Open AccessFrom Audit Findings to Fiscal Accountability: Tracing the Institutional Transmission of Public Debt Oversight in Ghana, 2016–2025
View Full Article|
Download PDFAbstract:
The institutional processes through which public debt audit findings are translated into corrective action remain insufficiently understood, particularly where fragmented administrative records and fiscal stress coexist with established debt-reporting practices. This study examines the evolution of public debt reporting in Ghana between 2016 and 2025 and traces the institutional response to a substantial discrepancy identified in the Auditor-General’s 2024 Whole-of-Government Accounts (WGA). A longitudinal embedded single-case study design was employed, combining systematic coding of Annual Public Debt Reports (APDRs) across eight disclosure domains with process tracing of the subsequent institutional response. Debt indicators were examined to contextualise the findings rather than to establish a causal relationship between audit intervention and fiscal outcomes. The analysis shows that annual debt reporting became an established practice, accompanied by broader disclosure of debt portfolio characteristics, costs, risks, guarantees, on-lending and contingent liabilities. Nevertheless, a GH¢138.91 billion difference was reported between the debt balance consolidated by the Controller and Accountant-General’s Department (CAGD) and the corresponding Ministry of Finance record. Subsequent parliamentary scrutiny indicated that a substantial part of this discrepancy arose from differences in consolidation boundaries, following which the institutions reconciled their figures. This sequence illustrates how an audit finding can progress through detection, disclosure, deliberation and correction, although limited publicly documented evidence was identified of sustained improvements in preventive controls. Meanwhile, Ghana’s debt-to-GDP ratio increased from 56.8% in 2016 to approximately 80% in 2020 before declining to 44.7% by the end of 2025, providing fiscal context without demonstrating that audit contributed to these changes. An Audit–Accountability Transmission Mechanism (AATM) is developed to distinguish the identification of audit findings from their institutional uptake and potential governance consequences. The framework highlights inter-agency reconciliation, parliamentary scrutiny and recommendation follow-up as critical mechanisms through which audit findings may be translated into corrective action and institutional accountability. The findings also demonstrate the importance of distinguishing discrepancies arising from different reporting boundaries from errors in underlying debt records. Effective public debt oversight therefore requires not only transparent reporting and independent scrutiny but also documented reconciliation procedures, traceable corrective actions and verifiable improvements in preventive controls.1. Introduction
Sovereign debt is more than a financing instrument. It is also a set of legal obligations and a claim on future public revenue. When it is well managed, borrowing can smooth expenditure, fund infrastructure and help governments absorb shocks. When debt records are incomplete, fragmented or weakly scrutinised, however, borrowing can conceal fiscal risk, shift costs to future generations and weaken confidence in government financial reporting. The debt distress that followed the pandemic in many emerging and developing economies has therefore renewed interest in the institutions that authorise, record, disclose, audit and oversee public borrowing (Horn et al., 2024; Rivetti, 2021). Evidence that external debt is often understated, and that unreported liabilities tend to emerge when a crisis makes them most costly, reinforces a basic point: a reliable debt figure is itself an outcome of good governance, not something that can simply be assumed (Brown, 2023; Horn et al., 2024).
Ghana provides a useful setting in which to examine this issue. The Public Financial Management Act, 2016 (Act 921) created statutory duties for debt management and reporting (Republic of Ghana, 2016), and the Ministry of Finance has submitted an Annual Public Debt Report (APDR) to Parliament for every financial year from 2016 to 2025. During the same period, Ghana experienced rapid debt accumulation, lost access to international capital markets, carried out a domestic debt exchange, restructured its external debt under the G20 Common Framework and entered an International Monetary Fund (IMF) programme (Abotebuno Akolgo, 2023; IMF, 2023; Karimu, 2024). Figures reported for earlier years were also revised, sometimes by several percentage points, in later editions of the same statutory report. In July 2025, the Auditor-General’s report on the 2024 Whole-of-Government Accounts (WGA) identified a GH¢138.91 billion difference between the public debt figure consolidated by the Controller and Accountant-General’s Department (CAGD) and the figure held by the Ministry of Finance (Arthur, 2025; Ghana News Agency, 2025). The Ghanaian case therefore raises the central question of this paper: how could extensive statutory reporting coexist with serious debt vulnerability and unreconciled official records, and what did external audit add?
The Auditor-General occupies a constitutionally protected position in Ghana’s accountability architecture. Article 187 of the 1992 Constitution requires the public accounts of Ghana and of all public offices to be audited and reported on to Parliament, and the Audit Service Act, 2000 (Act 584) gives operational form to that mandate (Republic of Ghana, 1992; Republic of Ghana, 2000). Internationally, the International Organization of Supreme Audit Institutions (INTOSAI) recognises in GUID 5250 that public debt may be examined through financial, compliance and performance audits and that useful audit reporting should contribute to sound debt management (INTOSAI, 2020). A mandate, however, is not an effect. Audit becomes consequential only when a finding is disclosed, understood by oversight actors, acted upon by the responsible institutions and absorbed into improved controls. The supreme audit institution (SAI) literature has long recognised that the link between formal audit arrangements and fiscal outcomes is weak and uneven (Blume & Voigt, 2011), that measuring audit outcomes remains an unresolved problem (Bonollo, 2019), and that recommendations travel to implementation by different institutional routes in different systems (Torres et al., 2019). What the literature lacks, particularly for African countries, is a mechanism-level account of how a public debt finding moves from an auditor’s working paper to a corrected national record.
