Ghana's Double Signal: The Ballot Box, the Barracks, and the Country Nobody Expected

The standard read on Ghana is reassuring. Two peaceful transfers of power in the past decade. An IMF programme that survived a sovereign default and is now in its third review. A professional military that has held to the barracks since the Rawlings era ended in 1992. V-Dem, Freedom House, and Afrobarometer all point the same direction: Ghana is the exemplar of democratic consolidation in West Africa.

QG Intelligence is a system that searches for structural resemblance between countries' economic trajectories and historical precedents, running the numbers across 131 countries and hundreds of indicators to surface risk patterns that conventional desk analysis does not reach. In its April 2026 scoring run¹⁵A specific computation run of QGI's full pipeline on the April 2026 data snapshot; scores are tied to a scoring date because indicator data and model outputs change as new data arrives.Read more →, Ghana scores at the 99th percentile of global election-stress¹Ghana sits higher than 99% of the 131-country cohort on its Elections & Voting Precedent Match score; the percentile is a rank, not a probability.Read more →globally. Its top short-horizon risk category is not elections; it is military coup. And the country whose current economic and political risk fingerprint most closely matches Ghana's, out of every country in the 131-country cohort, is not Mali, not Burkina Faso, not any of the countries that analysts already watch for coup risk in West Africa.

It is Mauritania.

That comparison is the story.

1. The Signal: Two Numbers That Don't Belong Together

Ghana's Elections & Voting Precedent Match²A per-recipe score (0–100) ranking how closely a country's current indicator trajectory resembles historical cases that preceded elections-related geopolitical events, within a cohort of countries that fire on that recipe.Read more → registers 99/100³The score is 99 out of 100; the bracketed range is a 95% bootstrap confidence interval showing the uncertainty around that estimate from repeated resampling.Read more → [95% bootstrap CI⁴A confidence interval calculated by resampling the data thousands of times to estimate how much the score would vary if measured on slightly different data; narrower intervals indicate more stable estimates.Read more →: 97–100] , placing it at the 99th percentile of a 131-country global cohort. Its top-salience risk category at the short-horizon window is not elections; it is Military Coup. And its closest structural analogue, at 99% cosine similarity¹¹A geometric measure of how similar two countries' 8-dimension L1 risk vectors are, regardless of scale; 99% cosine similarity means the two risk profiles point in nearly the same direction across all active dimensions.Read more → on the 8-dimension L1 risk profile¹²The specific eight master categories (A through H) used in the current QGI taxonomy: Political Power & Regime, Conflict & Insurgency, Governance & Rule of Law, Economic & Financial Stress, Social & Civil Unrest, Diplomatic & Geopolitical, Humanitarian & Displacement, Structural Vulnerability.Read more →, is not Mali or Burkina Faso or Niger. It is Mauritania.

Ghana's Elections & Voting score of 99/100 [CI 97–100] is not alarming in isolation. Elections-related structural stress is common in countries navigating economic transitions and competitive party politics; the signal at any given moment captures more than coup risk. The recipe's LOCO AUC 0.711⁵Leave-One-Country-Out Area Under the Curve: a cross-validation metric measuring how well the recipe discriminates countries that later experienced an event from those that did not, after removing each country from its own training set; 0.5 = chance, 1.0 = perfect.Read more → (MODERATE evidence strength⁶QGI's qualitative label for a LOCO AUC in the range that indicates the recipe performs meaningfully better than chance but is not in the highest confidence tier; the label travels with the recipe, not the country score.Read more →) and calibration error of 0.012⁷Expected Calibration Error (ECE): the average difference between a model's predicted probability and the actual observed event rate across score buckets; 0.012 means predictions are on average 1.2 percentage points off.Read more →tell you it is a well-performing model, one of the more reliable in QGI's current validated set.

What makes Ghana unusual is the S-tier / M-tier / L-tier / XL-tier⁸Pattern Length Tiers grouping historical precedent windows by duration: S = 3–8 years (policy cycles), M = 9–20 years (leadership eras), L = 21–40 years (structural shifts), XL = 41+ years; each tier captures a different type of structural dynamic.Read more →progression. At the S-tier (short time-horizon window), Military Coup is Ghana's top-ranked risk category with a salience z-score⁹A measure of how unusual a country's score for a given risk category is relative to its own historical average, scaled in units of standard deviation (σ); z = 2.76 means the score is 2.76 standard deviations above that country's mean.Read more →of 1.39. Elections & Voting is third. At the M-tier (medium window), Military Coup holds the top rank (salience 1.42); Elections & Voting climbs to second (salience 1.14, z-score 1.90). At the L-tier (long window), Elections & Voting surges to first (salience 1.66, z-score 2.76) with Military Coup holding second. At the XL-tier (longest window), Elections & Voting dominates at z-score 2.38; Military Coup has dropped out of the top tier entirely.

At short-horizon windows, the model is picking up coup-adjacent conditions as the dominant signal. At longer windows, the election-stress signature becomes primary. This is the structural fingerprint of a country where institutional stress is building across two timescales simultaneously, not a country with an acute crisis in one domain.

