key: cord-0049266-uit0nm20 authors: Arnold, Theresa; Gulati, Mitu; Panizza, Ugo title: How to restructure Euro area sovereign debt in the era of Covid-19 date: 2020-08-12 journal: nan DOI: 10.1093/cmlj/kmaa015 sha: 9ff151c8bf4fd06a95bf6d972217282b5090aef0 doc_id: 49266 cord_uid: uit0nm20 nan facing holdout issues on the local-law debt. This was also the case with Barbados' debt restructuring in 2018, which followed the Greek template. However, both Greece and Barbados faced significant holdout problems in their foreign-law-governed debt; many of the foreign-law creditors refused to take the same offers that were made to the local-law creditors and ended up receiving higher recoveries. 3 In the analysis that follows, we focus on three strategies a euro area sovereign might utilize in a possible future crisis to solve the primary source of disruption to modern day sovereign debt restructurings: the holdout creditor problem. 4 Therefore, in the analysis that follows, we assume that the sovereign and its official sector backers have been able to negotiate a restructuring deal with the majority of creditors, but those creditors do not want to enter into an agreement where either their recoveries are at risk of disruption from litigation by holdout creditors or they are in danger of being embarrassed by having suffered much bigger haircuts than the holdouts. We also put to the side the possibility of legal changes at the broader European level that might ameliorate the holdout problem more directly, 5 and abstract from the possibility of reforms that will create a clear seniority structure within euro area bonds. 6 Although the analysis that follows might be applied to other euro area sovereigns, we look specifically at Italy for the purposes of our analysis. By using Italy as an example, we do not mean to suggest that it needs to restructure its debt pre-emptively. 7 However, Italy did face a confidence crisis in late 2011 and the coronavirus has hit it particularly hard in 2020. 8 Italy's gargantuan debt stock of around e2.4 trillion and a debt to GDP ratio of 135 per cent make Italy vulnerable to another confidence crisis. 9 The economic recession and ). An alternative proposal which is fiscally neutral but could reduce the likelihood of rollover risk is the PADRE plan. See Pierre Pâris and Charles Wyplosz, PADRE: Politically Acceptable Debt Restructuring in the Eurozone, Geneva Special Report on the World Economy 3, ICMB and CEPR (2014) . 7 For a discussion of the risks associated with a pre-emptive Italian debt restructuring, see Giampaolo Galli, Why Restructuring of the Italian Public Debt Should be Avoided (2018) (Osservatorio CPI ); Ugo Panizza, 'Come Risolvere il Problema del Debito Pubblico Italiano: Un'analisi Critica Delle Soluzioni Facili' (forthcoming 2019) Rivista di Politica Economica. 8 For a prior discussion by one of us, see Ugo Paniza, Public Debt in Italy: Myths, Facts and Policies (2014 draft) 9 Ferdinando Giugliano, The Euro's $2.7 Trillion Dollar Problem, Bloomberg, April 4 (2019) accessed 10 July 2020; Peter S Goodman, 'Italy's Biggest Economic Problem? It's Still Italy', NY Times, August 9 (2019), Theresa Arnold, Mitu Gulati and Ugo Panizza • Euro area sovereign debt in the era of ballooning expenditure needs associated with Covid-19 are expected to bring the debt-to-GDP ratio well above 150 per cent. 10 Quite possibly, such a crisis could be dealt with through some combination of fiscal adjustment, economic reform and official (ESM/ OMT) financial support. This said, it is also conceivable that such a crisis might lead to a situation in which Italy might have to restructure its debts. We thus ask two related questions: in such a situation, what restructuring options might Italy have? And could Italy do something now to put itself in a better position to deal with such a situation in the future? Two considerations for any government in undertaking a debt restructuring are: (i) how to minimize the impact of that restructuring on its costs of future borrowing in situations where the sovereign needs to be able to tap those markets regularly; and (ii) how to minimize the domestic economic and political costs in situations where a large portion of the debt is held domestically. The balance of the foregoing considerations, along with the interests of official sector sponsors, will determine how the choice among restructuring options is made. We do not attempt the exercise of prescribing a specific restructuring plan for Italy. Instead, we put what we see as the most plausible options on the table. The options to prepare for a possible future debt restructuring fall into three categories: do nothing; do a little; or do a lot. One can think of these options in terms of the choice of what type of crisis insurance to buy: none at all; cheap, partial-coverage insurance; or expensive, full-coverage insurance. The article proceeds as follows. Section 2 describes the basic data on Italian sovereign debt. Section 3 works through the three options noted above. Section 4 concludes. According to Italian Treasury and Bank of Italy data, as of mid-2019, Italian public debt amounts to approximately e2.4 trillion (Table 1) . 11 The main holders of Italian public debt are domestic banks and other domestic financial institutions (e1.15 trillion, corresponding to 46 per cent of the total), non-resident entities (e750 billion, corresponding to 30 per cent of the total) and the Bank of Italy (e406 billion, corresponding to 16 per cent of the total). Italian households, which in the early 1990s held about 50 per cent of the Italian public debt, now hold less than 7 per cent of the total. About 83 per cent of the total stock of Italian debt (e2 trillion) consists of tradable instruments (government bonds and bills), which we refer to as "bonded debt". The data show that 43 per cent of bonded debt is held by domestic banks and other domestic financial institutions, 34 per cent by non-residents, 20 per cent by the Bank of Italy and the 10 For an early discussion of the economic impact of the Covid-19 crisis on Italy, see Ugo Panizza, 'Europe's Ground Zero', in Richard Baldwin and Beatrice Weder di Mauro, Mitigating the COVID-19 Economic Crisis: Act Fast and Do Whatever it Takes (CEPR Press 2020). 11 The most recent Bank of Italy data are available at . Treasury data on the composition of bonded debt are available at: used in this paper refer to the end of July 2019, while Italian Treasury data refer to the end of September 2019. There is a small discrepancy between Bank of Italy data, which report a total for bonded debt of e2,047 billion, and Treasury data which report a total for bonded debt of e2,020 billion. remaining 4 per cent by households. The remaining debt is held by Italian banks (e270 billion, corresponding to 65 per cent of the non-bonded debt), resident households (e81 billion, corresponding to 20 per cent of the non-bonded debt) and non-residents (e50 billion, corresponding to 12 per cent of the non-bonded debt). 12 More than 99 per cent of Italian bonded debt is denominated in euro (issues in US dollars amount to e4.9 billion, issues in British pounds amount to e2.5 billion and issues in Japanese yen amount to e200 million; Table 2 ). Most bonded debt consists of Treasury Bonds (BTP, 82 per cent of the total), Treasury Certificates (CCT, 7 per cent of the total) and Treasury Bills (BOT, 6 per cent of the total), all of which are governed by local Italian law. 13 A small share of bonds are listed on the Luxembourg exchange, under a Medium Term Note Program (EMTN). Most of these bonds are denominated in euro, but a small amount are in British pounds and Japanese yen. All of these Luxembourg-listed bonds, as best we can tell, are governed by local Italian law, although as we detail later, these bonds do have different (and better) legal protections for creditors than the aforementioned Treasury Bonds, Treasury Certificates and Treasury Bills. Finally there is a small amount of debt in the form of Global Bonds, issued under a US programme and governed by New York law (roughly e5 billion, denominated in US dollars). 