## 3.1    Reallocation

## Source details

**Canonical URL:** [3.1    Reallocation](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024076-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024076-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024076-print-pdf.pdf.json)

---

### Measurement approach
- Reallocation across trading and FDI partners is measured with the Lilien (1982) Index using annual bilateral import data and inward FDI for each country since 2000.
- Modified Lilien Index (MLI) definition:
  - ML I rt = sqrt( ∑ S irt ×(ln(x irt /x irt−1 )−ln(X rt /X rt−1 )) 2 )
  - ln(x irt /x irt−1 ) = growth of imports from trading partner i, in country r, in time period t.
  - ln(X rt /X rt−1 ) = growth in overall imports of country r.
  - S irt = x irt /X rt = average share of imports from partner i in total imports of country r.
  - The index equals 0 if there is no structural change between t−1 and t; higher values imply faster structural change (greater reallocation across partners).
- Data sources and notes:
  - Annual bilateral import data and inward FDI used since 2000.
  - Trade Data Monitor and fDi Markets used for empirical calculations (as indicated in figure notes).
  - The index captures reallocation across partners (structural change in partner composition), not firm-level events such as mergers and acquisitions or projects that do not create new jobs.

### Empirical timing and patterns
- Trade flows:
  - Reallocation across import partners accelerated with the intensification of the U.S.-China trade tensions in 2018.
  - The increase in the Lilien index for trade flows is especially pronounced among advanced economies (Table A1).
- FDI flows:
  - The uptick in reallocation across FDI sources is most pronounced in the past couple of years (Figure 2).
  - fDi Markets does not track mergers and acquisitions and other international equity investments, investment projects that do not create new jobs, or companies that establish a foreign subsidiary without a physical company presence.
- Aggregate-change observations:
  - Compared to the average observed over 2003-2019, import reallocation has increased by roughly 15 percent post pandemic in the full sample of countries.
  - Import reallocation has increased by almost 40 percent for the sample of advanced economies.
  - Foreign direct investment exhibits striking similarity, with reallocation across sources of inward FDI increasing significantly more for advanced economies (Table A1, panel B).
- Regression evidence (Table A1 — Panel A. Trade flows; coefficients from regressions of the Lilien Index on time-period indicators; excluded period 2003-2007):
  - 2008-2012 coefficients:
    - All: 0.340* (standard error (0.139))
    - AEs: 0.430** (0.126)
    - EMDEs: 0.274 (0.195)
    - U.S. bloc: 0.397** (0.123)
    - Others: 0.286 (0.206)
  - 2013-2020 coefficients:
    - All: 0.059 (0.113)
    - AEs: 0.470** (0.168)
    - EMDEs: -0.147 (0.143)
    - U.S. bloc: 0.275 (0.168)
    - Others: -0.085 (0.151)
  - 2021-2023 coefficients:
    - All: 0.706** (0.215)
    - AEs: 1.118*** (0.219)
    - EMDEs: 0.486 (0.300)
    - U.S. bloc: 1.157*** (0.216)
    - Others: 0.431 (0.315)
  - Observations:
    - All: 2,639
    - AEs: 785
    - EMDEs: 1,854
    - U.S. bloc: 893
    - Others: 1,746
  - R2:
    - All: 0.488
    - AEs: 0.432
    - EMDEs: 0.453
    - U.S. bloc: 0.375
    - Others: 0.463
- Regression evidence (Table A1 — Panel B. FDI flows):
  - 2008-2012 coefficients:
    - All: 0.0608 (0.602)
    - AEs: -1.16041 (0.766)
    - EMDEs: 1.0551 (0.867)
    - U.S. bloc: -0.66730 (0.774)
    - Others: 0.7149 (0.893)
  - 2013-2020 coefficients:
    - All: 0.3120 (0.624)
    - AEs: -0.48770 (0.855)
    - EMDEs: 0.9983 (0.886)
    - U.S. bloc: -0.38200 (0.804)
    - Others: 0.9438 (0.932)
  - 2021-2023 coefficients:
    - All: 1.8130** (0.779)
    - AEs: 2.8929** (1.311)
    - EMDEs: 1.03642 (0.913)
    - U.S. bloc: 2.2989** (1.114)
    - Others: 1.4429 (1.088)
  - Observations:
    - All: 1,660
    - AEs: 712
    - EMDEs: 948
    - U.S. bloc: 740
    - Others: 920
  - R2:
    - All: 0.278
    - AEs: 0.228
    - EMDEs: 0.301
    - U.S. bloc: 0.251
    - Others: 0.293
  - Country FE: Y Y Y Y Y

### Interpretation and caveats
- Main interpretation:
  - Underneath broadly stable aggregate trends of trade and FDI, there was a sharp uptick in the extent of reallocation across import and FDI sources.
  - The MLI measures structural change in the composition of imports across partners; higher MLI indicates faster structural change.
  - The analysis focuses on partner reconfiguration and does not capture certain types of cross-border equity flows or non-job-creating investment projects.
- Caveats:
  - fDi Markets limitations: does not capture mergers and acquisitions and other international equity investments, investment projects that do not create new jobs, or companies that establish a foreign subsidiary without a physical company presence.

