Data Visualization by Jona Wilke
Trade-restricting measures nearly tripled between 2019 and 2022. Fernández-Villaverde, Mineyama & Song distil 16 indicators into a single composite index and trace the causal link to GDP. The world has reversed three decades of integration in half the time it took to build them.
About the paper
“Geopolitical fragmentation is inherently multifaceted and challenging to define, let alone measure,” write Fernández-Villaverde (Penn), Mineyama (IMF), and Song (Johns Hopkins). The difficulty is that any single indicator, whether trade openness or sanctions counts, captures only one facet of a phenomenon that spans trade, finance, mobility, and politics simultaneously. A simple average fares no better, since it ignores the varying signal strength of each proxy.
The authors propose a dynamic hierarchical factor model with time-varying parameters and stochastic volatility. The model treats geopolitical fragmentation as an unobservable latent variable, using 16 empirical indicators as noisy proxies. Think of it as a filter that separates what all 16 indicators have in common from what is specific to trade, finance, mobility, or politics alone. Five measure trade (openness, restrictions, barriers, policy uncertainty, tariffs). Three capture finance (FDI, capital flows, capital controls). Three track mobility (migration, patents, migration fear). Five reflect politics (geopolitical risk, energy uncertainty, conflicts, sanctions, UN voting alignment).
The hierarchical structure is the central innovation. The model separates dynamics common to all groups (the common fragmentation factor) from dynamics specific to each dimension (group-specific factors), letting the data assign the weights rather than the researchers. Estimated via a Gibbs sampler with Bayesian priors, it accommodates missing observations, mixed frequencies, and non-stationarity.
The causality analysis pairs structural vector autoregressions (SVARs) with local projections (LPs) on quarterly panel data covering 89 economies, 36 advanced and 53 emerging. Three identification strategies (Cholesky decomposition, narrative restrictions drawn from 24 historical episodes, and external controls for military and monetary shocks) yield consistent results.
During the Cold War, fragmentation was the status quo.
The common factor distils 16 noisy indicators, from trade openness ratios and FDI flows to UN voting patterns and sanctions counts, into a single index via Bayesian estimation.
From 1975 to the early 1990s it barely moves. Trade openness, FDI ratios, and financial flows were rising, but the global economy remained split between market and socialist blocs. Integration was happening within blocs, not across them.
The Soviet collapse set off three decades of convergence.
The Berlin Wall falls in 1989. The Maastricht Treaty creates the EU. The WTO is established. China joins in 2001. The index plunges: by the mid-2000s it reaches its lowest point in the entire sample. The word “globalization” itself, the authors note, only gained currency beyond economics during this era.
By 2007, capital moved freely across borders, trade barriers had fallen to historic lows, and political cooperation was the presumed default. In retrospect, it was the high-water mark.
The trough around 2007 marks peak globalization, a level the world may never return to.
Then the index turned. It has not stopped climbing since.
The Global Financial Crisis. Crimea. The US-China trade war, which saw trade-restricting measures nearly triple between 2019 and 2022. COVID-19. Russia's full-scale invasion of Ukraine. Each shock ratcheted fragmentation higher; none was reversed.
By 2024 Q1 the common index has erased thirty years of integration gains in barely fifteen. The world is more fragmented than at any point since the dataset begins, and the trajectory shows no sign of inflecting.
Thirty years of globalisation, unwound in fifteen. The trend shows no sign of reversing.
During the Cold War, fragmentation was the status quo.
The common factor distils 16 noisy indicators, from trade openness ratios and FDI flows to UN voting patterns and sanctions counts, into a single index via Bayesian estimation.
From 1975 to the early 1990s it barely moves. Trade openness, FDI ratios, and financial flows were rising, but the global economy remained split between market and socialist blocs. Integration was happening within blocs, not across them.
The Soviet collapse set off three decades of convergence.
The Berlin Wall falls in 1989. The Maastricht Treaty creates the EU. The WTO is established. China joins in 2001. The index plunges: by the mid-2000s it reaches its lowest point in the entire sample. The word “globalization” itself, the authors note, only gained currency beyond economics during this era.