QGI's composite risk score 68/100¹³The country-level QGI Risk Score: a single percentile rank summarising the strength of the country's overall salience signal against the scoring cohort; distinct from individual recipe Precedent Match scores.Read more → [CI 60–75], Tier HIGH¹⁴A country-level severity label (HIGH / MODERATE / INDICATIVE) assigned based on maximum salience across all patterns, with additional composite-promotion rules; HIGH requires sustained cross-category signal strength.Read more →. The recipe-level breakdown from the current scoring run shows Elections & Voting at 99/100, Multilateral Agreement at 73/100 [CI 55–90], and Bilateral Treaty at 27/100 [CI 19–34]. The wide CI on Multilateral Agreement reflects either model uncertainty or data-coverage thinness for Ghana's multilateral treaty history; either reading is consistent with the structural argument. Civil War and Insurgency produces a null result for Ghana; the model does not see conflict-state conditions. This is not the DRC profile. It is not the Sudan profile. The stress is specifically in the political-economy intersection where electoral legitimacy meets institutional fragility.

The election-stress signal at the 99th percentile is available to any FT reader who follows Ghana. The simultaneous top-rank salience of Military Coup at short-horizon windows is what QGI adds. These two signals together constitute what I call the “coup-by-election” structural template: not a naked seizure of power, but a contested or destabilizing electoral outcome that becomes the pretext for military intervention. The framing matters because it points you toward a different historical template than the Burkina Faso or Mali junta model.

A methodological note: the Military Coup category appears as top salience through a statistical classification layer within QGI's scoring engine, not from a dedicated LOCO-validated recipe. Ghana's four validated recipe scores are Elections & Voting (99/100, LOCO AUC 0.711), Multilateral Agreement (73/100, LOCO AUC 0.869), Bilateral Treaty (27/100, LOCO AUC 0.668), and Civil War & Insurgency (null). The Military Coup signal is directional, not model-validated. I am using it as a structural indicator, not as a forecast. The election-stress signal is the ML-validated load-bearing component; Military Coup is the contextual reinforcement.

“Ghana is not showing one risk; it is showing the precise combination that precedes coup-by-election, where a contested or destabilising electoral outcome becomes the pretext for military intervention.”

2. The Mauritania Mirror: Why This Comparator Stops You

QGI identifies Mauritania as Ghana's closest structural analogue, with a cosine similarity of 99% on the L1 risk profile¹⁰An 8-dimension decomposition of a country's risk signal into the master categories of QGI's event taxonomy (Political, Conflict, Governance, Economic, Social, Diplomatic, Humanitarian, Structural); L1 is the first level of the taxonomy hierarchy.Read more →. Political Power & Regime accounts for 66% of the cosine similarity, Economic & Financial Stress for 24%, and Diplomatic & Geopolitical Relations for 9%. The three shared active dimensions (A, D, and F in QGI's taxonomy) form the entire explanatory structure of the match. Ghana and Mauritania's risk fingerprints are not generally similar across all eight dimensions; they are precisely similar on the three dimensions that matter most for regime-political economy interactions.

Mauritania does not appear on standard West Africa risk dashboards as a comparator for Ghana. The West Africa risk fraternity clusters Ghana comparisons around Senegal (similarly competitive electoral democracy), Côte d'Ivoire (larger Francophone economy), or, in the pessimistic register, Burkina Faso (post-2015 democratic fragility before the coups). Mauritania is off the standard mental map for Ghana analysis because the two countries sit in different subregional frames (Ghana Anglophone coastal, Mauritania Arabic-French Sahelian desert) and because their recent political trajectories appear to diverge. Ghana held credible elections in December 2024. Mauritania is run by a general-turned-president who came to power through a coup.

That apparent divergence is exactly why the structural similarity is analytically interesting.

In August 2008, General Mohamed Ould Abdel Aziz overthrew President Sidi Ould Cheikh Abdallahi, a civilian elected in 2007 in what international observers called Mauritania's most credible election to that point. The pretext was Abdallahi's attempt to dismiss officers loyal to Aziz. The AU condemned the coup and suspended Mauritania; the EU froze development assistance. In 2009, Aziz ran for president, won with 52% in a disputed but non-violently contested election, and was internationally recognized within months. His successor, Mohamed Ould Ghazouani, was his hand-picked former general.

This is the “soft coup with electoral normalization” template. It differs from the Burkina Faso/Mali/Niger junta model in a structural way: the Mauritania template involves military intervention followed by electoral legitimation, not an indefinite suspension of the constitutional order. The coup leader runs for election and wins. The international community accepts the result, resets bilateral relations, and moves on. The AU deterrence mechanism, already weakened by the Sahel coups, was tested and passed the Mauritania case, in part because Aziz moved fast toward an electoral framework.

The structural question for Ghana is not whether its military officer corps is currently planning a coup. Thirty years of civilian rule and the institutional memory of Rawlings-era interventions have created a genuine culture of military professionalism in the Ghana Armed Forces. The question is whether the cost-benefit calculus of intervention has shifted, whether the external deterrence that constrained the calculus has eroded, and whether the domestic conditions (fiscal deterioration, legitimacy stress, proximity to operational theaters) are accumulating.

