The State of Sovereign Debt Restructuring After the Meetings in Marrakech

The Common Framework is now three years old.

Zambia has been in default for that long too. Sri Lanka, technically outside the Common Framework but conceptually a twin case, would have been in default for most of this period too but for just over $1 billion in emergency fiscal and balance of payments support from China’s “commercial” state development bank, the China Development Bank (CDB). When that credit line was exhausted in early 2022, Sri Lanka also defaulted. Ghana slipped into default earlier this year after its domestic debt market could not absorb the government's increased domestic bond issuance and it entered a spiral of higher rates and growing deficits.

Over those three years, the initial pandemic shock faded.  Today, all parts of the global economy are settling into a new era marked by higher rates in the U.S. and Europe (though, for now, not in China or Japan) and higher borrowing costs across much of the so-called “Global South,” or, to use the term of optimistic bond salesmen from ten years ago, the world’s “frontier markets.”

It thus is a reasonable time to take stock of where the global debt restructuring architecture stands, start to distill lessons from recent experience, and offer a few suggestions for a path forward. Five points stand out.

Too Much Was Expected of the Common Framework

The Common Framework wasn't entirely new. It was essentially the standard debt restructuring architecture for lower middle-income countries with substantial debts to official bilateral creditors (i.e., “other governments” rather than the private markets).

Its basic features of that process are well established, if hard to apply in practice: the debtor country needs an IMF program, debt sustainability is assessed by the IMF (with input from creditors) on a case-by-case basis, and the standard Paris Club principle of comparability of treatment is meant to provide fair treatment across all buckets of the country’s external foreign currency debt (the Paris Club, non-Paris Club bilateral creditors, and private commercial lenders of various sorts – be they Glencore or private bond investors).

Most of the signatories to the Common Framework (19 of the 20) also agreed that the multilateral development banks (MDBs) should remain preferred creditors, and that a new initiative round of across-the-board debt relief after the Heavily Indebted Poor Countries (HIPC) Initiative wasn’t warranted. The IMF's analysis of low-income countries supports the decision to assess sustainability on a case-by-case basis rather than engage in an across the board write-down.

The only real innovation was the call for a single official creditors’ committee that would bring together all governments with claims on the country in distress. In such a committee, the new official creditors – Saudi Arabia is one, India is another, and China, of course, is the most important – would negotiate new payment terms together with the traditional “Paris Club” creditors from the world’s advanced economies. In many cases, the new creditors had most of the exposure – but the thought was that the traditional creditors’ accumulated experience would help smooth the negotiation of unavoidable adjustments to repayment terms.

Of course, there was a risk that negotiating together might slow things down, the creditors effectively gained a procedural veto. However, it also ensured fair treatment across the broader official creditor committee – as everyone would, in the end, agree on the same deal.

Three years on, it is fair to say that no one is extremely happy with the result.

The common official creditors’ committee has often proven to be a source of delay. The Common Framework didn’t spell out what would happen if a large bilateral creditor disagreed with the IMF’s debt sustainability assessment (as China did in Zambia), nor did it include a commitment for all the members of the expanded official creditors’ committee to provide the IMF with the traditional financing assurances quickly.

The lack of financing assurances from China – and thus the official bilateral creditors’ committee – was the main reason why Zambia's restructuring took so long.

In the case of Sri Lanka, which is just a bit too rich to qualify for the formal Common Framework or to fall under the low-income countries debt sustainability framework, the official creditors have chosen to organize themselves differently. There is a separate Chinese official creditors’ committee led by the Export-Import Bank of China (Exim) that negotiates apart from the Paris Club. And there almost certainly will be separate committees for China’s commercial creditors (including the CDB) and the bonds.*

But it was always probably unrealistic to assume that giving China a seat on the official committee would be enough to generate a smooth restructuring process in a world marked by increasingly geopolitical tensions – the G-20 itself has fractured after Russia’s invasion of Ukraine.

Moreover, China’s official creditors don’t necessarily share much in common with the traditional Paris Club creditors. The advanced economies have accepted that their lending to low-income countries should broadly be on concessional terms, while China’s main official lender, China Exim, often wants a commercial return (generally LIBOR plus). Exim can push out its maturities for a few years while keeping a market interest rate – but it generally is not authorized to provide deep NPV debt relief (more background).

