The G20's debt service suspension initiative (DSSI) is due to come into effect on 1 May. The DSSI is intended to provide a time-bound suspension of debt service payments owed to official bilateral creditors for the world's poorest countries that request forbearance, and private creditors are encouraged to do the same.
Private creditors are expected to participate on a voluntary basis on comparable terms, upon request from borrower countries, although there are a number of practical considerations that need to be ironed out. Countries will also need to weigh up the relative costs and benefits of seeking relief from bondholders, balancing immediate cash flow savings with higher potential costs further down the line, as well as possible risks, inter alia, related to future market access and the potential adverse impact on their sovereign rating.
Pakistan, according to the FT, may be the first country to seek such bilateral debt service relief, although there is no suggestion in the reports that the authorities are seeking the same relief from bondholders at this stage. The article says the authorities will not seek a repayment freeze from the private sector at present.
Signals from the bond market
Do bond prices tell us anything useful about the impact and implications of DSSI for eligible countries? There are 23 countries that have foreign bonds outstanding out of the list of countries that are eligible for DSSI (comprising a total of 76 IDA-eligible countries and 47 LDCs as defined by the UN). And within the group of eligible countries with bonds, what do bond markets tell us about the impact, if any, of DSSI on Sub-Saharan Africa (SSA) countries relative to others?
To look at the bond market signals, we divide sovereign issuers into three groups: one cohort being those from countries eligible for DSSI in SSA, a second cohort being eligible issuers outside Africa (non-African eligible issuers), and a third cohort comprising a selection of other small EM/frontier countries that are not eligible for DSSI (non-eligible issuers), which have similar sovereign ratings to the eligible issuers (a sort of control group). There are 13 countries in the first (SSA) cohort, 8 in the second (non-SSA) cohort (we don't include Fiji and Laos bonds as pricing isn't reliable), and we have 17 countries in our third cohort (control group).
We see that yields vary widely across the different cohorts, ranging from 8-33% among eligible SSA countries, 5-17% among eligible non-SSA countries, and 4-30% in the control group of small EM countries that aren't eligible for DSSI (Table 1, panels A-C). To generalise, some of the higher yielders and more distressed names are probably known for having pre-existing vulnerabilities (eg Angola, Suriname, Tajikistan and Zambia) while some other high yielders are new, due to the economic crisis caused by the coronavirus pandemic (eg Maldives and Sri Lanka), so some countries may be more affected by specific debt relief initiatives than others.
However, we should caution that comparisons of bond yields across these groups is made imperfect for three reasons. First, eligible countries span a range of sovereign ratings (from BB- down to CCC). However, the average ratings of each of our cohorts are similar, at around the B level. Second, the bonds have different maturities/durations. We take representative medium-term (10 year) bonds where available for each issuer, or the sole bond where a country only has one issue. This means that yields can differ markedly, even for issuers with the same rating. Third, the size of outstanding bond issues and bond stocks varies greatly and technical or liquidity features makes comparing yields potentially misleading.
We observe that, at first glance, there may be a yield premium on eligible SSA bonds, on an average basis, compared to bonds issued from countries that aren't eligible (the control group), which may not be a surprise, and to non-SSA eligible bonds, which may be more of a surprise. There isn't much discernible difference between average yields on bonds issued by non-SSA eligible countries and those issued by ineligible countries (the second and third cohorts), which may be a surprise. The African yield premium on our calculations (average yield on the first cohort compared to average yields in the other two cohorts) may amount to some 200bps.
