## wp17269

## Source details

**Canonical URL:** [wp17269](https://www.imf.org/-/media/files/publications/wp/2017/wp17269.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2017/wp17269.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2017/wp17269.pdf.json)

---

### Overall structure and objectives
- Objective: Investigate transmission of financial conditions to real economic activity in France and enhance accuracy of macroeconomic forecasts.
- Two-step analytic flow:
  - Step 1 (Section III): construct Financial Conditions Indices (FCIs) by selecting financial variables, building VAR-based FCIs, computing variable weights, performing Granger-causality tests, and illustrating evolution of FCIs versus macro variables.
  - Step 2 (Section IV): assess macro-financial linkages and quantify forecasting contribution of FCIs via forecasting economic variables with the quarterly FCI, evaluating in-sample and out-of-sample forecast performance, and monitoring activity within the quarter (nowcasting).
- Sample period: 2003:Q1–2015:Q4.

### Methodology: VAR-based, two-step approach
- Two-step VAR regression-based approach:
  - Step 1: Select financial variables consistent with economic theory and statistical criteria; VAR specifications always include the inflation rate as a control variable.
  - Step 2: Aggregate selected variables into an FCI using weights derived from historical co-movements and cumulative impulse response functions (IRFs) from the VARs; weights correspond to the average impact of a 1-unit shock on the macro variable over subsequent quarters.
- Key methodological choices:
  - Avoided PCA/factor methods and business-cycle or price purging to retain interpretability and enable scenario analysis.
  - FCIs normalized so that a 1-unit increase in the FCI corresponds to the impact of a 1 standard deviation increase in the long-term risk free rate (exception: Exports FCI normalized on the NFCs risk premium since long-term rate had very low impact on exports).
  - Weight of each financial variable defined as its average cumulative impact over eight quarters on the targeted variable’s growth in a VAR framework.

### Selection of financial variables and sector specificity
- Candidate categories (max one variable per category in each FCI): (i) interest rates, (ii) spreads or “risk premium”, (iii) equity market returns, (iv) credit standards, (v) credit quantities, (vi) exchange rate.
- Inclusion criteria:
  - Coefficient sign consistent with economic intuition.
  - Inclusion must enhance forecast performance of the FCI for GDP or the targeted component.
- Final set in each tailored FCI: a long-term risk-free rate, a sector-specific spread, sector-specific lending standards, and equity market returns.
  - Long-term risk-free rate chosen over short-term rate for greater robustness.
  - Credit quantities (credit flows) and exchange rates were tested but excluded: credit quantities behaved as lagging indicators and were often incorrectly signed; Granger-causality tests show GDP FCI predicts credit growth 2 and 3 quarters ahead.

### Impulse responses and average cumulative effects on GDP growth (8-quarter horizon)
- Table 1 — Average cumulative effect, over 8 quarters (Shock → Effect on GDP (%)):
  - Long-term risk free rate: 100 bps increase → -0.36
  - NFCs spread: 100 bps increase → -0.90
  - Enterprises’ credit standards tightening: 10 pp increase → -0.04
  - Mortgage credit standards tightening: 10 pp increase → -0.43
  - Equity market returns: 1 standard deviation change → 0.44
- Interpretation:
  - Effects are comparable to those found in prior studies (Guichard and Turner (2008) for US activity).
  - Equity market returns and NFCs spread have larger economic impacts on GDP growth than changes in credit standards.

### Weights of financial variables in tailored FCIs (standardized; FCI normalization)
- Weights computed as average cumulative impact of a 1 standard deviation shock over 8 quarters; standardized so 1 unit FCI = effect of 1 standard deviation increase in long-term risk free rate (except Exports FCI).
- Table 2. Weights in the GDP-linked FCI:
  - Long-term risk free rate: Original weight -0.28; Weight in the FCI 1.00
  - NFCs spread: Original weight -0.86; Weight in the FCI 3.03
  - Enterprises’ credit standards tightening: Original weight -0.09; Weight in the FCI 0.31
  - Mortgage credit standards tightening: Original weight -0.50; Weight in the FCI 1.77
  - Equity market returns: Original weight 0.98; Weight in the FCI -3.46
- Table 3. Weights in the Consumption-linked FCI:
  - Long-term risk free rate: Original weight -0.29; Weight in the FCI 1.00
  - Mortgage spread: Original weight -0.36; Weight in the FCI 1.23
  - Mortgage credit standards tightening: Original weight -0.17; Weight in the FCI 0.57
  - Equity market returns: Original weight 0.66; Weight in the FCI -2.26
- Table 4. Weights in the Investment-linked FCI:
  - Long-term risk free rate: Original weight -0.61; Weight in the FCI 1.00
  - NFCs spread: Original weight -0.50; Weight in the FCI 0.81
  - Enterprises’ credit standards tightening: Original weight -0.29; Weight in the FCI 0.47
  - Equity market returns: Original weight 2.01; Weight in the FCI -3.28
- Table 5. Weights in the Exports-linked FCI:
  - NFCs spread: Original weight -1.51; Weight in the FCI 1.00
  - Enterprises’ credit standards tightening: Original weight -1.63; Weight in the FCI 1.08
  - Equity market returns: Original weight 2.45; Weight in the FCI -1.62
  - Note: long-term interest rate excluded from Exports FCI due to very low impact on exports growth.

### Component-specific compositions and transmission channels
- Consumption FCI tightening when:
  - equity market returns decrease;
  - mortgage risk premium increases;
  - long-term risk free rate increases;
  - mortgage credit standards tighten (lesser extent).
- Investment FCI driven mostly by equity market returns and to a lesser extent by long-term risk free rate and NFCs’ risk premium; enterprises’ credit standards tightening has a small negative impact.
- Exports FCI driven more equally by equity market returns, NFCs’ credit standards, and NFCs’ risk premium; NFCs’ credit standards significantly impact export growth.
- Core transmission channels captured: equity market return, long-term risk-free rate, sector-specific interest-rate premium (spread), and credit standards.
- Equity market returns are the strongest leading indicator across GDP and components, interpreted as investor sentiment/expectations rather than direct equity-financing effects.
- Credit volumes are lagging indicators of real activity.

