## wpiea2023270-print-pdf - 8.5 percent when accounting for all three layers of fragmentation. Goes and Bekkers (2022)

## Source details

**Canonical URL:** [wpiea2023270-print-pdf - 8.5 percent when accounting for all three layers of fragmentation. Goes and Bekkers (2022)](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023270-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023270-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023270-print-pdf.pdf.json)

---

### Literature overview and context
- Prior estimates of welfare and output losses from geoeconomic fragmentation:
  - Losses of 0.4 percent of GDP (some countries) to 12 percent (most affected) under scenarios from mild fragmentation to full technological decoupling.
  - IMF (2022): ~1.2 percent of global GDP loss from eliminating trade in high-tech and energy between rival blocs; increases to 1.5 percent when barriers are extended to other sectors.
  - Felbermayr et al. (2022): welfare losses between 0 and 10 percent across countries under East-West trade “decoupling” scenarios that double existing NTBs.
  - Bolhuis et al. (2023): long-run output reduced by between 0.2 and 6.9 percent depending on scenario and elasticities.
  - Javorcik et al. (2022) and Attinasi et al. (2023): global output losses ranging from 0.1 to 5 percent using supply-chain decoupling frameworks.
- Empirical evidence: geopolitical alignment shapes bilateral FDI patterns and its effect strengthened since 2018 (Aiyar et al., 2023).

### Contribution and approach of this paper
- Provides a framework to discipline fragmentation counterfactuals across a large set of sectoral bilateral trade relationships by:
  - Using estimated sensitivity of trade barriers to geopolitical alignment across sectors from recent data.
  - Building scenarios that extrapolate from these empirical elasticities.
- Purposes:
  - Identify countries’ possible macroeconomic exposure to geoeconomic fragmentation ex ante.
  - Describe country characteristics that explain differences in exposure.
  - Spotlight impacts on economies outside major world markets.
- Novelty: First to quantify impact of geopolitical alignment on trade at a sectoral level and to integrate these estimates into standard quantitative trade models for counterfactual analysis.

### Structural gravity framework (methodology)
- Sector-level bilateral trade represented by structural gravity equation:
  - M_sn′n = [τ_sn′n O_sn′ P_sn]^{−θ_s} D_sn′ E_sn
  - P_sn and O_sn′ are inward and outward multilateral resistance terms as defined in the text.
- Two sufficient conditions to obtain the structural gravity form:
  1. Shares of spending v_sn′n are multiplicatively separable: v_sn′n = F_sn′ [τ_sn′n P_sn]^{−θ_s}, with P_sn defined accordingly.
  2. Market clearing for each origin country: D_sn′ = Σ_n M_sn′n = F_sn′ Σ_n [τ_sn′n P_sn]^{−θ_s} E_sn.
- Assumes ratio of external to internal trade barriers τ_sn′n / τ_snn is log-linear in country-pair characteristics:
  - ln(τ_sn′n / τ_snn) = β_0s + Σ_{l=1}^L β_ls u_l_{n′n}.

### Estimation strategy
- Two-step estimation:
  1. PPML estimate of full N×N bilateral expenditures including domestic expenditures:
     - M_sn′n = exp{Ω_sn′ + Π_sn + δ_sn′n} ζ_sn′n
     - Country-sector exporter fixed effect Ω_sn′, country-sector importer fixed effect Π_sn, undirected country-pair-sector fixed effect δ_sn′n.
     - From PPML, country-pair fixed effects map to −θ_s ln(τ_sn′n / τ_snn) = ˜δ_sn′n.
  2. OLS regression of estimated country-pair fixed effects on country-pair observables:
     - −˜δ_sn′n / θ_s = β_0s + Σ_{l=1}^L β_ls u_l_{n′n} + ε_sn′n
     - u_l includes: dist (log distance km), contig, lang, col, wto, rta, eu, and align (geopolitical alignment).
- Rationale: two-step approach accommodates missing country-pair characteristics (notably foreign-policy-alignment) while preserving theory-consistent identification of bilateral trade barriers.

### Data
- Trade flows:
  - Sector-level bilateral expenditure flows from the EORA global input-output database (Lenzen et al., 2013).
  - Uses simple average of 2017–19 EORA flows to avoid Covid-19 and Ukraine-war disruptions.
  - Aggregated to 10 goods sectors + 1 services sector; empirical analysis and counterfactuals restricted to goods sectors.
  - Dataset covers full matrix of sector-level bilateral trade flows between 185 economies (including “rest of the world”).
- Gravity controls:
  - Taken from the CEPII gravity dataset (Conte et al., 2022).
  - Sectoral trade elasticities: baseline from Caliendo and Parro (2015); robustness from Fontagné et al. (2022).
  - Elasticities vary from 0.69 in transport equipment to 15.72 in mining and quarrying.
- Bilateral geopolitical alignment:
  - Measured using ATOP-based alliance portfolio similarity (Signorino and Ritter, 1999) computed for 2018.
  - ATOP treaty coding treaty_{n′n} ∈ {0,1,2,3}: 3 = defense/offense; 2 = neutrality/consultation; 1 = nonaggression; 0 = no alliance obligation.
  - u_align_{n′n} bounded between −1 and 1; higher = greater foreign policy similarity.
  - 2018 values in practice range from -0.29 to 1; median 0.54; standard deviation 0.23.
  - Example: Germany–France alignment score 0.85; Germany–Angola alignment score 0.21.
- Sample size for regressions: 14,535 observations covering 185 countries.

### Key empirical estimates (impact of foreign policy alignment on bilateral trade barriers by sector)
- Table 1, Panel A (full controls) — foreign policy alignment (scale −1,1) coefficients by sector:
  - 1 – Agriculture and fishing: −0.014 (standard error 0.010) [not significant]
  - 2 – Mining and Quarrying: 0.023 (0.010)**
  - 3 – Food & Beverages: −0.249 (0.034)***
  - 4 – Textiles & Apparel: −0.148 (0.015)***
  - 5 – Wood & Paper: −0.073 (0.008)***
  - 6 – Petroleum, Chemical and Non-Metallic Mineral Products: −0.024 (0.006)***
  - 7 – Metal Products: −0.098 (0.014)***
  - 8 – Electrical Machinery: −0.025 (0.011)**
  - 9 – Transport Equipment: −0.340 (0.124)***
  - 10 – Other: −0.192 (0.016)***
  - R2 values across sectors range from 0.04 to 0.22.
- Table 1, Panel B (excluding economic-agreement controls) — foreign policy alignment (−1,1) coefficients by sector:
  - 1: −0.028 (0.010)***
  - 2: 0.014 (0.009) [not significant]
  - 3: −0.311 (0.035)***
  - 4: −0.193 (0.015)***
  - 5: −0.094 (0.008)***
  - 6: −0.045 (0.006)***
  - 7: −0.147 (0.015)***
  - 8: −0.066 (0.011)***
  - 9: −0.974 (0.130)***
  - 10: −0.236 (0.016)***
- Interpretation:
  - Greater foreign policy alignment is associated with lower bilateral trade barriers in most goods sectors; magnitude and significance vary by sector and by inclusion of controls for economic agreements.

