## _wp13194

## Source details

**Canonical URL:** [_wp13194](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13194.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13194.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13194.pdf.json)

---

### Executive Summary — Major themes and structure
- I. Introduction
- II. The GVAR Model – Model Structure and Data Set Used
  - A. GVAR Model —A Non-Technical Summary
  - B. The Data Set (2000–2011)
- III. Estimation of the GVAR Model
  - A. Specification and Estimation of the Country-Specific Models
  - B. Testing for Structural Breaks
  - C. Contemporaneous Effects of Foreign Variables on their Domestic Counterparts
  - D. Pair-wise Cross-Country Correlations: Variables and Residuals
- IV. Dynamic Analysis Using Generalized Impulse Response Functions and Generalized Forecast Error Variance Decomposition
  - A. Negative Shock to EURO-West Real GDP growth
  - B. Shock to Real GDP Growth in Nordic Countries
  - C. Shock to Real GDP Growth in Central Europe
  - D. Shock to Real GDP Growth in the Baltic countries
  - E. Shock to Euro-West Credit Growth
  - F. Shock to Interest Rate in ADV (the UK, Switzerland, Iceland, and Israel)
  - G. Negative Shock to Real Credit Growth in Central Europe
- V. Conclusions
- Appendices: Appendix I. GVAR Model Structure; Appendix II. Results of Unit Root Test and Weak Exogeneity Tests; Appendix tables and Appendix Figures

### Methodology and innovations
- Uses a modified Global VAR (GVAR) framework (Pesaran et al. (2004); Dées, di Mauro, Pesaran, and Smith (2007)) with country-specific VARX/VARX* models and foreign variables treated for weak exogeneity.
- Key innovation: combined trade and financial weights (based on BIS reporting banks’ external position data) to construct country-specific foreign variables.
- Weights for foreign-variable construction: fixed averages for the period 2005–11 (time-varying weights also tested; results qualitatively similar).
- GVAR dynamic analysis: Generalized Impulse Response Functions (GIRFs) over 20 quarters (emphasis on a two-year horizon) and Generalized Forecast Error Variance Decomposition (GFEVD).
- Country coverage: 33 European economies (includes 12 CESEE countries individually).
- Sample period for estimation: 2000Q2–2011Q4.
- Variable definitions preserved as in source:
  - real GDP growth: y_i,t = 400 * (ln(Y_i,t) − ln(Y_i,t−1))
  - inflation: π_i,t = ln(P_i,t) − ln(P_i,t−1) (CPI)
  - long-term interest rate: r_i,t = 1/4 * ln(1 + i_i,t/100)
  - real credit growth: Δ ln(credit_i,t / GDP_i,t)

### Data, weights, and coverage
- Primary data set period: 2000–2011; weights reported as averages for 2005–2011 (appendix A2a-b for 2000-2004).
- Weights composed from:
  - bilateral annual trade flows (IMF DOTS)
  - BIS locational banking statistics (external positions of international banks)
- Weight matrices shown as averages for 2005–2011; bilateral weights sum to one by column.
- Notable descriptive weight facts:
  - By 2011 Western Europe was the destination of 75 percent of exports from CESEE, while 68 percent of imports into CESEE were from Western Europe.
  - Share of foreign banks (assets of local banking system) in 2011 exceeded 70 percent in most CESEE countries (exceptions: European CIS countries and Turkey).
  - EURO-West group weights toward most sample countries range between 64% - 91% (except Baltic countries).
  - For Baltic countries, Nordic group joint weights range between 39%-67%.
  - ADV → Russia = 11% and ADV → Turkey = 17% (ADV group weights toward CESEE generally below 5% otherwise).

### Key empirical findings — co-movements and contemporaneous elasticities
- Co-movements:
  - Strong co-movements in output growth and long-term interest rates across European countries.
  - Weaker co-movements in inflation and real credit growth.
- Contemporaneous elasticities and examples:
  - For CESEE countries, a 1% increase in foreign output growth in a given quarter leads to an average 0.4% increase in domestic output growth within the same quarter.
  - Significant contemporaneous elasticities for credit growth observed in Hungary, Lithuania, Romania, Slovakia, and Turkey, and in ADV and NORD in advanced Europe.
  - High contemporaneous elasticity between domestic and foreign inflation for Lithuania; high contemporaneous interest-rate co-movement between EURO-West and NORD.
- Cross-section correlations (averages reported):
  - real GDP growth ~57%
  - interest rates ~52%
  - inflation ~27%
  - real credit growth ~36%
  - Correlations decline in first differences (real credit growth first-difference average ~6%).
- Residual diagnostics:
  - Pair-wise residual correlations generally small (absolute correlations generally < 10%), supporting foreign-variable construction approach.

### Stability, cointegration, and robustness tests
- Unit root, weak exogeneity, and co-integration tests performed (Appendix II and Tables A3–A6).
- Majority of country models show reasonable co-integration results (trace statistic at 95% significance level); selected r = 0 and r = 1 trace statistics reported for countries (examples preserved in Appendix).
- Structural break tests (PKsup, PKmsq, Nyblom , QLR, MW, APW, heteroskedasticity-robust variants):
  - PK tests do not reject structural stability in any cases.
  - Other tests reject parameter constancy in a small share (4–10%) of cases depending on test and variable.
  - Overall: structural instability is not a serious concern for the sample despite boom-bust and crises coverage.
- GVAR stability:
  - Modulus of every eigenvalue of the estimated GVAR is on or within the unit circle (model stable); some complex eigenvalues induce oscillating IRF features.
- Bootstrap IRF bands:
  - Rapidly widening IRF bands; mean IRF results are indicative and results beyond 6–8 quarters should be treated with caution.