14 In sum, about 99 per cent of Italian debt is governed by local law and, therefore, benefits from the local-law advantage (99.66 per cent if we include EMTN and 98.25 per cent excluding EMTN; Figure 1 ). About one-third (e597 billion) of outstanding BTPs were issued before 2013 and, hence, do not include collective action clauses (CACs) that are now required by the EU in order to provide an orderly path to restructuring. 15 Other bonds that do not include CACs include foreign-law bonds issued before 2004, ISPA bonds (these were bonds issued before 2013 to finance infrastructure projects), and BOT. In total, bonds and bills without Government bonds of the euro zone with maturities in excess of one year, the Treasury will albeit continue to issue, with its usual regularity, bonds without the above clauses, these being outstanding bonds at 01/01/2013 and off-the-run bonds, so as to guarantee the necessary liquidity.' Hence, the share of bonds without CACs is likely to be slightly higher than what is reported here. CACs amount to approximately e749 billion (ie, 37 per cent of total outstanding bonded debt; 33 per cent of the total if we exclude BOT, which amount to approximately e120 billion). Figure 2 shows the maturity profile of outstanding Italian bonds. During 2020-2021, the Italian Treasury will need to roll over about e60 billion per quarter, with large amounts of BTP without CACs that need to be rolled over in Q1 2020 and Q3 2021. 16 Figure 3 shows the residual stock of bonds without CACs. While the curve is steep, starting with e750 billion in Q3 2019 and dropping to e500 billion in Q4 2020, this is mostly due to maturing BOT. After the BOT effect ends, the curve flattens and by 2026, Italy will still have more than e300 billion of bonds without CACs. 17 The top panel of Figure 4 plots the expected composition of Italian bonded debt over the period 2019-2023. It shows that in 2023, more than 15 per cent of bonded debt will still consist of long-term bonds without CAC, if we add BOTs and hard-to-restructure foreign law and EMTN bonds, we find that in 2023, almost one-quarter of Italian bonds are hard to restructure. If to this debt we add loans (ie non-bond debt), we find that only two-thirds of Italian public debt consists of bonded debt with CACs (bottom panel of Figure 4 ). The do nothing approach is for the Italian authorities to literally do nothing. That is, to take no steps to utilize the local-law advantage in advance of a potential crisis. This is perhaps the most likely outcome, since governments are loath to admit that a debt crisis might be in the offing. The main reason for a country to take the do nothing approach is the concern that taking steps to prepare for a crisis will send a negative signal to the market and either lead to an increase in sovereign yields or, worse, cause the crisis to happen. 18 In most cases, this head-in-the-sand approach is far from ideal. But, as explained below, it may not be a terrible approach in the Italian case, given the advantages the legal terms governing the current Italian debt stock provide. As a starting point, Italy will have first-generation euro CACs in over 75 per cent of its bonded debt stock by mid-2020 (60 per cent of its total debt stock, see Figure 4 ). These clauses were put in place starting in January 2013, mandatorily, for all euro area sovereign debt issuances of over a year in maturity. Simplifying somewhat, the presence of these clauses means that all of the bonds containing them can be modified in a single transaction, so long as the approval of 66.67 per cent of the holders of each bond and 75 per cent of the holders in all the bonds put together (in principal amount) is obtained. However, there are issues. The requirement that the approval of 66.67 per cent of the creditors in each bond be obtained to effect a consensual debt restructuring gives rise to the risk of holdout creditors targeting a subset of the smaller bonds and accumulating blocking positions in them. In the Greek restructuring of 2012, where the majority of foreign-law bonds were governed by English law and had CACs that could only be activated with super-majority votes, there were holdouts in 25 separate sovereign and sovereign guaranteed bonds (amounting to about e6.4 billion in face amount of debt). The holdouts, after blocking the votes in their bonds, got paid in full. And this was in large part because Greece did not want to risk the type of litigation that Argentina was suffering at that same time. 19 For the Italian local-law bonds with CACs, the vote requirement is lower than what was required in the Greek foreign-law bonds (75 per cent, with a quorum requirement in the context of a bondholder meeting). But this is unlikely to be a significant barrier for a wellcapitalized hedge fund seeking to utilize the holdout strategy. In addition, if a creditor can figure out what the holdings of the European Central Bank (ECB) are in particular bonds (and they are substantial in many bonds), he/she will be in an even stronger position to hold out. The reason is that the ECB has asserted a 0 200,000 400,000 600,000 800,000 Source: Own elaboration based on IMF data and Italian Treasury data. In order to build the debt composition for 2020-2023, we started with IMF forecasts for the total stock of debt and then assumed that the shares of loans and BOT remain constant. Data for external debt, and domestic bonds are instead taken from the flow of maturing bonds data used for Figure 2 . The graph is built under the conservative assumption that no external bonds will be issued in the next three years; while Italy issued dollar bonds in October 2019, this was a small amount ($7 billion) and these issuances-and possible future external issuances-are unlikely to substantially change the figure given their relatively small amount. commitment to not voting in favour of any debt restructuring plan. 20 In a CAC that calls for an affirmative vote of the creditors of a certain percentage of principal amount, abstention is scored as a no vote. The ECB's commitment, therefore, effectively lowers the percentage the holdout would be required to purchase in order to block a modification. 21 The foregoing has already caused concern in policy circles-and has resulted in proposals for the euro area to reform its CACs starting in 2020. 22 That said, an argument could be made that the potential holdout problem is less serious than it has been made out to be. One way to see this is by detailing the steps a holdout creditor seeking to litigate against a distressed sovereign generally needs to take. Assuming that the bond in question has not yet come due but the sovereign is in default on other indebtedness, the holdout will typically utilize her bond's cross default provision to accelerate the payments owed on her bond, demanding immediate payment of principal and accrued interest. The ability to declare a default and accelerate is key in the litigation context because suing the sovereign on small unpaid coupon payments is unlikely to be worthwhile when measured against legal costs. Italian local-law bonds though, except for the sliver listed in Luxembourg under the EMTN program, have neither cross default nor acceleration provisions. In other words, the only bonds worth holding out on will be the ones where a substantial principal amount is due. Worse, from the holdout creditor's perspective, those bonds (except again for the Luxembourg-listed ones) lack explicit waivers of sovereign immunity from either suit or execution, which means that the Italian government lawyers can potentially use that as a defence. 