This paper addresses that gap through three questions. First, how did the scope and transparency of Ghana’s statutory public debt reporting change between 2016 and 2025? Second, how did a material difference in official debt records move from an external audit finding to institutional correction? Third, what does the Ghanaian experience suggest about the conditions under which SAI oversight can support fiscal sustainability? The argument is not that audit mechanically reduces debt. Fiscal sustainability reflects primary balances, growth, interest and exchange rates, contingent liabilities, financing choices, restructuring and political decisions. Instead, the paper develops and applies the Audit–Accountability Transmission Mechanism (AATM). The framework follows a finding through four stages: detection produces verified evidence; public disclosure makes that evidence available; legislative deliberation creates answerability; and corrective action changes data, controls, recoveries or policy. A feedback loop can then strengthen prevention and future risk-based audit planning. By separating audit output from audit impact, the framework helps explain why technically sound audits may still have limited effects when parliamentary follow-up, recommendation tracking or inter-agency data governance is weak.
The contribution has three main features. First, it treats the reconciliation of debt figures across reporting systems, rather than any one published number, as the issue that audit must test. Second, its within-case longitudinal design makes it possible to follow the sequence from audit testing to correction in a way that cross-country studies generally cannot. Third, Ghana’s first decade under a statutory debt-reporting regime covers a full debt cycle, including restructuring. This makes it possible to examine directly why greater transparency and stronger fiscal sustainability do not always move together. It should be stressed at the outset that the analysis traces these observable intermediate mechanisms, namely data reconciliation, parliamentary use and documented corrective action, rather than establishing a causal effect of audit on fiscal sustainability.
The remainder of the paper proceeds as follows. Section 2 reviews the literature on public debt governance, SAI effectiveness and accountability, and develops the AATM and three analytical propositions. Section 3 sets out the research design, data and coding. Section 4 reports the findings. Section 5 discusses their theoretical and practical implications, and Section 6 concludes with policy recommendations and a research agenda.
2. Literature and Theoretical Framework
Fiscal sustainability concerns a government’s capacity to meet current and future obligations without implausibly large adjustment, default or destabilising inflation. It is forward-looking and cannot be read from a single debt ratio. The vulnerability of a portfolio depends on its composition, currency exposure, interest-rate structure, maturity and refinancing profile, creditor concentration, guarantees, on-lending and contingent liabilities. Sound public debt management therefore seeks to meet financing needs at the lowest cost consistent with prudent risk, while keeping reliable records and a strategy aligned with macro-fiscal objectives (World Bank, 2021).
Transparency is a core input into that process. The IMF Fiscal Transparency Code requires fiscal reports to be comprehensive, relevant, timely and reliable, and gives specific attention to the disclosure of fiscal risks (IMF, 2018). The World Bank’s Debt Management Performance Assessment (DeMPA) examines governance, strategy, coordination, recording, operational risk and reporting across the debt-management cycle (World Bank, 2021). The instrumental case for transparency is well established: it reduces information asymmetry between government and its principals, enables legislative and public scrutiny, supports market confidence and reveals hidden liabilities. Fiscal transparency is associated with lower sovereign borrowing costs and greater foreign participation in domestic debt markets (Kemoe & Zhan, 2018), and the oversight of fiscal risks from public entities, more than the volume of disclosure, is what improves budget credibility in developing countries (ElBerry & Goeminne, 2021).
Recent scholarship has, however, moved the debate from the quantity of disclosure to the integrity of the disclosed number. The World Bank’s first comprehensive assessment of debt transparency in low-income developing countries found that public debt data published in different sources could differ by up to 30 percent of GDP, that many countries published no debt data at all, and that legal frameworks governing borrowing were frequently weak (Rivetti, 2021; Rivetti, 2022). Horn et al. (2024) digitised every vintage of the World Bank’s external debt statistics since the 1970s and showed that ex post revisions are frequent, systematically upward and largest in low-income countries with weak institutions; they estimate roughly one trillion US dollars of external sovereign borrowing that entered official statistics only in hindsight. Brown (2023) finds that governments hide loans strategically to avoid breaching debt sustainability thresholds, and hide less while under IMF scrutiny. A review of sixty legal frameworks concluded that weak reporting obligations, narrow coverage of the public sector perimeter, unclear borrowing authority and weak accountability mechanisms are the legal roots of debt opacity (Vasquez et al., 2024). The World Bank’s 2025 call for “radical” debt transparency singles out debt contracted outside central government, contingent liabilities and non-market instruments as the areas where risk is now most hidden, and urges joint data reconciliation between debtors and creditors (World Bank, 2025).
Taken together, this literature points to a problem that is central to the paper. Publication alone does not guarantee that debt figures are comparable. The figures may use different institutional boundaries, valuation methods, cut-off dates or treatments of arrears, guarantees, special-purpose vehicles and state-owned entities. The issue becomes especially important under accrual-based Whole-of-Government reporting, where debt statistics prepared for macroeconomic surveillance may legitimately differ from liabilities recognised in consolidated financial statements. Such a difference becomes an accountability concern when it is neither reconciled nor explained. An effective audit should therefore do more than choose between two totals. It should test classification, confirmation, elimination, completeness and reconciliation controls, and require a clear bridge for any difference that remains.
Agency theory frames public officials as agents entrusted by citizens and legislatures, the principals, with authority over public resources (Jensen & Meckling, 1976). Borrowing intensifies the agency problem because its benefits are immediate while repayment is deferred, the information is technically complex and part of the obligation may sit outside the central government boundary. Independent public audit reduces information asymmetry by testing management assertions and reporting deviations from law, standards or expected performance, and thereby supplies the information on which accountability forums depend (Bovens, 2007; Hay & Cordery, 2018).