### Dynamic experiments — shock types and main regional responses
- Shock categories analyzed: real shocks (GDP growth), financial shocks (interest rate, credit growth), inflation shocks.
- Regional aggregates and definitions used in GIRF experiments:
  - CE: Hungary, Poland, Czech Republic, Slovak Republic, Slovenia.
  - SE: Romania, Croatia.
  - Baltic: Estonia, Latvia, Lithuania.
  - Turkey and Russia analyzed separately.
  - Regional weights for GIRF based on GDP at PPP.
- Selected quantitative GIRF and GFEVD findings (values preserved exactly as reported):
  - Negative 1 percentage point (p.p.) shock to EURO-West real GDP growth:
    - CESEE countries: GDP growth drops by 0.65–1.25 p.p. in the same quarter.
    - Nordic countries: decline of about 0.5 p.p. in the same quarter.
    - ADV group: decline of about 0.5 p.p. in the same quarter.
    - One s.d. shock ≡ about 0.2 p.p. in annualized q-o-q growth rate for EURO-West on impact.
    - Country heterogeneity: Poland −0.4 p.p.; Slovakia and Lithuania around −1.5 to −1.6 p.p.
    - GFEVD (Table 9): EURO-West shocks contribute over half of rescaled total variance in first four quarters; EURO-West variance row: "62.8 62.5 57.8 58.3 56.3 55.3 54.8 54.3 53.8 53.5 53.3".
  - Positive 1 p.p. shock to Nordic real GDP growth:
    - Baltics: +1 p.p. same quarter, +1.5 p.p. in third quarter.
    - EURO-West: immediate +0.2 p.p., rises to ~0.3 p.p.
    - ADV: immediate +0.1 p.p.
    - One s.d. shock ≡ 0.14 p.p. increase in growth on impact to NORD.
    - GFEVD (Table 10): NORD output innovation is a main source; also importance of ADV interest rates and oil price.
  - 1 p.p. shock to CE real GDP growth:
    - CE own dynamics: declines gradually, settles in ~6 quarters.
    - EURO-West: +0.1–0.2 p.p. in first two quarters.
    - Baltics: +0.2 p.p. on impact, >0.4 p.p. in second quarter.
    - One s.d. shock ≡ 0.2 p.p. increase in CE growth on impact.
  - 1 p.p. shock to Baltic real GDP growth:
    - Baltics own response persistent: +0.6 p.p. in third quarter, +0.4 p.p. in fourth quarter, stabilizing ~0.4 p.p. in ~6 quarters.
    - Immediate bump: Nordics +0.02 p.p.; Russia +0.06 p.p.
    - One s.d. shock ≡ 0.67 p.p. increase in Baltic growth on impact.
    - GFEVD (Table 12): major external sources: EURO-West output and oil price; ADV important via interest rate, credit, and output.
  - One p.p. positive shock to EURO-West credit growth:
    - Limited impact on other sub-regions; dissipates in two to three quarters.
    - ADV: on impact decline in credit growth rate below 0.6 p.p., quickly dissipates.
    - GFEVD (Table 13): EURO-West innovations ~90 percent of forecast variance for real credit growth ("EURO-West variance 90.8 81.9 80.8 81.5 84.2 84.8 84.8 85.1 86.1 86.6 86.9").
    - For euro area, one s.d. shock ≡ 0.26 p.p. increase in annualized q-o-q growth rate.
  - One percent (100 basis points) increase in ADV group interest rate (UK ~80% of ADV GDP in PPP):
    - ADV interest rate path: after 100 bps increase, declines slightly by 25 bps by fifth quarter and continues to decline.
    - EURO-West: immediate +10 bps, rising to +60 bps by end of fourth quarter.
    - NORD: initial +10 bps, reaches +50 bps in fourth quarter.
    - CESEE: muted/negligible movements.
    - Real GDP growth responses by 6th quarter (q-o-q annualized):
      - EURO-West ~1.7 p.p.; ADV ~1.9 p.p.; NORD 3 p.p.; Baltics 6 p.p.; Russia 2 p.p.; SE 1.5 p.p.; Turkey and CE ~1.2 p.p.
    - One s.d. shock to ADV interest rate ≡ 4.5 basis points (bps) rise on impact.
    - GFEVD (Table 14): ADV interest-rate shocks dominate ("ADV variance 86.5 83.9 84.3 84.5 84.5 84.3 82.6 81.2 79.8 78.3 76.7").
  - Negative 1 p.p. shock to CE real credit growth (one s.d. ≡ −0.69 p.p. for CE):
    - CE real GDP: on impact −0.2 p.p.; second quarter −0.4 p.p., then dissipates with oscillation.
    - CESEE: Baltic and SE countries negatively affected ~0.3 p.p. average in first two quarters.
    - Russia and Turkey: small initial bump; Russia declines in second quarter.
    - GFEVD (Table 15): CE credit-growth own innovation dominates; external important sources include EURO-West output, oil price, and ADV output.
  - One p.p. shock to EURO-West inflation:
    - EURO-West inflation effect declines sharply; dissipates in three to four quarters.
    - ADV: on impact inflation declines by 0.05 p.p.
    - NORD: inflation hardly changes on impact.
    - CESEE: Baltics +0.4 p.p. on impact; SE +0.5 p.p. on impact.
    - One s.d. shock to euro area inflation ≡ 0.36 p.p. on impact.
    - EURO-West inflation variance entries from Table 16 preserved: "EURO-West variance 62.8 32.3 36.0 36.1 36.0 37.7 37.5 37.4 38.3 38.5 38.8".

### Interpretation and policy-relevant implications
- EURO-West group is the dominant source of shocks explaining large shares of forecast-error variance in output, inflation, interest rates, and credit across Europe.
- Financial shocks (long-term interest-rate shocks) propagate to real activity with rising effects over time; monitoring financial conditions is crucial for macroeconomic stability.
- Inflation pass-through across borders is limited relative to output and interest-rate co-movement; domestic inflation dynamics and oil-price shocks matter substantially for inflation transmission.
- Credit-growth spillovers from EURO-West to CESEE are muted; funding patterns (western European parent banks channeling funds) help explain limited transmission.
- CESEE benefited from integration (real per capita income rose by 4 percent annually in 1995–2007 for CESEE) but also faced rapid transmission of risks (capital flow reversals and parent-bank pressures in 2008 triggered sudden stops and deep crises).
- As CESEE economies grow via income convergence, their shocks have increasing effects on Western European partners (notably CE shocks affecting euro area GDP).
- Combining trade and financial linkages is important for accurately assessing regional spillovers; either dimension alone is incomplete.

### Appendices — model structure, tests, and data notes
- Appendix I: GVAR model structure; country-specific VARX(p,q) with foreign variables x_i,t* constructed using weights ω_ij (ω_ii = 0; sum_j ω_ij = 1); ECM representation and cointegration rank r_i details preserved from source.
- Appendix II: Unit-root test and weak exogeneity tests; variables generally assumed I(1) with examples and test-statistics preserved (ADF and WS statistics at 5% level reported in tables).
- Unit-root and weak exogeneity highlights (selected):
  - Real GDP growth: mostly I(1) or borderline I(0)/I(1) (examples include ADV, EURO, Lithuania, NORD, Poland); Slovakia I(0).
  - Inflation: Croatia, Latvia, Lithuania, Poland, Romania, Russia, Slovakia, Slovenia, Turkey I(1); ADV, Estonia, EURO, Hungary, NORD appear I(0).
  - Interest rates: I(1) except Hungary borderline I(0)/I(1).
  - Credit growth: mostly I(1) except Hungary, Poland, Slovakia appear I(0).
  - Weak exogeneity tests (Table A8): with the exception of real GDP growth, oil price for ADV, all foreign variables appear to be exogenous (selected F-statistics and Fcrit_0.05 preserved in Appendix).
- Weight matrices and changes:
  - Table A2a: Weight Matrix (2000-2004 averages) and Table A2b: Changes from 2000-2004 to 2005-2011 (changes > 1% highlighted in source).
- Data sources: Haver Analytics series used for CPI/HICP, GDP (SWDA), long-term interest rates, and various banking claims series (Table A1 excerpts preserved).

*Source: _wp13194 — Executive Summary, selected sections, experiments, tables and appendix extracts as provided.*

### Executive Summary ......................................................................................................

### Executive Summary

### Major themes and structure
- I. Introduction
- II. The GVAR Model – Model Structure and Data Set Used
  - A. GVAR Model —A Non-Technical Summary
  - B. The Data Set (2000–2011)
- III. Estimation of the GVAR Model
  - A. Specification and Estimation of the Country-Specific Models
  - B. Testing for Structural Breaks
  - C. Contemporaneous Effects of Foreign Variables on their Domestic Counterparts
  - D. Pair-wise Cross-Country Correlations: Variables and Residuals
- IV. Dynamic Analysis Using Generalized Impulse Response Functions and Generalized Forecast Error Variance Decomposition
  - A. Negative Shock to EURO-West Real GDP growth
  - B. Shock to Real GDP Growth in Nordic Countries
  - C. Shock to Real GDP Growth in Central Europe
  - D. Shock to Real GDP Growth in the Baltic countries
  - E. Shock to Euro-West Credit Growth
  - F. Shock to Interest Rate in ADV (the UK, Switzerland, Iceland, and Israel)
  - G. Negative Shock to Real Credit Growth in Central Europe
- V. Conclusions
- Appendices (structure and tests)
  - Appendix I. GVAR Model Structure
  - Appendix II. Results of Unit Root Test and Weak Exogeneity Tests
  - Appendix tables and Appendix Figures

