African Sovereign Refinancing and Banks: The Burden Returning Through Business Payments

Read African financing from both the government and bank balance sheets

The IMF’s October 15, 2026 African outlook highlights domestic bank financing of government and the transmission of public arrears to businesses and banks.[1] An assessment that ends when a bond is sold misses where the burden remains. Government refinancing and a business’s ability to pay suppliers need not improve on the same date.

Suppose a government sells new bonds to a bank to repay maturing debt. The bank acquires a new government claim, although the net change after redemptions needs separate checking. Its deposits and equity do not necessarily increase on matching terms. If the government instead postpones payments to suppliers, those businesses lose cash they would have used to repay banks or buy materials. Public financing should therefore be examined where bond issuance and payment delays meet on the same bank balance sheet, rather than treating them as unrelated events.

SG Group View: follow the burden that remains after refinancing

The central question for SG Group is whether the funding that supports government payments can continue to support business activity. Holding government bonds is not inherently dangerous: they earn interest and may serve as financing collateral when eligible. The difficulty arises when claims connected to the same government accumulate through several channels and alternative sources of payment become scarce when that government is under pressure. Holdings should be assessed alongside funding maturities, the ability to sell assets and the capacity to absorb losses.

Three distinctions guide the analysis. First, bond issuance is different from the additional money a government can spend. Second, a bank’s ability to retain an asset is different from its ability to turn that asset into cash on the required date. Third, unpaid government bills can return through business loan repayments and tax receipts. Together these show the conditions under which domestic borrowing can relieve a shortage of external funding. Conversely, stable deposit growth and continued business lending weaken the case for pessimism based on rising bank bond holdings alone.

Bond issuance and additional spending capacity are different quantities

Let M be maturing principal and B new issuance. Government obtains B−M in additional funds, excluding other transactions; the full issue amount is not new spending capacity. A bank reinvesting redemption proceeds may make a large auction purchase with little change in holdings. Here the next question is how funding clears public arrears or replaces supplier bridge loans. Issuance, bank holdings and business collections need to be connected over the same period.

Where does a bank find the money to finance government?

A bank cannot assess a government bond separately from the funding behind it. Deposits, borrowing from other financial institutions and sales of other assets have different costs and repayment dates. Identical holdings can imply different liquidity positions depending on whether the bank used existing cash, received new deposits or increased short-term borrowing. A high coupon does not assure a large profit when funding costs are also rising: the margin depends on the difference. The asset’s yield alone cannot establish whether the position is sustainable.

Payments can also affect the banking system differently from an individual bank. A bond purchase may initially create a government deposit, which later moves to households and businesses as the government spends. If recipients use another bank, settlement funds leave the purchasing bank. Domestic expenditure does not guarantee that money returns to the same institution. Government accounts, recipients’ accounts and interbank settlement must be linked to locate potential cash shortages; deposit growth across the system does not automatically cover every bank’s shortfall.

Related analysis: Read bank lending and corporate refinancing — Background on lending conditions, distinct from the fiscal feedback loop.

A saleable asset is different from a bank that can absorb losses

Liquidity asks whether cash can be obtained by the payment date. Capital asks whether enough value remains to support liabilities after an asset loss. Borrowing cash against government bonds can help a bank settle payments, but that borrowing does not eliminate a deterioration in the bonds’ value or expected recovery. Adding a liability to get past a payment date and replacing lost asset value require different responses.

This distinction matters when bank difficulties return to government. A bank needing temporary cash and one needing capital to cover losses imply different purposes and costs for public support. Capital injection is not always the least costly response to the former, while repeated short-term lending can leave the latter’s underlying gap unresolved. Confusing the two can prolong support, particularly when fiscal capacity is scarce. The choice depends on asset recovery, expected funding outflows and collateral terms, not on the mere fact that a bank holds government bonds.

Different borrowers do not always mean different underlying risks

A bank may lend to government, a construction company and a power utility without achieving as much diversification as the number of borrowers suggests. Suppose the contractor’s receivables are unpaid public works bills and the utility is waiting for government compensation. Delayed public payments can then affect bond values, business repayments and deposit withdrawals at the same time. A measure of direct government bond holdings is useful, but does not capture every exposure dependent on government finances.

Nor should every business dealing with government be assigned the same risk. Private customers, export earnings, cash reserves and the ability to switch counterparties can provide different periods of resilience. The relevant test follows the ultimate payer and alternative funding when a payment is delayed, rather than simply adding industry labels. Connecting a bank’s sectoral loan breakdown to its borrowers’ customers and collection periods helps distinguish exposures likely to deteriorate together from those that can behave independently.