### Model integration and counterfactuals (overview)
- Integrates sectoral empirical estimates into a dynamic many-country, many-sector quantitative trade model (special case of Cuñat and Zymek (2023)).
- Counterfactuals use gravity-estimated sensitivities to explore long-run macroeconomic impacts of fragmentation and distributional outcomes across countries.
- Distribution of losses across countries is highly contingent on scenario assumptions about size and incidence of new, fragmentation-induced trade barriers.

### 2.3 Results — Baseline findings
- Estimation approach: second-step OLS after purging origin and destination fixed effects from first-stage PPML.
- Main empirical findings (Table 1, Panel A):
  - Closer foreign policy alignment is associated with lower trade barriers in 9 out of 10 tradable goods sectors.
  - In 8 sectors the negative effect of alignment on trade barriers is statistically significant.
  - Gravity controls: greater bilateral distance increases trade frictions in all sectors; common border, common language, and membership in trade agreements decrease bilateral trade barriers.
  - Coefficients from Table 1 (Panel A) used in counterfactuals in Section 4.
- Magnitude and sectoral heterogeneity:
  - One standard deviation increase in foreign policy alignment = 0.233 points.
  - After accounting for trade elasticities, ranking of sectoral sensitivity:
    - Transport equipment: a one standard deviation increase in foreign policy alignment decreases trade barriers in the sector by almost 8 percent.
    - Food and beverages and other manufacturing are second- and third-most sensitive.
- Variance decomposition of bilateral trade barriers:
  - Share of total explained variance by sector ranges from 11 percent ("Mining and quarrying") to 21 percent ("electric machinery").
  - Explained variation dominated by geographic factors and economic agreements.
  - Differences in foreign policy alignments play a minor role, roughly on par with cultural variables.

### 2.3 Results — Robustness checks
- Effect persistence and sample checks:
  - Estimated elasticities stable across time (1993–2018) and across sub-samples (AE vs EMDE pairs).
- Alternative elasticity scaling:
  - Using Fontagné et al. (2022) elasticities yields broadly similar sector ranking; baseline scenarios run with both elasticity sets.
- Mechanism via economic/trade agreements:
  - Excluding economic-agreement controls increases estimated effect of foreign policy alignment, implying alignment also operates through propensity to enter economic agreements.

### Model structure and equilibrium (summary)
- Model class: dynamic trade model with forward-looking agents and sectoral Armington variety structure.
- Key calibration inputs:
  - Sectoral parameters {σ_sn}, {μ_s, θ_s}, capital share α.
  - Status-quo steady-state sectoral bilateral trade shares {v_{s n' n}} and world GDP shares {h_n}.
  - Capital share set to α = .406 (U.S. capital share for 2018 from PWT, edition 10.0).
- Steady-state comparison via exact-hat algebra; counterfactual change in bilateral trade barriers parameterized as (equation (29)):
  - ˆτ_{sn' n} = exp[ −1/θ_s ( ˆβ_align_s ˆu_align_{n' n} − 1 ) ̃β_align_s u_align_{n' n} ] if n' ≠ n; = 1 otherwise.

### Counterfactual scenarios (Section 4)
- General approach: change bilateral trade barriers via (i) changes in sensitivity ˆβ_align_s; (ii) changes in geopolitical alignments ˆu_align_{n' n}; or (iii) both.
- Scenario 1 — Geopolitical Polarization:
  - Blocs defined by treaty_{n USA} − treaty_{n CHN} ∈ {−3,...,3}.
  - Western: treaty_{n USA} > treaty_{n CHN}; Eastern: treaty_{n CHN} > treaty_{n USA}; Non-aligned: treaty_{n USA} = treaty_{n CHN}.
  - Polarization: in-bloc treaty strength +1 (capped at 3); out-of-bloc −1 (floored at 0). Non-aligned bilateral treaties unchanged except mechanically.
  - Within-bloc alignments ↑ → trade barriers ↓; between-bloc alignments ↓ → trade barriers ↑.
- Scenario 2 — Increased Geopolitical Sensitivity:
  - Set ˆβ_align_s = 2 for all s (doubling sensitivity).
  - This doubles share of variation in trade barriers attributable to foreign policy alignment.
- Scenario 3 — Combined Polarization and Increased Sensitivity:
  - Combines Scenarios 1 and 2; baseline geoeconomic fragmentation scenario.

### 4.2 Impact of Fragmentation on Real Incomes — Scenario outcomes
- Scenario 1 (“polarization only”):
  - Median economy: real-income loss ≈ 0.2 percent.
  - About one quarter of economies experience real-income gains.
  - Latin American and Caribbean economies are among biggest winners due to declines in trade barriers with the U.S.
- Scenario 2 (“increased geopolitical sensitivity of trade”):
  - Median economy: real-income loss ≈ 1 percent.
  - Regional median losses:
    - Emerging and developing Asia: 0.7 percent (smallest median loss).
    - Middle East and Central Asia: 1.5 percent (largest median loss).
- Scenario 3 (combined, baseline):
  - World median loss: 1.3 percent.
  - Median impacts by group:
    - Advanced economies: about 0.9 percent.
    - Emerging and developing Asia: median impact is 80 percent larger than for advanced economies.
    - Sub-Saharan Africa: median impact is more than 120 percent larger than for advanced economies.
    - Middle East and Central Asia: median impact is 150 percent larger than for advanced economies.
  - About one quarter of economies in sub-Saharan Africa and the Middle East and Central Asia experience real-income losses in excess of 3 percent.

### Drivers of differential exposure across countries
- Three determinants:
  1. Market size: smaller economies (by share of world GDP) suffer larger real-income losses under uniform trade-barrier increases.
  2. Comparative advantage / import basket: economies importing more in sectors sensitive to alignment face larger costs.
  3. Foreign policy alignments (and changes): trade barriers rise more for economies that are (or become) more geopolitically “distant” from major trading partners.
- Decomposition (equation (30)) separates:
  - ln ˆτ (weighted average change),
  - ln ˆτ^β_s (sector sensitivity component),
  - ln ˆτ^u_n'n (bilateral alignment level component),
  - ln ˆτ^{△u}_n'n (alignment change component),
  - η_sn'n (residual).

