GDP bonds are a really bad idea, part 3

By Felix Salmon
February 22, 2012
a bad idea when Jonathan Ford proposed it in August 2009, it was a bad idea when Shiller wrote about it in December of that year, and it's an even worse idea now, because Shiller's decided to kick it up another five notches or so.

" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

Can countries issue equity? Greece is making a stab at it, giving its bondholders GDP warrants which start paying out “in the event the Republic’s nominal GDP exceeds a defined threshold”. Chances are, the market won’t give the warrants much value; they’re more symbolic, really, of Greece’s good faith.

But Bob Shiller is much more ambitious when it comes to such things: “Countries should replace much of their existing national debt with shares of the “earnings” of their economies,” he writes in the Harvard Business Review.:

National shares would function much like corporate shares traded on stock exchanges. They would pay dividends regularly. Ideally, they’d be perpetual, although a country could always buy its shares back on the open market. The price of a share would fluctuate from day to day as new information about a country’s economy came out.

This was a bad idea when Jonathan Ford proposed it in August 2009, it was a bad idea when Shiller wrote about it in December of that year, and it’s an even worse idea now, because Shiller’s decided to kick it up another five notches or so:

Greece’s real GDP fell 7.4% in 2010. If its Trills were leveraged substantially—say, five to one—then the dividend paid on them would have fallen by about 40%. This would have done much to mitigate the crisis, making it easier for Greek taxpayers to bear.

This is almost literally incomprehensible. I spent a long time on the phone today with Shiller’s co-author Mark Kamstra, and even he had no real idea what Shiller was talking about here. I can see how an investor might try to leverage an investment in Greek Trills (a Trill being a bond paying one trillionth of GDP every year, in perpetuity) by buying those bonds with borrowed money. But I can’t see how Greece itself could do so. Shiller doesn’t spell it out, but these things would obviously be symmetrical: Greece would have to pay out five times its annual GDP growth in good years in order to get these large savings in bad years. And that seems like a clear recipe for unsustainable debt growth.

Even Kamstra concedes as much. “I think that a country would not issue a levered Trill,” he told me. “I think it gets you in a lot of trouble.”

But even if you put aside the insane concept of leveraged Trills, the idea behind them is still really bad. Kamstra tried to persuade me that the price of Trills would be less volatile than the S&P 500, and he might be right during periods of relatively normal interest rates. But when rates fall, it seems to me that he’s clearly wrong. A perpetual bond like a Trill is valued by adding up the present value of its income stream: how much is this year’s payment worth to me today, how much is next year’s worth, and so on. When you apply a discount rate, future coupon payments are worth less the more distant they are, and the sum of the total converges to the value of the bond.

If the coupons are steadily increasing, however, the math becomes very dangerous. The coupons will rise at the rate of nominal GDP growth, which in the US will probably be somewhere in the 4% to 5% range over the long term. As a result, if you’re a risk-averse person who wants a perpetual US government security and your discount rate is say 3%, then the expected value of a singe Trill is actually infinite. Of course, no security trades at a price of infinity. But the fact that valuations can get so high in a low-interest-rate environment is all you need to know about just how volatile Trill prices could get.

The point here is that Shiller seems to think that the price of Trills would be driven mainly by “new information about a country’s economy”. But he’s wrong about that. new information about a country’s economy tells you quite a lot about what its GDP might do in the next few years. But if you’re holding a perpetual bond, fluctuations of 1% or 2% in the value of short-term coupon payments are not going to make much difference to the value of the bond. What really makes a big difference is the interest rate you use to calculate net present value. In other words, while Trills are designed to respond to news about the economy, in fact they would be an incredibly noisy and volatile instrument reacting mainly to changes in long-term interest rates.

But what if I’m wrong and Kamstra’s right, and economic news is more important than discount rates? At that point, measuring GDP accurately becomes extremely important: the markets would care greatly about differences of just a percentage point or two.

