## _wp1452

## Source details

**Canonical URL:** [_wp1452](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1452.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1452.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1452.pdf.json)

---

### I. Introduction and context
- Over the past two decades, trade integration increased rapidly within the world economy, particularly within Asia.
- Gross trade in volume terms rose at an average rate of 8 percent per annum during 1990–2012, twice the average pace outside Asia.
- In value-added terms, trade increased at an average annual growth rate of over 10 percent during the same period, also double the average pace outside Asia.
- Asian economies became more vertically integrated as a tight-knit supply-chain network across the region was formed.
- Key research question: Have changes in trade patterns, in particular greater trade integration, led economies to move more in lockstep, and disproportionately so within Asia?

### II. Contributions, literature gaps, and methodological approach
- Literature and methodological gaps:
  - Earlier studies generally found trade intensity increases synchronization but many did not control simultaneously for fixed country-pair factors and common global shocks; omission can bias results.
  - When panel controls for country-pair heterogeneity and global shocks are included, the relationship between trade integration and BCS can weaken or become insignificant.
  - Gross trade data can misrepresent linkages in supply-chain networks; value-added trade data better capture true linkages.
- Contributions of the paper:
  - Accounts systematically for country-pair heterogeneity and common global shocks.
  - Uses value-added trade data (OECD-WTO TiVA) rather than gross trade.
  - Explores the nature of trade and specialization: vertical integration, trade specialization correlation, and intra-industry trade.
  - Controls for financial integration (bank, FDI, portfolio flows) and macro-policy synchronization.
  - Separately examines China shock spillovers and trade-channel propagation.
- Empirical strategy (baseline and augmented models):
  - Baseline: QCORR_{ij,t} = α_{ij} + α_t + f(TRADE_{ij,t-1}) + ε_{ij,t}.
  - Augmented: QCORR_{ij,t} = α_{ij} + α_t + f(TRADE_{ij,t-1}) + g(FINANCE_{ij,t-1}, POLICY_{ij,t-1}) + ε_{ij,t}.
  - Endogeneity addressed via country-pair fixed effects, lagged trade intensity, and instrumental variables (time-varying gravity variables; additional instruments for vertical integration: average import tariff on intermediate goods and difference in real per capita GDP).
- Core trade variables:
  - Trade Intensity (value-added), Vertical Integration, Intra-industry Trade (Grubel-Lloyd index), Trade Specialization Correlation (UNCTAD).
- Financial integration variables:
  - Banking integration (bilateral assets & liabilities share), Portfolio integration (CPIS), FDI integration (CDIS / UNCTAD-OECD historical).
- Policy synchronization variables:
  - Fiscal policy synchronization (quasi-correlation of structural fiscal balances), Monetary policy synchronization (−|r_it − r_jt| using short-term real rates purged of cycle), Exchange rate rigidity (negative of volatility of monthly nominal bilateral exchange rates).

### III. Stylized facts about business cycle synchronization (BCS)
- Measurement:
  - Default measure: instantaneous quasi-correlation (Abiad and others (2013)):
    - QCORR_{ij,t} = ( (g_{i,t} − ḡ_i) / σ_{i} ) * ( (g_{j,t} − ḡ_j) / σ_{j} )
    - g_{i,t} is output growth rate of country i in year t; ḡ_i and σ_i are mean and standard deviation of output growth over the sample. Growth measured as first difference of log real GDP.
  - Advantages: co-movement at every point in time; period mean converges to Pearson correlation; not bounded between −1 and 1; correlations on actual (not detrended) annual growth rates (~two decades).
- Empirical patterns (annual data for 63 countries: 34 advanced, 29 emerging; 13 Asian economies in dynamic factor model):
  - BCS spikes sharply in crisis times: largest spikes around the GFC outside Asia and around the Asian crisis of the late 1990s within Asia.
  - During normal times, BCS is much smaller but shows an upward trend since the 1990s globally, including in Asia.
  - Increase in BCS particularly large in Asia and Latin America, though still considerably lower than within the euro area during the 2000s.
  - Within Asia, BCS particularly high among ASEAN-5: Indonesia, Malaysia, the Philippines, Singapore, Thailand.
  - China’s output co-movements vis-à-vis the rest of Asia have increased, but Asian economies continue to co-move more with Japan and the United States.