### Key content elements (tables and figures inventory from the Executive Summary)
- Tables (selected)
  - 1. Countries and Regions in the GVAR Model
  - 2. Weight Matrix (average of weights for the period 2005-2011) 1/
  - 3. Trade Weight Matrix (average of weights for the period 2005-2011)
  - 4. Financial Weight Matrix (average of weights for the period 2005-2011)
  - 5. Number of rejections of the null of parameter constancy per variable across the
  - 6. Contemporaneous Effects of Foreign Variables on Their Domestic Counterparts
  - 7. Average Pair-wise Cross-Section Correlations: Variables and Residuals
  - 8. Regional Weights for the GIRF Exercise
  - 9–16. Generalized Forecast Error Variance Decompositions for specified shocks (indexed 9 through 16)
  - Appendix tables A1–A8 covering Data Source, alternative weight matrices, Unit Root Tests at the 5% significance level, VARX order, Cointegration Results, and Weak Exogeneity tests
- Figures (selected)
  - 1. Modulus of the Eigenvalues of the Estimated GVAR model
  - 2–10. Generalized Impulse Response Functions for specified shocks (indexed 2 through 10)
  - Appendix Figures A1–A9: Additional GIRFs for specified shocks

### Data coverage and period
- The primary data set period reported: 2000–2011
- Weight matrices reported as averages for the period 2005-2011 (and appendix A2a-b for 2000-2004)

### Methodological components highlighted
- Use of a GVAR (Global Vector Autoregression) model with country-specific models and foreign variables treated for weak exogeneity
- Estimation includes specification tests, structural break testing, contemporaneous effect estimation, pair-wise cross-country correlation analysis
- Dynamic analysis conducted using Generalized Impulse Response Functions (GIRFs) and Generalized Forecast Error Variance Decomposition for a range of shocks (real GDP, credit growth, interest rates, inflation), with regional/group-specific exercises (EURO-West, Nordic, Central Europe, Baltic, ADV)

### Analytical focus areas
- Transmission of shocks across countries and regions via trade and financial linkages captured by weight matrices
- Stability and parameter constancy testing across variables
- Role of foreign variables’ contemporaneous effects on domestic macroeconomic variables
- Decomposition of forecast error variance to attribute contributions of shocks across regions and variables

*Source: _wp13194 - Executive Summary (PDF) — contents and inventory as provided.*

### EXECUTIVE SUMMARY

### EXECUTIVE SUMMARY

### Overview
- Uses a modified version of the Global VAR (GVAR) model (Pesaran et al. (2004) framework as developed in Dées, di Mauro, Pesaran, and Smith (2007)) to study propagation of real and financial shocks across European countries, with expanded focus on Central Eastern and Southeastern Europe (CESEE).
- Focus variables: real GDP growth, inflation, real credit growth, and long term interest rates.
- Country coverage: 33 European economies including all Western European countries and a representative set of CESEE economies.

### Methodology and Innovations
- Key innovation: combined trade and financial weights (based on BIS reporting banks’ external position data) to construct country-specific foreign variables in the GVAR, capturing both trade and financial linkages.
- GVAR construction: individual country VAR models extended with weakly exogenous foreign variables (weighted averages of partners’ corresponding variables), estimated country-by-country, then assembled into one global system.
- Weights for foreign-variable construction are fixed averages for the period 2005–11 (time-varying weights also tested; results qualitatively similar).
- GIRF (generalized impulse response functions) and GFEVD (generalized forecast error variance decomposition) used for dynamic analysis; GIRF over 20 quarters but emphasis on a two-year horizon.

### Data and Weighting Details
- Sample period for estimation: 2000Q2–2011Q4.
- Variables definitions (as used in the model):
  - real GDP growth: y_i,t = 400 * (ln(Y_i,t) − ln(Y_i,t−1))
  - inflation: π_i,t = ln(P_i,t) − ln(P_i,t−1) (CPI)
  - long-term interest rate: r_i,t = 1/4 * ln(1 + i_i,t/100)
  - real credit growth: Δ ln(credit_i,t / GDP_i,t)
- Data sources referenced (full list in Appendix Table A1).
- Weights composed from:
  - bilateral annual trade flows (IMF DOTS)
  - BIS locational banking statistics (external positions of international banks)
- Weight matrices presented as averages for 2005–2011 (trade matrix, financial matrix, and combined matrix). Bilateral weights shown in columns and sum up to one.

### Key Empirical Findings
- Co-movements:
  - Strong co-movements in output growth and long-term interest rates across European countries.
  - Somewhat weaker co-movements in inflation and real credit growth.
- Regional shock transmission:
  - Shocks to euro area output growth reverberate strongly across European countries including Nordic countries and CESEE countries.
  - Shocks to the UK long-term interest rate have a strong impact on long-term interest rates in the euro area and the Nordic countries, but weak impact on CESEE countries.
  - Interest rate shocks have notable impacts on output across Europe.
  - Inflation pass-through from the euro area to CESEE and other Western European countries is weak.
  - Shocks to credit growth in the euro area have limited impact on credit growth in CESEE.
- Sub-regional ties:
  - Baltic countries are highly sensitive to output shocks from the Nordic countries (reflecting close trade and financial linkages).
  - Central European economies’ shocks to GDP and credit have some impact on euro area GDP growth.
  - Shocks to the Baltic countries’ real GDP growth have a small impact on growth of the Nordic countries and Russia; impact elsewhere is muted.
- Cross-country linkages and weights:
  - By 2011, Western Europe was the destination of 75 percent of exports from CESEE, while 68 percent of imports into CESEE were from Western Europe.
  - Share of foreign banks (in terms of assets of local banking system) in 2011 exceeded 70 percent in most CESEE countries (notable exceptions: European CIS countries and Turkey).
  - The euro area (EURO-West group) dominates as a foreign partner with weights ranging between 64% - 91% for most sample countries, except the Baltic countries.
  - For the Baltic countries the Nordic group joint weights range between 39%-67%, exceeding euro area influence.
  - ADV group weights toward CESEE generally below 5%, except ADV → Russia = 11% and ADV → Turkey = 17%.
  - Financial linkages accentuate euro-west and Nordic importance for CESEE beyond trade-only patterns.
- Contemporaneous elasticities:
  - For CESEE countries, a 1% increase in foreign output growth in a given quarter leads to an average 0.4% increase in domestic output growth within the same quarter.
  - Significant contemporaneous elasticities for credit growth observed in Hungary, Lithuania, Romania, Slovakia, and Turkey, and in ADV and NORD in advanced Europe.
  - High contemporaneous elasticity between domestic and foreign inflation for Lithuania; high contemporaneous interest-rate co-movement between EURO-West and NORD.
- Cross-country correlations and residual diagnostics:
  - Average cross-section correlations (levels): real GDP growth ~57%, interest rates ~52%, inflation ~27%, real credit growth ~36%.
  - Correlations decline when using first differences; real credit growth first-difference average ~6%.
  - Residuals from country VARX models have generally small pair-wise correlations (absolute correlations generally < 10%), supporting effectiveness of foreign-variable construction.

