1.03 so the current breakeven curve implies that you’ll hit the floor. However for other similar maturity TIPS the factor is around 1.20, and it’s not currently priced that we’ll have the cumulative 20% deflation over 5yrs to hit the floor. Hence the difference in pricing. You need to be very careful in this environment about a constant maturity chained index of TIPS yields because different TIPS have different characteristics.
Mankiw makes sense to me.
I had seen the big rate drop on tips and was excited to check my vanguard tips mutual fund. I was puzzled and dissapointed to find the fund price rose very little despite the big drop in the posted rate.
I guess this is bad news. Deflation remains a serious risk. Time to monetize the debt!
The deflation protection is not as asymmetric as Mankiw argues, because the par value is guaranteed only at maturity. During the life of the bond the principal can decline below $1000 if there is deflation. On longer maturities, this makes the par value protection fairly meaningless regardless of whether the bond is on the run or off the run.
ronmexico:
If you buy a TIPS at new issuance you cannot get back less principal than you paid. This is taken into account with the factor. At issuance the factor is 1. At maturity the factor needs to be at least 1. The inflation accrual occurs in the factor. For example, if you buy a TIPS and 12 months later the year over year CPI is 5% the factor will be 1.05, indicating a 5% annualized increase. For TIPS, all coupon payments are the real coupon times the factor to give you a real coupon rate. This is true for the principal also.
Older TIPS have higher factors reflecting the cumulative inflation since issuance. In the 5yr space of the TIPS curve there are two types of TIPS issues — recent 5yr issues and old 10yr issues that have rolled down the curve to be current 5yrs. The old 10yrs that now have 5yrs to maturity have approximately 20% of accrued inflation while the new 5yrs only have around 3% accrued inflation.
If we enter deflation the factor will start to decline. With the new TIPS the maximum principal that you can loose from deflation is 3% while it is around 20% for the older TIPS. The market is doing an efficient job of taking the embedded deflation in the TIPS curve and valuing the deflation floor to maturity. Because there is so much deflation embedded in the TIPS breakeven inflation curve it translates to a significant difference in pricing between issues because of the difference in expected principal at maturity.
TIPS manager: The data sources are indicated in links below the figure. These are synthetic constant-maturity calculations performed by the U.S. Department of Treasury that are sensitive to which bonds are used. There was evidently a change in the included bonds beginning on Monday, and it is the change in which bonds are used rather than a change in the yield that produces the drop in the series. That is Mankiw’s original point, which seems to be the same one you are making.
TIPS Manager:
I understand how the deflation floor works. I was saying that the perceived funding difference between TIPS and nominals accounts for a large portion of the yield spread between TIPS and nominals, and it has nothing to do with expected inflation. For 5yr TIPS, the funding difference is worth 185 bps. In other words, the 5yr breakeven rate of inflation is 185 bps higher than the yield spread you see on the screen. If the perceived funding spread returned to a more reasonable level, say 20 bps, breakevens would widen by 165 bps if inflation expectations remained unchanged. This is one reason why the market isn’t pricing the floor efficiently.
to MaynardGKeynes
on the run long bonds issued in 2008 have been issued at a fairly low real yield (1.75% for the 2028) giving high duration to the bond (at 1.75% over 20 years, 70% of the value of the bond is in the principal repayment). Moreover,they have traded down to 80% of par (I.e. 3% real yield) (in reality it is around 85 because inflation was high between january 08 and september 08). At such price close to 85% of the value benefits from the “deflation put”. Not meaningless for me.
to ronmexico
WIth the funding constraints that practically all market players are facing, it is very difficult to set-up the arbitrage that reverts prices to their “efficient”” level
Also, in the same fashion that forward rates are not expectations of future rates, break-even rates are not expectation of future inflation rates : there is an element of risk adjustment in them, one way or another. This being said, central bankers like the expectation idea, because it is for them an way to measure “long term inflation expectation” that sheds a different light than economic agents surveys.
In a quantitative easing environment, we could end up up with the tail moving the dog : In order to publicize its commitment to deliver inflation, and alter public expectation of inflation upwards, the Fed could put buying pressure on TIPS. After all, it is “only” a 500 bln market…
All well and good, this attribution to a statistical measurement change. But can one of the experts explain to a novice why the same 5yr TIP bill auctioned to a yield of 0.745% in April, and 3.270% just 6 months later? Same maturity, same coupon, quite a difference in bid.
The TIPs market is much smaller than the overall Treasuries market. As a result, with the unusually high volatility in all of the markets, its trends may not precisiely match the movements in the much bigger and more liquid markets with bigger institutional movements. Many of the large institutions don’t use TIPs for overall fast portfolio adjustments at all, so there are short periods of relative inefficiency that can occur.