### Granger-causality and predictive power (selected test statistics)
- Table A4 — Causality from Specific FCIs to Targeted Variables (chi2; df; p-value):
  - GDP-linked FCI → Current GDP growth: 21.3; df 4.0; p-value 0.000
  - GDP-linked FCI → 1 quarter ahead GDP growth: 37.6; df 2.0; p-value 0.000
  - GDP-linked FCI → 2 quarters ahead GDP growth: 25.1; df 2.0; p-value 0.000
  - GDP-linked FCI → 3 quarters ahead GDP growth: 14.3; df 2.0; p-value 0.001
  - Consumption-linked FCI → Current consumption growth: 0.3; df 2.0; p-value 0.874
  - Investment-linked FCI → Current Investment growth: 9.0; df 1.0; p-value 0.003
  - Investment-linked FCI → 3 quarters ahead investment growth: 10.1; df 2.0; p-value 0.006
  - Exports-linked FCI → 1 quarter ahead exports growth: 6.1; df 2.0; p-value 0.048
- Table A5 — Causality from the GDP-linked FCI to Credit Flows (chi2; df; p-value):
  - Current Credit flows: 4.398; df 2; p-value 0.111
  - 1 quarter ahead Credit flows: 4.012; df 2; p-value 0.135
  - 2 quarters ahead Credit flows: 7.245; df 2; p-value 0.027
  - 3 quarters ahead Credit flows: 8.820; df 2; p-value 0.012
- Interpretation:
  - GDP- and investment-FCIs Granger-cause their targeted variables at several horizons; consumption FCI lacks significant Granger-causality, consistent with contemporaneous movement.
  - GDP-linked FCI predicts credit flows 2 and 3 quarters ahead.

### Forecasting performance: in-sample RMSE ratios (Period 2003:Q1–2016:Q2)
- RMSE ratios (model RMSE relative to benchmark RMSE):
  - GDP growth — Standard FCI augmented model: current quarter 0.51, 1Q-ahead 0.63, 2Q-ahead 0.67, 3Q-ahead 0.75, Average 0.64.
  - GDP growth — Purged FCI augmented model: 0.60, 0.70, 0.72, 0.70, Average 0.68.
  - GDP growth — IMF FSI augmented model: 0.97, 0.92, 0.85, 0.91, Average 0.91.
  - GDP growth — Short-term rate augmented model: 0.86, 1.00, 0.87, 0.88, Average 0.90.
  - Consumption growth — Standard FCI augmented model: 0.95, 0.81, 0.78, 0.75, Average 0.82.
  - Consumption growth — Purged FCI augmented model: 0.93, 0.83, 0.76, 0.76, Average 0.82.
  - Consumption growth — IMF FSI augmented model: 1.01, 0.98, 0.98, 0.98, Average 0.99.
  - Investment growth — Standard FCI augmented model: 0.94, 0.82, 0.83, 0.83, Average 0.86.
  - Investment growth — Purged FCI augmented model: 0.83, 0.79, 0.88, 0.92, Average 0.86.
  - Investment growth — IMF FSI augmented model: 1.07, 0.96, 0.88, 0.90, Average 0.95.
  - Exports growth — Standard FCI augmented model: 0.88, 0.71, 0.81, 0.83, Average 0.81.
  - Exports growth — Purged FCI augmented model: 0.88, 0.78, 0.86, 0.90, Average 0.86.
  - Exports growth — IMF FSI augmented model: 0.99, 0.94, 0.93, 0.91, Average 0.94.
- Robustness and comparisons:
  - Purged FCIs do not systematically outperform standard FCIs; purged better at some horizons for GDP, consumption, investment, worse for exports.
  - IMF’s Financial Stress Index performs worse than tailored FCIs at all horizons.
  - Tailored FCI outperforms a model including a short-term interest rate for GDP at all horizons.

### Out-of-sample illustrative performance (2015:Q4–2016:Q4)
- Average absolute one-quarter-ahead forecast error for quarterly real GDP:
  - Benchmark model: 0.5 percentage points.
  - Model augmented with FCI: 0.18 percentage point.
- Note: comprehensive out-of-sample evaluation constrained by short estimation timespan and quarterly data frequency.

### Nowcasting: monthly FCIs and within-quarter monitoring
- Monthly FCIs constructed using same variables and weights as quarterly FCIs but at monthly frequency.
- Nowcast approach: three models per quarter corresponding to data availability after month 1, month 2, and month 3.
- Nowcast RMSEs relative to benchmark model without FCI:
  - GDP growth — FCI augmented model: 1st month 0.94, 2nd month 0.95, 3rd month 0.99, average 0.96.
  - Consumption growth — FCI augmented model: 0.89, 0.88, 0.90, average 0.89.
  - Investment growth — FCI augmented model: 0.81, 0.86, 0.97, average 0.88.
  - Exports growth — FCI augmented model: 0.88, 0.91, 0.93, average 0.91.
- Nowcast findings:
  - FCIs improve within-quarter now-cast performance in-sample; added value larger for GDP components than for GDP.
  - Financial conditions help more at the beginning of the quarter (first and second month models).
  - Tightening of financial conditions negatively impacts targeted real variable growth; magnitude and significance tend to decline over the quarter.
  - For investment, FCI-augmented early-month models can achieve explanatory power comparable to or better than later-month benchmark models.

### Nowcast model fit statistics (selected)
- GDP Nowcast models: Adjusted R2 examples across monthly models: 0.80, 0.83, 0.86, 0.88, 0.87, 0.89; Observations examples: 61, 49, 61, 50, 61, 50.
- Consumption Nowcast models: Adjusted R2 examples: 0.49, 0.57, 0.71, 0.78, 0.81, 0.88; Observations examples: 61, 49, 61, 50, 62, 50.
- Investment Nowcast models: Adjusted R2 examples: 0.73, 0.85, 0.80, 0.89, 0.84, 0.90; Observations examples: 61, 49, 61, 50, 61, 50.
- Exports Nowcast models: Adjusted R2 examples: 0.38, 0.52, 0.45, 0.54, 0.53, 0.58; Observations examples: 61, 49, 61, 50, 61, 50.
- Significance notation reported in tables: *** p<0.01, ** p<0.05, * p<0.1.