We find that:
1. The average yield on SSA eligible countries (the first cohort) is 13.1%, with a median of 11.6%, although the average falls to 11.4% if Zambia is excluded (a special case) and is even lower, at 10.4%, if Zambia and Angola are excluded. But we see a clear delineation within SSA bonds, with a small group of relatively strong countries with low yields (led by Senegal, and followed by Cote d'Ivoire, and Kenya; which arguably have more diversified export bases, and have achieved a strong track record of economic growth and generally sound economic policies), two names at the other end of the spectrum which are highly distressed (Angola) or already seeking to restructure (Zambia), and a set of weaker countries in the middle, with yields around 10-13%, which are either single commodity exporters (Nigeria, Mozambique, Republic of Congo) or more diversified but with weaker fundamentals (Cameroon, Ethiopia, Ghana). Within this classification, placing Benin and Rwanda is difficult (they share more of the characteristics of the first group than the middle group in our opinion.
2. The average yield on non-SSA eligible countries (the second cohort) is 10.6%, with a median of 9.6%, although the average falls to 9.6% if Maldives is excluded (yet there is no good reason, except its small size, for excluding it, as it seems a classic candidate to benefit from this treatment) and is even higher, at 12.0%, if we include only those issuers rated B or below, or that are unrated (ie we exclude BB- rated Honduras and Uzbekistan – again no good reason for this, except their low yields distort the average, and one could question why especially a BB-rated sovereign would seek debt relief which could jeopardise its rating and future market access).
3. The average yield on non-eligible countries (the third cohort, control group) is 10.4%, with a median of 8.3%, although the average falls to 9.1% if Suriname is excluded (yet there is no good reason for excluding it), and is even higher, at 11.3%, if we include only those issuers rated B or below, or that are unrated (ie we exclude BB/BB- rated Armenia, Dominican Republic and Vietnam).
On that basis, comparing our second and third cohorts, the average yield is roughly similar based on those with an average rating of B-rated or below (12% and 11.3%, respectively), or based on an average that removes outliers (9.6% and 9.1%).
However, problematically (for our thesis), if we remove the outliers, the average yield in SSA countries, at 10.4%, is lower than the average yield of B-rated and below sovereigns in both the non-SSA eligible group (12%) and the control group (11.3%).
We note also that the average yield for SSA bonds would be even higher if we adjusted the constituent yields better to match maturity and currency denomination across the issues. On the former, three of the bonds are quite short (with maturities over 2023-2026). On the latter, Benin is in EUR so the USD-equivalent would be higher. Having said that, the same arguments about yields being higher upon maturity-matching would apply to the non-SSA eligible bonds too (this group also has two short ones, 2022 and 2024).
To put this another way, SSA may be paying the price for debt service suspension initiatives in terms of giving it a higher cost of capital compared to other markets. Moreover, the impact on SSA yields may be amplified by two other factors; after all, if the DSSI really seeks NPV neutrality, the deferral of essentially just one coupon during the suspension period (May-December), which may amount, on average, to generally around 3-4pts, might only be expected to have a modest impact on prices, and one that is relatively small beer given price falls to date. Two other factors are, first, the concern that the official sector may seek to deepen the treatment under DSSI, either in terms of a needing a longer suspension period or moving towards an actual restructuring (which for some countries, may prove necessary at a later stage in any case). There might also be concern among investors over the calculation of NPV neutrality. That said, this factor should also be relevant to the second group (non-SSA eligible countries). Second, the existence of other calls for debt relief specifically in Africa that go further than DSSI may mean investors think this is the thin edge of the wedge. Indeed, if African leaders are calling for debt relief, we cannot be surprised if they take advantage of it if it's on the table, and we cannot be surprised if markets begin to factor this in. Put yet another way, it might be that African countries may be seen by investors as more likely to seek DSSI than their peers. Leaders (past and present) in some countries such as Angola, Ghana, Ethiopia and Rwanda have called for bilateral debt relief, although some countries such as Nigeria have stated they would not look to touch their bonds.