### IV. Dynamic factor model: global, regional, and sub-regional roles
- Model setup:
  - Dynamic factor model using quarterly GDP, consumption, and investment from 2000Q1 to 2012Q4; factors and residuals follow AR(2); estimated by Bayesian MCMC (Carter and Kohn (1994)).
  - Type I: three factors — global, regional (Asia, South America, Europe, North America), and country.
  - Type II: adds sub-regional factor for parts of Asia (China supply-chain vs others; ASEAN-5 vs others; advanced Asia vs others).
  - Dataset: 34 economies including 13 in Asia, 4 in South America, 14 in Europe, 3 in North America.
- Key factor findings:
  - Global factor: steady positive 2003–07, plummet during GFC, bounce back thereafter.
  - Asian regional factor: distinct peak in aftermath of GFC consistent with steep recovery amid policy stimulus.
- Variance decomposition (Type I and Type II):
  - Global factor explains on average about 35 percent of output fluctuations in individual economies.
  - Contribution of global factor is lower in Asia than in North America and Europe, yet still larger than regional or country factors — no evidence Asia decoupled from world over last two decades.
  - Type II: global factors explain a greater share of fluctuations in China supply-chain economies than in rest of Asia; global factor has higher explanatory power in advanced Asia than in emerging Asia.
- Complementary AR(1)/VAR(1) model results (summary table entries preserved):
  - Table A1.1 — Variance Decomposition in the Model with AR(1):
    - World 0.17 0.20 0.37 0.16 0.07 0.23 0.60
    - Asia 0.22 0.20 0.41 0.09 0.07 0.17 0.58
    - Oceania 0.03 0.11 0.15 0.27 0.10 0.37 0.52
    - Latin America 0.15 0.11 0.25 0.20 0.04 0.24 0.49
    - Europe 0.14 0.24 0.38 0.17 0.07 0.24 0.63
    - North America 0.22 0.20 0.42 0.19 0.09 0.28 0.70
    - Others 0.21 0.19 0.40 0.24 0.07 0.31 0.71

### V. Empirical results — trade, finance, policy, and crisis interactions
- Baseline (trade only) highlights (standard errors clustered at country-pair level):
  - Trade Intensity (Gross): coefficient 0.0399 (not significant as reported).
  - Trade Intensity (Value Added):
    - OLS 0.0488*** (0.0154)
    - IV 0.249*** (0.0738)
    - OLS 0.0632*** (0.0152), IV 0.295*** (0.0709), OLS 0.0652*** (0.0158) in other specs.
  - Intra-industry Trade: 0.00313*** (0.00116); 0.00326*** (0.00119); 0.00322*** (0.00120).
  - Trade Specialization Correlation: 1.261*** (0.157); 1.419*** (0.166); 1.252*** (0.160).
  - Vertical Trade Integration: −0.125*** (0.0242) in one specification, but result not robust; vertical integration discarded in remainder due to lack of robustness.
  - R-squared reported ~0.58; Observations reported (examples): 18224, 18619, 18614, 18606.
- Augmented models with finance and policy highlights:
  - Trade Intensity (Value Added) remains positive and significant:
    - OLS 0.117*** (0.0270); IV 0.575*** (0.0898); other estimates: 0.118*** (0.0270), 0.566*** (0.0889), 0.851*** (0.280), 0.466*** (0.180), 0.658*** (0.196).
  - Banking Integration: negative and significant in many specifications:
    - Examples: −0.0287* (0.0159); −0.0343*** (0.0127); −0.0488*** (0.0140); −0.0551*** (0.0122).
  - Portfolio Integration: −4.897* (2.620) in one column (weak).
  - FDI Integration: −1.338 (0.952) in one column (weak).
  - Fiscal Policy Synchronization: 0.0587*** (0.0127); 0.0584*** (0.0130).
  - Monetary Policy Synchronization: 0.00339** (0.00149); 0.00345** (0.00151).
  - Exchange Rate Rigidity: 0.136*** (0.0168); 0.137*** (0.0168).
  - R-squared up to 0.77 in one specification; observations varied (examples reported: 12186, 12159, 28609, 09095 as shown).
- Crisis vs non-crisis interactions (Table 4 highlights):
  - Trade integration has a greater impact during crisis times than in normal times; trade amplifies the synchronizing impact of common shocks.
  - Examples:
    - OLS: Trade Intensity 0.118*** (0.0270), interaction with GFC dummy: 0.0527** (0.0266).
    - IV: Trade Intensity 0.566*** (0.0889), interaction with GFC dummy 0.777*** (0.166).
  - Banking integration can be negative in tranquil times (e.g., −0.0282* (0.0158)), but interaction with GFC dummy is strongly positive: 0.397*** (0.0761) OLS; 0.351*** (0.0596) IV — banking integration turns into a contagion channel during crisis.
  - First-stage F-statistics reported (examples): 38.76, 41.67.
- Quantitative interpretation (as reported):
  - Moving a country pair from 25th to 75th percentile of trade intensity (~increase by about 12 percentage points) raises quasi-correlation by around 0.2 in normal times (median correlation ~0.1 across sample).
  - During crisis times, same movement raises quasi-correlation by almost 0.25.
  - Moving from 25th to 75th percentile of financial integration implies an increase in quasi-correlation by some 0.25 in crisis times, vs small negative association in tranquil times.
  - Figure 13 (as described): trend rise in trade intensity accounts for about a fourth of trend rise in BCS within Asia between 1999-2005 and 2006-12 (excl. 2009); other factors changed little over time.