### Historical illustrations and episodes
- All four FCIs capture tightening during 2008 global financial crisis and 2011 euro area crisis; show loosening after 2012 associated with ECB non-standard measures and 2014 QE.
- 2008: abrupt tightening from 2008:Q1 to 2008:Q4 driven by sharp fall in equity returns and deterioration in credit standards; loosening from 2009:Q1 driven by recovering equity markets and lower risk premiums.
- 2011: tightening from mid-2011 largely driven by declines in equity returns and increasing risk premia.
- Early 2016: moderate tightening explained by temporary downturn of the equity market.

### Practical implications, conclusions and suggested next steps
- Practical implications:
  - Tailored FCIs enhance short-term forecast and nowcast accuracy for GDP and components in France, and facilitate scenario analysis and monitoring of transmission channels.
  - Despite France’s bank-based system, equity market returns provide the strongest short-term signal about economic prospects; credit spreads and credit standards remain important sectoral indicators.
- Conclusions:
  - Financial conditions, via tailored FCIs, are useful leading indicators for real activity in France, primarily transmitting through investment and, to a lesser extent, exports; consumption moves more contemporaneously with financial conditions.
  - Equity market returns are the dominant component across FCIs.
  - Credit volumes are lagging indicators, consistent with demand-driven credit growth.
- Suggested next steps:
  - Real-time performance assessment or expanded out-of-sample framework.
  - Scenario analysis: apply shocks to financial variables entering FCIs and trace impacts on consumption, investment, exports, and GDP.

*Source: wp17269 (IMF working paper content: Appendix and sections on FCI construction, forecasting and nowcasting performance).*

### INTRODUCTION  ________________________________________________________________  5

### INTRODUCTION

### Overall structure of the chapter
- II. HOW WE DEPART FROM THE LITERATURE AND WHY
- III. FIRST STEP: CONSTRUCTING THE FCIS
  - A. Selection of Financial Variables
  - B. VAR-based FCIs
  - C. Weights of the Financial Variables in the FCIs
  - D. Granger-Causality Tests
  - E. Illustration: Evolution of FCIs and Macroeconomic Variables
- IV. SECOND STEP: ASSESSING MACRO-FINANCIAL LINKAGES AND THE FORECASTING CONTRIBUTION OF THE FCIS
  - A. Forecasting Economic Variables with the Quarterly FCI
  - B. In-sample Quarterly Forecast Performance
  - C. Out-of-sample Quarterly Forecast Illustration
  - D. Monitoring Activity During the Quarter
- V. CONCLUSION
- REFERENCES
- APPENDIX
  - A. Data Description
  - B. Construction of the FCIs Linked to Components of GDP
  - C. Granger Causality Tests
  - D. Nowcast Models

### Key empirical components and diagnostics (as listed)
- Figures
  - 1. Responses of GDP Growth to Financial Variables
  - 2. Granger Causality Tests
  - 3. Financial Conditions Indices and the Real Economy
  - 4. One Quarter-ahead Forecast Error
- Tables
  - 1. Effect of Financial Shocks on GDP Growth Derived from Impulse Responses
  - 2. Weights of the Financial Variables in the GDP-linked FCI
  - 3. Weights of the Financial Variables in the Consumption-linked FCI
  - 4. Weights of the Financial Variables in the Investment-linked FCI
  - 5. Weights of the Financial Variables in the Exports-linked FCI
  - 6. GDP Forecast Models: Root Mean Square Errors Relative to the Benchmark Model without FCI
  - 7. GDP Components’ Forecast Models: Root Mean Square Errors Relative to the Benchmark Model without FCI
  - 8. Nowcast models: RMSEs Relative to Benchmark Model without FCI

### Analytic flow implied by section headings
- Step 1 (Section III): construct Financial Conditions Indices (FCIs) by
  - selecting financial variables,
  - building VAR-based FCIs,
  - computing variable weights in FCIs,
  - performing Granger-causality tests,
  - illustrating evolution of FCIs versus macro variables.
- Step 2 (Section IV): assess macro-financial linkages and quantify forecasting contribution of FCIs via
  - forecasting economic variables with the quarterly FCI,
  - evaluating in-sample and out-of-sample forecast performance,
  - monitoring activity within the quarter (nowcasting).
- Appendix provides data description, construction details for GDP-component-linked FCIs, Granger causality test details, and nowcast model specifications.

*Source: wp17269 - INTRODUCTION*

### APPENDIX TABLES

### wp17269 - APPENDIX TABLES

### Overview
- Objective: Investigate transmission of financial conditions to real economic activity in France and enhance accuracy of macroeconomic forecasts.
- Sample period covered: 2003:Q1–2015:Q4 (includes pre-Global Financial Crisis and post-Global Financial Crisis years).
- Main innovation: Two-step VAR regression-based approach to construct four tailored Financial Conditions Indices (FCIs) for GDP, investment, private consumption, and exports, linking weights to the sign, magnitude and timing of each financial variable’s impact on the targeted macro variable.

### Methodology
- Two-step VAR approach:
  - Step 1: Select financial variables consistent with economic theory and statistical criteria, testing VAR specifications that always include the inflation rate as a control variable.
  - Step 2: Aggregate selected variables into an FCI using weights derived from historical co-movements and cumulative impulse response functions (IRFs) from the VARs; weights correspond to the average impact of a 1-unit shock on the macro variable over subsequent quarters.
- Avoided data-intensive PCA/factor methods to retain interpretability and permit scenario analysis; did not apply business-cycle or price purging to the financial variables.
- Nowcast methodology:
  - Use blocking approach for mixed-frequency data (monthly series broken into first, second, third month-of-quarter blocks).
  - Estimate three models corresponding to availability of one, two or three months of data within a quarter to balance simplicity and accuracy.