*Source: wpiea2024076-print-pdf - 3.1    Reallocation*

### 3.1    Reallocation

### 3.1    Reallocation

### Measurement approach
- Reallocation across trading and FDI partners is gauged using the Lilien (1982) Index, computed with annual bilateral import data and inward FDI for each country since 2000.
- The Modified Lilien Index (MLI) is defined in the source as:
  ML I rt = sqrt( ∑ S irt ×(ln(x irt /x irt−1 )−ln(X rt /X rt−1 )) 2 )
  where ln(x irt /x irt−1 ) is the growth of imports from trading partner i, in country r, in time period t, ln(X rt /X rt−1 ) is the growth in overall imports of country r, and S irt = x irt /X rt is the average share of imports from partner i in total imports of country r.  The index is equal to 0 if there is no structural change between t−1 and t.  The higher the value of the index, the faster is the structural change; in other words, the bigger the reallocation of imports across trading partners.
- Data sources and notes:
  - Annual bilateral import data and inward FDI used since 2000.
  - Trade Data Monitor and fDi Markets used for empirical calculations (as indicated in figure notes).
  - The index captures reallocation across partners (structural change in partner composition), not firm-level events such as mergers and acquisitions or projects that do not create new jobs.

### Empirical timing and patterns
- Trade flows:
  - Reallocation across import partners accelerated with the intensification of the U.S.-China trade tensions in 2018.
  - The increase in the Lilien index computed for trade flows is especially pronounced among advanced economies (Table A1).
- FDI flows:
  - The uptick in reallocation across FDI sources is most pronounced in the past couple of years (Figure 2).
  - fDi Markets does not track mergers and acquisitions and other international equity investments, investment projects that do not create new jobs, or companies that establish a foreign subsidiary without a physical company presence.
- Aggregate-change observations:
  - Compared to the average observed over 2003-2019, import reallocation has increased by roughly 15 percent post pandemic in the full sample of countries.
  - Import reallocation has increased by almost 40 percent for the sample of advanced economies.
  - Foreign direct investment exhibits striking similarity, with reallocation across sources of inward FDI increasing significantly more for advanced economies (Table A1, panel B).

### Interpretation and caveats
- Underneath broadly stable aggregate trends of trade and FDI, there was a sharp uptick in the extent of reallocation across import and FDI sources.
- The MLI measures structural change in the composition of imports across partners; higher MLI indicates faster structural change.
- The analysis of reallocation focuses on partner reconfiguration and does not capture certain types of cross-border equity flows or non-job-creating investment projects.

*Italic: Source: wpiea2024076-print-pdf - 3.1    Reallocation*

### 2022.  Moreover, while trade in primary goods accounted for more than 40 percent of total trade

### wpiea2024076-print-pdf - 2022.  Moreover, while trade in primary goods accounted for more than 40 percent of total trade

### Fragmentation and trade composition
- During 1947-1952, trade in primary goods accounted for "more than 40 percent" of total trade.
- During 2019-2023, trade in primary goods was "14 percent" of cross-country goods flows, reflecting the rise in trade in intermediates and final goods.
- Despite extreme fragmentation of trade between blocs during the Cold War, trade within blocs flourished due to policies and technological advances that "effecticely reduced trade costs between countries in the same geopolitical bloc."

### Role of nonaligned countries as connectors
- Gravity-equation results:
  - During the Cold War, trade with nonaligned economies was significantly lower ("-37 percent") than trade within the two blocs (Table 1, Panel B).
  - In the current period, trade with nonaligned economies has "kept up with within-bloc trade" (Panel A; PPML estimates in columns 4 and 8).
- U.S.-China tensions example:
  - China’s share of U.S. goods imports fell from "22 percent" in 2017 to "14 percent" in 2023 (U.S. Census Bureau data).
  - The 2018 hike in U.S. tariffs on Chinese imports "effectively curbed Chinese imports."
  - China lost rank as a destination for outward U.S. FDI; emerging markets such as India, Mexico, and UAE gained in announced FDI projects.
- Evidence of indirect links replacing direct U.S.–China ties:
  - As China lost U.S. import market share, Mexico, Canada, and several Asian economies (notably Vietnam) gained prominence.
  - Countries that gained most in U.S. import shares (e.g., Mexico and Vietnam) also gained in China’s export shares and are larger recipients of Chinese FDI.