### VI. China-specific spillovers
- Setup:
  - Panel regression on quarterly data: ∆y_{i,t} regressed on China growth shock_{t} interacted with trade linkages between China and country i. Trade variables interpolated to quarterly frequency using quarterly gross trade fluctuations.
  - China growth shock: residual from panel with country and time dummies over 1995Q1–2012Q4.
- Table 5 results (selected):
  - Bilateral Trade Intensity × Shock: 0.101** (F-statistic 3.05); Observations 27863; R-squared 0.201.
  - Trade Dependence on China × Shock (value added exported by country i to China): 0.119** (3.53); Observations 31693; R-squared 0.238.
  - Final Demand from China × Shock (value added exported by country i to China for final demand that “stays” in China): 0.148*** (3.76); Observations reported variably (table shows 3181 formatting); R-squared reported 0.210 in one column.
  - Coefficients are sums of contemporaneous and lagged coefficients; controls include VIX and oil prices; country fixed effects and time trends included.
- Interpretation (as reported):
  - Based on column (3), a 1 percentage increase in China’s growth is estimated to raise GDP growth in the median Asian economy by over 0.3 percentage point after a year, and in the median non-Asian economy by about 0.15 percentage point after a year.

### VII. Conclusions and policy implications
- Main empirical conclusions:
  - Using value-added trade data is key to finding a robust positive impact of trade intensity on BCS; gross trade measures often fail to capture true linkages in vertically fragmented production.
  - Trade integration increases BCS in normal times and acts as an amplification mechanism in crisis times.
  - Intra-industry trade and trade specialization correlation have positive effects on BCS (weaker than trade intensity); vertical integration shows no robust additional impact over trade intensity in these specifications.
  - Macro-policy synchronization (fiscal, monetary, exchange rate rigidity) matters for BCS.
  - Spikes in BCS coincide with major global/regional shocks (GFC, Asian crisis); trend rise in trade intensity contributed about one-fourth of the trend rise in BCS within Asia.
  - Financial integration contributed substantially to the spike in BCS during the GFC and can switch from a dampening role in tranquil times to a contagion channel in crisis times.
  - China growth shocks have sizeable international spillovers that depend on bilateral value-added trade linkages and final-demand dependence on China; spillovers stronger in Asia.
- Policy implications and projections:
  - All else equal, BCS among economies likely to continue rising as world economic integration increases.
  - Trade integration will be a driving force in normal times and an amplifier in crises.
  - Growth shocks in China will induce more synchronization as China’s role as source of final demand grows.
  - Future increases in financial integration—especially in emerging economies where financial integration lags trade—could strengthen spillovers and synchronization in crisis periods, even if they may reduce co-movement in tranquil times.

*Source: _wp1452 - Section IV concludes; _wp1452 - REFERENCES; APPENDICES I–III (IMF staff estimates and source material contained in the PDF).*

### References .............................................................................................................

### References

### I. Introduction and context
- Over the past two decades, trade integration increased rapidly within the world economy, particularly within Asia.
- Gross trade in volume terms rose at an average rate of 8 percent per annum during 1990–2012, twice the average pace outside Asia.
- In value-added terms, trade increased at an average annual growth rate of over 10 percent during the same period, also double the average pace outside Asia.
- Asian economies became more vertically integrated as a tight-knit supply-chain network across the region was formed.
- Key research question: Have changes in trade patterns, in particular greater trade integration, led economies to move more in lockstep, and disproportionately so within Asia?

### Literature and methodological gaps
- Earlier empirical studies (e.g., Frankel and Rose, 1997, 1998; Baxter and Kouparitsas, 2005; Imbs, 2004; Inklaar and others, 2008; Kumakura, 2006; Park and Shin, 2009) generally found that trade intensity increases synchronization, with varying magnitudes.
- Many prior studies did not account simultaneously for fixed country-pair factors and common global shocks; failing to control for both can cause omitted variable bias (Kalemli-Ozcan, Papaioannou and Peydro, 2013).
- When panel controls for country-pair heterogeneity and global shocks are included, the relationship between trade integration and BCS can weaken or become insignificant (examples: Kalemli-Ozcan, Papaioannou and Peydro (2013); Abiad and others (2013); Calderon, Chong and Stein, 2007; Shin and Wang, 2004).
- Gross trade data can misrepresent trade linkages amid supply-chain networks; value-added trade data better capture true linkages (Unteroberdoerster and others, 2011).

