Five Models, One Country: What QGI's Structural Scan Finds in Pakistan That Macro Analysis Misses

The standard country-risk read on Pakistan in mid-2026 is one of cautious stabilisation. The IMF's Extended Fund Facility, approved in September 2024, has held. Headline inflation, which peaked at 38 percent in 2023, has fallen sharply. Foreign exchange reserves have rebuilt to roughly three months of import cover after near-depletion in the crisis of 2022-23. India-Pakistan relations, following a ceasefire after the Pahalgam attack of April 2025, are on their most cooperative diplomatic footing in years. The elected government of Shehbaz Sharif is implementing a fiscal adjustment program, growth has resumed at around 3 percent in the current fiscal year, and the rupee has stabilised. Operationally, the conventional picture is: still fragile, still watching the debt-service calendar, but no longer in the acute danger zone.

QGI is a structural pattern-matching system trained on 25-year indicator trajectories across more than 170 countries. It does not read news or track election calendars; it reads the shape of how dozens of institutional, economic, and political indicators have been moving, relative to how they moved in other countries before the events those countries are known to have experienced. In its June 2026 scoring run, Pakistan ranks in the top 8 to 20 countries globally across five independent structural models simultaneously: labor action, religious and social movement, political repression, foreign intervention, and democratic transition. No single macro indicator produces this convergence. The five models use different indicator sets, were trained on different historical evidence bases, and capture different crisis types. The convergence across all five is visible only through a cross-national, multi-recipe structural scan.

And the closest structural analogue to Pakistan 2026, out of every country-year QGI scores, is not a peer in open crisis. It is Pakistan itself.

In 2011, at a match score of 0.9712²Analogue match scores are cosine similarity values between two indicator vectors in the recipe's feature space. Cosine similarity ranges from 0 (orthogonal, no structural resemblance) to 1 (identical configuration). QGI uses intersection-only cosine: only indicators present in both vectors contribute to the score. A score of 0.97 means the current configuration is nearly identical to the historical comparator along the recipe's indicator dimensions.Read more → in the foreign intervention model.

That comparison is the story.

1. The Structural Picture: Five Models, Five Top-20 Rankings

QGI's scoring vintage te-fleet-run-2, scored at 2026-06-01T22:58:47 UTC, places Pakistan as follows across the endogenous structural recipes:

The overall composite structural risk score is 81 out of 100 (confidence interval: 75 to 88), placing Pakistan approximately 30th globally across 156 scored countries. Countries carrying higher composite scores include Somalia, Lebanon, Myanmar, Ukraine, Sri Lanka, Ethiopia, Iraq, Sudan, Libya, and Israel. Pakistan's structural risk picture is not the highest on the board, but it is the most multi-dimensional: the simultaneous top-20 presence across five recipes that measure categorically different historical processes is a configuration that QGI's full scoring universe does not produce often.

Two recipes are deliberately switched off here. Natural disaster and epidemic-and-pandemic are exogenous shocks, things that happen to a country, not crises its own structure builds toward. They fail the endogenous-shape test that gives QGI's comparisons their value, so they do not count toward this profile.

“The same underlying institutional erosion is driving risk across three structurally independent models simultaneously.”

What makes this convergence analytically significant is the independence of the models. The labor action recipe uses a specific set of indicators about civil liberties, political freedoms, and institutional capacity. The foreign intervention recipe uses a separate set dominated by fiscal state capacity, gender empowerment, and demographic variables. The democratic transition recipe weights political participation, corruption, and neopatrimonialism. That all five models independently place Pakistan in the global top 8 to 20 means that the same underlying structural deterioration is legible across multiple different indicator windows.

