Trills
A recent post by Robert Shiller proposed a new type of government bond, called a trill. These bonds would pay the holder one trillionth of US nominal GDP each quarter. Today this would be about $3.50, or $14 per year. He suggests that these bonds might sell for as much as $1400 given current conditions in the financial markets.
Robert Shiller and I have shared an interest in NGDP futures for almost two decades, albeit for different reasons. In 1989 I published a paper advocating that the Fed buy and sell unlimited quantities of NGDP futures at the target price, and then use those transactions to adjust the monetary base. Shiller has focused on the financial angle; he argued that NGDP futures and/or NGDP bonds could make financial markets work more effectively. It seems to me that subsequent events have strengthened both of our arguments. If mortgage payments were linked to NGDP then mortgage defaults would have been lower in 2009, and if we had targeted NGDP expectations then any crisis that did occur would have had relatively little impact on AD.
Michael Pettis also has a longstanding interest in this idea, and comments on Shiller’s essay:
In my book, The Volatility Machine, I discuss the same things, arguing that developing countries typically put on what I called “inverted” debt structures, which automatically exacerbate volatility. The worst sources of this kind of inversion are either external debt, short-term domestic debt, or contingent liabilities arising out of the banking system. All of these forms of debt perform better than expected during good times and much worse than expected during bad times, and so they are an important part of the reason why developing countries, especially highly indebted ones, seem to veer so easily from boom to bust.
One of the things developing countries need to do to help break this cycle is to restructure their balance sheets in order to reduce embedded pro-cyclical structures and so reduce volatility. The best way would be somehow for countries to sell “equity”, the closest thing to which has been long-term, fixed-rate local currency debt. Schiller’s “trills” are an even better example. The main point is that these kinds of capital structures force the users of capital to pay more when times are good and less when times are bad. This provides an important cushion for when times are bad, and the very existence of this cushion not only will reduce the tendency for capital to flee a country just when it needs inflows most, but it should reduce the overall cost of capital by reducing financial distress costs.
I think “trills” are a great idea, and I remember writing a piece many years ago for the Financial Times (“A stake in Argentina’s future”, July 2, 2003) in which I praised the attempts – however minimal – to embed such a structure in bonds issued by Argentina as part of its 2003 debt restructuring. The Argentine structure was a tiny first step (it only involved a minimal amount of GDP warrants), but if a major developed country were to issue these “trills” and make them respectable, this would be very positive for developing countries who, like China, are much too volatile, tend to fly back and forth between periods of intense growth and intense despair, and have very few options for building hedges into their national balance sheet.
I would like to comment on Michael’s statement that “This provides an important cushion for when times are bad.” It may be obvious, but this protects borrowers in two ways; they are protected against falling RGDP, and falling price levels. Of course both deflation and depression make it harder to repay loans, which is why indexing bonds to NGDP is superior to indexing to the price level.
I hate to beat a dead horse, but you guys know by now that I am anti-inflation. Not against actual inflation, rather against using the concept of inflation in economics. Statisticians just pull inflation numbers out of thin air, and they are not reliable.
BTW, that was the point of my BU dorm post. Some commenters, such as Matthew Yglesias, thought I had some kind of right-wing agenda; that I was trying to undercut Johnston’s point about inequality. Actually my only point was that the CPI is unreliable, and also that dorms are the best way of evaluating changes in living standards over time, because they bring together a wide range of people. I do concede that dorms slightly understate the true rise in living standards, as an increasing share of Americans are going to college, and thus it is becoming less and less of an elite sample. College students now come from a wide enough range of backgrounds that changes in dorm quality, which reflect changing student expectations, are much more reliable than the CPI. We should abolish the CPI and put those tax dollars into creating GDP futures markets.
I want to put my name out there along with Pettis and Shiller as an enthusiastic supporter of NGDP bonds. The coupons could be sold separately, and their value, when compared to nominal zero coupon bonds, would provide an excellent real time estimate of NGDP growth expectations. The Fed would have no excuse for ever letting those expectations get more than 1% above or below the target level. And no more liquidity traps! Were we to adopt NGDP futures targeting then I believe that we could even surpass the Aussies in number of consecutive decades without a recession.
