## _cr05185 ## Source details **Canonical URL:** [_cr05185](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2005/_cr05185.pdf) ## Other formats - [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2005/_cr05185.pdf.md) - [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2005/_cr05185.pdf.json) --- ### Stress testing methodology and scope - Stress tests undertaken as part of the France FSAP to assess resilience of the financial system to a variety of potential strains. - Value of stress testing: - Yields quantitative results that relate to exposures to market risk or credit risk. - Broadly comparable across risk factors and, to some extent, across countries. - More forward looking and more adaptable than financial soundness indicators. - Limitations: - Focuses on impact effects and ignores feedback effects and general equilibrium reactions. - Cannot estimate medium-term responses to large disturbances not closely corresponding to historical experience. - Two test categories: - Risk factor tests: sudden shifts in one or more variables, revaluing balance-sheet items. - Macroeconomic scenarios: two-year projections based on movements in exogenous variables. - Execution and reporting: - Tests performed by banks (using agreed parameters) and by authorities; results checked for plausibility. - Results reported in detail with variables such as change in value of NPLs and change in provisioning. ### Banking sector: sample, tests, and key findings - Sample and coverage: - 7 banking groups generating 60 percent of banking sector revenue and holding 80 percent of total assets in 2003. - Groups: Banques Populaires; BNP-Paribas; Caisses d‘Epargne; Crédit Agricole-Crédit Lyonnais; Crédit Mutuel-Crédit Industriel et Commercial; HSBC-Crédit Commercial de France; Société Générale. - Definition and calibration: - Market risk shocks benchmarked on largest shift in any 20 working days over past two decades. - Single-factor tests F1–F9 and macroeconomic scenarios M1–M4 defined (see individual test summaries below). - Baseline positions taken as end-2003. - Macroeconomic scenarios projected over two years. - Selected single-factor tests (exact calibrations): - F1: overnight rates rise by 150 basis points; 10-year bond yield +50 basis points. - F2: overnight rates fall by 50 basis points; 10-year yield unchanged. - F3a: sustained increase in all rates of 300 basis points over two years. - F4: deterioration in loan quality—domestic nongovernment sector down one rating grade; volatile sectors down two rating grades. - F5: fall in equity prices by 30 percent (one-month fall). - F6a/F6b: 15 percent appreciation/depreciation of the euro vs. the U.S. dollar and Japanese yen. - F7: emerging-market credit rating downgrades (two-notch on largest claims representing 50 percent of exposure; one-notch on remainder). - F8: 30 percent increase in implied volatility of all financial market prices. - F9: combination of F1 and F5. - Metrics used for factor tests: - (i) After-tax impact relative to 2003 after-tax profits. - (ii) Percentage change in end-2003 own funds implied by before-tax impact. - (iii) Change in end-2003 risk-weighted capital adequacy ratio implied by before-tax impact. - Tax rate applied: 33 percent (marginal tax rate on bank profits). - Supplemental banking tests and results: - Residential real estate lending: 36.2 percent of bank lending by end-2003 (from 31.1 percent in 2000). - Initial average CAR: 11.34 percent at end-2003. - Examples (Change in Basel-II type CAR): - 0.2 percent increase in PD in each risk bucket → Increase in risk-weighted housing loans 18 percent → Change in CAR -0.30 percentage points. - Shift in distribution raising average 0.2 percent → Increase 6 percent → Change in CAR -0.10 percentage points. - Downgrade in all credit risk ratings one notch → Increase 27 percent → Change in CAR -0.45 percentage points. - Sustained interest-rate rise (Mascotte model: +150 basis points year 1 and +150 basis points year 2) projects: - reduce real GDP growth by 0.4 percentage points; - decline in credit growth to enterprises from projected 7.1 percent to 1.7 percent; - Aggregated banks’ estimates: Effect on profits -9.2 (percent of 2003 profits); Effect on CAR -1.19 (percentage points). - Commission Bancaire’s estimates: Effect on profits -13.0; Effect on CAR -1.22. - Liquidity test: - Authorities’ liquidity requirement used; ratio of liquid assets to liquid liabilities must be maintained above 100 percent. - Banks report monthly global liquid assets and liabilities; items weighted by estimated rollover/availability probabilities. - Liquidity shortfalls were rare and concentrated among disproportionately small banks. - Key sectoral findings: - French banking system well placed to cope with likely shocks short to medium term. - Under no test does any bank’s CAR fall below the minimum 8 percent level. - Assumed shocks can be absorbed out of one year’s profits. - Credit risk is largest and most widespread impact; F4 yields results similar to macroeconomic scenario M1. - Combination tests can have effects larger than sum of single-factor tests (e.g., F9 > F1 + F5). - Exchange rate moves or yield-curve flattening can benefit some banks and harm others. - Negative effects on NPLs become more prominent only after the two-year projection period. ### Macroeconomic scenarios (M1–M4) — definitions and key values - Scenario definitions: - M1: Cumulative fall of 20 percent in world demand for French exports → mild two-year contraction; output path comparable to recession of 1993-94. - M2: Sustained 50 percent increase in oil prices (to US$40 per barrel) → effect on real GDP growth on order of 0.2 percentage points; inflation rises. - M3: Sustained 50 percent increase in oil prices with anti-inflationary policy response (ECB reaction) → output loss on order of 0.3 percentage points. - M4: Sustained depreciation of the U.S. dollar vs. the euro of 32 percent → Real GDP reduced by about 0.9 percentage points each year; inflation lower. - Key values (annual percentage rates except where indicated): - Baseline Real GDP growth: 2004 = 1.7; 2005 = 2.4 - M1 Real GDP growth: 2004 = -0.5; 2005 = -0.2 - M2 Real GDP growth: 2004 = 1.5; 2005 = 2.2 - M3 Real GDP growth: 2004 = 1.4; 2005 = 2.1 - M4 Real GDP growth: 2004 = 0.8; 2005 = 1.5 - Baseline Real growth in household consumption: 2004 = 1.6; 2005 = 2.2 - World demand for French products: Baseline 2004 = 6.4; 2005 = 7.8; M1 2004 = -7.4; 2005 = -12.2 - Short-term interest rate: Baseline 2004 = 2.0; 2005 = 2.2 - Long-term interest rate: Baseline 2004 = 4.6; 2005 = 5.1 - Consumer price inflation: Baseline 2004 = 1.4; 2005 = 0.9 - Oil price (US$ per barrel): Baseline 2004 = 27.0; 2005 = 27.0; M2/M3 2004 = 40.0; 2005 = 40.0 - Growth in credit: Baseline 2004 = 4.4; 2005 = 7.2; M1 2004 = 3.0; 2005 = 1.6 - Reporting metrics for scenarios: - After-tax 2003 profits; risk-weighted CAR under current regulations (allowing for projected credit growth and accumulation of after-tax profits); risk-weighted CAR under Basel II-type rules (accounts for changes in risk weights). - Basel II-type CAR change likely to be especially severe because of risk-weight changes. ### Banking stress test results — selected numerical excerpts - Risk factor tests (bank estimates, relative impacts as percent of 2003 after-tax profits): - F1 Best 0.8; F1 Average -3.8; F1 Worst -20.2. - F2 Best 1.4; F2 Average 0.6; F2 Worst -2.9. - F3a Average -35.7; F3a Worst -54.7. - Macroeconomic scenarios (banks’ estimates, relative to 2003 after-tax profits and CAR): - M1 Relative to 2003 after-tax profits: Best -1.17; Average -23.6; Worst -46.9. - M1 Relative to end-2003 CAR: -1.06 -1.42 -1.40 (Best-Average-Worst formatting in source). - M2 Relative to 2003 after-tax profits: Best 25.2; Average 2.0; Worst -10.4. - M3 Relative to 2003 after-tax profits: Best 34.6; Average 8.7; Worst -6.5. - M4 Relative to 2003 after-tax profits: Best 13.8; Average -4.4; Worst 20.1. - Commission Bancaire estimates (examples): - M1 Commission Bancaire Average -26.0; Relative to end-2003 CAR -1.09; Relative to Basel II-type CAR -2.10. - M2 Commission Bancaire Average 7.0; Relative to end-2003 CAR -0.79; Relative to Basel II-type CAR -0.11. - M3 Commission Bancaire Average 13.2; Relative to end-2003 CAR -0.80; Relative to Basel II-type CAR -0.05. - M4 Commission Bancaire Average -9.2; Relative to end-2003 CAR -0.73; Relative to Basel II-type CAR -0.83. - Risk-weighted assets projections (Table 6 examples, projected cumulative percentage change from 2003): - Baseline: With volume effect Year 1 0.0 Year 2 0.0; Without volume effect Year 1 6.8 Year 2 8.7. - M1: With volume effect Year 1 25.6 Year 2 20.2; Without volume effect Year 1 34.1 Year 2 30.7. - M2: With volume effect Year 1 1.6 Year 2 1.6; Without volume effect Year 1 8.5 Year 2 10.4. - M4: With volume effect Year 1 7.5 Year 2 6.4; Without volume effect Year 1 14.7 Year 2 15.7. ### Insurance sector: sample, methods, regulatory features, and results - Sample coverage: - Life insurance: 26 companies holding 79 percent of sectoral insurance liabilities (≈ €766 billion). - Composition: 20 subsidiaries of the 7 bank groups (45 percent of sectoral liabilities); one state-controlled company (10 percent); five individual companies each ≥ 2 percent of total liabilities. - Nonlife insurance: groups/entities representing 75 percent of nonlife insurance liabilities; includes 52 companies forming 11 nonlife groups or single companies with >1.5 percent of total nonlife liabilities. - Test design: - Shock effects decomposed into stock effect (instantaneous asset value change) and flow effect (progressive change in asset income cumulated over two years). - In tests with actual data, stock effect more important than flow effect. - Outcomes presented as solvency margin changes, pre/post-shock solvency ratios, pre/post-shock policy yields, and resource shortfalls as share of main insurance liabilities. - Reinsurance included in alternative estimates; most reported results exclude reinsurance for conservatism. - Life insurance regulatory framework and implications: - Solvency ratio: net admissible asset value divided by minimum required solvency margin; must remain above minimum. - Guaranteed rate of return: industry average guaranteed rate at end-2002 was 2.6 percent (3.7 percent at end-1995). - 85 percent mandatory pay-out rule: companies must pay at least 85 percent of asset income minus administrative overhead. - 125 percent minimum prudence margin rule: if realized return falls below 125 percent of guaranteed minimum return, liabilities recalculated using realized return. - Deferred insurance benefit provision: gains in excess of guaranteed minimum return may be distributed over eight years. - Unrealized losses on fixed instruments ignored for solvency purposes; unrealized gains added to capital; if policies cashed in, gains and losses realized immediately. - For shocks considered and assuming policies not cashed in, companies can absorb shocks by changing current/future policy returns or adjusting solvency margin. - Insurance-specific stress tests: - F3b: 100 basis point decrease in all bond yields. - F10: fall in French real estate prices by 30 percent. - F11: 100 basis point increase in all interest rates combined with 30 percent fall in equity prices. - F12: natural disaster causing damage twice as large as worst previous experience (1999); tested without and with reinsurance (F12a and F12b). - Insurance stress-test overview and results: - Aggregate industry resilience: all companies remain within mandated regulatory limits under all tests, though regulatory constraints would bind in some cases (Table 7). - Largest effects: tests involving increases in interest rates (F1, F3a, F11). - Equity and real estate price falls generate relatively minor losses; a very large natural disaster would severely affect some nonlife companies if reinsurance did not reimburse (F12). - Extreme run (all life policies cashed in immediately): - All policy-holders reimbursed in full except in test F3a. - Sector shareholders would lose 79 percent of their capital, but remaining assets would exceed insurance liabilities by 1.1 percent. - 7 of 26 companies would be unable to meet all obligations; worst-case shortfall 9.1 percent of main insurance liabilities. - Net present value (NPV) test (present value of assets minus liabilities as percent of main insurance liabilities): - Initial level: Average 16.4; Worst 7.6. - F1 Change: Average -3.2; Worst -4.7. - F3a Change: Average -8.9; Worst -13.7. - F11 Change: Average -6.9; Worst -9.6. - Interpretation: large rise in interest rates would leave some companies relatively weak; industry-wide long-term effects of reasonable shocks appear manageable. ### International linkages through the financial sector - France foreign assets and liabilities (market values, end of period, in billions of euros): - French claims on abroad: 1999 = 2,310.6; 2000 = 2,667.3; 2001 = 2,785.9; 2002 = 2,554.0. - French liabilities to abroad: 1999 = 2,233.1; 2000 = 2,458.4; 2001 = 2,545.1; 2002 = 2,371.7. - Net position (in billions of euros): 1999 = 77.5; 2000 = 208.9; 2001 = 240.8; 2002 = 182.3. - GDP (in billions of euros): 1999 = 1,355.1; 2000 = 1,420.1; 2001 = 1,475.6; 2002 = 1,520.8. - Shares (in percent of GDP) — total claims on abroad: 1999 = 170.5; 2000 = 187.8; 2001 = 188.8; 2002 = 167.9. - Banking sector foreign exposures (BIS consolidated claims, end period, in billions of euros): - Total consolidated claims: Dec-99 = 835.6; Dec-00 = 848.9; Dec-01 = 929.1; Dec-02 = 994.1; Dec-03 = 1,071.6. - Developed countries claims: Dec-03 = 911.3. - Europe: Dec-03 = 530.4; United States: Dec-03 = 247.5. - Resident banks’ claims on nonresidents (components, end period, in billions of euros): - Lending: Dec-97 = 236.9; Dec-01 = 384.7; Jun-03 = 397.7. - Bonds: Dec-97 = 73.5; Jun-03 = 200.7. - Shareholdings: Dec-97 = 15.1; Jun-03 = 38.8. - Counterparty composition: - Loans to nonresident financial institutions (in billions of euros): Dec-97 = 302.0; Jun-03 = 292.4. - Loans to nonfinancial nonresidents (in billions of euros): Dec-97 = 60.1; Jun-03 = 88.4. - Customer deposits from nonresidents (in billions of euros): Dec-97 = 49.1; Jun-03 = 74.9. - Interpretation: - Total consolidated claims over €1,000 billion, equivalent to about one quarter of banking system’s total assets. - Claims on other industrialized countries in Europe constitute one half of the total; claims on the U.S. amount to more than 20 percent. - Exposures to noninvestment grade emerging markets are relatively small and widely diversified. ### Foreign banks’ operations in France and French banks abroad - Consolidated borrowing from foreign banks — total claims (in billion euros): - Dec-99 = 460; Dec-00 = 554; Dec-01 = 612; Dec-02 = 606; Dec-03 = 612. - Foreign banks’ presence in France (selected): - Number of institutions: Dec-97 = 291; Jun-03 = 277. - Metropolitan employment: Dec-97 = 21,913; Dec-01 = 41,716; Jun-03 = 41,370. - Total assets (in billions of euros): Dec-97 = 305.1; Jun-03 = 459.1. - Foreign banks’ market share: ~10 percent of sectoral totals. - French banks’ foreign operations (Box 2 highlights): - BNP Paribas: only 40 percent of lending to France; about a third of leasing activities abroad; more than half of retail revenue/profits generated abroad. - Société Générale: international operations in 2003 contributed 11 percent of revenue and 13 percent of profits. - Crédit Agricole: about a quarter of revenues abroad. ### Insurance sector international footprint and reinsurance - Foreign ownership: - Foreign insurers own subsidiaries in France with ~20 percent market share in nonlife business. - Premium income from outside France: - For domestically-incorporated companies, premium income from outside France amounts to less than 2 percent of total premiums (mostly within EU). - Reinsurance: - 2002 total reinsurance premiums = €19.3 billion (just under 15 percent of premium income). - Less than half of €19.3 billion went to French reinsurers. - Foreign earnings of reinsurers in France in 2002 ≈ €1 billion, two-thirds earned by French companies. - Asset allocation: - Vast bulk of insurers’ assets are claims on industrialized countries, especially euro-denominated government bonds from OECD countries. - Unit-linked policies represented 18 percent of life insurance policies in 2003. - Life insurance assets composition in 2003: fixed-income 75 percent; equity 22 percent; real estate 3 percent. ### Capital market interdependence and multivariate regime-switching results - Equity market stylized facts: - Foreign participation in French government bond trading ≈ 40 percent. - Equity holdings by non-French residents ≈ 35 percent. - Correlations between national equity markets risen; average equity market correlation around 0.70 since early 2000s. - Correlations between French and other major equity markets rarely decline below 0.50. - Multivariate Markov-switching model for equity returns (weekly data Aug 5, 1987–Aug 4, 2004; indices CAC40, DAX, S&P 500, FTSE; AEX and BEL20 included in Euronext tests): - Estimation results (group 1: CAC40, S&P 500, FTSE, DAX; Sample Size 888; Log Likelihood 5966.3): - δ1(1) = 4.49* (T-statistic = 12.57) - δ2(1) = 3.33* (T-statistic = 11.37) - δ3(1) = 3.66* (T-statistic = 11.04) - δ4(1) = 4.96* (T-statistic = 12.25) - σ1 = 1.23* (T-statistic = 15.18); σ2 = 1.83*; σ3 = 1.50*; σ4 = 1.63*. - Q = 0.98*; P = 0.92*. - δi(1) > δi(0) ⇒ regime one volatility greater than regime zero. - Example: French equity variance regime one = 20.25; regime zero = 6.44. - CAC40–S&P 500 correlation: regime one = 0.85; regime zero = 0.42. - Test for change in interdependence controlling for volatility: Likelihood-Ratio = 39 (chi-squared with 3 degrees of freedom) ⇒ degree of interdependence increased over time. - Euronext test (CAC40, AEX, BEL20; Sample Size 706; Likelihood 3792.2): - Β1 = -0.01 (T-statistic = 0.11); Β2 = 0.13 (T-statistic = 1.44); Β3 = -0.30* (T-statistic = 5.10). - Only Belgian dummy estimate Β3 significant and negative ⇒ Belgian market moderately less correlated with Dutch and French markets since Euronext start; effect not judged economically important. - Bond markets (France, Germany, U.S., U.K.; Sample Size 887; Likelihood 2408.6): - δ1(1) = 0.64*; δ2(1) = 0.56*; δ3(1) = 0.61*; δ4(1) = 0.58*. - σ1 = 0.67*; σ2 = 0.79*; σ3 = 0.71*; σ4 = 0.26*. - P0 = 0.92*. - For German bond returns, loading factor δ4(0) = 1.58* > δ4(1), implying regime zero is the more volatile regime for Germany. - Test statistic for changing interdependence across bond returns = 144.2 ⇒ reject null of no change; bond markets have become more integrated over time. ### Monetary transmission: structural features and empirical findings - Channels identified: interest rate, exchange rate, asset price, credit (plus expectations). - French idiosyncrasies affecting transmission: - Administered interest rates (Livret A, PEL), ni-ni (no remuneration of demand deposits/no check charges), usury legislation — impede full pass-through from market rates to retail deposit and lending rates (Channel A). - Administered savings and tax-favored products reduce interest-rate sensitivity of consumption and investment (Channels K and L). - Centralization of deposits, earmarking, presence of CDC and La Poste, and public policy-directed investment