### Stability and Robustness Diagnostics
- Unit root, weak exogeneity, and co-integration tests performed (results in Appendix II and Tables A3–A6). Majority of country models show reasonable co-integration results (trace statistic at 95% significance level).
- Structural break tests conducted (PKsup, PKmsq, Nyblom , QLR, MW, APW, and heteroskedasticity-robust variants). Summary:
  - PK tests do not reject structural stability in any cases.
  - Other tests reject parameter constancy in a small share (4–10%) of cases depending on test and variable.
  - Overall: structural instability is not a serious concern for the sample despite the severe boom-bust cycle and crises covered in the period.
- GVAR stability: modulus of every eigenvalue of the estimated GVAR is on or within the unit circle (model stable). Some complex eigenvalues induce oscillating IRF features.
- Bootstrap simulations show rapidly widening IRF bands; mean IRF results are indicative and results beyond 6–8 quarters should be treated with caution.

### Dynamic Experiments and Regional Response Patterns
- Shock categories analyzed: real shocks (GDP growth), financial shocks (interest rate, credit growth), and inflation shocks.
- Regional aggregation for GIRF experiments:
  - CE sub-region (“CE”): Hungary, Poland, Czech Republic, Slovak Republic, Slovenia.
  - SE sub-region (“SE”): Romania, Croatia.
  - Baltic sub-region (“Baltic”): Estonia, Latvia, Lithuania.
  - Turkey and Russia analyzed separately (not included in CE/SE/Baltic aggregates).
  - Regional weights for GIRF based on GDP at PPP (table provided in paper).
- Main dynamic results (selected):
  - Output shocks in EURO-West produce strong responses across EURO-West, NORD, and CESEE aggregates.
  - UK long-term interest rate shocks transmit strongly to EURO-West and NORD interest rates; weaker transmission to CESEE interest rates.
  - Inflation shocks in the euro area show weak pass-through to CESEE inflation.
  - Credit shocks in the euro area have limited propagation to CESEE credit growth.
- GFEVD used to quantify relative contributions of shocks to forecast error variance (rescaled GFEVD presented because GFEVD do not necessarily sum to one due to contemporaneous correlations).

### Policy-Relevant Implications (as presented in the paper)
- Close trade and financial linkages brought growth benefits to CESEE (real per capita income rose by 4 percent annually in 1995–2007 for CESEE), but also transmitted risks rapidly—capital flow reversals and parent-bank pressures in 2008 triggered sudden stops and deep crises in CESEE.
- With increasing size of some CESEE economies via income convergence, shocks originating in CESEE have increasingly notable impacts on Western European partners (notably central European economies’ shocks affecting euro area GDP).
- Combining trade and financial linkages in assessing regional spillovers is important for accurate representation of transmission channels; either dimension alone is incomplete.

*Source: _wp13194 - EXECUTIVE SUMMARY*

### 1. Spillover of Real Shocks: Shocks to Real GDP Growth

### 1. Spillover of Real Shocks: Shocks to Real GDP Growth

### A. Negative Shock to EURO-West Real GDP growth
- Experiment: a 1 percentage point negative shock to the EURO-West group’s real GDP growth.
- Generalized impulse response: immediate negative impact on growth across all countries and regions; oscillates and dissipates in about 12 quarters.
- Key quantitative responses:
  - CESEE countries: GDP growth drops by 0.65–1.25 percentage points (p.p.) in the same quarter.
  - Nordic countries: decline of about 0.5 p.p. in the same quarter.
  - ADV group: decline of about 0.5 p.p. in the same quarter.
- GFEVD findings (Table 9):
  - Shocks to variables in the EURO-West group together contribute over half of the rescaled total variance in the first four quarters.
  - Within EURO area variables, shock to real GDP growth is the dominant source of innovation; oil price (endogenous to euro area) is also an important source.
  - Within each country/region, shock to real GDP growth is the main source of innovation, though contributions of other variables rise over time.
- Note: a one s.d. shock is equivalent to about 0.2 percentage points (p.p.) in annualized quarter on quarter growth rate for the EURO-West group on impact.
- Country heterogeneity example: Poland’s response is a decline 0.4 p.p.; Slovakia and Lithuania around -1.5 to -1.6 p.p.

### B. Shock to Real GDP Growth in Nordic Countries
- Experiment: positive shock to real GDP growth in the Nordic countries.
- Regional profile: less severe regionwide than EURO-West shock, but strong transmission to the Baltics.
- Key quantitative responses:
  - Baltic countries: growth increases by 1 p.p. in the same quarter, rises to 1.5 p.p. in the third quarter, then declines.
  - EURO-West group: immediate effect of 0.2 p.p. increase, rising to about 0.3 p.p. in the next quarter.
  - Central Europe, Russia, Turkey: same-quarter impact around 0.15--0.2 p.p., rising further in the next 2-3 quarters.
  - ADV group: immediate effect only 0.1 p.p.
- GFEVD findings (Table 10):
  - Real GDP growth innovation in NORD is a main source of influence.
  - Other important sources: interest rate shocks in the ADV group, oil price shocks, and EURO-West output shocks.
- Note: a one s.d. shock is equivalent to 0.14 p.p. increase in growth rate on impact to the NORD group.

### C. Shock to Real GDP Growth in Central Europe (CE)
- Experiment: one p.p. shock to CE group real GDP growth (CE includes Czech R., Hungary, Poland, Slovakia, and Slovenia).
- CE own response: real GDP growth declines gradually and settles in about six quarters after the shock.
- Cross-regional impacts:
  - Euro-West group: 0.1 - 0.2 p.p. increase in growth in the first two quarters, dissipating quickly.
  - Nordic countries: 0.1 p.p. rise on impact, then declines.
  - ADV countries: similar profile to Nordics.
  - SE group: rise of below 0.15 p.p. on impact, 0.25 p.p. in second quarter.
  - Baltic countries: rise by 0.2 p.p. on impact and over 0.4 p.p. in the second quarter.
- GFEVD findings (Table 11):
  - CE real output growth is very sensitive to shocks to EURO-West output, oil price shocks, and shocks to ADV output and interest rate.
  - CE’s domestic inflation and output are main sources of domestic shocks.
- Note: a one s.d. shock to CE real GDP growth is equivalent to 0.2 p.p. increase in its own GDP growth rate on impact.

### D. Shock to Real GDP Growth in the Baltic countries
- Context: Baltic experience of boom-bust-recovery; close links to Nordic countries and Russia.
- Own-country impact: largest on Baltics’ own growth; persistent effects.
- Key quantitative responses:
  - After small dip in second quarter, growth increases: 0.6 p.p. in the third quarter, 0.4 p.p. in the fourth quarter, stabilizing around 0.4 p.p. in about 6 quarters.
  - Immediate bump in growth for Nordic countries: 0.02 p.p.; for Russia: 0.06 p.p.
  - Average of first six quarters growth impact on other countries: around 0.01 - 0.05 p.p.
- GFEVD findings (Table 12):
  - Major external sources of shocks to Baltic real GDP growth: EURO-West group’s output and the oil price (endogenous to EURO-West).
  - ADV group: important via interest rate shocks, credit, and output.
  - Smaller contributions from Nordic, CE, SE groups concentrated in output and inflation.
  - Russia’s shocks mainly from output.
- Note: the one s.d. shock to Baltic GDP results in a 0.67 p.p. increase in its real GDP growth rate on impact.

### E. Shock to Euro-West Credit Growth (Financial shock)
- Experiment: positive one percentage point (p.p.) shock to euro area credit growth.
- Regional credit response: limited impact on credit growth in other sub-regions; impact diminishes in two to three quarters.
- Key quantitative responses:
  - ADV: on impact, decline in credit growth rate below 0.6 p.p., quickly dissipates.
  - CESEE: generally weaker impact with declines appearing with some lags.
- Interpretation:
  - Credit growth responses are more country specific and do not follow common international dynamics.
- GFEVD findings (Table 13):
  - Innovations in EURO-West variables are main source of forecast variance for real credit growth (close to 90 percent).
  - Within EURO-West, innovation in interest rate to real credit growth dominates, followed by shocks to credit growth itself.
  - Credit shock in ADV group is also an important source.
  - In other countries, innovation in output growth is generally most important for forecast variance in credit growth.
- Note: for the euro area, the one s.d. shock is equivalent to about 0.26 p.p. increase in annualized quarter on quarter growth rate.