Those readers interested in my China posts should look at Michael Pettis’ entire post. He has the best China blog that I have been able to find. He is more pessimistic about China that I am, partly because of a difference in outlook, but partly because of the old glass half full/half empty issue. I like to be a contrarian, and since there is so much skepticism about China in the US press, I like to emphasize the positive changes that I have seen. But I defer to his expertise about the microeconomic/banking problems in China.
PS. The new blog still has a few bugs to work out, for instance today it slowed to a crawl at times. But we do have a full RSS feed. I have only a vague idea of what RSS means, but if it is important to my readers, it is important to me. (I apologize to one or two commenters who mentioned the RRS issue last spring. It turns out you were right, but I was constantly dealing with so many issues that I only now got a handle on it.) To give you an idea of how bad I am with computers, it took me 20 minutes to get my office computer on today. I have no idea why.
PPS. Replying to a recent comment I cast doubt on the view that the US economy is recovering. I should have said 3rd quarter data (2.2% growth) did not indicate a recovery. Given the recent performance of the stock market I believe that the recovery probably began in the 4th quarter, and that RGDP growth is probably now well over 3%. So I guess that is some sort of progress.
PPPS. If the direct link to Shiller doesn’t work, link through the Pettis post.
Tags:
4. January 2010 at 18:22
I can see that trills would encourage a country like Japan to become more active in boosting its NGDP.But what about “competition” for financing between countries? Wouldn´t it end up encouraging “excessive” inflation?
4. January 2010 at 19:25
I should just like to render a plea that Trills, like TIPS, be limited to only a small part of federal financing. The consequences of using them as a major source of financing are potentially frightening – it offers cheap financing now in exchange for creating a perpetual obligation. For a government that can’t control its addiction to deficits, I have many fears.
Felix Salmon comments:
http://blogs.reuters.com/felix-salmon/2009/12/28/the-return-of-gdp-bonds/
And I pretty much agree with him… Also:
http://seekingalpha.com/article/179973-not-so-cheap-trills
4. January 2010 at 19:41
I am glad Shiller’s trills are getting a bit more play, initially they seemed to be dismissed out of hand because financial innovation has fallen into such disrepute.
On the BTW, I didn’t take your dorm/qualitative argument as an inequality argument. But I do think college dorms as part of a sector of the economy near the end of a secular boom make a poor substitute even for the BLS’s fancy stats work. Maybe food or a 2000 calorie diet as a percent of income would be a better indication of living standards, the New Yorker had an article that mentioned that food consumed 40% of the average American’s income a hundred years ago while it’s only 8% today. If one insists on being whimsical there is always the Big Mac, which could be indexed to US incomes over time.
4. January 2010 at 19:49
Statsguy said: “I should just like to render a plea that Trills, like TIPS, be limited to only a small part of federal financing. … Also: http://seekingalpha.com/article/179973-not-so-cheap-trills”
heh… earlier today (before this post) I commented on the Aleph blog:
Merkel: “Here’s the question: what do you think nominal GDP growth will be on average forever? If it is above 4.4%, one should be willing to pay an infinite amount to buy it. At lower rates of nominal GDP growth, the security will have a finite value that declines rapidly with lower nominal GDP growth.”
ahem… risk premium? but yes, why shouldn’t they fluctuate in value since they are effectively a claim on bond, stock and real property markets.
Merkel: “They suck a lot of money in, and do not consider what it will do to the government in future years. I can say with confidence that a large issuance of trills would lead to the demise of the US Government. There is no way that the government could keep up with the payments, because most finance today relies on the idea that the economy can grow out of the debt burden. With trills, that is not possible.”
as the same logic applies, I guess developing nations should stop issuing US dollar denominated debt too?
I would like to think that both the buyers and sellers of such securities wouldn’t take the implications lightly. In any case, provided the govt does not issue vast sums, I can’t see the harm. Moreover, their pricing might yield useful signals to policy makers (especially the Fed)
http://alephblog.com/2009/12/27/not-so-cheap-trills/#comment-24113
4. January 2010 at 19:53
Thanks for the full RSS feed. MUCH easier to read now.
Shiller has a lot of good ideas set forth in his (very brief) Subprime Solution book. While he’s so skeptical of humans behaving rationally and markets being efficient, all of his ideas are of products that would make the markets more efficient by providing more information.