reduce banks’ funding and asset flexibility. - Mortgage market features (high transaction costs, limited refinancing, predominance of fixed-rate mortgages) weaken housing/wealth channels. - Administered savings scale and composition (December 2003; amounts in billions of euros; percent of total deposits; percent of GDP): - Administered deposits: 528.7 — 49.3 percent of total deposits — 34.0 percent of GDP. - Livrets A: 112.0 — 10.4 percent — 7.2 percent of GDP. - PEL: 216.5 — 20.2 percent — 13.9 percent of GDP. - Of which centralized deposits: 217.0 — 20.2 percent — 14.0 percent of GDP. - Demand deposits: 342.1 — 31.9 percent — 22.0 percent of GDP. - Total deposits: 1,073.2 — 100.0 percent — 69.1 percent of GDP. - Livret A specifics: - Since July 2003 remuneration set automatically at average of inflation and ECB short-term interest rate, plus 25 basis points. - Adjustments every six months starting August 1, 2004. - Under new system a lag of between 2 and 3 months and 8 and 9 months remains before ECB policy changes are reflected. - Ni-ni requirement: - Banks cannot remunerate demand deposits nor charge for checks; majority of nonadministered deposits excluded from potential remuneration. - Combined with administered savings, >80 percent of deposits insensitive to market/policy rates in short run. - European Court of Justice ruling on October 5, 2004 found prohibition to remunerate demand deposits contrary to freedom of establishment; expected to mark end of ni-ni. - PEL features: - Largest administered savings by volume; requires monthly minimum saving; deposit and lending rates locked in at plan setup. - December 2002 reform eliminated interest-rate subsidy for plans not resulting in a loan for new plans; existing stock remains important. - Usury legislation: - Caps interest charged to individuals/overdrafts at four thirds of average observed rate in the loan category; quarterly backward-looking calculation by Banque de France. - Can be binding, especially at very low market rates, introducing delay in upward repricing when policy tightens. - Centralization (Fonds d’Epargne — FdE) and CDC: - €217 billion centralized in FdE at end-2003. - FdE assets end-2002 (in euros, Table 29): - Loans: 111.5 (49.2 percent of FdE assets) — housing-related loans 85.2 (37.6 percent). - Securities: 112.5 (49.6 percent) — public-sector securities 66.5 (29.3 percent); bonds and other fixed-income 40.8 (18.0 percent). - Total assets: 226.7. - FdE lending largely determined by public policy demand and relatively insensitive to interest rates; surplus funds invested in financial markets. - Centralization can reduce banks’ resource base and limits monetary transmission when administered rates do not adjust. - Banking sector balance-sheet contrasts (end-2002): - Noncentralized deposits available to banking system = 51 percent of GDP in France vs 91 percent in rest of euro-area. - Lending by French banks (excluding CDC) = 69 percent of GDP vs 101 percent for rest of euro-area banks. - Aggregate loan-to-deposit ratio France = 1.14; rest of euro-area = 1.11 (excluding CDC improves France to ~1.35). - In percent of total assets (Table 30): credit to euro-area non-MFI borrowers 27.8 percent in France vs 38.3 percent non-French euro-area banks. - Mortgage and housing market features (costs, enforcement, and effects): - Mortgage transaction and enforcement costs: - Notary honorarium between 0.66 percent and 3.9 percent of loan amount. - Tax 0.61 percent (waived for some loans). - Typical total costs between 1.5 percent and 5.0 percent of loan. - Early mainlevée d’hypothèque fee between 0.5 percent and 2.0 percent if property sold within two years after loan maturity. - Realizing a mortgage requires a court procedure routinely taking more than a year. - Refinance and product features: - 14 percent of mortgage loans variable rate (ECB (2003)). - France mortgage debt = 22 percent of GDP in 2001 (only Greece, Italy, Finland lower in euro-area). - High transaction costs and predominance of fixed rates reduce households’ sensitivity of mortgage costs to market rates. - Notary transaction cost examples (Table 32): for transaction amount €100,000 transaction costs €7,240; taxes €4,990 (7.2 percent). - Household financial structure (end-2002, in billions of euros): - Deposits: 865.7; Insurance claims: 835.1; Equity: 680.7; Mutual funds: 241.3; Credits liabilities: 587.3. - Financial net worth: 2,087.4. - Cross-country leverage (2000, selected): - France: Financial liabilities 54 (percent of gross disposable income); Financial wealth 308; Leverage 18. - Netherlands: 177; 583; 30. - United Kingdom: 116; 456; 26. - Composition of household financial assets (2000, percent): - Currency and deposits: France 30.4 percent. - Insurance claims: France 23.3 percent (unit-linked 4.9 percent). - Composition of household liabilities: - Mortgage debt: France 61.0 percent; Consumer credit 20.0 percent; Professional loans 19.0 percent. - Implication: low household leverage and asset composition imply monetary policy transmission via balance-sheet channels is weaker; consumption more sensitive to income effects than substitution effects. ### Empirical evidence on interest-rate transmission and consumption - Econometric approach: - Three-step evaluation for six credit market rates: long-term relationship, error-correction models, and simulation of 100 basis point policy rate hike under four Livret A adjustment assumptions. - Four Livret A adjustment scenarios: 1. No change. 2. Adjustment by 50 basis points with a 6-month delay (post-August 2003 system mimic). 3. Full adjustment with a six-month delay. 4. Full and instantaneous adjustment. - Simulation results (100 basis point shock to policy rate): - Consumer credit rates: - Additional reaction under full and immediate adjustment vs post-August 2003 system: - After one year: additional 26 to 44 basis points. - After two years: additional 31 to 79 basis points. - Effect on small consumer loans limited. - Credit to enterprises: - After one year: ranges from 75 to 121 basis points. - After two years: ranges from 89 to 129 basis points. - Contribution of Livret A non-adjustment to spread widening (2000Q3–2003Q3): - Rise in spread between Livret A and policy rate explains 143 basis points of the 236 basis point increase in the spread between the consumption rate and the policy rate over that period. - Long-term consumption equation (quarterly 1990–2003): - Real disposable income and spread between consumption credit rate and policy rate significant. - Level of policy rate not significant in long-term consumption equation. - Spread between consumption credit rate and policy rate negatively associated with consumption. - Static computation suggests up to 3¼ percentage points of potential consumption growth were forgone over the last three years because of slow adjustment of consumption credit rates. - Short- and long-term rate relationships (selected econometric diagnostics): - Long-term market credit rates related to nominal policy rate and administered-rate spreads in many specifications. - Medium- and long-term credit rates (including mortgage rates) strongly related to 10-year government bond rate; coefficients often close to or above 0.9 and frequently significant. - Unit-root tests (ADF statistics, selected): - Consumption (1979:01–2003:04): ADF without trend = -3.27. - Real disposable income: ADF without trend = -2.71. - Real policy rate: ADF without trend = -8.22. - Real Livret A rate: ADF without trend = -7.09. - Error-correction and Phillips-Loretan estimates (Appendices VI–VIII) report detailed cointegration and dynamic coefficients for overdraft rates, discount rates, mortgage fixed and variable rates, and term loans to enterprises (see source for full tables). ### Policy recommendations and reform options (as presented) - Improve administered-rate responsiveness: - Argue for instantaneous adjustment of the Livret A rate and other administered rates to changes in ECB rates; August 2003 formula (six-month adjustment, possible half-adjustment) helps but still hampers transmission. - Reassess PEL rate setting to enable more rapid, preferably automatic, adjustment to monetary environment and long-term market rates; eliminate/reduce absolute interest rate commitments in new PEL contracts (e.g., define commitments relative to market rates at loan origination). - Reassess government interventions that bias bank pricing: - Consider elimination of ni-ni requirement (checks free/zero interest bearing accounts quid pro quo). - Reassess usury legislation given potential interference with transmission during rapid policy-rate changes. - Improve mortgage market functioning: - Introduce mechanisms such as home equity loans and reforms to mortgage liens to enhance refinancing, reduce transaction costs, and increase rate sensitivity. - Broader objectives: - Aim for a better functioning mortgage market and reduce distortions from earmarking and centralization to enhance monetary transmission. - Consider reforms to reduce lock-in effects from tax-favored long-term savings and life insurance provisions to increase households’ ability to adjust portfolios. *Italic: Content derived from _cr05185 (France FSAP stress testing methodology, banking and insurance sector tests, international linkages, market interdependence, and monetary transmission analysis).* ### References.............................................................................................................. ### _cr05185 - References ### Stress testing methodology and scope - Stress tests were undertaken as part of the France FSAP to assess resilience of the financial system to a variety of potential strains. - Value of stress testing: - Yields quantitative results that relate to specific aspects of financial sector exposures, such as exposure to various forms of market risk or credit risk. - Broadly comparable across risk factors and, to some extent, even across countries. - More forward looking and more adaptable than an analysis of financial soundness indicators. - Limitations of stress testing: - Focuses on impact effects and ignores feedback effects through changes in behavior. - Does not involve estimation of effects in general equilibrium where financial institutions react over time by adjusting portfolios and operations. - Cannot provide estimates of how a financial system would respond over the medium term to large disturbances, especially those that do not correspond closely to historical experience. - Two broad categories of tests: - Risk factor tests: assume a relevant variable (e.g., exchange rate) shifts suddenly while all else remains the same; magnitude estimated by revaluing balance-sheet items. - Macroeconomic scenarios: based on assumptions about movements in exogenous variables that influence projections of major macroeconomic variables; involve a time dimension and were projected over two years for the France FSAP. - Methodological approach: - Tests exploited stress testing capacity available at financial institutions and supervisory authorities. - Performed both by individual institutions (using agreed parameters and scenarios) and at an aggregate level by authorities; results checked for consistency. - Results reported in detail to illuminate sector behavior and dispersion across institutions. - Several relevant variables were reported for each test (for example, change in value of NPLs and change in provisioning). ### Banking sector: sample and coverage - Participants comprised 7 banking groups, which together generated 60 percent of banking sector revenue and held 80 percent of total assets in 2003. - The groups were: - Banques Populaires - BNP-Paribas - Caisses d‘Epargne - Crédit Agricole-Crédit Lyonnais - Crédit Mutuel-Crédit Industriel et Commercial - HSBC-Crédit Commercial de France - Société Générale - Market structure notes: - Four groups are mutuals and two are incorporated French commercial banks. - HSBC-Crédit Commercial de France is much smaller than the others but included to represent a foreign-owned bank and a medium-sized bank. - High sectoral coverage was achieved, aided by domination of the French banking market by a number of large groups. - Banks’ own estimates were prepared by central units responsible for overall risk management, enabling a high level of technical sophistication. ### Definition of tests and procedures - Risk factor tests: - Provide estimates of the impact of a sudden change in one or more variables. - Impact estimated by calculating effect on end-2003 positions. - For changes in market prices or yields: relevant assets and liabilities are revalued and changes netted. - For changes in asset quality: impact estimated as the change in provisions required by the change in rating. - Single- and multiple-factor tests: - Multiple-factor tests consider several variables shifting at once; effects need not be additive because of hedging and nonlinearities (for example, derivatives introduce complex offsetting influences). - Execution: - Tests performed by the banks themselves based on specifications provided by authorities, allowing inclusion of off-balance-sheet items such as derivative positions. - Banks accounted for effects on themselves from losses accruing to their insurance subsidiaries, under the (extreme) assumption that policies are cashed in and asset prices fall very sharply, potentially requiring additional capital from parent banks. - Results from banks were checked by authorities for plausibility; in some instances banks were dropped from the sample when estimates appeared unreliable or too favorable. ### Metrics and interpretation of results - Impacts of shocks estimated as absolute amounts in euros; results presented relative to several metrics for interpretation. - For factor tests, three metrics used: - (i) After-tax impact relative to 2003 after-tax profits. - (ii) Percentage change in the end-2003 own funds implied by the before-tax impact. - (iii) Change in the end-2003 risk-weighted capital adequacy ratio implied by the before-tax impact. - Rationale for metrics: - After-tax impact relative to profits measures ability to absorb shock out of current income without capitalization being affected. - The before-tax approach is considered because in times of stress there are not necessarily profits from which taxes can be deducted, and because objective is to estimate capacity of banks to cope with instantaneous shocks without help from profits or tax credits. - The tax rate applied to the impact is 33 percent, which is the marginal tax rate on bank profits. - Impact on capital ratio measures effect on capitalization itself and helps assess whether any bank becomes under-capitalized even temporarily; useful for international comparisons. - Baseline for tests: - Starting point for all tests was institutions’ financial positions at end-2003, even if data were preliminary at the time tests were undertaken. - Time horizon for macroeconomic scenario projections: - Projections were made over two years for the France FSAP. - Reporting detail: - Results were reported with sufficient detail to show behavior of respective sector and dispersion across institutions; variables such as change in value of NPLs and change in provisioning were reported for each test to aid interpretation. *Italic: Content derived from _cr05185 - References (France FSAP stress testing methodology and banking sector tests).* ### 12. Because of the small number of banks in the sample and the legal requirement to ### _cr05185 - 12. Because of the small number of banks in the sample and the legal requirement to ### Definition and calibration of tests - Individual tests cover major sources of risk and are calibrated to correspond to the very largest shocks experienced in the relevant historical period. - For market risks (interest rates, equity prices, etc.), the benchmark is the largest shift experienced in any period of 20 working days (which corresponds to four calendar weeks) over the past two decades. - Magnitudes of shocks are comparable to those employed in FSAP stress testing for other major European countries. ### Single-factor tests (F1–F9) - F1: Flattening and upward shift in the yield curve: - overnight rates rise by 150 basis points - the 10-year bond yield increases by 50 basis points - intermediate rates rise by smoothly interpolated amounts - F2: Steepening of the yield curve: - overnight rates fall by 50 basis points - the 10-year bond yield is unchanged - intermediate rates smoothly interpolated - F3a: Sustained increase in all rates of 300 basis points over two years. - Direct effect on banks’ balance sheets not included in factor test results reported in the FSSA. - Effects on banks’ life insurance subsidiaries considered under extreme assumption that all life policies are cashed in. - F4: Deterioration in loan quality: - domestic nongovernment sector by one rating grade - volatile energy, commercial real estate, telecommunications, and transportation sectors by two rating grades - downgrade increases expected losses and provisioning - F5: Fall in equity prices by 30 percent in European, American, and Japanese markets (one-month fall). - Accounts for impact on banks of losses from their insurance subsidiaries - F6: Exchange-rate movements: - F6a: 15 percent appreciation of the euro against the U.S. dollar and the Japanese yen - F6b: 15 percent depreciation of the euro against the U.S. dollar and the Japanese yen - F7: Emerging-market credit rating downgrades: - two-notch downgrade for banks’ largest claims on emerging market countries (which represent 50 percent of banks’ total exposure to emerging markets) - one-notch downgrade applied to the remainder - F8: 30 percent increase in the implied volatility of all financial market prices (applies mostly to derivatives positions). - F9: Combination of F1 and F5. - This especially severe test does not correspond to historical experience from recent decades (central banks historically react to large equity price falls by lowering rates). ### Macroeconomic scenario tests - Scenarios use authorities’ large econometric models; projection period is two years starting at end-2003. - Scenarios are defined as deviations from a baseline projection; banks estimated effects using internal models (only one bank supplied supplementary projections). - Results reported using three metrics: - after-tax 2003 profits - risk-weighted capital adequacy ratio (CAR) under current regulations, allowing for projected credit growth and accumulation of after-tax profits - risk-weighted CAR calculated according to Basel II-type rules (accounts for changes in risk weights from increased variance/unexpected loss) - Change in the Basel II-type CAR is likely to be an especially severe test. ### Supplemental banking sector tests (Box 1) - Residential real estate: - Residential real estate lending reached 36.2 percent of bank lending by end-2003 (from 31.1 percent in 2000). - Authorities used a Basel II-type internal ratings based approach calibrated on historical experience. - Results (Change in Basel-II type CAR; initial average CAR was 11.34 percent at end-2003): - An 0.2 percent increase in the probability of default in each “risk bucket” → Increase in risk-weighted housing loans (percent) 18 → Change in Basel-II type CAR (percentage points) -0.30 - Shift