23 Put that together with the dysfunction of the Italian legal system, 24 and it is hard to imagine one of the specialist hedge funds that won large recoveries against Argentina in the New York courts wanting to invest in holding out on the non-Luxembourg listed Italian local-law bonds. In contrast to Italian creditors, the foreign-law holdout creditors in the Greek restructuring in 2012 had the rights to declare cross defaults, accelerate the debt and sue for recovery in English, Swiss or Japanese courts under the explicit consents to jurisdiction and waivers of immunities from suit and execution in the contracts. Absent those rights to A key question will be whether there is an implicit waiver of sovereign immunity under Italian law. Ordinarily, we would presume that an Italian court would find such an implicit waiver as a matter of customary international law since sovereign bond issuances are considered a commercial activity. However, Italian courts have shown a willingness to reinstate immunity for actions of the state that are necessary to tackle an extreme crisis (and Covid-19 appears to have created just such a situation). 190-2. 24 In the most recent Doing Business ranking from the World Bank, Italy ranks 122nd out of 190 countries in terms of ease of contract enforcement. Average enforcement time is 1,120 days and the cost of litigation, as a fraction of the claim is 27.6%. Those do not strike us as numbers that would attract litigation oriented holdout creditors. For details, see Theresa Arnold, Mitu Gulati and Ugo Panizza • Euro area sovereign debt in the era of declare defaults, accelerate, bring suit and execute against the state's assets in a jurisdiction where the legal system moves quickly and efficiently, a sovereign such as Italy might hope that it would simply be unattractive for any investors in its bonds to hold out. If so, doing nothing might work. That said, the costs of default for a country that needs to regularly tap the borrowing markets can be substantial. And potential holdout creditors, especially if they watched what happened in Greece a few years prior, might recognize their potential to impose costs on Italy even in the absence of substantial litigation rights. 25 This is because during the restructuring period the country is likely to be given a Selective Default rating by the credit rating agencies, which will make new borrowing hard. 26 Plus, its bonds are unlikely to be eligible for delivery at the central bank's discount window, making them an unattractive investment. The end result will probably be an increase in sovereign borrowing costs beyond the usual default penalty. All in all, a country like Italy may not want to have a number of defaulted bonds outstanding after its restructuring. Let us assume that the prospect of being in a long-term payment default vis-à-vis a substantial number of creditors, even if their litigation options are not attractive, is a prospect that the Italian government does not relish. The question then is whether there are some non-obvious options available under existing Italian law. We see two: (a) an extension of maturities of those bonds that were issued at low rates in prior years and that are coming due; and (b) the imposition of a withholding tax equivalent to the haircut taken voluntarily by non-holdouts. As we discuss further below, these options pose some litigation risk, and Italy might choose to 'do a little' by taking steps to mitigate that risk. As a threshold question to examining these options, we must ask whether euro area countries contemplating a restructuring are constrained to utilizing only the current euro CACs and no other technique. The answer here, we think, is no for a number of reasons, including the explicit language of the CACs and the policy intent behind them. 27 To clarify the question, let us say that the Italian government decides that it wants to reduce its debt burden by imposing a withholding tax (more on this later). Do the euro CACs in its existing bonds bar the use of this strategy? No. There is nothing in the mandate requiring the inclusion of euro CACs that constrains any individual euro area country from using its tax authority. Indeed, there could not be, as Italy has retained the authority to levy taxes, even as it has delegated other aspects of sovereignty to the euro area. Italy is constrained in terms of the contract provisions it has agreed to with bondholders, but the vast majority of its local-law bonds (except those pesky Luxembourg-listed ones) do not contain the protections against the sovereign imposing withholding taxes that are standard in foreign-law bond issuances (including in those issued by Italy). A similar analysis applies to other routes to enable an easier restructuring, such as making it harder to sue in Italian courts; increasing the costs of litigating a contracts case, for example. That would not be particularly good for the Italian legal system and economy generally, but it would deter holdout creditors with debt instruments whose holders were required to litigate in Italian courts under Italian law. And there is nothing in the body of European law respecting CACs that constrains Italy from taking such actions. The language of the CACs themselves signals the possibility of other restructuring techniques being used. In relevant part, they say: In the case of a cross-series modification, the terms and conditions of the Bonds and debt securities of any other series . . . may be modified in relation to a reserved matter with the consent of the Issuer and [the requisite proportion of bondholders]. 28 The word 'may' is crucial; it tells us that the CACs are a debt restructuring option that the sovereign may use. But just one option. 29 If these CACs were meant to say that sovereigns were constrained from using other restructuring options, they would have said 'must' instead of 'may'. One might ask whether there is anything in the ESM Treaty, whose Article 12 mandated that all euro area nations utilize CACs, suggesting that future debt restructurings of the bonds with CACs could only proceed by utilizing the CACs. There is not. Nor is there anything in the history of why these CACs were put in place by the European authorities that suggests so. The history tells us that, starting with the infamous Deauville beach walk on 19 October 2010, the utilization of CACs was meant to set up an orderly mechanism for future debt restructurings in Europe and to signal to investors that the purchasers of euro area sovereign bonds that those bonds could be restructured in the future. 30 Mark Weidemaier, in his 2019 paper, sums this up: [A euro area] sovereign that has issued local-law debt [with CACs] remains free to alter its law to facilitate restructuring, although it will encounter various legal constraints in doing so. These constraints, however, are not absolute; there is room for the prudent exercise of the local law advantage. Having determined that the existence of the CACs does not preclude the sovereign from using the local-law advantage to facilitate other paths to a restructuring, we turn to a closer examination of the two options mentioned above: a maturity extension and the imposition of a withholding tax. A paper from 2011 (Edelen et al.) , and two follow-ups from 2019 (Cervantes et al. and Cramer et al.) , suggest that there might be a strategy to unilaterally extend maturities of debt with especially low coupon rates. 