4. January 2010 at 19:53
@StatsGuy
Perpetual obligations are exactly what we have now, the government simply rolls over its debt. If you keep rolling over your debt, that’s exactly like a perpetual obligation. The analogy to ‘paying down the debt’ would be to repurchase Trills on the open market (using tax money).
4. January 2010 at 20:14
Thanks for the full-text feed.
4. January 2010 at 20:51
Being able to read your full post within the confines of my blog reader is awesome!
4. January 2010 at 21:37
I like the idea of the Tril, however that being said. Perpetual obligations are very dangerous. A single bad congress could spell disaster. Also, the Tril could function as an effective GDP future for free banking based on such.
5. January 2010 at 00:03
Unfortunately, college dorms attract a very narrow range of people: undergrads. Some of us believe there is a bubble in undergraduate education, and the fluffy dorms could be a result of that. I would say houses are better, but with the current generation renting well into their 30’s and the recent bubble, that’s clearly not true. If we indexed to something like hard drive capacity, we’d be deflating like crazy. So I haven’t found an answer. The CPI may be better than I used to think, since it puts nice-looking loglinear growth on things like the S&P 500. I’ll half-joke that we should just set the CPI equal to whatever puts NGDP on a loglinear trend and be done with it.
5. January 2010 at 05:17
Scott,
Any comments on David Merkel’s rationale for why “trills” are a bad idea?
http://alephblog.com/2009/12/27/not-so-cheap-trills/
5. January 2010 at 08:13
Marcus, Why would it lead to excessive inflation? If inflation rises the government would have to pay higher rates of interest on the trills. If anything, wouldn’t it make it harder to inflate away our debt?
Statsguy. I see the perpetuity aspect as a side issue. I’d be happy if we had 1, 2, 3, 4 and 5 year maturities.
OGT, I like dorms because it is the only housing market where all classes mix. Big macs are primarily consumed by people with modest incomes. And Big Mac quality is fixed, so it doesn’t show a change in living standards from quality adjustments. Dorm quality changes, and that’s what we are trying to measure.
thruth, Yes, I am not so worried. And to address Statsguy’s point, if there were problems then I doubt they would ever become a large share of the debt.
Thanks JTapp. Shiller is a creative thinker.
Jsalvatier, I agree.
You’re welcome Ian and Dilip.
Doc, Yes, and again I don’t see why they must be perpetuities.
Carl, You said;
“Unfortunately, college dorms attract a very narrow range of people: undergrads.”
Actually, dorms are the only part of the housing market where all classes mix.
Mark, I am pretty sure the price would not be infinite. I don’t see how they are different from the current situation, where the government just keeps roling over the debt. If they became a burden then just pay them off and go back to issuing ordinary debt instead.
By the way, I am not saying critics like this guy and statsguy are wrong, it is a complex issue. But if people think there is a problem, then limit maturities to 5 or 10 years. We don’t have TIPS perpetuities.
5. January 2010 at 08:15
I don’t think the perpetual aspect of trills is critical to their functioning. You could easily cap it at 10 or 20 years and still preserve their countercyclical and nominal GDP indexing nature. Because the government has the abilitity to tax, they are never truly constrained by their debts. Trills automatically index to monetary inflation, so a government would have to explicitly default on them rather than printing their way out. I see this as an advantage, since I believe inflationary default to be more damaging than “force majeure” default.
5. January 2010 at 08:45
The nice thing about a perpetuity is you don’t have multiple maturities to worry about: looking forward a newly issued trill is identical to a past issue in the same way that newly issued company stock has the same rights as old stock. Trading on a single security should maximize liquidity (perhaps Shiller’s desire for a perpetuity comes from his experience with the illiquidity of multiple housing futures contracts). From that single security, it would be easy to create derivative contracts, such as strips, futures and options.
azmyth said “Trills automatically index to monetary inflation, so a government would have to explicitly default on them rather than printing their way out.”
Yep, there’s a good reason so much sovereign debt is US denominated.
5. January 2010 at 09:23
A set of 5 year rolling maturity gdp bonds seems like a much better idea. If the issues were sufficiently large, there should be enough liquidity to permit the type of targeting you advocate, and significantly deter gaming of the system. Still, would be best to limit total value to perhaps 20% of US debt at most. Trills are very much like “equity”, in which case the US debt market (trills vs. bonds) will begin to take on characteristics of the private capital markets (stocks vs. bonds). I wonder what the political impact would be of having a large set of trill-owners who are NOT incented to prefer deflationary policies (like the US bond market).