in the distribution of default toward riskier “buckets” such that the average rises 0.2 percent → Increase in risk-weighted housing loans (percent) 6 → Change in Basel-II type CAR (percentage points) -0.10 - Downgrade in all credit risk ratings of one notch → Increase in risk-weighted housing loans (percent) 27 → Change in Basel-II type CAR (percentage points) -0.45 - Sustained increase in interest rates: - Mascotte model simulated a 150 basis point rise in all interest rates in the first year, and another 150 basis point rise in the second. - Model projects: - reduce real GDP growth by 0.4 percentage points - decline in growth in credit to enterprises from projected 7.1 percent to 1.7 percent - Effect on profits and CAR (percent of 2003 profits; percentage points change): - Aggregated banks’ estimates: Effect on profits -9.2; Effect on CAR -1.19 - Commission Bancaire’s estimates: Effect on profits -13.0; Effect on CAR -1.22 ### Procedures and estimation - Authorities’ projected paths of exogenous variables were changed per scenario, the models rerun, and new projections obtained. - Projections under scenarios were used in authorities’ models of bank profitability, sector-by-sector loan quality, and provisioning. - Models used include linear econometric models, nonlinear models of loan quality and transitions between credit ratings, and the SAABA “intelligent system” model of corporate finances. - Data sources included Quantitative Impact Studies (QIS) data and Banque de France detailed data on corporate finances and credit ratings. ### Defined macroeconomic scenarios (M1–M4) and key parameter values - M1: Cumulative fall of 20 percent in world demand for French exports → mild two-year contraction; output path comparable to recession of 1993-94. - M2: Sustained 50 percent increase in oil prices (from end-2003 level to US$40 per barrel) → effect on real GDP growth on order of 0.2 percentage points; inflation rises. - M3: Sustained 50 percent increase in oil prices with an anti-inflationary policy response (ECB policy reaction; transmission mainly through exchange rate) → output loss on order of 0.3 percentage points. - M4: Sustained depreciation of the U.S. dollar against the euro of 32 percent → Real GDP is reduced by about 0.9 percentage points each year; inflation lower. - Key values from Table 1 (annual percentage rates except where indicated): - Baseline scenario: Real GDP growth 2004 = 1.7; 2005 = 2.4 - Scenario M1: Real GDP growth 2004 = -0.5; 2005 = -0.2 - Scenario M2: Real GDP growth 2004 = 1.5; 2005 = 2.2 - Scenario M3: Real GDP growth 2004 = 1.4; 2005 = 2.1 - Scenario M4: Real GDP growth 2004 = 0.8; 2005 = 1.5 - Real growth in household consumption (selected): Baseline 2004 = 1.6; 2005 = 2.2 - World demand for French products (selected): Baseline 2004 = 6.4; 2005 = 7.8; M1 2004 = -7.4; 2005 = -12.2 - Short-term interest rate (selected): Baseline 2004 = 2.0; 2005 = 2.2 - Long-term interest rate (selected): Baseline 2004 = 4.6; 2005 = 5.1 - Consumer price inflation (selected): Baseline 2004 = 1.4; 2005 = 0.9 - Oil price (US$ per barrel): Baseline 2004 = 27.0; 2005 = 27.0; M2/M3 2004 = 40.0; 2005 = 40.0 - Growth in credit (selected): Baseline 2004 = 4.4; 2005 = 7.2; M1 2004 = 3.0; 2005 = 1.6 ### Liquidity test - Authorities’ existing liquidity requirement used rather than constructing a new test. - Banks report monthly data on global liquid assets and liabilities, including cash positions, claims (including repo-related claims with at most one month remaining maturity), negotiable securities, and off-balance-sheet commitments and available liquidity lines. - A ratio of liquid assets to liquid liabilities is constructed with items weighted to reflect estimated probabilities of rollover or availability in a liquidity squeeze. - Ratio must be maintained above 100 percent at all times. - Data allowed recording of number of banks with liquidity ratios below or close to 100 percent and their market share. - Account taken of distortions from exceptional circumstances (e.g., merger activity depressing ratio for transitory technical reasons). ### Results and key findings - The French banking system is well placed to cope with likely shocks over the short to medium term (Tables 2–6 referenced). - Under no test does the CAR of any bank fall below the minimum 8 percent level. - The impact of the assumed shocks can be absorbed out of one year’s profits. - Credit risk has the largest and most widespread impact; single-factor test F4 yields results similar to macroeconomic scenario M1. - No single sector dominates; transportation and commercial real estate lending risk are relatively important. - Combination of factors can have effects larger than the sum of single-factor tests (e.g., F9 vs F1 and F5). - Other risk factors may be important for individual banks (e.g., large equity price drop primarily affects one bank under F5). - Exchange rate movements (F6a and F6b) or flattening yield curve (F1) can benefit some banks and harm others. - Credit risk distribution is uneven across banks; a bank with low NPLs in a sector could still be worst affected by sectoral deterioration. - Macroeconomic scenarios: - Dollar depreciation (M4) has effect similar to large decline in external demand but milder. - Two scenarios with oil price rise (M2, M3) yielded very small or even positive effects on banks due in part to deposit-rate stickiness and yield-curve steepening from policy reaction. - Negative effects on NPLs become more prominent only after the two-year projection period. - Results cannot be extended simply to larger oil price shocks combined with larger policy reactions, which might have disproportionate and nonlinear impacts. *Source: _cr05185 - 12.* ### 26. The results of the macroeconomic scenario tests are predicated on the quality of the ### _cr05185 - 26. The results of the macroeconomic scenario tests are predicated on the quality of the ### Macroeconomic scenario tests and bank projection quality - Results are predicated on the quality of the projections for bank profitability, NPLs, etc., which "do indeed appear reasonable," as do the underlying estimated coefficients (in terms of magnitudes, standard errors, R2 statistic, etc.). - The starting point for the projections was the 2003 outcome, "which was well within the range of what has been typical for French banks in the recent past." - The estimation of the NPL equation was "econometrically most problematic" because during the 1990s banks worked off a large stock of NPLs acquired during the 1993-94 crisis, producing a distinct downward trend that appears to have leveled off in the most recent years. These estimates were performed mainly for illustrative purposes; stress test estimates were obtained using the more detailed and robust models available to the authorities. ### Comparison of bank-level and Commission Bancaire estimates - The effects estimated by individual banks broadly correspond to those obtained by the Commission Bancaire based on aggregate data, but: - Bank estimates tend to show more moderate effects (Tables 4 and 5). - The effects of a deterioration in the quality of loans to the commercial real estate sector estimated by the banks are "much smaller" than those estimated by the Commission Bancaire, but bank estimates of effects for other sectors are larger. - One possible explanation: banks may use a different sectoral classification than the Commission Bancaire. ### Basel II-type CAR and risk-weighted assets - Estimation of the effect on Basel II-type CAR shows that banks remain resilient even according to this more severe measure. - Changes in the Basel II-type CAR reflect both changes in capital and important changes in risk-weighted assets. - Not only does the stock of assets grow over time, but the risk weightings vary across scenarios; "The latter effect can be quite large (Table 6)." ### Impact of insurance subsidiary losses on banks - Losses in life insurance subsidiaries can be "sometimes significant, at least for individual institutions." - Those losses were estimated under the extreme assumption that insurance contracts are cashed in and "therefore, large capital losses must be realized immediately and in full." - Under more moderate assumptions, the banks "would not have to provide any additional capital for their insurance subsidiaries." ### Liquidity test results - The liquidity test shows that banks have "rarely failed to observe the required minimum ratio of liquid assets to liquid liabilities." - Those banks that do suffer occasional liquidity shortfalls are "disproportionately small banks." - Liquidity statistics reported (percent, 1999-2003): - Share of number of banks: Average 1.4, Standard error 0.7, Share of number of banks 7.9, Standard error 0.8 (table formatting in source). - Share of total bank assets 4/: Average 1.0, Standard error 1.2, Share of total bank assets 32.3, Standard error 8.1. - Note: 4/ Excluding effect of one large merger in 2000. ### Results from selected stress test tables (as reported) - Table 2 (Risk factor tests — bank estimates, relative impacts shown as percent of 2003 after-tax profits; relative to end-2003 own funds in percentage points; relative to end-2003 CAR in percentage points): - Example entries (banks' estimates, selected): F1 Best 0.8; F1 Average -3.8; F1 Worst -20.2; F2 Best 1.4; F2 Average 0.6; F2 Worst -2.9. - Extreme scenario F3a Average -35.7 Worst -54.7; impacts on own funds and CAR included (e.g., F3a: -2.58, -2.99, -10.01; -0.34-0.98). - Table source: French authorities, and staff estimates. - Table 3 (Macroeconomic scenario and liquidity stress tests — bank estimates and Commission Bancaire estimates): - Banks' estimates (macroeconomic scenarios M1–M4, impacts relative to 2003 after-tax profits and CAR): - M1 Relative to 2003 after-tax profits: Best -1.17, Average -23.6, Worst -46.9; Relative to end-2003 CAR: -1.06-1.42-1.40; Relative to Basel II-type CAR: -2.10-2.25. - M2 Relative to 2003 after-tax profits: Best 25.2, Average 2.0, Worst -10.4; Relative to end-2003 CAR: -0.83-0.97; Relative to Basel II-type CAR: 1.15 -0.15-0.42. - M3 Relative to 2003 after-tax profits: Best 34.6, Average 8.7, Worst -6.5; Relative to end-2003 CAR: -0.84-0.96; Relative to Basel II-type CAR: 1.30 -0.10-0.42. - M4 Relative to 2003 after-tax profits: Best 13.8, Average -4.4, Worst 20.1; Relative to end-2003 CAR: -0.77-1.06-0.11; Relative to Basel II-type CAR: -0.80-0.98. - Commission Bancaire estimates (examples): - M1 Commission Bancaire Average -26.0; Relative to end-2003 CAR -1.09; Relative to Basel II-type CAR -2.10. - M2 Commission Bancaire Average 7.0; Relative to end-2003 CAR -0.79; Relative to Basel II-type CAR -0.11. - M3 Commission Bancaire Average 13.2; Relative to end-2003 CAR -0.80; Relative to Basel II-type CAR -0.05. - M4 Commission Bancaire Average -9.2; Relative to end-2003 CAR -0.73; Relative to Basel II-type CAR -0.83. - Liquidity test (Percent, 1999-2003) details and notes are provided; impact on risk-weighted capital adequacy ratio allows for accumulated after-tax profits, credit growth, and change in risk weights, in percentage points. - Table 5 (Details of Macroeconomic Scenario Tests — projected cumulative percentage change from 2003 level, banks' and Commission Bancaire's estimates, selected entries): - Banks' estimates — Net profits Year 1 / Year 2 (Best, Average, Worst examples): - M1 Banks' Net profits: Best -3.70, Average -9.02, Worst -19.95; Bank estimates M1 in other columns: -1.17 -23.60 -49.93. - M2 Banks' Net profits: Best 23.44, Average 1.01, Worst -3.96; Bank estimates M2: 25.20 1.95 -10.37. - M3 Banks' Net profits: Best 9.73, Average -5.41, Worst -6.90; Bank estimates M3: 34.61 8.71 -6.45. - M4 Banks' Net profits: Best 0.23, Average -1.47, Worst 6.06; Bank estimates M4: 13.83 -4.37 -20.10. - Commission Bancaire's estimates for Net profits Year 1/Year 2 and NPL ratio Year 1/Year 2/Year 3 are reported (examples): - Commission Bancaire baseline NPL ratio Year 1 -1.58, Year 2 12.52, Year 3 5.96 (table formatting in source). - Commission Bancaire M1 Net profits: -6.02 -26.02 and NPL ratio Year 1 5.96, Year 2 16.56, Year 3 21.22 (table formatting in source). - Table 6 (Risk-Weighted Assets Projections — projected cumulative percentage change from 2003 level): - Example entries (with and without volume effect, Year 1 / Year 2): - Baseline: With volume effect Year 1 0.0 Year 2 0.0; Without volume effect Year 1 6.8 Year 2 8.7. - M1: With volume effect Year 1 25.6 Year 2 20.2; Without volume effect Year 1 34.1 Year 2 30.7. - M2: With volume effect Year 1 1.6 Year 2 1.6; Without volume effect Year 1 8.5 Year 2 10.4. - M3: With volume effect Year 1 4.8 Year 2 1.6; Without volume effect Year 1 11.9 Year 2 10.4. - M4: With volume effect Year 1 7.5 Year 2 6.4; Without volume effect Year 1 14.7 Year 2 15.7. - Source: Commission bancaire. Note: Projected increase in risk-weighted assets under Basel II-type rules. ### Insurance sector: sample, procedures, and key regulatory features - Sample coverage: - Life insurance: 26 companies holding 79 percent of total sectoral insurance liabilities (approximately €766 billion). - Composition: 20 subsidiaries of the 7 bank groups included in the banking sector stress tests (with 45 percent of sectoral liabilities); one state-controlled company (with 10 percent of liabilities); five individual companies that each hold at least 2 percent of total liabilities (three of these are subsidiaries of large foreign insurers). - For bank subsidiaries, stress tests were conducted on a group basis; others on an individual basis. - Nonlife insurance: tests applied to groups or entities representing 75 percent of nonlife insurance liabilities; includes 52 companies forming the 11 nonlife insurance groups or single companies that had insurance liabilities greater than 1.5 percent of total nonlife liabilities. None is owned by a bank. - Stress tests conducted on end-2003 data. - Test design and estimation: - Tests were run by the insurance supervisory agency, the CCAMIP. - Tests structured analogous to banking stress tests where possible; starting point for estimates: financial situation at end-2003. - A shock can have two effects: - Stock effect: instantaneous change in the value of the company’s assets (capital gains or losses). - Flow effect: progressive change in the return on the company’s assets (asset income level changes); the flow effect on profitability was cumulated over two years. - In tests with actual data, the stock effect was "much more important than the flow effect." - Presentation of outcomes: - Effects presented as: (i) changes to companies’ solvency margin (ratio of net admissible asset value to total main liabilities); (ii) pre- and post-shock solvency ratio relative to the required solvency margin; (iii) pre- and post-shock policy yields. - If a company cannot meet regulatory and contractual obligations from immediately available resources, the resource shortfall is reported as a share of the company’s main insurance liabilities. - Confidentiality: authorities reported only the average outcome, and the best and worst individual outcomes, without identifying individual institutions; they also reported the number of institutions that would fail prudential requirements and the magnitude of shortfalls. - Results were also estimated including reinsurance reimbursables, but most reported results exclude reinsurance to be conservative. Results for the large natural disaster test are presented with and without reinsurance. ### Life insurance regulatory framework relevant to stress tests - Key rules and features: - Solvency ratio: companies must maintain a solvency margin (capital and unrealized gains, "net admissible assets") above the minimum required solvency margin; solvency ratio is actual solvency margin divided by minimum required solvency margin. - Guaranteed rate of return: industry average guaranteed rate at end-2002 was 2.6 percent (was 3.7 percent at end-1995); guaranteed return is "much lower than actual policy yields." - The 85 percent mandatory pay-out policy rule: companies required to pay out at least 85 percent of asset income minus administrative overhead ("net income"). - The 125 percent minimum prudence margin rule: if realized rate of return on assets falls below 125 percent of the guaranteed minimum return, insurance liabilities must be recalculated using the realized rate of return rather than historical yield, affecting actual and required solvency ratios. - Deferred insurance benefit provision: companies are allowed to distribute gains in excess of the guaranteed minimum return over an eight year period. - Accounting treatment of unrealized losses on fixed instruments (bonds): unrealized losses are ignored for solvency purposes (asset-liability matching premise); unrealized gains are added to capital. - Unrealized losses (gains) on other assets (e.g., equity and real estate) are (not) booked on balance sheet. - If policies are cashed in, gains and losses must be realized immediately. - Implications for stress absorption: - For shocks of the magnitudes considered and assuming policies are not cashed in, companies have discretion to absorb shocks by: - Changing current or future returns on policies; and/or - Adjusting the available solvency margin. - A company’s policy mix choice depends on business strategy and objectives (smoothing profits, retaining market share, building reserves, financing expansion, etc.). - Presentation of results includes estimates of the maximum effect on current policy returns and, alternatively, the maximum effect on the solvency ratio to capture indeterminacy in company responses. ### Interpretation framework (illustrated by Figure 1 in source) - Conceptual mapping: - Vertical axis: solvency ratio (percent) relative to minimum required solvency margin. - Horizontal axis: current yield on policies (percent). - Regulatory and contractual constraints depicted (85 percent return rule and 125 percent income rule) bound combinations of policy yield reductions and solvency reductions that an insurer can choose. - Small shocks move firms along segments (e.g., A1-B1) where the shock can be absorbed by reducing current yield or solvency or a mix; larger shocks move firms further (e.g., A2-B2). - Very large shocks could exhaust room for maneuver (A3-B3), where firms would no longer meet solvency requirements nor the 125 percent income rule and/or the 85 percent return rule. - Note: Depiction of constraints as linear in the diagram is for convenience; actual constraints need not be linear. Normally, either the return rule or the income rule will bind before the other. _Italic: Source — French authorities, Commission Bancaire, CCAMIP, and staff estimates as presented in the cited stress test section._ ### 43. If a shock reduces a company’s solvency ratio below the required level (if the ### _cr05185 - 43. If a shock reduces a company’s solvency ratio below the required level (if the ### Measurement of solvency and yield shortfalls - If a shock reduces a company’s solvency ratio below the required level (a point such as B2), the solvency shortfall is measured relative to its main insurance liabilities (Column 8 of Table 7). - If the 85 percent return rule is binding or if a company is unable to meet the 125 percent income rule (a point such as A2), the shortfall is