- Weak Banks, Rising Government Debt and Delayed Action Amplified Ghana’s Financial Crisis — Dr Opoku-Afari
Ghana’s banking-sector crisis was years in the making, with persistent undercapitalisation, weak governance and increasingly concentrated exposure to government debt creating vulnerabilities long before regulators embarked on the sweeping financial-sector clean-up in 2017, according to former First Deputy Governor of the Bank of Ghana, Dr Maxwell Opoku-Afari.
His assessment suggests that the eventual cost of stabilising the financial system cannot be understood simply as the price of resolving poorly managed banks. Rather, it was part of a wider cycle in which fiscal weakness migrated into bank balance sheets, financial-sector problems migrated back into the government budget, and the sovereign ultimately became responsible for repairing institutions that had themselves become increasingly exposed to sovereign risk.
In his August 2026 policy note, How Not to Miss a Crisis: Lessons from Ghana, Dr Opoku-Afari describes Ghana’s 2022 debt crisis not as an unexpected accident but as the culmination of vulnerabilities that had accumulated over more than a decade. Strong economic growth, continued access to financing and repeated macroeconomic surveillance helped mask structural weaknesses including weak revenue mobilisation, persistent fiscal deficits, contingent liabilities and increasingly expensive borrowing.
The banking system became one of the principal channels through which those risks were eventually transmitted.
Ghana’s banks, according to the paper, were undercapitalised for much of the decade before the 2017 clean-up, while governance and risk-management weaknesses persisted across parts of the industry. The restructuring between 2017 and 2019 was therefore necessary to resolve distressed institutions, restore confidence and strengthen prudential standards.
But fixing the problem imposed a substantial fiscal cost. Drawing on the International Monetary Fund’s 2024 review of Ghana’s Extended Credit Facility-supported programme, the paper estimates that resolution and clean-up costs reached approximately 7.10% of GDP between 2017 and 2021, with much of that burden absorbed by the state.
The absence at the time of an effective deposit-insurance framework capable of absorbing the full resolution burden meant government effectively became the backstop required to protect depositors and maintain confidence in the financial system.
Allowing insolvent institutions to collapse without sufficient safeguards could have imposed severe losses on households, businesses and other creditors, potentially triggering wider financial instability. But protecting depositors required the government to absorb losses that had originated within private and state-owned financial institutions. The immediate objective was financial stability. The longer-term consequence was additional pressure on public debt.
The banking clean-up stabilised the system, but the fiscal burden associated with doing so became part of the same sovereign balance sheet that was already under strain from high deficits, expensive borrowing, energy-sector liabilities, cocoa-sector obligations and other contingent exposures.
The cost did not end with the 2017–2019 intervention. Following the Domestic Debt Exchange Programme, the government issued additional recapitalisation bonds equivalent to about 2.60% of GDP to support undercapitalised banks and strengthen the financial sector after losses on sovereign securities weakened bank capital.
That sequence illustrates what the paper calls a “spillback” from the financial system to the budget.
Banks had accumulated government securities. When the sovereign’s debt became unsustainable and those securities were restructured, losses and valuation effects weakened the same institutions that had helped finance government. The state then had to provide new support to parts of the financial system.
This is the essence of the sovereign-bank nexus. Banks regard government securities as core financial assets and governments rely on banks as major buyers of domestic debt. When sovereign finances are sound, the relationship can function without difficulty.
But when fiscal conditions deteriorate, the connection becomes dangerous. A government that borrows heavily from its banking system can progressively concentrate sovereign risk on bank balance sheets. If confidence in government debt subsequently weakens, the resulting losses can erode bank capital, constrain lending and threaten financial stability.
Government may then have to support those banks — adding further liabilities to the sovereign balance sheet that caused the problem in the first place.
That feedback mechanism is sometimes described as a sovereign-bank “doom loop”.
Ghana’s experience provides a particularly striking illustration because the move towards domestic borrowing had originally been partly justified as a risk-management strategy.
Reducing excessive dependence on foreign-currency borrowing was intended to limit exchange-rate exposure and deepen Ghana’s domestic capital market. The country developed the Ghana Fixed Income Market, strengthened its primary-dealer framework and expanded the range of securities available to domestic investors.
Pension funds, insurers and banks became increasingly important financiers of government.