An unpaid public invoice can reappear as business borrowing

A company’s wage and materials bills do not wait for a government payment. Working-capital credit can bridge that delay, leaving the bank indirectly financing a public bill. An increase in such lending should not immediately be read as stronger investment. Borrowing to bridge a payment gap has different implications for future production and earnings from borrowing to expand capacity. Collection periods and loan purposes help reveal public payment delays hidden within growing business credit.

Bridge finance assumes that payment will arrive and repay the loan. Repeated postponement can turn successive short-term renewals into a permanent source of funding. Stopping renewal may interrupt the company’s payments, while continuing it ties up the bank’s resources. The absence of a formally overdue loan can obscure this change. Contractual maturities, actual collection dates and renewal histories distinguish an isolated extension from funding that has become effectively long term.

Follow payment delays beyond the first supplier

A public contractor awaiting payment may respond before missing a bank repayment. It might buy fewer materials, delay subcontractor payments or cut overtime. Cash shortages can therefore reach small businesses to which the first bank has no direct exposure. Examining only the contractor may suggest little trouble while its bank loan remains current. Following who waits for payment and who reduces production reveals how pressure can surface elsewhere first.

The same unpaid bill must not be counted repeatedly. Adding the government’s unpaid amount, the contractor’s unpaid subcontractor bill and the subcontractor’s bank borrowing can count financing needs generated by one delay several times. Tracking each entity is useful when asking where cash is needed. Measuring total government arrears requires linking the invoices and avoiding double counting. The location of a cash blockage and the ultimate loss are different measurement tasks.

Bond interest does not translate one for one into lending capacity

Government interest payments support both bank income and cash, but the entire amount is not necessarily available for new business loans. Deposit interest, operating expenses and maturing borrowing also require cash. A loss provision is not itself a cash payment, but can reduce earnings or capital and affect lending decisions. Higher government yields can improve bank income while changing the benchmark against which a business loan is judged. If the gap between what a company can pay and what the bank requires narrows, lending decisions can change through maturities and collateral terms as well as price.

Weak business lending cannot all be attributed to attractive government bonds. Poor orders can reduce borrowing demand even when banks have funds. Applications, approvals, unused commitments and offered terms are needed alongside loan balances. A crowding-out explanation requires evidence that firms wanted financing but could not obtain it because of funding or return constraints. Two balance-sheet stocks moving together do not establish that causal relationship.

Domestic-currency borrowing still has a maturity schedule

Domestic-currency debt can reduce foreign-funding needs while leaving a repayment schedule. Portfolios with the same average maturity can have dispersed or concentrated payments. The relevant interaction here is whether sovereign maturities, supplier collections and bank funding dates coincide. When government refinancing and bridge-loan renewal fall together, a bank needs funding for both. Separate average maturities do not reveal that overlap.

Related analysis: Emerging-market foreign funding and maturities — Distinguish domestic refinancing from foreign-currency repayment.

An improvement for government need not be an improvement for banks

Changing interest or principal terms can reduce a government’s future budget burden. For banks holding the debt, the same change can reduce receipts or asset value. Government savings do not establish that the region’s overall burden has disappeared. If bank losses lead to capital needs, reduced lending or public support, some of the fiscal relief can return under another budget heading. Debt changes therefore need to be examined across the borrower’s and holders’ accounts.

This does not imply that debt adjustment should always be avoided. Trying to maintain unsustainable payments can generate more government arrears and weaken bank assets, so delay also has a cost. The relevant comparison is not merely revised terms against an unchanged contract on paper, but against feasible payments and the consequences of continued delay. Planning for bank capital and business payments alongside debt adjustment permits comparison without hiding where the burden moves.

How bank support can return pressure to government

Public bank support can take the form of a repayable loan, a guarantee or a capital injection. The amount advanced is not necessarily the final loss, and a guarantee is not an immediate cash payment. Each can nevertheless affect future fiscal capacity. Committed payments should be distinguished from contingent ones, while recoveries or subsequent sales of equity can change the final cost. Adding announced support amounts can misstate both cash requirements and eventual expense.

The value of a guarantee also depends on the guarantor’s capacity. If the government is both the payer behind bank assets and the institution expected to rescue the bank, those promises cannot be added as independent safety nets. A guarantee reallocates a loss without necessarily increasing the resources available to meet it. Distinguishing fresh external capital or reliable revenue from additional promises backed by the same resources makes the strength of support easier to assess.