Pakistan offers an interesting comparison, as while it seems the authorities there have requested DSSI from bilateral creditors (seemingly ahead of SSA countries), its bonds haven't really reacted (with yields still about 9%), perhaps because investors think the authorities won't seek comparability from bondholders, as they have indicated, despite its government being one of the key proponents of the initiative. Pakistan may judge the reputational impact and potential loss of market access as too great a cost, being a regular issuer over recent years and with a longer history of issuing (having restructured before back in 1999), and be keen to protect its sovereign rating (which could take time to recover if it does fall into default by seeking the same relief from bondholders). And the relative share of the debt stock and potential savings may support its intention not to include the bond. According to the article, savings from debt service relief on bilateral debt could amount to US$1.8bn. Pakistan has around US$24bn in bilateral debt (about a third of its MLT external debt), according to central bank figures, with US$10.9bn in outstanding Paris Club debt and US$13.2bn in other bilateral debt. According to the FT, US$1bn is owed to non Paris Club members of the G20 including China. Pakistan has US$5.3bn in foreign bonds outstanding (about 8% of MLT external debt), with annual interest payments of US$362mn on our calculations.
Still, for African sovereign issuers, if this regional premium exists because of debt relief efforts, it could be important for all the issuers in the region, as spillovers (real or perceived) affect borrowing costs for all its issuers, potentially limiting market access and reducing investor demand. Or maybe this regional premium (to the extent it exists) simply reflects Zambia's announcement that it plans to restructure, which sent its own negative signal to investors elsewhere in the region. However, Zambia's vulnerabilities and eventual restructuring was well flagged and may be seen as an isolated case, with little spillover; this may have been the case were it not for the fact that it has been joined by others, albeit outside the scope of DSSI, with Ecuador deferring coupons until August under a consent solicitation and Seychelles announcing it was seeking to restructure, highlighting the problems facing some countries in this environment, especially those with pre-existing vulnerabilities, some of whom are in SSA. Moreover, investors may be concerned that some borrowers may use international debt relief efforts as cover to mask opportunistic calls for restructuring.
Or maybe it's just fundamentals: Other relevant factors
The question is, if there is an Africa yield premium is it because of debt relief initiatives? SSA yields may on average be higher than peers for other reasons and isolating the specific impact of debt relief initiatives is difficult.
First, SSA may comprise a disproportionate share of countries that are dependent on a single or just a few commodity exports (energy – oil and gas, metals or soft commodities) relative to the other two cohorts here. The impact of lower oil prices on the balance of payments and debt sustainability of oil exporters in SSA has been particularly acute, although – with the possible exception of Angola – arguably in line with the experience of other oil exporters that aren't eligible for DSSI (eg Gabon and Iraq in our control group), or in line with other single-commodity exporters among the eligible non-SSA group of countries (eg Mongolia, on a maturity matched basis).
Second, coming into the crisis, SSA (as a region) may have had weaker fundamentals (higher or rising public debt ratios, lower reserves, large external financing needs, and more limited policy space) compared to other regions, so that yields may have risen disproportionately more (Figure 1).
Figure 1: Public debt and bond yields in SSA
Source: Tellimer Research, IMF, Bloomberg. X-axis = Public debt/GDP (%) projected for 2020 from IMF SSA REO. Y-axis = yield on representative US$ bond yield. *Benin is EUR.
Moreover, some of these macro challenges and vulnerabilities are not unique to SSA countries that are eligible for DSSI, and are not even limited only to those countries that are eligible for DSSI. Other countries also face specific challenges caused by the global crisis, evident from their high yields, arising from potential near-term refinancing problems or economic shocks that impact key sectors (such as tourist destinations).
For instance, among non-SSA eligible countries, we have Maldives (indicative yield 17%), impacted by the negative shock on tourism earnings and with potential refinancing challenges ahead of bond maturities coming due (2022-2023), and Tajikistan (15.5%), where more challenging financing conditions may frustrate its large dam-related financing needs while slower demand growth could reduce demand for the energy that the dam will produce. There are also some high yielders in the control group. Sri Lanka (16%) may, like Maldives, also be impacted by the negative shock on tourism earnings and has a bond maturity in October, while Suriname (30%) may also suffer from tighter liquidity conditions.