### Quantitative role of determinants (variance comparison, Table 2)
- Variance decomposition results (covariance contributions scaled by total variance):
  - All (baseline): 2.467 1.000
  - Size: ˆy_n({ˆτ}_sn′n): 1.264 0.512
  - Import basket: ˆy_n(ˆτβs_sn′n): 0.238 0.097
  - Alignments: ˆy_n({ˆτu_sn′n}_sn′n): 0.541 0.219
  - Alignment changes: ˆy_n(nˆτΔu_sn′n o_sn′n): 0.332 0.135
- Interpretations:
  - Uniform (average) trade-barrier change across all sectors and pairs (size channel) generates more than half of the variation in real-income changes — economy size is primary driver.
  - Differences in alignments and alignment changes together generate ~ one third of baseline variation.
  - Differences in import baskets account for < one tenth of baseline variance.

### Additional and alternative counterfactuals — mitigation implications
- Economic agreements determined by foreign policy alignment (endogenous RTAs):
  - If economic-agreement membership fully determined by alignment, median regional losses increase by between 60 percent and 100 percent relative to baseline — baseline is conservative.
- Alternative trade elasticities (Fontagné et al. (2022) vs Caliendo and Parro (2015)):
  - Using Fontagné et al. (2022) reduces magnitude of income losses somewhat; distribution across groups remains broadly unchanged.
- “Neutral” economies respond:
  - Neutral economies sign new RTAs regionally:
    - Regional integration offsets some losses; for most economies in Emerging and Developing Asia, Middle East and Central Asia, and Sub-Saharan Africa avoided losses are < 0.5 percentage points.
    - Global distribution of losses hardly altered.
    - Implication: need more ambitious integration agreements than average RTA to materially offset fallout.
  - Neutral economies join major blocs opportunistically:
    - Moderates losses by 0.8 percentage points on average for neutral economies.
    - Gains from one bloc insufficient to offset overall losses when sensitivity of trade barriers increases.

### Policy implications (summary)
- Avoiding or limiting fragmentation is economically and distributionally important: higher trade barriers imply efficiency losses and fall disproportionately on smaller and poorer economies.
- Non-aligned economies may limit losses by seeking closer regional integration with non-aligned partners, but such efforts must be wide-reaching and ambitious to materially offset likely income losses.
- If geoeconomic fragmentation cannot be avoided, EMDEs may need to brace for a decade in which global trade trends act as a headwind rather than the tailwind observed in recent decades.

### Appendix highlights — robustness, calibration, decomposition, and alternative counterfactuals
- Robustness:
  - Year-by-year estimates (1993–2018) retain expected negative sign and sector ranking stable.
  - Sub-sample estimates (AE vs EMDE exporters/importers) preserve sector ranking; effects larger in samples with at least one advanced economy.
  - Alternative elasticities: notable difference for transport equipment (CP 2015 θ = 0.69; FGO 22 θ = 3.27) but headline findings robust.
- Calibration:
  - Baseline data: EORA 2017–19 averages for {v_sn'n0} and {h_n0}; capital stocks per worker for 2018 from PWT ({k_n0}).
  - Core calibration uses system (31)–(34) and assumes U.S. near steady state in 2018; capital share α = .406.
- Decomposition of trade-barrier changes:
  - Average rise in import barriers for country n:
    - ln ˆτ_n = − Σ_{n'≠n} Σ_s ω_sn'n [ ˜β_align_s/θ_s h(ˆβ−1) u_align_n'n + ˆβ △u_align_n'n ].
  - Three driving factors for country n’s import-barrier increase:
    1. Exposure of import basket to sensitive sectors: Σ_s Σ_{n'≠n} ω_sn'n ˜β_align_s/θ_s.
    2. Initial alignment with partners: Σ_s Σ_{n'≠n} ω_sn'n u_align_n'n.
    3. Change in alignment with partners: Σ_s Σ_{n'≠n} ω_sn'n △u_align_n'n.
- Alternative counterfactuals:
  - A.4.1: incorporate indirect effects of alignment through WTO, RTA, EU participation (˜β_align,T2_s).
  - A.4.2: neutral economies sign new RTAs regionally (△u_RTA_n'n = 1 for specified regional-neutral pairs).
  - A.4.3: neutral economies join major blocs based on trade-share rule (18 join Western; 1 joins Eastern).
- Sector input shares and trade elasticities (calibration highlights):
  - Sector 1 Agriculture and fishing: μ = .4348 ; CP 2015 θ = 1.14 ; FGO 22 θ = 4.78
  - Sector 2 Mining and quarrying: μ = .40515 ; CP 2015 θ = 7.21 ; FGO 22 θ = 13.97
  - Sector 3 Food and beverages: μ = .6972 ; CP 2015 θ = 2.55 ; FGO 22 θ = 4.16
  - Sector 4 Textiles and wearing apparel: μ = .6915 ; CP 2015 θ = 5.56 ; FGO 22 θ = 4.83
  - Sector 5 Wood and paper: μ = .6679 ; CP 2015 θ = 9.95 ; FGO 22 θ = 5.01
  - Sector 6 Petroleum, chemical and non-metallic mineral products: μ = .71615 ; CP 2015 θ = 6.05 ; FGO 22 θ = 5.01
  - Sector 7 Metal products: μ = .7386 ; CP 2015 θ = 1.15 ; FGO 22 θ = 7.03
  - Sector 8 Electrical and machinery: μ = .6858 ; CP 2015 θ = 8.19 ; FGO 22 θ = 6.80
  - Sector 9 Transport equipment: μ = .743.69 ; CP 2015 θ = 0.69 ; FGO 22 θ = 3.27
  - Sector 10 Other manufacturing: μ = .6675 ; CP 2015 θ = 5.00 ; FGO 22 θ = 4.64
  - Sector 11 Services: μ = .4325 ; CP 2015 θ = 5.00 ; FGO 22 θ = 5.00

*Source: wpiea2023270-print-pdf — Divided We Fall: Differential Exposure to Geopolitical Fragmentation in Trade — Working Paper No. WP/23/270.*

### 8.5 percent when accounting for all three layers of fragmentation. Goes and Bekkers (2022)

### wpiea2023270-print-pdf - 8.5 percent when accounting for all three layers of fragmentation. Goes and Bekkers (2022)

### Literature overview and context
- Prior studies estimate welfare and output losses from geoeconomic fragmentation across a wide range:
  - Losses of 0.4 percent of GDP (some countries) to 12 percent (most affected) under scenarios from mild fragmentation to full technological decoupling.
  - IMF (2022): ~1.2 percent of global GDP loss from eliminating trade in high-tech and energy between rival blocs; increases to 1.5 percent when barriers are extended to other sectors.
  - Felbermayr et al. (2022): welfare losses between 0 and 10 percent across countries under East-West trade “decoupling” scenarios that double existing NTBs.
  - Bolhuis et al. (2023): long-run output reduced by between 0.2 and 6.9 percent depending on scenario and elasticities.
  - Javorcik et al. (2022) and Attinasi et al. (2023): global output losses ranging from 0.1 to 5 percent using supply-chain decoupling frameworks.
- Empirical evidence shows geopolitical alignment shapes bilateral FDI patterns and its effect strengthened since 2018 (Aiyar et al., 2023).