The value of SAIs, however, does not follow automatically from their design. Blume & Voigt (2011) found limited and non-uniform cross-country effects of SAI organisational features on fiscal and governance outcomes, while Schelker and Eichenberger (2010) found that greater auditor competence was associated with better fiscal outcomes in Swiss cantons, suggesting that the effectiveness of auditing depends on its integration into fiscal decision-making rather than on its mere existence. Gustavson & Sundström (2016) argue that good auditing requires independence, professionalism and recognition of citizens as the ultimate principal, and Cordery and Hay (2019) locate SAI relevance in the creation and demonstration of public value rather than in the volume of reports issued. Bonollo (2019) reviewed the literature on measuring SAI outcomes and found that it concentrated on performance rather than financial auditing, rarely used quantitative indicators and pointed above all to the need for follow-up procedures that monitor the effect of recommendations. Comparative evidence confirms that recommendations produce impact by different routes: an auditee-led route based on management response and audit follow-up, and a parliamentary route based on committee action (Johnsen et al., 2019; Reichborn-Kjennerud & Johnsen, 2018; Torres et al., 2019). Cracel Viana et al. (2022) show, for public-private partnerships, that an SAI can reduce information asymmetry through disclosure while leaving doubts about whether the auditee actually changed its practice. A global mapping of 189 SAIs finds no single institutional model that dominates, and marked heterogeneity shaped by state capacity (Romão et al., 2026). Power (2003) reminds us that audit can also produce legitimacy rather than control, which is precisely why the distinction between audit output and audit impact matters.
In Ghana specifically, Afadzinu (2026) shows that the appointment and tenure of the Auditor-General remain exposed to political control, with consequences for audit transparency; the present paper takes the resulting formal independence as a background condition and asks instead what happens once a finding exists. INTOSAI GUID 5250 applies the general logic to public debt, encouraging SAIs to address financial statement assertions, legal compliance, debt-management systems, portfolio risk and the economy, efficiency and effectiveness of debt-management activity (INTOSAI, 2020). The INTOSAI Development Initiative’s handbook translates that guidance into audit questions covering governance, borrowing, guarantees, on-lending, debt recording and reporting (INTOSAI Development Initiative, 2025), and the United Nations has recently urged that SAI findings be integrated into fiscal decision-making if external audit is to support sustainable fiscal systems (United Nations Department of Economic and Social Affairs, 2025). Public debt auditing therefore extends beyond the verification of year-end debt balances to encompass the assessment of strategy implementation, risk identification, data governance and the adequacy of information provided to decision-makers.
Bovens (2007) defines accountability as a relationship in which an actor is obliged to explain and justify conduct to a forum that can pose questions, pass judgement and impose consequences. An audit finding creates the first of these, answerability, by documenting a condition against a criterion and identifying the responsible actor. The consequences may be corrective rather than punitive: reconciliation of records, recovery of amounts, amendment of procedures, improved disclosure, system integration or changes in borrowing practice. Parliamentary committees are the critical intermediary because they convert technical findings into public questions and formal recommendations, and because in Westminster-derived systems the Public Accounts Committee (PAC) is the forum before which the executive must account for the Auditor-General’s report.
Much of the existing research treats audit quality, transparency or fiscal performance as an end result, leaving the connections between them underexplored. The AATM instead presents audit impact as a sequence of four observable stages: detection, public disclosure, legislative deliberation and corrective action, followed by feedback into prevention and future audit planning (Figure 1). A break at any point weakens the connection between audit and fiscal governance. A strong finding that never enters the public domain has little accountability effect. A published finding that Parliament does not use may create no meaningful answerability. A hearing without monitored implementation may produce an appearance rather than lasting change. The framework is intentionally modest about causality. It does not predict the debt ratio; it helps identify where a debt finding is likely to stall and, therefore, where reform is most needed.
The AATM builds on the audit-impact and accountability literature by drawing a clear line between audit output and audit impact. Detection and disclosure produce an audit output. Impact begins when oversight bodies use the finding and responsible institutions alter a record, control, recovery process or policy. Framing the issue in this way turns the general claim that audit matters into a process that researchers and practitioners can trace, compare and diagnose stage by stage. Relative to existing audit-impact and accountability research, which tends to stop at audit output or to treat fiscal performance as a single end result, what the AATM adds is an explicit account of the output-to-impact transition: it specifies the stage at which a debt finding is converted into public answerability and institutional correction, and the stage at which that conversion fails, so that the framework can be traced, compared and diagnosed rather than asserted.
Three propositions guide the empirical analysis.
Proposition 1. Public debt audit improves transparency when its findings test not only whether debt is published but also whether debt data are complete, consistent across reporting systems, correctly valued and reconciled.
Proposition 2. Audit findings are more likely to produce corrective action when they are publicly disclosed, taken up by Parliament and assigned to identified institutions through time-bound recommendations.
Proposition 3. The contribution of audit to fiscal sustainability is indirect and mediated by improved data integrity, risk controls, recommendation implementation and the use of audit evidence in borrowing and fiscal decisions.
Proposition 3 states a mediated theoretical relationship rather than a directly estimated causal effect. Within the study period, the analysis traces observable intermediate mechanisms, including data reconciliation, parliamentary use and documented corrective action. It does not establish that audit caused changes in Ghana’s debt ratio or fiscal sustainability, and it treats preventive control redesign as observable only where public evidence is available.