By 2007, capital moved freely across borders, trade barriers had fallen to historic lows, and political cooperation was the presumed default. In retrospect, it was the high-water mark.
The trough around 2007 marks peak globalization, a level the world may never return to.
Then the index turned. It has not stopped climbing since.
The Global Financial Crisis. Crimea. The US-China trade war, which saw trade-restricting measures nearly triple between 2019 and 2022. COVID-19. Russia's full-scale invasion of Ukraine. Each shock ratcheted fragmentation higher; none was reversed.
By 2024 Q1 the common index has erased thirty years of integration gains in barely fifteen. The world is more fragmented than at any point since the dataset begins, and the trajectory shows no sign of inflecting.
Thirty years of globalisation, unwound in fifteen. The trend shows no sign of reversing.
Three times the magnitude of any other dimension. The political fracture (sanctions, conflicts, collapsing UN voting alignment) is not merely tracking the other dimensions. It is pulling them forward.
Scroll to see how the four dimensions have diverged.
Beneath the common index, four dimensions tell different stories.
The hierarchical model decomposes the common signal into four group-specific factors: trade, financial, mobility, and political. Each is driven by its own subset of indicators, and each follows a distinct trajectory.
Trade fragmentation, for instance, declined more slowly than the common factor during globalisation, falling below zero only around 2005. Trade-to-GDP ratios have been flat since 2008, even as restrictions have multiplied, a phenomenon the IMF has labelled “slowbalisation.”
Financial fragmentation tracks the common factor almost exactly.
Of the four dimensions, the financial factor comoves most tightly with the overall index. Cross-border capital flows, FDI ratios, and the financial flow ratio all surged during hyperglobalization and all have reversed sharply since 2008. Sanctions regimes, capital controls, and the weaponisation of dollar-based payments have accelerated the decoupling.
Financial fragmentation moves nearly one-to-one with the common factor, the tightest correlation of any dimension.
The political dimension, however, has broken away entirely.
Unlike the other three factors, the political dimension “sharply diverges from the common factor, surging post-2008,” the authors write. Sanctions have multiplied. Violent conflicts tracked by the Uppsala Conflict Data Program have intensified. The UN General Assembly kappa score, a measure of bilateral voting alignment, has collapsed.
At 3.48, the political factor is three times the magnitude of any other dimension and serves as the leading indicator. Trade and financial fragmentation follow where politics leads.
Political fragmentation running at 3x every other dimension is a leading indicator for economic costs yet to materialise.
Beneath the common index, four dimensions tell different stories.
The hierarchical model decomposes the common signal into four group-specific factors: trade, financial, mobility, and political. Each is driven by its own subset of indicators, and each follows a distinct trajectory.
Trade fragmentation, for instance, declined more slowly than the common factor during globalisation, falling below zero only around 2005. Trade-to-GDP ratios have been flat since 2008, even as restrictions have multiplied, a phenomenon the IMF has labelled “slowbalisation.”
Financial fragmentation tracks the common factor almost exactly.
Of the four dimensions, the financial factor comoves most tightly with the overall index. Cross-border capital flows, FDI ratios, and the financial flow ratio all surged during hyperglobalization and all have reversed sharply since 2008. Sanctions regimes, capital controls, and the weaponisation of dollar-based payments have accelerated the decoupling.
Financial fragmentation moves nearly one-to-one with the common factor, the tightest correlation of any dimension.
The political dimension, however, has broken away entirely.
Unlike the other three factors, the political dimension “sharply diverges from the common factor, surging post-2008,” the authors write. Sanctions have multiplied. Violent conflicts tracked by the Uppsala Conflict Data Program have intensified. The UN General Assembly kappa score, a measure of bilateral voting alignment, has collapsed.
At 3.48, the political factor is three times the magnitude of any other dimension and serves as the leading indicator. Trade and financial fragmentation follow where politics leads.
Political fragmentation running at 3x every other dimension is a leading indicator for economic costs yet to materialise.
Fragmentation splits along bloc lines, not national ones.