### Contribution and approach of this paper
- Provides a framework to discipline fragmentation counterfactuals across a large set of sectoral bilateral trade relationships by:
  - Using estimated sensitivity of trade barriers to geopolitical alignment across sectors from recent data.
  - Building scenarios that extrapolate from these empirical elasticities.
- Purpose:
  - Identify countries’ possible macroeconomic exposure to geoeconomic fragmentation ex ante.
  - Describe country characteristics that explain differences in exposure.
  - Spotlight impacts on economies outside major world markets.
- Novelty: First to quantify impact of geopolitical alignment on trade at a sectoral level and to integrate these estimates into standard quantitative trade models for counterfactual analysis.

### Structural gravity framework (methodology)
- Sector-level bilateral trade represented by structural gravity equation:
  - M_sn′n = [τ_sn′n O_sn′ P_sn]^{−θ_s} D_sn′ E_sn
  - P_sn and O_sn′ are inward and outward multilateral resistance terms as defined in the text.
- Two sufficient conditions to obtain the structural gravity form:
  1. Shares of spending v_sn′n are multiplicatively separable: v_sn′n = F_sn′ [τ_sn′n P_sn]^{−θ_s}, with P_sn defined accordingly.
  2. Market clearing for each origin country: D_sn′ = Σ_n M_sn′n = F_sn′ Σ_n [τ_sn′n P_sn]^{−θ_s} E_sn.
- Assumes ratio of external to internal trade barriers τ_sn′n / τ_snn is log-linear in country-pair characteristics:
  - ln(τ_sn′n / τ_snn) = β_0s + Σ_{l=1}^L β_ls u_l_{n′n}.

### Estimation strategy
- Two-step estimation:
  1. PPML estimate of full N×N bilateral expenditures including domestic expenditures:
     - M_sn′n = exp{Ω_sn′ + Π_sn + δ_sn′n} ζ_sn′n
     - Country-sector exporter fixed effect Ω_sn′, country-sector importer fixed effect Π_sn, undirected country-pair-sector fixed effect δ_sn′n.
     - From PPML, country-pair fixed effects map to −θ_s ln(τ_sn′n / τ_snn) = ˜δ_sn′n.
  2. OLS regression of estimated country-pair fixed effects on country-pair observables:
     - −˜δ_sn′n / θ_s = β_0s + Σ_{l=1}^L β_ls u_l_{n′n} + ε_sn′n
     - u_l includes: dist (log distance km), contig, lang, col, wto, rta, eu, and align (geopolitical alignment).
- Rationale: two-step approach accommodates missing country-pair characteristics (notably foreign-policy-alignment) while preserving theory-consistent identification of bilateral trade barriers.

### Data
- Trade flows:
  - Sector-level bilateral expenditure flows from the EORA global input-output database (Lenzen et al., 2013).
  - Uses simple average of 2017–19 EORA flows to avoid Covid-19 and Ukraine-war disruptions.
  - Aggregated to 10 goods sectors + 1 services sector; empirical analysis and counterfactuals restricted to goods sectors.
  - Dataset covers full matrix of sector-level bilateral trade flows between 185 economies (including “rest of the world”).
- Gravity controls:
  - Taken from the CEPII gravity dataset (Conte et al., 2022).
  - Sectoral trade elasticities: baseline from Caliendo and Parro (2015); robustness from Fontagné et al. (2022).
  - Elasticities vary from 0.69 in transport equipment to 15.72 in mining and quarrying.
- Bilateral geopolitical alignment:
  - Measured using ATOP-based alliance portfolio similarity (Signorino and Ritter, 1999) computed for 2018.
  - ATOP treaty coding treaty_{n′n} ∈ {0,1,2,3}: 3 = defense/offense; 2 = neutrality/consultation; 1 = nonaggression; 0 = no alliance obligation.
  - u_align_{n′n} bounded between −1 and 1; higher = greater foreign policy similarity.
  - 2018 values in practice range from -0.29 to 1; median 0.54; standard deviation 0.23.
  - Example: Germany–France alignment score 0.85; Germany–Angola alignment score 0.21.
- Sample size for regressions: 14,535 observations covering 185 countries.

### Key empirical estimates (impact of foreign policy alignment on bilateral trade barriers by sector)
- Table 1 reports coefficient estimates from equation (9) across 10 goods sectors. Selected results (Panel A: full controls):
  - Foreign policy alignment (scale −1,1) coefficients by sector:
    - 1 – Agriculture and fishing: −0.014 (standard error 0.010) [not significant]
    - 2 – Mining and Quarrying: 0.023 (0.010)** 
    - 3 – Food & Beverages: −0.249 (0.034)***
    - 4 – Textiles & Apparel: −0.148 (0.015)***
    - 5 – Wood & Paper: −0.073 (0.008)***
    - 6 – Petroleum, Chemical and Non-Metallic Mineral Products: −0.024 (0.006)***
    - 7 – Metal Products: −0.098 (0.014)***
    - 8 – Electrical Machinery: −0.025 (0.011)** 
    - 9 – Transport Equipment: −0.340 (0.124)***
    - 10 – Other: −0.192 (0.016)***
  - R2 values across sectors range from 0.04 to 0.22.
- Panel B (excluding controls for economic agreements) shows generally larger negative coefficients on foreign policy alignment:
  - Foreign policy alignment (−1,1) coefficients by sector (Panel B):
    - 1: −0.028 (0.010)*** 
    - 2: 0.014 (0.009) [not significant]
    - 3: −0.311 (0.035)***
    - 4: −0.193 (0.015)***
    - 5: −0.094 (0.008)***
    - 6: −0.045 (0.006)***
    - 7: −0.147 (0.015)***
    - 8: −0.066 (0.011)***
    - 9: −0.974 (0.130)***
    - 10: −0.236 (0.016)***
- Interpretation:
  - Greater foreign policy alignment is associated with lower bilateral trade barriers in most goods sectors, with statistically significant negative effects in Food & Beverages, Textiles & Apparel, Wood & Paper, Petroleum/Chemical/Non-Metallic, Metal Products, Transport Equipment, and Other.
  - The magnitude and statistical significance vary by sector and by inclusion of controls for economic agreements.