Putting China’s official creditors on the official creditor committee also did not solve the related but distinct question of how to coordinate different Chinese creditors. China isn’t just a single lender – there is a large stock of debt from several entities: the Ministry of Commerce, China Exim, CDB, the Industrial Commercial Bank of China (ICBC), and sometimes other state banks. In some cases, big Chinese firms themselves have extended buyers’ credits to their customers.

China seems to have decided that only Exim will participate in the official creditors’ committee – and thus narrowed the scope of the restructuring coordinated through it: the Sinosure-guaranteed ICBC loans should also be covered (based on Paris Club precedent), but they dropped out of the Zambia deal at the last minute.

The Common Framework Hasn’t Been all Bad

Zambia’s restructuring is the defining restructuring done through the Common Framework. It clearly took far too long, but at the end of the day, Zambia will get some real debt relief – at least from the official creditors.

True, the amount of that debt relief isn’t yet fixed, and the relief provided by the bondholders if Zambia's payment capacity is upgraded amounts to forgiving a couple years missed interest payments and a 6 percent rather than an 8 percent coupon for a few years (assuming the proposed restructuring is completed. The bondholders also get an amazing amount of cash upfront – over $500 million in early amortization of the $2 billion base bond.  That is something that should raise concerns about the comparability of the two deals.

But the deal with China Exim clearly results in real NPV relief. If Zambia’s debt servicing capacity is still judged by the IMF to be poor in 2027, Zambia pays Exim a concessional 1 percent rate for ten years, and then the interest rate owed to Exim steps up to a still concessional 2.5 percent as the loan amortizes over the following ten years. If Zambia’s debt capacity is upgraded, the coupon steps up to a still somewhat concessional 4 percent rate – and more worryingly, the amortizations also jump. As a result, the loan would be paid off in full after fifteen years (in 2038). With the Exim deal (unlike with the bond deal) there is unambiguous NPV debt reduction in both cases (40 percent relative to the accumulated stock of principal and interest arrears in the base case, 20 percent in the case of an upgrade).**

Sri Lanka has fairly similar debts relative to the size of its economy, but it fell just outside the Common Framework, and it doesn’t currently have a clear path to securing real debt relief – and runs a high risk of needing a second restructuring. For all its faults, the Common Framework did deliver a better outcome for Zambia than looks likely in Sri Lanka.

The Current “Architecture” for Evaluating Debt Sustainability Has Too Many Cliff Effects

The architecture for the evaluation of debt sustainability all hinges on a single institution, the IMF.

That should make consistency across cases easier.

But the IMF itself actually has two fundamentally different frameworks for debt sustainability.

The key feature of Common Framework cases is that only low-income countries are eligible, and they are thus evaluated based on the joint IMF/World Bank low-income country debt sustainability framework. The IMF and the World Bank both tend not to like this framework because it makes heavy use of balance of payments-based metrics rather than more familiar fiscal measures. It does, however, have some virtues: it sets clear limits on external debt-to-GDP (30 or 40 percent as countries in trouble usually aren’t judged “strong”), external debt-to-exports (140 to 180 percent, though 140 falls to 85 percent in certain conditions), and external debt service.

The market access country debt sustainability framework rests on fundamentally different analytic foundations. Most obviously, it completely ignores external public debt.*** That, I think, would trouble the IMF’s founders. It certainly reduces my confidence in the model’s output.

The model itself focuses on public fiscal debt – and in restructuring cases, it sets limits on total public debt (domestic and external) and the gross financing need.**** The model also considers the risk that a country’s fiscal debt profile will fall off the IMF’s forecast path over time, and thus the so called “width” of the fan chart showing the path of public debt to GDP is part of the IMF’s overall assessment. See the IMF’s argument for why Argentina’s current public debt is sustainable.***** In Sri Lanka’s case, this framework – which the IMF considers to be state of the art technically – allowed for a very high public debt-to-GDP ratio. All that matters for the model is the size of the gross financing need.

I suspect that the belief that the gross financing need (the fiscal deficit plus amortization) is a better measure for debt sustainability than, say, fiscal debt relative to fiscal revenues or external foreign exchange debt relative to external foreign exchange reserves, is hardwired into the model (the IMF just released their template, I haven't had to digest it in full). The market access framework was designed to assess the risk of debt distress in countries as poor as Sri Lanka and as rich as Japan, and work for Uruguay as well as the United States. That intrinsically favors some variables over others.