3. Methodology
The study uses a longitudinal embedded single-case design (Yin, 2018). Ghana is the main case, with two units of analysis. The first is the development of statutory annual public debt reporting from the 2016 to the 2025 financial year. This ten-year record provides the evidence used to assess Proposition 1. The second is the sequence of events surrounding the debt discrepancy identified in the Auditor-General’s audit of the 2024 WGA, which provides the process evidence for Propositions 2 and 3. A single-case design suits the research questions because the mechanism operates within Ghana’s particular legal, accounting and political setting. Ghana is also a revealing case: it offers a complete statutory reporting series, a full debt cycle that includes restructuring and an audited discrepancy that Parliament examined within the study period.
The analysis is empirical but not econometric. Debt indicators are used to describe the fiscal setting, while documentary analysis and process tracing identify the sequence of events, the actors involved and the changes that can be observed. This design avoids attributing movements in the debt ratio to audit when restructuring and macroeconomic conditions provide the main explanations. The study therefore draws conclusions about the operation of the accountability mechanism, not about the size of any causal effect of audit on debt-to-GDP.
The evidence corpus consists of official documents and corroborating public records, summarised in Table 1. The ten APDRs form a complete annual series beginning with the first report required under section 72 of Act 921. Contemporaneous editions are retained so that revisions between editions can be identified, while later editions supply the revised historical series. The 2025 report, the tenth in the series, was published in 2026 and is used for the 2021 to 2025 ratios; the 2020 and 2024 reports supply the earlier vintages against which its restatements are measured.
Source Group | Coverage | Analytical Use |
Annual Public Debt Reports (Ministry of Finance) | Financial years 2016 to 2025 (ten reports) | Debt stock, composition, service, strategy, guarantees, on-lending, fiscal risks, restructuring and the evolution of disclosure scope; revisions between editions |
Statutory and constitutional framework | 1992 Constitution; Audit Service Act, 2000 (Act 584); Public Financial Management Act, 2016 (Act 921) and its 2025 amendment | Audit mandate, reporting duties, fiscal rules and the debt anchor |
Auditor-General evidence | Report on the Public Accounts of Ghana (Whole-of-Government Accounts) for the year ended 31 December 2024, and official public summaries | Debt-data discrepancy, audit criteria, components of the difference and recommendation |
Parliamentary evidence | PAC hearings of August 2025 as reported by the state news agency and national press | Deliberation, answerability and post-report reconciliation |
International standards | INTOSAI GUID 5250; IDI public debt audit handbook; IMF Fiscal Transparency Code; World Bank DeMPA | Normative criteria for audit scope, transparency, risk and data governance |
Macro-fiscal context | IMF programme documents (2023 to 2026) | Contextual interpretation only; no causal attribution to audit |
A structured codebook (Appendix A) was applied to each APDR. The review covered eight areas: legal basis and reporting timeliness; debt stock and institutional perimeter; portfolio composition and debt service; cost and risk indicators; strategy targets and performance; guarantees, on-lending and contingent liabilities; debt sustainability and stress information; and data quality, reconciliation and audit follow-up. Each area received a score of 0 for absent disclosure, 1 for partial disclosure and 2 for substantive disclosure. Coding took place in two passes. The first identified the structure and main content of each report, while the second checked the tables, appendices and narrative against the code definitions. Where the two passes produced different results, the lower score was retained unless the report contained clear evidence for the higher category. For example, in the data quality, reconciliation and audit follow-up domain, a report scored 0 when it gave no information on reconciliation, revisions or audit follow-up; 1 when it simply marked figures as provisional or revised without explaining the change; and 2 when it identified the affected figures, quantified and explained the reconciliation or revision, and reported the status of relevant audit action. To take a concrete case from the corpus, in the guarantees, on-lending and contingent liabilities domain the 2024 APDR was scored 2 because it quantified outstanding guarantees at GH¢3.15 billion and an on-lent portfolio at GH¢16.21 billion and set out a dedicated fiscal-risk discussion; a report that noted only that such exposures existed, without amounts, would score 1, and one silent on them would score 0 (Ministry of Finance, 2025).
Process tracing was used for the embedded reconciliation episode (Beach & Pedersen, 2019). Evidence was organised against the four AATM stages: the Auditor-General’s detection of a difference; public disclosure of the finding; PAC deliberation with the CAGD, the Ministry of Finance and audit representatives; and the reported inter-agency reconciliation. For each stage the analysis asked what observable evidence would be expected if the mechanism operated, and whether it was found. A negative-case test asked whether a decade of routine publication had itself prevented the discrepancy. It had not, which supports the paper’s distinction between publication and audited data integrity.
The descriptive debt series combines three vintages: ratios reported contemporaneously in the 2020 APDR for 2016 to 2020, the revised series in the 2024 APDR for 2020 to 2024, and the further revised series in the 2025 APDR for 2021 to 2025. The overlap years are reported in every vintage in which they appear so that the size and direction of revision are visible. The spliced series is shown only to illustrate the broad trajectory and is not used for estimation.
Construct validity was strengthened by triangulating statutory requirements, debt-management reports, audit evidence, parliamentary follow-up and international criteria, and by requiring that any press-reported figure be corroborated by a second independent outlet. An explicit audit trail links each result to a document, year and table or reported event. Rival explanations for debt outcomes, including exchange-rate movements, primary balances, GDP rebasing and restructuring, were retained in interpretation rather than treated as noise.
The study uses publicly available documents only. It involves no human participants, confidential audit working papers or personal data, and institutional ethics approval was not required. The author is employed by the Ghana Audit Service; to limit the risk of institutional bias, only published reports and public hearing records were used, no privileged information was drawn upon, and the analysis records where audit-side evidence was itself incomplete or contested. The principal limitations are that a single-country study cannot estimate average cross-country effects, that public documents may omit internal implementation activity, and that the 2024 reconciliation episode was still developing when the evidence window closed. The paper therefore makes analytical generalisations about mechanisms rather than statistical generalisations about all SAIs.