31 Assuming this strategy is workable, the idea would be to ask the holders of local-law bonds with especially low coupons to voluntarily extend maturities. The addendum to that polite request would be that those who refused would have their maturities unilaterally extended by law. The provision that the aforementioned papers hone in on is Article 3 of the Italian Public Debt Consolidated Act (the "Debt Act"). Under Article 3: [ The Ministry] has the authority, within the annual limits established by the budgetary law, to issue framework decrees that allow the Treasury to . . . proceed, in order to restructure the national and external public debt, to the reimbursement before maturity of bonds, to the transformation of maturities. 32 That language seems clear on its face. It says that the Ministry of Economy and Finance (the 'Ministry') has the authority to issue decrees to allow the Treasury to either redeem bonds at par (maybe because interest rates have dropped considerably since issuance) or transform (extend) maturities (maybe because interest rates have increased considerably since issuance). In both cases, investors lose and would presumably impose a reputational cost on the issuer. But the options to either reduce or extend the duration of a loan are familiar contractual provisions in the debt world (particularly the former 33 ). One might question whether, if Italy does hit turbulent times, this power to extend maturities will be of much value. After all, countries in deep sovereign debt crises typically need to impose principal haircuts as well as maturity extensions. The objection is a fair one. But, given the large amount of the Italian debt that is held domestically and the negative impact on the economy that will likely be caused by imposing brutal principal haircuts to the holdings of domestic financial institutions, maturity extensions may be an attractive option for the government. This is especially so if a significant portion of the Italian debt stock was incurred when borrowing rates were especially low (as is the case). 34 Under the structure of Article 3, the Treasury department receives the power to call bonds or extend maturities only if a framework decree authorizing such actions is issued by the Ministry. These decrees are not law; they are administrative actions implementing legal authority. They are also of limited duration. Each decree lasts for a year, no more. 35 In prior years, they have been used to authorize the Treasury to do both debt exchanges and redemptions. 36 It was also under Article 3's structure that a framework decree was issued by the Ministry to impose euro CACs starting in 2013. Let us say therefore that the Italian government decides to issue a framework decree in 2020 to authorize the Treasury, if need be, to extend the maturities of all Italian local-law bonds by five years. There are at least two possible barriers to the Treasury utilizing this authority. We take the potential barriers in turn. First, we consider Article 8 of the Debt Act. This is perhaps the most significant barrier to the use of Article 3. It requires that 'payments of public debt [cannot be] reduced, paid late or subject to any special levy, not even in case of public necessity'. 37 The question then is whether extending maturities or reducing them (as Article 3 seems to allow the Treasury to do) violates the '[may not be] paid late' requirement. At first cut, it does not appear so, since that reading would reduce Article 3 to meaninglessness. 38 That said, the argument could also be made that our reading of Article 3 reduces Article 8 to meaninglessness as well. And courts generally do not like to negate statutory provisions. The reality, therefore, is that if Italy decides to go down this path, a court will likely have to decide which of the clauses to favour, while putting a gloss on the other. And Italy's hope would have to be that the court chooses to interpret Article 8 in a manner that weakens it, such as by reading it to require simply that whatever the maturity date iswhether shortened, lengthened or kept the same-it has to be complied with. The foregoing is not implausible in the context of the euro CACs. The reason being that if Article 8 were to be interpreted to bar extensions of maturity, that would also negate crucial provisions of the CACs that, by Italian law, explicitly allow for the sovereign to extend maturities of the bonds with euro CACs with a super majority approval of the creditors. The question for the Italian government will be whether it wishes to go down this path of litigation. 40 Earlier we discussed the question of whether the existence of the euro CACs as a restructuring mechanism constrains the use of Article 3's power generally. Now, we address the question of whether the Treasury might use its Article 3 power in a specific bond that contains a euro CAC as part of its contract terms. After all, if a crisis were to occur at any time over the next few years, there would be more Italian local-law debt with CACs than without. The euro CACs have an explicit mechanism by which the maturities of the bonds can be extended, requiring a pre-specified vote of the creditors. Our view is that, for those bonds, the right of the sovereign to unilaterally extend the maturities is dubious. 41 The reason being that these bonds already have a term explicitly governing how maturities may be extended. But for the remaining debt-which we estimate in the years 2020-2024 to range between 45 per cent and 35 per cent of the total debt stock ( Figure 3 )-there are no CACs. For that latter subset, there is a strong argument that Italy can use the power that it already has by law to unilaterally extend maturities. We should caution though that such a power is unlikely to be unlimited. If the sovereign, for example, tried to extend the maturities of its one-year bonds by 50 years, there would likely be an immediate expropriation challenge. 42 Because the option of unilaterally extending maturities is likely to only work for locallaw bonds that do not contain CACs and the relief that Italy would obtain would be limited compared to a more significant restructuring that reduces principal amounts as well as extending maturities, Italy may wish to pursue other options presented in this article. However, if Italy wanted to make this tool a more feasible option for a future restructuring, it could reduce the litigation risk associated with this option by proactively clarifying the apparent discrepancy between Article 3 and Article 8. Indeed, repealing Article 8 might be even better. 40 Another objection that we have heard informally from some in the Italian government has to do with the Ministry's annual decrees issued between 2012 and 2019, which say that restructurings in those years will occur via 'mutual consent'. This guidance, we suspect, was a signal to the markets that restructurings would be done via a consultative process. There is no specification of what that consultative process would be, nor have we been unable to uncover evidence of the intent behind the words 'mutual consent'. But we have reason to think that 'mutual consent' does not mean that unanimous approval of creditors is required. There are two points worth considering here. Given that bonds issued after 2013 contain euro CACs, "mutual consent" cannot possibly mean a requirement of unanimous creditor approval because that would contradict the explicit super majority requirements of the euro CACs for maturity extensions. Further, the mutual consent requirement, whatever it means, comes out of an annual decree. It is up to the Ministry, every year, to decide whether it wishes to use that language in its decrees. If it happens to be a year where the Ministry wishes to give the Treasury to power to unilaterally extend maturities, it can delete the 'mutual consent' language from that year's decree, as it has in many prior years. 