Sovereign Debt Restructuring

Sovereign Debt Restructuring:

The Role of Institutions for Collective Action[1]

Richard Portes

London Business School, Haas School of Business, and CEPR

March 2000

The problem

‘Most Directors agreed that there would be considerable benefits from the introduction of collective action provisions in new bond contracts…Many Directors felt that this could best be achieved if industrial countries included such terms in their own bond issues. In addition, provisions authorizing a trustee to negotiate with the debtor on behalf of bondholders, but without authorizing the trustee to legally bind them to any agreement, could also contribute to an orderly and speedy restructuring process.’

(International Monetary Fund, 1999, para. 30)

‘Creditors’ committees could have a role to play in effectively resolving financial crises. While the creation of a single standing committee was generally not considered practical, consideration could be given to ad hoc arrangements in appropriate cases…’

(International Monetary Fund, 1999, para. 33)

These recent statements by the Managing Director of the IMF suggest that the argument is over. The proposals of a framework for orderly workouts of sovereign debt put by Eichengreen and Portes (1995), Macmillan (1997) and others have essentially been accepted. All that remains is implementation.

I shall maintain that this is unfortunately not the case. Market participants – the lenders – still vigorously oppose any official action along these lines. Their opposition is even stronger when it comes to official sanction of payments standstills and IMF lending into arrears, other measures that might be used as part of an orderly workout procedure (see further discussion in para. 33 of IMF 1999). More broadly, market participants and the US Treasury appear to reject any rule-based procedures and any effective action by the official community to bring about collective action clauses and creditors’ committees.

The banks and securities houses, as represented by the Institute of International Finance, continue to argue against measures that they interpret as making it easier for debtors to default, as well as any official intervention into creditor-debtor relationships – except ‘relatively large but temporary official support’ (IIF 1999, p. 7), i.e. IMF-led bailouts.

In 1996 they supported ‘the market-based approach [with] the following key characteristics:

case-by-case resolution
direct contact between debtor and creditors as necessary rather than official intervention through negotiations directed by bodies such as the IMF…
normal levels of public support through operations of the IMF and other multilaterals
if arrears have accumulated, delay of IMF disbursements until agreements in principle are reached with creditors on a schedule for elimination of arrears.

It should also be presumed that if contract revision becomes an issue, it will be on the basis of rescheduling rather than forgiveness, and with no interruption in interest payments.’

(Institute for International Finance, 1996, pp. ii-iii)

Their view has been unaffected by the subsequent financial crises in Asia, Russia, Brazil, and debt servicing difficulties elsewhere. Their 1999 report argues that ‘crisis resolution based on restoring private sector confidence…(the 1990s approach) seems both more feasible and more conducive to maintenance of the robust capital market of the 1990s than alternative approaches emphasizing standstills, reschedulings, and concerted lending (the 1980s approach).’ (IIF, 1999, p. 7)

The principle evoked to justify this stance is that debtors repay only when the pain caused by default is unacceptable (loss of trade credit and capital market access, legal harassment, etc.). Thus anything that might mitigate these consequences would shift the supply curve of funds to emerging markets downwards (to the left) and the demand curve upwards, as borrowers perceive less likely damage if things go wrong, and lenders understand that reaction. The price (interest rate) would rise, and the effect on the ‘volume of capital flows is ambiguous but would seem more likely to fall…than to rise.’ (IIF 1996, p. 30).

Ignoring the totally unsubstantiated assertion about the volume of flows, the underlying assumption may be wrong. As Rogoff (1999) points out, there are alternatives: the debtor may be concerned only with its reputation for repayment, or more broadly, its standing in the international economic community. Those concerns would be unaffected by measures that would facilitate debt restructuring where necessary and make it less damaging to the debtor country’s trade and growth as well as its trade partners. So the borrowers’ demand might not rise. Moreover, such measures might in fact make creditors as a class better off by overcoming the coordination failures involved in crises[2], just as bankruptcy codes do in cases of domestic illiquidity or insolvency. In that case, the effect on the supply of lending would be positive rather than negative. Thus the effects of, for example, collective action clauses on market terms is an empirical question. I return below to some empirical evidence.

First, however, I look at broad evidence on how lenders’ perceptions of the likelihood of crisis and default, as well as their assessment of the authorities’ determination to achieve ‘bailins’, have been affected by recent events. Here I find that just as after the Mexican crisis of 1994-95, the international markets have been ‘resilient’ and show relatively little sign of concern that a fresh wave of debt-servicing problems might lie ahead. This suggests either very short memory or moral hazard effects of the big bailout packages.