Connect sovereign holdings to the capacity to absorb a loss

Government claims as a share of bank assets and banks’ holdings as a share of government debt use different denominators. The first measures a bank’s dependence on government; the second measures government’s dependence on banks as buyers. One can be high while the other is not. Substituting one for the other under the shared label of bank sovereign holdings confuses the direction of dependence. When both are high, alternatives on both sides deserve attention.

A simple independent identity helps explain the connection. Let A be total assets, K equity, E the relevant government claims and h the loss fraction on E. Holding everything else fixed and assuming the loss reduces assets and equity equally, and provided remaining assets A−hE are positive, the post-loss equity ratio is (K−hE)/(A−hE). This is neither an estimate for an actual bank nor a regulatory capital ratio. It shows why a loss that looks small relative to assets can be large relative to the smaller equity base. Claims and capital must be measured on a compatible basis.

Related analysis: Fiscal policy and the term premium — A separate channel linking bond supply to asset valuation.

Stable reported values can coexist with changing collateral capacity

Recognised losses, sale prices and borrowing capacity against collateral need not change together; accounting classifications, contracts and facility terms matter. The interaction here arises if both bonds and supplier bridge loans become difficult to turn into cash. A bank unable to collect from a supplier awaiting government payment, and unable to raise funds against bonds, needs alternatives. This is a conditional combination, not evidence that all regional banks face additional collateral demands.

Take growing private credit seriously as counterevidence

The observation in Figure 5 does not fit a simple claim that more government financing must reduce private credit.[2] Growth in one part of bank assets and contraction in another require separate evidence. A wider funding base can allow lending to both government and businesses. An assessment of pressure must retain actual observations pointing in the opposite direction.

Possible explanations for simultaneous credit growth include more deposits or external funding, a different allocation of bank resources and recovering business demand. Establishing their actual contributions requires separate evidence. Nor does growing lending prove that every willing borrower obtained the amount it needed. The appropriate response is not to add conditions merely to dismiss contrary evidence, but to revise the certain claim: public and private credit can grow together, so an assertion of crowding out needs additional observations.

Related analysis: Household support costs and comparison conditions — A related analysis of support endpoints and national budgets.

Identify the fiscal link rather than infer it from loan growth

One test compares firms heavily dependent on public customers with firms serving other customers. Longer collection periods and more short-term borrowing in the first group during government payment delays would support the proposed channel. Differences in industry, size or seasonality can still provide alternative explanations. Comparing earlier trends among similar firms and connecting public payment dates to borrowing dates helps separate government arrears from a general downturn.

At the bank level, lending terms can be compared across institutions with different government exposures at the same date. Size and deposit bases can explain differences, however. A highly exposed bank with stable funding that continues lending would challenge a simple ranking of deterioration by exposure. Funding sources, maturities and business customers need matching to test whether another factor drives both holdings and lending. The objective is to locate the transmission, not merely rank banks by their sovereign portfolios.

International transmission runs through import payments and investment

If a company awaiting public payment postpones imported machinery or materials, the pressure reaches foreign suppliers. Exporters may see longer collection periods, a shift to advance-payment terms or smaller orders. Regional growth can remain steady while companies selling to public contractors experience different conditions. Knowing whether the ultimate customer is public or private, and how many payment intermediaries stand between buyer and seller, makes international sales and cash forecasts more specific than country labels alone.

More bank holdings of government debt do not immediately imply capital flight from global markets. Securities may simply move between domestic owners; alternatively, new external funding may increase both deposits and holdings. Cross-border effects require a distinction between new financial flows and changes in asset ownership. Delayed import payments, investor bond sales and banks’ foreign-currency funding can move in different directions and on different schedules.

How does resumed government payment move through the system?

Clearing government arrears need not immediately increase business borrowing. A recipient may use the cash to repay bridge loans, reducing outstanding credit even as the bank recovers funds and the company saves interest. Reading that decline as worsening credit would reverse the interpretation. After public payments resume, repayments of bridge loans, new investment lending and payments to suppliers should be tracked separately. Recovery and fresh lending have different functions.

If government funds the payment by selling more bonds to banks, a business claim may effectively be replaced by a government claim. The company’s cash position improves without necessarily reducing the banking system’s dependence on government. Clearing arrears through stable revenue or external grants creates a different combination of recovered bank cash and new sovereign exposure. The result depends on the funding source and assets left behind, as well as the amount of arrears paid.

Do not turn a research stress test into a current banking crisis

A June 2026 IMF working paper studies banks’ sovereign exposure and capital effects under assumed changes to debt terms. Its authors’ views do not necessarily represent IMF management or its Executive Board.[3] Such analysis serves a different purpose from statistics showing actual bank failures. It tests where the capacity to absorb losses becomes limited under specified conditions. A severe result does not establish that those conditions have already occurred.