Except, you really can’t measure GDP to within that degree of accuracy. Here’s a recent paper from the Bureau of Economic Analysis:

Measuring the accuracy of national accounts esti­mates is a long-standing challenge for three main rea­sons. One, the early GDP and GDI estimates are based on partial data and are intended to provide an “early read” on the general picture of economic activity for decision-makers. These early estimates are revised as more complete and accurate source data become avail­able. Two, the source data for the national accounts come from a mix of survey, tax, and other business and administrative data; these source data are subject to a mix of sampling and nonsampling errors and biases that cannot be measured in terms of standard errors. Three, the national accounts are regularly revised to re­flect the changes in the economic concepts and meth­ods necessary for these accounts to provide a picture of the evolving U.S. economy that is relevant and accurate for today’s economy. These updates range from ex­panding the definition of investment from investments in plant and equipment to include investments in computer software to updating seasonal adjustment factors to reflect the most recent seasonal patterns.

What does all this mean in practice? Thomas Dall at the BEA helped me out, taking one recent datapoint as an example: nominal GDP at the end of the first quarter of 2009.

When it was first reported, that number was $14.097 trillion. But then three months later, in July 2009, it was revised upwards, to $14.178 trillion. A year after that it was revised back down, to $14.05 trillion, and a year after that, in July 2011, it came down further, to $13.894 trillion. In other words, between July 2009 and July 2011, the GDP figure for the first quarter of 2009 was revised down by $284.3 billion, or 2% of GDP.

And Dall didn’t pick that datapoint because it was particularly noisy: it’s the only one we looked at.

This is bad news for any government thinking of issuing Trills. Governments, after all, go to great lengths to issue easily-understandable series of bonds with fixed coupons, so that the financial markets can price them easily and have a transparent yield curve. The only people welcoming GDP bonds with open arms would be in futures markets, where traders love volatility and try to make lots of money off it.

Which, of course, is the whole reason that Shiller is pushing this idea so aggressively. Shiller is a principal in a company called MacroMarkets, which exists to create “innovative financial instruments to facilitate investment and risk management” — a/k/a volatile new derivatives.

If Trills existed, you can be quite sure that MacroMarkets would immediately create futures and options based on Trills, trying to make money off their volatility. The volatility would depress the price that governments could sell the Trills for, but at the same time it could make a fortune for Bob Shiller. “Bob’s experience in the markets is that if there isn’t enough volatility in the price of the contract, the speculators lose interest in the contracts,” says Kamstra.

So let’s discount Shiller, here, as someone who’s way too conflicted to take at face value about such things. GDP bonds are like most financial innovations: they’re much more likely to do harm than they are to do good. And no country should even dream about issuing such things until some big corporation has blazed the trail first, as a kind of proof of concept. Lots of companies, from Walmart to ExxonMobil, do better in good economies and worse in bad economies: it might make sense for them to issue GDP bonds. Let’s wait until one of them does, so that we can get a feel for how such bonds behave, before we ask our governments to follow suit.

I feel we’ll be waiting a long time. If it’s true that the price of a GDP bond can skyrocket when interest rates fall, that bond would be extremely dangerous for any company issuing it. The market value of the company’s outstanding bonds could easily exceed the company’s enterprise value, with the result that technically shares in the company would be worthless. I can’t imagine any CFO or corporate treasurer risking it. And neither should any finance minister.


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see

Felix – you misrepresent me, and I am disappointed. I did not speak of Bob’s experience with macro markets, I did not speak definitely on some of Bob’s ideas because I do not speak for him, not because I have “no real idea what Shiller was talking about here” and your implication that Bob’s monetary interests drive him only reveal a cynic’s perception of the world, and is no substitute for actual understanding of this effort.

Posted by MarkKamstra | Report as abusive

“As a result, if you’re a risk-averse person who wants a perpetual US government security and your discount rate is say 3%, then the expected value of a singe Trill is actually infinite.”

This statement doesn’t make sense – discount rates should differ based on the riskiness of the cash flow they are discounting, and 3% is way too low for an equity-like instrument like a GDP bond. One better way might be to calculate an appropriate discount rate like you would calculate a cost of equity, e.g. with CAPM, beta is close to 1, etc.