The authors re-estimate the model using eleven bloc-level indicators for three geopolitical groupings: the US-EU alliance, the China-Russia axis, and the heterogeneous “Others” bloc. Before 2008, all three moved roughly together. Globalization was a shared tide.
The “Others” bloc is instructive. Post-crisis regional cooperation (ASEAN, the African Union, China's Belt and Road Initiative) and trade substitution toward Mexico and Vietnam have kept its fragmentation relatively stable. The Global South, in aggregate, has not fragmented. It has been fragmented upon.
The China-Russia bloc is decoupling fastest, and the shocks compound.
Crimea in 2014 was the first inflection point: sanctions and counter-sanctions began severing this bloc from Western financial networks. The 2018-19 US-China trade war pushed fragmentation levels back to those of the early 1990s. Then in 2022, Russia's full-scale invasion of Ukraine triggered the most comprehensive sanctions regime in modern history, accelerating economic decoupling from the West and deepening dependence on China for trade and financial partnerships.
By 2024, the China-Russia bloc sits at roughly three times the fragmentation level of the US-EU bloc. The gap is accelerating.
The gap between blocs is now the widest on record.
The shaded area shows the growing divergence between the two major blocs. The asymmetry that matters most for the global economy is this: US-EU fragmentation shocks have immediate, significant effects on global GDP. China-Russia shocks, by contrast, show no significant impact, likely because China's global expansion has partially substituted for lost Western linkages.
When the US-EU bloc fragments, the world economy contracts. When the China-Russia bloc fragments, the costs are absorbed internally or redirected. The architecture of global finance still runs through the West.
US-EU shocks hit global GDP immediately; China-Russia shocks are absorbed. The damage is asymmetric because global finance still runs through the West.
Fragmentation splits along bloc lines, not national ones.
The authors re-estimate the model using eleven bloc-level indicators for three geopolitical groupings: the US-EU alliance, the China-Russia axis, and the heterogeneous “Others” bloc. Before 2008, all three moved roughly together. Globalization was a shared tide.
The “Others” bloc is instructive. Post-crisis regional cooperation (ASEAN, the African Union, China's Belt and Road Initiative) and trade substitution toward Mexico and Vietnam have kept its fragmentation relatively stable. The Global South, in aggregate, has not fragmented. It has been fragmented upon.
The China-Russia bloc is decoupling fastest, and the shocks compound.
Crimea in 2014 was the first inflection point: sanctions and counter-sanctions began severing this bloc from Western financial networks. The 2018-19 US-China trade war pushed fragmentation levels back to those of the early 1990s. Then in 2022, Russia's full-scale invasion of Ukraine triggered the most comprehensive sanctions regime in modern history, accelerating economic decoupling from the West and deepening dependence on China for trade and financial partnerships.
By 2024, the China-Russia bloc sits at roughly three times the fragmentation level of the US-EU bloc. The gap is accelerating.
The gap between blocs is now the widest on record.
The shaded area shows the growing divergence between the two major blocs. The asymmetry that matters most for the global economy is this: US-EU fragmentation shocks have immediate, significant effects on global GDP. China-Russia shocks, by contrast, show no significant impact, likely because China's global expansion has partially substituted for lost Western linkages.
When the US-EU bloc fragments, the world economy contracts. When the China-Russia bloc fragments, the costs are absorbed internally or redirected. The architecture of global finance still runs through the West.
US-EU shocks hit global GDP immediately; China-Russia shocks are absorbed. The damage is asymmetric because global finance still runs through the West.
A fragmentation shock cascades through the entire macroeconomy.
The authors trace the transmission mechanism using structural VARs with Cholesky identification. A one-standard-deviation fragmentation shock depresses four macroeconomic variables: GDP per capita, industrial production, fixed investment, and stock prices. All four decline significantly, but the timing and magnitude differ.
Stock prices react first and most sharply, dropping roughly 0.9% within three quarters as markets price in the expected damage to corporate earnings and cross-border capital flows. The real economy follows with a lag.
GDP per capita declines for two years, peaking at about -0.4%.
Unlike stock prices, which partially recover after their initial drop, GDP per capita continues deteriorating for eight quarters before stabilising. Even at quarter 16, four years after the shock, the economy has not returned to its pre-shock level.