41 For discussions of the argument that the power of unilateral extension might apply to the bonds with CACs as well, see Cervantes et al (n 31); Cramer et al (n 31). 42 We do not, however, think that an expropriation objection would work generally against all uses of the Art 3 power to unilaterally extend maturities. Expropriation objections apply to a state taking away property rights that creditors have been granted (eg, the Greek retrofit of 2012). However, Art 3 has been in place since 2003 at least. See Cervantes et al (n 31). That said, Italy would be on stronger ground in the future if it issued a clarification of Art 3's powers earlier rather than later. Inherent to sovereignty is the power to tax. Bondholders who are owed moneys by a sovereign have, therefore, since time immemorial, worried about the sovereign deciding to impose a new tax on the payments it owes as a way to avoid payment. As a result, almost every international sovereign bond governed by the laws of New York or England contains what is called the tax gross up clause. This clause protects the bondholder against the sovereign deciding to levy an extra tax on the payments it owes on the bond as a way of doing a back door restructuring. 43 Most local-law-governed bonds, however, do not contain such a protection. And that is the case for all of Italy's local-law bonds, except the small fraction of Luxembourg-listed ones mentioned earlier. A simple strategy to deal with holdouts, therefore, is for the Minister of Finance to give a speech to local bankers in which she says that holdout bonds will be taxed so that the net (after tax) return will equal the payout on the new bonds. This was tried in Jamaica's restructuring of local-law bonds in 2010. 44 The minister's threat was successful and he never in fact had to introduce the tax. The fly in the ointment is if the country has bilateral investment treaties and double taxation treaties that bar such actions against the citizens of other countries. 45 An Italian government considering such a strategy would need to consider the risk of future litigation under these treaties. Indeed, it could reduce the risk of this litigation by being careful about the provisions it agrees to in any new treaties. 46 Taking steps today to reduce the future litigation risk associated with an exercise of Italy's power to unilaterally extend maturities or impose a withholding tax on holdouts would improve Italy's future restructuring options. But neither of these options provides Italy with as much flexibility for significant debt relief as the option of retrofitting more effective CACs, described in the next section. Governments, as we noted at the start, are generally reluctant to do anything that might signal to the outside world that they are concerned about the possibility of a financial crisis; even if everyone recognizes that it is a real possibility. So, 'Do a Lot', is not a strategy we expect to be embraced. Nevertheless, let us assume for the purposes of this section that a sovereign such as Italy would want to minimize the future costs of potential holdouts in a possible restructuring. 43 Buchheit explains how the clause works: 'Any deduction as a result of the imposition of withholding taxes thus becomes the borrower's responsibility because the borrower's payment to the lender must [by virtue of the tax gross up clause] be topped up to compensate for the deduction.' Lee C Buchheit, How to Negotiate Eurocurrency Loan Agreements (1st edn 1995). 44 We have found no mention of this technique in the sovereign debt literature. Unsurprisingly, sovereign debt restructuring guru, Lee Buchheit, advised Jamaica at the time. 45 We have not delved into the complexities of Italy's tax agreements with other countries. 46 The aforementioned Art 8 could pose a barrier here as well, since it bars any 'special levy'. As mentioned earlier, clarifying that Art 8 was not meant to pose such a barrier in advance of a crisis might be advisable. Theresa Arnold, Mitu Gulati and Ugo Panizza • Euro area sovereign debt in the era of Operating under that assumption, we think that the strategy of enhancing the existing CACs in bonds that have them and retrofitting them in the debt that does not already have them is a good option for a number of reasons that we explain below. As noted earlier, all euro area sovereign bonds issued since 2013 with a maturity greater than one year already have CACs in them. But these CACs contain the requirement that a 66.67 per cent vote of the holders, in principal amount, be obtained for each bond in order to use the aggregation feature to modify all of the bonds containing the CACs. And this creates a vulnerability to holdout creditors. Further, there is a whole set of debt instruments that are going to be even more vulnerable to holdouts, which are the bills with maturity under a year, the long-term bonds issued before 2013, government guaranteed instruments, non-bond debt (eg, syndicated loans) and so on. The reason being that CACs were not made mandatory for these sovereign debt instruments. Euro area sovereigns could have inserted them voluntarily; but, as far as we know, none did. Realistically though, if the crisis is an especially serious one the government is going to need to restructure some of these instruments as well. 47 Thus far, we have tackled the easy question of whether the sovereign might use one of its other powers-such as its taxing authority-to engineer a debt reduction without running afoul of the mandatory CACs. The answer was yes. The trickier, albeit closely related, question is whether the sovereign can enhance the existing CACs in its bonds by using its local-law authority. By enhancing, we mean making changes to the existing model CACs provided by the European authorities to make the bonds even less vulnerable to holdout problems than they are now. In particular, two modifications that have already been the subject of debate among policy makers are: (a) introducing a mechanism by which the entire debt stock of a country can be modified at a single shot; for example, via a vote of 75 per cent of all the holders across all the bonds, so long as some non-discrimination criterion is satisfied (the so-called single shot or single-limbed feature); and (b) an addendum to the voting rules for modifications that disenfranchises the ECB from having its votes counted (assuming that the ECB continues to hold to the position that it is legally constrained from voting). 48 47 Conventional wisdom in sovereign debt restructurings is that short maturity instruments (under a year), trade credits, and guaranteed debt, generally get paid in full. But that assumption only holds where the amount of debt in these instruments is small. In the recent Barbados restructuring of 2018-2019, the amount of short-maturity debt was so large that it had to be restructured. And, since those instruments had no CACs, they had to be retrofit into them in the same manner as was Both changes to the existing euro CAC structure correct for design flaws in the original 2013 version and enable the CACs to better achieve their goals of deterring holdouts while providing a clear and predictable mechanism by which creditors can participate in a burden sharing exercise. At the time the original euro CACs were designed, it was thought that a bond-by-bond vote requirement of 66.67 per cent would be enough to deter holdouts. And the implications of the ECB not being able to vote in a restructuring (or having to always vote no, regardless of the restructuring proposal) had not been fully considered. So, the question is whether an individual nation, such as Italy, concerned about the possibility of a crisis, can act on its own via its 'local law' advantage to legislate improvements in its CACs consistent with the original goals of those CACs. There are two potential barriers here. First, for new bonds issued with the improved CACs, there is the question of whether deviating from the form mandated in 2013 violates European law. Secondly, for already-issued bonds where these provisions are retrofit via legislative action, there is the potential for creditor lawsuits claiming an interference with property rights. We take the possibilities in turn. Neither poses a significant barrier. A country including enhanced anti-holdout provisions in its new bonds potentially violates the instruction in the ESM Treaty that CACs be introduced in a manner such that their 'legal impact is identical'. 