The following are independent examples for comparing simulations, not a description of a particular study’s method. A simulation needs a defined debt perimeter, capital measure and treatment of recovery or support. Applying a change to bonds alone differs from including loans to public enterprises, even at an identical percentage. Losses allocated only to domestic banks differ from those shared with other holders. A common capital threshold in a study is also different from each country’s actual supervisory requirement. Holding assumptions fixed permits comparison of resilience without presenting the exercise as a crisis forecast.

Regional banking networks cross national aggregates

A banking group operating across countries need not hold only debt issued where it is headquartered. A bank in one country can hold another government’s debt while lending to businesses in a third. Pressure can therefore cross borders in ways a simple table pairing each government with domestic banks misses. Group-wide exposure and funds available at individual entities should be distinguished to locate losses and assess where resources can move.

Funds somewhere in a group do not establish unrestricted access at every entity. Local depositor needs, contracts or regulatory conditions can constrain transfers. Conversely, strain at one entity does not automatically make the entire group unable to pay. National balances, consolidated balances and intragroup transactions need reconciliation without double counting. Assessing regional resilience requires the conditions under which ordinary funding transfers can continue during stress, rather than group size alone.

Does cutting expenditure necessarily reduce pressure on banks?

Less additional government borrowing may reduce the bonds banks need to absorb. But expenditure cuts can also reduce business cash receipts. Postponing a new project differs from withholding payment for completed deliveries: one reduces future orders, while the other delays recovery of costs already incurred. Equal reductions in spending can therefore reach business loans differently and at different times. Borrowing figures should be read alongside the expenditure cut and arrears remaining, rather than treated as a complete measure of bank risk.

Payment priorities matter too. Income support, maintenance of working infrastructure, settlement of existing contracts and new investment produce different consequences when interrupted. Choosing among them requires national institutions and resources, not a universal ranking. Market analysis can clarify how a choice reaches bank assets. If deferred maintenance interrupts business operations, a near-term budget saving must be considered alongside future tax receipts and repayment capacity. A change in total spending cannot measure that trade-off.

Separate early signals from later confirmation

Auction terms, government payment schedules, business collection periods and short-term bank funding can move before realised losses or defaults. Waiting only for the latter risks delay, while treating one early change as a crisis invites false alarms. Several observations should point in the same direction. If auction terms worsen while arrears decline and bank funding remains stable, a story of deterioration across the entire chain needs revision.

Observation dates and publication dates must also remain separate. Bank statistics available in October may describe end-June holdings rather than October payments. Different government and bank reporting cut-offs can make the same payment appear outstanding in one document and collected in another. Updates should connect the stages of a transaction instead of simply selecting the latest publication date. A reporting delay is evidence of neither unchanged conditions nor deterioration.

Specify the evidence that would change the assessment

A broader buyer base, dispersed maturities and stable bank funding would weaken the assessment that government refinancing is concentrating pressure in banks. Growth in nonbank investors should be tested through ultimate holdings and maturities. A bank purchase followed by a transfer within the same group may change appearances without changing who ultimately bears the burden. Auction participation, continued ownership and the allocation of price and payment risk together determine whether concentration has really diminished.

Shorter government payment delays and fewer bridge-loan renewals alongside stable operations would weaken the proposed business transmission. If credit rises while materials purchases or wage payments weaken, it is worth testing whether financing is bridging old receivables rather than funding new activity. Establishing that requires firm-level collections and uses of funds, not regional averages. Specifying the evidence in advance reduces the temptation to select only convenient figures afterwards.

The same issuance can have different outcomes when alternatives differ

With stable deposits and timely public payments, new government bonds can help connect settlement and business activity. A second case arises when refinancing succeeds but arrears persist, leaving banks holding both sovereign debt and bridge loans to suppliers. Dependence on the same payer and the timing of cash recovery then matter more than the sum of balances. A third case combines funding outflows or losses with mutual support needs between banks and government. These are conditions to distinguish through evidence, not probability-weighted forecasts.

Connecting the October outlook to financing requires moving beyond the successful sale of a bond to the payments it funds and the assets and liabilities it leaves at banks. Domestic banks can provide an important stabilising source of demand while also becoming more dependent on the same payer. The assessment turns on payment speed, funding stability and the capacity to absorb losses, alongside refinancing volumes. Following those three connects apparently different developments in sovereign markets and business operations within one financing chain.