This persistence is the central finding. Fragmentation does not produce a temporary dip. It shifts the level of output permanently downward, consistent with models in which trade disruptions reduce long-run productivity through lost specialisation gains and slower technology diffusion.
GDP declines for two full years and never fully recovers. Fragmentation acts as a permanent output loss.
The investment channel amplifies the damage: fixed investment falls 1.4%.
Fixed investment is hit hardest, declining 1.4% at its trough, more than three times the GDP response. Industrial production falls 1.0%. The mechanism is intuitive: fragmentation raises uncertainty about future market access, increasing the option value of waiting. Firms delay or cancel cross-border investment projects. Supply chain reconfiguration absorbs capital that would otherwise fund expansion.
This investment channel explains why fragmentation compounds over time. Each year of reduced capital formation lowers the productive capacity available for future growth, a feedback loop the authors identify as central to the persistent GDP effects.
Fixed investment falls 3.5x more than GDP. The investment channel turns a trade shock into a growth trap.
A fragmentation shock cascades through the entire macroeconomy.
The authors trace the transmission mechanism using structural VARs with Cholesky identification. A one-standard-deviation fragmentation shock depresses four macroeconomic variables: GDP per capita, industrial production, fixed investment, and stock prices. All four decline significantly, but the timing and magnitude differ.
Stock prices react first and most sharply, dropping roughly 0.9% within three quarters as markets price in the expected damage to corporate earnings and cross-border capital flows. The real economy follows with a lag.
GDP per capita declines for two years, peaking at about -0.4%.
Unlike stock prices, which partially recover after their initial drop, GDP per capita continues deteriorating for eight quarters before stabilising. Even at quarter 16, four years after the shock, the economy has not returned to its pre-shock level.
This persistence is the central finding. Fragmentation does not produce a temporary dip. It shifts the level of output permanently downward, consistent with models in which trade disruptions reduce long-run productivity through lost specialisation gains and slower technology diffusion.
GDP declines for two full years and never fully recovers. Fragmentation acts as a permanent output loss.
The investment channel amplifies the damage: fixed investment falls 1.4%.
Fixed investment is hit hardest, declining 1.4% at its trough, more than three times the GDP response. Industrial production falls 1.0%. The mechanism is intuitive: fragmentation raises uncertainty about future market access, increasing the option value of waiting. Firms delay or cancel cross-border investment projects. Supply chain reconfiguration absorbs capital that would otherwise fund expansion.
This investment channel explains why fragmentation compounds over time. Each year of reduced capital formation lowers the productive capacity available for future growth, a feedback loop the authors identify as central to the persistent GDP effects.
Fixed investment falls 3.5x more than GDP. The investment channel turns a trade shock into a growth trap.
Three identification strategies point to a causal link with GDP.
Using structural VARs and local projections on panel data covering 89 economies (36 advanced, 53 emerging), the authors estimate the causal effect of fragmentation shocks under three identification strategies: Cholesky decomposition, narrative restrictions drawn from 24 historical episodes, and external controls for military and monetary shocks.
The baseline Cholesky estimate puts the GDP per capita loss at about 0.4% per one-standard-deviation shock. When instrumented with the narrative series, the effect rises to 0.7%. Industrial production, fixed investment, and stock prices all decline, with negative effects peaking one to two years after the initial shock.
The countries least responsible for fragmentation bear the heaviest burden.
The disaggregated results sharpen the picture. Advanced economies lose 0.2 to 0.3% of GDP per shock. Emerging markets lose 0.6 to 0.9%, roughly three times as much.
The logic is straightforward. Emerging markets depend on open trade networks for catch-up growth, on foreign capital inflows for investment, and on technology transfer from advanced economies. Sever those links, and the periphery suffers disproportionately while the core endures.
Emerging markets lose 3x more GDP than advanced economies. The structure of fragmentation is regressive by construction.
The sectoral breakdown shows where the damage concentrates.