### Selection of financial variables (design rules)
- Candidate categories considered (maximum one variable per category in each FCI): (i) interest rates, (ii) spreads or “risk premium”, (iii) equity market returns, (iv) credit standards, (v) credit quantities, (vi) exchange rate.
- Inclusion criteria:
  - Coefficient sign must be consistent with economic intuition.
  - Inclusion must enhance forecast performance of the FCI for GDP or the targeted component.
- Final set included in each tailored FCI: a long-term risk-free rate, a sector-specific spread, sector-specific lending standards, and equity market returns.
  - Long-term risk-free rate chosen over short-term risk-free rate because the latter’s coefficient was less robust.
  - Only one private-sector spread/risk premium included per FCI to avoid redundancy.
- Sector specificity:
  - Investment and exports FCIs: spread on loans to non-financial corporations (relative to the 10-year Treasury bond interest rate) and enterprises’ credit standards.
  - Consumption FCI: mortgage spread and mortgage credit standards.
- Excluded variables after testing:
  - Credit quantities (credit flows) — identified as lagging indicators of financial conditions and often incorrectly signed in predictive relationships; Granger-causality tests indicate the final GDP FCI (without quantity variables) contains information to predict credit growth 2 and 3 quarters ahead.
  - Exchange rates — not retained after testing.

### Key empirical findings and transmission channels
- Core transmission channels captured: equity market return, the long-term risk-free rate, a sector-specific interest-rate premium (spread), and credit standards.
- Leading-indicator properties:
  - Financial conditions are robust leading indicators primarily via their effect on enterprise investment, and to a lesser extent on exports.
  - Changes in financial conditions impact households’ consumption contemporaneously (possible channels: wealth effects, expectations, or common shocks on consumer and market sentiment).
- Relative importance of components:
  - Equity market returns are the strongest leading indicator across all three macroeconomic components and GDP itself, interpreted as capturing investor sentiment and market expectations of future economic activity rather than direct financing ease via equity markets.
  - Credit spreads are strong leading indicators particularly for investment and net exports.
  - Credit standards provide additional leading information on short-term economic prospects.
- Forecast horizons where FCIs help:
  - FCI-augmented VARs can improve predictions of future real GDP, consumption, investment and real exports one to three quarters ahead.
  - Monthly FCIs help improve high-frequency “now-cast” models particularly during the first month of the quarter.

### Forecasting and nowcast model integration
- VAR specifications:
  - FCI-augmented VARs take into account macroeconomic and price dynamics and include controls such as oil prices, world growth, the real effective exchange rate, or the change in unemployed labor force depending on the targeted macro component.
  - FCIs are treated as endogenous variables within VARs to capture feedbacks without imposing exogeneity.
- Nowcast applications:
  - Monthly nowcast models incorporate early-available financial information via the monthly FCIs.
  - The monthly FCIs could be used for scenario analysis by applying shocks to FCIs and assessing their impact through specific channels (consumption, investment, exports).
  - The chosen approach balances simplicity (three models for the three data-availability states within a quarter) and nowcast accuracy.

### Comparative and literature context
- Contrast with common approaches:
  - PCA/factor-based FCIs: incorporate large sets of variables but are harder to interpret; often use many variables (examples cited: 18, 45, 62, 100 variables in various studies).
  - Equally-weighted FCIs: simple but arbitrary weighting.
  - Regression-based/VAR-based FCIs (this paper): weights reflect the marginal contribution of financial variables to growth of targeted macro variables, offering interpretability and suitability for scenario analysis.
- On business-cycle purging:
  - The paper follows a standard VAR-based approach without purging financial variables of their endogenous macroeconomic components, noting mixed evidence in the literature on purged-vs-non-purged FCI forecast performance.

### Practical implications and potential uses
- Policy and forecasting:
  - Tailored FCIs can enhance short-term forecast and nowcast accuracy for GDP and its components in France.
  - FCIs enable monitoring of how changes in specific financial conditions translate into macroeconomic outcomes—useful for central banks, finance ministries, and forecasters in scenario analysis and stress testing.
- For France specifically:
  - Despite a bank-based financial system where bank lending predominates, equity market returns provide the strongest short-term signal about economic prospects.
  - Credit standards and spreads remain essential to capture sectoral financing conditions affecting investment and exports.

*Source: wp17269 - APPENDIX TABLES (PDF “wp17269 - APPENDIX TABLES”), material covering methodology, variable selection, construction and empirical findings on Financial Conditions Indices for France.*

### Appendix C).

### Appendix C)

### VAR-based FCIs: methodology and VAR specification
- The weight of each financial variable in the FCI is defined as its average impact over eight quarters on the targeted variable’s growth in a VAR framework.
- The VAR equation is estimated with X as a vector of jointly determined stationary variables: real growth of GDP, consumption, investment or exports, harmonized CPI inflation, and all selected financial variables included in the FCI.
- Ordering by relative sluggishness:
  - Real economy variables’ growth and inflation are ranked first and second.
  - Credit standards follow (banks unlikely to revise lending standards rapidly).
  - Risk free long-term interest rates ranked after credit standards (assumed to evolve less rapidly than private sector risk premiums and stock market returns).
  - Private-sector risk premiums and stock market returns are ranked last (expected to respond very quickly).
- Illustration: the GDP-linked VAR includes:
  - (i) long-term risk free rate (10-year Treasury bond interest rate);
  - (ii) NFCs risk premium (spreads for NFCs loans over 2 years, to 10-year Treasury bond);
  - (iii) change in enterprises’ and mortgage credit standards;
  - (iv) equity market returns (CAC40 total returns).
- Notes from experiments:
  - The long-term risk free rate performed better than slope-of-yield-curve alternatives in terms of sign.
  - French and European volatility indices (vcac and vstoxx) were tested but not retained because they did not raise in-sample performance.

