A proposal for an auction-based sovereign debt restructuring mechanism

Tim Willems 17 November 2020



This year’s Nobel Memorial Prize in Economic Sciences to Paul Milgrom and Robert Wilson forms a useful reminder that auction theory has great practical relevance. The Nobel Committee highlighted applications to radio frequencies, electricity, and natural resources. There is however another potential application, to sovereign debt restructurings. 

Improving the debt restructuring process has great relevance at present, where the pandemic-induced surge in public debt is raising concerns about sustainability. This has produced calls for ‘standstills’ (Bolton et al. 2020, Gelpern et al. 2020), with the official sector initiating the Debt Service Suspension Initiative (DSSI) (suspending debt service on official-bilateral debts for eligible countries requesting it). While the DSSI has helped (Lang et al. 2020), it might not suffice for the most indebted/affected countries who may need to approach creditors to obtain further relief. 

Unfortunately, the current negotiation-based debt restructuring process is not efficient. First, the process is complex and time-consuming, typically taking years to resolve (IMF 2020a) at great cost to all parties. Second, human actors are unlikely to spot (and realise) all gains from trade. Consequently, negotiation-based outcomes won’t typically be located at the efficient frontier, meaning that both debtor and creditors could be made better off. Within creditors, some are also likely to be better served by any negotiated deal than others—harming inter-creditor equity. 

Deploying an auction mechanism can improve upon negotiated outcomes. Auctions are an effective way of managing many players, perfectly able to spot (and realise) efficiency gains. This should appeal to both debtors and creditors, as this can benefit all parties involved (with much less drama and delay). While negotiations become more complex as the number of players grows, auctions converge to the optimal outcome.

The proposal starts by noting that an efficient debt restructuring should optimally ‘fit’ debtor/creditor preferences. Any debt relief that is deemed necessary to restore sustainability should take a form that is least painful to the country’s creditors. Such design allows the country to obtain the greatest amount of relief, while inflicting the lowest possible damage upon creditors (thus preserving creditor support). 

Creditor preferences are likely to display significant heterogeneity, particularly regarding the trade-off between granting relief through face value haircuts, or via maturity extensions.1 Different creditors can have different preferences for various reasons: they can have a different outlook on the country, while balance sheet considerations may be important too. Each creditor’s regulatory/tax environment can also affect the appeal of the various options. An optimal restructuring should allow each creditor to self-select into those options that fit their situation best, as that will maximise the relief a debtor can find subject to maintaining creditor support – hence restructurings typically offer a menu to creditors (Van Wijnbergen 1991). 

Reflecting all these factors – which differ across creditors, of which there are typically thousands – in a negotiation is impossible and better handled through a price mechanism. A simple example might prove insightful. Imagine a firm with 100 workers, needing to dismiss five. Those five will receive severance pay, but how to determine whom should leave? While a negotiation-based approach is likely to be chaotic and lengthy (leaving many bitter about the result ex post), a price-based strategy would ask workers only one question: At what level of compensation would you leave your job voluntarily? Workers who hate their job and were thinking of quitting anyway, will file a much lower price than content employees. By bidding in that way: 

  • The unhappy workers can release themselves from their much-hated job while being compensated for it 
  • The firm gets to restructure its employee base at the lowest-possible cost.

An auction-based strategy, which follows price-based corporate resolution proposals (Bebchuk 1988, Aghion et al. 1992, Hausch and Ramachandran 2000), can generate restructuring outcomes with similar efficiency-properties. But in order to determine the optimal shape of the restructuring, one first needs to learn about creditor-preferences (each creditor’s relative willingness for granting relief via a maturity extension versus a face value haircut). Such information can be solicited through Klemperer’s Product-Mix Auction (PMA) (Klemperer 2010). The PMA offers a single-round method to sell goods which are considered imperfect substitutes. It allows bidders to submit multiple, mutually exclusive bids – enabling them to approximate their demand curve.2

The PMA can be applied to sovereign debt restructurings via the following sequential process: 

  1. Each creditor enters the auction with budgets proportional to the market value of their existing claims. 
  2. Like in the current framework (Buchheit et al. 2019, IMF 2020b), the process starts by determining a sustainable debt service-profile. Suppose that this leads to zero debt service obligations for 2021, $1 billion in debt service over 2022, $3 billion over 2023, and so forth (say, up to 2039). For simplicity, one can think of zero-coupon bonds. Given a discount rate, this step determines the amount of NPV-relief embedded in the restructured profile. The question that the next step will answer is: how to optimally distribute this aggregate NPV reduction over creditors?
  3. Creditors bid for the various restructured bonds announced in Step 2. They do so by informing the auctioneer about their yield curve via bids like: “I bid my budget on {the bonds due in 2022 OR 104% of that amount for the bonds due in 2023 OR 110% of that amount for the bonds due in 2024 OR etc.}.3 Creditors with a pessimistic outlook (or other reasons to prefer quick repayment) will specify a steeper yield curve, which increases their odds of obtaining a shorter-dated claim (but at a larger haircut). Creditors with a more optimistic outlook (or other reasons to prefer maturity extensions) will specify a flatter yield curve; they are fine with a longer-dated claim, which enables them to avoid a large haircut. Haircuts are determined endogenously to ensure market clearing; haircuts will be such that creditor preferences end up being consistent with the pre-specified repayment profile (making sure that all restructured holdings ‘fit’ with the schedule announced under Step 2). Under competitive bidding, the auction allocates each creditor a combination of restructured bonds which they would have chosen voluntarily given the haircuts that ultimately materialise.

