This edition's structure assumes a connection. The section is called The Ballot and the Bill, and the premise is that elections and borrowing costs are one subject. When we drew up the article list, this piece carried the working argument that an election year is an expensive year to go to market.

We went looking for the evidence. It is not there, at least not in the form the assertion takes, and the honest thing to do with a premise that fails is to publish the failure rather than quietly write around it.

The standard reference is Block and Vaaler, published in the Journal of International Money and Finance in 2004, which examined sovereign credit rating actions across nineteen developing countries and daily secondary market bond spreads across eleven countries covering fourteen presidential elections.

Their finding on ratings supports the conventional view. Rating agencies downgrade developing country sovereigns more often in election years, by roughly one notch. Of 64 downgrades in their sample, 30 fell in election years, a difference significant at the one per cent level.

Their finding on spreads does not. Average spreads in the sixty days before an election were higher than in the sixty days after. But within that pre-election window the slope was negative: spreads fell as polling day approached, as electoral uncertainty resolved. The estimated slope was minus 0.0025.

Two honest caveats before anybody builds a position on this. The paper's text reports that slope as significant at the one per cent level while its own table marks it at ten per cent, and we flag the discrepancy rather than choosing. And the sample runs from 1987 to 1999, which is a different world: before the euro, before the commodity supercycle, before quantitative easing, and before most of the frontier sovereigns in this edition had ever issued a bond.

Later work narrows the claim further rather than restoring it. Vaaler, Schrage and Block, writing in the Journal of International Business Studies in 2005, found that elections do not unconditionally raise spreads at all: pre-election spreads widen when a market-friendly incumbent looks likely to lose to a left-leaning challenger, and narrow when the reverse is expected. The direction depends on who is expected to win, not on the fact of an election.

And an IMF working paper by Csonto and Ivaschenko found that short-run movements in emerging market spreads are dominated by global risk aversion and United States interest rates rather than by any discrete election-year jump.

So the premise fails as stated. Elections do not reliably make borrowing more expensive. Something else does, and the figures on it are stark.

The World Bank's International Debt Report 2026, published on 4 November, puts the total external debt stock of low and middle income countries at 9,445 billion dollars at the end of 2025. That is up 5.5 per cent on the year and up 59.7 per cent since 2015.

Debt service paid in 2025 came to 1,398.8 billion dollars: 973.3 billion of principal and 425.5 billion of interest. China alone accounted for 419.8 billion of that, about thirty per cent, so the figure for everybody else was 979 billion, which was down 2.5 per cent on the year.

Sub-Saharan Africa's external debt stock rose 10.8 per cent in a single year to 1.02 trillion dollars. Low income countries as a group rose 8.9 per cent.

Between 32 and 33 of the low income countries assessed under the IMF and World Bank debt sustainability framework are in debt distress or at high risk of it. Of the 33 in the September 2025 count, nine were already in distress and twenty-one of the total were African.

Here is the number that explains more than the election calendar ever could.

In 2025, new public and publicly guaranteed commitments from private creditors to IDA-eligible countries carried an average interest rate of 8.2 per cent. The equivalent rate from official creditors was 2.7 per cent.

That 8.2 per cent is the highest level since 1989. The gap between what private and official lenders charge has widened to 5.5 percentage points, against 3.4 points in 2016. For IDA-eligible borrowers going to the bond market specifically, the average coupon on new commitments in 2025 was 9.4 per cent.

Set that beside the maturity question and the picture gets worse, because the terms are deteriorating on both axes at once.

Coupons rose from 8.0 per cent in 2025 to 8.8 per cent in the first seven and a half months of 2026. Average maturity shortened from 10.4 years to 8.9. Paying more for less time is the definition of a borrower losing bargaining power, and it happened without any election.

When conventional issuance becomes unavailable or prohibitively expensive, sovereigns do not stop borrowing. They borrow differently, and the way they have chosen is the most significant development in this data.

Between 2024 and 2026, five sovereigns executed collateralised total return swaps: Angola, Argentina, Colombia, Nigeria and Senegal. The facilities ranged from one billion to 9.3 billion dollars, with maturities of one to three years, and collateralisation ratios between 133 and 198 per cent.

Read that ratio carefully. A government raising a billion dollars is posting up to two billion in securities against it, on a one to three year term, because the alternative was a bond market that would not have it at a price it could pay.

This edition has already met two of these arrangements from the inside. Nigeria's five billion dollar facility with an Abu Dhabi bank, collateralised at roughly 133 per cent in naira securities, converts currency weakness directly into a demand for hard currency. Angola carries 2.5 billion dollars across two such swaps maturing in 2028, at 148 and 140 per cent collateralisation.

These are not bonds. They are secured, short-dated, and they sit outside the architecture built over forty years to handle sovereign distress. If any of them is called in a stressed market, there is no restructuring framework that knows what to do with it.

One genuinely good figure sits in this report and it deserves to be stated properly.