### Impulse responses and cumulative effects on GDP growth
- Figure 1 shows impulse responses of GDP growth to each financial variable in the VAR; Table 1 reports average cumulative effects over 8 quarters.
- Table 1 — Average cumulative effect, over 8 quarters (Shock → Effect on GDP (%)):
  - Long-term risk free rate: 100 bps increase → -0.36
  - NFCs spread: 100 bps increase → -0.90
  - Enterprises’ credit standards tightening: 10 pp increase → -0.04
  - Mortgage credit standards tightening: 10 pp increase → -0.43
  - Equity market returns: 1 standard deviation change → 0.44
- The effects of selected financial variables on GDP growth are of the same order of magnitude as those found by Guichard and Turner (2008) for US activity.

### Weights of financial variables in the FCIs: computation and normalization
- Weights are computed as the average cumulative impact of a 1 standard deviation shock on the financial variables on GDP growth over the 8 following quarters.
- Weights are standardized such that an increase of 1 unit in the FCI corresponds to the impact of a 1 standard deviation increase in the long-term risk free rate.
- With this normalization, an increase in the FCI corresponds to a tightening of financial conditions and is expected to have a negative impact on the real macroeconomic variable considered.
- General findings on relative importance:
  - Equity market returns play the most economically significant role to anticipate GDP growth, followed by the spread on corporate loans (NFCs spread).
  - NFC spread captures cost of external finance for firms.
  - Credit standards’ tightening has a negative effect; statistically and economically significant for mortgages but not for lending to enterprises.
  - The composition and weights differ across FCIs (GDP, consumption, investment, exports), reflecting distinct channels.

### Component-specific FCI compositions and interpretations
- Consumption FCI:
  - Tightening (negative impact on consumption growth) when:
    - equity market returns decrease,
    - mortgage risk premium increases,
    - long-term risk free rate increases,
    - mortgage credit standards tighten (to a lesser extent).
  - Channels: confidence effects from equity returns; mortgage repayment cost and availability affect disposable income and access to housing-related consumption.
- Investment FCI:
  - Driven mostly by equity market returns and to a lesser extent by long-term risk free rate and NFCs’ risk premium (both reduce investment growth when they increase).
  - NFCs’ credit standards tightening has a negative but small impact on investment (interpretations: larger firms less affected; bond issuance financing).
- Exports FCI:
  - Driven more equally by equity market returns, NFCs’ credit standards, and NFCs’ risk premium.
  - NFCs’ credit standards significantly impact export growth (funding availability and cost affect firms’ export dynamics).
- Consumption moves more concurrently with its FCI; GDP, investment, and exports are anticipated by their tailored FCIs (see Granger-causality tests).

### Tables — standardized weights (Original impact; Standardized weight)
- Table 2. Weights in the GDP-linked FCI:
  - Long-term risk free rate: Original weight -0.28; Weight in the FCI 1.00
  - NFCs spread: Original weight -0.86; Weight in the FCI 3.03
  - Enterprises’ credit standards tightening: Original weight -0.09; Weight in the FCI 0.31
  - Mortgage credit standards tightening: Original weight -0.50; Weight in the FCI 1.77
  - Equity market returns: Original weight 0.98; Weight in the FCI -3.46
- Table 3. Weights in the Consumption-linked FCI:
  - Long-term risk free rate: Original weight -0.29; Weight in the FCI 1.00
  - Mortgage spread: Original weight -0.36; Weight in the FCI 1.23
  - Mortgage credit standards tightening: Original weight -0.17; Weight in the FCI 0.57
  - Equity market returns: Original weight 0.66; Weight in the FCI -2.26
- Table 4. Weights in the Investment-linked FCI:
  - Long-term risk free rate: Original weight -0.61; Weight in the FCI 1.00
  - NFCs spread: Original weight -0.50; Weight in the FCI 0.81
  - Enterprises’ credit standards tightening: Original weight -0.29; Weight in the FCI 0.47
  - Equity market returns: Original weight 2.01; Weight in the FCI -3.28
- Table 5. Weights in the Exports-linked FCI:
  - NFCs spread: Original weight -1.51; Weight in the FCI 1.00
  - Enterprises’ credit standards tightening: Original weight -1.63; Weight in the FCI 1.08
  - Equity market returns: Original weight 2.45; Weight in the FCI -1.62
  - Note: the long-term interest rate was not integrated into the exports-linked FCI because it had a very low impact on exports growth.

### Granger-causality and predictive power
- Granger-causality tests indicate:
  - GDP-FCI and investment-FCI movements precede respectively GDP and investment growth.
  - The FCIs also precede 1 quarter-ahead exports growth.
  - Consumption moves more concurrently with its FCI.
- Implication: the constructed FCIs provide statistically significant information that helps forecast future values of investment, GDP and one-quarter ahead exports.

*Appendix C).*

### Appendix C for the results of the tests). The finding that consumption reacts less to past

### Appendix C for the results of the tests

### E. Illustration: Evolution of FCIs and Macroeconomic Variables
- All four FCIs capture the tightening of financial conditions during the 2008 global financial crisis and the 2011 euro area crisis, and illustrate more recent evolution up to early 2016.
- 2008 global financial crisis:
  - Financial conditions were very loose before the crisis and tightened gradually due to increasing long-term rates and widening private sector spreads.
  - Tightening became abrupt from 2008:Q1 to 2008:Q4 driven by a sharp fall in equity returns and deterioration in credit standards.
  - Loosening from 2009:Q1, supported by recovering equity markets and a notable decrease in risk premiums, peaking at the end of 2009.
- 2011 euro area crisis:
  - Tightening before and intensification from mid-2011, largely driven by declines in equity returns and increasing risk premia; less so by tightening credit standards.
- Post-2012 and ECB measures:
  - After 2012, ECB announcements of non-standard measures and implementation of quantitative easing in 2014 were followed by loosening of financial conditions: relaxed credit standards, low long-term risk free rates, increased equity returns, and reduced corporate spreads.
  - Moderate tightening in early 2016 explained by temporary downturn of the equity market.
- Interpretations of consumption-FCI contemporaneity:
  - Consumption and the tailored FCI may be impacted by common unobserved shocks (consumer confidence and market sentiment).
  - Consumption could be directly and quickly impacted by financial conditions through wealth effects, though likely small given the small share of stocks in financial wealth in France.
  - Changes in financial conditions (notably stock prices) could cause immediate household revisions of expected future income and wealth, affecting consumption via intertemporal smoothing.

