## _wp13252

## Source details

**Canonical URL:** [_wp13252](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13252.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13252.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13252.pdf.json)

---

### Introduction
- Purpose: construct a financial conditions index (FCI) for Poland to explore the link between financial conditions and real economic activity.
- FCI composition: combines a wide range of high-frequency financial variables, including external variables, to assess aggregate financial conditions.
- Intended uses:
  - identify and assess macro-financial linkages;
  - provide historical perspective on tightness/looseness of financial conditions;
  - inform monetary policy by capturing channels of monetary transmission and gauging impact on the real economy;
  - capture external spillovers on domestic financial conditions and activity by including external financial variables;
  - potentially forecast near-term economic activity and serve as input in macroeconomic forecasting models.

### Methodologies
- Two complementary construction methods:
  - Vector auto-regression (VAR): expresses the FCI in terms of its contribution to near-term GDP growth; weights ω_j equal cumulative two-quarter impulse response of GDP growth to a one-unit shock to x_{j,t}, estimated from a recursive VAR including all financial variables, quarterly GDP growth, and the GDP deflator; structural identification via Cholesky decomposition.
  - Factor analysis (FA): statistical extraction of common factor F_t from X_t − μ = λ F_t + ε_t; factor regressed on current and lagged GDP growth and residual υ_t used as factor-based FCI capturing exogenous financial developments.
- VAR advantage: accounts for interdependent relationships among financial variables (direct and indirect channels).
- FA advantage: allows wider coverage of financial market variables; downside—common factor may be difficult to interpret economically.

### Key findings and contributions
- Correlation between VAR FCI and Factor FCI = 0.78.
- FCI–growth correlations (2004Q1–2012Q4):
  - VAR-based FCI contemporaneous correlation with annualized q/q growth = 0.56.
  - VAR-based FCI correlation with two-quarter-ahead growth = 0.67.
  - Factor-based FCI uncorrelated with contemporaneous growth by construction; correlation with four-quarter-ahead growth rate = 0.52.
- Major domestic contributors to the overall FCI:
  - the WIBOR rate;
  - bank lending standards;
  - the corporate loan spread.
- Important external variables, especially during the global financial crisis:
  - VIX volatility index;
  - EURIBOR-OIS spread.
- External variables have both direct impact on growth and indirect impact via domestic financial conditions.

### VAR-based FCI — Variables, identification, and impulse responses
- Domestic variables included:
  - Lending standards (Senior Loan Officer Opinion Survey, average across corporate loan categories), measured as percent balance.
  - 3-month WIBOR rate.
  - Corporate loan spread (spread between lending rate for non-financial corporations and policy interest rate).
  - 5-year government bond yield.
  - Real effective exchange rate (REER).
- External variables included:
  - EURIBOR-OIS spread.
  - VIX index.
- Cholesky ordering: VIX, EURIBOR-OIS, GDP growth, GDP deflator, lending standards, WIBOR, REER, corporate loan spread, government bond yield.
  - Assumptions: domestic financial conditions do not have contemporaneous effects on growth and inflation; domestic developments do not contemporaneously affect external variables.
- Selected impulse response and cumulative effects:
  - A 20 basis point increase in WIBOR → roughly -0.3 percentage points annualized QoQ growth in first quarter; accumulates to over 1½ percentage points after two years.
  - Tightening of average lending standards by one standard deviation (roughly 10 percentage points) → cumulatively reduces growth by 1¼ percentage points in two years.
  - Government bond yield increase and REER appreciation hurt growth with lags (two to three quarters).
- Correlations (preserved as presented; horizons t+1 through t+5):
  - FCI: t+1 0.56, t+2 0.66, t+3 0.67, t+4 0.55, t+5 0.31
  - Lending standards: 0.46, 0.40, 0.30, 0.10, 0.13
  - 3-mo WIBOR: -0.35, -0.46, -0.45, -0.32, -0.08
  - Real exchange rate: 0.03, -0.06, -0.10, -0.31, -0.26
  - Corporate loan spread: -0.37, -0.27, -0.26, -0.24, -0.25
  - Government 5-yr bond yield: -0.04, -0.26, -0.38, -0.36, -0.25
  - EURIBOR spread: -0.36, -0.46, -0.57, -0.64, -0.51
  - VIX: -0.37, -0.42, -0.50, -0.43, -0.40