### Model integration and counterfactuals (overview)
- The paper integrates the sectoral empirical estimates of the sensitivity of trade barriers to geopolitical alignment into a dynamic many-country, many-sector quantitative trade model (a special case of Cuñat and Zymek (2023) and compatible with general equilibrium trade modeling literature).
- Counterfactuals use the estimates from the gravity analysis to explore long-run macroeconomic impacts of fragmentation scenarios and to discuss distributional outcomes across countries.
- The distribution of losses across countries is highly contingent on scenario assumptions about the size and incidence of new, fragmentation-induced trade barriers.

### Paper structure (as stated)
- Section 2: structural gravity framework and estimates of impact of geopolitical alignment on trade.
- Section 3: dynamic many-country, many-sector quantitative trade model for trade-cost counterfactuals.
- Section 4: counterfactuals introducing Section 2 estimates into the model to explore long-run macroeconomic impacts across fragmentation scenarios.
- Section 5: concluding remarks.

*Source: wpiea2023270-print-pdf - 8.5 percent when accounting for all three layers of fragmentation. Goes and Bekkers (2022).*

### 2.3  Results

### 2.3 Results

### Baseline Results
- Estimation approach:
  - Second-step estimates obtained by estimating equation (9) with OLS after purging bilateral expenditure flows of origin and destination fixed effects in the first step (Section 2.1.2).
  - Expectation: countries more aligned in foreign policy have lower trade barriers (β_align_s > 0), conditional on gravity controls.
- Main empirical findings (Table 1, Panel A):
  - Results reported across 10 tradable goods sectors with a full set of bilateral control variables.
  - Closer foreign policy alignment is associated with lower trade barriers in 9 out of the 10 sectors analyzed.
  - In 8 sectors the negative effect of alignment on trade barriers is statistically significant.
  - Standard gravity controls behave as expected: greater bilateral distance increases trade frictions in all sectors; common border, common language, and membership in trade agreements decrease bilateral trade barriers.
  - Coefficients on foreign policy alignment from Table 1 (Panel A) are used in counterfactual exercises in Section 4.
- Magnitude and sectoral heterogeneity (Figure 2):
  - A one standard deviation increase in foreign policy alignment equals 0.233 points.
  - After accounting for trade elasticities, the sectoral sensitivity ranking:
    - "Transport equipment" is the most sensitive: a one standard deviation increase in foreign policy alignment decreases trade barriers in the sector by almost 8 percent.
    - "Food and beverages" and "other manufacturing" are the second- and third-most sensitive sectors, respectively.
- Variance decomposition of bilateral trade barriers (Figure 3):
  - Share of total explained variance of estimated bilateral trade barriers by sector ranges from 11 percent ("Mining and quarrying") to 21 percent ("electric machinery").
  - Explained variation is dominated by geographic factors and economic agreements.
  - Differences in foreign policy alignments currently play a minor role, roughly on par with cultural variables (language and colonial history).

### Robustness Checks
- Effect persistence and sample checks:
  - Estimated elasticities of sectoral trade barriers with respect to foreign policy alignment are fairly stable across time, supporting the interpretation as inherent, unchanging sector characteristics.
  - Findings are replicated across different sub-samples of country pairs, addressing concerns that estimates are driven by particular bilateral relationships.
- Alternative elasticity scaling:
  - First-stage regression output also scaled using an alternative set of trade elasticities from Fontagné et al. (2022).
  - This yields a broadly similar ranking of sectors in sensitivity to foreign policy alignment, with a notable difference for one sector.
  - Baseline geoeconomic fragmentation scenarios in Section 4 are performed with both sets of elasticities; sector-level differences do not materially impact headline macro outcomes.
- Mechanism via economic/trade agreements (Table 1, Panel B):
  - Excluding controls for economic agreements (u_wto, u_rta, u_eu) increases the estimated effect of foreign policy alignment on trade barriers, with coefficients almost doubling in some cases.
  - Interpretation: geopolitical alignment also operates through countries’ propensity to enter economic or trade agreements; baseline estimates may understate total effect.

### Model (overview of structure and equilibrium)
- Model class and purpose:
  - A special case of the dynamic trade model in Cuñat and Zymek (2023).
  - Economies differ in reliance on and productivity in multiple sectors; sectoral inputs are Armington-differentiated.
  - Forward-looking agents with constant probability of death (Blanchard 1985) yield a steady state independent of initial conditions, enabling steady-state counterfactuals via exact-hat algebra.
- Key assumptions and components:
  - Agents:
    - Unit mass of agents per economy n = 1,...,N.
    - Constant probability of death each period: ξ.
    - Net population growth zero due to exogenous births of mass ξ.
    - Time preference: discount rate ρ.
    - Human capital endowment: H_nt, grows at gross rate γ (H_nt+1 = γ H_nt).
    - Period utility is logarithmic in final consumption.
    - Life-cycle budget constraint and optimal-savings problem given by equations (10)–(11).
  - Technologies:
    - Non-traded aggregate "all-purpose" good: X_nt = ∏_{s=1}^S (X_snt^{σ_sn})^{σ_sn} (equation (12)), with σ_sn ∈ (0,1) and ∑_s σ_sn = 1.
    - Sector-s input assembled from tradable, place-specific varieties per equation (13) with trade elasticity parameter θ_s ≥ 0.
    - Production of economy-n variety in sector s: Cobb-Douglas form Q_snt = z_sn [ K_{snt}^{α} H_{snt}^{1−α} ]^{1−μ_s} [ J_{snt}^{μ_s} ]^{μ_s} (equation (14)), with α, μ_s ∈ (0,1).
    - Investment efficiency parameter η_n relates final good to investment (X_nt = C_nt + η_n I_nt + ∑_s J_snt); capital accumulation follows K_{nt+1} = I_nt + (1−δ) K_nt (equation (15)).
  - Market structure:
    - Perfectly competitive markets.
    - Iceberg transport costs τ_{sn' n} ≥ 1.
    - International trade in a riskless bond with nominal return R_t; cohort wealth defined as A_{nt}(t′) ≡ η_n P_{nt−1} K_{nt}(t′) + B_{nt}(t′).
- Equilibrium conditions:
  - Prices: final consumption, intermediates, and investment price equal P_nt defined by equation (16) with p_snt defined in equation (17).
  - Optimal consumption/investment Euler equation (18) and no-arbitrage condition between capital and bond (20).
  - Market clearing for sectoral imports M_{sn' n t} as in equation (21) and resource constraints as in (22)–(23).
  - Sectoral trade flows obey a gravity representation implied by model conditions.
- Steady-state and exact-hat algebra:
  - Unique stable steady state with constant prices and real variables growing at rate γ.
  - Comparison of steady states under alternative trade-barrier configurations performed via exact-hat algebra (equations (24)–(28)).
  - Counterfactual change in bilateral trade barriers parameterized as (equation (29)):
    - ˆτ_{sn' n} = exp[ −1/θ_s ( ˆβ_align_s ˆu_align_{n' n} − 1 ) ̃β_align_s u_align_{n' n} ] if n' ≠ n; = 1 otherwise.
- Calibration inputs:
  - Two types of inputs required:
    1. Sectoral parameters: {σ_sn}, {μ_s, θ_s}, and capital share α.
    2. Status-quo steady-state sectoral bilateral trade shares {v_{s n' n}} and world GDP shares {h_n}.
  - Parameter sourcing:
    - Sectoral expenditure shares by country from EORA for 2017–19.
    - Sectoral input shares computed from global spending on inputs in each sector from EORA divided by global output.
    - Sectoral trade elasticities aggregated from Caliendo and Parro (2015); resulting sectoral parameters reported in Appendix Table A3.
    - Capital share set to α = .406 (U.S. capital share for 2018 from PWT, edition 10.0).
  - Status-quo steady-state shares constructed from observed EORA trade and GDP shares and PWT capital stocks (Appendix A.2).