Posted by PandR | Report as abusive

GDP bonds are an excellent idea because they would provide a market check on politicians’ and monetary bodies’ estimates of a country’s future economy. Prices of traded GDP bonds can be used to estimate future economy growth rates.

The future is always uncertain and there is no reason to think that governing bodies in enacting policies that affect a country’s future are any more knowing than a trading market. In every test of market based predictions versus consensus estimates from experts, the markets are more accurate.

Posted by MiltonRecht | Report as abusive

When Mr. Shiller refers to leverage with respect to Trills, he is indicating that the country in question has debt and therefore is considered to be “levered”, similar to a publicly traded company that has debt.

Subtracting some proxy for costs from GDP has the effect of “leveraging” Trills as follows: (i) in prosperous times, the interest on the debt will be lower b/c you can pay down your debt. This means that the dividend on the Trills will be even higher (GDP is rising ; interest cost is declining).

(ii) in recessionary times, the interest on the debt will be higher b/c you are taking on more debt. This means that the dividend on the Trills will be even lower (GDP is falling ; interest cost is rising)

Dividend on Trills = GDP – Interest Cost / 1 trillion

Posted by benjaminw | Report as abusive

Companies do issue securities with a lot of the features of GDP bonds – they’re called common stock.

Posted by realist50 | Report as abusive

Regarding your confusion on the example Shiller gave about how a dividend could be cut to avoid default. He wasn’t saying governments would issue leveraged Trills or that investors could leverage their purchase of Trills (though undoubtedly they could), his example was referring to if the entirety of the Greek economy’s balance sheet was leveraged at 5:1 i.e., $5 in bonds to every $1 in trills. Given that every government’s balance sheet is financed almost exclusively through debt, 5:1 leverage would actually be a reduction in the amount of leverage used to finance the country’s balance sheet/cumulative deficits.

So if GDP slipped -7.4%, they could cut the dividend by approximately -40% (or roughly 5x -7.4%, based on that 5:1 ratio) to preserve payments to bond holders. Bond holders could breathe easy and it would avoid being a global default circus, but people who held trills would take it on the shins. The good news is that it would hopefully leave critical fixed income investors e.g., pension funds, insurance companies unscathed because they wouldn’t own the trills anyway.

Trillholders (which, frankly, is kind of an awesome name – thrillholders would be even better!) would be holding equity, not debt, and would have to account for that kind of risk in the price they pay, i.e., their discount rates would be substantially higher. If the government 10-year bonds yield 3%, an equity holder is unlikely to use a 3% discount rate on a long-duration asset like equity. Country market risk premia in North America are running in the mid-high single digit range, so you’d likely be looking at an equity discount rate of 8% or so, my guess is people would even use higher rates if they believe the company is likely to clip the dividend to finance deficits during a recession (and most will).

The other great thing it could be used for as well is raising money to rapidly reduce debt or build rainy day funds when times are good. Have debt running at an above-average level for a sustained period of time, but showing good growth in annual GDP above-trend? – It should get reflected in a rising trill price. You could issue trills to the market to bring debt back under control or build a rainy day fund for later for a recession or an infrastructure project. It would be dilutive to the current trillholders (caps their upside on their share price) in the short run, but prevents the government from having to issue expensive capital (high-yield bonds, or low-priced trills) when times are bad. Good investors that specialize in buying long duration assets like real estate (like some REITs) will do this all the time.

The big concerns I see, in addition to the statistical issues you cite, are that a) it introduces way more complexity into government finance (finance ministers couldn’t just worry about the bond market anymore and would really have to think about their overall cost of capital); b) companies with strong reps could get away with structural deficits for longer with more financing alternatives; c) depending on how well trills are valued, it could actually ramp up the volatility in the greater stock market (as the trill would probably be seen as a good proxy for the market return, if its volatile – it will really throw everything else around much quicker); d) as always, it could potentially crowd out private investment in equity, certainly it would seem like country ETFs would be really susceptible to massive Trill issuance.