measured relative to both its main insurance liabilities and required solvency margin (last two columns of Table 7). - On Figure 1, distance S-B2 corresponds to the quantity of assets needed to restore solvency to a minimum acceptable level. - On Figure 1, distance R-A2 corresponds to the shortfall in resources needed to pay the 85 percent required yield on policies. ### Graphical representation: solvency ratio and realized yield (Figure 2) - Axes and thresholds: - Solvency ratio should be above 100 percent. - Prudential provisions apply if the realized investment yield is below 125 percent of the guaranteed minimum return on policies (the “minimum prudential guideline”). - Typical company movements under shocks: - Starting point O. - Modest shock → position along A1-B1; firm can choose a point on the vertical line depending on profit-smoothing and realized payouts. The vertical line length depends on policy provisions and regulatory constraints. - Larger shock → position A2-B2; the “minimum prudence guideline” is binding; firm must choose payout rate and comply with prudential requirements to stay above the horizontal axis. - Very large shock → position A3-B3; firm cannot meet several prudential requirements and owners’ capital would necessarily be reduced. - A shock to liabilities is represented as a downward shift of all points (for example, O to O’), directly linking a realized shock to solvency. ### Policyholder behavior and extreme liquidity scenarios - Many buffers available to life insurance companies can be vitiated if policyholders cash in early. - Early cash-in is unlikely unless there is a large, sudden, and persistent rise in interest rates that induces transfers out of life insurance products into higher return instruments. - As an extreme test, assume the sector collapses and all policyholders cash in their policies immediately: this evaluates capital shortfalls if companies had to meet all obligations immediately. - Authorities can prevent such runs by mandating suspension of repurchases of policies or by transferring the portfolio of a distressed company. ### Net present value (NPV) approach for life insurers - Alternative approach: measure net worth where all assets and liabilities are measured in present value terms. - This ignores current accounting and prudential smoothing (including ability to smooth shocks over eight years) and assumes shocks are entirely absorbed by capital. - Given the long-term nature of life insurance, the NPV test better captures long-term mismatch between assets and liabilities and is analogous to bank stress testing. - Results of this approach are presented in Table 8. ### Definition and calibration of stress tests - Insurance-sector tests parallel single and multiple factor tests used for banks; assumed shocks under tests F1, F2, F3a, and F5 are the same as for banks. - Tests F4, F6, F7, and F8 are not applicable because insurance companies have negligible exposure to these risk factors. - Insurance-specific tests: - F3b: a 100 basis point decrease in all bond yields. - F10: a fall in French real estate prices by 30 percent. - F11: a 100 basis point increase in all interest rates combined with a 30 percent fall in equity prices. - F12: a natural disaster causing damage twice as large as the worst previous experience (1999); applies only to the nonlife sector. Effects calculated without and with reinsurance receivables (F12a and F12b). - Notes: - F3b is somewhat implausible given low current rates but illustrates asymmetries. - F11 is severe and implies an unlikely reaction to a large fall in equity prices. - F12 is regarded as very severe but conceivable; natural disaster scenarios do not take into account possible second-round effects on asset liquidity. ### Stress test results — overview (Tables 7 and 8) - Aggregate industry resilience: - All insurance companies would remain within mandated regulatory limits under all tests, although regulatory constraints would be binding in some cases, forcing changes to solvency position or policy yield or both (Table 7). - Directional effects: - Largest effects seen in tests involving increases in interest rates (F1, F3a, F11). - A fall in interest rates benefits companies (F2, F3b) because capital gains (asset values rise) more than offset negative earnings over the test duration. - Increases in interest rates can hurt insurers, especially if they must sell bonds at a loss when policyholders cash in to seek higher returns. - Specific outcomes: - In test F3a (permanent increase in interest rates), two (life) companies and thirteen (unspecified) companies would not be able to absorb the shock by reducing solvency (policy yield). - Regulatory constraints on solvency are less often binding than constraints on yield; in test F1 only two life companies could not absorb the shock by reducing solvency margin, whereas six life companies could not absorb the full effect by reducing yields on current policies. - Even when a constraint is binding, the shortfall is small relative to main insurance liabilities. - Equity, real estate, and natural disaster impacts: - Equity and real estate price falls generate relatively minor losses. - A very large natural disaster would severely affect some nonlife companies if reinsurance did not reimburse (F12). - The sign of effects across insurance companies is consistently the same (unlike banking, where shocks can benefit some banks and harm others). - Extreme run scenario (all life policies cashed in immediately): - All policy-holders could be reimbursed in full except in test F3a. - For the sector as a whole, shareholders would lose 79 percent of their capital, but remaining assets would exceed insurance liabilities by 1.1 percent. - 7 of the 26 companies would be unable to meet all obligations to policy-holders; these seven are mostly bank subsidiaries with larger bond shares. - In the worst case, the shortfall would amount to 9.1 percent of main insurance liabilities. - These results suggest system-level resilience over the test horizon, even under extreme assumptions such as sector collapse. ### Net present value (NPV) test results (Table 8) - Initial level: - Average: 16.4 - Worst: 7.6 - Changes (difference between present value of assets and liabilities, as a percent of main insurance liabilities): - F1: - Levels: Average 13.2, Worst 3.0 - Change: Average -3.2, Worst -4.7 - F2: - Levels: Average 17.0, Worst 8.6 - Change: Average 0.7, Worst 2.7 - F3a: - Levels: Average 7.4, Worst -5.3 - Change: Average -8.9, Worst -13.7 - F3b: - Levels: Average 17.7, Worst 6.5 - Change: Average 1.3, Worst 5.4 - F5: - Levels: Average 12.7, Worst 3.8 - Change: Average -3.7, Worst -7.3 - F10: - Levels: Average 15.0, Worst 4.8 - Change: Average -1.3, Worst -3.1 - F11: - Levels: Average 9.5, Worst -0.6 - Change: Average -6.9, Worst -9.6 - Interpretation: - NPV results confirm previous findings: a large rise in interest rates would leave some companies relatively weak; an equity price collapse would have an important impact on some companies. - Firms with negative estimated present value would need to transfer some impact to policyholders or receive capital injections from shareholders. - From an industry-wide perspective, long-term effects of reasonable shocks appear manageable. *Source: French authorities, and staff estimates (as presented in the supplied content).* ### 56. Such connections through investment portfolios are most immediately relevant to the ### _cr05185 - 56. Such connections through investment portfolios are most immediately relevant to the ### International linkages through the financial sector (scope and focus) - The note concentrates on linkages that arise within the financial sector rather than on international linkages through other sectors. - Attention focuses on: - The international operations of French financial institutions—especially banks and insurance companies—through direct exposure and through the activities of their subsidiaries. - The activities of foreign institutions in providing financial services in France. - The connections between securities markets in France and elsewhere. ### France: Foreign assets and liabilities (summary of Table 9; market values, end of period) - French claims on abroad (in billions of euros): - 1999: 2,310.6 - 2000: 2,667.3 - 2001: 2,785.9 - 2002: 2,554.0 - Components of French claims on abroad (in billions of euros): - Direct investment: 1999: 908.8; 2000: 1,133.5; 2001: 1,051.1; 2002: 833.1 - Securities: 1999: 587.5; 2000: 713.3; 2001: 806.0; 2002: 825.7 - Monetary financial institutions' claims: 1999: 455.6; 2000: 461.8; 2001: 514.2; 2002: 513.5 - Other: 1999: 358.7; 2000: 358.7; 2001: 414.6; 2002: 381.7 - French liabilities to abroad (in billions of euros): - 1999: 2,233.1 - 2000: 2,458.4 - 2001: 2,545.1 - 2002: 2,371.7 - Components of French liabilities to abroad (in billions of euros): - Direct investment: 1999: 645.9; 2000: 643.2; 2001: 604.1; 2002: 522.1 - Securities: 1999: 850.6; 2000: 993.8; 2001: 1,041.6; 2002: 974.2 - Monetary financial institutions' liabilities: 1999: 498.5; 2000: 569.9; 2001: 622.1; 2002: 626.2 - Other: 1999: 238.1; 2000: 251.5; 2001: 277.3; 2002: 249.2 - Net position (in billions of euros): 1999: 77.5; 2000: 208.9; 2001: 240.8; 2002: 182.3 - GDP (in billions of euros): 1999: 1,355.1; 2000: 1,420.1; 2001: 1,475.6; 2002: 1,520.8 - Select shares (in percent of GDP) — French claims on abroad: - Total claims on abroad (In percent of GDP): 1999: 170.5; 2000: 187.8; 2001: 188.8; 2002: 167.9 - Direct investment: 1999: 67.1; 2000: 79.8; 2001: 71.2; 2002: 54.8 - Securities: 1999: 43.4; 2000: 50.2; 2001: 54.6; 2002: 54.3 - Monetary financial institutions' claims: 1999: 33.6; 2000: 32.5; 2001: 34.8; 2002: 33.8 - Other: 1999: 26.5; 2000: 25.3; 2001: 28.1; 2002: 25.1 - Select shares (in percent of GDP) — French liabilities to abroad: - Total liabilities to abroad (In percent of GDP): 1999: 164.8; 2000: 173.1; 2001: 172.5; 2002: 156.0 - Direct investment: 1999: 47.7; 2000: 45.3; 2001: 40.9; 2002: 34.3 - Securities: 1999: 62.8; 2000: 70.0; 2001: 70.6; 2002: 64.1 - Monetary financial institutions' liabilities: 1999: 36.8; 2000: 40.1; 2001: 42.2; 2002: 41.2 - Other: 1999: 17.6; 2000: 17.7; 2001: 18.8; 2002: 16.4 - Net position (In percent of GDP): 1999: 5.7; 2000: 14.7; 2001: 16.3; 2002: 12.0 ### Banking sector: scale and geographic composition of foreign exposures - French banks play a major role in channeling cross-border financial flows; international lending is on a scale comparable to that of other major industrialized countries. - Lending patterns: - Particularly active in lending to other industrialized countries, primarily via the EMU-wide interbank market. - Lending to developing countries is less in absolute terms and smaller relative to some other major industrialized countries, though amounts are large relative to recipient economies. - BIS-based consolidated claims (Table 10; end of period, in billions of euros): - Total consolidated claims: - Dec-99: 835.6 - Dec-00: 848.9 - Dec-01: 929.1 - Dec-02: 994.1 - Dec-03: 1,071.6 - Developed countries claims: - Dec-99: 660.8 - Dec-00: 682.4 - Dec-01: 752.5 - Dec-02: 832.3 - Dec-03: 911.3 - Europe: Dec-99: 453.2; Dec-00: 428.4; Dec-01: 459.1; Dec-02: 486.9; Dec-03: 530.4 - United States: Dec-99: 130.5; Dec-00: 169.3; Dec-01: 181.8; Dec-02: 232.3; Dec-03: 247.5 - Offshore centers: Dec-99: 58.8; Dec-00: 59.4; Dec-01: 61.3; Dec-02: 55.6; Dec-03: 48.0 - Africa & Middle East: Dec-99: 32.2; Dec-00: 32.4; Dec-01: 35.7; Dec-02: 35.9; Dec-03: 35.7 - Asia & Pacific: Dec-99: 35.6; Dec-00: 32.8; Dec-01: 29.4; Dec-02: 27.8; Dec-03: 29.0 - Latin America/Caribbean: Dec-99: 25.2; Dec-00: 22.5; Dec-01: 23.6; Dec-02: 17.6; Dec-03: 15.1 - Shares of total (in percent) — selected: - Developed countries: 79.1; 80.4; 81.0; 83.7; 85.0 - Europe: 54.2; 50.5; 49.5; 49.0; 49.5 - United States: 15.6; 19.9; 19.6; 23.4; 23.1 - Offshore centers: 7.0; 7.0; 6.6; 5.6; 4.5 - Latin America/Caribbean: 3.0; 2.7; 2.5; 1.8; 1.4 - Interpretation: - Total consolidated claims currently over €1,000 billion, equivalent to about one quarter of the banking system’s total assets. - Claims on other industrialized countries in Europe constitute one half of the total; claims on the U.S. amount to more than 20 percent. - Claims on offshore financial centers and on other countries constitute smaller and declining shares; claims on Latin America and the Caribbean declined from 2002, likely reflecting financial crises in that region. ### Resident (nonconsolidated) banks’ claims on nonresidents (Table 11 and Figures 5–7) - French banks’ claims on nonresidents — components (in billions of euros; end of period): - Lending: - Dec-97: 236.9 - Dec-98: 219.5 - Dec-99: 254.4 - Dec-00: 240.4 - Dec-01: 384.7 - Dec-02: 391.5 - Jun-03: 397.7 - Bonds: - Dec-97: 73.5 - Dec-98: 93.1 - Dec-99: 120.2 - Dec-00: 143.5 - Dec-01: 147.6 - Dec-02: 185.3 - Jun-03: 200.7 - Shareholdings: - Dec-97: 15.1 - Dec-98: 17.4 - Dec-99: 29.8 - Dec-00: 33.0 - Dec-01: 28.6 - Dec-02: 30.7 - Jun-03: 38.8 - Share of total assets (in percent): - Lending: 8.7; 8.0; 8.5; 8.2; 11.8; 11.8; 11.3 - Bonds: 2.7; 3.4; 4.0; 4.9; 4.5; 5.6; 5.7 - Shareholdings: 0.6; 0.6; 1.0; 1.1; 0.9; 0.9; 1.1 - Geographic pattern (narrative points from figures and text): - Lending dominated by Europe (notably the U.K.), significant exposures to the U.S. and East Asia (mainly Japan, Korea and China). - Relatively large exposures to (near) investment-grade emerging markets such as the Czech Republic, Morocco, and Poland. - Exposures to noninvestment grade emerging markets are relatively small and widely diversified. - Holdings of foreign bonds have been growing in importance, with European issuers predominating (Italy and the Netherlands relatively important for bond holdings). - Equity holdings by banks are modest and roughly stable since end-1999; geographic distribution is wide. - Some claims reflect loans to or equity stakes in banks’ own subsidiaries and joint ventures. ### Counterparty composition and concentration (Table 12) - Claims on nonresidents are concentrated in claims against other financial institutions: - Loans to nonresident financial institutions (in billions of euros): - Dec-97: 302.0 - Dec-98: 267.4 - Dec-99: 277.9 - Dec-00: 275.6 - Dec-01: 330.1 - Dec-02: 314.6 - Jun-03: 292.4 - Loans to nonfinancial nonresidents (in billions of euros): - Dec-97: 60.1 - Dec-98: 66.8 - Dec-99: 75.2 - Dec-00: 86.6 - Dec-01: 95.6 - Dec-02: 88.1 - Jun-03: 88.4 - Customer deposits from nonresidents (in billions of euros): - Dec-97: 49.1 - Dec-98: 50.2 - Dec-99: 55.3 - Dec-00: 64.9 - Dec-01: 74.3 - Dec-02: 70.9 - Jun-03: 74.9 - Selected ratio outcomes (percent): - Customer deposits from nonresidents/Total customer deposits: 6.5; 6.4; 6.8; 7.8; 8.3; 7.8; 1.9 - Loans to nonfinancial nonresidents/total loans to nonfinancials: 6.0; 6.5; 6.8; 7.1; 7.5; 6.8; 19.8 - Loans to nonresident financial institutions/Total loans to financial institutions: 39.7; 34.5; 31.6; 31.1; 31.3; 30.0; 27.6 - Sectoral concentration (share of banks in sectoral total; percent): - Customer deposits from nonresidents: 90.5; 89.3; 88.4; 88.2; 89.6; 88.6; 90.0 - Loans to nonfinancial nonresidents: 84.0; 83.1; 86.0; 86.1; 87.7; 85.4; 83.8 - Loans to nonresident financial institutions: 79.1; 82.5; 85.9; 88.4; 85.6; 90.1; 89.0 - Interpretation: - Loans to nonresident financial institutions amount to three quarters of financial institutions’ total loans to nonresidents, and about 30 percent of total loans to all financial institutions. - Claims on nonfinancial nonresidents are much smaller and only slightly exceed deposit liabilities toward nonfinancial nonresidents. - 80 to 90 percent of foreign assets and liabilities are concentrated in the commercial banks; mutuals and other credit-granting institutions are less involved in European money markets. ### Foreign operations of major French banking institutions (Box 2 summary) - Major banks vary greatly in extent and type of foreign operations; all large groups are engaged in the European-wide interbank market and wholesale investment banking. - BNP Paribas: - More than half its revenue and profits from retail operations are generated abroad. - Only 40 percent of its lending goes to France; a quarter goes to North America and a sixth to Western Europe. - Acquired two retail branch networks in the United States; presence in WAMU region, the Maghreb and the Mashrak. - Significant presence in consumer lending markets in the U.K., Spain, Italy, and Hungary. - Operations in Asia mainly through joint ventures. - About a third of leasing activities are abroad. - Société Générale: - Focused on acquiring retail networks in transition countries (notably the Czech Republic and Slovenia) and some developing countries (Ghana, Tunisia). - Recent ROE in those networks in excess of 30 percent. - International operations in 2003 contributed 11 percent of group revenue and 13 percent of profits. - Crédit Agricole: - Acquired Indosuez and Crédit Lyonnais, increasing foreign operations. - Earns about a quarter of revenues abroad (mainly in the Americas and Western Europe). - Minority shareholdings in banks abroad: Greece, Italy, Poland, Portugal. - Wrote off investments in Argentina and handed over operations there following the crisis in 2002. - Note: consolidated accounts do not always distinguish clearly between domestic and foreign business, especially for wholesale activities. ### Subsidiaries, branches and employment abroad (Table 13) - Number of subsidiaries abroad: - 1997: 257 - 1998: 275 - 1999: 316 - 2000: 345 - 2001: 410 - Number of branches abroad: - 1997: 207 - 1998: 221 - 1999: 227 - 2000: 208 - 2001: 200 - Employees abroad (levels): - 1997: 21,160 - 1998: 23,827 - 1999: 21,308 - 2000: 16,875 - 2001: 16,131 - 2002: 20,337 - Employees abroad — banks: - 1997: 20,976 - 1998: 23,585 - 1999: 21,043 - 2000: 16,393 - 2001: 15,572 - 2002: 19,808 - Three largest credit institutions — employees abroad: - 1997: 9,018 - 1998: 9,921 - 1999: 7,647 - 2000: 8,967 - 2001: 7,642 - 2002: 10,629 - Employees abroad/total employment (in percent): - 1997: 5.5; 1998: 6.2; 1999: 5.6; 2000: 4.4; 2001: 4.1; 2002: 5.1 - Memorandum items: - Number of domestic branches: 1997: 9,204; 1998: 9,269; 1999: 9,109; 2000: 8,495; 2001: 8,493; 2002: 8,745 - Metropolitan employment: 1997: 361,597; 1998: 359,767; 1999: 362,039; 2000: 368,663; 2001: 380,475; 2002: 380,545 - Interpretation: - Much foreign activity undertaken by subsidiaries and therefore not booked to the metropolitan head office. - Number of subsidiaries abroad has increased; direct employment abroad has tended to decrease with year-to-year fluctuations. ### Foreign banks’ operations in France (Tables 14 and 15) - Consolidated borrowing from foreign banks — total claims (in billion euros; end of period): - Dec-99: 460 - Dec-00: 554 - Dec-01: 612 - Dec-02: 606 - Dec-03: 612 - Breakdown (in billion euros): - Public sector: 43; 59; 57; 64; 87 (Dec-99 to Dec-03) - Nonbank private sector: 98; 143; 153; 152; 152 (Dec-99 to Dec-03) - Net local currency claims on residents: 24; 44; 50; 42; 27 (Dec-99 to Dec-03) - Interpretation: - Total rose by a third in the early years of the EMU, then leveled off around €600 billion. - Interbank lending within the Eurozone constitutes much of the total and accounts for most of the increase. - Claims on the public sector increased and now constitute about one eighth of the total; claims on the nonbank private sector are about double that. - Foreign banks’ presence in France (Table 15; selected items): - Number of institutions: Dec-97: 291; Dec-98: 320; Dec-99: 307; Dec-00: 328; Dec-01: 312; Dec-02: 287; Jun-03: 277 - Metropolitan employment: Dec-97: 21,913; Dec-98: 26,371; Dec-99: 26,703; Dec-00: 39,013; Dec-01: 41,716; Dec-02: 41,370 - Total assets (in billions of euros): 305.1; 295.6; 373.7; 509.5; 495.8; 472.9; 459.1 (Dec-97 to Jun-03) - Loans to nonfinancial sector (in billions of euros): 59.9; 72.3; 132.6; 175.3; 179.1; 183.1; 173.2 - Deposits (in billions of euros): 35.6; 40.1; 44.2; 68.9; 75.2; 79.0; 77.2 - Share of sectoral totals (percent) — total assets: 10.1; 9.7; 11.0; 14.8; 13.1; 12.5; 11.5 - Share of sectoral totals (percent) — loans to nonfinancial sector: 5.9; 7.1; 12.0; 14.5; 14.0; 14.2; 13.3 - Share of sectoral totals (percent) — deposits: 4.7; 5.1; 5.4; 8.3; 8.4; 8.7; 8.3 - Interpretation: - Foreign banks have about a ten percent market share, comparable to other major European countries. - Foreign banks have a higher share of lending than deposit-taking. - Number of foreign banks roughly constant; volume of activity jumped in 2000 due to a major takeover. - Cross-border retail business remains limited; branch networks and mutualist ownership structures affect patterns of entry and acquisitions. *Source: Banque de France, Bank for International Settlements, French authorities, and staff estimates.