49 The argument against Italy being able to fix the design flaws in its CAC would be that all European countries are required to have the same flaws until the European authorities decide it is time to fix those flaws for everyone. 50 There is a certain formalist appeal to the foregoing. 'Identical legal impact' arguably means the same in every way. But, the very manner in which these euro CACs were designed tells us that the creators could not have meant to use the term 'identical legal impact' in such a broad fashion. The more likely meaning of 'identical legal impact' is much narrower. That is, that every country would have an identical baseline of the same basic CAC framework that would signal to holders of sovereign bonds that a debt restructuring was possible. And that one way (not the only way) by which this restructuring could be implemented would be via the basic euro CAC. And everyone had to have that identical baseline. There was no prohibition, however, on the sovereign enhancing the operation of the CACs by including additional provisions to its debt contracts. One way to see the foregoing is to compare the operation of CACs in the Italian foreign-law (New York law) versus local-law bonds. 49 Art 12(3) of the ESM Treaty says: 'Collective action clauses shall be included, as of 1 January 2013, in all new euro area government securities, with maturity above one year, in a way which ensures that their legal impact is identical'. 50 Some commentators do take this view. For example, Klaus-Albert Bauer writes: 'A natural understanding of [ESM Art 12] this clause would seem to prohibit not only the issue of bonds with non-conforming collective action clauses but also the later unilateral amendment of Euro Area CACs to suit a particular issuer's needs in times of crisis.' Klaus-Albert Bauer, 'The Euro Area Collective Action Clause-Some Questions and Answers at 10', in Collective Action Clauses and the Restructuring of Sovereign Debt (n 30). Theresa Arnold, Mitu Gulati and Ugo Panizza • Euro area sovereign debt in the era of The CACs in the Italian foreign-law bonds, such as the one issued in October 2019, are buttressed with all sorts of potent investor protections including acceleration clauses, cross default provisions, waivers of immunity and consents to jurisdiction. What this means is that CACs in the foreign-law bonds, if the bondholder has the votes to block a restructuring attempt, can be turned into a potent litigation weapon. Assets can be seized, payments to other bondholders can be blocked and so on. Conversely, in the local-law bonds that lack these protections, there is no weaponry for the holdout creditor other than the hope that the sovereign does not want to suffer the reputational cost of default. The point being that the issuer, by including a stand-alone CAC in the local law bonds and not adding extra creditor protections, has an effective anti-holdout weapon, rather than the possible creditor weapon that the CAC tends to be in the foreign law bonds. Both the localand foreign-law bonds contain identical CACs, but they can potentially have radically different effects as a function of the other contractual bells and whistles that the issuer adds (or chooses not to add) to the baseline. And this was explicitly envisioned right from the start by the European authorities, who understood that sovereign bonds would be issued under both local and foreign law and that the sovereigns may want to include enhancements such as acceleration clauses or trustee structures. 51 Plus, the actual CACs for foreign-law bonds were designed differently in one crucial fashion right from the start-the governing law is a reserved matter that cannot easily be changed in the foreign-law bonds whereas it is not a reserved matter at all in the local-law bonds. 52 That is a significant difference, as Weidemaier (2019) points out. 53 In other words, the CACs were designed such that they could not have identical impact (in its most literal sense) right from the start. Given the foregoing, we think that the sensible interpretation of Article 12(3)'s dictates is that every euro area nation needs to implement identical model CACs. But those serve as a baseline; ensuring that creditors all are forewarned that a restructuring of sovereign debt in the future is possible and that one method that might be used is the basic euro CAC mechanism. Nothing, however, stops the sovereign from either adding in more investor protections (making the CAC a weaker tool from the sovereign's perspective) or adding improvements to the voting mechanism (making the CAC a stronger tool from the sovereign's perspective) as long as the basic model CACs are kept in place. To reiterate, the euro CACs serve as an identical baseline for all euro area sovereigns. From a policy perspective, it is easy to see why the foregoing makes sense. Euro area sovereigns vary vastly in terms of their economic strength and the types of markets they are able to tap. That means that they will likely need to offer different contractual rights to investors in order to borrow effectively; especially in a world where the European authorities have made it clear that they are not guaranteeing the payment of each individual sovereign's claims. To the extent a sovereign wants to prepare for a future crisis by 51 experimenting with using enhanced restructuring tools that build on what the European authorities have already provided, and particularly if this is done in consultation with the European authorities, it is hard to see why those authorities would not wish that. The experimentation by different sovereigns with various enhancements can only serve to provide the authorities with helpful information. Last but not least, there is the question of whether, even if there was a hypothetical violation of the Article 12(3) 'identical legal impact' dictate, there would be any practical impact. With the caveat that the matter of international treaty enforcement is beyond our expertise, our understanding is that violations of treaty obligations generally only give the signatories to the treaty the right to bring legal actions. The presumption is that there are no private rights of action. 54 Here, if the European authorities are happy with the actions of the individual state, there is not going to be any complaining; just the clinking of champagne glasses. If faced with a debt crisis situation and the need to protect against holdouts, Italy will likely wish to retroactively add the two enhancements mentioned above-the option of using single-shot CACs and the disenfranchisement of the ECB-to all of its already-issued and outstanding local-law-governed debt. The legal barrier to making such a retroactive change, we believe, is non-trivial. The starting point in analysing this question has to be the recognition that most modern legal systems disfavour retroactive changes to contract rights. 55 Moreover, scepticism about the government's need to make such changes is likely to be especially heightened in contexts where the government is seeking to reduce its own debt obligations. 