Regarding other questions: rolling maturity non-indexed bonds are fundamentally different because they are non-indexed, and therefore can be partly inflated rather than explicitly defaulted on. A default only system (even for a small portion of debt) is very dangerous; when a bad negative shock hits a heavily indebted entity without an inflation option, the choice is between generational debt slavery, and utter ruin.
[…waiting for the ensuing arguments in favor of utter ruin by gold standardbearers]
5. January 2010 at 09:46
[…waiting for the ensuing arguments in favor of utter ruin by gold standardbearers]
heh… some people just like regime change 🙂
5. January 2010 at 11:08
Statsguy: You make a false dichotemy between total default and debt slavery. In the event of a U.S. debt default, it is highly unlikely that it would be a 100% default. The U.S. would likely pay off a percentage of the value of the bond, in exactly the same way they would by inflating away a percentage of the bond.
5. January 2010 at 11:29
Azmyth:
Imagine that the US were to declare – tomorrow – that it was formally defaulting on some portion its obligations (5%). It wasn’t printing more money (which has a stimulative effect). Just defaulting, perhaps through a unilaterally determined 5% reduction in value of outstanding bonds, or perhaps going into binding international arbitration (process TBD) to determine exactly how payments would be distributed.
What do you think the ‘multiplier effect’ of that event would be? How much would the value of existing debt drop? A mere 5%? And how much would interest rates rise? A mere 5%? (Would the US create debt serniority tiers to fund existing operations?) One might argue some degree of monetary equivalence (but not really, because formal restructuring of govt debt doesn’t alter private nominal debt value directly, while inflation does), but in the real world formal debt default (of any magnitude) is a very discrete event.
The consequences of this discrete event are so heavy (and the owners of debt – including some very strong voting blocs – would be so threatened) that the central government would endure massive pain to avoid it – paying increasingly high interest rates to rollover ever more onerous debt. There’s a very real probability of initiating a death spiral.
Observe what Iceland is going through, and Iceland had virtually no meaningful impact on global macroeconomic demand.
Yet I suppose it’s possible that market perceptions could be well managed and reactions tame – but I’m a very risk averse individual when it comes to national policy.
5. January 2010 at 12:28
Sumner- I’d wager a larger cross section of people eat Big Macs than go to residential four year schools. Though I have exactly no data for this.
What you are trying to do is adjust for quality. What I am trying to do is index nominal income to a good that does not change in quality. BTW, do PPP adjustments include qualitative calculations?
I see that I haven’t convinced you of the teeth gnashing to come in academia. Ah, well.
5. January 2010 at 15:08
Scott wrote:
>Actually, dorms are the only part of the housing market
>where all classes mix.
So what? Whoever they are, they share the fact that they’re paying tuition. If you believe tuitions are too high – which, true or not, is certainly a valid inflation concern – then the dorm index goes out the window.
-Carl
* I read one analysis that found tuitions stopped paying for themselves in terms of future salary deltas somewhere around 1997. If true, they’re supported instead with cheap government loans and by putting kids in daycare while mothers work (inflation in positional goods like college education has been a hallmark of the two-income era). Maybe we should be looking at the quality of daycares.
5. January 2010 at 16:08
“Scott wrote:
>Actually, dorms are the only part of the housing market
>where all classes mix.”
No, there are still a small handful of prisons where that continues to hold true.
5. January 2010 at 18:30
azmyth and thruth, I agree with both of your comments.
Statsguy, Why does default mean utter ruin? Donald Trump went bankrupt once, and he seems to be doing fine. (I can picture you rolling your eyes, but I am curious how you’ll respond.)
I agree that default is a big problem, but so is inflation. If a country inflates away its debts that can also cause a higher risk premium next time they want to borrow.
OGT, That’s not the issue, I agree that a lot of people consume Big Macs, I should have answered your question differently. A Big Mac tells us little about living standards. One’s housing unit tells us a lot. The size, amenities, etc. If I go to a new country I could tell a lot about how rich it was by visiting people’s homes. Even in rich places where housing units are small (Japan?) I imagine the quality of the rooms and appliances would be pretty good. But sandwiches would tell me little. India might have great sandwiches, their bread is tastier than ours.