### Counterfactuals: Scenarios
- General approach:
  - Counterfactuals change bilateral trade barriers via: (i) changes in sensitivity of trade barriers to geopolitical alignment (ˆβ_align_s); (ii) changes in countries' geopolitical alignments (ˆu_align_{n' n}); or (iii) both.
  - Other determinants of trade barriers are held constant in most analyses.
- Scenario 1 — Geopolitical Polarization (Section 4.1.1):
  - Definition of blocs:
    - Measure of treaty strength: treaty_{n' n} ∈ {0,1,2,3} from ATOP; relative strength toward U.S. vs China defined as treaty_{n USA} − treaty_{n CHN}, ranging from −3 to 3.
    - Three blocs:
      - Western bloc: treaty_{n USA} > treaty_{n CHN}.
      - Eastern bloc: treaty_{n CHN} > treaty_{n USA}.
      - Non-aligned bloc: treaty_{n USA} = treaty_{n CHN}.
  - Polarization implementation:
    - Members of Western and Eastern blocs increase treaty strength with economies in their bloc by 1 (capped at 3) and decrease treaty strength with all other economies by 1 (floored at 0).
    - Bilateral treaties of non-aligned bloc countries are unchanged by assumption, but their bilateral alignment relative to other blocs changes mechanically due to others’ treaty changes.
  - Expected trade-barrier effects:
    - Within-bloc alignments increase → lower trade barriers within blocs.
    - Between-bloc alignments decrease → higher trade barriers across blocs.
- Scenario 2 — Increased Geopolitical Sensitivity of Trade Barriers (Section 4.1.2):
  - Assume ˆβ_align_s = 2 for all s (doubling the sensitivity of trade barriers to foreign policy alignment).
  - Rationale: doubling implies doubling the share of variation in trade barriers attributable to foreign policy alignment, moving it ahead of cultural factors but still behind geography and economic agreements in explanatory power.
  - Implication by construction (equation (29)): relative increase in trade barriers for sectors with high pre-existing sensitivity to alignment and for country pairs with relatively low bilateral alignment.
- Scenario 3 — Combined Polarization and Increased Sensitivity (Section 4.1.3):
  - Combines scenarios 1 and 2 and serves as the baseline geoeconomic fragmentation scenario.
  - Captures both more clearly delineated geoeconomic blocs and stronger translation of geopolitical divides into effective trade barriers.

_Italic: Content summarized from "2.3 Results" in the source PDF._

### 4.2  Impact of Fragmentation on Real Incomes

### 4.2  Impact of Fragmentation on Real Incomes

### Scenario outcomes and headline magnitudes
- Scenario 1 (“polarization only”):
  - Median economy experiences a modest real-income loss of about 0.2 percent.
  - About one quarter of economies experience real-income gains.
  - Latin American and Caribbean economies are among the biggest winners in this scenario due to significant declines in trade barriers with the U.S.

- Scenario 2 (“increased geopolitical sensitivity of trade”):
  - Median economy experiences a real-income loss of about 1 percent.
  - Impacts are relatively uniformly distributed across regions:
    - Smallest median income loss: emerging and developing Asia, 0.7 percent.
    - Largest median income loss: Middle East and Central Asia, 1.5 percent.

- Scenario 3 (combined scenario, baseline):
  - Delivers the largest median loss across the world as a whole: 1.3 percent.
  - Largest variation in real-income losses across economies.
  - Median impacts by group:
    - Advanced economies: about 0.9 percent.
    - Emerging and developing Asia: median impact is 80 percent larger than for advanced economies.
    - Sub-Saharan Africa: median impact is more than 120 percent larger than for advanced economies.
    - Middle East and Central Asia: median impact is 150 percent larger than for advanced economies.
  - About one quarter of economies in sub-Saharan Africa and the Middle East and Central Asia experience real-income losses in excess of 3 percent.

### Drivers of differential exposure across countries
- Three determinants of a country’s exposure to geoeconomic fragmentation:
  1. Market size: smaller economies (by share of world GDP) suffer larger real-income losses even under a uniform increase in trade barriers, because smaller economies rely more on international trade.
  2. Comparative advantage / import basket: economies that import more in sectors whose trade barriers are especially sensitive to geoeconomic alignment see larger increases in the cost of imported goods.
  3. Foreign policy alignments (and changes): trade barriers rise more for economies that are (or become) more geopolitically “distant” from major trading partners.

- Method: decomposition of trade-barrier changes as in equation (30), separating:
  - ln ˆτ (weighted average change across all sectors and country pairs),
  - ln ˆτβs (sector sensitivity component),
  - ln ˆτu_sn′n (bilateral alignment level component),
  - ln ˆτΔu_sn′n (alignment change component),
  - η_sn′n (residual uncorrelated to first order).