The authors run the same local projection analysis across OECD sectors. Industries tied to global markets (manufacturing, construction, wholesale and retail trade, finance and insurance) show large, significant GDP declines. Domestically oriented sectors (agriculture, real estate, public administration) barely register a response.
Fragmentation does not affect all industries equally. It concentrates on the sectors that benefited most from globalisation: those built on cross-border supply chains, capital flows, and specialisation.
Manufacturing loses over 2% of output. Public services lose essentially nothing. The damage is concentrated, not diffuse.
Fragmentation breaks things quickly. Globalisation heals them slowly.
The impulse response functions reveal a final asymmetry. Rising fragmentation shocks produce “immediate adverse effects on the global economy.” GDP, industrial production, and investment all decline within two to three quarters. But declining fragmentation shocks, the benefits of renewed globalisation, “unfold gradually over 2 to 3 years, exhibiting greater persistence.”
Even if every geopolitical tension resolved tomorrow, the economic recovery would be slow. Supply chains, once broken, do not reassemble overnight. Trust between trading partners takes years to rebuild. The costs are front-loaded; the benefits are back-loaded. Destroying integration is far easier than rebuilding it.
Costs arrive in quarters. Benefits take years. Each turn of the ratchet is quick; each reversal is painfully slow.
Three identification strategies point to a causal link with GDP.
Using structural VARs and local projections on panel data covering 89 economies (36 advanced, 53 emerging), the authors estimate the causal effect of fragmentation shocks under three identification strategies: Cholesky decomposition, narrative restrictions drawn from 24 historical episodes, and external controls for military and monetary shocks.
The baseline Cholesky estimate puts the GDP per capita loss at about 0.4% per one-standard-deviation shock. When instrumented with the narrative series, the effect rises to 0.7%. Industrial production, fixed investment, and stock prices all decline, with negative effects peaking one to two years after the initial shock.
The countries least responsible for fragmentation bear the heaviest burden.
The disaggregated results sharpen the picture. Advanced economies lose 0.2 to 0.3% of GDP per shock. Emerging markets lose 0.6 to 0.9%, roughly three times as much.
The logic is straightforward. Emerging markets depend on open trade networks for catch-up growth, on foreign capital inflows for investment, and on technology transfer from advanced economies. Sever those links, and the periphery suffers disproportionately while the core endures.
Emerging markets lose 3x more GDP than advanced economies. The structure of fragmentation is regressive by construction.
The sectoral breakdown shows where the damage concentrates.
The authors run the same local projection analysis across OECD sectors. Industries tied to global markets (manufacturing, construction, wholesale and retail trade, finance and insurance) show large, significant GDP declines. Domestically oriented sectors (agriculture, real estate, public administration) barely register a response.
Fragmentation does not affect all industries equally. It concentrates on the sectors that benefited most from globalisation: those built on cross-border supply chains, capital flows, and specialisation.
Manufacturing loses over 2% of output. Public services lose essentially nothing. The damage is concentrated, not diffuse.
Fragmentation breaks things quickly. Globalisation heals them slowly.
The impulse response functions reveal a final asymmetry. Rising fragmentation shocks produce “immediate adverse effects on the global economy.” GDP, industrial production, and investment all decline within two to three quarters. But declining fragmentation shocks, the benefits of renewed globalisation, “unfold gradually over 2 to 3 years, exhibiting greater persistence.”
Even if every geopolitical tension resolved tomorrow, the economic recovery would be slow. Supply chains, once broken, do not reassemble overnight. Trust between trading partners takes years to rebuild. The costs are front-loaded; the benefits are back-loaded. Destroying integration is far easier than rebuilding it.
Costs arrive in quarters. Benefits take years. Each turn of the ratchet is quick; each reversal is painfully slow.
Sixteen indicators, 197 quarters, 89 economies. The picture is unambiguous: globalisation is unwinding, political fractures are leading the way, and emerging markets bear the largest costs.
Fernández-Villaverde, Jesús, Tomohide Mineyama, and Dongho Song. “Are We Fragmented Yet? Measuring Geopolitical Fragmentation and Its Causal Effect.” NBER Working Paper 32638 (2024).
Visualization by Jona Wilke