## Executive Summary ## Source details **Canonical URL:** [Executive Summary](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023124-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023124-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023124-print-pdf.pdf.json) --- ### Key question and approach - Investigates how bilateral trade patterns are altered by rising trade policy uncertainty (TPU) and whether geopolitical alignments determine bilateral trade reactions. - Uses a structural gravity framework augmented with: - a text-based TPU index (World Trade Uncertainty Index, WTUI), - a geopolitical distance measure based on UN General Assembly voting records (ideal point distance, IPD). - Data cover 186 countries between 2002 and 2019. ### Principal empirical finding - Geopolitical distance by itself does not affect contemporaneous bilateral trade in normal times. - During periods of elevated TPU, geopolitical closeness matters: countries trade relatively more with geopolitically closer partners (“friends”). - Quantified effects (preferred specification: GDP weighted WTUI, Column (2)): - A one standard deviation hike in global trade policy uncertainty leads to: - approximately a 1.0 percent increase in bilateral trade between countries at the 25th percentile of geopolitical distance (friends) relative to the mean, - a 0.7 percent decrease in bilateral trade for countries at the 75th percentile of geopolitical distance (geopolitical rivals) relative to the mean, - a 3.1 percent decrease in bilateral trade for countries at the 99th percentile of geopolitical distance (close to maximum rivalry) relative to the mean. - Example magnitudes: - The 2019 shock to global trade policy uncertainty was 4.9 standard deviations and the geopolitical distance between China and the U.S. was 2.5, implying a predicted (symmetric) -12.7 percent decrease in bilateral trade relative to neutral country pairs. - Geopolitical distance between Germany and Spain was -1.2 yielding predicted relative increases of 6.8 percent for the same shock. - Summing absolute predicted relative changes across all country pairs yields a total rearrangement of $1.1 trillion or 5.1 percent of global trade along geopolitical alignments. ### Sectoral and product insights - Effects hold for agriculture, manufacturing, and energy sectors, but not for services (services are negatively affected by geopolitical distance even in times of low uncertainty). - Mining and energy trade is more elastic to uncertainty than agriculture and manufacturing (statistically significant difference via Wald test). - Using a list of “strategic” products from the April 2023 World Economic Outlook (IMF, 2023), suggestive evidence that geopolitical distance is a more important determinant of bilateral trade in high-uncertainty periods for these products; elasticity in strategic manufacturing sectors approximately 15% larger (difference not robustly statistically significant). - Granular manufacturing exercise: average elasticity estimate of -0.01004 for 27 strategic sectors versus -0.00836 for 91 non-strategic sectors (difference around 20%). Dispersion: standard deviation 0.01612 for strategic group versus 0.03895 for non-strategic group. ### Robustness - Excluding China and the United States: interaction coefficient remains significant but about half the size. - Results driven by post-2016 period of elevated TPU. - Including controls for RTAs, various non-tariff measures, and WTO disputes does not materially alter results. - Robust to origin/destination-specific policy uncertainty specifications and a semi-parametric quartile (bin) approach. ### Policy implication highlighted - Finding underscores the importance of the multilateral trade system and the World Trade Organization in providing institutional underpinning for a stable and predictable trade environment that can foster peaceful relations; low trade policy uncertainty may be required for trade to bring together countries with different geopolitical alignments. *IMF Working Paper — Executive Summary (wpiea2023124-print-pdf)* ### Empirical methodology and data - Estimation framework: - Structural gravity model (Anderson and Van Wincoop, 2003) estimated with Poisson pseudo-maximum likelihood (PPML) estimator. - Specification includes exporter-time (δit), importer-time (δjt), and exporter-importer (δij) fixed effects. - Core estimating equation (symbolic form preserved): - TradeFlowijt = exp(β0 IPDijt + β1 Uncertaintyt × IPDijt + δit + δjt + δij) × εijt - Data sources and coverage: - Trade flows: ITPD-E bilateral trade database (USITC); regression sample covers 186 countries/regions, 2002–2019. - Geopolitical distance (IPDijt): “ideal point distance” based on UN General Assembly voting data (Bailey, Strezhnev, Voeten, 2017). - Uncertainty measures (Uncertaintyt): textual frequency-based WTUI (Ahir, Bloom, Furceri, 2022); four standardized measures used (simple average WTUI, GDP-weighted WTUI, simple average WUI, GDP-weighted WUI). - Summary statistics (regression sample, N = 571,204 observations for listed variables): - TradeFlowijt: Mean 513.9; Std. Dev. 5,618; Min 0; Max 575,075. - IPDijt: Mean -0.0499; Std. Dev. 0.973; Min -1.265; Max 4.724. - WTUIt: Mean 0.0906; Std. Dev. 1.129; Min -0.406; Max 4.519. - Weighted WTUIt: Mean 0.0802; Std. Dev. 1.133; Min -0.291; Max 4.692. - WUIt: Mean 0.470; Std. Dev. 0.946; Min -0.581; Max 2.285. - Weighted WUIt: Mean 0.450; Std. Dev. 0.929; Min -0.979; Max 3.160. ### Results (benchmark, magnitudes, heterogeneity) - Benchmark results: - IPD alone: not a significant contemporaneous predictor of trade flows. - Interaction IPD × global uncertainty: significant and negative across specifications, stronger with WTUI measures specific to trade. - Preferred specification uses GDP-weighted WTUI. - Average Marginal Effects (AME) over 34,410 exporter-importer pairs: - 25th percentile IPD: +1.0 percent bilateral trade per one standard deviation rise in global TPU (relative to mean). - 75th percentile IPD: -0.7 percent bilateral trade per one standard deviation rise in global TPU (relative to mean). - 99th percentile IPD: -3.1 percent bilateral trade per one standard deviation rise in global TPU (relative to mean). - Aggregate rearrangement implied by a one standard deviation TPU shock: $1.1 trillion or 5.1 percent of global trade reallocated along geopolitical alignments. ### Robustness and sensitivity (detailed) - Excluding China and the United States: interaction term significant but roughly half the size. - Pre-/post-2016 split: results driven by post-2016 period of elevated trade uncertainty. - Controls for RTAs, Anti-Dumping (AD) actions, Countervailing Duties (CVD), SPS and TBT measures, Specific Trade Concerns (STCs), and WTO disputes do not materially alter the interaction result. - Origin- and/or destination-specific policy uncertainty measures or their product yield similar patterns for non-systemic countries. - Semi-parametric quartile (bin) approach supports significance of higher-order interactions (e.g., Q3 WTUI × Q2 IPD and Q4 WTUI × Q4 IPD negative coefficients). ### Sectoral results (detailed) - Mining and energy: more elastic to uncertainty than agriculture and manufacturing (Wald test significant). - Services: coefficient not significant; services negatively affected by geopolitical distance even at low uncertainty. - Strategic manufacturing: - Suggestive evidence that strategic manufacturing sectors are more affected by TPU-mediated geopolitical effects; elasticity roughly 15% larger (difference not robustly statistically significant). - Manufacturing split: strategic sectors average elasticity -0.01004; non-strategic average -0.00836; difference ~20%. ### Conclusion and policy recommendations - Interpretation: - Evidence supports a “trading with friends” response to elevated trade policy uncertainty: trade shifts toward geopolitically closer partners when TPU rises. - High TPU limits trade’s ability to bring geopolitically different countries closer; low TPU is conducive for trade to help align interests across geopolitical divides. - Policy recommendations: - Reduce trade policy uncertainty and trade tensions via: - Constructive multilateral engagement on thorny trade issues (e.g., subsidies and dispute settlement reform). - Renewed efforts to re-establish a fully functioning dispute settlement mechanism at the WTO. - Implications: - Systemic increases in TPU can materially reallocate trade ($1.1 trillion or 5.1 percent of global trade in the quantified shock), affecting global supply chains, strategic sectors, and sector-specific