### Empirical associations shown in figures (Figure 4 summaries)
- Panel A (nonaligned countries): change in U.S. import shares (2018-23 vs 2013-17) vs change in Chinese export shares (same periods)
  - Weighted regression slope = "1.634" (p-value = "0.000"); n = "57".
  - Interpretation: a 1 percent increase in the U.S. import share between 2013-17 and 2018-23 is associated with a 1.6 percent higher share of Chinese exports over the same period.
- Panel C (nonaligned countries): change in U.S. import shares (2018-23 vs 2013-17) vs change in Chinese outward FDI (same periods)
  - Weighted regression slope = "0.719" (p-value = "0.003"); n = "44".
  - Interpretation: the same 1 percent increase in U.S. import share is associated with a 0.7 percent increase in the share of FDI from China.
- Panel B (Cold War period, nonaligned countries): change of Western bloc import shares (1952-56 vs 1934-38) vs change of Eastern bloc export shares (same periods)
  - Weighted regression slope = "−0.026" (p-value = "0.628"); n = "81".
  - Interpretation: the two series are essentially orthogonal—no evidence that nonaligned economies served as connectors during the Cold War.
- Robustness: results are robust to i) using Chinese exports or FDI as weights; ii) using weights computed in the 2018-23 period; iii) excluding outliers; and for panels A and C, robust to controlling for average real GDP growth in 2018-2023.

### Why nonaligned economies differ today versus the 1950s
- Relative economic footprint in 1950:
  - Western and Eastern blocs together accounted for "roughly 85 percent" of global GDP and "more than half" of the world’s population.
  - Nonaligned countries were mostly developing economies receiving foreign aid, technical assistance and military equipment driven by geopolitical considerations.
- Indicators of greater integration and heft for today's nonaligned economies:
  - Median nonaligned economy trade-to-GDP ratio: "80 percent" in 2019 vs "40 percent of GDP" in 1960.
  - Average most favored nation import tariff of the median nonaligned country: "12 percent" in 2019 vs "40 percent" in 1960.
  - Median nonaligned economy free trade agreements with partners equivalent to "one fifth of global GDP" in 2015 vs "0 percent" in 1960.

### Conclusions and policy implications
- Stylized facts:
  - Trade and investment between blocs is decreasing relative to trade and investment within blocs, similar to the Cold War pattern.
  - The current decoupling is smaller than during the Cold War but is in an early stage and "could worsen significantly" if geopolitical tensions persist and restrictive trade policies continue to mount.
- Deglobalization and resilience:
  - Increasing geoeconomic fragmentation has not produced deglobalization, then or now, but for different reasons:
    - During the Cold War, within-bloc integration and technological improvements drove a surge in trade.
    - Today, resilience reflects re-routing of flows via connector (nonaligned) countries.
  - Nonaligned connector countries could benefit from rising geoeconomic fragmentation.
- Fundamental conundrum:
  - The global economy is more resilient partly because it substitutes away from tariffed or sanctioned trade; however, such substitution "does not necessarily increase diversification, resilience, or lessen strategic dependence."
- Policy path forward:
  - Outcomes hinge on whether policymakers choose to preserve gains from an integrated global economy (potentially tolerating re-routed flows) or pursue "more severe forms of decoupling."