The proposed mechanism has several intuitive and desirable properties: 

  • Logical consistency. Creditors who are pessimistic on the country’s future will want to be repaid soon, the odds of which can be increased by bidding a steeper yield curve – bringing larger haircuts to shorter-dated claims. This is consistent with the bearish outlook on the debtor, as the logical implication of that view is that debt relief is needed. More optimistic creditors (thinking the country doesn’t need much debt relief) will bid a flatter yield curve – accepting that they will likely end up with longer-dated claims. But given the lower haircut, and given their optimistic outlook, they are still happy with this outcome.
  • Efficient sorting. The auction allows creditors to self-select into that restructured instrument to which they attach most relative value (be that for reasons relating to outlook, balance sheet, or regulatory/tax implications). This feature enables the debtor to maximise the relief it can obtain, as the auction shapes the losses (embedded in the repayment profile specified under Step 2) in a way that is least painful to creditors. 
  • Voluntariness. By relying on a price-mechanism to tailor the restructured debt stock to creditor-preferences, no creditor will be forced into a deal. Creditors wishing to take a haircut (maturity extension), can get their haircut (maturity extension). Creditors who end up with a restructured claim they favoured less a priori, will be compensated by a more favourable price (ensuring they will willingly hold their restructured position).
  • Inter-creditor equity. In the resulting allocation, no creditor will prefer the allocation of any other creditor over its own.4 This makes the mechanism reflect inter-creditor equity, which should help generate creditor support. 

An auction-based resolution may also weaken the ‘holdout problem’ (creditors trying to free-ride on each other’s contributions towards restoring sustainability). Creditors know that a restructuring will only go ahead if a sufficient majority of bondholders agrees to the proposed terms.5 But if other creditors are willing to provide debt relief, they must have information that doing so is indeed necessary and beneficial. This converse to the ‘winner’s curse’ makes creditors realise that any relief they offer, will only come into effect if enough other creditors agree that providing relief is the correct course of action. This makes creditors more willing to contribute to the public good of debt reduction and decreases the likelihood of creditor coordination issues getting in the way of a deal (Detragiache and Garella 1996).

The PMA can also be enriched to put state-contingent debt on offer in the restructuring (alongside conventional bonds). This may help generate creditor support in the face of creditors who believe that the pre-specified repayment profile is providing too much relief. Such creditors should be attracted to GDP-linked bonds, since those can be expected to flourish following superfluous debt relief.6 Other types of bonds (green, Islamic, domestic, etc.) could be put on offer as well, giving the debtor more options to tailor the post-restructuring debt stock to creditor-preferences – further raising prospects of reaching agreement.7 

An auction-based process is likely to improve upon the status quo, particularly since the latter is not well suited to handle today’s dispersed creditor base. By responding to individual creditor-preferences, an auction will be able to serve creditor-interests better, minimising costs to the debtor. Although an auction-based resolution will not solve all difficulties (parties still need to agree on a repayment profile under ‘Step 2’), creditors might become more willing to accept any repayment-proposal if they know that it will be followed by a price-based allocation mechanism (which will yield an outcome they would have chosen voluntarily, had they seen the terms in advance). Ultimately, all parties will benefit from restructured claims taking a form and allocation that is least painful to creditors, as that distributes the losses which need to be incurred to restore sustainability in the most efficient way. Herewith, the current proposal should help mobilise creditor support. The entire process can furthermore be extremely quick and cheap to implement, again benefitting both debtors and creditors. 

Note: This proposal greatly benefited from many discussions with Paul Klemperer, who provided numerous detailed and essential inputs. Further thanks to Craig Beaumont, Wolfgang Bergthaler, Lee Buchheit, Jeremy Bulow, Marcos Chamon, Enrica Detragiache, Mitu Gulati, Adnan Mazarei, Eriko Togo, Felix Vardy, Sweder van Wijnbergen, and Jeromin Zettelmeyer for useful comments. The views expressed here are those of the author and should not be attributed to the International Monetary Fund, its Executive Board, or its management. Any errors are my own.


Aghion, P, O Hart and J Moore (1992), “The Economics of Bankruptcy Reform”, in: Olivier J. Blanchard, Kenneth A. Froot, and Jeffrey D. Sachs (eds), The Transition in Eastern Europe, Chicago: University of Chicago Press. 

Bebchuk, L (1988), “A New Approach to Corporate Reorganization”, Harvard Law Review 101: 775-804. 