### F. Shock to Interest Rate in ADV (the UK, Switzerland, Iceland, and Israel)
- Experiment: one percent (100 basis points) increase in ADV group interest rate (UK dominates the group, ~80% of group GDP in PPP terms).
- Interest rate responses:
  - ADV group: after 100 bps increase, interest rate declines slightly by 25bps by the fifth quarter and continues to decline subsequently.
  - EURO-West: immediate increase of 10bps, rising to 60 bps by end of fourth quarter, then diminishing.
  - Nordic countries: initial 10bps rise, reaches 50 bps in the fourth quarter before diminishing.
  - CESEE: much muted, generally negligible movements.
- Real GDP growth responses to the ADV interest rate shock (q-o-q annualized rate) by the 6th quarter:
  - EURO-West group: around 1.7 p.p.
  - ADV group: 1.9 p.p.
  - NORD group: 3 p.p.
  - Baltics: 6 p.p.
  - Russia: 2 p.p.
  - SE: 1.5 p.p.
  - Turkey and CE: around 1.2 p.p.
- Interpretation: sudden increase in long-term interest rates can have significant output effects across regions.
- GFEVD findings (Table 14):
  - Main source of innovation: shocks to interest rate in the ADV group (mainly the UK), followed by shocks to inflation in the ADV group.
  - Oil price shock is another important source of innovation to ADV interest rate.
- Note: a one s.d. shock to interest rate is equivalent to 4.5 basis points (bps) rise in interest rate on impact.

### G. Negative Shock to Real Credit Growth in Central Europe
- Experiment: negative one p.p. shock to real credit growth in CE (negative one s.d. shock equivalent to 0.69 p.p. decline in credit growth for CE).
- CE real GDP growth responses:
  - On impact: 0.2 p.p. decline in real GDP growth rate in CE.
  - Second quarter: impact increases to -0.4 p.p., then dissipates with some oscillation.
- Cross-regional impacts:
  - Western Europe (Euro-West, ADV, Nordic): declines less than 0.05 p.p. on impact; decline increases in second quarter but offset in following quarters.
  - CESEE: Baltic countries and Southern European countries negatively affected with average impact around 0.3 p.p. in the first two quarters.
  - Russia and Turkey: small initial bump in growth rate, though Russia declines in the second quarter.
- GFEVD findings (Table 15):
  - CE credit growth’s own innovation dominates as the major source.
  - External important sources: EURO-West output shocks, oil price shocks, and ADV output shocks.
  - Baltic countries’ credit and inflation also notably influence CE credit growth.

### 3. Shock to EURO-West Inflation
- Experiment: one p.p. shock to EURO-West inflation.
- Regional inflation responses:
  - EURO-West group: inflationary impact declines sharply; effect diminishes in three to four quarters.
  - ADV group: on impact, inflation rate declines by 0.05 p.p.
  - Nordic countries: inflation hardly changes on impact.
  - CESEE economies: more varied impacts on impact:
    - Baltic countries: inflation increases by 0.4 p.p.
    - Southeastern Europe: inflation increases by 0.5 p.p.
- Interpretation: inflation spillovers are weaker than output spillovers, possibly due to market rigidities, differences in consumer baskets (CESEE countries tend to have a higher food component in the CPI), and other factors reducing inflationary pass-through.

*Source: _wp13194 - 1. Spillover of Real Shocks: Shocks to Real GDP Growth*

### 0.2 p.p. for central Europe, and it quickly diminishes afterwards.

### _wp13194 - 0.2 p.p. for central Europe, and it quickly diminishes afterwards.

### GVAR model design and focus
- Developed a Global VAR (GVAR) model to study cross-border linkages among countries in Europe.
- Model weights foreign variables using both trade and financial data to reflect close trade and financial linkages.
- Strong focus on CESEE countries: 12 CESEE countries included individually.
- Model used to study transmission of both real and financial shocks across European regions.

### Main cross-border transmission findings
- Output growth and long-term interest rates:
  - Strong co-movement in domestic output growth and long-term interest rates with foreign partners.
  - Shocks to the EURO-West group’s output growth are felt in strong ripples across Europe; output growth in other regions significantly affected on impact.
  - Shocks to long-term interest rate in the UK produce effects on long-term interest rates across Europe; the response is initially small but gradually rises.
  - Shocks to long-term interest rates have a rising and strong impact on growth over time, indicating that financial shocks affect real sector activity.
- Inflation and credit growth:
  - Co-movement in inflation and credit growth is weaker across countries.
  - Shocks to EURO-West group’s inflation have more subdued impacts on inflation elsewhere, indicating smaller inflation pass-through across borders.
  - Shocks to credit growth in the EURO-West group have relatively muted impact on credit growth in CESEE, reflecting funding patterns channeled through western European parent banks and less dependence on parent country credit growth.

### Regional heterogeneity and notable interdependencies
- CESEE countries:
  - Generally respond strongly to shocks originating from advanced Europe.
  - As CESEE economies grow, their impact on the rest of Europe is notable though still small; influence expected to grow as income convergence resumes.
- Nordic and Baltic clusters:
  - Nordic countries exert enormous influence on the Baltic countries: a shock to real GDP growth in the Nordic countries has a large impact on the Baltic countries and also affects EURO-West and elsewhere.
- Russia and Turkey:
  - Russia: an observed small but increasing decline in inflation in the quarters following an EURO-West inflation shock—possibly reflecting a strengthening of Russia currency if EURO-West inflation is driven by higher oil prices.
  - Turkey: the effect on inflation is a small increase of less than 0.2 p.p. on impact, which gradually rises in the following quarters.

### Key quantitative points and illustrative magnitudes
- One s.d. shock to euro area inflation is equivalent to a higher inflation rate of 0.36 percentage points (p.p.) in the euro area on impact.
- EURO-West group plays a dominant role in forecast-error variance of inflation and output:
  - From Table 16 (one s.d. shock to EURO-West inflation), EURO-West variance entries include: "EURO-West variance 62.8 32.3 36.0 36.1 36.0 37.7 37.5 37.4 38.3 38.5 38.8" (as reported in the table).
  - From Table 9 (negative one s.d. shock to EURO-West real GDP growth), EURO-West variance row shows: "EURO-West variance 62.8 62.5 57.8 58.3 56.3 55.3 54.8 54.3 53.8 53.5 53.3" (as reported in the table).
- Selected variance-decomposition highlights (preserving table values):
  - From Table 13 (one s.d. shock to EURO-West real credit growth): "EURO-West variance 90.8 81.9 80.8 81.5 84.2 84.8 84.8 85.1 86.1 86.6 86.9".
  - From Table 14 (one s.d. shock to interest rate in the ADV group): "ADV variance 86.5 83.9 84.3 84.5 84.5 84.3 82.6 81.2 79.8 78.3 76.7".
- Forecast-error variance shares indicate important roles for oil prices and ADV/EA variables in several regional variance decompositions (see tables for explicit variable-level percentages).

