Friday, February 02, 2007

Accrued Interest has TIPophobia

My recent post on my aversion to TIPS resulted in a torrent of comments. OK, 10. OK, half of them were mine. OK, so we don't get a lot of comments here at Accrued Interest. Anyway...

I thought we'd do a little investment math on the subject, and really get down to it. Now, I normally try to avoid lots of formulas on this blog, because people hate to read that shit. But in real life, the math is what bonds are all about. If you follow closely to what is written below, you will see exactly why I don't buy TIPS.

Based on Bloomberg's survey, the median prediction from economists for CPI in 2007 is 2.0%. So we'll use that as a baseline. First I'll compare a 10-year TIP vs. a nominal Treasury bond. To make things easy, we'll assume that real interest rates don't change. Nominal interest rates will change based entirely on shifts in inflation. We'll run inflation as 1%, 2%, and 3%. We'll also run a 1-year horizon.

Inflation Rate = 3%: TIP 5.32%, TSY -1.88%
Inflation Rate = 2%: TIP 4.42%, TSY 4.82%
Inflation Rate = 1%: TIP 3.47%, TSY 11.89%
AVERAGE = TIP 4.40%, TSY 4.94%

So if you have equal views on inflation, the TIP is the wrong investment. But if you have even a small view that inflation is more likely to rise than fall, the TIP might make sense. In fact, the weighting to the higher inflation scenario needs to be just 38% to even the score.

Now, let's assume the Fed gets involved when inflation changes. Of course, why inflation changes is a significant question here. If energy prices are weak, and it causes CPI to print low, the Fed might not do anything. But for the sake of argument, let's assume that the Fed hikes 100bps in the 3% scenario and cuts 100bps in the 1% scenario. In both cases the effect on 10-year rates will be exactly half that of short rates. I.e., real rates shift by 50bps in addition to the inflation effect.

Inflation Rate = 3%: TIP 0.95%, TSY -5.11%
Inflation Rate = 2%: TIP 4.42%, TSY 4.82%
Inflation Rate = 1%: TIP 7.52%, TSY 15.56%
AVERAGE = TIP 4.30%, TSY 5.09%

The average difference in return is 79bps. But you only need to increase the weighting of the 3% scenario to 41% for the trade to be breakeven on average.

But we live in a world of choices. Instead of buying the 10-year Treasury by itself, what if you added a 50% position in a floating-rate bond? Again, assuming the same Fed action. We'll assume the Fed moves quarterly, the FRN resets at LIBOR flat, that LIBOR moves exactly in line with the Fed, and starts at 5.30%. Here are your returns.

Inflation Rate = 3%: TIP 0.95%, TSY/FRN 0.28%
Inflation Rate = 2%: TIP 4.42%, TSY/FRN 5.06%
Inflation Rate = 1%: TIP 7.52%, TSY/FRN 10.24%
AVERAGE = TIP 4.30%, TSY 5.20%

The breakeven probability for the 3% inflation scenario is 72%! That means that unless you view the 3% inflation scenario to be more than 72% likely, you are better off owning a combination of floaters and traditional Treasury bonds.

But wait, it gets better.

Remember that if your economic forecast is for more inflation, its entirely possible that the Fed recognizes the problem early enough to prevent any actual inflation. The result would be that rates rise, but the rate hikes mute and/or prevent any acceleration in inflation. In that case, the real rates impact on both the TIP and Treasury would be the same, but your FRN position performs extremely well.

Also, bear in mind that we've only run a 1-year scenario. Let's say that at the end of 1-year the inflation spike has resulted in several Fed hikes, but the market now figures inflation is whipped. So while you got a year's worth of extra inflation-adjustment on your TIP, when you go to sell the thing, less expected inflation is priced in, so you've suffered some price loss there. Meanwhile the TSY/FRN portfolio has performed much better, because the subsiding of inflation pressure has allowed the Treasury to appreciate (or not depreciate as much), while the floater is benefiting from higher short-term rates.

Finally, it tends to be that the curve flattens as the Fed hikes rates. The classic play on a curve flattening is to own a position in longer bonds and a position in floaters. So the TSY/FRN portfolio benefits from this tendency as well.