### Summary findings on transmission and components
- Financial conditions as measured by tailored FCIs are leading indicators for real GDP growth in France.
- Transmission to real activity primarily occurs through investment, and to a lesser extent via exports.
- Private consumption tends to move contemporaneously with its tailor-made FCI.
- Transmission occurs through:
  - Equity market returns (strongest effect across GDP and components),
  - Risk free rate,
  - Sector-specific risk premium,
  - Credit standards.
- Credit volumes appear to be a lagging indicator of real activity.
- Equity market returns give the strongest effect across GDP, consumption, investment, and exports.

### IV. SECOND STEP: Assessing macro-financial linkages and forecasting contribution of the FCIs
- Approach:
  - Augment standard forecast and nowcast models with constructed FCIs.
  - Four target variables: GDP, consumption, investment, exports.
  - Specification includes dependent variable lags, inflation, tailored FCI, and exogenous Z (differs by target).
  - Lag order chosen by information criteria; at most 4 lags for nowcasting current GDP and 2 lags for multi-quarter ahead forecasts.
- In-sample quarterly forecast performance:
  - All FCI-augmented models improve forecast performance versus benchmark without FCI at all horizons, especially for GDP.
  - RMSE ratios (model RMSE relative to benchmark RMSE) reported for Period 2003:Q1–2016:Q2:
    - GDP growth — Standard FCI augmented model: current quarter 0.51, 1Q-ahead 0.63, 2Q-ahead 0.67, 3Q-ahead 0.75, Average 0.64.
    - GDP growth — Purged FCI augmented model: 0.60, 0.70, 0.72, 0.70, Average 0.68.
    - GDP growth — IMF FSI augmented model: 0.97, 0.92, 0.85, 0.91, Average 0.91.
    - GDP growth — Short-term rate augmented model: 0.86, 1.00, 0.87, 0.88, Average 0.90.
    - Consumption growth — Standard FCI augmented model: 0.95, 0.81, 0.78, 0.75, Average 0.82.
    - Consumption growth — Purged FCI augmented model: 0.93, 0.83, 0.76, 0.76, Average 0.82.
    - Consumption growth — IMF FSI augmented model: 1.01, 0.98, 0.98, 0.98, Average 0.99.
    - Investment growth — Standard FCI augmented model: 0.94, 0.82, 0.83, 0.83, Average 0.86.
    - Investment growth — Purged FCI augmented model: 0.83, 0.79, 0.88, 0.92, Average 0.86.
    - Investment growth — IMF FSI augmented model: 1.07, 0.96, 0.88, 0.90, Average 0.95.
    - Exports growth — Standard FCI augmented model: 0.88, 0.71, 0.81, 0.83, Average 0.81.
    - Exports growth — Purged FCI augmented model: 0.88, 0.78, 0.86, 0.90, Average 0.86.
    - Exports growth — IMF FSI augmented model: 0.99, 0.94, 0.93, 0.91, Average 0.94.
- Robustness checks:
  - Purged FCIs do not perform significantly better than standard FCIs overall; purged versions perform better only at some horizons for GDP, consumption, and investment, and perform worse for exports.
  - Models augmented with IMF’s Financial Stress Index (FSI) perform much worse than those with tailored FCIs at all horizons.
  - GDP model with tailored FCI outperforms a simple macro-model including a short-term interest rate at all horizons.
- Historical performance:
  - FCI-augmented models explain real economy dynamics between 2006 and 2009 much better than benchmark models for GDP, investment, and exports.
  - FCI tailored for consumption enhances consumption forecasts at the beginning of the Euro area crisis in 2011.
- Out-of-sample illustrative performance (2015:Q4–2016:Q4):
  - Average absolute one-quarter-ahead forecast error for quarterly real GDP:
    - Benchmark model: 0.5 percentage points.
    - Model augmented with FCI: 0.18 percentage point.
  - Note: comprehensive out-of-sample analysis limited by short estimation timespan and quarterly data frequency.

### D. Monitoring activity during the quarter (Nowcasting)
- Construct monthly FCIs using same financial variables and weights as quarterly FCIs but at monthly frequency.
- Nowcast models estimate quarterly growth from monthly indicators and monthly FCIs; three models per quarter using data available after month 1, month 2, and month 3.
- Benchmark monthly indicators:
  - GDP: industrial production and households’ consumption expenditure of manufactured goods.
  - Investment: industrial production of investment goods and NFCs’ investment (quarterly variable).
  - Consumption: consumption expenditure.
  - Exports: exports volume.
- Nowcast RMSEs relative to benchmark model without FCI:
  - GDP growth — FCI augmented model: 1st month 0.94, 2nd month 0.95, 3rd month 0.99, average 0.96.
  - Consumption growth — FCI augmented model: 0.89, 0.88, 0.90, average 0.89.
  - Investment growth — FCI augmented model: 0.81, 0.86, 0.97, average 0.88.
  - Exports growth — FCI augmented model: 0.88, 0.91, 0.93, average 0.91.
- Nowcast findings:
  - FCI-augmented models improve within-quarter now-cast performance in-sample.
  - Added value higher for GDP components than for GDP (GDP benchmark without FCI already performs well).
  - Financial conditions improve nowcasts more at beginning of quarter (first and second month models) than at end of quarter.
  - Explanatory power is higher for each model augmented with an FCI compared to benchmark.
  - In all FCI models, tightening of financial conditions has a negative impact on targeted real variable growth; magnitude and significance decline over the quarter (especially for GDP and exports).
  - For investment, benchmark models of second and third months have lower explanatory power than augmented models of first and second months, implying FCI allows earlier accurate nowcasts of investment.