### Impulse response patterns (qualitative summary)
- Real GDP:
  - Negative one p.p. shock to EURO-West real GDP growth generates immediate declines in other regions with varying magnitude and persistence; responses are largest for ADV and CE in some horizons (see GIRFs and Table 9).
  - One p.p. shock to Nordic real GDP growth produces pronounced effects in Nordic and EURO-West regions and noticeable spillovers to ADV and Baltic groups (see GIRFs and Table 10).
  - One p.p. shock to Central European real GDP growth yields notable effects locally (CE) and spillovers to EURO-West and ADV (see GIRFs and Table 11).
  - One p.p. shock to Baltic real GDP growth mostly affects Baltic group and EURO-West to a lesser extent (see GIRFs and Table 12).
- Credit:
  - One p.p. shock to real credit growth in EURO-West generates strong persistence in EURO-West credit growth and limited spillovers to CESEE and other regions (see GIRFs and Table 13).
  - Negative one s.d. shock to real credit growth in Central European countries produces significant local variance contributions from CE variables and moderate spillovers elsewhere (see Table 15 and Figure 9).
- Interest rates:
  - One percent shock to interest rates in ADV group raises ADV rates significantly and transmits to other regions with gradually rising responses in some cases (see Figures 7–8 and Table 14).
- Inflation:
  - One p.p. shock to EURO-West inflation generates immediate and regionally heterogeneous inflation responses; EURO-West inflation variance has large contributions from oil price shocks and domestic inflation (see Figure 10 and Table 16).

### Interpretation and implications
- EURO-West group dominates as a source of shocks that explain a large share of forecast-error variance in output, inflation, interest rates, and credit across Europe.
- Financial shocks (long-term interest rate shocks) propagate to real activity with rising effects over time, underlining the importance of monitoring financial conditions for macroeconomic stability.
- Cross-border inflation pass-through is limited relative to output and interest-rate co-movement, implying domestic inflation dynamics and oil-price shocks play substantial roles.
- Funding patterns (western European parent banks channeling funds) help explain the muted transmission of EURO-West credit shocks to CESEE credit growth.

*Source: Author's calculations and tables/figures from the supplied IMF GVAR chapter.*

### References

### References

### Appendix I. GVAR model structure
- Model scope:
  - Assume N number of countries are to be modeled, and data such as real GDP, inflation, interest rates etc. are collected in x_i,t (a k_i vector of country specific variables) where i = 1,...,N, and t = 1,2,..., T.
  - For each x_i,t, the corresponding foreign variables x_i,t* (a k_i* vector of country specific foreign variables) can be constructed using weights ω_ij with ω_ii = 0 and sum_j ω_ij = 1 for j = 1,...,N.
  - Footnote: The foreign variables generally also include some global variables such as oil prices. (35)

- Country model specification:
  - A VAR(2, 2) model with foreign variables (VARX or VARX* model) allowing for deterministic trend is fitted for country i:
    - x_i,t = δ_i + Γ_i,1 t + Γ_i,2 t^2 + Φ_i,1 x_i,t-1 + Φ_i,2 x_i,t-2 + Λ_i,0 x_i,t* + Λ_i,1 x_i,t-1* + Λ_i,2 x_i,t-2* + ε_i,t
    - δ, Γ, and Λ are coefficients (matrices) to be estimated.
  - Interpretation: Foreign variables are treated as exogenous and explicitly included to capture inter-country linkages.

- Error-correction (ECM) representation:
  - The model can be rewritten in error correction terms:
    - Δx_i,t = α_i + β_i z_i,t-1 + Ψ_i,1 Δx_i,t-1 + ... + Π_i,0 Δx_i,t* + ... + ε_i,t
    - Define z_i,t = (x_i,t′, x_i,t*′)′ and α, β, γ (and relevant co-integration rank) are estimated.
  - Co-integration specifics:
    - If co-integration rank for country i is r_i between x_i and x_i*, then α_i is a k_i × r_i matrix of rank r_i and β_i is a (k_i + k_i*) × r_i matrix of rank r_i.
    - Partition β_i as β_i = (β_i,d′, β_i,f′)′ conformable to z_i,t.
    - The r_i error correction terms are β_i′ z_i,t-1 = β_i,d′ x_i,t-1 + β_i,f′ x_i,t-1*.

- Error assumption and cross-country shocks:
  - Error terms ε_i,t are assumed serially uncorrelated with zero mean and a non-singular covariance matrix Ω_ii = (σ_tt) where Ω_t0 ε_t = (ε_1,t, ..., ε_N,t).
  - In particular, ε_i,t ~ i.i.d. N(0, Ω_ii).
  - The model allows cross-country correlation of idiosyncratic shocks:
    - Cov(ε_i,t, ε_j,t) = Ω_ij ≠ 0, implying an additional channel of cross-country spillovers beyond the VARX structure.

- Flexibility of GVAR:
  - No requirement to impose uniform variable sets or identical lag structures across country-specific models (no need to impose same lag structure for the VARX(p, q)).
  - This flexibility helps capture diverse inter-country relationships.

### Appendix II. Results of unit root test and weak exogeneity tests (introductory note)
- Variable integration assumption:
  - Following PSW and DdPS, variables included in the country-specific models are assumed to be integrated of order one (I(1)).
  - Rationale: This permits the model to distinguish short-run and long-run relations through co-integration relations.
- Empirical step:
  - Unit root tests are carried out for the domestic and foreign variables; results are presented subsequently.

*Source: _wp13194 - References*

### Appendix Tables A2-A4. The results reported include both the conventional augmented

### _wp13194 - Appendix Tables A2-A4. The results reported include both the conventional augmented

### Unit-root test results (domestic variables)
- Tests used: augmented Dickey-Fuller (ADF) and weighted symmetric (WS) estimation of ADF-type regressions (Park and Fuller (1995)). Lag length selected by Akaike Information Criterion (AIC).
- Real GDP growth (dy):
  - "mostly I(1) or borderline I(0)/I(1) (e.g. for ADV, EURO, Lithuania, NORD, Poland), except for Slovakia which is I(0)."
  - Selected ADF and WS statistics (5% significance level) include (examples shown column-wise for country groups): dy (with trend) ADF: -3.5, -3.0, -2.7, -2.6, -3.1, -2.8, -1.6, -3.3, -2.8, -3.4, -2.9, -2.3, -2.9, -4.4, -3.5, -2.6.
  - Δ dy ADF and WS statistics show strong rejection in differences: Δ dy ADF examples: -2.9, -6.5, -5.1, -3.9, -5.5, -5.6, -6.0, -7.3, -4.7, -3.6, -3.2, -9.6, -5.9, -7.7, -6.2, -5.2.
- Inflation (π):
  - "Inflation for Croatia, Latvia, Lithuania, Poland, Romania, Russia, Slovakia, Slovenia, and Turkey is I(1), while ADV, Estonia, EURO, Hungary, NORD appears to be I(0) (some with a trend)."
  - Examples of π (with trend) ADF: -3.5, -4.3, -3.5, -4.0, -3.7, -6.8, -4.6, -2.3, -2.3, -4.7, -3.9, -2.3, -3.0, -3.5, -2.7, -3.3.
  - Δπ ADF and WS show stationarity in first differences: Δπ ADF examples: -2.9, -4.7, -6.1, -4.0, -6.2, -6.8, -6.3, -4.7, -4.1, -5.4, -6.4, -3.4, -7.0, -4.9, -5.6, -4.2.
- Interest rates (lr):
  - "Interest rates are I(1) except for Hungary which is borderline I(0)/I(1)."
  - Selected lr (with trend) ADF/WS show mixed values (e.g., lr (with trend) ADF: -3.5, -2.1, -3.0, -2.4, -3.5, -3.2, -4.6, -2.5, -2.1, -3.9).
- Credit growth (dcredit / dCR):
  - "Credit growth (dCR) are mostly I(1) except for Hungary, Poland, and Slovakia which appears to be I(0)."
  - dcredit (with trend) ADF examples: -3.5, -3.8, -5.8, -2.3, -0.5, -2.5, -5.5, -2.5, -1.7, -0.5, -3.6, -3.2, -3.2, -4.8, -0.8, -3.4.
  - Δ dcredit ADF examples: -2.9, -8.7, -3.6, -4.5, -7.3, -6.0, -5.7, -7.7, -6.3, -5.6, -7.8, -3.2, -6.2, -8.2, -2.4, -6.0.
- Notes:
  - WS statistics are weighted symmetric estimation of ADF-type regressions introduced by Park and Fuller (1995).
  - Critical values used are at the 5% significance level.