*Source: IMF staff analysis in wpiea2024076-print-pdf (excerpt).*

### References

### wpiea2024076-print-pdf - References

### Key themes in cited literature
- Geoeconomic fragmentation and the economic risks from a fractured world economy are central topics (e.g., Aiyar et al. (2023a); Aiyar, Presbitero, and Ruta (2023b)).
- The role of geopolitical alignment in economic outcomes is addressed, including effects on FDI flows and "Investing in Friends" (Aiyar, Malacrino and Presbitero (2024)).
- Global supply chain dynamics, including "de-globalisation", nearshoring, friend-shoring, and reallocation, are examined across multiple contributions (Antràs (2020); Alfaro et al. (2024); Alfaro and Chor (2023); Attinasi et al. (2023); Fajgelbaum et al. (2024); Javorcik et al. (2022); Lovely (2023)).
- The economic consequences of the U.S.–China trade war and related policy shocks are documented (Bown (2021); Dang, Alicia and Zhao (2023); Fajgelbaum and Khandelwal (2022); Fajgelbaum et al. (2024); Rotunno et al. (2023); Utar et al. (2023); Xue (2023)).
- Measurement and empirical methods commonly used include PPML, gravity models, Lilien Index regressions, and geopolitical risk indices (Caldara and Iacoviello (2022); Silva and Tenreyro (2006); Correria, Guimarães and Zylkin (2020); Yotov (2022); Bailey, Strezhnev and Voeten (2017)).
- Historical parallels and comparisons to the Cold War era are used to frame current fragmentation (Cooper (2010); Huntington (1998); Leffler (1984); Schiller (1955); Churchill (1946)).

### Appendix: Figures — Main analytical notes and metrics
- Figure A1: Rising Fragmentation Pressures
  - Panel A plots the number of harmful restrictions on trade and investment per year.
  - Panel B plots the geopolitical risk developed by Caldara and Iacoviello (2022) and the fragmentation risk index, which measures the average number of sentences, per thousand earnings calls, that mention at least one of: deglobalization, reshoring, onshoring, nearshoring, friend-shoring, localization, regionalization.
  - Sources listed: Caldara and Iacoviello (2022); Global Trade Alert; NL Analytics; and authors’ calculations.
- Figure A2: Timing of Trade Fragmentation: The Cold War and Now
  - Charts plot semi-elasticity of trade for flows between blocs and with nonaligned, with associated 90 percent confidence bands, estimated with PPML and a fully saturated gravity model as in equation 1 in the main text.
  - Cold War results use yearly data from 1920 to 1990—with 1947 as excluded year—and bloc definition based on Gokmen (2017).
  - Recent period results use quarterly trade data from 2017:Q1 to 2023:Q3 (with 2021:Q4 as excluded quarter), with bloc definition based on the Ideal Point Distance (Bailey et al. (2017)).
  - Sources: Trade Data Monitor; TRADHIST (Fouquin and Hugot (2016)); and IMF staff calculations.
- Figure A3: Change in Russia’s Trade Flows
  - Plots the share of Russia’s goods trade with China, the United States and the Euro area, before and after Russia’s invasion of Ukraine.
  - Sources: Trade Data Monitor; and authors’ calculations.