QGI's 99% cosine match with Mauritania says: the structural conditions that produce the soft-coup-with-normalization pattern are present in Ghana's indicator trajectory at a level of similarity the model finds historically meaningful. Mali and Burkina Faso offer the junta-with-no-timetable template: indefinite constitutional suspension, foreign security contractors, Russia-backed legitimation architecture. That is a different and more destabilizing outcome. The Mauritania template (intervention, quick elections, institutional continuity) is exactly the scenario where preventive engagement has the most leverage and the narrowest window. The distinction matters for policy more than it matters for risk scoring.

“The Mauritania template (intervention, quick elections, institutional continuity) is exactly the scenario where preventive engagement has the most leverage and the narrowest window. The distinction matters for policy more than it matters for risk scoring.”

3. The Forward Call: A Six-Month Trigger, Not a Year Target

The forward call is an event-trigger window, not a calendar target. The structural stress QGI is measuring does not wait for election day; in the Mauritania analogue, it materialized in the 18 months betweena credible election and a coup. The near-term trigger runs through the IMF Extended Credit Facility review cycle. Ghana entered the ECF in 2023 and is in its third review, with disbursements contingent on continued fiscal consolidation. The Mahama government inherited Bawumia's debt restructuring architecture and must sustain it against an NDC base that ran on relief from austerity. The six-month watch window is May–November 2026: if Ghana misses an ECF disbursement trigger or requests a substantial program modification, accompanied by public-sector wage arrears or fuel subsidy rollbacks, the fiscal-legitimacy stress will register in QGI's next scoring cycle. That is the leading indicator.

The falsification scenario.The call is off if the Mahama government navigates the third and fourth ECF reviews without significant program modification, and if ECOWAS reconstitutes a credible deterrence framework through the Accra Initiative, giving Ghanaian military officers a named reputational cost for intervention. Both conditions holding by November 2026 would signal that QGI's L-tier election-stress peak has decayed before reaching the Mauritania threshold.

The moderate form is most probable: not a junta, not a clean democratic handover, but a period of elevated civil-military tension where the military's role as a fiscal-crisis stabilizer becomes politically central , exercised through informal veto power over economic policy rather than an overt seizure of government. A disputed 2028 electoral outcome resolved through military mediation (the Mauritania 2009 pattern) would confirm this scenario. A coup before November 2026 would confirm the extreme form. A clean second electoral cycle with no institutional crisis would falsify it.

“A coup before November 2026 would confirm the extreme form. A clean second electoral cycle with no institutional crisis would falsify it.”

4. What QGI Doesn't See, And Why the Case Still Holds

Three named gaps warrant disclosure.

Foreign security presence.QGI's indicator base (World Bank development indicators, V-Dem electoral quality scores, SIPRI arms transfer data) cannot see informal private military company (PMC) approaches. Wagner Group / Africa Corps has operated in Burkina Faso, in the operational theater that borders Ghana's northern corridor directly. The commercial logic of PMC deployments (mining concession revenue in exchange for security services) maps precisely onto Ghana's gold-exporting economy. This is not an assertion that PMC discussions are occurring; it is an observation that QGI would not surface them if they were. Field knowledge has to carry this.

AES bloc-level political coherence.Niger's Elections & Voting score is 93/100, placing it #10 globally in the same recipe cohort as Ghana. Niger sits directly north of Ghana's northern corridor. The Alliance des États du Sahel has explicitly positioned itself as a peer legitimacy-granting structure, and a Ghanaian military government recognized by Bamako and Niamey would have a form of regional standing that did not exist before 2023. This bloc-level political reality is invisible to QGI's per-country scoring. It changes the expected consequences of a coup attempt from near-certain AU isolation (the 2008 Mauritania model) to a more ambiguous regional response.

Civil society resilience. Ghana's civil society (the Commission on Human Rights and Administrative Justice, the Ghana Centre for Democratic Development, the major media houses) proved robust during the 2008 election dispute, resolving a contested outcome within constitutional frameworks. QGI cannot measure institutional resilience directly; it measures indicator trajectories. The Mauritania analogue involved weaker civil society. This is the most important reason Ghana's structural risk, conditional on the indicator conditions being met, may be lower than the cosine similarity alone implies.

Despite these gaps, the structural case holds because of what QGI can see: the deteriorating fiscal-political trajectory in Ghana's Economic & Financial Stress dimension (salience 1.01 on the L1 profile), the Political Power & Regime dimension at maximum salience in the current scoring run (1.66, the same dimension that accounts for 66% of the Ghana-Mauritania cosine match), and the election-stress recipe at the 99th percentile of a validated global cohort. The gaps do not negate the structural reading; they bound the probability and shape the intervention options.

Published 2026-05-19· QG Intelligence · By Amin Al-Ait · Salience values cited from the 2026-04-30 V1.9.1 scoring run. This piece reports structural pattern resemblance and historical analogue comparison. It is not a forecast.