The difference in how the IMF’s two debt sustainability frameworks specify restructuring targets sounds technical, but it has real-world consequences.

Zambia faced with a clear constraint on the NPV of its total external debt. The bondholders may have gamed that framework with early amortizations ahead of the 2027 target, but it did at least try to anchor the level of debt that Zambia would face going forward.

Sri Lanka, whose per capita GDP of about $4,000 is above the low-income country threshold, was evaluated using the same metrics as advanced economies and wasn’t provided an NPV target for external debt to anchor its restructuring.

Talk about path dependence.

That said, the low-income country debt sustainability framework has cliff effects of its own.

Zambia’s debt restructuring had to be anchored entirely by external debt-to-exports because Zambia’s GDP data was under review (GDP itself isn’t reliably measured in most low-income countries). And bondholders accurately pointed out that there was a huge difference between the $15 billion in external debt allowed for countries with weak debt carrying capacity – resulting from low estimated debt-carrying capacity – and the $24 billion in external debt allowed for countries with higher estimated capacity (exports were estimated to be $17 billion).

Zambia started with about $20 billion in external debt (including $3 billion in non-resident holdings of local bonds and about $2.5 billion in multilateral debt) so there was a huge practical difference between a $15 billion target and a $24 billion target. One implied debt relief, the other would have allowed a Sri Lanka-like outcome. The lower target was the right given how little of Zambia’s copper export proceeds actually flow through Zambia’s coffers, but the bondholders were right to note that far too much depended on a single rather obscure calculation done by the IMF.

A bit more common sense in setting targets – while hard to defend in a world where both private and official creditors generally want the results to come out of a model – would actually help. Judgement matters.

There is no Consensus on How to Treat Domestic Debt When an External Debt Restructuring is Required

This is clear from Zambia and Sri Lanka.

In Zambia, the influx of money from non-resident investors in local currency debt (many of the biggest investors also held its foreign currency bonds) provided Zambia a lifeline during the extended period after it defaulted but before it was able to borrow from the IMF. Both out of loyalty – and out of concerns about financial stability – Zambia wanted to exclude the non-resident holders of local debt from its otherwise comprehensive debt restructuring.

That’s fair enough; foreign investors in local currency debt are indeed exposed to different risks than foreign investors in foreign currency debt – they can only exit if they can both sell their bonds and sell the local currency and, in general, they can only exit en masse at a deep discount. There is a bit of automatic risk-sharing. But the non-resident holders had an impact on both the stock of external debt (relevant for the restructuring given that the NPV of external debt was a binding constraint) and, more significantly, on estimates of external debt service.

Their impact on the flow of external debt service was strange.

Suppose non-residents held a $1 billion stock of local currency treasury bills that matured after a year, and they were expected to roll over those bills every year for the next ten years. That $1 billion stock of bills would thus count as $1 billion in the external debt service calculations every year over the next ten years even though the debt could, as a practical matter, only roll off once.

That doesn’t make sense – the stock of external debt bills is best viewed as a one-time drain on reserves, not a continuous draw on fiscal resources.

Zambia's local currency bonds had a bit longer maturity in practice (see Theo Maret). But Zambia ultimately solved this problem with a trick – they assumed the bills will roll off, and thus only impact the debt once. That leaves the deeper questions unanswered.

Sri Lanka, by contrast, didn’t have foreign investors in its local currency debt market. It still ended up doing a “soft” domestic debt restructuring – largely to free up space in the IMF’s calculations to make a better offer to external creditors. To reduce its gross financing need, the central bank swapped out of bills, for example – a transaction that had essentially no impact of Sri Lanka’s true risk going forward but helped reduce its total fiscal financing need.

So, the current frameworks had the effect, even if not the intent, of forcing Zambia to stop issuing local currency debt to foreign investors and of forcing Sri Lanka to restructure debts held by its central bank for no substantive reason.

There is, I believe, a need for a fundamental rethink of the incentives for domestic debt restructuring created by both frameworks going forward.

There is no doubt that domestic debt restructurings are in fact necessary in some cases. In Ghana, for example, domestic fiscal dynamics put the government on a treadmill that it could only stop by changing the terms of its domestic debt. Ghana historically has covered a large part of its fiscal deficit with external bond issuance. But when it lost external market access and didn’t start collecting more tax revenue, it turned to the domestic market. And its issuance outpaced the capacity of the local banks to absorb the debt, and thus it was issuing ever more debt at ever higher rates simply to pay interest on its existing domestic borrowing.