Carl Lumma, I’m not interested in the tuition issue, I am interested in student perceptions about what sort of quality is appropriate. Colleges have to cater to student taste; and as students get richer they demand plusher dorm rooms.
Your ‘positional’ comment makes me wonder how much of this is regional. In Texas you can buy a big house for $300,000. In NYC and SF there is almost nothing in that price range. I’d guess people in Texas would agree with me, but not NYC and SF.
Statsguy. Most prisons skew strongly toward the poor and minorities. You must be referring to the Club Feds, the sort of place they’d put Madoff.
5. January 2010 at 19:00
StatsGuy: I don’t think that an outright default would be a walk in the park, but neither would inflationary default. You mention Iceland about how bad it can get, but I’d rather go through their experience than Weimar Germany, Mugabe’s Zimbabwe, or any number of hyperinflating economies.
When someone buys government debt, they are explicitly taking on the risk that the sovereign might default. Why should society place the burden of the debt on people would did not accept the risk in order to spare those who did? As for the multiplier effect, bonds are a substitute for reserves and cash, and so defaulting is by itself contractionary. However, monetary policy could offset the contractionary effect by partially monetizing the remaining debt or by OMOs. A contractionary shock does not need to lead to a vicious cycle. It only does if the monetary authority screws up (which unfortunately happens fairly often).
In the event of a default (semi or total), rates would likely be so high that the government would choose to finance its operations out of tax revenue. I would love to live in a world where our government were responsible enough not to back itself into a corner out of which hyperinflation or default were the only ways out. Either scenario is quite frightening, but I still think default is the lesser of the two evils. I think your position is quite plausible, I just don’t agree.
Sorry for being so off topic. Dorms are much better these days.
5. January 2010 at 20:58
Azmyth – I will concede that if default were managed incredibly well, it could conceivably be no more (perhaps less) disruptive. Almost by definition, if policy makers have more instruments available (monetary policy + default mechanisms) they can theoretically outperform policy that uses fewer instruments.
I simply have no faith that in practice these instruments would be managed particularly well given public perceptions and political forces. You cite some cases of bad inflation, but if you review the past 50 years and observe all the inflation events (most years for any given country) vs. all the formal default events (few and far between), I believe we observe that the formal default events are on average much worse.
Re: Donald Trump… well, I suppose the US could make a go of it, Trump-style. We could just borrow a ton more money and buy some huge casinos to try to recreate our wealth.
Actually, come to think of it, that’s pretty much what we are doing… 🙂
6. January 2010 at 06:47
Sumner- The Big Mac doesn’t tell us about living standards, but the comparison of Big Macs to income does tell us something. It tells us how much people in different place or times have left over to make other purchases.
Unfortunately we can’t use Indian Dosa because not enough people in the US eat them, yet.
The qualitative points are interesting and important in daily life, but largely unmeasurable.
I am a bit confused by your stance, however, in light of your enthusiasm for PPP GDP, which surely does as poor a job adjusting for Greek and German housing/dorm quality as the CPI does.
6. January 2010 at 15:38
Statsguy, I would have been disappointed in you if you had fallen for my Trump example.
OGT, I do think the Big Mac is much better for comparing PPP between countries than it is for comparing real incomes over time.
I wouldn’t say I am a big fan of PPP either, its just that with countries like China the nominal prices are often different by 10 to 1, so you need a rough estimate at least. I haven’t thought about dorm quality between countries. In some countries only the elites go to college.
27. January 2010 at 09:50
One great thing about the Trills idea is that it really highlights the strangeness of the short-term rate-centric approach to money we have under the status quo. Imagine a world in which we had a deep Trills market and Trills were the primary asset purchased and sold in OMOs. Nothing else about monetary policy changed; i.e. the target continued to be some subjective amalgam of inflation and activity indicators as opposed to Scott’s favored market-based NGDP expectations.