### Contributions of the paper
- Accounts systematically for country-pair heterogeneity and common global shocks throughout the analysis.
- Uses value-added trade data (building on the OECD-WTO initiative on trade in value added) instead of gross trade data.
- Explores BCS impact of the nature of trade and specialization, including vertical integration, trade specialization correlation, and intra-industry trade.
- Controls for other influences on BCS such as financial integration (bank, FDI, portfolio flows) and macro-economic policy synchronization.
- Separately examines the impact of a China shock on other economies and the propagation of that shock through trade channels.

### Main findings and stylized facts
- BCS spikes across the globe during crises; using the quasi-correlation indicator proposed by Abiad and others (2013):
  - Output correlations outside of Asia peaked during the 2008/09 Global Financial Crisis (GFC).
  - Within Asia the biggest spikes in BCS occurred during the Asian Crisis of the late 1990s.
- During normal times, BCS is typically much lower but has been on an upward trend over the past two decades, particularly in Asia.
- Dynamic factor analysis results:
  - Global factors play a major role in driving business cycles in Asia and elsewhere.
- Macro panel analysis (sample of 63 advanced and emerging economies spanning the last two decades), using value-added trade data:
  - Bilateral trade intensity has a significant, positive, and robust effect on bilateral BCS among country pairs.
  - Significant, though smaller, effects are found for intra-industry trade and correlation of trade specialization.
- Crisis vs normal times:
  - The impact of trade integration on BCS is greater in crisis times than in normal times; trade amplifies the synchronizing impact of common shocks.
  - This is qualitatively consistent with findings that financial integration increases BCS in crisis times, even though it typically reduces it during normal times.
- China-specific spillovers:
  - Growth spillovers from China are significant and sizeable.
  - Spillovers are larger in economies that are more dependent on final demand from China in value-added terms.
  - China is a growing source of BCS, especially within Asia.

### Structure of the remainder of the paper (as presented)
- Section II: Stylized facts about BCS, with focus on Asia; includes factor analysis on global, regional, and sub-regional roles.
- Section III: Role of intensity and nature of trade for BCS; covers effects of financial integration; includes related analysis of spillovers of growth shocks in China to other economies.

*Source: _wp1452 - References.*

### Section IV concludes.

### _wp1452 - Section IV concludes.

### Stylized facts about business cycle synchronization (BCS) in Asia
- Measurement:
  - Default measure: instantaneous quasi-correlation proposed by Abiad and others (2013):
    - QCORR_{ij,t} = ( (g_{i,t} − ḡ_i) / σ_{i} ) * ( (g_{j,t} − ḡ_j) / σ_{j} )
    - g_{i,t} denotes output growth rate of country i in year t; ḡ_i and σ_i are mean and standard deviation of output growth rate of country i over the sample period. Growth rate measured as first difference of log real GDP.
  - Advantages emphasized:
    - Enables calculation of co-movement at every point in time (avoids rolling-window autocorrelation).
    - Period mean asymptotically converges to Pearson correlation; measure not necessarily bounded between -1 and 1.
    - Correlations calculated on actual (not detrended) growth rates using annual data (~two decades).
- Empirical patterns:
  - Using annual data for 63 countries (34 advanced, 29 emerging; 13 of the 63 in Asia in the dynamic factor model; other sample breakdowns used elsewhere):
    - BCS spikes sharply in crisis times: largest spikes around the GFC outside Asia and around the Asian crisis of the late 1990s within Asia.
    - During normal times, BCS is much smaller but shows some upward trend since the 1990s globally, including in Asia.
    - Increase in BCS particularly large in Asia and Latin America, though still considerably lower than within euro area during the 2000s.
    - Within Asia, BCS particularly high among ASEAN-5: Indonesia, Malaysia, the Philippines, Singapore, Thailand.
    - China’s output co-movements vis-à-vis the rest of Asia have increased, but Asian economies continue to co-move more with Japan and the United States.

### Role of global, regional, and sub-regional factors (dynamic factor model results)
- Model setup:
  - Dynamic factor model (following Hirata, Kose, and Otrok (2013)) using quarterly GDP, consumption, and investment from 2000Q1 to 2012Q4; factors and residuals follow AR(2); estimated by Bayesian MCMC (Carter and Kohn (1994)).
  - Two model types:
    - Type I: three factors — global, regional (Asia, South America, Europe, North America), and country.
    - Type II: adds sub-regional factor for parts of Asia (China supply-chain vs others; ASEAN-5 vs others; advanced Asia vs others).
  - Dataset for factor model: 34 economies including 13 in Asia, 4 in South America, 14 in Europe, 3 in North America.
- Key factor findings:
  - Global factor: steady positive 2003–07, plummet during GFC, bounce back thereafter.
  - Asian regional factor: distinct peak in aftermath of GFC, consistent with steep recovery amid policy stimulus.
- Variance decomposition (Type I and Type II):
  - Global factor explains on average about 35 percent of output fluctuations in individual economies.
  - Contribution of global factor is lower in Asia than in North America and Europe, yet still larger than regional or country factors — no evidence Asia decoupled from world over last two decades.
  - Type II: global factors explain a greater share of fluctuations in China supply-chain economies than in rest of Asia; global factor has higher explanatory power in advanced Asia than in emerging Asia.