2. The Primary Driver: What the Indicators Are Actually Saying

Political Civil Liberties as the Through-Line

One indicator appears across multiple recipe models as a primary or significant driver: V-Dem's Political Civil Liberties Index (v2x_clpol), which captures freedoms of assembly, association, and political expression. In the June 2026 vintage, Pakistan's Political Civil Liberties Index carries a z-score change of -2.968³SHAP (SHapley Additive exPlanations) values decompose each country's score into indicator-level contributions. The percentage figures cited in this report represent each indicator's share of the total absolute signal. Positive SHAP pushes the score toward pre-event resemblance; negative SHAP offsets it. The SHAP decomposition comes from the scoring-refit XGBoost model, not the LOCO-validated training model, so driver percentages should be read as explanatory rather than as out-of-sample validated features.Read more →, a steep and fast deterioration relative to the cross-national distribution. This single indicator threads through the labor action model (18.2% of signal), the religious and social movement model (6.7% of signal), and the foreign intervention model as a contextual signal.

The real-world context beneath that z-score is specific. Since the removal of Imran Khan as Prime Minister in April 2022, Pakistan Tehreek-e-Insaf (PTI) has been subjected to a systematic judicial and administrative campaign: the party's election symbol removed by the Election Commission; Khan convicted on multiple charges including a 14-year sentence under the Official Secrets Act; hundreds of PTI leaders and supporters charged under anti-terrorism provisions following the events of May 2023, when supporters briefly ransacked military installations. The February 2024 elections produced PTI-affiliated independents as the largest single bloc, but a coalition of PML-N and PPP formed the government. As of mid-2026, Khan remains imprisoned and PTI's organisational capacity is severely degraded.

The Labor Action Model: Why Pakistan Is Globally 8th

The labor action model's top driver is population size (SP.POP.TOTL, 22.9% of signal, z-score +3.097): Pakistan has crossed 250 million people, and scale is itself a structural amplifier. But the operative driver is the Political Civil Liberties Index decline at -2.968, combined with a Private Civil Liberties Index (v2x_clpriv, covering property rights and private freedoms) moving in the opposite direction at +1.275. This divergence (political liberties contracting while private liberties hold) is the structural fingerprint of a state where collective action is being squeezed but individual private life remains relatively intact. QGI's model has learned, from historical cross-national data, that this configuration is associated with organised labour mobilisation becoming a vehicle for political expression precisely because the formal political channels are closing.

The momentum in this score has been declining gradually and is now slightly accelerating downward: 65.5 in 2023, 64.3 in 2024, 64.7 in 2025, 63.7 in 2026. The structural score at the 96th percentile reflects accumulated trajectory history, not just the most recent point.

The Foreign Intervention Model: Fiscal Capacity and the Self-Rhyme

The foreign intervention model identifies the following top drivers for Pakistan in 2026. Fiscal Capacity⁴The foreign intervention model's top analogue is Pakistan 2011. Several of the model's primary drivers (v2x_gender, v2stfisccap, v2xeg_eqprotec) are V-Dem indicators that carry a one-to-two year coding lag relative to real-world events. This means the 2026 vintage reflects V-Dem coding through approximately 2024-25. Developments in Pakistan's gender empowerment environment and fiscal capacity after that window are not yet captured.Read more →(v2stfisccap, 26.7% of signal) is declining at a z-score change of -1.961, a significant deterioration in the state's ability to extract revenue relative to its size and structural requirements. Women's Political Empowerment (v2x_gender, 24.4% of signal) is declining at -2.154, a sharper move that functions in QGI's model as a proxy for the quality of state-society relations and institutional inclusiveness. Two partially offsetting signals are also present: Equal Protection Index (v2xeg_eqprotec, 21.6% of signal, improving at +0.576 z-score) and GDP growth (NY.GDP.MKTP.KD.ZG, 20.1% of signal, slightly negative SHAP contribution). The net configuration still produces a rank-20 score because the two declining signals outweigh the offsetting ones.

Life expectancy improving at +2.746 z-score (SP.DYN.LE00.IN, 7.2% of signal) is also present as a positive signal in the model, and it matters for interpretation: Pakistan is not a collapsing state. Human development metrics are improving even as institutional quality erodes. This is precisely the structural configuration (a state with rising population health outcomes and a demographic dividend, combined with fiscal weakness and constrained political freedoms) that QGI's model associates with external military or security involvement in the historical evidence base.