Suddenly, there would be no natural stopping point between the conventional and the unconventional. Every monetary operation would be unconventional, like FDR’s silver purchase program or purchasing shares in the S&P, but seemingly with less risk of distortion given the nature of Trills as a universal equity claim on the economy. Admittedly, this wouldn’t itself solve Scott’s primary complaint about the failure to target a more meaningful variable and to target the forecast. But it would make it a lot harder to suddenly stop easing and start slowly crawling up the quantitative easing ladder at an arbitrary point of 0% FFR. The zero bound problem doesn’t go away in the sense that it isn’t really there in the first place, but it becomes a less sensible excuse and forces Fed Policy to focus on its target as opposed to its mechanism.
27. January 2010 at 10:01
dlr said “Imagine a world in which we had a deep Trills market and Trills were the primary asset purchased and sold in OMOs.”
If we ever get to the point where Trills are the primary asset traded in OMOs, I think we are done and Scott should be happy. How could the Fed credibly defend allowing a large drop in the value of Trills?
27. January 2010 at 10:45
thruth – Exactly. If trills are the main thing traded in OMOs, the currency essentially becomes convertible for NGDP futures, as Bill Woolsey has suggested. If the stock of money is too high, people buy trills because they’re inflation proof by definition, and vice versa. The fed doesn’t even really need to fix the price, although it would probably help if it did.
27. January 2010 at 12:44
How could the Fed credibly defend allowing a large drop in the value of Trills?
I think I may be missing something. The Trills pay a fraction of NGDP. A drop in the value of Trills might simply reflect a change in risk premiums (as in the stock market or with TIPS) — not necessarily a change in expectd NGDP. Or, a drop in Trills might reflect a decline in expected RGDP despite unchanged NGDP expectations. Say we had a real problem like the housing bust, but the Fed optimally followed Scott’s advice and kept the NGDP level on the same path despite lower expected real growth for a bit. Trills might decline in value if real interest rates didn’t also decline, because the prospective real returns would decline. This wouldn’t indicate an NGDP problem.
Shiller’s own paper shows the Trill valuation hypothetically varying through time or across models with different attempts at guessing what how the Trill risk premium might come out.
In this post Scott does mention NGDP bonds with the coupons stripped, which would essentially create an NGDP futures security, but that is not a Trill. A Trill is essentially an equity instrument. You can’t strip the coupons of a Trill any more than you can strip the permanent dividends from a stock. You’d have stripped it bare.
27. January 2010 at 14:06
dlr said “drop in the value of Trills might simply reflect a change in risk premiums (as in the stock market or with TIPS) “” not necessarily a change in expectd NGDP.”
If the Fed was going to credibly prevent large NGDP declines and increases (or stabilize inflation and employment) in one period they are going to do it in every period. In other words, by ensuring there aren’t large swings in NGDP in each period, the Fed will stabilize the value of Trills. I think this sufficiently captures the essence of what Scott/Bill Woolsey and others are advocating, though they might consider OMOs in NGDP futures/Trills too radical…
“Trills might decline in value if real interest rates didn’t also decline, because the prospective real returns would decline. This wouldn’t indicate an NGDP problem.”
I can accept that there might be some variation in the Trills that the Fed deems acceptable (perhaps necessary to actually extract a useful signal about expectations), but large swings would demand policy action because it would just be too much of an elephant in the room.
“Shiller’s own paper shows the Trill valuation hypothetically varying through time or across models with different attempts at guessing what how the Trill risk premium might come out.”
Shiller isn’t imagining a world where Trills are used in OMOs.
“A Trill is essentially an equity instrument. You can’t strip the coupons of a Trill any more than you can strip the permanent dividends from a stock. You’d have stripped it bare.”
I’m indifferent re targeting NGDP futures or trills, but overall I’m not sure what you are arguing here… Why couldn’t Trills be stripped? e.g. one could just strip the year 1 coupon and sell the residual trill to someone else. That residual trill becomes a regular trill in a year’s time.
27. January 2010 at 14:40
Everyone, Thanks for your interest in this old thread. I like the perspective in dlr’s first post. The problem raised in his second is real, but he answers with the idea of stripping off coupons. I think thruth is right that even a perpertuity could be stripped of the first few years. You really just need a one (or perhaps 2) year NGDP contract for monetary policy purposes. Elsewhere I’ve argued that the risk premium of a one year would be fairly small, and the time-varying part of the risk premium (which is what matters for macro stabilization) still smaller. So I think it would be good enough to avoid big swings in NGDP.