GDP-linked Bonds: Can Argentina's Failure Become Greece's Success?
The new Greek government has floated the idea of swapping the official Greek debt that is now held by EU countries and the European Stability Mechanism (ESM) with GDP-indexed bonds. The proposal has received support from some in the economics profession in various forums, for example here.
But if precedent is any guide, the idea that Greece should be issuing GDP-indexed bonds should be treated cautiously. GDP-indexed bonds have the sensible theoretical rationale of making payments proportional to, or even contingent on, the country's capacity to pay. This reduces the probability of a costly default and smoothes out the debt burden for the issuing country, to the benefit of the country's taxpayers.
However, Argentina's GDP-indexed warrants, the only relevant real world experience with this type of instrument, proved very costly for that country. In this note we describe the details of the Argentinean experience, the lessons for Greece, and an alternative strategy to restructure Greece's official loans. We propose that the current European Union and ESM loans be modified so that Greece will be able to pause servicing them if growth falls below a predetermined threshold. Such a step would accomplish most of the benefits of GDP-indexed bonds while avoiding most of their pitfalls.
There are several problems associated with GDP-indexed bonds. First, GDP-linked bonds introduce the temptation by the issuing country to alter the GDP statistics to reduce payments. In addition, a country that knows that its growth determines its payments may seek to grow less. These two factors add to the risk premium imposed on the GDP-indexed bonds, increasing their cost and reducing the savings associated with them.
For example, Argentina's inflation and national accounts statistics have been dubious for the past eight years. Inflation has been significantly understated for political reasons (low inflation is "good"), but real GDP growth has been overstated (slow growth is "bad"). Thus, short-term political incentives to exaggerate growth have led Argentina to overpay on its debt. It paid the 2008 coupon, although later revisions to national accounts data showed that its trigger, the real growth rate in 2007, had not exceeded the threshold.
An even more important problem with GDP-indexed bonds is their financial structure. A GDP-indexed bond can be designed in many ways. It can index the nominal value of the security (as in Shiller's growth-linked bonds), or it can index its return, more in line with the idea of a contingent, pay-in-good-times contract. Or it can appear on both components of the bond, as in the Argentine case, which included a nominal GDP-indexed payment contingent on exceeding a real growth threshold. The more complex the design, the more difficult to price and the lower the value.
By reducing the debt service in bad times (and the reverse), GDP growth risk is transferred from the debtor country to the creditor, who in turn needs to hedge it, or pool it under some particular investment benchmark. The problem is that growth risk is exotic, difficult to price and hedge properly (no financial instrument has a high correlation to real growth) and therefore costly for investors.1 As a result, the investor universe for these bonds is narrow and highly speculative—which, again, adds to the cost of the debt.
The Argentinean case is a good example of the potential downside of this complex financial engineering. In June 2005, at the time of the debt exchange, when the GDP-indexed warrants were issued, pricing models and consensus expectations pointed at a fair value for the warrants of roughly 4 cents. At issuance, however, the market priced the instrument with a large discount; indeed, when the detached warrant started trading on its own six months later, Argentina's sovereign spread had already compressed by about 400 basis points, but the warrants were worth just 2 cents. In other words, the country paid at least a hefty 50 percent discount for the warrant, well beyond the 12 percent exit yield (post exchange) of the sovereign bonds. This fiscal cost outweighed any smoothing benefit offered by the warrant, casting doubt on the design of the operation.
All these considerations must be applied to Greece. The Argentinean experience suggests that if Greece were to issue a GDP-indexed bond, the market would undervalue it because of its exotic nature. Greece would overpay as Argentina did. A recent post by PIIE Senior Fellow Paolo Mauro highlighted statistical issues (how to ensure that statistical offender Greece does not make up growth figures), but disregarded valuation problems, arguing that valuation issues would be irrelevant because Greece would not be issuing a market-traded bond but a GDP-indexed loan to an official creditor.
That risk aversion may well be lower for an official creditor, but valuation should still be the relevant concern and could become an obstacle if the idea turns into a real exchange proposal. The loans to Greece are large and have fiscal implications for the creditor countries. In the Argentine case the warrant was detached so that it could trade independently and the bond could be easier to price separately. How would GDP-indexed bonds be valued and booked by a European government, or the ESM?
Given the long Greek history of tampering with the national accounts, the very large forecasting errors in Greek GDP in the last few years, and the tremendous uncertainty about Greece's potential growth after the shock to its economy, any valuation based on the future GDP path will be subject to significant error. Were these bonds to be adopted, European countries would be well advised to demand that the Greek statistical office be put under European management to minimize the data risk, something that would carry obvious political cost.
The interest rate of the European loans to Greece is already very low. The gains from indexing future interest rate payments to the future growth path will therefore be quite small. In fact, given that a large portion of Greece's official debt is indexed to the 3-month Euribor (Euro Interbank Offered Rate), and that Greece should, under most scenarios, grow faster than the euro area for the foreseeable future, it is not clear that indexing interest payments to Greek growth would have a positive expected value (as the 3-month Euribor is a function of euro area growth and thus likely to increase much less).
A much simpler intermediate option would be a clause that automatically postpones by two years (the expected duration of a "normal" recession) the debt service on current official loans when year-on-year GDP growth in Greece becomes negative (or falls below some threshold that is clearly below potential growth, say 1 percent).2 This structure would achieve three objectives: (1) it would preserve the face value of the loans and make them easier to value than a complex GDP-indexed bond, because a potential rescheduling would only marginally affect the net present value of the loan (akin to increasing the maturity of the debt in expected terms based on the probability of recessions over the maturity of the loans) but not affect the nominal amounts paid; (2) it would reduce the procyclicality of fiscal policy by automatically creating fiscal space for Greece when the next recession hits, while eliminating most of the moral hazard concerns; and (3) it would reinforce the message that servicing official debt will not be an impediment for Greece's ability to service private debt. This combination could make the arrangement politically palatable and economically effective while facilitating Greece's return to markets.
This alternative, while preferable to Argentina-style GDP-indexed loans, is not a substitute for a more balanced policy mix and a stable political outlook that can set Greece into a sustainable growth path, as suggested in Ubide (2015).
Angel Ubide is senior fellow at the Peterson Institute for International Economics and codirector of global economics for the D. E. Shaw Group. Eduardo Levy Yeyati is visiting professor at the Harvard Kennedy School of Government and president of the Center for the Implementation of Public Policies Promoting Equity and Growth (CIPPEC) in Buenos Aires, Argentina.
1. For example, in the case of Argentina, the bond returns were a function of growth and real exchange rate risk, a combination almost impossible to price with any confidence. This generated a very large risk premium and very low market value.
2. Because the GDP statistics are always subject to revisions, an alternative would be to create a business cycle dating committee, similar to the National Bureau of Economic Research's committee for the United States or the Center for Economic and Policy Research's for the euro area, that would determine, based on high frequency data, if Greece was in recession.