### VAR-based FCI — Contributions and crisis dynamics
- Decomposition: contribution of each variable = cumulative contribution to growth after 2 quarters; computed for 2004Q1–2013Q1 (weights based on 2004Q1–2012Q4).
- Major domestic contributors over sample:
  - WIBOR, corporate loan spread, and lending standards most significant.
  - Between 2005Q2 and 2007Q3 these three together contributed on average almost 3 percentage points per quarter in annualized QoQ GDP growth (cumulative two-quarter impact).
  - NBP policy: policy rate reduced by 250bps between 2005Q2 and 2007Q1; WIBOR closely followed.
- Crisis period (2007Q4–2009Q2):
  - WIBOR, corporate loan spread and lending standards together contributed on average -3¾ percentage points per quarter to growth.
  - WIBOR increased by 90bps between 2008Q1 and 2008Q4.
  - Lending standards tightened dramatically; benchmark: lending standards tightened by about 86pp in 2009Q1.
  - NBP responded with policy rate cuts totaling -250bps between end-2008 and end-2009.
- External factor impacts:
  - VIX and EURIBOR-OIS spread together contributed on average -3¼ percentage points per quarter to Poland’s GDP growth between 2008Q3 and 2009Q1.
  - Forecast error variance decomposition: about 10 percent of error in Poland growth forecast eight quarters out is due to shocks to VIX and EURIBOR-OIS spread.
  - EURIBOR-OIS spread strongly influences WIBOR; VIX affects lending standards, corporate loan spread, REER (depreciation), and government bond yield but does not significantly affect WIBOR.
- Recent dynamics (up to March 2013):
  - Conditions favorable early-2010 to mid-2011 due mainly to low WIBOR and improved external conditions.
  - 2011Q3–2011Q4 deterioration due almost entirely to external factors (widening VIX and EURIBOR-OIS spread).
  - Aggregate financial conditions began easing again; NBP easing cycle began November 2012 with policy rates cut by cumulative 150bps by March 2013.

### Factor-based FCI — Composition and interpretation
- Variables included (domestic): lending standards, WIBOR, corporate loan rate, REER (y/y growth), WIG stock index (y/y growth).
- Variables included (external): VIX, EURIBOR-OIS spread, S&P 500 stock index.
- Inclusion criteria: potential impact on growth (from VAR) and correlation with common factor (at least 30 percent).
- Factor loadings:
  - High loadings (>80 percent) for S&P 500, VIX, EURIBOR-OIS spread.
  - Example interpretation: factor loading 0.9 for WIG implies common factor explains 90 percent of variance in stock prices.
- Directional associations:
  - Higher S&P 500, low VIX, low EURIBOR-OIS spread → better financial conditions (higher factor FCI).
  - Higher WIG, easy lending standards, REER appreciation → pull up factor FCI.
  - High WIBOR and corporate lending rate → associated with more adverse financial environment.
- Comparison with VAR-FCI:
  - FA model does not explicitly capture relationships between financial variables and GDP; factor may reflect equity market movements even if those do not matter for growth.