* ### 66. The French insurance sector is among the six largest in the world. It has extensive ### The French insurance sector is among the six largest in the world. It has extensive ### Insurance sector linkages and international footprint - The sector’s international linkages take the form mostly of ownership connections, reinsurance, and financial investments, rather than the writing of cross-border policies. - Foreign insurance companies own subsidiaries in France with a combined market share of about 20 percent in nonlife business, and rather less than that in the life sector. - For domestically-incorporated companies, premium income from outside France amounts to less than two percent of total premiums, and that comes overwhelmingly from within the EU. - Premium income from non-French business is almost all in the nonlife sector. - French insurers make extensive use of foreign reinsurers; French commercial reinsurers are comparatively small and concentrate on business in France. ### Ownership, subsidiaries, and diversification (Axa example) - Axa, the largest French insurer, has extensive insurance operations in the U.S., Europe (mainly the U.K., Germany, and Belgium), Japan, Hong Kong, Singapore, and Australia, conducted through subsidiaries. - Axa also has smaller operations in other countries (for example Morocco) and operates a medium-sized bank in Belgium; in 2003 it pulled out of insurance business in Argentina and Brazil. - Less than a fifth of Axa’s group revenue is generated in France; North America, or Germany, and the U.K. combined are equally important. - In recent years, Axa has earned more net income on life business in the U.S. than in France. - Diversification across countries and lines of business tends to stabilize group profits; any one member company could suffer large losses, but “in principle Axa group’s liability is limited.” ### Asset allocation, currency and country risk - Data are not available on the allocation of assets by currency or country. - Prudential regulations and companies’ incentives ensure that the vast bulk of assets take the form of claims on industrialized countries, and in particular euro-denominated government bonds from OECD countries. - Hence, the vulnerability of the insurance sector to country or exchange rate risk is small. - Evidence suggests French insurance companies’ involvement in credit derivatives markets is “not yet large relative to their balance sheet size.” ### Reinsurance exposure and implications - In 2002, total reinsurance premiums amounted to €19.3 billion (just under 15 percent of premium income). - Less than half of €19.3 billion went to French reinsurers. - Foreign earnings of reinsurers in France in 2002 amounted to just over €1 billion, two-thirds of which was earned by French companies. - Implication: France is transferring abroad a significant portion of insured risk, while not taking on the concentration of risk that is characteristic of reinsurance; the French financial system as a whole has little exposure to vulnerabilities of the reinsurance sector. - Progressive difficulties at foreign reinsurers would be reflected first in higher reinsurance costs, then in the nonavailability of some types of reinsurance, and, ultimately, in the nonreimbursement of reinsured claims. ### Equity and bond market interdependence — stylized facts - France’s capital markets have become closely integrated into the global marketplace; French government bonds serve as a benchmark for short- to medium-term Euro-denominated interest rates. - Foreign participation and holdings: - Roughly 40 percent of trading activity in French government bonds is accounted for by foreign investors. - Equity holdings by non-French residents are roughly 35 percent. - By comparison, 9 percent of U.S. domestic equity is held by foreigners; 22 and 15 percent of domestic Japanese and Canadian equity (respectively) is held by foreigners. - For most EU countries, the range of foreign-held domestic equity is between 25 and 38 percent. - Correlation and volatility patterns: - Correlations between national equity markets have been rising; estimated correlation coefficients between French and other major equity markets rarely decline below the 0.50 mark. - Since the early 2000s, French equity price movements have become even more correlated with other major equity markets; the average equity market correlation rises sharply from late 1999, reaches new highs in 2002, and plateaus thereafter at around 0.70. - The correlation between the French market and the Netherlands and Belgium increased markedly in late 2001, roughly coinciding with implementation of the common Euronext trading platform in October 2001. - Correlations across bond markets are generally positive, ranging on average between 0.2 and 0.8; bond correlations are lower on average than equity correlations and seem more stable over time. - Methodological note on measures: - Correlation coefficient and volatility measures in the study are based on an exponentially weighted moving average of past returns, where weights decay by a factor of 0.90. - Academic work indicates traditional (unconditional) correlation measures may be biased and dependent on the level of volatility observed. - Volatility observations: - Volatility has risen across most asset markets over the second half of the sample period. - French equity and bond returns have exhibited higher volatility than those in the U.S., broadly in line with other European markets; the higher equity volatility may reflect the relatively heavy weight in the CAC40 of financial and technology firms. - Correlations tend to be higher during periods of heightened volatility; after conditioning for changes in volatility, much of the observed increase in correlations may be attributable to the rise in volatility rather than solely to macroeconomic or institutional factors. ### Conditional correlation analysis and Euronext impact - A multivariate model that estimates return correlations conditional on market volatility was used to assess changes in interdependence and detect regime changes. - Findings: - Markets have become more integrated even when controlling for changes in volatility, supporting the influence of globalization on correlations. - For the French and Dutch markets, little empirical support is found that the move to a unified Euronext trading platform on October 29, 2001 resulted in greater cross-market correlation in equity returns. - The Belgian stock market became somewhat less correlated with the French and Dutch markets since Euronext started, though not to an economically significant degree; this may reflect market composition differences and acquisitions of Belgian firms by multinationals. ### Data and sample used for multivariate regime-switching model - Weekly data period: August 5, 1987 to August 4, 2004. - Equity indices analyzed: CAC40, DAX, S&P 500, FTSE; additionally AEX and BEL20 (BEL20 data spans January 1, 1991 till August 4, 2004). - Bond indices analyzed: 7- to 10-year indices for France, Germany, the U.S. and U.K., chosen to reduce durational differences across sovereign benchmarks. *Prepared by Daniel Hardy and Toni Gravelle (content as presented in the source chapter).* ### 87. We postulate a multivariate model for returns that are driven by idiosyncratic and ### _cr05185 - 87. We postulate a multivariate model for returns that are driven by idiosyncratic and common shocks ### Model specification - Postulate a multivariate model for returns r1t, r2t, ... , rnt driven by idiosyncratic and common shocks. - Each variable follows a distributed lag process: (for i = 1, 2, 3...) - rit = αi + Φi(l)rit−l + uit (equation (1)) - αi is the intercept; Φi(l) is the lag polynomial specific to variable i; uit is the disturbance term. - Disturbances are correlated: E(uit ujt) ≠ 0 for i ≠ j. - Disturbance decomposition into common and idiosyncratic structural shocks for i = 1,...,4: - u1t = σ1 z1t + δ1 zct + gc - u2t = σ2 z2t + δ2 zct + gc - u3t = σ3 z3t + δ3 zct + gc - u4t = σ4 z4t + δ4 zct + gc (equation (2), notation as in source) - idiosyncratic shocks zit: uncorrelated with each other and with the common shock zct, zero mean, variances normalized to one. - Loading coefficients δi’s and σi’s interpreted as standard deviations. ### Regime-switching specification - Common shock zct is drawn from two distinct regimes: regime 0 and regime 1. - Regime one assumed to have a smaller mean for the common shock and, in principle, a larger variance (and higher covariance) for equity returns. - State variable Sct ∈ {0, 1} governs regime-dependent parameters. - State-dependent disturbance form (for i = 1,..,4): - uit = σi zit + δi(Sct) zct + gc(Sct) (equation (3) notation) - Markov-switching dynamics for Sct: - Pr[St = 0 | St−1 = 0] = q - Pr[St = 1 | St−1 = 1] = p (equation (5)) - High-variance states are modeled as relatively persistent at the weekly frequency. ### Variance-covariance structure and implications - Variance-covariance matrix Σ(Sct) entries derive from idiosyncratic variances σi2 and cross-products of loading factors δi,S: - Variance of uit in regime 0: σi2 + δi,02 - Covariance of ui and uj in regime 0: δi,0 δj,0 - Different δi allowed across regimes; thus variance and covariance can be greater in either regime for a given series. - Example interpretation from text: - State 0 = high mean regime; State 1 = low mean. - Correlations can change sign across pairs and regimes (e.g., CAC 40–S&P vs CAC40–DAX in same regime). - Endogenous designation of high-volatility periods via the Markov-switching estimation (no ex post rule needed). ### Incorporation of Euronext dummy (three-equation framework) - δi parameters adjusted to include dummy DS for Euronext (value 0 before October 29, 2001 and 1 thereafter): - δi(t) = δi(Sit) + βi DS (equation (6) notation) - Assumed that βi impact is constant across states. - Under this specification the diagonal variances are not affected by the dummy; correlations across French, Dutch, and Belgian equity markets can be influenced if βi significant. ### Estimation results — Equity markets (group 1: CAC40, S&P 500, FTSE, DAX) - Sample Size: 888 - Log Likelihood value: 5966.3 - Parameters, Estimates, and T-statistics (variables i = 1, 2, 3, 4 correspond to CAC40, S&P 500, FTSE, DAX respectively): - δ1(1) = 4.49* ; T-statistic = 12.57 - δ2(1) = 3.33* ; T-statistic = 11.37 - δ3(1) = 3.66* ; T-statistic = 11.04 - δ4(1) = 4.96* ; T-statistic = 12.25 - σ1 = 1.23* ; T-statistic = 15.18 - σ2 = 1.83* ; T-statistic = 38.12 - σ3 = 1.50* ; T-statistic = 35.71 - σ4 = 1.63* ; T-statistic = 28.57 - Q = 0.98* ; T-statistic = 32.50 - P = 0.92* ; T-statistic = 121.25 - δ1(0) = 2.22* ; T-statistic = 11.66 - ∆2(0) = 0.99* ; T-statistic = 8.78 - ∆3(0) = 1.13* ; T-statistic = 10.38 - ∆4(0) = 1.87* ; T-statistic = 12.33 - gc(0) = 0.24* ; T-statistic = 3.58 - gc(1) = -0.11 ; T-statistic = -0.69 - * implies estimate is significant at 5% levels. - Key findings from this estimation: - All coefficients except gc(1) significant at the 5 percent level. - Factor loadings in regime one, δi(1), are larger than δi(0) ⇒ regime one volatility greater than regime zero. - Example volatility comparison for French equity: - Regime one variance: σ1^2 + δ1,1^2 = 20.25 - Regime zero variance: σ1^2 + δ1,0^2 = 6.44 - gc(1) below gc(0) ⇒ high-volatility regime characterized by small or negative stock price movements. - Correlations higher in regime one: - CAC 40–S&P 500 covariance in regime one: δ1(1) δ2(1) = 14.52 - CAC 40–S&P 500 covariance in regime zero: δ1(0) δ2(0) = 2.18 - Corresponding correlation coefficients: 0.85 for regime one and 0.42 for regime zero. - Estimated probabilities of being in regime one are more frequent and persistent in the latter part of the sample (visualized in Figure 17 in source). - Much of the increase in unconditional correlation measures between French and other major equity markets since the mid-1990s is attributable to the higher level of volatility observed in that period; only part is attributable to changes in fundamental linkages. - Test for change in interdependence over time (controlling for volatility): - Likelihood-Ratio test statistic = 39 - Asymptotic distribution: chi-squared with three degrees of freedom - Conclusion in text: statistic exceeds chi-squared critical values for standard confidence levels ⇒ degree of interdependence between the four equity markets has increased over time even after accounting for volatility changes. ### Estimation results — Euronext test (three-variable system: CAC40, AEX, BEL20) - Purpose: test whether correlations among Euronext exchanges changed after move to common trading platform (October 29, 2001) via dummy in δs. - Sample Size: 706 - Likelihood value: 3792.2 - Parameters, Estimates, and T-statistics (variables i = 1, 2, 3 correspond to CAC40, AEX, BEL20 respectively): - δ1(1) = 4.47* ; T-statistic = 13.75 - δ2(1) = 4.12* ; T-statistic = 12.87 - δ3(1) = 5.53* ; T-statistic = 14.10 - σ1 = 1.51* ; T-statistic = 25.00 - σ2 = 1.63* ; T-statistic = 32.00 - σ3 = 0.93* ; T-statistic = 12.85 - Q = 0.99* ; T-statistic = 99.02 - P = 0.96* ; T-statistic = 32.01 - δ1(0) = 2.22* ; T-statistic = 11.50 - δ2(0) = 0.99* ; T-statistic = 9.66 - δ3(0) = 1.13* ; T-statistic = 11.33 - gc(0) = 0.17 ; T-statistic = 1.02 - gc(1) = 0.23* ; T-statistic = 2.87 - Β1 = -0.01 ; T-statistic = 0.11 - Β2 = 0.13 ; T-statistic = 1.44 - Β3 = -0.30* ; T-statistic = 5.10 - * shows significance at 5% levels. - Key findings from Euronext estimation: - Only the Belgian dummy parameter estimate, Β3, is statistically significant and negative. - Interpretation in text: since inception of Euronext trading, Belgian equity market became moderately less correlated with Dutch and French markets. - Effect size measured by dummy not judged economically important by authors and may reflect other unobserved factors coinciding with Euronext consolidation timing (e.g., Belgian market becoming more local due to EMU effects, merger and acquisition activity, declining liquidity and listings). *Italic: Source content from the supplied IMF PDF content unit.* ### 104. Estimation results indicate that, in periods of either high or low volatility, the French ### _cr05185 - 104. Estimation results indicate that, in periods of either high or low volatility, the French ### Equity market interdependence (paragraph 104) - Estimation results indicate that, in periods of either high or low volatility, the French equity market is positively correlated (on average) with its Dutch and Belgian counterparts, indicating a persistent degree of interdependence. - The timing of high volatility (and high correlation) periods generally overlaps with those estimated for the four equity markets displayed in Figure 17. - The test for no change in the degree of interdependence is strongly rejected, indicating that, after controlling for changes in volatility, the French equity market’s linkages to the Dutch and Belgian markets has grown over time. - Reference: Figure 18. Estimated Regime 1 Probabilities (CAC 40, AEX, BEL20). ### Bond markets: model, specification, and estimation (paragraphs 105–106, footnote 58) - A multivariate Markov-switching model was estimated for bond returns across France, Germany, the U.S. and U.K. - In estimating the independent-switching model, the specification for the transition probabilities used unconditional probabilities: - P0 = Pr[St = 0] and P1 = Pr[St = 1] = 1-P0. (equation (8)) - These probabilities are unconditional and capture the idea that the high-variance shocks are unpredictable (and not very persistent in this case). - Footnote 58: Lagged parameter coefficient estimates estimated from equation (1) were all insignificant. As was the case for equity returns, actual bond returns were used in equations (3) and (4) in the place of the residuals derived from equation (1). Idiosyncratic means were also added to equations (3) to allow for the individual returns to have different average returns. But these were found to be insignificant and the original specification is used in the analysis. ### France: Bond market estimation results (Table 19 and paragraph 106) - Sample Size: 887 - Likelihood value: 2408.6 - Parameters, estimates, and T-statistics (as presented): - δ1(1) = 0.64* T-statistic = 19.84 - δ2(1) = 0.56* T-statistic = 16.32 - δ3(1) = 0.61* T-statistic = 19.06 - δ4(1) = 0.58* T-statistic = 27.38 - σ1 = 0.67* T-statistic = 35.42 - σ2 = 0.79* T-statistic = 39.40 - σ3 = 0.71* T-statistic = 37.11 - σ4 = 0.26* T-statistic = 8.35 - P0 = 0.92* T-statistic = 43.81 - δ1(0) = 0.38* T-statistic = 3.32 - δ2(0) = 0.26* T-statistic = 2.14 - δ3(0) = 0.42* T-statistic = 3.43 - δ4(0) = 1.58* T-statistic = 7.21 - gc(1) = 0.08* T-statistic = 3.26 - gc(0) = -0.05 T-statistic = 0.54 - Note: *shows significance at 5 percent levels. Variables i = 1, 2, 3, 4 correspond to the returns from French, U.S., U.K., and German bond indices respectively. ### Regime interpretation and cross-market comparisons (paragraphs 107–108) - For bond markets, regime one is, except for Germany, the high-variance regime. - For German bond returns: - The loading factor for the common shock is found to be 1.58 in regime zero, which is greater than the regime one loading factor. - This implies that for German bond returns, the more volatile regime is regime zero rather than one. - German bond returns are generally more correlated to other bond returns in regime zero than in regime one. - Example: German bond market covariance with the French market: - regime zero covariance = 0.60 - regime one covariance = 0.37 - Pair-wise correlation estimates across the U.S., French and UK bond markets are higher in regime one. - The results reveal a lack of regime dependency over the sample period: - As indicated in Figure 19, the estimated probabilities