4. Findings
The first finding is the continuity of the reporting regime. Ghana produced an APDR for every financial year from 2016 to 2025 under section 72 of Act 921. The 2020 report describes itself as the fifth in the series, the 2024 report as the ninth and the 2025 report as the tenth (Ministry of Finance, 2021; Ministry of Finance, 2025; Ministry of Finance, 2026; Ministry of Finance, n.d.). All ten reports disclosed the gross debt stock, the domestic and external split, government financing and debt-service information. Even the earlier editions discussed on-lending, guarantees and selected contingent liabilities. Over time, however, the reports moved beyond a year-end stock figure towards a fuller account of portfolio performance and risk. Later editions included Medium-Term Debt Management Strategy benchmarks, the holders and composition of domestic debt, cost and risk indicators, credit-risk assessments of on-lent facilities, public-private partnership exposures and, after 2022, detailed milestones in the restructuring process. The 2020 report, for example, quantified financial-sector resolution costs and energy-sector obligations and estimated their effect on the debt ratio. The 2024 report disclosed GH¢3.15 billion in outstanding guarantees and an on-lent portfolio of GH¢16.21 billion, together with a dedicated discussion of fiscal risk (Ministry of Finance, 2021; Ministry of Finance, 2025).
The improvement was not linear in every domain, and this is the analytically important qualification. Definitions, provisional figures and historical series were revised between editions. The 2024 APDR records that the provisional 2023 debt ratio of 72.3 percent published in the 2023 edition was revised to 68.7 percent, and that end-2020 was restated from the 76.1 percent reported contemporaneously to 80.4 percent (Ministry of Finance, 2021; Ministry of Finance, 2025). The same 2024 report notes that data reconciliation and validation with commercial and bilateral creditors was still under way as part of the external restructuring. Some contingent-liability disclosures depended on data supplied by beneficiary entities, and the relationship between APDR debt statistics and the liabilities recognised in the consolidated Whole-of-Government statement of financial position was nowhere set out. The evidence therefore supports a conclusion of broader disclosure, not a conclusion of complete or reconciled debt information. Proposition 1 is supported in its negative form: a decade of publication did not by itself produce consistency across reporting systems.
The second finding is that broader disclosure did not prevent debt vulnerability (Figure 2). When Figure 2 shows more than one value for the same year, the observations represent revisions published in later APDR vintages, not estimates issued simultaneously by competing sources. The contemporaneous series reports debt-to-GDP ratios of 56.8 percent in 2016, 55.5 percent in 2017, 57.6 percent in 2018, 62.4 percent in 2019 and 76.1 percent in 2020 (Ministry of Finance, 2021). The revised series in the 2024 APDR places the ratio at 80.4 percent in 2020, 78.5 percent in 2021 and 72.9 percent in 2022, when Ghana lost market access and announced the domestic debt exchange (Ministry of Finance, 2025). After the domestic exchange, the Eurobond exchange and the agreement with official creditors, the ratio fell to 68.7 percent in 2023 and 61.8 percent at end-2024, when central government and guaranteed debt stood at GH¢726.7 billion (US\$49.4 billion). The 2025 APDR revised the earlier ratios again, to 78.7 percent for 2021, 73.1 percent for 2022, 68.9 percent for 2023 and 61.6 percent for 2024. It also revised the end-2024 stock to GH¢728.6 billion (US\$49.5 billion) and reported a ratio of 44.7 percent at end-2025 on a stock of GH¢641.1 billion (US$61.3 billion) (Ministry of Finance, 2026). The debt stock rose in US dollar terms while falling in cedi terms because the cedi appreciated by 40.7 percent against the US dollar during 2025, reaching an end-period rate of GH¢10.46. This reduced the external share of the portfolio from 57.5 to 47.9 percent and brought the ratio to the 45 percent statutory anchor nine years before the 2034 deadline (Citi Newsroom, 2025; Ministry of Finance, 2026).
The post-2022 decline should not be read as an effect of the audit. It reflects restructuring, fiscal consolidation under the IMF-supported programme, nominal GDP growth and, particularly in 2025, exchange-rate appreciation that reduced the cedi value of external debt (IMF, 2026). Nor does the pre-2022 rise prove audit ineffectiveness, because the Auditor-General neither sets fiscal policy nor authorises borrowing. What the trajectory demonstrates is that reporting and audit are institutional safeguards operating within a wider political economy of borrowing (Abotebuno Akolgo, 2023; Karimu, 2024); they cannot substitute for credible fiscal rules, prudent financing choices and active management of contingent liabilities. Two further features of Figure 2 bear directly on the paper’s argument. The 4.3 percentage-point upward restatement of 2020 between the 2020 and 2024 editions is of the same order as the underreporting that Horn et al. (2024) identify as typical for low-income countries, and it occurred within a regime of annual statutory reporting. And revision did not stop once restructuring was complete: every ratio for 2021 to 2024 moved again between the 2024 and 2025 editions, by 0.2 percentage points in each case, and the end-2024 stock rose by about GH¢1.9 billion, changes that the 2025 report marks only with an asterisk for revised figures. The most recent figures are therefore the ones most exposed to future revision, and the absence of a revision note is itself a finding for the eighth coding domain.