### Trade, financial integration, policy coordination: theory and empirical strategy
- Theory overview:
  - Trade integration: ambiguous theoretical effects on BCS (specialization vs intra-industry trade vs demand-driven channels).
  - Financial integration: ambiguous theoretical effects (capital reallocation and specialization could reduce BCS; global banking shocks could increase BCS); empirical literature mixed.
  - Policy coordination: similar fiscal/monetary/exchange rate policies across countries expected to raise BCS; empirical support from OECD and Asian studies.
- Empirical strategy:
  - Core elements: panel regressions with country-pair fixed effects and time fixed effects; primary use of trade in value-added (OECD-WTO Trade in Value Added database) instead of gross trade.
  - Four trade variables: trade intensity (value-added), vertical integration, intra-industry trade (Grubel-Lloyd index), trade specialization correlation (UNCTAD).
  - Financial integration variables: banking integration (bilateral assets & liabilities share), portfolio and FDI integration (shorter sample).
  - Policy synchronization variables: fiscal policy synchronization (structural fiscal balance purged from cycle), monetary policy synchronization (real policy rate purged), exchange rate rigidity.
  - Baseline model (equation 1): QCORR_{ij,t} = α_{ij} + α_t + f(TRADE_{ij,t-1}) + ε_{ij,t}.
  - Augmented model (equation 2): QCORR_{ij,t} = α_{ij} + α_t + f(TRADE_{ij,t-1}) + g(FINANCE_{ij,t-1}, POLICY_{ij,t-1}) + ε_{ij,t}.
  - Endogeneity addressed via: country-pair fixed effects, lagged trade intensity, instrumental variables (time-varying gravity variables; for vertical integration additional instruments: average import tariff on intermediate goods and difference in real per capita GDP).

### Empirical results — trade, finance, policy and crisis interactions
- Baseline (trade only) — Table 2 highlights (standard errors clustered at country-pair level):
  - Trade Intensity (Gross): coefficient 0.0399 (not significant as reported).
  - Trade Intensity (Value Added): OLS 0.0488*** (0.0154); IV 0.249*** (0.0738); OLS in other specs 0.0632*** (0.0152), IV 0.295*** (0.0709), OLS 0.0652*** (0.0158).
  - Intra-industry Trade: 0.00313*** (0.00116); 0.00326*** (0.00119); 0.00322*** (0.00120) in various columns — positive and significant at 1 percent.
  - Trade Specialization Correlation: 1.261*** (0.157); 1.419*** (0.166); 1.252*** (0.160) — positive and significant at 1 percent.
  - Vertical Trade Integration: -0.125*** (0.0242) in one specification, but result not robust; authors discard vertical integration in remainder due to lack of robustness across specifications and instruments.
  - R-squared reported ~0.58; Observations varied (e.g., 18224, 18619, 18614, 18619, 18614, 18606 as shown).
- Augmented models with finance and policy — Table 3 highlights:
  - Trade Intensity (Value Added) stays positive and significant: e.g., OLS 0.117*** (0.0270); IV 0.575*** (0.0898); other IV/OLS estimates 0.118*** (0.0270), 0.566*** (0.0889), 0.851*** (0.280), 0.466*** (0.180), 0.658*** (0.196).
  - Banking Integration: negative and significant in many specifications: e.g., -0.0287* (0.0159); -0.0343*** (0.0127); -0.0282* (0.0158); -0.0336*** (0.0127); -0.0488*** (0.0140); -0.0543*** (0.0125); -0.0551*** (0.0122).
  - Portfolio Integration: -4.897* (2.620) in one column (weak).
  - FDI Integration: -1.338 (0.952) in one column (weak).
  - Fiscal Policy Synchronization: 0.0587*** (0.0127); 0.0584*** (0.0130).
  - Monetary Policy Synchronization: 0.00339** (0.00149); 0.00345** (0.00151).
  - Exchange Rate Rigidity: 0.136*** (0.0168); 0.137*** (0.0168).
  - R-squared up to 0.77 in one specification; observations varied (e.g., 12186, 12159, 28609, 09095 etc. as reported).
- Crisis vs non-crisis interactions — Table 4 highlights:
  - Interaction results indicate trade integration has greater impact during crisis times.
  - Example coefficients: Trade Intensity 0.118*** (0.0270), interaction with GFC dummy: 0.0527** (0.0266) in OLS; IV: Trade Intensity 0.566*** (0.0889), interaction with GFC dummy 0.777*** (0.166).
  - Banking Integration main effect can be negative in tranquil times (e.g., -0.0282* (0.0158)), but interaction with GFC dummy is strongly positive: 0.397*** (0.0761) OLS; 0.351*** (0.0596) IV — implying banking integration turns into a contagion channel during crisis.
  - First-stage F-statistics reported (e.g., 38.76, 41.67).
- Quantitative interpretation (figures referenced):
  - Moving a country pair from 25th to 75th percentile of trade intensity (~increase by about 12 percentage points) raises quasi-correlation by around 0.2 in normal times (median correlation ~0.1 across sample).
  - During crisis times, same movement raises quasi-correlation by almost 0.25.
  - Moving from 25th to 75th percentile of financial integration implies an increase in quasi-correlation by some 0.25 in crisis times, vs small negative association in tranquil times.
  - Figure 13: trend rise in trade intensity accounts for about a fourth of trend rise in BCS within Asia between 1999-2005 and 2006-12 (excl. 2009); other factors changed little over time.