Bolton, P, L Buchheit , P-O Gourinchas, M Gulati, C-T Hsieh, U Panizza and B Weder di Mauro (2020), “Necessity is the Mother of Invention: How to Implement a Comprehensive Debt Standstill for COVID-19 in Low- and Middle-income Countries”, VoxEU.org, 21 April. 

Buchheit, L, G Chabert, C DeLong and J Zettelmeyer (2019), “The Restructuring Process”, in: S. Ali Abbas, Alex Pienkowski, and Kenneth S. Rogoff (eds), Sovereign Debt: A Guide for Economists and Practitioners, Oxford: Oxford University Press. 

Das, U S, M G Papaioannou and C Trebesch (2012), "Sovereign Debt Restructurings 1950–2010: Literature Survey, Data, and Stylized Facts”, IMF Working Paper No. 12/203. 

Detragiache, E and P G Garella (1996), “Debt Restructuring with Multiple Creditors and the Role of Exchange Offers”, Journal of Financial Intermediation 5, 305–336. 

Gelpern, A, S Hagan and A Mazarei (2020), “Debt Standstills Can Help Vulnerable Governments Manage the COVID-19 Crisis”, in: Maurice Obstfeld and Adam Posen (eds), How the G20 Can Hasten Recovery from COVID-19, PIIE Briefing 20-1. 

Hausch, D and S Ramachandran (2000), “Corporate Debt Restructuring: Auctions Speak Louder Than Words”, in: Stijn Claessens, Simeon Djankov, and Ashoka Mody (eds), Resolution of Financial Distress: An International Perspective on the Design of Bankruptcy Laws, World Bank Institute, 91-106.

IMF (2020a), “The International Architecture for Resolving Sovereign Debt Involving Private-Sector Creditors”. 

IMF (2020b), “Argentina: Technical Assistance Report-Staff Technical Note on Public Debt Sustainability”. 

Klemperer, P (2009), “Central Bank Liquidity and “Toxic Asset” Auctions”, VoxEU.org, 25 September. 

Klemperer, P (2010), “The Product-Mix Auction: A New Auction Design for Differentiated Goods”, Journal of the European Economic Association 8: 526-36. 

Lang, V, D Mihalyi and A Presbitero (2020), “Borrowing Costs After Debt Relief”, VoxEU.org, 14 October. 

Van Wijnbergen, S (1991), “Mexico and the Brady Plan”, Economic Policy 6: 13-56.


1 Das et al. (2012) report that, in general, retail investors tend to prefer maturity extensions, while institutional investors prefer to provide relief through face value haircuts (but at shorter maturities). 

2 The PMA was originally designed for the Bank of England in 2007/8 (Klemperer 2009). During the GFC, the BoE wanted to provide liquidity against a widened range of collateral, raising the question how the interest rate should depend on collateral quality. Banks ideally wished to obtain liquidity against weak collateral (like mortgage-backed securities) but could be enticed to put up strong collateral at a sufficient discount (lower interest rate). The PMA enables banks to submit bids like “I wish to receive £1 billion {at 5% against weak collateral OR at 4% against strong collateral}”. Similarly, the BoE specifies how its willingness to provide liquidity against weak collateral depends on the premium (higher interest rate) it comes with. The auction then determines how much the BoE should lend against weak/strong collateral, and at what interest rates.

3 The PMA can also handle fractional bids: “I wish to bid 60% of my budget on {the bonds due in 2022 OR 104% of that etc.}; for the remaining 40% of my budget, I wish to buy {the bonds due in 2030 OR 107% of that etc.}” 

4 Imagine a country with a 5% discount rate. However, for whatever reason, investor A is applying a 0% discount rate to this country—thus bidding a flat yield curve. As a result, the restructuring will boil down to a pure maturity extension for A (no face value reduction). Note that, given its 0% discount rate, A doesn’t feel an NPV reduction following a maturity extension. But the country (discounting at 5%) does! Moreover, no other investor (all applying positive discount rates) will look at A’s deal and think of it as attractive relative to their own; still A is happy, considering themselves as the only creditor to have avoided an NPV reduction in the restructuring (while all others saw face value haircuts). 

5 These days, many sovereign bond contracts contain Collective Action Clauses (CACs) through which all bondholders agree ex ante to accept restructuring terms if they are endorsed by a sufficient supermajority. The underlying supermajority-threshold (often two-thirds or three-quarters) acts like a minimum tendering requirement. 

6 Given the implementation difficulties associated with GDP-linked instruments, offering local currency bonds instead could be an alternative. If the country recovers strongly from its crisis, the USD-value of local currency bonds is likely to rally (e.g. on the back of an appreciating local currency)—thus proxying the payoff structure of GDP-linked instruments, while avoiding its practical complications (such as statistical challenges associated with accurate and timely GDP measurement).

7 The PMA can also allow the debtor to submit its own preferences expressing its willingness to substitute among such options, finding the efficient solution given all creditors’ (and the debtor’s) bids—analogously to the process in footnote 1.



Topics:  Covid-19 International finance

Tags:  debt restructuring, auctions, sovereign debt, COVID-19

Economist, IMF


CEPR Policy Research