### V. Conclusion and next steps
- Financial conditions, via tailored FCIs, are useful leading indicators for real economic activity in France, improving accuracy of forecasts and nowcasts at quarterly and monthly frequencies for GDP and its components.
- Key transmission picture for France:
  - Stock market and credit spreads are strong leading indicators acting primarily through enterprise investment, and to a lesser extent through exports.
  - Household consumption is contemporaneously impacted by financial conditions (wealth effects, expectations, or common shocks on sentiment).
- Equity market returns are the strongest component across tailored FCIs, possibly reflecting equity prices as informative signals of near-term real activity and investment.
- Private risk premiums, interest rates, and credit standards are also significant, with varying relative weights across GDP and components.
- Credit volumes are lagging indicators, suggesting credit growth in France is demand-driven rather than supply-driven.
- Suggested next steps:
  - Real-time performance assessment or out-of-sample framework evaluation (limited so far by short data span).
  - Scenario analysis: apply shocks to financial variables entering FCIs and assess impact on real activity.

*Source: IMF working paper content (Appendix and sections on FCI construction, forecasting and nowcasting performance).*

### REFERENCES

### wp17269 - REFERENCES

### Key bibliographic sources
- References include working papers, IMF reports, central bank research, and academic articles on financial conditions indices (FCIs), financial stress, nowcasting, and forecasting. Notable entries:
  - Allard-Prigent C., Audenis C., Berger K., Carnot N., Duchene S., Pesin F., 2002. Présentation du modèle MESANGE, Modèle Économétrique de Simulation et d’Analyse Générale de l’Économie. Document de travail du Ministère de l’Économie, des Finances et de l’Industrie, No. 326640.
  - Altavilla, C., Darracq Paries, M., Nicoletti, G., 2015. Loan Supply, Credit Markets and the Euro Area Financial Crisis. European Central Bank Working Paper Series, No. 1861.
  - Hatzius, J., Hooper, P., Mishkin, F.S., Schoenholtz, K.L., Watson, M.W., 2010. Financial Conditions Indexes: A Fresh Look after the Financial Crisis. NBER Working Paper, No. 16150.
  - International Monetary Fund, 2017. “Are Countries Losing Control of Domestic Financial Conditions?”, Global Financial Stability Report, Chapter 3, Spring 2017.
- Additional empirical and methodological contributions span studies on FCI construction, factor models, financial stress indexes, short-term GDP forecasting using large monthly datasets, and country-specific analyses for France, Canada, the United States, and euro-area economies.

### A. Data description — variables and transformations
- Quarterly variables (financial and macroeconomic):
  - Financial variables (standardized): 10-year Treasury note interest rate — transformation: y-o-y difference; Spread on loans to NFCs over 2 years (to 10Y T-note interest rate) — y-o-y difference; Spread on new housing loans (to 10Y T-note interest rate) — y-o-y difference; Change in credit standards to enterprises (over the past quarter) — None (Bank Lending survey); Change in credit standards for mortgages (over the past quarter) — None (Bank Lending survey); CAC40 total return index — y-o-y growth.
  - Macroeconomic variables: Real GDP — y-o-y growth; Real private consumption (household and nonprofit institutions) — y-o-y growth; Real gross fixed capital formation — y-o-y growth; Real exports of goods and services — y-o-y growth; Harmonized consumer price index — y-o-y growth; Unemployed labor force — y-o-y growth; World average crude oil price — y-o-y growth; Real world GDP — y-o-y growth; Real effective exchange rate — y-o-y growth; Real gross fixed capital formation for NFCs (used in nowcast models for Investment) — y-o-y growth.
  - Sources include Haver Analytics (Bank of France), IMF WEO, IMF Information Notice System, INSEE Labor force survey.
- Monthly variables:
  - Financial variables (standardized): 10-year Treasury note interest rate — y-o-y difference; Spread on MFI new lending to domestic NFCs (to 10Y T-note interest rate) — y-o-y difference; Spread on new housing loans to domestic households (to 10Y T-note interest rate) — y-o-y difference; CAC40 total return index — y-o-y growth.
  - Macroeconomic variables: Manufacturing industrial production — y-o-y growth; Industrial production: investment goods — y-o-y growth; Household consumption expenditure: manufactured goods — y-o-y growth; Export volume: goods — y-o-y growth.

### B. Construction of FCIs linked to components of GDP — methodology and normalization
- VAR specification:
  - The VAR estimates a vector X of jointly determined stationary variables: real growth in the economic variable (GDP, C, I or E), harmonized CPI inflation, and all selected financial variables (previously transformed to avoid non-stationarity and standardized).
  - Lag length of the VAR is determined by information criterion lag-order selection statistics; if different criteria point to different lag lengths, the lag length that minimizes autocorrelation in residuals is chosen. Stability tests are conducted.
- Weight computation:
  - The weight of each financial variable in the FCI equals the average cumulative impact of a 1-unit shock of this variable on the real economic variable’s growth over the following 8 quarters, based on cumulative orthogonalized impulse response functions from the VAR.
  - The 8-quarter horizon is chosen to: i) capture the complete effect transmitted with lags; ii) align with the use of current and lagged FCI values to improve forecasts from current to three quarters ahead.
- Normalization:
  - FCIs are normalized such that a 1-unit increase in the FCI corresponds to the effect on the economic variable’s growth of a one-unit change in the long-term risk free rate included in the VAR.
  - Exception: Exports FCI is standardized based on the risk premium’s weight (the long-term interest rate was not integrated into the Exports FCI because it had a very low impact on exports growth).
- Table A3 (component inclusion summary):
  - Interest rate: Long-term risk free rate included for GDP, Investment, Consumption, Exports (note exception for Exports normalization).
  - Spread/risk premium: NFC spread (to long-term risk free rate) included for GDP, Investment, Exports; Mortgage spread (to long-term risk free rate) included for Consumption.
  - Equity market: CAC 40 returns included for GDP, Investment, Consumption, Exports.
  - Credit standards: Enterprise and mortgage bank lending standards (BLS) included in various FCIs as specified.