of being in regime zero (the low volatility regime) at time t never climbs above 0.5, implying that volatility and correlation estimates do not jump from one level to another. - Nevertheless, the independent-regime switching specification shows brief, transitory periods in which bond returns are characterized as being in a low volatility and correspondingly low correlation regime (see Figure 20). ### Tests of changing interdependence and interpretation (paragraph 109) - The test with null hypothesis of no change in “economic” interdependence across bond returns yields a test statistic of 144.2, indicating the null can be rejected. - Implication: The degree of interdependence between the four bond markets has changed over time, even after accounting for the effects of changes in volatility. - Interpretation: This supports the hypothesis that bond markets, including the French bond market, have become more highly integrated as globalization effects have taken hold in the late part of the 1990s. - Footnote 59: Technically, hypothesis testing should be carried out to ascertain the correct number of regimes (one versus two) and/or the correct type of switching process (Markov-switching versus independent-switching). However, this would require the calculation of bootstrapping or Monte Carlo methods and lies outside of the scope of this study. ### Monetary transmission: introduction and scope (paragraphs 110–112) - The liberalization of France’s financial sector since the second half of the 1980s should have improved the transmission of monetary policy to the real economy. - The financial landscape in France remains characterized by a large number of idiosyncrasies—many caused by government policies—that affect monetary transmission. - Aim: Provide a comprehensive overview of these idiosyncrasies and their likely or potential impact on the transmission of monetary policy; not exhaustive. - Empirical analysis focuses on two aspects: - Transmission from policy interest rates to the interest rates faced by economic agents. - Interest rate sensitivity of household consumption. - Sectional roadmap: - Section II: anchors paper in literature on cross-country heterogeneity of monetary transmission in the euro-area. - Section III: reviews relevant French idiosyncrasies and discusses potential effects. - Section IV: econometrically quantifies interest rate transmission and the effect of interest rates on private consumption. - Section V: concludes and discusses options for reform. ### Literature overview and transmission channels (paragraphs 113–117) - Four main channels usually identified in the monetary transmission mechanism: - Interest rate channel - Exchange rate channel - Asset price channel - Credit channel - Interest rate channel details: - Substitution effect: modifies relative price of current and future goods and services; induces agents to postpone (contractionary) or advance (expansionary) consumption and investment, provided they do not face liquidity constraints. - Income effect: changes disposable income of households; sign depends on household sector net position (net borrowers/net savers). - Exchange rate channel: - In flexible exchange rate and open capital account, monetary loosening tends to put downward pressure on domestic currency, stimulating exports; but capital flows and other asset market developments can mitigate this. - Asset price channel: - Affects equity prices and Tobin’s Q; impacts firms’ investment incentives via equity issuance. - Wealth effect on household savings and consumption. - Credit channel: - Monetary policy changes affect asymmetric information in bank lending via collateral values and expected cash flows, altering banks’ willingness to lend, especially to small and medium enterprises. - Expectations channel noted as important but not discussed in detail here. - Cross-country empirical approaches: - Macroeconometric models: tend to point to a somewhat smaller reaction to monetary policy in France than in other large euro-area economies. - Example: van Els and others (2001): French economic growth reduced by 0.28 percentage points after a 100 basis point hike sustained for two years; euro area average = -0.38 percentage points. - Structural VARs: suggest relatively higher sensitivity of the French economy but raise cross-country comparability issues. - Mojon and Peersman (2001): impact on GDP of a one standard deviation monetary policy shock in France close to euro-area average; large confidence bands preclude strong cross-country conclusions. - Common findings across models: - Exchange rate effect dominates in the short run. - Domestic channels become more important in the medium run. - Main impact is on investment rather than on consumption, which reacts little and slowly. - Mc Adam and Morgan (2001): out of total impact of -0.31 percent on activity after two years from an interest rate increase, impact on cost of capital = -0.10 percent; direct interest effect on consumption = -0.03 percent; remainder from income, wealth and exchange rate effects. - Peersman and Smets (2001): effect on investment three times larger than effect on GDP. - Consumption responsiveness in France: - Evidence mixed; consumption in France could be even less reactive to monetary conditions than in other euro-area countries. - van Els and others (2001): monetary policy’s impact on consumption in France is stronger than euro-area average in their exercise, noting changes compared with earlier results (BIS, 1995). - VARs by Bouscharain and others (1999): aggregate effect larger for France, but reactivity of private consumption is, with Germany, the lowest of the sample. *Italic: IMF staff selected issues material as provided in the supplied content unit.* ### 118. This section identifies structural and institutional features of the French ### This section identifies structural and institutional features of the French economy and financial sector that could impact monetary policy transmission ### Framework and schematic - The analysis uses the schematic presentation of Kuttner and Mosser (2002), modified for the euro-area/French context (Figure 21). - Main modification: separation of intermediated interest rates (deposit and lending rates) from nonintermediated ones to identify impacts of administered interest rates and idiosyncrasies that affect intermediated rates only. - Figure 21 maps ECB policy instruments through channels labeled A–P to outcomes such as nominal deposit and lending rates, asset price levels, loan supply, aggregate demand (I, C, G, NX), etc. ### Channels most affected by French idiosyncrasies (summary) - The arrows labeled with capital letters in Figure 21 mark channels potentially affected by French idiosyncrasies. Table 23 links specific idiosyncrasies to channels A–P. - The most prominent government interventions interfering with transmission (core effects marked with bold Xs in Table 23) are: - (i) Administered interest rates, the ni-ni, and the usury legislation interrupt transmission from euro-area market rates to nominal deposit and lending rates (Channel A). - (ii) A wide range of government interventions affect the interest rate sensitivity of consumption and investment (Channel L and, to a lesser extent, Channel K). ### Idiosyncrasies and their channel-specific roles (high-level) - Channel A (market interest rates → nominal lending and deposit rates) - Affected by: administered savings schemes, ni-ni, usury legislation, competitive situation in financial system, and presence of publicly owned/non-profit-maximizing institutions. - Historical note: France’s cooperative banks were seen as insufficiently profit-oriented until recently. - Channel B (market interest rates → asset prices) - Slowed by: centralization of deposits, presence of public financial institutions, high transaction costs or delays in asset markets (housing and mortgage market features), administered savings schemes, tax advantages for life insurance and long-term holdings, earmarking of assets. - Channels C and D (bank reserves/money market conditions → bank loan supply) - Responsiveness reduced by: less competitive/market-oriented system, centralization of deposits, public financial institutions, government interventions affecting cost of funding (ni-ni, administered rates). - Channel E (deposit/lending rates deviations → exchange rate) - If deposit rates do not adjust downward with market rates, limited incentive to reallocate toward foreign exchange; scope limited since advent of the euro. - Channel F (euro-area market rates → French market rates) - Generally full and timely adjustment due to close integration, but divergences conceivable in small/domestic-dominated segments or where distortions exist (example: CDC investment preferences). - Channel G (bank rates deviating → asset pricing) - Same idiosyncrasies as Channel B plus usury legislation. - Channel H (first leg monetarist channel) - Idiosyncrasies similar to Channels B and G affect portfolio adjustments when monetary operations change base money relative to other assets. - Channels I and J (broad credit channel: asset values → collateral → borrowing) - Affected by: balance sheet structures, households’ portfolio composition, rate of homeownership, importance of non-usable-as-collateral assets (life insurance), degree of household leverage, housing/mortgage market features, banking structure and behavior, role of public financial institutions, centralization of deposits. - Note: Aggregate household leverage is low in France, so asset price variations do not disproportionately affect free assets. - Channels K and L (interest rate channel: real rates → consumption and investment) - Reduced by “lock-in” features: administered savings schemes, life insurance policies, special tax regimes; also influenced by fixed transaction costs (notary fees, taxes), housing/mortgage market features, competitive situation in banking system, balance sheet structures, and factors that reduce loan-supply flexibility (centralization of deposits, earmarking). - Channel M (exchange rate → aggregate demand) - Relevance depends on openness toward non-euro-area world; France’s relatively low openness reduces this channel’s importance versus rest of euro area. - Channel N (second leg monetarist channel: asset market relative prices → output markets) - Affected by flexibility in reallocating portfolios across financial and real assets and by the structure and openness of the economy. - Channel O (bank loan supply → aggregate demand) - Depends on dependence on bank financing, degree of disintermediation, balance sheet structure of nonbank agents, competition, and constraints steering loan supply (e.g., PEL resources use). - Channel P (wealth → consumption) - Depends on perceivability and convertibility of wealth, importance of wealth vs income, composition of household portfolios; influenced by tax-favored life insurance and housing/mortgage market features. - Across all channels, the structure of GDP and production (sectoral sensitivity, e.g., industry more capital intensive than services) affects aggregate outcomes. ### Econometric analysis: interest rate transmission and consumption (key findings) - Focus: interest rate transmission to retail rates and its effect on aggregate demand, particularly consumption—areas where administratively set interest rates play a large role. - Two most important administratively set interest rates have historically deviated significantly and during sustained periods from the policy rate (see Figure 22). - Empirical results summarized: - Market credit rates are sensitive to administered rates, because administered rates influence banks’ cost of resources. - Half of the opening of the spread between the consumption credit rate and the policy rate between 2000 Q3 and 2003 Q3 could be explained by the lack of adjustment of the Livret A rate to the easing of monetary policy rates. - In line with other studies, consumption is found to be more sensitive to income effects than to substitution effects. - A static computation suggests that over the last three years, up to 3¼ percentage points of potential consumption growth appears to have been forgone because of the slow adjustment of consumption credit rates. ### Data and methodological caveats - Data limitations: - Data kindly provided by the Banque de France only go back to the beginning of the 1990s and are quarterly → relatively low number of observations. - No weighted average series of deposit rates was available; the policy rate was used as a proxy in the consumption analysis. - Structural change caveat: - The French economy and financial system changed substantially over the sample period due to liberalization, privatization, multiple reforms of administered savings schemes and other government interventions, structural reduction in inflation, and EMU adoption → added uncertainty about the validity and present/future relevance of results. *Source: IMF staff section identifying structural and institutional features of the French economy and financial sector affecting monetary policy transmission (as presented in the supplied content)._* ### 137. Most of the short-term bank credit rates in the dataset are sensitive to the spread ### _cr05185 - 137. Most of the short-term bank credit rates in the dataset are sensitive to the spread between the policy rate and the main administered interest rate, likely reflecting the impact of the administered savings schemes on the cost of banks’ resources. ### Short-term bank credit rates: sensitivity to administered rates - Most short-term bank credit rates in the dataset are sensitive to the spread between the policy rate and the Livret A rate, reflecting the impact of administered savings schemes on banks’ cost of resources. - Exception: the Livret A rate does not play a role for the rate on small loans to consumers (see Appendix VI). ### Econometric approach - Three-step evaluation: - Step 1: Test impact of administered interest rates (through the Livret A rate) on six credit market rates using a long-term relationship with specification: policycredit = α + β(policy − LivretA) (spread used to minimize multicolinearity). - Step 2: Estimate full dynamics of interest rate transmission using error-correction models (following Mojon (2001)); long-term coefficients for consumption credit rates are sometimes < 1 and sometimes > 1; for rates on credit to enterprises, coefficients are in all cases closer to 2 than to 1 (estimated equations in Appendix VII). - Step 3: Simulate reaction of the six short-term credit rates to a 100 basis point hike in the policy rate under four assumptions about Livret A adjustment: 1. No change (Livret A unchanged). 2. Adjustment by 50 basis points with a 6-month delay (post-August 2003 system mimic). 3. Full adjustment with a six-month delay. 4. Full and instantaneous adjustment. ### Simulation results (100 basis point shock to the policy rate) - General conclusion: Lack of full and instantaneous adjustment of the Livret A rate hampers monetary transmission (Tables 24 and 25). - Consumer credit rates: - Additional reaction under full and immediate adjustment vs. post-August 2003 system: - After one year: additional 26 to 44 basis points. - After two years: additional 31 to 79 basis points. - Effect on small consumer loans is limited. - Credit to enterprises: - Responsiveness higher than for consumer credits: - After one year: ranges from 75 to 121 basis points. - After two years: ranges from 89 to 129 basis points. - Detailed numerical simulation outcomes are reported in Table 24 (consumer credit rates) and Table 25 (enterprise credit rates) for specific loan categories and timing horizons (3 months, 6 months, 1 year, 2 years, 5 years, long term) under the four Livret A adjustment scenarios. ### Contribution of Livret A non-adjustment to spread widening (2000Q3–2003Q3) - The absence of full adjustment of the Livret A rate to the policy rate explains more than half of the increase in the spread between the consumption credit rate and the policy rate over the last two years. - Quantified contribution: the rise in the spread between the Livret A rate and the policy rate explains 143 basis points of the 236 basis point increase in the spread between the consumption rate and the policy rate between 2000Q3 and 2003Q3 (Table 26). - Note: a large part of the widening is due to other factors. ### Medium- and long-term credit interest rates - None of the long-term credit rates appears sensitive to administered saving rates. - Mortgage loans are not influenced by the PEL rate; medium- and long-term loans to enterprises are not influenced by the Livret A rate. - Long-term credit rates consistently exhibit a high degree of correlation with the long-term market rate proxied by the 10-year government bond rate (Appendix VIII). - The PEL rate has only a marginal impact on short-term dynamics of mortgage rates. - Variable mortgage rates: - Correlated with long-term market rates, with a speed of adjustment marginally higher than for fixed mortgage rates. - Majority of mortgage loans are contracted at fixed rates in France, rendering fixed-rate dynamics relatively more important for monetary transmission. - Since 1998, the difference between variable and fixed mortgage rates has consistently been less than 100 basis points. ### Consumption response to interest rates - Estimation period and specification: - Long-term consumption equation estimated with quarterly data from 1990 to 2003 using real disposable income, real policy interest rate, and the spread between the consumption credit rate and policy rate as explanatory variables. - Preferred estimation reported as Equation (1) in Table 27. - Key empirical findings: - The level of the policy rate is not significant in the long-term consumption equation. - The spread between the consumption credit rate and the policy rate is significant and negatively associated with consumption (the spread matters). - Interpretation: - If all rates move simultaneously, consumption is not affected (income and substitution effects offset). - If consumption credit rate does not adjust to a change in the policy rate, a relaxation of monetary policy can dampen consumption (negative income effect and muted substitution effect). - Given imperfect and not instantaneous transmission from policy rate to consumption credit rate in practice, the overall impact of a relaxation of monetary policy on consumption in France tends to be slightly negative. - Quantified macro effect of incomplete transmission: - The long-term estimated equation suggests that because banks’ credit rates did not adjust fully to the monetary loosening, about 3¼ percentage points of potential consumption growth could have been forgone over the last two to three years. ### Role of the Livret A rate in consumption equations - The Livret A rate does not appear to directly influence consumption when added to the preferred regression (its spread with the policy rate is not significant in Equations (2) and (5) of Table 27). - Over the longer 1980–2003 sample (Equation (6) in Table 27), the spread between Livret A and the policy rate appears significantly positive, reflecting historical context in the 1980s when Livret A was often below the policy rate and real and nominal dynamics differed. ### Concluding remarks and policy options - Econometric evidence: administered rates exert a significant influence on some bank credit rates, slowing monetary transmission and reducing the impact of policy rate changes. - Policy recommendation 1: instantaneous adjustment of the Livret A rate and other administered rates to changes in ECB rates is argued for; the August 2003 formula (adjusts every six months and potentially only by half) helps but still hampers transmission. - Policy recommendation 2: reassess other government interventions and policies that may bias bank pricing, notably: - The obligation to provide checks free of charge as quid pro quo for zero interest-bearing checking accounts (ni-ni policy) — consider elimination. - Inertia in setting deposit and credit PEL rates — consider more rapid, preferably automatic, adjustment of PEL rates to the monetary environment and long-term market rates. - Eliminate or reduce absolute interest rate commitments in new PEL contracts (for example, define commitments relative to market rates at loan origination). - Reassess usury legislation, as in its current form it could interfere with transmission in case of rapid policy rate changes. - Additional objective: aim for a better functioning mortgage market. *Source: IMF staff calculations and analysis as presented in the IMF document.