56 To illustrate, let us say that the current euro CACs allow the issuer the following options: (a) If holders of 75 per cent or more of an individual series of bonds (in principal amount) approve of a restructuring offer, that offer becomes binding on all holders. (b) In a cross-series modification, if holders of 66.67 per cent or more (in principal amount) approve of the restructuring offer, and 75 per cent or more (in principal amount) of all the holders of the various series aggregated also approve of the offer, the offer is binding on all holders. 57 To these existing options, let us now say that Italy wishes to add a third and different option, using its local-law power. That option enhances the existing CACs by permitting, in the same fashion that is standard for bonds on the international (foreign-law) market, a single-shot restructuring option. (c) In a cross-series modification, if the restructuring offer satisfies the condition that the terms are Uniformly Applicable to all holders, and 75 per cent or more of the holders of all the series approve the offer, it will be binding on all holders. 58 Let us say that Italy wishes also to add in a provision to disenfranchise the ECB or any other supra-national institutions that might be mandatorily required to vote one way or the other on a restructuring plan without consideration of the merits of the situation. For bonds that were already issued, prior to the hypothetical Italian debt crisis, with either no CACs or the first-generation euro CACs, these modifications will likely face legal challenge as a violation of property rights. Both the Italian constitution and European law provide protections for expropriations of property rights. Article 42 of the Italian Constitution protects against interference with property rights unless the action is for reasons of 'general interest' and the government makes 'provisions for compensation'. And Article 13 of the European Convention on Human Rights is similarly protective with protections for the peaceful enjoyment of property except under conditions of 'public necessity'. In such situations, the interference with property rights needs to be proportional to the public need. 59 The two questions to ask, therefore, are (i) whether the inclusion by Italy of an additional option to the existing model euro CAC and a provision disenfranchising the ECB, in order to better tackle a financial crisis, would be ruled as the kind of action that was in the general public interest and proportional to what was needed, and (ii) what compensation might be required, if there was a violation. 60 The question of whether the retrofit inclusion of a single-shot CAC would survive the kind of proportionality analysis that was done by courts in the aftermath of the Greek 2012 restructuring has already been the subject of two excellent papers by Manuelides (2019) and Weidemaier (2019) . Both suggest that a retrofit of a single-shot CAC option would likely pass muster in the European courts. And although they do not analyse the 58 This condition that the offer must be 'Uniformly Applicable' is basically a non-discrimination condition that was put in place in international bonds in 2014 when the single-shot CAC restructuring option was added to the standard terms. question of adding in a provision that disenfranchises the ECB, we think that that would similarly be ruled to not be an undue interference with property rights. The key case that helps predict how the retrofit would fare is the European Court of Human Rights' decision in Mamatas and Others v Greece. Others such as Manuelides (2019) have parsed the implications of the various portions of the court's discussion and we will not repeat that analysis here. 61 However, it is helpful to boil down the core of the court's analysis in the Mamatas case and indeed a variety of other cases from the corporate context where corporations have put in place contractual modifications to help ameliorate a holdout problem. 62 As we see it, the crucial question for the court will boil down to whether the sovereign has, in its use of the local-law advantage, acted opportunistically to take value away from the creditors and benefit itself. 63 Or, alternatively, whether it acted in the public interest in trying to fix flaws in the existing mechanism for restructurings so as to enable an orderly restructuring mechanism that benefits both the general public and the majority of creditors. Assuming that the steps that Italy takes are in the vein of correcting flaws in the existing restructuring mechanism because of the possibility of an imminent crisis and, moreover, that they are similar to those that the European authorities have indicated are in the works for implementation across Europe anyway, it is hard to see how or why a court would rule the changes invalid. This is essentially what Greece did with its CAC retrofit in March 2012. At the time of the Greek restructuring in 2012, European policy makers had already given the green light for a committee to design a CAC suitable for inclusion in all euro area bonds. But that committee process took time, given that agreement from all the Member States had to be obtained, and was not ready as of March 2012, by which time Greece was unable to wait any longer. Therefore, Greece's restructuring lawyers designed a CAC that basically was an advance version of what would subsequently be designed at the European level. 64 Indeed, the experience with the use of the Greek designed clause has been useful in informing discussions of what kind of clause was appropriate for euro-wide inclusion. More importantly for our purposes, it was important to the European Court of Human Rights-in the context of its approval of Greece's retrofit-that Greece had retrofit the kind of provision that was standard. 65 Finally, there is the matter of damages. If it so happens that courts rule that Italy has violated property rights protections, the question to ask is what the monetary damages would be. 66 After all, if it turns out that the impact on the prices of the bonds from adding in features that improve their operation is an increase in the price of the bonds, then damages should be small (zero). And if damages are negligible, it should not matter if the courts find a technical legal violation. 67 Different legal systems analyse contract and expropriation damages differently. If viewed as a pure contract violation, the legal remedy in both civil and common law systems is what is called expectation damages; which is to give the aggrieved party the monetary amount that puts her back in the position she would have been had there not been a violation. 68 And one way to measure that would be to look at the prices of Italian sovereign bonds before and after the retrofit. Based on the existing empirical work on the pricing impact of including CACs in sovereign bonds, it is safe to predict that the impact of the retrofit will either be to increase the market price of the bonds (after all, they are less subject to the cost of holdouts now) or to have no price impact at all. 69 That is, damages would be zero. An alternative possibility would be for the court to ask what the value to the creditor would have been to be a holdout from the restructuring. We think it unlikely a court would go down this path, given how speculative the calculations would have to be. That is, unless the crisis has hit and the market begins pricing the holdout-friendly bonds at a premium to bonds that are easier to restructure. 