The Democratic Transition Signal: Between Brazil 1974 and Uganda 2021

The democratic transition model places Pakistan at the 79.9th percentile (rank 36 of 174). This recipe carries an Experimental tier and a HARD_FAIL verdict in the current walk; treat the score as directional orientation only. The top drivers are Political Participation (v2x_partip, 27.4% of signal, falling at -1.846 z-score), population scale (SP.POP.TOTL, 22.8% of signal), and Political Corruption Index (v2x_corr, 21.5% of signal, declining at -1.902). Neopatrimonialism (v2x_neopat, 17.1% of signal) is also rising at +2.098 z-score, and the Pakistani military's dominant position in the country's political economy maps precisely onto that signal. The momentum in this score has edged upward in the most recent period: 67.6 in 2023, 67.4 in 2024, 67.9 in 2025, 68.6 in 2026.

3. The Historical Analogue: What Pakistan 2011 Implies

The foreign intervention model's top analogue for Pakistan 2026 is Pakistan itself in 2011, with a match score of 0.9712⁵The foreign intervention recipe has LOCO AUC of 0.7227 (Directional tier) but has not yet completed beta calibration, meaning the mapping between score_pp and event-base-rate probability has not been formally validated. The Pakistan score of 89.0 (rank 20 of 172) is valid as a structural resemblance signal; it should not be read as an event probability. Self-analogues can inflate cosine similarity by sharing country-fixed structural characteristics that are not event-predictive.Read more →, near-perfect structural rhyme to its own most turbulent year.

2011: The Year Pakistan's Civil-Military Architecture Was Stress-Tested at Every Joint

The twelve months of 2011 delivered, in rapid succession: the arrest and subsequent diplomatic standoff over CIA contractor Raymond Davis; the US Navy SEAL raid on the Abbottabad compound killing Osama bin Laden in May, conducted without Pakistani knowledge or consent, in a garrison town housing a major Pakistani military academy; the Salala checkpoint attack in November, in which US-led NATO forces killed 24 Pakistani soldiers on the Afghan border, prompting Pakistan to close NATO supply lines and boycott the Bonn Conference; and the Memogate affair, in which a Pakistani-American businessman was accused of secretly seeking US help to prevent a military coup. All of this occurred against the backdrop of the highest TTP attack tempo on record and the drone strike campaign operating at its most intensive.

The Pakistani state's fiscal capacity in 2011 was stretched by exactly the same dynamics visible today: heavy security spending, IMF-contingent external borrowing, and a revenue base inadequate to the demands being made of the state. Women's political empowerment was already under structural pressure from both internal dynamics and the deteriorating Afghan spillover. These are the variables the foreign intervention model is measuring. The 0.9712 match score says: Pakistan's current structural configuration is nearly identical to the configuration it held in 2011 along these indicator dimensions.

What the Second and Third Analogues Add

Haiti 1994 (match 0.9497) and Somalia 1994 (match 0.9268) are the second and third analogues in the foreign intervention model. Both are cases of external military presence responding to state fiscal fragility and institutional failure: Haiti 1994 was the US-led multinational intervention to restore President Aristide after the military coup; Somalia 1994 was the catastrophic conclusion of the UNOSOM II mission following Black Hawk Down, with US and UN forces withdrawing.

Pakistan is not Haiti 1994 or Somalia 1994 in any direct sense. Its institutions are far more robust, its military is a full-spectrum professional force, and it has nuclear weapons. But the structural signal the model is detecting (the specific combination of fiscal weakness, political empowerment deterioration, and scale) is what those cases had in common with Pakistan 2011, and what Pakistan 2026 now shares with all three.

The Russia 1989 Labour Comparator

In the labor action model, the closest non-self analogue is Russia 1989 (match 0.8449). In 1989, Soviet coal miners in the Donbas and Siberia launched the first major independent labour action in Soviet history, demanding better conditions and, increasingly, political reforms that the Gorbachev government could not fully deliver. The second analogue is Brazil 1978 (match 0.8166): the ABC metalworkers' strikes in the industrial suburbs of Sao Paulo under the military government, which launched Luiz Inacio Lula da Silva's political career and began the long transition that produced democracy in 1985.