### Unit-root test results (foreign variables)
- Summary: "Results for foreign variables are mostly similar to domestic variables, but with some notable differences. For example, foreign inflation variables are I(0), except for Lithuania which is slightly more borderline I(0)/I(1). Interest rates for many countries (ADV, Croatia, the Czech Republic, Hungary, Poland, Romania, Russia, Slovakia, Slovenia, and Turkey) appear to be I(2), with the rest I(1). Credit growth is across the board I(1)."
- Selected foreign-variable statistics (5% level) include:
  - dy* Δ and Δ2 ADF and WS: Δ dy* ADF examples: -2.9, -5.3, -5.7, -3.4, -4.9, -5.5, -5.8, -3.7, -4.7, -5.2, -5.4, -5.3, -5.0, -5.2, -5.5, -5.5.
  - π* (with trend) ADF examples: -3.5, -3.8, -3.6, -3.6, -3.9, -3.8, -3.9, -3.9, -3.9, -3.8, -3.9, -3.8, -3.9, -3.9, -3.9, -3.8.
  - dcredit* (with trend) ADF examples: -3.5, -0.8, -1.1, -0.7, -2.1, -0.6, -1.0, -1.1, -1.3, -2.9, -1.0, -0.8, -1.2, -1.4, -1.0, -2.7.
  - Δ dcredit* ADF examples: -2.9, -7.1, -6.8, -6.1, -8.1, -7.1, -6.8, -5.9, -5.9, -2.8, -6.8, -6.6, -5.9, -6.0, -6.6, -6.3.
  - lr* tests show mixed ADF/WS values with some weakly stationary and some trending results.

### Weak exogeneity of foreign variables (Table A8)
- Concept: Foreign variables should be weakly exogenous with respect to long-run parameters of the error-correction model; weak exogeneity implies x does not affect x* in the long run ("long-run forcing").
- Test results (F test; Fcrit_0.05 shown) and significance at 5% indicated by §:
  - ADV: F(3,28) = 2.95; Fcrit_0.05 = 3.97; oil = 1.47; § denotes significance at 5% where applicable. (Table shows multiple numbers; exact reported row: "ADVF(3,28)2.953.971.474.51")
  - Czech Rep.: F(2,14) = 3.74; row: "Czech Rep.F(2,14)3.742.092.070.39"
  - Estonia: F(1,17) = 4.45; row: "EstoniaF(1,17)4.450.441.620.941.05"
  - EURO-West: F(2,10) = 4.10; row: "EURO-WestF(2,10)4.100.330.310.48"
  - Croatia: F(1,31) = 4.16; row: "CroatiaF(1,31)4.162.630.050.62"
  - Hungary: F(2,14) = 3.74; row: "HungaryF(2,14)3.742.831.860.52"
  - Lithuania: F(1,15) = 4.54; row: "LithuaniaF(1,15)4.540.751.850.011.354.56"
  - Latvia: F(1,17) = 4.45; row: "LatviaF(1,17)4.450.190.020.410.92"
  - NORD: F(3,11) = 3.59; row: "NORDF(3,11)3.590.030.570.101.02"
  - Poland: F(2,17) = 3.59; row: "PolandF(2,17)3.591.423.621.23"
  - Romania: F(2,14) = 3.74; row: "RomaniaF(2,14)3.740.440.100.80"
  - Russia: F(1,15) = 4.54; row: "RussiaF(1,15)4.543.930.020.00"
  - Slovakia: F(2,29) = 3.33; row: "SlovakiaF(2,29)3.330.050.050.03"
  - Slovenia: F(1,31) = 4.16; row: "SloveniaF(1,31)4.160.070.252.92"
  - Turkey: F(1,31) = 4.16; row: "TurkeyF(1,31)4.160.120.011.19"
- Overall conclusion stated: "With the exception of real GDP growth, oil price for ADV, all foreign variables appear to be exogenous."

### Weight matrices and changes (Tables A2a and A2b)
- Table A2a: Weight Matrix (average of weights for the period 2000-2004). Bilateral weights shown in columns and sum up to one. Weights are average annual weights for the period 2000-2004; example entries:
  - ADV column sample: 0.00 (ADV), 0.05 (Czech Rep.), 0.03 (Estonia), 0.63 (EURO), 0.03 (Croatia), 0.05 (Hungary), 0.07 (Lithuania), 0.07 (Latvia), 0.22 (NORD), 0.05 (Poland), 0.07 (Romania), 0.13 (Russia), 0.03 (Slovakia), 0.04 (Slovenia), 0.18 (Turkey).
  - EURO column sample: 0.92 (ADV), 0.79 (Czech Rep.), 0.20 (Estonia), 0.00 (EURO), 0.84 (Croatia), 0.80 (Hungary), 0.35 (Lithuania), 0.31 (Latvia), 0.70 (NORD), 0.77 (Poland), 0.74 (Romania), 0.61 (Russia), 0.68 (Slovakia), 0.78 (Slovenia), 0.68 (Turkey).
- Table A2b: Changes in Weight Matrix from 2000-2004 to 2005-2011 (changes larger than 1% in absolute terms highlighted in bold). Example entries:
  - ADV column changes: 0.00, -0.01, -0.01, -0.03, -0.01, -0.02, -0.04, -0.04, -0.01, 0.00, -0.02, -0.02, 0.00, -0.01, -0.01.
  - EURO column changes: -0.01, -0.01, -0.05, 0.00, 0.05, -0.03, -0.10, -0.08, -0.02, -0.02, 0.06, 0.03, 0.00, 0.01, -0.03.
- Note on calculation: Weights for a specific year are calculated based on the total of trade flow and BIS reporting banks' external position between countries for that year.

### Data sources (Table A1 excerpts)
- Note: "All data are sources from Haver Analytics, the series codes and information are from Haver Analytics database."
- Country-level series include CP I (Harmonized Consumer Price Index, SA, 2005=100, or country-specific CPI), GDP (SWDA, 2005=100), Interest Rate (Long-term Interest Rate: Rolling 12 Month Average (%)), and credit series (banking institutions' claims or deposit money banks' claims on private sector in local currency units as specified per country).
- Specific country examples listed in Appendix (selected):
  - Austria: HICP (SA, 2005=100); GDP (SWDA, 2005=100); Long-term Interest Rate: Rolling 12 Month Average (%); Bkg Insts: Claims on Oth Res Sectors in Cty (Mil.Euros).
  - Netherlands: Harmonized CPI (SA, 2005=100); Gross Domestic Product (SWDA, 2005=100); Long-term Interest Rate: Rolling 12 Month Average (%); Bkg Insts: Claims on Oth Res Sectors in Cty (Bil.Euros).
  - Russia: Consumer Price Index (SA, 2005=100); Gross Domestic Product (SA, Bil.Chn.2008.Rubles); Zero Coupon Yield Curve: 10-Year (AVG, %); Dep Corps {MFSM}: Claims on Private Sector (Bil.Rubles).
  - Turkey: HICP: Monetary Union Index: Consumer Prices (SA, 2005=100); Gross Domestic Product (SWDA, Thous.98.TL); Interest Rates: 12 Month Time Deposits (% p.a.); Dep Corps {MFSM}: Claims on Private Sector (Mil.New Liras).