70 This may happen for two reasons. First, clauses that have to do with restructuring are likely to be more salient at a time of crisis and hence may be taken more seriously and even misinterpreted during these periods. In tranquil times, however, markets may end up barely noting such an innovation. Secondly, if the sovereign takes action such as retrofitting CACs in the midst of a crisis, it is likely to be viewed as a signal that the sovereign's situation is worse than the market had anticipated. Furthermore, there may be a price differential simply because turbulent times are characterized by high price volatility. Assume that the price of a group of bonds happens to drop the day a retrofit is included in these bonds, a court may interpret this drop in price as linked to the inclusion of the retrofit even if this drop was purely driven by the high price volatility that characterizes turbulent periods. Put simply, the greater the amount of daylight between when the retrofit is done and when the crisis hits, the smaller the likelihood of significant damages. The choice our article poses is between doing something and doing nothing. This choice can be better understood if we frame the problem in the same way as we would frame the decision of whether we should buy insurance to protect ourselves from a particular event that may or may not occur in the future. In this case, the uncertain event is the need to restructure one's sovereign debt. From the point of view of the borrowing country, the expected cost of a restructuring is P Â C; where P is the probability that a debt restructuring will be needed and C is the political and economic cost of the restructuring. P and C, respectively, measure the probability and the cost of restructuring if the country does not do anything to protect itself. If the country buys insurance (ie, if it does something along the lines described above), it will pay an insurance premium s, it will need to restructure with probability aP, and, in case of restructuring, it will pay a cost C 0 (with C 0 P, where a captures the increase in the probability of restructuring associated with doing something. Doing something will be convenient if the expected cost of implementing a reform is lower than the expected cost of doing nothing. That is, if: (s þ aPÂC) 0 < (PÂC). This condition can be written as: s < PðC À aC 0 Þ Theresa Arnold, Mitu Gulati and Ugo Panizza • Euro area sovereign debt in the era of We already discussed that C is likely to be substantially smaller than C. Therefore, if the probability of a future restructuring in non-zero (ie, P > 0), and if moral hazard is not very important (ie if a is not much larger than one) P(CÀaC 0 ), will be positive. But what do we know about s? The existing evidence suggests that s is likely to be very small, or even negative. For instance, multiple papers find that that the introduction of CAC in European bonds was associated with a reduction (and not an increase) in yields: pointing to the idea that s could be negative (or, for sure, not positive and large). Note that this is view also seems to be shared by the Italian authorities. Italy recently issued a dollar bond which included all sorts of creditor protections, but according to the Director General for Public Debt of the Italian Treasury if the bond had been issued under local terms and priced in euros the yield would have been broadly in line with the current yield on Italy's benchmark 10-year BTP bonds. 71 In fact, the spread over US Treasuries of the Italian dollar bonds (150 basis points) was basically identical to the spread of Italian BTPs over German Bunds. Taken together the evidence on bonds with CACs indicates that the market does not penalize bonds that are easier to restructure and anecdotal evidence on the recent Italian bond issuance suggests that the market does not seem to reward bonds that have clauses that make them especially creditor friendly. This all suggests that s is likely to be very close to zero. But if s is close to zero, then the moral hazard problem (which contributes to s) is unlikely to be very important, making a close to one. Hence, as long as the probability of restructuring is non-zero the condition that s< P(CÀaC 0 ) is likely to hold and the option of doing something now dominates that of doing nothing. In theory, that is. 71 See Giuseppe Fonte, Italy Plans to Issue More Foreign Currency Bonds Next Year, Reuters, 10 October (2019) The Economics and Law of Sovereign Debt and Default We draw from Mark Weidemaier and Yannis Manuelides on this point. See MC Weidemaier, Restructuring Italian (or Other Euro Area) Debt: Do Euro CACs Constrain or Expand the Options Road to Nowhere: The Legal and Remedial Futility of Challenges to a Restricted Single-Limb Retrofit of Local-Law Italian Debt Stock On private rights of actions under international treaties more generally, see Paul B Stephan For discussions, see Buchheit and Gulati Restructuring Sovereign Debt under Local Law: Are Retrofit Collective Action Clauses Expropriatory' (2012) 2 Harv Bus L Rev 163 This is one of the lessons of the famous gold clause cases of the 1930s, where the US government used the local law advantage to abrogate the gold clauses in its bonds and in the bonds of private parties. See Boudreau, ibid. For additional detail, see Sebastian Edwards, American Default: The Untold Story of FDR, the Supreme Court and the Battle Over Gold Financial Crises and Constitutional Compromise In the actual euro CACs, there are additional options and different voting thresholds and quorum requirements for when the votes are taken at a physical meeting of the holders versus in writing Mitu Gulati and Ugo Panizza • Euro area sovereign debt in the era of Covid-19 Manuelides also analyses the important precedent of Germany retrofitting CACs into its local corporate instruments some years prior. See Manuelides (n 2). Numerous others have also analysed the implications of the litigation that followed the Restructuring Government Debt Under Local Law: The Greek Experience and Implications for Investor Protection Under European Law The Future of Involuntary Sovereign Debt Restructurings, Mamatas and Others v. Greece and the Protection of Holdings of Sovereign Debt Under the ECHR' (2019) 14 Cap Mkts International Arbitration and Greek Sovereign Debt' (2018) 19 Oregon Rev Int'l L 180 The Greek Bond Haircut: Public and Private International Law and European Law Limits to Unilateral Sovereign Debt Restructuring Exit Consents in Sovereign Bond Exchanges For reflections on the design of the Greek restructuring from its primary architect, see Lee C Buchheit, The Greek Debt Restructuring of 2012', Global Restructuring Rev Mitu Gulati and Ugo Panizza • Euro area sovereign debt in the era of Covid-19 The fact that the voting thresholds that Greece saw in its retrofit CACs in 2012 were lower than those used in international bonds at the time did not trouble the courts (it had a simple % quorum and 66.67 % aggregate vote requirement). Nor, for that matter, did the fact that Greece used the single-shot CAC that aggregated votes across all the series-a type of provision that is now standard we are assuming a scenario where the litigation gets resolved some years after the restructuring has been conducted. That puts the remedy of specific performance out of the realm of consideration since a court is unlikely to be willing This was the outcome in the Gold Clause case of the 1930s (Perry v United States) that involved a constitutional challenge to the US government changing the terms of its own bonds retroactively. See Vanberg and Gulati For a discussion of comparative contract law remedies, see, eg, Luca Ficetola, Comparing Remedies for Contract Violations in Italian and English Law But it is a possibility. See, eg, Paolo Colla et al Pricing Terms in Sovereign Debt Contracts: A Greek Case Study' (2011) 6 Cap Mkts L Hidden Holdouts: Contract Arbitrageurs and the Pricing of Collective Rights 18 For the argument that whatever may make a debt restructuring process smoother is optimal ex post but not ex ante, see