### Forecast evaluation — Methods and results
- Forecast equation: y_{t+h} = α + Σ_{i=1}^p β_i y_{t-i+1} + γ X_t + ε_{t+h}, where X_t denotes indicator (FCI or CLI). VAR-based FCI used as four-quarter moving average. Number of lags p = 3 (chosen by AIC).
- In-sample predictive tests (partial R-squared examples, h=2):
  - Real GDP: VAR FCI 0.48; Factor FCI 0.42; CLI 0.22.
  - Industrial production: VAR FCI 0.64; Factor FCI 0.54; CLI 0.21.
  - At h=4, factor-based FCI tends to dominate.
- Out-of-sample predictive tests (relative RMSE to AR model, forecast sample 2010Q1–2012Q4, estimation up to 2009Q4):
  - Real GDP relative RMSE:
    - VAR FCI 0.88 (h=2), 0.88 (h=4)
    - Factor FCI 0.98 (h=2), 0.81 (h=4)
    - CLI 0.72 (h=2), 1.09 (h=4)
  - Industrial production relative RMSE:
    - VAR FCI 0.94 (h=2), 1.42 (h=4)
    - Factor FCI 0.97 (h=2), 1.23 (h=4)
    - CLI 1.10 (h=2), 1.03 (h=4)
  - Autoregressive model remains superior for forecasting employment growth and the unemployment rate in the tested specifications.
- Conclusion from forecasts: VAR-based and factor-based FCIs generally outperform CLI and the AR model for two- and four-quarter-ahead GDP growth in the tested samples.

### Conclusion — Uses, drivers, and caveats
- Two alternative FCIs for Poland constructed using VAR and FA; VAR decomposition clarifies macro-financial linkages.
- Main drivers of aggregate financial conditions in Poland:
  - WIBOR rate, lending standards, and corporate loan spread due to strong co-movement with the business cycle.
  - External indicators (VIX, EURIBOR-OIS spread) important given Poland’s links to Europe and global funding conditions.
- Uses: FCI is useful as an analytical tool to inform monetary policy analysis and near-term growth forecasting.
- Caveat: FCI is conditioned on the static historical relationship between real and financial variables; if this relationship changes, the model would need to be reassessed.

*Source: _wp13252 (PDF chapter/section).*

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

### _wp13252 - References

### Introduction
- The paper presents a financial conditions index (FCI) for Poland to explore the link between financial conditions and real economic activity.
- The FCI combines a wide range of high-frequency financial variables, including external variables, to assess aggregate financial conditions in the economy.
- The FCI is intended to:
  - identify and assess macro-financial linkages;
  - provide a historical perspective in comparing the relative tightness and looseness of financial conditions;
  - be useful for the conduct of monetary policy by capturing important channels of monetary transmission and gauging potential impact of monetary policy decisions on the real economy;
  - capture external spillovers on domestic financial conditions and economic activity by including external financial variables;
  - potentially have forecasting power for near-term economic activity and serve as input in macroeconomic forecasting models.

### Methodologies
- Two complementary methods are used to construct the FCI for Poland:
  - Vector auto-regression (VAR): allows expressing the FCI in terms of its contribution to near-term GDP growth.
  - Factor analysis (FA): a purely statistical exercise aiming at extracting the “common factor” from a wide range of financial variables.
- The methods follow recent work on FCIs.2
- The FCI’s forecasting performance is evaluated against a composite leading indicator (CLI) developed by the OECD.3

### Key findings and contributions
- The FCI constructed by either VAR or FA is highly correlated with GDP growth.
- Among domestic financial variables, the following contribute most to the overall FCI:
  - the WIBOR rate;
  - bank lending standards;
  - the corporate loan spread.
- External variables are also important, particularly during the global financial crisis:
  - the VIX volatility index;
  - the EURIBOR-OIS spread.
- External variables have both a direct impact on growth and an indirect impact through their influence on domestic financial conditions.

### Forecasting evaluation
- In-sample and out-of-sample forecasting exercises indicate that the FCI can outperform the CLI in predicting near-term GDP growth.