Domestic debt restructuring thus was necessary in Ghana’s case. The dynamics around domestic debt alone were unsustainable – and a domestic debt restructuring was needed no matter what happened to the external debt.

There is still an issue, though, with the IMF’s targets. The low-income country framework allows a country like Ghana to have 40 percent of GDP in external debt and 55 percent of GDP in total fiscal debt at the end of its restructuring and, frankly, it will be hard to squeeze Ghana’s current domestic debt down to 15 percent of GDP. The 55 percent of GDP target thus may require a deeper external restructuring than the 40 percent of GDP external debt to GDP target on its own requires.

That may or may not be the right outcome. I tend to think the 40 percent external debt to GDP level is too high if most of the debt is high-interest rate market debt rather than debts owed to other governments. But there is clearly a major disconnect between the debt to GDP target for Ghana and that of Sri Lanka, which bondholders will be sure to note, especially as "low-income" Ghana has had more recent bond market access than the “market access” Sri Lanka.

The Set of Restructuring Cases it Too Small

Far fewer low-income countries have stepped up for the Common Framework than expected back in 2020.

And fewer middle-income countries have sought restructuring than the market expected last year (about 20 countries have been trading at spreads that imply a high risk of default for some time, and most have paid despite no real near-term prospects of returning to market access).

That’s in part because the Common Framework has been perceived as a roach motel – you check in but don’t check out – that is only useful for countries that have run out of funds and literally have no choice.

But it is also because the IMF has provided financing to a set of countries that are at high risk of default on the IMF’s own metrics – Kenya and Pakistan mostly obviously, but Egypt would fit the bill even if it isn’t an International Development Association (IDA) blend country and thus isn’t on the IMF/World Bank list of countries at high risk of distress.

These countries aren’t clearly insolvent.

External debt of over 50 percent of GDP means effective insolvency for most frontier markets in my book, even if the IMF’s market access debt sustainability model says otherwise in Sri Lanka’s case.

Kenya isn't clearly over the threshold for insolvency. The total external debt of the government is just over 30% of GDP. But Kenya has some very problematic specific debts. Namely, its $2 billion bond maturing in 2024 (a ten-year that was Kenya’s first sovereign issuance), and the $2 billion China Exim Bank LIBOR + 360 bps loan for the standard gage railway. With an 8 percent coupon and steady amortizations (it was a 15-year loan, and the 5-year grace period has expired), its payment terms are now quite onerous.

Do Kenya’s other debts need to be restructured so Kenya gets a comprehensive overhaul of its stock? It can be debated – but there clearly is a case for dealing with the jumbo bond and the high interest rate railway loan by extending maturities at reasonable coupons.

Pakistan’s external debts aren’t that high compared to its GDP, though export revenues are low, which does suggest Pakistan's carrying capacity for external debt (and especially external debt at commercial interest rates) is limited. It also has a potential domestic debt problem – though more because of persistently low revenues than an unusually high domestic debt stock. 

Yet even if Pakistan's external debt doesn't obviously exceed Pakistan's long-term payment capacity, there is still a compelling case for Pakistan to restructure its external debts. It has essentially no reserves, and it thus has more debt coming due than it can really sustain: $10 billion a year in general government “loans” from bilateral and commercial creditors now mature every year. 

Remember that all of Pakistan’s reserves (and then some) are borrowed, and it has already used all its SDRs. Rescheduling the Chinese policy loans coming due in the next few years would help give Pakistan breathing room on the external side to rebuild its reserves – and the interest rate, of course, should be capped to avoid turning a current problem of external illiquidity into a future problem of insolvency.

Egypt is somewhat similar. It has a significant stock of domestic debt and high interest costs relative to revenue. External debt on its own isn’t excessively high.  But it is large relative to Egypt’s reserves (also all borrowed) and the markets are now effectively closed to new external bond issuance at reasonable rates. Proactively terming the debt out has to be an option if neither the Gulf Cooperation Council (GCC) countries nor the IMF are willing to provide long-term financing to anchor fiscal sustainability (the cost of IMF borrowing right now is also an issue).