### C. Granger causality tests — results (Tables A4 and A5)
- Purpose: Tests indicate whether past values of target-specific FCIs help predict the targeted variable, given its past values. Tests executed using vargranger in Stata after fitting quarterly VARs.
- Table A4 — Causality from the Specific FCIs to the Targeted Variables (chi2, df, p-value):
  - GDP-linked FCI:
    - Current GDP growth: 21.3; df 4.0; p-value 0.000
    - 1 quarter ahead GDP growth: 37.6; df 2.0; p-value 0.000
    - 2 quarters ahead GDP growth: 25.1; df 2.0; p-value 0.000
    - 3 quarters ahead GDP growth: 14.3; df 2.0; p-value 0.001
  - Consumption-linked FCI:
    - Current consumption growth: 0.3; df 2.0; p-value 0.874
    - 1 quarter ahead consumption growth: 3.4; df 2.0; p-value 0.182
    - 2 quarters ahead consumption growth: 0.3; df 2.0; p-value 0.844
    - 3 quarters ahead consumption growth: 3.2; df 2.0; p-value 0.200
  - Investment-linked FCI:
    - Current Investment growth: 9.0; df 1.0; p-value 0.003
    - 1 quarter ahead investment growth: 4.5; df 2.0; p-value 0.108
    - 2 quarters ahead investment growth: 5.6; df 2.0; p-value 0.062
    - 3 quarters ahead investment growth: 10.1; df 2.0; p-value 0.006
  - Exports-linked FCI:
    - Current exports growth: 3.4; df 2.0; p-value 0.182
    - 1 quarter ahead exports growth: 6.1; df 2.0; p-value 0.048
    - 2 quarters ahead exports growth: 2.0; df 2.0; p-value 0.373
    - 3 quarters ahead exports growth: 3.4; df 2.0; p-value 0.186
- Table A5 — Causality from the GDP-linked FCI to Credit Flows (chi2, df, p-value):
  - Current Credit flows: 4.398; df 2; p-value 0.111
  - 1 quarter ahead Credit flows: 4.012; df 2; p-value 0.135
  - 2 quarters ahead Credit flows: 7.245; df 2; p-value 0.027
  - 3 quarters ahead Credit flows: 8.820; df 2; p-value 0.012
- Interpretation highlights:
  - GDP-linked FCI Granger-causes GDP growth at current and up to 3 quarters ahead with p-values 0.000, 0.000, 0.000, and 0.001 respectively.
  - Investment-linked FCI shows significant Granger-causality for current investment growth (p-value 0.003) and 3 quarters ahead (p-value 0.006); 2 quarters ahead is marginal (p-value 0.062).
  - Exports-linked FCI shows predictive power for 1 quarter ahead exports growth (p-value 0.048).
  - GDP-linked FCI predicts credit flows at 2 and 3 quarters ahead (p-values 0.027 and 0.012).

### D. Nowcast models — specification and selected statistics (Tables A6–A9)
- General notes:
  - Models are specified for first, second, and third month nowcasts within a quarter. Variable notation: X_i^m refers to the observation of variable X released right after the end of the i-th month of quarter; data availability lags determine which months’ observations are used in each monthly model.
  - Significance notation: *** p<0.01, ** p<0.05, * p<0.1.
- Table A6 — GDP Nowcast Model (selected coefficients and fit):
  - Key coefficients (examples across months): 0.225***, 0.226***, 0.115***, 0.123***, 0.150***, 0.159***; other coefficients include 0.273***, 0.212***, -0.0443*, 0.120***, 0.109**, 0.291***, 0.260***, -0.0339**, 0.0910***, 0.0842***, 0.230***, 0.226***, -0.00987.
  - Constants: 0.964***, 0.960***, 0.973***, 0.958***, 1.060***, 0.991***.
  - Observations: 61, 49, 61, 50, 61, 50 (across models).
  - Adjusted R2: 0.80, 0.83, 0.86, 0.88, 0.87, 0.89.
  - Notes: ܲܫ refers to manufacturing Industrial production; ݏ݊݋ܥ to Household consumption expenditure of manufactured goods.
- Table A7 — Consumption Nowcast Model (selected coefficients and fit):
  - Key coefficients (examples): 0.413***, 0.309***, 0.165***, 0.0739, -0.123***, 0.424***, 0.424***, 0.252***, 0.188***, -0.0911***, 0.327***, 0.329***, -0.0509***.
  - Constants: 0.852***, 0.846***, 0.662***, 0.681***, 0.707***, 0.663***.
  - Observations: 61, 49, 61, 50, 62, 50.
  - Adjusted R2: 0.49, 0.57, 0.71, 0.78, 0.81, 0.88.
  - Note: ݏ݊݋ܥ refers to Household consumption expenditure of manufactured goods.
- Table A8 — Investment Nowcast Model (selected coefficients and fit):
  - Key coefficients (examples): 0.415***, 0.481***, 0.329***, 0.446***, 0.308***, 0.417***, 0.217***, 0.153***, -0.303***, 0.305***, 0.213***, 0.166***, 0.126**, -0.265***, 0.186***, 0.150**, -0.164***.
  - Constants: 0.427**, 0.464**.
  - Observations: 61, 49, 61, 50, 61, 50.
  - Adjusted R2: 0.73, 0.85, 0.80, 0.89, 0.84, 0.90.
  - Notes: ݒ݊ܫ refers to previous quarter’s nonfinancial corporations’ investment; ܲܫ to industrial production in investment goods.
- Table A9 — Exports Nowcast Model (selected coefficients and fit):
  - Key coefficients (examples): 0.377***, 0.350***, 0.282***, 0.265***, 0.187**, 0.204**, -0.760***, 0.210***, 0.193***, 0.168**, 0.167**, -0.647**, 0.227**, 0.174**, -0.551**.
  - Constants: 2.244***, 1.853***, 2.070***, 1.828***, 1.924***, 1.811***.
  - Observations: 61, 49, 61, 50, 61, 50.
  - Adjusted R2: 0.38, 0.52, 0.45, 0.54, 0.53, 0.58.
  - Note: ݌ݔܧ refers to exports volume of goods; exports volume data is available with a two-month lag and FCI with a one-month lag, affecting which months’ data enter each monthly nowcast model.

*Content derived from "wp17269 - REFERENCES" (IMF PDF).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17269.pdf_