### China spillovers
- Setup:
  - Panel regression on quarterly data estimating cross-country growth spillovers from China: ∆y_{i,t} regressed on China growth shock_{t} interacted with trade linkages between China and country i (trade variables interpolated to quarterly frequency using quarterly gross trade fluctuations).
  - China growth shock computed as residual from panel with country and time dummies over 1995Q1–2012Q4.
- Results (Table 5):
  - Bilateral Trade Intensity x Shock: coefficient 0.101** (F-statistic 3.05); Observations 27863; R-squared 0.201.
  - Trade Dependence on China x Shock (value added exported by country i to China): 0.119** (3.53); Observations 31693; R-squared 0.238.
  - Final Demand from China x Shock (value added exported by country i to China for final demand that “stays” in China): 0.148*** (3.76); Observations 3181? (table shows 3181 then 81 formatting — reported Observations 31693/3181/…; R-squared 0.210 as shown). Coefficients are sums of contemporaneous and lagged coefficients; controls include VIX and oil prices; country fixed effects and time trends included.
- Interpretation:
  - Based on column (3), a 1 percentage increase in China’s growth is estimated to raise GDP growth in the median Asian economy by over 0.3 percentage point after a year, and in the median non-Asian economy by about 0.15 percentage point after a year.

### Concluding remarks and policy implications
- Main empirical conclusions:
  - Using value-added trade data is key to finding a robust positive impact of trade intensity on BCS; gross trade measures often fail to capture true linkages in vertically fragmented production.
  - Trade integration increases BCS in normal times and acts as an amplification mechanism in crisis times.
  - Intra-industry trade and trade specialization correlation have positive effects on BCS (weaker than trade intensity); vertical integration shows no robust additional impact over trade intensity in these specifications.
  - Macro-policy synchronization (fiscal, monetary, exchange rate rigidity) matters for BCS.
  - Spikes in BCS coincide with major global/regional shocks (GFC, Asian crisis); trend rise in trade intensity contributed materially to the modest upward trend in BCS, especially in Asia (about one-fourth of the trend rise in Asia).
  - Financial integration contributed substantially to the spike in BCS during the GFC and can switch from a dampening role in tranquil times to a contagion channel in crisis times.
  - China growth shocks have sizeable international spillovers that depend on bilateral value-added trade linkages and final-demand dependence on China; spillovers stronger in Asia.
- Policy implications and projections:
  - All else equal, BCS among economies likely to continue rising as world economic integration increases.
  - Trade integration will be a driving force in normal times and an amplifier in crises.
  - Growth shocks in China will induce more synchronization as China’s role as source of final demand grows.
  - Future increases in financial integration—especially in emerging economies where financial integration lags trade—could strengthen spillovers and synchronization in crisis periods, even if they may reduce co-movement in tranquil times.