### Tables and Figure (listed)
- Tables:
  - I. Correlations Between Financial Variables and Real Activity, 2004Q1–2012Q4
  - II. In-Sample Predictive Tests
  - III. Out-of-Sample Predictive Tests, Relative RMSE
- Figure:
  - I. Contributions to FCI, 2004Q1–2013Q1

*Source: _wp13252 - References (PDF chapter/section).*

### Section V concludes.

### Section V concludes.

### II. OVERVIEW OF METHODOLOGY — Vector Auto-Regression (VAR)
- FCI formulation: FCI_t = Σ_j ω_j (x_{j,t} − x̄_j), where x̄_j denotes the variable mean over the sample period (2004Q1–2012Q4).
- Financial variables enter the FCI as deviations from the mean, representing shocks at each point in time.
- Weight ω_j: cumulative two-quarter impulse response of GDP growth to a one-unit shock to x_{j,t}, estimated from a recursive VAR including all financial variables, quarterly GDP growth, and the GDP deflator.
- Structural identification: Cholesky decomposition; weights reflect the impact of “pure” structural shocks to the extent the recursive identification scheme is successful.
- Advantage: VAR accounts for interdependent relationships among financial variables (direct and indirect channels, e.g., monetary policy affects other financial market variables which in turn affect growth).

### II. OVERVIEW OF METHODOLOGY — Factor Analysis (FA)
- FA formulation: X_t − μ = λ F_t + ε_t, where F_t is the common (unobserved) factor and ε_t is orthogonal error.
- Common factor summarizes greatest common variation in financial variables chosen for potential impact on growth (from VAR) and based on factor loading (correlation with the common factor).
- To purge influence of past economic activity, regress common factor on current and lagged GDP growth: F_t(A(L)) y_t + υ_t; the residual υ_t is the factor-based FCI capturing exogenous financial developments that predict future activity.
- FA advantage: fewer constraints on number of variables included, enabling wider coverage of financial market developments; downside—factor may be difficult to interpret economically.

### III. A FINANCIAL CONDITIONS INDEX FOR POLAND — Overview of Constructed FCIs
- Correlation between measures:
  - Correlation(VAR FCI, Factor FCI) = 0.78.
- Interpretation:
  - Increasing index = easing financial conditions; decreasing index = tightening.
- Correlations with GDP growth (2004Q1–2012Q4):
  - VAR-based FCI contemporaneous correlation with annualized q/q growth = 0.56.
  - VAR-based FCI correlation with two-quarter-ahead growth = 0.67.
  - Factor-based FCI uncorrelated with contemporaneous growth by construction; correlation with four-quarter-ahead growth rate = 0.52.
- Example magnitudes (VAR-based, contribution to annualized q/q growth):
  - Between 2007Q4 and 2009Q3: average contribution = -5¼ percentage points per quarter.
  - At 2008Q4 trough: economy contracted by 1.6 percentage points (annualized QoQ), of which adverse financial conditions contributed almost -13 percentage points.
- VAR FCI measured in contribution-to-growth units (cumulative two-quarter impact).

### III. A FINANCIAL CONDITIONS INDEX FOR POLAND — VAR-Based FCI: Variables and Identification
- Domestic variables included:
  - Lending standards (Senior Loan Officer Opinion Survey, average across corporate loan categories), measured as percent balance.
  - 3-month WIBOR rate.
  - Corporate loan spread (spread between lending rate for non-financial corporations and policy interest rate).
  - 5-year government bond yield.
  - Real effective exchange rate (REER).
- External variables included:
  - EURIBOR-OIS spread.
  - VIX index.
- Cholesky ordering (assumptions): VIX, EURIBOR-OIS, GDP growth, GDP deflator, lending standards, WIBOR, REER, corporate loan spread, government bond yield.
  - Assumes domestic financial conditions do not have contemporaneous effects on growth and inflation; domestic developments do not contemporaneously affect external variables.
- Empirical findings from impulse responses and cumulative effects:
  - A 20 basis point increase in WIBOR → roughly -0.3 percentage points annualized QoQ growth in first quarter; accumulates to over 1½ percentage points after two years.
  - A tightening of average lending standards by one standard deviation (roughly 10 percentage points) → cumulatively reduces growth by 1¼ percentage points in two years.
  - Government bond yield increase and REER appreciation hurt growth with lags (two to three quarters), impacts not immediate.
- Correlations (Table-style summary preserved):
  - FCI: t+1 0.56, t+2 0.66, t+3 0.67, t+4 0.55, t+5 0.31
  - Lending standards: 0.46, 0.40, 0.30, 0.10, 0.13
  - 3-mo WIBOR: -0.35, -0.46, -0.45, -0.32, -0.08
  - Real exchange rate: 0.03, -0.06, -0.10, -0.31, -0.26
  - Corporate loan spread: -0.37, -0.27, -0.26, -0.24, -0.25
  - Government 5-yr bond yield: -0.04, -0.26, -0.38, -0.36, -0.25
  - EURIBOR spread: -0.36, -0.46, -0.57, -0.64, -0.51
  - VIX: -0.37, -0.42, -0.50, -0.43, -0.40