vulnerabilities (e.g., mining and energy, strategic manufacturing). - Suggestions for future research: - Build more precise measures of trade policy uncertainty that better capture the bilateral nature of this uncertainty. - Develop measures that allow for a more disaggregated look at impacts across trading partners and more precise predictions of expected changes to the global trade network. *IMF Working Paper — Trading with Friends in Uncertain Times (wpiea2023124-print-pdf)* ### Executive Summary ...................................................................................................... ### Executive Summary ### Key question and approach - Investigates how bilateral trade patterns are altered by rising trade policy uncertainty (TPU) and whether geopolitical alignments determine bilateral trade reactions. - Uses a structural gravity framework augmented with: - a text-based TPU index (World Trade Uncertainty Index, WTUI), - a geopolitical distance measure based on UN General Assembly voting records (ideal point distance, IPD). - Data cover 186 countries between 2002 and 2019. ### Principal empirical finding - Geopolitical distance by itself does not affect contemporaneous bilateral trade in normal times. - During periods of elevated TPU, geopolitical closeness matters: countries trade relatively more with geopolitically closer partners (“friends”). - Quantified effects (preferred specification: GDP weighted WTUI, Column (2)): - A one standard deviation hike in global trade policy uncertainty leads to: - approximately a 1.0 percent increase in bilateral trade between countries at the 25th percentile of geopolitical distance (friends) relative to the mean, - a 0.7 percent decrease in bilateral trade for countries at the 75th percentile of geopolitical distance (geopolitical rivals) relative to the mean, - a 3.1 percent decrease in bilateral trade for countries at the 99th percentile of geopolitical distance (close to maximum rivalry) relative to the mean. - Example magnitudes: - The 2019 shock to global trade policy uncertainty was 4.9 standard deviations and the geopolitical distance between China and the U.S. was 2.5, implying a predicted (symmetric) -12.7 percent decrease in bilateral trade relative to neutral country pairs. - Geopolitical distance between Germany and Spain was -1.2 yielding predicted relative increases of 6.8 percent for the same shock. - Summing absolute predicted relative changes across all country pairs yields a total rearrangement of $1.1 trillion or 5.1 percent of global trade along geopolitical alignments. ### Sectoral and product insights - Effects hold for agriculture, manufacturing, and energy sectors, but not for services (services are negatively affected by geopolitical distance even in times of low uncertainty). - Mining and energy trade is more elastic to uncertainty than agriculture and manufacturing (statistically significant difference via Wald test). - Using a list of “strategic” products from the April 2023 World Economic Outlook (IMF, 2023), suggestive evidence that geopolitical distance is a more important determinant of bilateral trade in high-uncertainty periods for these products; elasticity in strategic manufacturing sectors approximately 15% larger (difference not robustly statistically significant). - Granular manufacturing exercise: average elasticity estimate of -0.01004 for 27 strategic sectors versus -0.00836 for 91 non-strategic sectors (difference around 20%). Dispersion: standard deviation 0.01612 for strategic group versus 0.03895 for non-strategic group. ### Robustness - Excluding China and the United States: interaction coefficient remains significant but about half the size. - Results driven by post-2016 period of elevated TPU. - Including controls for RTAs, various non-tariff measures, and WTO disputes does not materially alter results. - Robust to origin/destination-specific policy uncertainty specifications and a semi-parametric quartile (bin) approach. ### Policy implication highlighted - Finding underscores the importance of the multilateral trade system and the World Trade Organization in providing institutional underpinning for a stable and predictable trade environment that can foster peaceful relations; low trade policy uncertainty may be required for trade to bring together