If you view inflation as incipient, floating-rate bonds are a good investment. They perform well whether short-term rates rise because of Fed activity or actual inflation. The weighting you give floaters in a portfolio depends on how aggressive you want to be with your inflation bet. In this piece, I gave the example of a very simple portfolio with just 10-year Treasuries and floating rate notes. In reality, you'd probably own various bonds, but you'd probably do well to make something of a barbell out of your portfolio, increasing the 10 yr+ bucket and the >1 year/FRN bucket.

Unless you have a very specific view of inflation/Fed activity which you want to play, or you have a CPI-based liability to match, I don't think TIPS are the right investment.

11 comments:

Anonymous said...

One thing though...You don't want to look at this year's expected inflation. You want to look at 10 year expected inflation if you're looking at a 10 year bond. You need Michigan Household for that.

Anonymous said...

I guess we have to wait until next March to see if things have changed. Of course, they may be completely wrong....From July 06.

http://www.federalreserve.gov/boarddocs/hh/2006/july/ReportSection2.htm

"On average over the first half of 2006, the median respondent to the Michigan SRC survey continued to expect the rate of inflation during the next five to ten years to be just under 3 percent. In June, the Survey of Professional Forecasters, conducted by the Federal Reserve Bank of Philadelphia, reported expected inflation at a rate of 2-1/2 percent over the next ten years, an expectation that has been roughly unchanged for the past eight years."

Anonymous said...

TIPs are inflation insurance. Right now, you are paying the premiums on that policy and not seeing a benefit. People hold gold for a related kind of monetary insurance. I don't see either as an investment but, both are useful diversifiers in moderation.

Anonymous said...

Looking at Lehman TIPS Index and Lehman Long Treasuries Index (Nominal), TIPS outperformed since 1997 but greatly underperformed over the past year. (Overall, TIPS outperformed.)

I would agree that TIPS should be used in moderation if at all, depending on how close one is to high real liquidity needs like retirement.

Accrued Interest said...

OK lots to address here.

You make a fair point about long-term inflation expectations vs. 1-year expectations. I was trying to show the potential return over 1-year, not over the life of the bond. You could easily do the same analysis over the life of the bond. I don't buy bonds expecting to hold them to maturity. I mean, sometimes I do, but usually not. So I always use a shorter horizon.

Surveys and accuracy are basically antonyms in the investment market. I put very little weight on any survey. But if we're going to do analysis on TIPS we have to start with some inflation assumption. We should just be aware that any assumption has a low probability of being exactly correct.

Don't you always over pay for insurance? Isn't that how Buffet made his billions?

I'm aware of the long-term performance of TIPS indices. That's something that Western Asset (who seems to worship TIPS) points to a lot. I think there was some period where TIPS was gaining acceptance in the market. Any time you are willing to be an early adopter in the bond market, you'll get paid a premium in yield. If later the asset class becomes widely used, you'll add alpha. Look at hybrid-ARMS. Anyway, I think that period has come and gone. Unless the Treasury vastly expands the TIP auctions to where they rival Treasury auctions in size, there won't be any further reduction in the liquidity premium.

Anonymous said...

I see massive inflation in 10 plus years due to 1. Massive budget deficits (Baby Boomer retirement, exploding Medicaid/Medicare costs). 2. Massive oil price increases (peak oil). If the Fed were to deal with this by monetary means it would result in huge unemployment. So I think the Fed will choose the inflation instead. (Or more precisely there will be a compromise resulting in stagflation (high unemployment, high inflation)

Vivek Vish said...

This paper by the Richmond Fed shows that the Michigan Household Survey's estimate is unbiased and appears to be efficient with respect to predictive economic variables. (The Survey of Professional Forecasters is neither unbiased nor efficient.)

http://www.richmondfed.org/publications/economic_research/economic_quarterly/pdfs/summer2002/mehra.pdf

I understand what you mean about the early adopter premium, if I may coin the term. But even if that has played itself out, I don't think that you can assume that the TIPS market is efficiently priced with reflect to expected inflation. You SHOULD always pay for insurance, but I wouldn't immediately assume from the get-go that the market has priced these probabilities correctly.