Posted by TDMFIN | Report as abusive

I am befuddled….

If Exxon wanted to issue securities that entitle the holders to a share of profits, then why wouldn’t they call it “stock”? If somehow they wanted to issue these securities without diluting voting control, they could conceivably issue multiple classes of stocks with different voting rights (I think?).

A government can’t issue equity ownership, since the government doesn’t own the economy, so this is a way to get around that issue. Otherwise, call a Trill what it is — it is effectively a stock.

Posted by TFF | Report as abusive

Also, I don’t understand how Felix comes up with an infinite valuation. A slow-growth (but essentially stable) stock with a $15 dividend would likely trade between $350 and $450 right now. High, not infinite.

Posted by TFF | Report as abusive

dude, you are nuts to suggest that Shiller introduced Trills because “it could make a fortune for Bob Shiller.”

That is *so* wrong it is not *even* wrong.

Dislike Shiller’s idea? — take it apart. But your riposte is shrill & unenlightening.

Posted by dedalus | Report as abusive

Your article should have been entitled, “Da Trill Is Gone”.

Posted by samadamsthedog | Report as abusive

In Latest Greek Bailout, Warning Signs for Europe

Posted by kaylabi | Report as abusive

No it wont be Thrill holders or Trill holders if it is Greek trills. It will be Grills!! And Belgium will have Brills, France Frills! Of course Greece and Germany can duke it out for Grills or Grrills! depending on who is more annoyed!

Since calculating GDP is fraught with measurement issues, government can issue trills based on tracking “GDP sectors”. There can be trakkr or ETF like features so investors can better understand the basis behind them. At sector level, volatility is more likely to be discerned.

Posted by ishamon | Report as abusive

From the post, the point that the expected value of a “singe [sic] Trill is actually infinite” seems off-base (or at least misleading), because the probabilities of default or significant interest rate moves are both non-zero.

Here’s an interesting video on UK perpetual gilts from Britain-should-pay-off-war-loans

Posted by wtn | Report as abusive

Felix, it seems odd to write a whole post about GDP bonds without discussing Argentina, which issued GDP warrants as a sweetener during its own restructuring. These warrants have been trading since 2005 and therefore provide an obvious reference point for any discussion on the topic.

The Argentine experience is supportive of your argument that GDP bonds are not an especially promising means of borrowing for most countries. These instruments are necessarily complex, with payouts contingent on both the level and growth rate of GDP as well as inflation (via the GDP deflator). Perhaps because of this complexity, Argentine GDP warrants have persistently traded at very cheap valuations relative to reasonable assumptions about future growth. While this has made them very appealing as an investor, it obviously reduces their utility as a source of funds in the future.

Posted by NYC_Economist | Report as abusive

I am impressed by the quality of most of the comments here. Your post is not completely useless if only because it has drawn such good commentary.

I have only two things to add to them. First, your worries about measurement error are misguided. Are these errors unbiased? In that case they will nearly completely cancel, requiring only a small convexity adjustment to compensate. Are they biased? Then a deterministic adjustment will compensate.

Second, there is no need to cower fearfully, waiting for some bold private company to blaze the trail. If the Treasury wants to see the results of a prudential experiment, it can conduct one itself just by doing a small issue. No matter how bad these things are compared to straight bonds, they just can’t be very bad in absolute terms if there aren’t very many of them.

Posted by Greycap | Report as abusive

The elasticity of tax revenue (which will be used to actually pay those dividends) to GDP is not one (and can vary quite a bit) so the idea needs refining, even if the principle is interesting.

Posted by Th.M | Report as abusive

“If the Treasury wants to see the results of a prudential experiment, it can conduct one itself just by doing a small issue.”

Not a bad idea, but you might see different dynamics with a small float than if you have a large, active market. Even with TIPS, I get the impression that the market is noticeably less liquid than with conventional Treasury bonds.