### III. A FINANCIAL CONDITIONS INDEX FOR POLAND — VAR-Based FCI: Contributions and Dynamics
- Decomposition approach: contribution of each variable = cumulative contribution to growth after 2 quarters; decomposition computed for 2004Q1–2013Q1 (2013Q1 weights based on 2004Q1–2012Q4).
- Major domestic contributors over sample:
  - WIBOR, corporate loan spread, and lending standards are most significant.
  - Between 2005Q2 and 2007Q3 these three together contributed on average almost 3 percentage points per quarter in annualized QoQ GDP growth (cumulative two-quarter impact).
  - NBP policy move: NBP reduced policy rate by 250bps between 2005Q2 and 2007Q1; WIBOR closely followed.
- Crisis period contributions:
  - Between 2007Q4 and 2009Q2, WIBOR, corporate loan spread and lending standards together contributed on average -3¾ percentage points per quarter to growth.
  - WIBOR increased by 90bps between 2008Q1 and 2008Q4, reflecting interbank market freeze.
  - Lending standards tightened dramatically; benchmark: lending standards tightened by about 86pp in 2009Q1.
  - NBP responded with policy rate cuts totaling -250bps between end-2008 and end-2009, aiding recovery.
- External factors:
  - VIX and EURIBOR-OIS spread together contributed on average -3¼ percentage points per quarter to Poland’s GDP growth between 2008Q3 and 2009Q1.
  - Forecast error variance decomposition: about 10 percent of error in Poland growth forecast eight quarters out is due to shocks to VIX and EURIBOR-OIS spread.
  - EURIBOR-OIS spread strongly influences WIBOR; VIX does not significantly affect WIBOR but affects lending standards, corporate loan spread, REER (depreciation), and government bond yield.
- Recent dynamics (up to March 2013):
  - Conditions favorable early-2010 to mid-2011 due mainly to low WIBOR and improved external conditions.
  - 2011Q3–2011Q4 deterioration due almost entirely to external factors (widening VIX and EURIBOR-OIS spread).
  - Aggregate financial conditions started to ease again; NBP easing cycle began November 2012 with policy rates cut by cumulative 150bps by March 2013.

### III. A FINANCIAL CONDITIONS INDEX FOR POLAND — Factor-Based FCI
- Variables included (domestic): lending standards, WIBOR, corporate loan rate, REER, WIG stock index.
- Variables included (external): VIX, EURIBOR-OIS spread, S&P 500 stock index.
- REER and stock price indices enter as year-on-year growth.
- Inclusion criteria: potential impact on growth (from VAR) and correlation with common factor (at least 30 percent).
- Factor loadings (illustrative ordering preserved):
  - High loadings (>80 percent) for S&P 500, VIX, EURIBOR-OIS spread.
  - Example interpretation: factor loading 0.9 for WIG implies common factor explains 90 percent of variance in stock prices.
- Directional associations:
  - Higher S&P 500, low VIX, low EURIBOR-OIS spread → better financial conditions (higher factor FCI).
  - Higher WIG, easy lending standards, REER appreciation → pull up factor FCI.
  - High WIBOR and corporate lending rate → associated with more adverse financial environment.
- Comparison with VAR-FCI:
  - FA model does not capture relationships between financial variables and GDP directly; factor may reflect equity market movements irrespective of whether stocks matter for growth.