### VARX orders, co-integration, and cointegration test results
- VARX model orders and number of cointegrating relations (Table A6 excerpt):
  - Example entries: ADV p q = 1 1 3; Czech Rep. p q = 3 2 2; Estonia 3 2 1; EURO-West 3 2 2; Hungary 3 2 2; NORD 3 2 3; Poland 3 1 2; etc.
  - Note: p and q are the orders of domestic and foreign variables respectively.
- Cointegration results (Table A7, Trace statistic at 5% significance; * denotes rejection at 5%):
  - r = 0 trace statistics (examples): ADV 137.4*, Czech Rep. 129.8*, Estonia 74.8*, EURO-West 197.0*, Croatia 71.8*, Hungary 122.1*, Lithuania 117.5*, Latvia 83.6*, NORD 159.2*, Poland 131.7*, Romania 137.2*, Russia 150.3*, Slovakia 130.6*, Slovenia 77.1*, Turkey 70.4*.
  - r = 1 trace statistics (examples): ADV 81.3*, Czech Rep. 70.9*, Estonia 36.3, EURO-West 117.1*, Croatia 29.1, Hungary 67.7*, Lithuania 46.9*, Latvia 30.2, NORD 88.8*, Poland 82.2*, Romania 65.3*, Russia 49.3, Slovakia 62.9*, Slovenia 35.3*, Turkey 41.8*.
  - r = 2 and r = 3 reported where applicable; selected r = 2 examples: ADV 35.3*, Czech Rep. 34.8*, EURO-West 60.5*, Poland 39.2*, Romania 46.1*.
- Note: 5% significance level based on MacKinnon, Haug, and Michelis (1999).

### Global variables and oil price unit-root tests (Table A5)
- poil tests (5% level):
  - poil (with trend) ADF: Critical Value -3.45; Statistic -2.3.
  - poil (with trend) WS: Critical Value -3.24; Statistic -2.6.
  - poil (no trend) ADF: Critical Value -2.89; Statistic -1.5.
  - poil (no trend) WS: Critical Value -2.55; Statistic -0.3.
  - Δpoil ADF: Critical Value -2.89; Statistic -6.2.
  - Δpoil WS: Critical Value -2.55; Statistic -6.2.
  - Δ2 poil ADF: Critical Value -2.89; Statistic -7.4.
  - Δ2 poil WS: Critical Value -2.55; Statistic -7.7.

### Impulse-response function notes (Appendix figures)
- GIRFs computed from estimated GVAR model (see paper).
- Figure A1: Generalized Impulse Response Function of Real GDP Growth to a negative one p.p. shock to Real GDP Growth in the Euro-West group.
  - Note: "One s.d. shock to Euro group real GDP growth is equivalent to about -0.2 percentage points in annualized quarter on quarter growth rate for the EURO group."
- Figure A2: Generalized Impulse Response Function of Real Credit Growth to a one p.p. shock to Real Credit Growth in the EURO-West group.
  - Note: "One s.d. shock to Euro group credit growth is equivalent to about 0.3 percentage points in annualized quarter on quarter growth rate for the Euro group."
- Figure A3: Generalized Impulse Response Function of Inflation to a one p.p. shock to Inflation in the EURO-West group.
  - (Figures show country-level GIRFs for CROATIA, CZECH R., ESTONIA, HUNGARY, LATVIA, LITHUANIA, POLAND, ROMANIA, SLOVAK R., SLOVENIA; axes labeled in percent and quarters.)

*Source: Appendix Tables A2–A4 and accompanying tables and figures in the supplied IMF appendix content.*

### 0.4 percentage points in the EURO group.

### _wp13194 - 0.4 percentage points in the EURO group.

### Generalized impulse responses: Interest rate shock in the ADV group (the UK, Switzerland, Iceland, and Israel)
- Figure A4: Generalized Impulse Response Function of Interest Rate to a One Percent Shock to Interest Rate in the ADV Group.
- Notes: GIRF calculated based on the estimated GVAR model, see paper.
- A one s.d. shock to interest rate in the ADV group is equivalent to 0.0 basis points (bps) rise in interest rate.
- Source: Author's calculations.

### Generalized impulse responses: Real GDP growth to an interest rate shock in the ADV group
- Figure A5: Generalized Impulse Response Function of Real GDP Growth to a One Percent Shock to Interest Rate in the ADV Group (the UK, Switzerland, Iceland, and Israel).
- Real GDP growth measured QoQ annualized, in percent.
- Notes: GIRF calculated based on the estimated GVAR model, see paper.
- A one s.d. shock to interest rate in the ADV group is equivalent to 0.0 basis points (bps) rise in interest rate.
- Source: Author's calculations.

### Generalized impulse responses: Real GDP growth to a one p.p. shock to Real GDP Growth in the Nordic countries
- Figure A6: Generalized Impulse Response Function of Real GDP Growth to a One p.p. Shock to Real GDP Growth in the Nordic countries.
- One s.d. shock to Nordic group real GDP growth is equivalent to 0.1 percentage points in quarater on quarter annualized growth rate.
- Real GDP growth measured QoQ annualized, in percent.
- Notes: GIRF calculated based on the estimated GVAR model, see paper.
- Source: Author's calculations.

### Generalized impulse responses: Real GDP growth to a one p.p. shock to Real GDP Growth in the Central European countries
- Figure A7: Generalized Impulse Response Function of Real GDP Growth to a One p.p. Shock to Real GDP Growth in the Central European countries (Czech R., Hungary, Poland, Slovakia, and Slovenia).
- One s.d. shock to CE group real GDP growth is equivalent to 0.2 percentage points in quarater on quarter annualized growth rate.
- Real GDP growth measured QoQ annualized, in percent.
- Notes: GIRF calculated based on the estimated GVAR model, see paper.
- Source: Author's calculations.

### Generalized impulse responses: Real GDP growth to a negative one p.p. shock to Real Credit Growth in the Central European countries
- Figure A8: Generalized Impulse Response Function of Real GDP Growth to a Negative One p.p Shock to Real CREDIT Growth in the Central European countries (Czech R., Hungary, Poland, Slovakia, and Slovenia).
- One s.d. shock to CE group real CREDIT growth is equivalent to -0.5 percentage points in quarater on quarter annualized growth rate.
- Real GDP growth measured QoQ annualized, in percent.
- Notes: GIRF calculated based on the estimated GVAR model, see paper.
- Source: Author's calculations.

### Generalized impulse responses: Real GDP growth to a one p.p. shock to Real GDP Growth in the Baltic countries
- Figure A9: Generalized Impulse Response Function of Real GDP Growth to a One p.p. Shock to Real GDP Growth in the Baltic countries (Estonia, Latvia, and Lithuania).
- One s.d. shock to BALTICS group real GDP growth is equivalent to 0.7 percentage points in quarater on quarter annualized growth rate.
- Real GDP growth measured QoQ annualized, in percent.
- Notes: GIRF calculated based on the estimated GVAR model, see paper.
- Source: Author's calculations.

*Source: Author's calculations.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13194.pdf_