And you don't have to be spot-on in your inflation prediction. You just have to be wrong on the upside--or if your inflation prediction is greater than inflation implied by the TIPS / Treasury differential, then not TOO wrong on the downside.

I'm not even implying that I'd be long TIPS. (I have $75,000 in tuition debt at 9% interest. The only thing I'm willing to go long on right now is myself.) Just that you shouldn't dismiss TIPS on the expectation that the inflation insurance discount is efficiently priced. You should dismiss TIPS at any given time given one's expected inflation versus the implied inflation of TIPS plus whatever you would pay to get rid of inflation risk (if anything)or how much payment you would DEMAND if you have many nominal liabilities and don't want to get rid of inflation risk.

As far as the above post by anonymous about high inflation, I really wonder to what extent budget deficits and government debt correlate with higher inflation. I am familiar with the theoretical relation but it's surprising how often theory fails to adequately explain the economy. Furthermore, the only people who are really predicting massive spikes in oil is Goldman. Even if there are massive spikes, oil prices are losing their ability to feed into inflation as energy consumption per unit GDP declines (due to increases in efficiency).

I would however like to see the evidence that would refute that (or any of my claims). I think we're here for the same reason--finding the right answer and then making some cash from it.

Accrued Interest said...

This debate is getting fun.

It does seem that if the Michigan survey of just under 3% is correct, the 10-year TIP would outperform the 10-year Treasury. But as I said in previous posts, you have more choices than two, and I think there are better choices.

As far as oil goes, that is the best argument for TIPS, in my opinion. Because if headline CPI is relatively high solely because of oil, and core PCE remains manageable, the Fed will do nothing. Therefore the TIP benefits from rising headline inflation while short-term bond rates do not rise. While I believe a long commodity position is a better choice given an outlook for rising oil, it may be that the TIP is a more viable option for some investors.

As far as deficits causing inflation, that's a controversial link at best. I tend toward the neo-classical/monetarist school of economics, which dismisses any link between government spending and inflation. I haven't read any academic data which empirically supports a strong deficit -> inflation link within reasonable bounds. I agree with Milton Friedman: inflation is always and everywhere a monetary phenomenon. You know P=M/Y and all that.

The deficit thing is more of a Keynesian idea, but honestly I don't think most neo-Keynesians view the link to inflation as particularly strong. You have to explain the generally dysinflationary period of 1981-1992 if you are going to make some claim of a deficit=inflation theory. Good luck with that.

All these comments have got me thinking about whether there is some paired trade possible here. Short TIPS at some maturity, long Treasury + floater in some combination. I'll have to think about it more.

Anonymous said...

Vivek, I think we have "dueling papers" with different conclusions. See this 2006 paper from NBER and the Fed. Survey of Professional Forecasters and Livingston do best:

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=770947

Re: TIPS, I agree with Tom's points that the liquidity discount has disappeared since their introduction, and index returns reflect that. This is likely to be non-repeatable. Second, Tom's comment about TIPS only in tax-exempt accounts is a further drag on any liquidity increase. Have you tried explaining TIPS or TIPS mutual fund performance to a retail client? Their typical answer is "All I know is that I'm paying taxes for income I didn't get to put in my pocket."

Accrued Interest said...

MN Investor is right. Individual investors hate paying taxes on cash they don't actually see. Now of course, they happily reinvest dividends/capital gains on their mutual funds. But some how that's different.

A wise bond salesman once told me that in the muni market, the best hedge is a retail out. By that he meant that retail investors (and their brokers) don't understand the first thing about bonds, so if the trading desk buys a bond and the market moves against them, they just sell it to retail.

Vivek Vish said...

Blue Chip Economic Indicators is forecasting 2.3% inflation in 2007, implying TIPS are fairly priced (for this year) to expectations assuming no inflation volatility premium on nominal bonds. Of course, as you mentioned, expectations and surveys are just that.

http://news.yahoo.com/s/nm/20070210/bs_nm/usa_economy_bluechip_dc;_ylt=AiTReLRYKV17M16n2_TWh3CyBhIF