The clearest evidence of SAI value during the study period is the public debt discrepancy reported in the audit of the 2024 WGA. According to public summaries of the Auditor-General’s report, the CAGD had consolidated a public debt figure of GH¢876.08 billion, while the Ministry of Finance record showed GH¢737.17 billion, leaving a difference of GH¢138.91 billion, or about 18.8 percent of the Ministry’s figure (Arthur, 2025; Ghana News Agency, 2025). The audit report described the components as an ‘overstatement’ of GH¢1.7 billion in external debt, GH¢132.9 billion in domestic debt and GH¢4.1 billion in Ghana Cocoa Board (COCOBOD) domestic debt, together with a small ‘understatement’ of a COCOBOD external balance. This paper uses the more neutral terms “difference” or “discrepancy” for the GH¢138.91 billion total because the later parliamentary evidence showed that much of it resulted from different consolidation perimeters rather than a deliberate inflation of central government debt. The audit nevertheless treated the unreconciled difference as inconsistent with Act 921 and with International Public Sector Accounting Standards 1 (IPSAS 1) and IPSAS 35 on financial-statement presentation and consolidation. It recommended a comprehensive reconciliation between the CAGD and the Ministry of Finance covering both the reported overstatements and omissions (Ghana News Agency, 2025). The audit also reported weaknesses in consolidation controls, including inter-entity balances of GH¢7.99 billion at the beginning and GH¢8.55 billion at the end of the year that should have been eliminated to zero, and GH¢74.24 billion in provisions classified as investment provisions rather than impairment losses (Ghana News Agency, 2025).
Two further features of the episode are analytically significant. First, the public record carries more than one figure for each side of the comparison. The Ministry of Finance’s own 2024 APDR reports central government and guaranteed debt at the end of 2024 of GH¢726.7 billion, restated a year later to GH¢728.6 billion; the audit report’s Ministry comparator was GH¢737.17 billion; and at the PAC hearing the CAGD cited a whole-of-government figure of GH¢861.4 billion against GH¢711.4 billion for central government and guaranteed state-owned enterprise (SOE) debt (Arthur, 2025; Ghanaian Times, 2025; Ministry of Finance, 2025; Worlanyo, 2025). Rather than an embarrassment to be explained away, this multiplicity is the finding: the same end-2024 sovereign debt carried at least five published values within eighteen months, each defensible on its own perimeter, and no published statement bridged them. Second, the finding should not be simplified into a claim that one institution deliberately overstated the national debt. The PAC hearing established that the larger part of the gap arose from the inclusion of unguaranteed SOE liabilities in the consolidated accounts, liabilities that do not add to central government debt unless the enterprise defaults (Worlanyo, 2025). That is a perimeter difference of the kind that the transparency literature identifies as the most common source of divergence between debt statistics and financial statements (Rivetti, 2022; Vasquez et al., 2024). It becomes an accountability concern because it was neither disclosed nor reconciled before the audited accounts were submitted.
The audit added value by changing the status of the difference. Before detection it was an inter-system inconsistency known, if at all, to the two compiling agencies. Once reported, it became a matter requiring public answerability and institutional correction. The episode demonstrates the first two AATM stages, detection and disclosure, and shows why a public debt audit must test the bridge between debt-management records and financial reporting rather than only the arithmetic within either system.
The PAC hearings of August 2025 supplied the third stage of the mechanism. The Controller and Accountant-General, a Deputy Minister of Finance, the Director of the Budget Office, the Ministry’s Director of Internal Audit and representatives of the Auditor-General appeared before the Committee to explain the discrepancy and the status of reconciliation (Worlanyo, 2025). An Assistant Auditor-General confirmed that the figures had been reconciled after the report was issued, that the reconciliation had validated the CAGD’s consolidated position on its own perimeter, and that the audit report had been issued under time pressure (Ghanaian Times, 2025; Worlanyo, 2025). The Committee concluded that the CAGD had not overstated central government debt, while affirming that the two records should have been reconciled and their perimeters explained before submission. This created an observable sequence from an audit finding to an institutional response rather than leaving the finding as a paragraph in an annual report. Proposition 2 is supported: disclosure followed by parliamentary uptake produced answerability and an explicit institutional commitment to reconcile.
The fourth stage, corrective action, was present but incompletely documented in the public domain. The institutions reported that the figures had been reconciled, but no public schedule of reconciling items, no statement of the corrected closing figure on each perimeter, and no account of system-control changes, responsible officers or follow-up dates were evident in the material reviewed. The immediate data issue appears to have been addressed. The durability of the correction depends on whether recurring quarterly reconciliation, common instrument identifiers, cut-off controls and version governance were instituted, and the public record does not yet show that they were. Proposition 3, which holds that audit’s contribution to sustainability runs through data integrity and control redesign, is therefore only partly testable within the window: the data-integrity channel operated, the control-redesign channel remains unobserved.
Table 2 organises the evidence against the AATM stages and the fiscal context. The pattern is one of a mechanism that operated through its first three stages and stalled, or at least became invisible, at the point where correction should have been converted into prevention. That is the stage at which the SAI literature has most consistently located the weakness of audit impact (Bonollo, 2019; Torres et al., 2019), and it is the stage at which the reform recommendations in Section 6 are directed.