### Quantitative role of determinants (variance comparison)
- Partial counterfactuals show relative importance of components in generating cross-country variance of real-income changes. Table 2 (variance comparison) results:
  - All (baseline) — Cov[x,ˆy_n({ˆτ_sn′n}_sn′n)] / V ar[ˆy_n({ˆτ_sn′n}_sn′n)]
    - Allˆy_n({ˆτ_sn′n}_sn′n): 2.467 1.000
  - Size
    - ˆy_n({ˆτ}_sn′n): 1.264 0.512
  - Import basket
    - ˆy_n(ˆτβs_sn′n): 0.238 0.097
  - Alignments
    - ˆy_n({ˆτu_sn′n}_sn′n): 0.541 0.219
  - Alignment changes
    - ˆy_n(nˆτΔu_sn′n o_sn′n): 0.332 0.135

- Interpretations:
  - Introducing the same (average) trade-barrier change across all sectors and country pairs generates more than half of the variation in real-income changes from the baseline scenario — economy size is the primary reason for differential exposure.
  - Differences in countries’ alignments and alignment changes together generate another one third of the baseline variation.
  - Differences in import baskets account for less than one tenth of the baseline income-change variance.

### Additional and alternative counterfactuals — implications for mitigation
- Economic agreements determined by foreign policy alignment (endogenous RTAs):
  - If membership in economic agreements is fully determined by foreign policy alignment, median losses for regions increase by between 60 percent and 100 percent relative to the baseline — implying the baseline is conservative by assuming existing economic-agreement patterns remain unchanged.

- Alternative trade elasticities (Fontagné et al. (2022) vs. Caliendo and Parro (2015)):
  - Using elasticities from Fontagné et al. (2022) somewhat reduces the magnitude of income losses across countries.
  - Reason: larger elasticities in sectors like "Food" and "Transport Equipment" imply smaller welfare losses from trade-barrier increases in standard gravity trade models.
  - Nevertheless, real-income losses remain sizeable and their distribution across country groups remains broadly unchanged.

- “Neutral” economies respond to fragmentation:
  - Scenario: “Neutral” economies sign new RTAs with other “neutral” economies in their region:
    - Regional trade integration can offset some losses, but for most economies in Emerging and Developing Asia, Middle East and Central Asia, and Sub-Saharan Africa the income losses avoided are smaller than 0.5 percentage points.
    - Global distribution of real-income losses is hardly altered compared with the baseline.
    - Reason: presence of existing RTAs and the estimated modest barrier-reducing effect of the average RTA.
    - Implication: to substantially limit fallout, countries would need more ambitious integration agreements than the average RTA.

  - Scenario: “Neutral” economies join major blocs opportunistically (join the bloc with which they enjoy strongest pre-existing trade ties):
    - Geopolitical realignment moderates losses compared to the baseline by 0.8 percentage points on average for neutral economies.
    - However, gains from lower trade barriers with one bloc generally are insufficient to offset overall losses from increased sensitivity of trade barriers combined with increased distances; the least aligned economies still experience significant welfare losses.

### Policy implications (summary)
- Avoiding or limiting fragmentation of the global trade landscape is economically and distributionally important: higher trade barriers imply efficiency losses and fall disproportionately on smaller and poorer economies.
- Non-aligned economies may limit losses by seeking closer integration with non-aligned partners regionally, but such efforts must be wide-reaching and ambitious to materially offset likely income losses.
- If geoeconomic fragmentation cannot be avoided, EMDEs may need to brace for a decade in which global trade trends act as a headwind rather than the tailwind observed in recent decades.

*Source: wpiea2023270-print-pdf — https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023270-print-pdf.pdf*

### References

### Appendix

### A.1 Robustness Checks

- A.1.1 Different Time Periods
  - Estimations of the effect of foreign policy alignment on bilateral trade barriers for different years between 1993 and 2018 mostly retain their expected negative sign.
  - The ranking of sectors in terms of responsiveness of trade barriers to foreign policy alignment is stable over time.
  - Implication: sensitivity of trade barriers to geopolitics is an inherent sector characteristic that varies little over decades; supports assumption used in geoeconomic fragmentation scenarios in Section 4 that a sector’s place in the ranking of trade sensitivity remains unchanged even as geopolitics becomes more important.

- A.1.2 Different Country-Pair Samples
  - Re-estimation for sub-samples where exporting or importing countries belong only to advanced economies (AE) or EMDEs yields negative estimated effects of foreign policy alignment on bilateral trade barriers across most sectors.
  - The ranking of sectors by trade sensitivity to foreign policy alignment is broadly preserved across sub-samples.
  - Estimated effects are larger in absolute terms when the sample is restricted to country pairs in which at least one partner is an advanced economy; this is attributed in part to better data quality for trade flows involving advanced economies.
  - Sample size and coverage note: Every regression has 14,535 observations covering 185 countries.

- A.1.3 Different Trade Elasticities
  - Baseline conversion of bilateral fixed-effect estimates into trade-barrier-equivalent values uses trade elasticities from Caliendo and Parro (2015) (CP 2015).
  - Alternative conversion uses elasticities from Fontagné et al. (2022) (FGO 22).
  - Elasticities are comparable for most sectors; notable exception:
    - Transport equipment elasticity: CP 2015 = 0.69; FGO 22 = 3.27.
  - Result: Except for transport equipment, effects of foreign policy alignment on trade barriers are similar regardless of elasticity set.
  - Given importance of transport equipment (large share of trade flows and sensitivity under baseline elasticities), main geoeconomic fragmentation scenarios in Section 4 are run with both elasticity sets; choice of elasticities does not materially affect headline findings.

### A.2 Calibration

- Data and baseline periods
  - Average sectoral bilateral trade shares and world GDP shares taken from EORA for the period 2017–19, denoted {v_sn'n0} and {h_n0}.
  - Capital stocks per worker for 2018 taken from PWT, denoted {k_n0}.
  - 2018 treated as “period 0” and 2019 as “period 1” when computing certain steady-state ratios.
- Core calibration approach
  - System of equations (31)–(34) relates changes in sectoral bilateral trade shares and factor price-related terms to known observed data, with the only unknown being {[R−(1−δ)]/R0 −(1−δ) P_n1/P_n0}_n.
  - Assumption: U.S. economy is close to steady state as of 2018, allowing use of equation (20) to compute the needed ratio via observable PWT series:
    - R−(1−δ)/R0 −(1−δ) P_n1/P_n0 = P_USA,1 Y_USA,1 / (P_I_USA,0 K_USA,1) × P_n1 Y_n1 /(P_I_n0 K_n1).
  - These right-hand-side elements computed from real GDP, capital stock and investment price data from PWT.