But there will always be financial crises, both national and international, no matter how much effort is devoted to prevention. Yet the bailout packages cannot and should not be repeated. And the record on ‘bailing in’ is so far very unsatisfactory (Eichengreen and Rühl, 2000). What is to be done? This is the underlying issue in the debate on the future role of the IMF (Council on Foreign Relations, 1999; Portes, 2000; International Financial Institutions Advisory Commission, 2000). The official community is concerned to bail in the private sector, to achieve a degree of 'burden-sharing' that would be politically defensible and would discourage irresponsible lending. Yet so far it has not succeeded in establishing mechanisms that work.

It will not be feasible – economically or politically – to cut back Fund bailouts and moral-hazard-creating activities without an adequate substitute. The historical evidence that I shall discuss shows clearly the need for official intervention and appropriate institutions to deal with international debt problems. Indeed, to get the Fund out of the big bailout package business, there has to be an alternative that will not involve as much pain and suffering as debt default does today or did in the pre-IMF period.

That alternative is an orderly workout negotiated between creditors and the debtor. But organizing such workouts efficiently requires standing committees of creditors, which did exist in the earlier period of bond finance, as well as collective action clauses in debt contracts. These should provide for collective representation of creditors and qualified majority voting, so that no rogue creditor can block a settlement; for sharing debt service pro rata among creditors, so no creditor could derive advantage over the others from taking the debtor to court; and eliminating acceleration provisions that require immediate repayment of all debts if there is default on a single scheduled payment. These clauses were not common in the earlier period, but there is ample subsequent experience with them. Together with an appropriate negotiating framework, they could substantially ease the pain of default for both debtor and creditors, while not unduly encouraging this outcome. Bailing in and burden-sharing are the language of a zero-sum game. We can do better.

We cannot sensibly restrict the role of the Fund without structures that can replace it. The historical evidence – and developments since the Mexican crisis – indicate that the markets will not spontaneously achieve these outcomes. The equivalent of law is required, although not the kind of law that would be needed to establish an international bankruptcy court.[3],[4]

I discuss these institutional changes and deal with the objections that have been raised against them. Finally, I argue that the official sector will have to take a much more assertive stance in order to implement these measures.

Market reactions

Investors should take risks, and risks mean that some proportion of investments will fail. This is why we have domestic bankruptcy laws that provide for orderly workouts when investments and the firms that made them do fail. Similarly, if the international capital markets are functioning well, mistakes will be made. Then one or another country (or its private sector borrowers) will fail: a crisis. History records many. We should not expect or even wish to prevent them all. That would be at the cost of insufficient, excessively risk-averse investment. So such crises will always recur - but capital markets forget.

What did the markets learn from the Mexican crisis of 1994-95? Spreads above comparable US Treasury instruments for Eurobonds of emerging market economies fell considerably during the period mid-1995 to mid-1997 – but no more than spreads for high-yield US corporate bonds (Cline and Barnes, 1997). So it is hard to argue on that evidence that the markets became more circumspect about emerging market lending (or that the Mexican bailout simply encouraged them). The volume of lending, too, does not show any negative reaction (Figure 1).

Did the Asian crisis and other subsequent debt-servicing difficulties have noticeable effects on the willingness to lend to emerging markets? Market participants might have concluded that bailouts on the 1997-98 scale would be infeasible in future, because either the resources or the political support would not be forthcoming; or they might have taken fright at the expressed determination of the official community to achieve 'private sector bailins' and 'burden-sharing'.

It is again hard to see in the aggregate data any evidence for this perception. The volume of gross flows to emerging countries in 1998 was not substantially below that of 1996 and was greater than any previous year, as are the bank loan and bond investment components of the total. The Emerging Markets Bond Index spread jumped sharply in 1998Q3 and by 1999Q1 was at its 1995 previous peak, but it then fell by 300 basis points, despite the problems of Ecuador, Pakistan, Ukraine and Romania. And this index mainly tracks Brady bonds, which are not representative, according to Kamin and von Kleist (1999). Their calculations for a wider range of issues, disaggregated by credit rating category, are shown in Figure 2. The spread on AA issues is above its 1997Q1 low, but still below the 1992 level, and that for BBB issues is also not dramatically higher. All spreads are down from their 1999Q1 peak. Note also that the emerging market spreads have been moving substantially in parallel to those on US corporate bonds (Figure 3). And finally, the performance of emerging market stock indices in 1999 suggests that investors have gone back in with enthusiasm (Figure 4).

Market participants seem to have been misled by their perception of the success of the Mexican bailout in restoring market access. They asserted in September 1996 that there would be no need for ‘extraordinary assistance’ to limit contagion in future potential crises (IIF 1996, pp. 12, 35). They were not much better at forecasting in January 1999, when they argued that ‘the risk of unwarranted retreat from emerging markets finance at least temporarily overshadows any contrary risk of prospective overlending…’, and they expressed ‘concern that a major case of…disruption of debt servicing could impose a severe shock on emerging capital markets globally’ (IIF 1999, pp. 10, 50). As we saw, the peak of spreads was precisely in 1999Q1, and the problem countries cited above have apparently not created significant spillovers.