AATM Stage or Theme | Observed Ghana Evidence | Inference |
Detection | The Auditor-General identified a GH¢138.91 billion difference between the debt balance consolidated in the Whole-of-Government Accounts and the Ministry of Finance record, and related consolidation-control weaknesses. | Independent testing revealed a data-integrity and perimeter risk that a decade of routine publication had not resolved. |
Disclosure | The finding entered the public domain through the Auditor-General’s report and official and press summaries in July and August 2025. | The discrepancy entered the accountability arena and could no longer remain a bilateral technical matter. |
Deliberation | The PAC examined the CAGD, the Ministry of Finance and audit representatives; the perimeter explanation was tested in public. | Parliament converted evidence into answerability and demanded explanation. |
Correction | The institutions reported reconciliation after the audit report; the Auditor-General’s office confirmed it. No public bridge statement or control-redesign evidence was found. | Audit contributed to the correction of the record, but the preventive feedback loop is not yet observable. |
Fiscal context | Debt reporting expanded while the debt ratio rose to about 80 percent of GDP before restructuring, then fell to about 45 percent by the end of 2025. | Audit and transparency are necessary governance conditions, not sufficient determinants of sustainability. |
5. Discussion
The findings support the paper’s main theoretical claim: having the institutions and authority to audit is not the same as achieving audit impact. This distinction offers one explanation for the mixed results in the SAI literature. Formal institutional features may show weak relationships with fiscal outcomes (Blume & Voigt, 2011; Romão et al., 2026) because publicity, legislative uptake, implementation and follow-up vary considerably even among SAIs with similar mandates. Ghana’s experience also suggests that the parliamentary and auditee-led routes identified by Torres et al. (2019) may work together rather than compete. Parliament creates public answerability, but audited entities must still implement and sustain the response through their own systems and controls.
The case also gives a more concrete meaning to public value in audit (Cordery & Hay, 2019). In debt governance, value does not always appear immediately as a cash recovery or a lower debt ratio. It may begin with something more basic: establishing whether authoritative records agree, exposing unexplained differences and improving the evidence available to Parliament, debt managers and other users of the public balance sheet. Reliable information about public obligations is itself a governance asset because pricing, strategy, budgeting, restructuring negotiations and risk assessment all depend on it (Horn et al., 2024; Kemoe & Zhan, 2018). For SAIs, the lesson is that credibility rests not only on identifying a difference but also on examining the bridge between the systems that produced it. That work helps distinguish an error from a legitimate methodological difference or a control failure and, in doing so, strengthens both accountability and the legitimacy of audit (Power, 2003). For the wider debate, the Ghana case suggests that the value of public debt audit lies less in moving the headline debt ratio than in converting a hidden inter-system inconsistency into a matter of public answerability, and that fiscal transparency and public accountability rest as much on reconciled, version-controlled data as on the frequency of publication.
A central implication is that fiscal transparency should be judged by data lineage as well as publication frequency. Regular publication is valuable, but it cannot by itself show how debt statistics relate to liabilities in consolidated financial statements or to alternative institutional perimeters. Transparency therefore requires a documented bridge that identifies the perimeter, valuation basis, cut-off date and treatment of arrears, guarantees, SOEs and restructured instruments. This extends the World Bank’s call for joint data reconciliation and oversight of debt outside central government (World Bank, 2025) and gives operational content to legal reforms proposed by Vasquez et al. (2024): the debt perimeter and the responsibility for reconciling it should be explicit, reviewable and enforceable.
Version governance is equally important because legitimate revisions can still weaken accountability when their provenance is unclear. Rebased GDP, updated exchange rates, reclassification and finalised restructuring terms may all justify a change, but users must be able to identify the authoritative vintage, the affected observations and the reason for revision. Machine-readable publication with stable release dates, revision notes and common instrument identifiers would create a traceable chain from source records to published totals. It would also allow the SAI to test revision controls and recurrence over time, shifting audit attention from isolated annual figures to the reliability of the reporting process.
The evidence supports Proposition 3 in its mediated form. Accurate debt records improve debt-service forecasts, risk metrics, borrowing plans and consolidated financial statements. Compliance audits can test borrowing authority, statutory ceilings, guarantees and reporting deadlines; the 2025 amendment to Act 921, with its 45 percent debt anchor, primary-balance rule and independent Fiscal Council, now provides explicit criteria against which such audits can be conducted (Citi Newsroom, 2025). Performance audits can examine whether the Medium-Term Debt Management Strategy is implemented, whether refinancing and currency risks remain within benchmarks, and whether on-lent loans and guarantees are assessed and recovered. These improvements reduce the probability that risks remain hidden or unmanaged, which the hidden-debt literature shows to be precisely where crises originate (Brown, 2023; Horn et al., 2024).
They do not determine the primary balance, the exchange rate, global interest rates, commodity prices or the political willingness to comply with fiscal rules. Ghana’s restructuring and the large 2025 decline in the ratio illustrate the distinction. A robust audit can warn, verify and recommend, but elected authorities and debt managers decide whether fiscal and borrowing practice changes are needed. The SAI’s influence is therefore strongest where its recommendations are integrated into budget scrutiny, approval of the debt strategy and assessment of debt-management performance, a conclusion consistent with the finding that oversight of fiscal risk, rather than disclosure volume, is what improves budget credibility (ElBerry & Goeminne, 2021). The distributional consequences of the domestic debt exchange for pensioners and other bondholders (Oduro, 2026) also underline that sustainability restored through restructuring is not costless, and that the earlier detection of risk which audit can provide has a social as well as a fiscal value.
The AATM contributes to theory by specifying the mediators between SAI activity and fiscal governance. It moves beyond a binary audit/no-audit variable and provides observable process tests that answer Bonollo’s (2019) call for outcome measures. Researchers can operationalise each stage: the materiality and risk orientation of findings; publication timeliness and accessibility; parliamentary hearings and recommendations; implementation rate and time to correction; and recurrence of the underlying issue. Comparative studies can then test which institutional configurations allow findings to traverse the chain, and whether the auditee-led or parliamentary route dominates in different administrative traditions.