### Appendix: Tables — Key regression coefficients and statistics
- Table A1: Reallocation across Import and FDI Sources: Evolution over Time Periods
  - Panel A. Trade flows (coefficients from regressions of the Lilien Index on time-period indicators; excluded period 2003-2007)
    - 2008-2012 coefficients:
      - All: 0.340* (standard error (0.139))
      - AEs: 0.430** (0.126)
      - EMDEs: 0.274 (0.195)
      - U.S. bloc: 0.397** (0.123)
      - Others: 0.286 (0.206)
    - 2013-2020 coefficients:
      - All: 0.059 (0.113)
      - AEs: 0.470** (0.168)
      - EMDEs: -0.147 (0.143)
      - U.S. bloc: 0.275 (0.168)
      - Others: -0.085 (0.151)
    - 2021-2023 coefficients:
      - All: 0.706** (0.215)
      - AEs: 1.118*** (0.219)
      - EMDEs: 0.486 (0.300)
      - U.S. bloc: 1.157*** (0.216)
      - Others: 0.431 (0.315)
    - Observations:
      - All: 2,639
      - AEs: 785
      - EMDEs: 1,854
      - U.S. bloc: 893
      - Others: 1,746
    - R2:
      - All: 0.488
      - AEs: 0.432
      - EMDEs: 0.453
      - U.S. bloc: 0.375
      - Others: 0.463
  - Panel B. FDI flows
    - 2008-2012 coefficients:
      - All: 0.0608 (0.602)
      - AEs: -1.16041 (0.766)
      - EMDEs: 1.0551 (0.867)
      - U.S. bloc: -0.66730 (0.774)
      - Others: 0.7149 (0.893)
    - 2013-2020 coefficients:
      - All: 0.3120 (0.624)
      - AEs: -0.48770 (0.855)
      - EMDEs: 0.9983 (0.886)
      - U.S. bloc: -0.38200 (0.804)
      - Others: 0.9438 (0.932)
    - 2021-2023 coefficients:
      - All: 1.8130** (0.779)
      - AEs: 2.8929** (1.311)
      - EMDEs: 1.03642 (0.913)
      - U.S. bloc: 2.2989** (1.114)
      - Others: 1.4429 (1.088)
    - Observations:
      - All: 1,660
      - AEs: 712
      - EMDEs: 948
      - U.S. bloc: 740
      - Others: 920
    - R2:
      - All: 0.278
      - AEs: 0.228
      - EMDEs: 0.301
      - U.S. bloc: 0.251
      - Others: 0.293
    - Country FE: Y Y Y Y Y
- Table A2: Trade and investment flows between blocs: Excluding the U.S. and China
  - Estimation approaches: OLS (columns 1-2 and 5-6) and PPML (columns 3-4 and 7-8) with dependent variables bilateral trade in US dollars (panels A and B) or number of FDI projects (panel C).
  - Sample periods:
    - Panel A: 2017:q1-2023:q4 (quarterly)
    - Panel B: 1920-1990 (excluding 1939-1945)
    - Panel C: 2010:q1-2023:q4 (quarterly)
  - Key variable definitions:
    - Post War = 1 from 2022:q1 onwards (period following Russia’s invasion of Ukraine)
    - Cold War = 1 for years 1947-1991
    - Between Bloc = 1 if source and destination not in same bloc
    - Nonaligned = 1 if at least one country in the pair is nonaligned
    - Wider bloc definition uses Ideal Point Distance (Bailey et al. (2017)); narrower bloc definition uses a hypothetical Western vs Eastern bloc list (details in notes).
  - Panel A. Trade around the Russian invasion of Ukraine — Wider vs Narrower bloc definitions (selected coefficients)
    - Between Bloc * Post War:
      - OLS (Wider): -0.0676*** (0.012)
      - OLS (Narrower): -0.0959** (0.043)
      - PPML (Wider): -0.0483*** (0.014)
      - PPML (Narrower): -0.0876 (0.058)
      - Wider PPML (alternative columns): -0.4535*** (0.052); -0.2902*** (0.063)
      - Narrower PPML (alternative columns): -0.6162*** (0.176); -0.6852*** (0.150)
    - Nonaligned * Post War:
      - OLS (Wider): 0.0005 (0.008)
      - OLS (Narrower): 0.0624* (0.038)
      - PPML (Wider): -0.0124 (0.016)
      - PPML (Narrower): 0.1237*** (0.029)
      - Additional columns report: -0.0282*** (0.006); -0.0293** (0.014); 0.1071*** (0.028); 0.1601*** (0.034)
    - Observations (Panel A):
      - Wider OLS: 230,935
      - Wider PPML: 230,817
      - Narrower OLS: 253,544
      - Narrower PPML: 253,428
  - Panel B. Trade during the Cold War (selected coefficients)
    - Between Bloc * Cold War:
      - OLS (Wider): -0.4258*** (0.061)
      - PPML (Wider): -0.5033*** (0.166)
      - PPML (Narrower): -1.0146*** (0.119)
    - Nonaligned * Cold War:
      - OLS (Wider): -0.2463*** (0.044)
      - PPML (Wider): -0.2621** (0.116)
      - PPML (Narrower): -0.5736*** (0.222)
    - Observations (Panel B): 795,793; 795,385; 738,340; 661,277 (across columns)
  - Panel C. FDI around the Russian invasion of Ukraine (selected coefficients)
    - Between Bloc * Post War:
      - OLS (Wider): 0.7472*** (0.143)
      - PPML (Wider): -0.5259*** (0.151)
      - OLS (Narrower): -2.7213*** (0.317)
      - PPML (Narrower): -1.7384*** (0.317)
      - Additional columns: -0.1991*** (0.055); -0.2265*** (0.077); -0.2158*** (0.076); -0.0806 (0.069)
    - Nonaligned * Post War:
      - OLS (Wider): -0.0455 (0.050)
      - PPML (Wider): -0.2111* (0.115)
      - OLS (Narrower): 0.0056 (0.034)
      - PPML (Narrower): -0.0277 (0.061)
      - Additional columns: -0.0309 (0.057); -0.21520 (0.186); 0.0004 (0.039); -0.1218* (0.069)
    - Observations (Panel C): 119,616; 118,272; 119,616; 109,864; 110,768; 109,480; 110,768; 101,260 (across columns)
  - Fixed effects and controls:
    - Country-pair FE: Y in all columns
    - Time FE: present in various specifications (indicated in table notes)
    - Source x Time FE and Destination x Time FE: included in some specifications (N/Y patterns as indicated)

*Italic: Source — wpiea2024076-print-pdf - References (appendix notes and tables).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024076-print-pdf.pdf_