Net transfers on long-term external debt to low and middle income countries, which is disbursements minus everything paid back, turned positive in 2025 at plus 45.4 billion dollars. That reverses three consecutive years of outflows including minus 118.4 billion in 2024. Excluding China the 2025 figure is plus 83.1 billion.

The composition matters more than the headline. The positive number is driven by public and publicly guaranteed debt at plus 49.9 billion, which largely means multilateral lending. Private non-guaranteed flows were still negative at minus 4.5 billion. And net transfers from official bilateral creditors were negative in 2025: governments lending to other governments took back more than they put in.

The institution keeping the aggregate positive is the multilateral system. That is worth remembering in a year when the countries that fund it are cutting their development budgets, as this edition has already documented.

The G20 Common Framework was established in November 2020 to coordinate debt treatment for low income countries. In the six years since, four countries have received an official treatment through it: Chad in 2022, Zambia, Ghana, and Ethiopia as the fourth in 2025.

That is not a throughput problem that better administration fixes. Four cases in six years against thirty-two countries in or near distress is a ratio that tells a government in difficulty to avoid the process rather than enter it, which is precisely why collateralised swaps have become attractive.

The specific maturities falling in the election year are smaller than the framing of this section implied, and we should say so.

Nigeria has a 1.5 billion dollar Eurobond at a 6.5 per cent coupon maturing on 28 November 2027, which is its only Eurobond maturity between late 2026 and 2029, and it falls ten months after the January election. Kenya has a final amortising instalment of 108 million dollars in May 2027, having cut that from 300 million through its buyback. Angola has no bullet Eurobond maturity in 2027 at all; its wall is 8.3 billion dollars of external amortisations in 2028. Argentina has semi-annual step-up instalments in January and July rather than a single date. Papua New Guinea's only Eurobond matures in October 2028.

In other words, the countries voting in 2027 have mostly already dealt with 2027. The liability management operations documented elsewhere in this edition worked. What they did was move the problem into 2028 and 2029, which is a different article and probably next year's.

Having checked, our position is this.

The election calendar is not what makes 2027 expensive. The documented evidence on election-year spreads is mixed at best, conditional on partisan expectations, and drawn from a sample three decades old. We are not going to assert a relationship the literature does not support, and we have adjusted the premise of this section accordingly.

What makes borrowing expensive is a structural split in the price of credit: 8.2 per cent from private lenders against 2.7 from official ones, a gap that has widened by more than two percentage points in a decade and now sits at its highest private level since 1989. That gap exists whether or not anybody votes.

And the thing to watch in 2027 is not a maturity date. It is whether any of the five collateralised swaps comes under stress, because that is where the new borrowing has gone and it is the one part of the system nobody has a procedure for.

First, the terms on new African issuance. Coupon up and maturity down was the 2026 pattern. If it continues into 2027 the squeeze is structural rather than cyclical.

Second, bilateral net transfers. They were negative in 2025. With donor budgets contracting, a second negative year would mean official creditors as a whole have become a net drain rather than a net source.

Third, any fifth or sixth case entering the Common Framework. Four in six years is the current rate. Any acceleration would be the first evidence that the architecture works at scale.

Fourth, the collateralised swaps. Angola's 2.5 billion matures in 2028, Argentina's three billion in 2028. Watch for the first one that is rolled rather than repaid, because a rolled swap is a swap that could not be repaid.

The records room in a finance ministry does not look like a crisis. It looks like shelving. The obligations sit in order, each one serviceable on its own terms, and the question is never whether any single file can be paid. It is what happens when the money to pay them costs three times what it costs the country lending it.

Data notes: External debt stocks, debt service, creditor composition, net transfers, interest rates on new commitments, sub-Saharan African bond terms and the total return swap data are all from the World Bank, International Debt Report 2026, published 4 November 2026, covering 119 low and middle income countries through the Debtor Reporting System; all figures are observed outturns for 2025 unless stated, and the 2026 African bond terms cover 1 January to mid-August 2026. Forward-looking aggregate debt service projections for 2026 and 2027 are in the statistical appendix of that report rather than the chapter consulted, and are not estimated here. Debt distress counts are IMF and World Bank debt sustainability framework classifications as cited in 2026 publications; two different counts are reported and not reconciled. Common Framework caseload from the Paris Club Annual Report 2025. Election-year spread evidence from Block and Vaaler, Journal of International Money and Finance, 2004, verified against the published paper, which covers 19 countries and 18 presidential elections for ratings from 1987 to 1998 and 11 countries and 14 elections for spreads from 1994 to 1999; the paper's text and its own table report different significance levels for the pre-election slope and both are noted. Further evidence from Vaaler, Schrage and Block, Journal of International Business Studies, 2005, and Csonto and Ivaschenko, IMF Working Paper WP/13/164. Country maturity schedules from national debt management offices, S&P Global Ratings and Fitch Ratings as cited. The J.P. Morgan EMBI index level is proprietary and is not quoted.