* ### introduction of home equity loans and reforms of the system of mortgage liens. Finally, a ### _cr05185 - introduction of home equity loans and reforms of the system of mortgage liens. Finally, a ### Public intervention in the financial sector - Government interventions affect both liability (deposit) and asset sides of the financial system’s balance sheet. - Main interventions on the liability side: administered savings schemes, differential tax treatment of savings products, the ni-ni requirement, and usury legislation. - Main interventions on the asset side: centralization of deposits to be invested by a state-owned entity, restrictions on the use of some administered deposits, and differential tax treatment of investments. - Continued presence of government-owned financial institutions (La Poste and the Caisse des Dépôts et Consignations) provides the government a foothold in financial intermediation. ### Administered savings - Definition: government-designed savings products with features set by the government along seven dimensions: (i) eligibility requirements; (ii) determination of the remuneration; (iii) tax treatment; (iv) quantitative deposit limits; (v) withdrawal rules; (vi) distribution channels; and (vii) the destination and/or purpose of the collected funds. - Popularity and scale: - Administered savings schemes encompass half of all resident bank deposits in France, for an amount in excess of a third of GDP. - Most popular schemes: Plan d’Épargne-Logement (PEL) in volume and the Livret A in number. - Livret A: - Historically set in practice by the government; since July 2003 the remuneration rate to be set automatically at the average of the inflation rate and the ECB’s short-term interest rate, plus 25 basis points. - Adjustments are planned to take place every six months, starting on August 1, 2004. - The Banque de France implements the arrangement but may deviate under exceptional circumstances and is expected to keep the Livret A rate positive in real terms at all times. - Under the new system a lag of between 2 and 3 months and 8 and 9 months will remain before a change in the ECB’s policy rates is reflected in the Livret A rate. - Inflation may become the sole determinant of the Livret A rate when the ECB’s policy rate becomes negative in real terms. - Plan d’Épargne-Logement (PEL): - Largest administered savings scheme in volume, representing one fifth of total bank deposits. - Characteristics: requires monthly minimum saving, deposit interest rate and lending rate are locked in for the duration of the plan at the rate established when the PEL is set up. - December 2002 reform eliminated the interest-rate subsidy for plans that do not result in a loan, reducing attractiveness and new inflows, but existing stock ensures PEL remains important for years. - Lending rate varies with amounts saved under the plan and time the plan has been maintained. ### Deposit composition and table figures (as presented) - Amount (In billions of euros) / Percent of Total Deposits / Percent of GDP: - Administered deposits: 528.7 49.3 34.0 - Livrets A: 112.0 10.4 7.2 - Livrets bleu: 15.5 1.4 1.0 - Codevi: 43.2 4.0 2.8 - Livrets jeunes: 5.7 0.5 0.4 - Livrets d'Epargne Populaire (LEP): 53.8 5.0 3.5 - Comptes d'Epargne-Logement (CEL): 36.0 3.4 2.3 - Plans d'Epargne-Logement (PEL): 216.5 20.2 13.9 - Plans d'Epargne Populaire (PEP): 46.0 4.3 3.0 - Of which: centralized deposits: 217.0 20.2 14.0 - Demand deposits: 342.1 31.9 22.0 - Other deposits: 202.4 18.9 13.0 - Total deposits: 1,073.2 100.0 69.1 - Sources noted: Banque de France and IMF staff calculations (Table 28, December 2003). ### The ni-ni requirement - Definition: provision that banks can neither remunerate demand deposits nor charge for providing checkbooks; in practice banks forgo any charge for the use of checks. - Effects: - Together with households’ tendency to maintain high levels of demand deposits, the ni-ni means the majority of nonadministered deposits is excluded from potential remuneration. - Combination of ni-ni and administered savings schemes implies that more than 80 percent of deposits is insensitive to changes in market and policy rates, at least in the short run. - Alters monetary transmission: reduces sensitivity of incomes, cash flows, and banks’ resource costs to interest rate changes; interferes with the trade-off between cash and demand deposits when policy rates change. - In euro-area context, uniform reserve requirement for deposits with maturity up to two years reduces the risk that shifts among deposit types materially change banks’ free reserves. - Legal change noted: On October 5, 2004, the European Court of Justice ruled that the prohibition to remunerate demand deposits was contrary to the freedom of establishment within the EU; this ruling is expected to mark the end of the ni-ni. ### Tax treatment of savings products and life insurance - Tax incentives favor long-term savings instruments: life insurance policies, PEL, Plan d’Epargne en Actions (PEA), Plan d’Epargne Populaire (PEP), Plan d’Epargne-Retraite (PERP), etc. - Favorable treatment typically requires holding the product for a minimum period, usually between 4 and 8 years; life insurance commonly requires over 8 years. - Life insurance specifics: - Unit-linked policies represented 18 percent of life insurance policies in 2003. - Life insurance assets composition in 2003: fixed-income instruments 75 percent, equity 22 percent, real estate 3 percent. - Many life insurance contracts specify a minimum level of periodic savings and insurers often charge costs up front, which locks in savings behavior and penalizes early changes. - Insurance companies can use hidden reserves to smooth returns; the linkage between market interest rates/asset prices and returns on life insurance is less tight than for bonds, stocks, or bank deposits. - Borrowing against a life insurance contract is possible and provides some flexibility. - Effect on monetary transmission: - Favorable tax treatment and penalties for early withdrawal reduce households’ ability to adjust savings in response to monetary policy changes, potentially slowing asset price adjustments, wealth effects, and the speed of savings-behavior changes. ### The usury legislation - Rule: caps the interest rate financial institutions can charge on loans to individuals or overdrafts to companies at four thirds of the average rate observed in the relevant loan category. - Implementation: Banque de France monitors market rates and calculates quarterly average rates; application of the four thirds ratio provides the usury rate for the next quarter—backward-looking. - Consequences: - At four thirds of observed rates, the usury rate is often binding, especially at very low market interest rates, and may not allow adequate pricing of risk. - Backward-looking nature can make it more binding when monetary policy tightens, because it can take 6 months or more before market rate changes are fully reflected in the usury rate. - Possible outcomes: - If banks refuse to lend when they cannot price risk adequately, lending to risky borrowers should decline when monetary policy tightens. - If banks continue lending without fully charging risk, monetary policy may be blunted because lending rates for risky borrowers adjust only with a delay of 3–7 months. - Effect asymmetry: loosening of monetary policy should not face the same downward adjustment delay. ### Centralization of deposits - A significant share of funds from administered savings products is centralized in the Fonds d’Epargne (FdE). - Scope: centralization applies to all funds collected in the Livrets A and Livrets bleu, virtually all other deposits collected by La Poste, and parts of funds collected in Codevi and some other accounts. - Scale as of end-2003: - €217 billion centralized in the FdE, out of about €529 billion in administered deposits and €1,073 billion in bank deposits. - Distribution remuneration: collecting institutions receive a remuneration to cover collection costs, expressed as a percentage of their stock of centralized deposits of a given instrument. - Monetary transmission implications: - Centralization can create instability in banks’ deposit bases depending on the spread between the Livret A rate and market rates. - If monetary policy loosens and market rates fall while the Livret A rate remains unchanged, savers may shift towards administered savings and banks may experience a net loss of resources. - A rising spread between the Livret A rate and the policy rate will put downward pressure on banks’ resources; a declining spread will have the opposite effect, partially offsetting normal effects of monetary policy on banks’ liquidity. *Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2005/_cr05185.pdf* ### 165. In practice, however, the liquidity effects have been contained because banks have ### _cr05185 - 165. In practice, however, the liquidity effects have been contained because banks have ### Liquidity effects, Livret A, and bank behavior - Banks have sought to maintain the rates on their other deposit products at levels that are competitive with the Livret A rate, so deviations between the Livret A and market rates have impacted more on banks’ cost of resources than on their liquidity. - The most sophisticated and proactive savers are often the ones who have already reached the limit of their Livret A deposits, which limits liquidity effects. ### Fonds d’Epargne (FdE): objectives, asset composition, and implications - Management and objectives - The FdE is managed by the Caisse de Dépôt et Consignation (CDC) in function of certain public policy objectives, the most important being the financing of social housing projects; funds are also invested in loans to local public authorities and some other projects. - Half of the assets of the FdE are invested in financial markets, mainly in government securities. - Asset composition (Table 29: France: Assets of Fonds d’Epargne, end-2002) - Loans: 111.5 (49.2 percent) - Housing-related loans: 85.2 (37.6 percent) - Equipment loans: 6.0 (2.6 percent) - Other loans: 20.3 (9.0 percent) - Securities: 112.5 (49.6 percent) - Shares: 5.2 (2.3 percent) - Public-sector securities and assimilated: 66.5 (29.3 percent) - Bonds and other fixed-income securities: 40.8 (18.0 percent) - Other assets: 2.7 (1.2 percent) - Total assets: 226.7 (100.0 percent) - Transmission and behavioral implications - Demand for loans from the FdE is largely determined by public policy needs and is relatively insensitive to interest rates. - Supply of funds adjusts passively to demand; surplus funds are invested in financial markets. - An increased spread between administered and market rates that leads to increasing inflows into the FdE is unlikely to increase loans from the FdE; social housing companies may borrow elsewhere, reducing FdE lending. - This mechanism could lead to a reduction in the supply of bank credit in times of a loosening of monetary policy not accompanied by a reduction in the Livret A rate, as depositors move savings toward centralized deposits while FdE borrowers are discouraged from taking new loans. - Result: increased share of deposits invested in euro-area financial markets instead of being used to finance loans. ### Restrictions on use of administered deposits (earmarking) - Most deposits collected through administered savings schemes can only be used for specific purposes: - Funds centralized in the FdE: only for certain public policy objectives. - PEL and CEL deposits: only invested in mortgage loans, some other real estate-related assets, and energy-saving projects. - Codevi deposits: only used for loans to small and medium sized companies. - Effects of earmarking - Reduces asset-side flexibility in the banking system balance sheet (including FdE). - May create incentives not aligned with changes in monetary policy. - Example: If monetary policy is loosened and households increase deposits in older, higher-yielding PELs, banks may increase mortgage lending potentially at the expense of other loans—blunting the overall impact of monetary policy changes, especially if favored sectors have inelastic demand for loans. ### Differential tax treatment of investments - The tax system favors certain investments (examples in text): - Tax breaks for equity investments in small and medium sized enterprises. - Differences in tax treatment between bonds and shares. - Tailored tax breaks to promote specific investments. - Effects - Tax incentives increase the cost of reallocating investments in response to changing circumstances and reduce economic agents’ ability to react to monetary policy adjustments, especially where tax advantages are contingent on maintaining investments for a prescribed period. - Note: This is not unique to France; many countries (example cited: United States encourages housing investment via tax system). ### Banking sector structural features and balance sheet statistics - Reduced role of deposits - Noncentralized deposits available to the banking system amount to only 51 percent of GDP in France, against 91 percent in the rest of the euro-area. - Lower overall bank intermediation - Lending by French banks (excluding CDC) amounts to about 69 percent of GDP, as against 101 percent for the banks of the rest of the euro-area. - Aggregate loan-to-deposit ratios - Aggregate loan-to-deposit ratio in France: 1.14 - Aggregate loan-to-deposit ratio in the rest of the euro-area: 1.11 - Excluding the CDC improves the French ratio to about 1.35. - Selected balance-sheet structure highlights (Tables 30 and 31) - Table 30 (In percent of total assets, end-2002) — French banks vs Non-French Euro-area banks vs Euro-area banks: - Interbank loans to euro-area MFIs: 23.0 / 20.8 / 21.3 - Credit to euro-area non-MFI borrowers: 27.8 / 38.3 / 36.0 - Of which: lending for house purchases: 8.4 / 12.4 / 11.5 - Of which: lending by CDC: 2.7 / 0.0 / 0.6 - Euro-area securities: 25.0 / 17.1 / 18.9 - Deposits of euro-area non-MFI residents: 24.5 / 34.3 / 32.1 - Of which: demand deposits: 7.0 / 11.5 / 10.5 - Of which: deposits centralized at the CDC: 5.8 / 0.0 / 1.3 - Debt securities: 20.8 / 18.1 / 18.7 - Of which: money market funds: 7.3 / 1.6 / 2.8 - Capital and reserves: 6.8 / 5.6 / 5.9 - Table 31 (In percent of GDP, end-2002) — French banks vs Non-French Euro-area banks vs Euro-area banks: - Interbank loans to euro-area MFIs: 62.9 / 55.2 / 56.8 - Credit to euro-area non-MFI borrowers: 76.0 / 101.3 / 95.9 - Of which: lending for house purchases: 23.0 / 32.9 / 30.8 - Of which: lending by CDC: 7.3 / 0.0 / 1.6 - Euro-area securities: 68.4 / 45.4 / 50.3 - Non-euro-area and other assets: 66.0 / 62.9 / 63.6 - Total assets: 273.3 / 264.8 / 266.6 - Interbank borrowing from euro-area MFIs: 72.1 / 54.7 / 58.5 - Deposits of euro-area non-MFI residents: 66.9 / 90.9 / 85.7 - Of which: demand deposits: 19.1 / 30.5 / 28.1 - Of which: deposits centralized at the CDC: 15.9 / 0.0 / 3.4 - Debt securities: 56.8 / 47.9 / 49.8 - Of which: money market funds: 20.0 / 4.1 / 7.5 - External and other liabilities: 58.9 / 56.4 / 56.9 - Capital and reserves: 18.7 / 14.9 / 15.7 - Implication: The relatively low level of bank intermediation implies that transmission channels depending on the banking system will tend to be weaker in France than in the rest of the euro-area. ### Competitive situation and binding customer strategies - Market structure and bank strategies - Significant concentration: sector dominated by six large banks. - Banks use strategies to bind customers: aggressive pricing on mortgage loans, sale of product packages, branch network and personalized service. - Fees and commissions constitute a significant part of bank profitability; clients face disincentives to changing banks. - Implication: These factors may contribute to relatively low price flexibility and interest-rate sensitivity in the retail banking market, reducing the impact of the interest rate channel of monetary policy. ### Role of remaining public financial institutions (CDC and La Poste) - Caisse de Dépôt et Consignation (CDC) - Collects certain legally protected deposits (€37 billion at end-2003). - Own funds: €12.5 billion. - Investments include long-term loans for urban renewal, public housing, credit to “social economy” projects, seed capital for SMEs, and equity stakes in small and medium-sized companies. - Public-policy orientation of investments reduces sensitivity of CDC’s asset portfolio to interest rates and the economic environment, potentially reducing sensitivity to monetary policy. - CDC manages a significant part of France’s pension funds and maintains large stakes in the Caisses d’Epargne and the insurance company CNP. - La Poste - Provides limited banking services; can collect most deposits but can only provide PEL or CEL-based mortgage loans and associated unregulated mortgage loans. - Most deposits collected are centralized in the FdE and managed by the CDC. - Branch network: 13,000 post offices providing financial services — largest branch network in French banking industry. - Advantages: reduced distribution costs per financial product (network shared with mail activities), certain tax and other advantages. - Market impact: strong competition on specific products; major contributor to success of administered savings products; presence reinforces centralization impact. - Past behavior: offered very competitive mortgage rates and gained market share in part because remuneration was based on volume of loans sold, with loans remaining on FdE balance sheet and La Poste receiving fixed remuneration. - Reforms affecting La Poste (convention signed February 26, 2004; effective retroactively from January 1, 2004) - Remuneration change: from fixed commission to receiving profits on its mortgage portfolio (portfolio remains on FdE balance sheet). - La Poste requested to distribute mortgage loans not connected to a PEL and consumer loans; government conditionally agreed to the first request provided: - Financial services reorganized into a separate banking entity, licensed by the CECEI, supervised by the Commission Bancaire, subject to same legal and regulatory framework as other banks. - New postal bank must have its own accounts separate from postal services to increase transparency and avoid cross-subsidization. - With the exception of Livret A and a few other administered savings schemes, the postal bank will manage its own balance sheet rather than passing most deposits to the FdE. ### Idiosyncrasies related to asset markets — Housing market - Characteristics - Relatively low turnover and some rigidity in financing arrangements. - Typical French households buy a house only once or twice in their lifetime, if at all. - Home ownership rates are relatively low in France (Eurostat figure cited in source material). - Transaction costs and charges for real property sale (notary and other costs) - Registration rights (tax): 4.89 percent - Notary honorarium: between 0.825 (on the amount of the transaction above a threshold of €16,769.40) and 5 percent (on the part of the transaction amount up to €3,049) - Diverse costs: between €458 and €1,525 - Real estate agent commissions: typically 4–10 percent, depending on transaction size. *Source: IMF staff report content from the provided PDF excerpt.