In order to publicize its commitment to deliver inflation, and alter public expectation of inflation upwards, the Fed could put buying pressure on TIPS.
Does it require a tin foil hat to see a connection to the announcement, 1 year ago today, in the reduction of individual annual purchase limits of I Bonds?
David, I’m not a bond expert, but I think the short answer to your question is yes, you can lose on a pre-owned TIP. When you buy, the seller takes the inflation adjustment to date, scaled by the price. You get the ‘flation effects thereafter, which may net to a reduction in maturity value to less than you paid. If the coupon interest you receive until maturity is less than the reduction in bond value, you lose. I saw a post on CR that suggested that short life TIPs were a safe way to get good interest, but I wonder if the poster really understood his chart, and the potential downside.
I used to think like you about the convenience of purchasing new tips vs old ones in the sense that you describe. This is not the first deflation scare since 1998 when they started to be sold and I started dealing with them.
However, on second thought, this is not a very reasonable asumption.
Unless there is a very long deflation, or you have to sell the tip before inflation overcomes the past deflation, the results of purchasing new or old tips seem to be the same.
Say you purchase an old one. When you purchase, you pay an accumulated inflation of say 20%, since the tip is old. You pay 120 (lets say the price is 100)
Then follows say 8 months of deflation (a long time) and the accumulated deflation is 4%. OK, now you are losing 4% because (I am simplifying a bit), if you sell, you will get 116 (Lets say the price is 100 again). But if you do not sell and keep, lets say 16 more months, in which inflation of 5% is accumulated. The inflation coefficient is now 121 (again simplifying), then after 2 years, you are above 1%.
Now, consider what would have happened if you purchase a new one. And keep it also for 2 years. After the first period of 8 months, the coefficient is 96. And if you sell, you will as in the other case, loose 4%. (The guarantee of 100% payment by the Treasury, is, crucially, at maturity).
Then you wait 16 months. Inflation acumulates, and the coefficient is at 101. Same results.
There is a difference only if all through the life of the bond, the results become negative. Then, it can be said that your original decision of purchasing a new tip was better than to buy an old one.
I supose that somehow this advantage is priced in the different issues.
I hope this is useful.
Jack
TIPS went ape-crazy durign the last weeks of November and this has nothing to do with changing any government calculations. In particular, the 1/15/2026 bond was up to 3.5% yield-to-maturity about 2 weeks ago and now is back to about 2.7%, which is more or less where it was before that sudden spike in late November. I know this because I was urging a friend to buy when yields were high (unlike me, she can’t take the risk of stocks, which were also a great deal in late November, when the SP500 hit 750). My theory is that some hedge fund blew up when stocks crashed and they had a margin call to meet. They responded by dumping a lot of TIPS on the market at once. There’s lot of these crazy things going on the markets these past few months. Opportunities for active traders abound.
Hence the problem with owning TIPS through a mutual fund, as became evident in late 2006 – by definition these portfolios hold mostly off-the-run issues, so the price “floor” isn’t really there quarter to quarter.
And, btw, kudos to whomever coined the phrase “pre-owned TIPS”.
maynardGkeynes posted:
The old 10yrs that now have 5yrs to maturity have approximately 20% of accrued inflation while the new 5yrs only have around 3% accrued inflation.
If we enter deflation the factor will start to decline. With the new TIPS the maximum principal that you can loose from deflation is 3% while it is around 20% for the older TIPS.
Is this accurate? Take my example below:
If I bought 15Apr 2010 TIP today @ 92.9406, and an Inflation Factor of 1.1534, my Principal investment would be $107,197.69. On 15Apr2010, I will get a payout of the MAX ($100,000, Inflation Factor*$92,940.60? So my max loss is not 15.34% of Principal, which is what it seems maynardGkeynes would say it is, but rather 6.7143% of Principal ($-7,197.69/$107,197.60). Am I thinking about this correctly?
A subtle point is that the inflation factor on a TIP can fall below 1 prior to maturity, which matters for the calculation of the nominal coupon payments (pls correct me if I am wrong).
The “floor” of 1 (or par) matters only at maturity, when the TIP is sure to return par.
So, the yield differential between the newer 5-yr TIP and the older (off-the-run) 5-yr TIP should reflect the risk of the CPI five years from now being lower than it is today. To me, that risk seems negligible.
(Alternatively, these bonds should have similar holding period returns if there is a drop in the CPI next year, followed by subsequent increases.)
Idea: Trade puts and calls (*) or futures based on terminal inflation adjusted principle values?
Wouldn’t that provide the information needed?
* European style