The strategy had genuine benefits. Domestic borrowing could reduce direct foreign-currency exposure, draw on Ghana’s growing institutional savings pool and give government a more flexible funding source than international capital markets.
But those benefits carried costs that were not sufficiently internalised.
“Domestic borrowing costs increased, and private-sector credit was crowded out as banks reallocated portfolios toward government securities. This deepened the sovereign–bank nexus and heightened systemic risk within the financial sector,” the paper states.
By the end of 2021, banks held more than 30% of Ghana’s domestic debt, according to data cited in the analysis. Domestic debt had risen from about 31% of GDP in 2019 to more than 40% during 2020–2021 before declining following the debt exchange.
Government securities were not merely another investment category in banking portfolios. They had become central to the financial system’s asset allocation. Policies aimed partly at reducing Ghana’s vulnerability to foreign creditors simultaneously increased the dependence of domestic banks, pension funds and insurers on the financial health of the sovereign.
The country changed the location of some of its risk without eliminating it.
“As fiscal financing increasingly shifted toward Cedi-denominated instruments, banks and other institutional investors became the primary absorbers of government issuance, tightening the sovereign–bank nexus and increasing the macro-financial costs of any loss of market confidence,” Opoku-Afari writes.
The DDEP eventually exposed precisely that weakness. A sovereign restructuring that was necessary to restore government debt sustainability simultaneously created losses and income pressures for institutions whose portfolios contained large quantities of government securities.
The financial system was therefore not standing outside the sovereign crisis.
It was embedded within it. That is why Opoku-Afari’s analysis challenges a narrow reading of Ghana’s financial-sector clean-up. The crisis cannot simply be divided into one episode involving badly managed banks between 2017 and 2019 and a separate sovereign-debt crisis beginning in 2022.
The two were connected through the government balance sheet. Indeed, the paper argues that fiscal vulnerabilities were already visible long before the eventual default.
Fiscal deficits averaged 8.30% of GDP between 2010 and 2024, while gross public debt rose from about 38.90% of GDP in 2010 to 92.70% in 2022 before declining following the debt restructurings.
Debt service increasingly constrained government finances. Between 2018 and 2022, interest payments absorbed roughly 29.00% of total public expenditure. Combined with compensation of employees, those rigid expenditure items consumed nearly 60.00% of the budget, leaving considerably less room for investment in health, education, water and other development priorities.
By 2018–2020, debt service was taking more than 45.00% of government revenue, according to the paper’s analysis.
These were not invisible warning signs. Nor were they limited to headline debt-to-GDP ratios. Liquidity indicators were deteriorating sharply. Debt service-to-revenue had breached its benchmark as early as 2013 and climbed to more than 40.00% by 2022, while the present value of public debt-to-GDP exceeded relevant sustainability thresholds for prolonged periods.
Yet Ghana continued to obtain financing. International bond offerings attracted investors. Domestic banks continued purchasing securities. Budgets continued to be approved. External partners continued providing financing and completing programme reviews.
That is precisely what makes Opoku-Afari’s warning uncomfortable. Continued access to money can create the appearance that debt remains sustainable even when the underlying ability to service that debt is deteriorating.
A sovereign that can refinance maturing obligations can postpone the moment at which vulnerabilities become visible.
But liquidity is not the same thing as solvency. And market access is not proof of fiscal sustainability. The paper argues that political incentives and optimistic assumptions contributed to delayed recognition and corrective action, even though persistent deficits, weak revenue mobilisation, quasi-fiscal operations and rising sovereign exposure in banks were already observable.
Successive IMF-World Bank Debt Sustainability Analyses did identify growing risks. The 2015 DSA classified Ghana as being at high risk of debt distress but considered the debt sustainable under assumptions including continued market access and fiscal consolidation. The 2019 assessment drew attention to financial-sector clean-up costs, energy liabilities and weak domestic revenues, while subsequent analyses placed greater emphasis on contingent liabilities.
The problem was therefore not necessarily the absence of warning. It was whether those warnings were strong enough, comprehensive enough and translated into action quickly enough.
Opoku-Afari argues that conventional surveillance did not sufficiently internalise rollover and liquidity risk, feedback between domestic debt and financial institutions, foreign-exchange exposure and the wider public-sector balance sheet.
Optimistic baseline assumptions further reduced the apparent urgency. This is a critical distinction for future policy. A bank can satisfy conventional capital requirements while simultaneously carrying a dangerous concentration of sovereign exposure.