Bailouts and bailins

The 1980s debt strategy did not include the large bailout packages that we have seen in the 1990s. The 1990s strategy for dealing with debt crises has involved commitments to provide $34 bn in support of Thailand, $40 bn for Indonesia, $57 bn for Korea, and $42 bn for Brazil, not to mention the short-lived commitment of $22 bn to Russia. These amounts range from 50% to 130% (Korea) of annual export revenues, from 6% (Brazil) to 30% (Indonesia) of GDP, and up to 19 times IMF quota (for Korea).

At the very least, this magnitude of IMF-led intervention is politically unsustainable, and we hear rising calls to cut the Fund back or even abolish it. The resources may not be there - whether because G-7 governments will not be able to provide the 'second-line' funding behind the IMF allocations, or because the size of package necessary the next time around may be yet another order of magnitude higher. With full capital account convertibility, and in a world of free capital mobility, bailout lending can never be big enough itself to stem a panic - residents and speculators can short the currency up to the full, IMF-augmented value of reserves, and indeed all of M2 can be exported by residents.

Nor should these large bailouts be continued. There has been much debate about moral hazard, the undesirable effect of insurance on the incentives of those insured. Much of that, in my view, is ‘hazardous moralizing’ - as when market participants solemnly talk about the ‘sanctity of contracts’ or the dangers of contractual changes that would, they say, encourage debtors to default. They are unwilling to concede that lenders may regard official support in crises as an important safety net and that this moral hazard may lead to overlending.[5]

It is difficult to identify moral hazard effects empirically. In my own view, bailouts have indeed reduced the perceived risks in international lending and have thereby encouraged excessive risk-taking. Investor moral hazard has been more important in recent years than borrower moral hazard. The danger now, however, is that my (long-standing) view is shared by a wide range of opponents of bailouts and by extension, of the IMF itself. But all insurance generates moral hazard, yet the existence of insurance doesn’t destabilize markets (and actuarially fair insurance does not involve subsidies). We simply need a better balance between the costs of moral hazard and the benefits of intervention that may create it.

Yet efforts to limit moral hazard and achieve burden-sharing by bailing in private-sector creditors have so far been unsuccessful. Eichengreen and Rühl (2000) discuss the cases of Pakistan, Ecuador, Romania, and Ukraine. All involve bonds rather than bank debt and sovereign rather than private-sector debt. They criticize the IMF-World Bank strategy, arguing that attempts to condition official assistance on 'private-sector participation' does not change the payoffs to the private sector, nor does it change their set of possible strategies - and hence it cannot affect the equilibrium outcome of the game between the private sector and the IFIs.

Moreover, since these are not economically big countries with large outstanding debt, their debt problems have no systemic implications. Thus there was never any opportunity for their cases to set credible precedents that could affect lender behavior with systemically important countries and thereby reduce moral hazard significantly. Instead, the IFIs committed themselves to time-inconsistent policies - that they would stand aside, despite the costs of default, if the markets did not 'participate' in a resolution - and since these policies were correctly perceived by the markets to be time-inconsistent, they were not credible.

The historical evidence[6]

The evidence most relevant to the issue of how to manage debt crises without bailouts may come from the 50 or so years preceding World War I. This was an era of financial globalization, in many respects not unlike the present day. Large portfolio capital flows were intermediated by the bond market, as has has been the trend since the early 1990s. Sovereigns and other borrowers experienced crises, halted payments, and restructured their debts. There was no IMF involved in the process.

Initially, bondholders formed ad hoc committees to represent their interests. The committee would negotiate the best deal it could and recommend to the bondholders with whom it was in contact that they accept its terms. If a sufficient majority did so, then the loan was restructured. Generally, committees insisted on full payment of interest arrears but might agree to some write-down of principal and stretch out of interest. There was no analogue to IMF conditionality: the committee could not insist that the country go back on the gold standard or adopt other reforms, although it might be more inclined to accept the settlement offer of a government that seemed so disposed.[7]

This approach had shortcomings. Ad hoc committees had high administrative costs. Their organizers found it difficult to establish their credentials with investors. A committee that could not claim adequate investor participation had difficulty getting the debtor to take it seriously in negotiations, since it was unlikely to induce adequate participation in the conversion. Insiders sometimes manipulated committees, either by using them to cut a deal that worked to their advantage or by exaggerating administrative costs and skimming the proceeds of membership subscriptions.