Argentina also obviously needs to redo the terms of its 2020 restructuring, which were designed for a low-for-long world that no longer exists.

No doubt there are other cases. But the broader point is, in a world of limited new issuance, there isn't a viable equilibrium where the market receives continuous coupons on bonds that are priced for default – particularly if the IMF isn’t willing to chip in with large amounts of net new money.

Where do we Stand?

Zambia looks like it’s close to being resolved, after a lengthy – and at times difficult – debate. But the bondholders look to have scored a better deal (way more cash upfront, a higher coupon) than China's Export-Import Bank, which I suspect will generate problems in future cases.

Sri Lanka will likely try to secure a restructuring deal with its creditors even though it really shouldn’t be in a rush to sort out its debt if its creditors aren't offering terms that provide a fair chance of restoring sustainability. Extracting every penny allowed under the IMF's framework with extra debt service if GDP growth exceeds target is far too narrow a frame if a world where the IMF's framework isn't a good baseline for real sustainability.******

Ghana is likely to be somewhat difficult because of the interaction between its politics and its current debt position – it needs a bit of relief from its external bonds after borrowing way too much both internally and externally during the pandemic.

None of these cases, though, are likely to define a new architecture on their own. In fact, the biggest problem they’ve highlighted is arguably not in the institutions for coordination with China, but rather in the IMF’s own analytic framework for debt sustainability. If Sri Lanka emerges with too much debt, it won’t really be down to China, as the IMF hasn’t insisted that Sri Lanka even try to negotiate serious debt reduction. It is being treated as a case of temporary illiquidity, not of insolvency.

And, well, there is a set of much more consequential cases looming on the horizon that will weigh on the market until there is clarity one way or another.

* The NPV reduction in the bond deal (judged using the IMF's uniform 5 percent discount rate for low-income countries) depends if the NPV is assessed relative to the bond's current legal claim ($3 billion face plus 840 million of past due interest) or relative to the face value of the bond at the time of default ($3 billion). Lazard (two years ago) suggested that comparability of treatment should be assessed according to the face value of claim at the time of default to avoid rewarding high coupon claims. The bondholders didn't agree, and won. But as a result the actual NPV reduction relative to Zambia's pre-default debt stock isn't clear. This is a topic that warrants its own blog.

** The Chinese official creditors' committee announced a deal, on something – though no one seems to know what – just ahead the Marrakech meetings.

*** External private debt too – the Asian financial crisis falls outside the IMF's current conceptual framework.

**** The Sri Lanka targets don't pass muster on the IMF's preferred fiscal terms (look at debt v. revenue) and I have not seen a convincing analysis that Sri Lanka can support external det to GDP of 50 percent or more with a large fraction of its debt amortizing and thus needing to be refinanced in the market from 2027 on.

***** Amortizing fiscal debt + the forecast fiscal deficit

****** See the appendix of Argentina's Fifth and Sixth Reviews Under the Extended Arrangement Under the Extended Fund Facility, Request for Rephasing of Access, Waivers of Nonobservance of Performance Criteria, Modification of Performance Criteria and Financing Assurances Review, 

“The debt fanchart module points to a moderate risk of sovereign stress as in the fourth review. Gross public debt is projected to reach 89.5 percent of GDP at end-2023, around 17 percentage points of GDP higher than projected at the fourth review, reflecting mainly the step devaluation in August and the revision to the growth forecast (GDP measured in U.S. dollars is projected to be 18 percent lower in 2023 compared to the fourth review). Debt is projected to decline to around 69 percent of GDP by end–2028, and the probability of debt stabilization under the baseline continues to be high (above 99 percent). Uncertainty as proxied by the fanchart width is significantly lower than in the fourth review (58 vs. 73 percent) ...The fanchart index remains at a moderate level ... A long-term fan chart analysis points to debt sustainability, albeit with substantial risks. The probability of debt stabilization in a fan chart ending in 2032 is close to 83 percent, lower than at the 5-year horizon but sufficiently high to be consistent with debt sustainability, although with substantial risks.”

******* The Sri Lanka targets don't really pass muster of the IMF's standard fiscal measures (look at debt v. revenue) and I have not seen a convincing analysis that Sri Lanka can support external debt to GDP of 50 percent or more with a large fraction of its debt amortizing, and thus in need of refinancing in the market from 2027 and beyond.