The model is not predicting a Soviet-style collapse or a Lula-style labour-led political transformation in Pakistan. But both analogues are cases where a large young population with constrained political liberties found in organised labour, in the absence of open political channels, the vehicle for expressing political discontent that then became structurally transformative. Pakistan has a population of 250 million, over 60 percent under age 30, rapidly improving child health outcomes, and a political civil liberties environment that has contracted at -2.97 standard deviations in the current vintage.

The Uganda 2021 Repression Mirror

Like Democratic Transition, the political repression model carries an Experimental tier and a HARD_FAIL verdict in the current walk, so its score is structural orientation only, not a validated probability. The political repression model's top analogue is Uganda 2021 (match 0.9057, secondary: elections and voting). Uganda 2021 was Yoweri Museveni's post-election systematic persecution of Bobi Wine's National Unity Platform: hundreds of supporters killed in pre-election violence, Wine himself shot and arrested, widespread detention of opposition figures after a heavily manipulated vote. Rwanda 2021 (match 0.8885) and Zimbabwe 2019 (match 0.8805) follow as second and third analogues.

The Zimbabwe analogue deserves particular attention for Pakistan's current circumstances. Zimbabwe 2019 combined post-coup political consolidation under Mnangagwa with economic crisis and a currency collapse. The structural co-occurrence of political repression and currency or fiscal distress in the same episode is the signature. Pakistan's current configuration of PTI suppression coinciding with IMF program conditionality and structural fiscal weakness carries the same dual-stress signature.

The political repression model's cosine similarity, measuring how closely the current indicator trajectory matches the historical pre-repression movement pattern, is at 0.304, rated HIGH. The Liberal Democracy Index (v2x_libdem) is falling at -2.841 z-score. Legislature oversight capacity (v2lgqstexp) is declining at -1.283. Freedom from Political Killings is improving at +1.627, which means the model is not reading Pakistan as in active violent state repression, but as in the institutional degradation that precedes and enables it.

“The crackdown has been sustained long enough to have normalised in coverage. The structural conditions that enabled it remain fully in place and have actually deepened.”

4. The News-versus-Structure Picture: Five Divergences

QGI's high-frequency news layer (hf-topics.json, built 2026-06-09) provides live news pulse scores for Pakistan across 14 topic categories. The most analytically significant finding from this layer is not what the news is covering but what it is not covering relative to what the structural models show.

The calibration check is important before reading the divergences. Three topics (Trade, Sanctions and Investment at pulse 100; Diplomacy and Alignment at pulse 100; and Energy, Resources and Environment at pulse 99) are fully ACTIVE. These map precisely to what is actually dominating coverage of Pakistan in mid-2026: the IMF EFF negotiations and structural benchmarks, the India-Pakistan ceasefire diplomatic track, the CPEC renegotiation with Beijing, and the energy sector pricing reforms required under the IMF program. The HF layer is calibrated correctly. The silence on other topics is not a data artifact.

The Repression Divergence: Structural Score 82.6, News Pulse 25

The most extreme structural-versus-news divergence in Pakistan's full profile is on Repression and Human Rights: structural score 82.6, news pulse 25. The structural score places Pakistan in a globally elevated zone on the institutional conditions that historically co-occur with state crackdowns on civil and political rights. The news pulse at 25 (roughly one-third of what would be expected given that structural backdrop) means international media is running well below the coverage volume that a structural backdrop this elevated would typically accompany.

This divergence is not primarily explained by the PTI crackdown being unknown or unreported. The issue is rather that the crackdown has been sustained long enough to have normalised in coverage: the international media cycle that follows an acute event has largely passed, while the structural conditions that enabled the crackdown remain fully in place and, by QGI's indicators, have actually deepened. The v2x_libdem decline (-2.841 z-score) and the v2lgqstexp decline (-1.283 z-score) suggest this is not a resolved situation being correctly deprioritised. It is an ongoing institutional erosion that coverage has underweighted because the acute flashpoints are less visible than they were in 2023.

The Terrorism Divergence: Structure at 82.9, Pulse at 50

The terrorism and targeted violence structural score of 82.9 against a news pulse of 50 points to a different divergence. TTP attacks in Khyber Pakhtunkhwa and Balochistan have been running at elevated levels since the ceasefire with the government collapsed in late 2022. The QGI structural score measures the institutional and political configuration that historically co-occurs with elevated terrorism frequency. The Afghan border situation, the fiscal constraints on security spending, and the political fragmentation of governance in KPK combine to sustain a structural risk that international coverage is not fully reflecting at current pulse 50.