Government debt can similarly appear manageable under a baseline assumption of continuing refinancing even though modest changes in interest rates, exchange rates or investor sentiment could fundamentally alter the trajectory.
Ghana experienced both problems. The risk structure of the debt was itself deteriorating. Opoku-Afari’s assessment of public debt between 2016 and 2024 identifies a weighted average interest rate of approximately 10.70%, while 17.50% of debt was scheduled to mature within one year. Foreign-currency debt averaged 54.50% of the stock, adding another layer of vulnerability to exchange-rate movements.
That combination expensive debt, refinancing requirements and foreign-currency exposure created a self-reinforcing cycle.
Higher interest payments increased borrowing requirements. New borrowing occurred at increasingly expensive rates. Currency depreciation inflated the cedi value of foreign debt.
And greater domestic borrowing pushed banks further towards government securities while potentially crowding out private-sector lending.
The eventual crisis was therefore not one isolated failure. It was a system interacting with itself. That makes the institutional lessons especially important.
The paper notes that domestic accountability institutions identified irregularities and civil society organisations including the Africa Centre for Energy Policy, Institute for Fiscal Studies and Oxfam raised concerns about debt, energy-sector inefficiencies and revenue management.
Their weakness was not necessarily an inability to identify risk. It was a lack of authority to force corrective action.
“Accountability institutions including the Auditor-General and anti-corruption bodies documented irregularities, but follow-through was slow, contested, or incomplete,” the paper says.
This changes the question policymakers should be asking. The issue is not simply: who saw the crisis coming? It is whether Ghana had institutions capable of converting warning signals into binding policy discipline while adjustment was still relatively inexpensive.
Opoku-Afari proposes several reforms. Debt reporting should extend beyond conventional central-government obligations to encompass state-owned enterprises, arrears, public-private partnerships, guarantees and other contingent liabilities.
Fiscal rules should be enforced more credibly, supported by an independent and sufficiently empowered Fiscal Council.
Most significantly for the banking system, the paper recommends a “sovereign-bank firewall” prudential limit on banks’ holdings of government securities, incentives for greater portfolio diversification and resolution mechanisms that do not depend on sovereign guarantees.
It also argues for stress tests that explicitly model what happens when domestic yields rise sharply, sovereign bonds lose value and banks subsequently require recapitalisation that increases public debt.
That is precisely the feedback mechanism Ghana has already experienced. The larger lesson is that banking supervision, fiscal policy and debt management cannot be treated as separate disciplines.
A banking regulator concerned only about individual institutions may overlook the systemic consequences of their common exposure to sovereign debt.
A finance ministry concerned principally with finding buyers for government securities may underestimate the financial-stability consequences of concentrating those securities in domestic banks.
And debt sustainability analysis focused excessively on headline debt ratios can miss risks hidden in maturity structures, refinancing requirements, contingent liabilities and the financial system itself.
Ghana’s banking clean-up therefore offers a broader lesson than the 7.10% of GDP estimated cost. It demonstrates the price of waiting until vulnerabilities crystallise. The first intervention repaired banks weakened by poor capitalisation, governance and risk management.
The second major shock came when the sovereign’s own debt restructuring damaged institutions that had become major creditors to the state. Government then had to provide additional support to protect financial stability.
The taxpayer effectively stood at both ends of the cycle. For Ghana, preventing a repeat will require more than ensuring that banks are currently well capitalised.
It requires questioning how those banks deploy their capital, how concentrated their sovereign exposure becomes, whether government borrowing is crowding out productive private credit and whether the fiscal position underlying sovereign securities is itself sustainable.
Opoku-Afari’s central conclusion is consequently broader than banking. Ghana’s 2022 crisis, he argues, emerged from a prolonged accumulation of fiscal, macro-financial and public-sector balance-sheet vulnerabilities that were visible but not addressed “with sufficient speed or depth”.
The country’s post-crisis architecture will therefore be judged not simply by whether it can repair the damage already done.
It will be judged by whether the next warning is acted upon before another expensive rescue becomes necessary.
As the paper concludes, Ghana’s experience shows that strong growth and continued market access can coexist with deteriorating debt quality and hidden vulnerabilities. The real policy challenge is to recognise those risks while adjustment remains a choice rather than waiting until banks, investors, pension funds, businesses and taxpayers are forced to absorb the cost.