Posted by TFF | Report as abusive

I suggested something like this during the US debt ceiling showdown. The Treasury was prevented by law from issuing any additional debt, but there was no law against issuing equity, right?

Dividends are certainly attractive to investors, and would get a better offering price and subscription to the IPO. But they certainly aren’t essential. Investors would have to think about what “a claim on the future profits” of country X really means; but they would still come up with a value. The Government of X would get some operating funds, without being tied to a stream of future payments.
Any subsequent issues, of course, would dilute current holders. So countries would not be able to return to investors every year like they can with debt issues.

Also, no one has yet mentioned Gilts — perpetual obligations of the UK government. The coupon on Gilts is periodically raised by parliament.

Posted by engineer27 | Report as abusive

30-year TIPS already have some of the characteristics of bonds that pay a percentage of nominal GDP. Commenters above have already addressed some of areas where you are confused about this issue. You did get one thing right, which is that measuring nominal GDP is not an exact science. Same problem as with CPI-inflation which drives TIPS. Governments issuing GDP bonds or inflation-linked bonds have an incentive to manipulate statistics, ala Argentina.

Before we experiment with perpetual GDP bonds, I think we should see how perpetual TIPS work, since that is a more natural evolution from what we already have.

Posted by revelo | Report as abusive

“Governments issuing GDP bonds or inflation-linked bonds have an incentive to manipulate statistics, ala Argentina.”

You don’t think that happens here?

I thought about TIPS as well. Like GDP bonds, they are fundamentally pro-cyclic in their payout (they pay very little interest in a deflationary environment), yet still increase in value in a recession (due to flight to safety and falling real interest rates).

Could also argue that inflation will dominate real GDP growth over the coming decades in both the US and Europe. Technological innovation will have to fight a strong demographic headwind to make any progress.

Which brings me back to the top — does anybody SERIOUSLY believe that the official CPI represents the on-the-ground reality? Supposedly we saw just an 8.3% *cumulative* increase in the CPI from June 2007 to December 2011. My impression would be something closer to 15%.

Of course, that would make the economy look like it is in pretty bad shape. Much happier to pretend that boom times are here again…

Posted by TFF | Report as abusive

“But Bob Shiller is much more ambitious when it comes to such things”

The value of these things depends on how you want to use them, surely?
For example, I forget the details, but I seem to remember that part of John Geanakoplos’ social security plan involved this sort of bond.

Geanakoplos’ plan was to defuse the Republican claim that “your social security money doesn’t REALLY belong to you and can be taken away any time the government wants” while retaining the security of Social Security (none of the volatility of the stock market, and indexing to the general wealth of the nation). There are various additional bells and whistles to solve various specific problems, but the large scale plan, as I recall, involved “individual accounts” — ie a specific record in a ledger somewhere of what you are owed — but money spent against GDP bonds rather than in the stock market or some other financial market.

Posted by handleym | Report as abusive

““As a result, if you’re a risk-averse person who wants a perpetual US government security and your discount rate is say 3%, then the expected value of a singe Trill is actually infinite.”
This statement doesn’t make sense – discount rates should differ based on the riskiness of the cash flow they are discounting, and 3% is way too low for an equity-like instrument like a GDP bond. ”

It’s worse than that. It’s the same as valuing ecological assets — there is uncertainty in the appropriate “interest” rate, in the one case the relevant discount rate, in the other case the relevant growth rate. This uncertainty is extremely important because of the projection to infinity. If your range of possible valid values for this rate includes 0, then you get strange things happening that depend on exactly how you take your limits.

The bottom line is that, in a context of the real uncertainties of the situation, projecting all the way to t=infinity makes no sense — all that makes sense is to project as far as you are confident in projecting and then make a reasonable assumption about what happens after that. There are various reasonable assumptions one might make, but plenty of them do not value such a financial instrument as having a price of infinity.

Posted by handleym | Report as abusive

If some one needs expert view regarding running a blog then i advise him/her to pay a quick visit this weblog, Keep up the fastidious work.