5. The Forward Look: Resemblance, Precedent, and the Bifurcation

QGI's framework requires a falsifiable forward statement. It does not produce timing or probability estimates. What the structural pattern-match supports is a statement about the configuration and the historical precedents for how configurations of this type have resolved.

Pakistan's structural configuration in 2026 has, across five independent models, produced historical analogues that cluster in two distinct resolution categories.

The democratic transition trajectory.Pakistan 2013 (match 0.9777 in the democratic transition model) is the structural self-analogue in that recipe: Pakistan's first-ever civilian-to-civilian transfer of power, from Asif Ali Zardari's PPP government to Nawaz Sharif's PML-N. Brazil 1974 (match 0.9186) is the cross-national comparator: Geisel's controlled abertura, a managed liberalisation under military pressure that eventually produced full democracy in 1985. Nigeria 1998 (match 0.9262) is the final year of the Abacha dictatorship before the transition to Obasanjo.

The authoritarian consolidation trajectory. Uganda 2021 (match 0.9057 in the political repression model) is the downside case: a contested election followed by systematic suppression of the opposition, with Museveni extending his now-four-decade rule. Zimbabwe 2019 (match 0.8805) combined political consolidation with currency crisis, producing a years-long economic deterioration alongside political stasis.

The structural evidence does not discriminate between these trajectories. Both are consistent with the current configuration.

“The structural evidence does not discriminate between these trajectories. Both are consistent with the current configuration.”

The falsifiable forward call has three observable components. First: if Pakistan's Political Civil Liberties Index (v2x_clpol) stabilises and reverses over the next two V-Dem coding cycles, the labor action and political repression scores should fall materially from their current positions. Second: if Fiscal Capacity (v2stfisccap) improves as the IMF program builds structural state revenue capacity, the foreign intervention score should recede. Third, and most immediately: the Repression and Human Rights divergence (structural score 82.6, news pulse 25) is the configuration to watch. If repression intensifies and news coverage rises to match it, that combined configuration has historically preceded the Uganda 2021 resolution. If instead the PTI ban is lifted, Khan released, and the February 2024 election results are more fully honoured, the democratic transition model should see its closest analogue shift from Nigeria 1998 (Abacha's last year) toward Nigeria 1999 (Obasanjo's inauguration).

6. The Multi-Recipe Convergence: What This Means for Risk Analysis

The central methodological argument behind this report is this: no desk analyst reviewing Pakistan from any single source of data would find this convergence. An analyst reviewing V-Dem data would see the institutional deterioration but might attribute it to a temporary election cycle. An analyst reviewing fiscal data would see the IMF program as the stabilising frame. An analyst following news would see the diplomatic progress with India as an offsetting signal.

The point of the five-model convergence is that it is not visible from any single vantage point. The same Political Civil Liberties deterioration (one indicator, one trajectory) appears in five different crisis models because five different historical crisis types are structurally associated with that indicator configuration. Each model sees it through a different historical lens. The labor model sees it as the configuration before organised workers act collectively in the absence of political outlets. The political repression model sees it as the institutional degradation before state crackdown. The foreign intervention model sees it through the fiscal and empowerment co-occurrence. The democratic transition model sees it through the neopatrimonialism frame. The religious and social movement model sees it as the pressure-cooker configuration that precedes mobilisation.

The mass protest finding is itself informative.Pakistan ranks globally last (174th of 174) in the mass protest structural model. The structural configuration most closely associated with classical street protest movements is not what Pakistan 2026 structurally resembles. The model identifies Bangladesh 2015, Zimbabwe 2016, and Chad 2021 as the closest analogues to what Pakistan's structural profile does resemble in the protest recipe: repressive-environment, limited-turnout, constrained-channel protest, not the full-spectrum mass mobilisation type. An analyst looking for signs of Tahrir Square-style protest as the leading indicator of Pakistan instability is looking at the wrong signal.