*Italic: Source — _wp1452 - Section IV concludes.*

### REFERENCES

### _wp1452 - REFERENCES

### References (selected)
- Abiad, A., D. Furceri, S. Kalemli-Ozcan, and A. Pescatori, 2013, “Dancing Together? Spillovers, Common Shocks, and the Role of Financial and Trade Linkages,” in World Economic Outlook, (Washington: International Monetary Fund, October), pp. 81–111.
- Ahuja, A. and M. Nabar, 2012, “Investment-led Growth in China: Global Spillovers,” IMF Working Paper 12/267, (Washington: International Monetary Fund).
- Backus, D., P. Kehoe and F. Kydland, 1992, “International Real Business Cycles,” Journal of Political Economy, Vol. 100, pp. 745–775.
- Baxter, M. and M. Kouparitsas, 2005, “Determinants of Business Cycle Co-movement: A Robust Analysis,” Journal of Monetary Economics, Vol. 52, pp. 113–157.
- Bordo, M., and T. Helbling, 2010, “International Business Cycle Synchronization in Historical Perspective,” NBER Working Paper No. 16103, (Massachusetts: National Bureau of Economic Research).
- Calderon, C., A. Chong, and E. Stein, 2007, “Trade Intensity and Business Cycle Synchronization: Are Developing Countries any Different?” Journal of International Economics, Vol. 71, pp. 2–21.
- Carter, C. K. and R. Kohn, 1994, “On Gibbs Sampling for State Space Models,” Biometrika, Vol. 81, pp. 541–553.
- Di Giovanni, J. and A. Levchenko, 2010, “Putting the Parts together: Trade, Vertical Linkages and Business Cycle Co-movement,” American Economics Journal: Macroeconomics, Vol. 2, pp. 95–124.
- Frankel, J. A., and A. K. Rose, 1997 and 1998, on economic structure, common currency adoption, and endogeneity of OCA criteria.
- Grubel, H. G. and P. J. Lloyd, 1975, Intra-industry Trade: the Theory and Measurement of International Trade in Differentiated Products, (New York: Wiley).
- Hirata, H., M. Kose, and C. Otrok, 2013 “Regionalization vs. Globalization,” IMF Working Paper 13/19, (Washington: International Monetary Fund).
- Imbs, J., 2004 and 2006, on trade, specialization, synchronization, and financial integration.
- Inklaar, R., R. J-A. Pin, and J. Haan, 2008, “Trade and Business Cycle Synchronization in OECD countries—A re-examination,” European Economic Review, Vol. 52, pp. 646-666.
- International Monetary Fund, 2013, German-Central European Supply Chain—Cluster Report, IMF Country Report No. 13/263 (Washington, August).
- Kalemli-Ozcan, S., B.E. Sorensen, and O. Yosha, 2003; Kalemli-Ozcan et al., 2013 on global banks, crisis transmission, and financial globalization.
- Kim, C-J., and C. R. Nelson, 1999, State-Space Models with Regime Switching: Classical and Gibbs-Sampling Approaches with Applications, (Massachusetts: The MIT Press).
- Kumakura, M., 2006, “Trade and Business Cycle Co-movement in Asia-Pacific,” Journal of Asian Economics, Vol. 17, pp. 622–645.
- McKinnon, R. I., “Optimum Currency Areas,” The American Economic Review, Vol. 53, pp. 717–725.
- Mohammad, A., O. Unteroberdoerster, and J. Vichyanond, 2011, “Implications of Asia’s Regional Supply Chain for Rebalancing Growth,” in Regional Economic Outlook: Asia and Pacific, (Washington: International Monetary Fund, April), pp. 47–60.
- Morgan, D., B. Rime, and P. Strahan, 2004, “Bank Integration and State Business Cycles,” Quarterly Journal of Economics, Vol. 119, pp. 1555–1585.
- Mundell, R. A., 1961, “A Theory of Optimum Currency Areas,” The American Economic Review, Vol. 51, pp. 657–665.
- OECD and WTO, 2012, “Trade in Value-Added: Concepts, Methodologies, and Challenges,” (OECD-WTO online document).
- Otto, G., G. Voss, and L. Willard, 2001, “Understanding OECD Output Correlations,” Research Discussion Paper No. 2001–05, (Sydney: Reserve Bank of Australia).
- Park, Y-C. and K. Shin, 2009, “Economic Integration and Changes in the Business Cycle in East Asia: Is the Region Decoupling from the Rest of the World,” Asian Economic Papers, Vol. 8:1, pp. 107–140.
- Shin, K., and Y. Wang, 2003 and 2004, on trade integration and business cycle synchronization in East Asia.
- Xing, Y., and N. Detert, 2010, “How the iPhone widens the United States Trade Deficit with the People’s Republic of China”, ADBI Working Paper Series No. 257, (Manila: Asian Development Bank).

### Appendix I — Data
- Source of expenditure-side real GDP components: IMF WEO data.
- Sample period: 2000Q1 to 2012Q4.
- Countries covered: 34 countries partitioned into four geographical regions and subgroups:
  - Asia: China, Hong Kong SAR, India, Indonesia, Japan, Korea, Malaysia, Philippines, Singapore, Taiwan POC, Thailand, Australia, and New Zealand
    - China supply-chain: China, Korea, Malaysia, Philippines, Taiwan POC, and Thailand
    - ASEAN-5: Indonesia, Malaysia, Philippines, Singapore, and Thailand
    - Advanced Asian economies: Hong Kong SAR, Japan, Korea, Singapore, Taiwan POC, Australia, and New Zealand
  - South America: Brazil, Argentina, Chile, and Venezuela
  - Europe: France, Germany, Greece, Italy, Poland, Portugal, Spain, Sweden, United Kingdom, Russia, Czech Republic, Hungary, Latvia, and Slovenia
  - North America: Canada, Mexico, and United States