countries with different geopolitical alignments. --- ### Introduction ### Context and motivation - Global TPU has reached new heights in recent years, driven by rising geopolitical tensions and “geo-economic fragmentation” (Aiyar et al., 2023). - TPU has been documented to negatively impact international trade (e.g., Constantinescu, Mattoo, Ruta, 2020); geopolitical alignment has been associated with lower trade barriers (Hakobyan, Meleshchuk, Zymek, 2023). - Hypothesis: heightened TPU increases the salience of geopolitical differences, prompting re-optimization of sourcing, FDI, and market entry decisions toward less risky, geopolitically closer partners. ### Conceptual mechanism - Without multilateral rules, differences in perceived risk across partners matter more; “friendshoring” and reallocation toward geopolitically close partners can occur via: - policy choices favoring close partners, - firm and household behavioral changes driven by risk aversion, expectations, or preferences. --- ### Empirical Methodology and Data ### Estimation framework - Structural gravity model (Anderson and Van Wincoop, 2003) estimated with Poisson pseudo-maximum likelihood (PPML) estimator. - Specification includes exporter-time (δit), importer-time (δjt), and exporter-importer (δij) fixed effects to control for multilateral resistance and time-invariant trade costs. - Regression focuses on time-varying trade costs: IPDijt and its interaction with global uncertainty Uncertaintyt. - Core estimating equation (symbolic form preserved): - TradeFlowijt = exp(β0 IPDijt + β1 Uncertaintyt × IPDijt + δit + δjt + δij) × εijt ### Data sources and coverage - Trade flows (TradeFlowijt): ITPD-E bilateral trade database (USITC), covering historical countries/regions for 1986–2019; regression sample covers 186 countries/regions, 2002–2019. - Geopolitical distance (IPDijt): “ideal point distance” based on UN General Assembly voting data (Bailey, Strezhnev, Voeten, 2017). - Uncertainty measures (Uncertaintyt): textual frequency-based WTUI (Ahir, Bloom, Furceri, 2022); four standardized measures used (simple average WTUI, GDP-weighted WTUI, simple average WUI, GDP-weighted WUI). - Summary statistics (regression sample, N = 571,204 observations for listed variables): - TradeFlowijt: Mean 513.9; Std. Dev. 5,618; Min 0; Max 575,075. - IPDijt: Mean -0.0499; Std. Dev. 0.973; Min -1.265; Max 4.724. - WTUIt: Mean 0.0906; Std. Dev. 1.129; Min -0.406; Max 4.519. - Weighted WTUIt: Mean 0.0802; Std. Dev. 1.133; Min -0.291; Max 4.692. - WUIt: Mean 0.470; Std. Dev. 0.946; Min -0.581; Max 2.285. - Weighted WUIt: Mean 0.450; Std. Dev. 0.929; Min -0.979; Max 3.160. --- ### Results ### Benchmark results - Geopolitical distance (IPD) alone: not a significant contemporaneous predictor of trade flows. - Interaction IPD × global uncertainty: significant and negative across specifications, stronger when using WTUI measures specific to trade than overall WUI. - Preferred specification uses GDP-weighted WTUI due to specificity to trade and larger economies’ influence. ### Economic magnitudes and heterogeneity - Average Marginal Effects (AME) calculated over 34,410 exporter-importer pairs: - 25th percentile IPD: +1.0 percent bilateral trade per one standard deviation rise in global TPU (relative to mean). - 75th percentile IPD: -0.7 percent bilateral trade per one standard deviation rise in global TPU (relative to mean). - 99th percentile IPD: -3.1 percent bilateral trade per one standard deviation rise in global TPU (relative to mean). - Aggregate rearrangement implied by a one standard deviation TPU shock: $1.1 trillion or 5.1 percent of global trade reallocated along geopolitical alignments (sum of absolute predicted relative changes). ### Robustness and sensitivity - Excluding China and the United States: interaction term significant but roughly half the size. - Pre-/post-2016 split: results driven by post-2016 period of elevated trade uncertainty. - Controls for RTAs, Anti-Dumping (AD) actions, Countervailing Duties (CVD), SPS and TBT measures, Specific Trade Concerns (STCs), and WTO disputes do not materially alter the interaction result. - Using origin- and/or destination-specific policy uncertainty measures or their product yields similar patterns for non-systemic countries. - Semi-parametric