For practice, the framework supplies a diagnostic. If findings are technically weak or contestable, investment is needed in specialised debt-audit and information-systems audit capacity. If reports are delayed or inaccessible, disclosure systems require attention. If Parliament does not use findings, committee capacity and scheduling are the constraint. If recommendations are repeatedly accepted but not implemented, public tracking, escalation and follow-up audit become the priority. This diagnostic orientation is more actionable than treating SAI independence as the sole reform lever, important though independence remains in Ghana (Afadzinu, 2026).
6. Conclusion and Policy Implications
This paper examined how public debt auditing relates to transparency, accountability and fiscal sustainability in Ghana between 2016 and 2025. Over that decade, Ghana established a regular system of annual debt reporting and disclosed increasingly detailed information on the debt portfolio and its risks. Yet debt vulnerability worsened before restructuring, historical figures changed across editions of the statutory report, and the Auditor-General found a material difference between the debt records of two central institutions. The case shows that publishing a figure is not the same as reconciling it, just as issuing an audit report is not the same as achieving impact.
The AATM helps explain how an audit finding can produce change. Audit first detects a condition; public reporting brings it into view; Parliament creates answerability; and the responsible institutions correct data, controls or policy. If the process works fully, the lessons feed back into prevention and future audit planning. Ghana’s reconciliation episode passed through all four stages, but the public record says little about whether preventive controls were redesigned. The existence of several published debt figures also confirms that transparency and audit must focus on the bridge between different reporting perimeters. More broadly, Ghana’s debt trajectory shows that SAI oversight is an essential part of debt governance, but it cannot by itself deliver fiscal sustainability.
Table 3 sets out seven reforms derived from the AATM and the Ghana evidence. They are designed to be implemented within existing constitutional and statutory mandates, including the 2025 amendment to Act 921.
Policy Action | Lead Institutions | Operational Implication |
1. Standalone public debt audit programme | Ghana Audit Service | Establish a multi-year, risk-based programme combining financial, compliance, performance and information-systems audit under GUID 5250; publish a thematic public debt audit at least every two years. |
2. Quarterly inter-agency debt reconciliation | Ministry of Finance, CAGD, Bank of Ghana, COCOBOD and relevant SOEs | Use common instrument identifiers and formally reconcile principal, accrued interest, arrears, guarantees, on-lending, eliminations and restructuring adjustments before quarterly and annual reporting. |
3. Published bridge statement | Ministry of Finance and CAGD | Publish, with each APDR and each set of consolidated accounts, a reconciliation between APDR debt statistics and Whole-of-Government liabilities explaining perimeter, valuation, exchange-rate, consolidation and cut-off differences. |
4. Public recommendation tracker | Parliament (through the Public Accounts Committee) and audited entities | Assign each public debt audit recommendation an owner, due date, status, evidence link and follow-up outcome; disclose overdue high-risk actions. |
5. Audit of strategy and fiscal rules | Ghana Audit Service and Parliament | Test compliance with MTDS benchmarks, borrowing authority, guarantees, PPP fiscal commitments, SOE exposures and the statutory debt anchor and primary-balance rule, not only the year-end balance. |
6. Machine-readable debt data with version control | Ministry of Finance and CAGD | Publish instrument-level or appropriately aggregated open data with metadata, revision histories and release calendars, while protecting legitimately confidential terms. |
7. Specialist capability | Ghana Audit Service | Build multidisciplinary teams in sovereign debt, IPSAS consolidation, macro-fiscal analysis, valuation, derivatives, data analytics and debt-management information systems. |
Future research should test the AATM comparatively across SAIs and debt-distress episodes. A panel could measure the share of debt-related recommendations implemented, time to correction, recurrence rates, PAC hearing delays, accessibility of audit reports and the quality of reconciliation disclosures, and examine whether stronger transmission predicts better debt-data quality, smaller ex post revisions, fewer unreported liabilities or lower borrowing costs, controlling for macroeconomic and political factors. For Ghana, a follow-up study should trace whether the 2024 reconciliation produced durable systems change and whether the corrected figures were consistently reflected in the 2025 APDR, the 2025 WGA and the fiscal statistics used by the new Fiscal Council.
The publicly available documents used in this study are accessible through the sources cited in the reference list. The compiled debt data and documentary coding records supporting the findings are available from the corresponding author upon reasonable request.
The author declares no conflicts of interest.
Table A1. Disclosure domains and evidence sought in each Annual Public Debt Report
Domain | Evidence Sought |
Legal basis and timeliness | Statutory authority, reporting deadline, evidence of submission to Parliament and publication |
Stock and perimeter | Debt definition, central or general government boundary, guaranteed debt, consolidation scope |
Portfolio and service | Domestic and external split, creditor, currency, tenor, interest structure, debt service |
Cost and risk | Average time to maturity, average time to refixing, variable-rate exposure, refinancing and currency risk, stress indicators |
Strategy performance | MTDS targets, actual performance and explanation of deviations |
Fiscal risks | Guarantees, on-lending, SOE and PPP exposures, recoveries and credit-risk assessment |
Sustainability | Debt sustainability analysis, sensitivity, debt-service burden, restructuring and forward outlook |
Integrity and follow-up | Reconciliation, data-quality caveats, revisions between editions, audit findings and recommendation status |
Note: MTDS = Medium-Term Debt Management Strategy; SOEs = state-owned enterprises; PPP = public-private partnership. Coding scale: 0 = absent; 1 = partial or aggregate disclosure; 2 = substantive disclosure with definitions, quantified detail and performance explanation.
Views: -
Crossref: 0