### Appendix I — Model and estimation (baseline model without sub-regions)
- Variables and notation (as presented):
  - y_{i,t} stands for GDP, consumption, and investment growth rates of countries.
  - c_{i,t} stands for factors; and e_{i,t} represents residuals.
  - β_{.,i}, γ_{.,i}, ρ_{.,i} are parameters, with β_{.,i} called factor loadings.
- Identification and normalization:
  - Factors and factor loadings are only identified by their product (β_{.,i} c_{.,t}); scale and sign not independently identified.
  - Following Hirata and others (2013), a positive sign is imposed to the factor loading of the first country in a group.
  - Variance (σ^2_{f,1}) of factor equations is assigned the value estimated during initialization via principal component approach and classical estimation.
- Estimation method: Bayesian Markov Chain Monte Carlo (MCMC) with the Carter and Kohn (1994) algorithm.
- MCMC drawing steps (as numbered):
  1) Draw AR coefficients (φ_{.,j}) of the factor equation from a multivariate normal distribution conditional on factor (c_{.,t}) and variance of shocks (σ^2_{f,j}). Priors: mean = Zero and variance = 10.
  2) Draw coefficients (β_{.,i}) of the observation equation from a multivariate normal distribution conditional on factor (c_{.,t}) and the error term equation (Ω_{1,i}, σ^2_{e,i}). Serial correlation removed by multiplying both sides with (I − Ω_{1,i})^{-1}. Priors: mean = Zero and variance = 10.
  3) Draw variance (σ^2_{e,i}) of the error term equation from an inverted gamma distribution conditional on coefficients (β_{.,i}, Ω_{.,i}) and factors (c_{.,t}). Prior distribution parameters: α = 1 and β = 1/10.
  4) Draw coefficients (Ω_{.,i}) of the error term equation from a multivariate normal distribution conditional on factor (c_{.,t}) and the other parameters (β_{.,i}, σ^2_{e,i}). Priors: mean = Zero and variance = 10.
  5) Draw factors (c_{.,t}) from a multivariate normal distribution conditional on parameters. The Carter and Kohn (1994) algorithm is applied in calculating parameters of the multivariate normal distribution.
  6) Return to step (1).
- State-space representation components (observation and transition equations) are specified with observation vectors y_{i}(t), factor vector α_{i}(t), loading matrix B, covariance R, transition matrices, and factor covariance structures as given in the source.

### Appendix I — Complementary model (VAR(1) / Bordo and Helbling approach)
- Complementary model features:
  - Combines a VAR approach with the baseline model to allow interactions between economic variables.
  - Simplifying assumptions: VAR(1) model; only output considered (not consumption or investment); no sub-regional factors included; dynamics of the error term dropped outside of VAR(1).
  - To reduce dimensionality, non-zero lag parameters (Ψ_{k,i}) allowed only for country i’s own lagged GDP growth rate, the United States, and the largest economy in each region (China, Brazil, Germany, and Australia, respectively).
- Complementary model equation structure and factor dynamics are described similarly to the baseline but with VAR(1) term included.
- Estimation method: almost the same as baseline model.

### Additional results, figures, and variance decomposition summary
- Figure A1.1 (Model with sub-regions): Estimated regional factors show sub-regional factors mostly not significantly different from zero except for a few years; China supply-chain and advanced economies show relatively distinguished sub-regional factors.
- Figure A1.2 (Model with AR(1)): Global factor is estimated tightly and similar to baseline; Asian regional factor consistent over the second half of the period; European factor shows larger downward trend after the GFC than in baseline.
- Interpretation notes:
  - The transmission of the global factor affects economies through both their own lags and those of major countries.
  - In the AR(1) complementary model, the influence of the global factor on Asian economies is not smaller than its impact on European economies.
  - The explanatory power of the Asia regional factor is relatively small, suggesting global factors continue to play a key role in driving the regional cycle over the full period 2000–12.
- Table A1.1 — Variance Decomposition in the Model with AR(1) (entries preserved exactly as presented)
  - Row labels and corresponding numeric entries (preserved verbatim):
    - World 0.17 0.20 0.37 0.16 0.07 0.23 0.60
    - Asia 0.22 0.20 0.41 0.09 0.07 0.17 0.58
    - Oceania 0.03 0.11 0.15 0.27 0.10 0.37 0.52
    - Latin America 0.15 0.11 0.25 0.20 0.04 0.24 0.49
    - Europe 0.14 0.24 0.38 0.17 0.07 0.24 0.63
    - North America 0.22 0.20 0.42 0.19 0.09 0.28 0.70
    - Others 0.21 0.19 0.40 0.24 0.07 0.31 0.71
  - Column heading context (as in source): OwnTrasmission Sum OwnTrasmission Sum Global Regional Sum

*Source: _wp1452 - REFERENCES (IMF staff estimates and source material contained in the PDF).*

### APPENDIX II. FURTHER DETAILS ON THE DATA

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp1452.pdf_