### IV. FORECAST EVALUATION — Method and Results
- Forecast equation estimated: y_{t+h} = α + Σ_{i=1}^p β_i y_{t-i+1} + γ X_t + ε_{t+h}, where X_t denotes the indicator (FCI or CLI). VAR-based FCI used as four-quarter moving average to match year-on-year variables. Number of lags = 3 (chosen by AIC).
- In-sample predictive tests (partial R-squared and F-statistics):
  - VAR-based FCI improves predictive power more than CLI across horizons; factor-based FCI also improves forecasts.
  - Example partial R-squared (h=2):
    - Real GDP: VAR FCI 0.48; Factor FCI 0.42; CLI 0.22.
    - Industrial production: VAR FCI 0.64; Factor FCI 0.54; CLI 0.21.
  - At h=4, factor-based FCI tends to dominate (consistent with FA construction capturing exogenous developments).
- Out-of-sample predictive tests (RMSE relative to AR model, forecast sample 2010Q1–2012Q4, estimation up to 2009Q4):
  - VAR-based and factor-based FCIs perform better than CLI and the autoregressive model for two- and four-quarter-ahead GDP growth.
  - Relative RMSE (h=2, h=4) examples:
    - Real GDP: VAR FCI 0.88 (h=2), 0.88 (h=4); Factor FCI 0.98 (h=2), 0.81 (h=4); CLI 0.72 (h=2), 1.09 (h=4).
    - Industrial production: VAR FCI 0.94 (h=2), 1.42 (h=4); Factor FCI 0.97 (h=2), 1.23 (h=4); CLI 1.10 (h=2), 1.03 (h=4).
  - Autoregressive model remains superior for forecasting employment growth and the unemployment rate in the tested specifications.

### V. CONCLUSION — Key Findings and Caveats
- Construction:
  - Two alternative FCIs for Poland constructed using VAR and FA.
  - Decomposition of VAR-based FCI into contributions of individual financial indicators clarifies macro-financial linkages.
- Main drivers of aggregate financial conditions in Poland:
  - WIBOR rate, lending standards, and corporate loan spread play major roles due to strong co-movement with the business cycle.
  - External indicators (VIX, EURIBOR-OIS spread) are important, reflecting Poland’s links to Europe and global funding conditions.
- Uses:
  - FCI useful as an analytical tool to inform monetary policy analysis and near-term growth forecasting.
- Caveat:
  - FCI is conditioned on the static historical relationship between real and financial variables; if this relationship changes, the model would need to be reassessed.

*Source: _wp13252 - Section V concludes.*

### References

### _wp13252 - References

### References list

- Gumata, Nombulelo, Nir Klein, Eliphas Ndou, 2008, “A Financial Conditions Index for South Africa,” IMF Working Paper 08/161 (Washington: International Monetary Fund).

- Hatzius, Jan, Peter Hooper, Frederic Mishkin, Kermit L. Schoenholtz, and Mark W. Watson, 2010, “Financial Conditions Indexes: A Fresh Look after the Financial Crisis,” NBER Working Paper No.16150.

- Hofman, David, 2011, “A Financial Conditions Index for Russia,” IMF Country Report 11/295 (Washington: International Monetary Fund).

- Osorio, Carolina, Runchana Pongsaparn, D. Filiz Unsal, 2011, “A Quantitative Assessment of Financial Conditions in Asia,” IMF Working Paper 11/170  (Washington: International Monetary Fund).

- Swiston, Andrew, 2008, “A U.S. Financial Conditions Index: Putting Credit Where Credit is Due,” IMF Working Paper 08/161 (Washington: International Monetary Fund).

*Source: _wp13252 - References.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13252.pdf_