Table 1: Comparison of bond yields in DSSI eligible countries and other EM/frontier markets
|Angola||ANGOL 8 11/26/29||42.8||23.0||CCC+|
|Benin||BENIN 5 3/4 03/26/26 EUR||83.5||9.5||B+|
|Cameroon||REPCAM 9 1/2 11/19/25||87.8||12.6||B-|
|Congo Rep||REPCON 6 06/30/29||80.3||11.7||Caa2|
|Cote d'Ivoire||IVYCST 6 3/8 03/03/28||90.6||8.0||B+|
|Ethiopia||ETHOPI 6 5/8 12/11/24||88.3||9.8||B|
|Ghana||GHANA 7 5/8 05/16/29||75.7||12.1||B|
|Kenya||KENINT 7 1/4 02/28/28||90.0||9.1||B+|
|Mozambique||MOZAM 5 09/15/31||72.7||11.6||Caa2u|
|Nigeria||NGERIA 7.143 02/23/30||72.7||11.9||B-|
|Rwanda||RWANDA 6 5/8 05/02/23||91.7||9.9||B+|
|Senegal||SENEGL 6 1/4 05/23/33||88.9||7.6||B+|
|Zambia||ZAMBIN 8.97 07/30/27||34.8||33.2||CCC|
|Average ex Zambia||11.4|
|Average ex Zambia and Angola||10.4|
|Grenada||GRENAD 7 05/12/30||87.0||10.7||-|
|Honduras||HONDUR 6 1/4 01/19/27||92.6||7.7||BB-|
|Mongolia||MONGOL 8 3/4 03/09/24||98.3||9.3||B|
|Pakistan||PKSTAN 6 7/8 12/05/27||87.4||9.2||B-|
|Papua New Guinea||PNGIB 8 3/8 10/04/28||90.9||10.0||B|
|Tajikistan||TAJIKI 7 1/8 09/14/27||64.0||15.5||B-|
|Uzbekistan||UZBEK 5 3/8 02/20/29||102.2||5.1||BB-|
|Average ex Maldives||9.6|
|Average of B rated and below, and unrated||12.0|
Panel C: Selected countries not included in DSSI (control group)
|Armenia||ARMEN 7.15 03/26/25||106.4||5.6||Ba3|
|Bahrain||BHRAIN 7 10/12/28||96.9||7.5||B+|
|Barbados||BARBAD 6 1/2 10/01/29||88.7||8.2||B-|
|Belarus||BELRUS 6.2 02/28/30||90.2||7.6||B|
|Belize||BELIZE 4.9375 02/20/34||51.5||12.3||CCC|
|*Bolivia||BOLIVI 4 1/2 03/20/28||81.0||7.8||B+|
|Costa Rica||COSTAR 4 3/8 04/30/25||79.8||9.6||B+|
|Dom Rep||DOMREP 5.95 01/25/27||87.9||8.3||BB-|
|Egypt||EGYPT 7 1/2 01/31/27||95.8||8.3||B|
|El Salvador||ELSALV 6 3/8 01/18/27||77.8||11.2||B-|
|Gabon||GABON 6 5/8 02/06/31||69.4||11.7||Caa2|
|Iraq||IRAQ 5.8 01/15/28||71.9||11.4||B-|
|Jamaica||JAMAN 6 3/4 04/28/28||99.7||6.8||B+|
|*Sri Lanka||SRILAN 7.55 03/28/30||58.6||16.0||B|
|Suriname||SURINM 9 1/4 10/26/26||42.0||30.1||CCC+|
|Ukraine||UKRAIN 9 3/4 11/01/28||98.0||10.1||B|
|*Vietnam||VIETNM 4.8 11/19/24||104.4||3.7||BB|
|Average ex Suriname||9.1|
|Average of B rated and below, and unrated||11.3|