quartile (bin) approach supports significance of higher-order interactions (e.g., Q3 WTUI × Q2 IPD and Q4 WTUI × Q4 IPD negative coefficients). ### Sectoral results - Sectors estimated using preferred TPU measure (GDP-weighted WTUI) in sector-specific regressions: - Mining and energy: more elastic to uncertainty than agriculture and manufacturing (statistically significant via Wald test). - Services: coefficient not significant; services negatively affected by geopolitical distance even at low uncertainty (consistent with services requiring closer bilateral policy alignment). - Strategic manufacturing (merged with IMF April 2023 strategic sector indicator): - Trade in strategic manufacturing sectors appears more prone to TPU-mediated geopolitical effects; elasticity ~15% larger (difference not robustly statistically significant). - Detailed manufacturing split: strategic sectors average elasticity -0.01004; non-strategic average -0.00836; difference ~20%. --- ### Conclusion ### Interpretation - Evidence supports a “trading with friends” response to elevated trade policy uncertainty: countries and firms shift trade toward geopolitically closer partners when TPU rises. - Complementary to literature positing that trade increases friendship: this paper suggests high TPU limits trade’s ability to bring geopolitically different countries closer; low TPU is conducive for trade to help align interests across geopolitical divides. ### Policy relevance - Reinforces the role of stable, multilateral trade institutions—particularly the World Trade Organization—in maintaining predictable trade environments that reduce incentives to reorient trade along geopolitical lines and that may support peaceful relations. - Highlights that systemic increases in TPU can materially reallocate trade ($1.1 trillion or 5.1 percent of global trade in the quantified shock) across geopolitical alignments, with implications for global supply chains, strategic sectors, and sector-specific vulnerabilities (e.g., mining and energy, strategic manufacturing). *IMF Working Paper — Executive Summary (wpiea2023124-print-pdf)* ### Conclusion ### Conclusion ### Main findings - The paper examines how bilateral trade patterns are affected by rising trade policy uncertainty between country pairs at different geopolitical distances using a structural gravity model. - Geopolitical alignments do not systematically affect contemporaneous trade patterns. - When trade policy uncertainty is high: - Trade with “friends” increases relative to neutral countries. - Trade with rivals declines relative to neutral countries. - The result holds across sectors, with the exception of services, where conditions for trade are a priori more sensitive to geopolitical alignments. - The effect is stronger in manufacturing sectors of strategic importance, which are more likely to be in the crosshairs of policy reversals. ### Implications for welfare and global trade networks - These results highlight an important channel for the amplification of welfare losses due to the increased risk of geoeconomics fragmentation and associated uncertainty facing the global economy, even in the absence of any new trade barriers. - As firms take precautions, the network of global trade flows may start shifting towards close clusters of “friends”, generating winners and losers at the cost of overall efficiency. ### Policy recommendations - Negative effects can be mitigated through actions aimed at reducing trade policy uncertainty and trade tensions. These include: - Constructive multilateral engagement on thorny trade issues (e.g., subsidies and dispute settlement reform). - Renewed efforts to re-establish a fully functioning dispute settlement mechanism at the WTO. ### Suggestions for future research - Build more precise measures of trade policy uncertainty that better capture the bilateral nature of this uncertainty, since trade policy reversals are often targeted at specific partners, sectors, or firms. - Develop measures that would allow for a more disaggregated look at the impacts across trading partners and more precise predictions of the expected changes to the global trade network. *IMF Working Paper — Trading with Friends in Uncertain Times* --- _Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023124-print-pdf.pdf_