### A.3 Decomposition of Trade-Barrier Changes

- Baseline formulation
  - Baseline counterfactual trade-cost changes have two components:
    - A uniform increase in sensitivity of trade barriers to foreign policy alignment across sectors, ˜β_align_s = ˆβ for all s.
    - Changes in bilateral alignments, △u_align_n'n ≡ [ˆu_align_n'n − 1] u_align_n'n.
  - Equation (36) expresses ln ˆτ_sn'n as a function of these components.
- Country-average import-barrier change
  - Average rise in import barriers faced by country n:
    - ln ˆτ_n = − Σ_{n'≠n} Σ_s ω_sn'n [ ˜β_align_s/θ_s h(ˆβ−1) u_align_n'n + ˆβ △u_align_n'n ].
  - Where ω_sn'n ≡ M_sn'n / Σ_s Σ_{n'≠n} M_sn'n ; M_sn'n is the 2017–19 average import value by country n from n' in sectors.
- Three driving factors behind increase in import barriers for country n
  1. Exposure of country n’s import basket to sectors sensitive to geopolitical alignment, represented by Σ_s Σ_{n'≠n} ω_sn'n ˜β_align_s/θ_s.
  2. Initial alignment of country n with its trade partners, represented by Σ_s Σ_{n'≠n} ω_sn'n u_align_n'n.
  3. Change in alignment of country n vis‑à‑vis its trade partners, represented by Σ_s Σ_{n'≠n} ω_sn'n △u_align_n'n.
- Aggregation and decomposition definitions
  - Define ω_n, ln ˆτ (average change across countries), β/θ (alignment sensitivity of average country’s import basket), u (average bilateral foreign policy alignment), and △u (average change in bilateral foreign policy alignment) as in equations (38)–(41).
  - Decomposition for each bilateral-sector ln ˆτ_sn'n:
    - ln ˆτ_sn'n = ln ˆτ + ln ˆτ^β_s + ln ˆτ^u_n'n + ln ˆτ^{△u}_n'n + η_sn'n, with ln ˆτ^β_s, ln ˆτ^u_n'n, ln ˆτ^{△u}_n'n defined in equations (43)–(45).
  - Interpretation:
    - ln ˆτ^β_s: relative effect due to sector sensitivity difference from average sector.
    - ln ˆτ^u_n'n: relative effect due to bilateral alignment deviation from world average.
    - ln ˆτ^{△u}_n'n: relative effect due to bilateral alignment change deviation from world average.
  - By construction, the weighted average of the decomposition terms and residual η is zero.

### A.4 Additional and Alternative Counterfactuals

- A.4.1 Economic Agreements Based on Geopolitical Alignment
  - Finding in Section 2.3.2: estimated responsiveness ˜β_align_s generally rises if economic agreement controls (e.g., WTO, RTA, EU participation) are omitted.
  - Interpretation: foreign policy alignment may directly reduce barriers and indirectly promote joint participation in economic agreements that further reduce barriers.
  - Formal relationships posited (equations 46–48):
    - u_WTO_n'n = υ_WTO + λ_WTO u_align_n'n + ε_WTO_n'n.
    - u_RTA_n'n = υ_RTA + λ_RTA u_align_n'n + ε_RTA_n'n.
    - u_EU_n'n = υ_EU + λ_EU u_align_n'n + ε_EU_n'n.
  - Link between estimates (equation 49): ˜β_align,T2_s = ˜β_align,T1_s + ˜β_WTO,T1_s ˜λ_WTO + ˜β_RTA,T1_s ˜λ_RTA + ˜β_EU,T1_s ˜λ_EU.
  - Counterfactual: repeat baseline fragmentation experiment using {˜β_align,T2_s}_s (which encapsulate direct and indirect effects via economic agreements) and interpret resulting income effects accordingly.

- A.4.2 “Neutral” Economies Sign New RTAs
  - Counterfactual modification to equation (29) to include RTA dummy changes (equation 50), with △u_RTA_n'n representing changes in RTA dummy.
  - First counterfactual explored in Section 4.4.3: set △u_RTA_n'n = 1 if:
    - (i) economies n' and n both belong to the “neutral” bloc of countries; and
    - (ii) economies n' and n both belong to either Emerging and Developing Asia, or Middle East and Central Asia, or Africa regions.
    - Otherwise △u_RTA_n'n = 0.
  - Assumption: trade-promoting impact of these newly signed RTAs equals the average RTA in sample, given by estimates {˜β_RTA_s}_s from Section 2.3.1.

- A.4.3 “Neutral” Economies Join Major Blocs
  - Assignment rule: for each “neutral” economy compute trade share with Eastern and Western blocs (2017–19 trade); assign economy to a bloc if its trade share with that bloc is at least 10 percentage points larger than with the other bloc; if within 10 percentage points, remain “neutral”.
  - Outcome under this rule: 18 “neutral” economies join the Western bloc; 1 economy joins the Eastern bloc.
  - After assignment, perform the baseline fragmentation scenario as in Section 4.1.3.

### A.5 Sector Input Shares and Trade Elasticities (calibration table highlights)

- Sectoral input shares (μ_s) and two alternative trade-elasticity sets (θ_s) used in calibration (CP 2015 vs FGO 22):
  - Sector 1 Agriculture and fishing: input share .4348 ; CP 2015 θ = 1.14 ; FGO 22 θ = 4.78
  - Sector 2 Mining and quarrying: input share .40515 ; CP 2015 θ = 7.21 ; FGO 22 θ = 13.97
  - Sector 3 Food and beverages: input share .6972 ; CP 2015 θ = 2.55 ; FGO 22 θ = 4.16
  - Sector 4 Textiles and wearing apparel: input share .6915 ; CP 2015 θ = 5.56 ; FGO 22 θ = 4.83
  - Sector 5 Wood and paper: input share .6679 ; CP 2015 θ = 9.95 ; FGO 22 θ = 5.01
  - Sector 6 Petroleum, chemical and non-metallic mineral products: input share .71615 ; CP 2015 θ = 6.05 ; FGO 22 θ = 5.01
  - Sector 7 Metal products: input share .7386 ; CP 2015 θ = 1.15 ; FGO 22 θ = 7.03
  - Sector 8 Electrical and machinery: input share .6858 ; CP 2015 θ = 8.19 ; FGO 22 θ = 6.80
  - Sector 9 Transport equipment: input share .743.69 ; CP 2015 θ = 0.69 ; FGO 22 θ = 3.27
  - Sector 10 Other manufacturing: input share .6675 ; CP 2015 θ = 5.00 ; FGO 22 θ = 4.64
  - Sector 11 Services: input share .4325 ; CP 2015 θ = 5.00 ; FGO 22 θ = 5.00
- Note: Input shares computed from EORA for 2017–19; CP 2015 aggregated from Caliendo and Parro (2015) and Costinot and Rodríguez-Clare (2014); FGO 22 aggregated from Fontagné et al. (2022).

*Divided We Fall: Differential Exposure to Geopolitical Fragmentation in Trade — Working Paper No. WP/23/270*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023270-print-pdf.pdf_