* ### 181. In terms of financing, the French mortgage system (hypothèque) is relatively ### _cr05185 - 181. In terms of financing, the French mortgage system (hypothèque) is relatively ### Mortgage financing: costs, procedures, and enforcement - Establishing a mortgage typically includes: - Notary honorarium between 0.66 percent and 3.9 percent of the loan amount (0.66 percent for very large amounts; 3.9 percent for very small amounts). - A tax of 0.61 percent (waived for some loans, including PEL loans). - A few hundred euros in various costs. - In most cases, these costs total between 1.5 percent and 5.0 percent of the loan amount. - If the mortgaged property is sold before two years after the end of the original loan maturity, an additional mainlevée d’hypothèque fee between 0.5 percent and 2.0 percent is due. - Realizing a mortgage: - Requires a court procedure. - Courts “tend to sympathize with debtors and are reluctant to evict families from their house,” according to bankers. - The procedure routinely takes more than a year. - Practical consequences: - Mortgages are relatively expensive and inflexible. - Mortgages do not provide the lender with full security. ### Alternative guarantee mechanisms (caution and guarantee companies) - Banks often accept a third-party guarantee (caution) instead of a mortgage or PPD. - Typical guarantor: specialized financial company (example: Crédit Logement), owned by several large banks. - Fee structure for guarantee companies: - An upfront fee paid by the borrower, typically consisting of two parts: a regular fee and participation in the guarantee company’s reserve fund. - Of the reserve fund contribution, 75 percent is reimbursed at the end of the guarantee arrangement (regardless of whether this end comes on or ahead of schedule). - Advantages of the caution relative to mortgage/PPD: - Typically costs about a third less than a PPD. - Does not penalize early repayments or sale of property before loan maturity. - Allows a more flexible approach to borrower financial difficulties; guarantee companies seek solutions in consultation with borrower and lender. - If sale of the property is required, guarantee companies allow the borrower to sell voluntarily rather than forcing an auction sale. ### Other mortgage loan costs and refinancing rules - Lender one-time processing cost (frais de dossier): typically about 1 percent of the loan amount. - Insurance for death or disability: premium about 0.4 percent of the loan amount; in many cases not adjusted for declining principal. - Refinancing: - Happens in France but significantly less frequently than in the United States. - High fixed costs make refinancing profitable only after significant declines in interest rates. - By law, early repayments must be allowed with a penalty cap equal to the lower of 3 percent of the amount repaid early or six months of interest. ### Mortgage market composition and transaction costs - 14 percent of mortgage loans in France are at variable rates (ECB (2003) observation). - France’s level of mortgage debt: 22 percent of GDP in 2001. - Within the euro-area, only Greece, Italy, and Finland scored lower. - Notary transaction cost examples (Table 32: Typical Transaction Costs Charged by the Notary on the Sale of Existing Residential Real Estate): - For amount 10,000: Transaction costs 1,794; Taxes 499; (In percent of transaction amount) 17.9 - For amount 20,000: Transaction costs 2,458; Taxes 998; 12.3 - For amount 50,000: Transaction costs 4,251; Taxes 2,495; 8.5 - For amount 100,000: Transaction costs 7,240; Taxes 4,990; 7.2 - For amount 250,000: Transaction costs 16,205; Taxes 12,475; 6.5 - For amount 500,000: Transaction costs 31,146; Taxes 24,950; 6.2 - For amount 1,000,000: Transaction costs 61,030; Taxes 49,900; 6.1 - For amount 10,000,000: Transaction costs 598,933; Taxes 498,100; 6.0 - Low turnover and high transaction costs: - Contribute to housing being perceived more as a service-providing asset than a financial asset, reducing wealth effects. - Fixed-rate prevalence and expensive refinancing make household mortgage interest costs largely insensitive to market rate changes. - Absence of cash-out options and high transaction costs limit households’ ability to use housing as a financial adjustment tool. ### Comparative mortgage-market features (Low, Sebag-Montefiore, and Dübel (2003)) - France characterized by: - low profitability (the lowest in their sample); - significant government involvement; - low loan-to-value (LTV) ratios; - relatively short loan terms (partly due to requirement to pay off loan within working lifetime and expected prior savings); - good product variation in some respects; - limited product variation in other respects (limited availability of second mortgages; no possibility for cash-out mortgage borrowing/home equity loans); - mortgage lending focused on people in their 30s and 40s; - relatively low transparency; - the existence of usury legislation; - branch-driven distribution; - high transaction costs. ### Households’ financial structure, leverage, and asset composition - French households are net creditors; income and balance sheet leverage are low compared to peers. - Lower leverage implies: - Income effects of interest rate changes tend to be stronger (once fully reflected in asset and liability rates). - Interest rate changes are less likely to create balance-sheet problems among French households. - France: Overall balance sheet of the household sector, end-2002 (Provisional) (In billions of euros; total assets and liabilities sum to 2,846.4): - Cash: 32.6 (Assets) - Deposits: 865.7 (Assets) - Debt securities: 63.3 (Assets); 0.3 (Liabilities) - Credits: 19.9 (Assets); 587.3 (Liabilities) - Equity: 680.7 (Assets) - Mutual funds: 241.3 (Assets) - Insurance claims: 835.1 (Assets) - Other: 107.9 (Assets); 171.4 (Liabilities) - (Financial) net worth: 2,087.4 - Total: 2,846.4 (Assets) and 2,846.4 (Liabilities) - Cross-country household debt and financial wealth, 2000 (select figures from Table 36; in percent of gross disposable income or percent): - France: Financial liabilities 54; Financial wealth 308; Leverage 18 - Netherlands: 177; 583; 30 - United Kingdom: 116; 456; 26 - Italy: 45; 338; 13 - Spain: 72; 286; 25 - Germany: 104; 222; 47 - United States: 90; 405; 22 - Japan: 100; 439; 23 - Composition of household financial assets, 2000 (Table 38, in percent of total financial assets): - Currency and deposits: France 30.4 - Directly-held listed shares: France 5.0 - Directly-held unlisted shares: France 19.8 - Mutual funds (other than money market mutual funds): France 9.0 - Insurance claims: France 23.3 (of which unit-linked policies 4.9) - Pension funds: France 1.5 - Liabilities composition (Table 38, in percent of total liabilities): - Mortgage debt: France 61.0 - Consumer credit: France 20.0 - Professional loans: France 19.0 ### Administered savings products (overview highlights) - Livret A: - Eligibility: Individuals or legal entities; maximum one Livret A or livret bleu per person. - Remuneration: Determined by BdF on a government-designed formula (2.25%). - Tax treatment: Tax exempt. - Deposit limits: €15,300 (individuals); €76,500 (legal entities). - Withdrawal: Freely withdrawable; interest paid until end of most recent quinzaine. - Livret bleu: - Same as Livret A (2.25%); 1/3 of interest is taxed, but tax paid by Crédit Mutuel. - Deposit limits: €15,300 (individuals); €76,500 (legal entities). - Codevi: - Eligibility: Individuals with tax residence in France; maximum one per person and two per household. - Remuneration: Same as Livret A (2.25%). - Tax treatment: Tax exempt. - Deposit limit: €4,600. - Withdrawals: Freely withdrawable; interest paid until end of most recent quinzaine; withdrawals only in branch where account is held. - LEP (Livret d'Épargne Populaire): - Eligibility: Low-income individuals with tax residence in France; maximum one per person and two per household. - Remuneration: Set by the government (3.25%). - Tax treatment: Tax exempt. - Deposit limit: €7,700. - CEL (Compte Épargne Logement): - One per individual. - Remuneration: Set by government. Currently 1.5%, plus interest rate subsidy of 0.75% paid by the state if and when a loan is taken up. - Tax: Tax exempt except social taxes (10%). - Deposit limit: €15,300. - Gives right, after 18 months, to a mortgage loan of up to €23,000 at a predetermined low interest rate. - PEL (Plan d'Épargne Logement): - One per individual. - Remuneration: Set by government. Currently 2.5% plus an interest rate subsidy of 1% paid by the state if and when a loan is taken up. - Tax: Tax exempt except social taxes (10%). - Deposit limit: €61,200 (excl. capitalized interests). - Term: Min 4, max 10 year term. Gives right to mortgage loan up to €92,000 at a predetermined interest rate. - Notes on administered products: - Quinzaine = standard two-week period used for calculation of interest. - As of December 31, 2001, deposit limits reported in euros. *Source: _cr05185 PDF chapter content provided.* ### 4.25 percent to 3.25 percent on August 1, 2004. ### _cr05185 - 4.25 percent to 3.25 percent on August 1, 2004. ### Interest rate subsidy and PEL contracts - For plans opened before December 12, 2002, the interest rate subsidy is unconditional and paid on a regular basis, together with the interest paid by the bank. - For plans opened after that date, the interest rate subsidy is paid as a lump sum at the closing of the PEL-based mortgage loan. ### Data sources and variable definitions (Table 40) - Consumption: Quarterly Accounts - INSEE, 1978:01 - 2003:04, Variable TD_P3M_D1. - Real disposable income: Quarterly Accounts - INSEE, 1978:01 - 2003:04. Deflated by consumption price. Variable TD_B6_S143 (TD_P3M_D3/TD_P3M_D1). - Inflation rate: Quarterly Accounts - INSEE. Derived from consumption price index. - Nominal policy rate: International Financial Statistics, 1974:01 - 2004:01. French repurchase agreement rate up to 1998:04; European overnight interbank rate from 1999:01. - 10-Year government bond rate: International Financial Statistics, 1970:01 - 2003:04. - Nominal Livret A rate: Banque de France 1/, 1971:01 - 2004:01. - Nominal PEL rate: Banque de France 1/, 1971:01 - 2004:01. (Deposit rate for new contracts.) - Nominal consumption credit rates: Banque de France 1/, 1990:02 - 2004:01. Definitions derived: Rate for short-term loans below €1524 (from the usury legislation), Overdraft rate, Rate for short-term loans above €1524, Rate for medium- and long-term loans. - Nominal mortgage rates: Banque de France 1/, 1990:02 - 2004:01. Rates for mortgages at fixed and variable rates. - Nominal credit rates to enterprises: Banque de France 1/, 1984:02 - 2004:01. Definitions derived: Overdraft rate (from the usury legislation), Discount rate, Rate for other short-term loans, Rate for medium- and long-term loans. - All real rates: Deflated using contemporary inflation. - Note: The most recent observations of these series are available on the Banque de France website (http://www.banque-france.fr/). ### Unit-root (ADF) test results (Appendix V) — ADF statistic for H0: ρ = 0 (presence of a unit root) - Critical values: 5 % Critical Value = -2.89 (with trend), 5 % Critical Value = -1.95 (without trend). - Consumption (y): Lags = 1; Estimation Period 1979:01 - 2003:04; ADF with trend = -1.34; ADF without trend = -3.27. - Real disposable income (y): Lags = 3; Estimation Period 1979:01 - 2003:05; ADF with trend = -1.23; ADF without trend = -2.71. - Inflation (∆y without trend): Lags = 3; Estimation Period 1979:01 - 2003:04; ADF without trend = -5.59. - Real policy rate (∆y without trend): Lags = 2; Estimation Period 1979:01 - 2003:04; ADF without trend = -8.22. - Real consumption credit rate (∆y without trend): Lags = 1; Estimation Period 1990:02 - 2003:04; ADF without trend = -5.31. - Real Livret A rate (∆y without trend): Lags = 3; Estimation Period 1979:01 - 2003:04; ADF without trend = -7.09. - Nominal policy rate (∆y without trend): Lags = 4; Estimation Period 1979:01 - 2003:04; ADF without trend = -7.09. - Nominal consumption credit rate (∆y without trend): Lags = 8; Estimation Period 1990:02 - 2003:04; ADF without trend = -3.79. - Nominal mortgage rate (∆y without trend): Lags = 8; Estimation Period 1990:02 - 2003:04; ADF with trend = 0.13; ADF without trend = -3.10. - Nominal Livret A rate (∆y without trend): Lags = 8; Estimation Period 1990:02 - 2003:04; ADF with trend = -0.37; ADF without trend = -6.28. - Nominal PEL rate (∆y without trend): Lags = 8; Estimation Period 1979:01 - 2003:04; ADF with trend = -0.24; ADF without trend = -6.35. - Source: IMF staff calculations. - Notes: - 1/ The number of lags is determined through the maximization of the Schwarz criteria. - 2/ ADF statistic is for H0: ρ = 0, presence of a unit root. - ADF regression forms used: ∆y_t = constant + trend + ρ y_{t-1} + Sum(δ_i ∆y_{t-i}) + ε_t and ∆y_t = constant + ρ y_{t-1} + Sum(δ_i ∆y_{t-i}) + ε_t. ### Long-term equations for short-term market credit interest rates (Appendix VI, Phillips-Loretan nonlinear estimator) - General formulation: Market Credit Interest Rate = Cointegration Relation + Sum (Lagged Variables) + Sum (Forward Variables). Correction term ρ = (Market Credit Interest Rate(-1) – Constant – Cointegration Relation(-1)). - Selected coefficient estimates (figures in parenthesis are t-statistics; significance at the 5-percent level signaled with an asterisk): - Overdraft Rate for Loans Below €1,524: - Real policy rate coefficient = 0.23 (0.15) - Spread between the Livret A rate and the policy rate = -0.72 (-0.34) - Constant/other long-run coefficients include 0.92 (12.3)*, 0.93 (13.0)*, 0.84 (9.61)*, 0.67 (7.09)*, 0.61 (5.29)*, 0.70 (6.34)*, 0.54 (3.74)* across specifications. - Estimation period = 1990:03-2003:03 (degrees of freedom 35; standard error of dependent variable 2.23; standard error of estimate 0.51; DW 2.02). - Discount Rate / Overdraft Rate for Other Short-Term Loans / Overdraft Rate Above €1,524: - Real policy rate coefficients range from 0.9 to 2.34 with t-statistics reported (e.g., 0.9 (2.88)*; 1.89 (2.29)*; 2.26 (11.2)*; 2.33 (7.38)*; 2.34 (4.86)*; 1.87 (4.97)*). - Spread between Livret A and policy rate shows mixed signs and significance (e.g., 1.66 (1.41), 1.49 (5.51)*, 1.76 (3.93)*, 1.69 (2.43)*, 1.05 (1.93)*). - Estimation periods generally 1990:03-2003:03 or 1990:02-2003:03; degrees of freedom and standard errors vary by equation (examples: degrees of freedom 43, 35, 36; standard error of dependent variable 1.77–2.7; standard error of estimate 0.26–0.5; DW 1.64–2.02). - All variables in levels. Figures in parenthesis are t-statistics; significance at the 5-percent level is signaled with an asterisk. ### Error-correction models for market credit interest rates (Appendix VII) - Short-term market credit interest rates estimated with error-correction specifications; conventions: - r_policy = Nominal Policy Rate - r_LivA = Nominal Livret A Rate - spread_LivA = Spread between the Livret A Rate and the Policy Rate - Error-correction model forms reported for: - Rate for consumer loans below €1,524 (including estimation periods, standard errors, DW statistics; coefficients and short-term dynamics reported in equation form). - Overdraft rate on consumer accounts. - Rate for consumer loans over €1,524. - Discount rate for enterprises. - Overdraft rate on company accounts. - Rate for other short-term loans to enterprises. - Medium- and long-term market credit interest rates modeled with conventions: - R_GB = Nominal 10-Year Government Bond Rate - R_PEL = Nominal Deposit Rate on PEL Contracts - Error-correction specifications reported for: - Fixed mortgage rate. - Variable mortgage rate. - Rate for medium and long term loans to enterprises. - Note: T-statistics in the long-term dynamic cannot be interpreted; their significance was assessed in previous steps (Appendices VI and VIII). ### Long-term equations for medium- and long-term market credit interest rates (Appendix VIII, Phillips-Loretan nonlinear estimator) - Selected coefficient estimates for mortgage loans and term loans to enterprises (figures in parenthesis are t-statistics; significance at the 5-percent level signaled with an asterisk): - 10-year government bond rate coefficients: - Mortgage loans at fixed rate = 0.94 (2.95)* - Mortgage loans at variable rate = 0.22 (0.10) - Medium- and long-term loans to enterprises = 0.31 (0.34) - Mortgage loans at fixed rate (alternative specification) = 1.34 (16.0)* - Mortgage loans at variable rate (alternative) = 1.32 (6.99)* - Term loans to enterprises (alternative) = 1.44 (8.41)* - Spread between the PEL rate and the 10-year government bond rate: -0.78 (-1.37) and -2.10 (-0.52) in some specifications. - Spread between the Livret A rate and the 10-year government bond rate: -1.83 (-1.14) in one specification; other specifications show coefficients 0.81 (10.90)*, 0.93 (10.10)*, 0.87 (12.40)*, 0.80 (8.46)*, 0.91 (11.35)*, 0.87 (13.20)*. - Estimation periods and diagnostics: - Estimation periods generally 1990:02/03/04 - 2003:03 (examples: 1990:04-2003:03, 1990:02-2003:3, 1990:03-2003:3). - Degrees of freedom: 34–43 across specifications. - Standard error of dependent variable: 2.16–2.45. - Standard error of estimate: 0.17–0.28. - DW: 1.77–2.35. - Two sets of results reported: "With Administered Rates" and "Without Administered Rates". - All variables in levels. Figures in parenthesis are t-statistics; significance at the 5-percent level is signaled with an asterisk. ### Key empirical implications - Short-term market credit rates show significant long-run relationships with the nominal policy rate and with spreads involving administered rates (Livret A and PEL) in many specifications; several long-run coefficients are statistically significant at the 5-percent level. - Medium- and long-term market credit rates, including mortgage rates, are strongly related to the 10-year government bond rate in multiple specifications, with coefficients often close to or above 0.9 and frequently significant at the 5-percent level. - Administered-rate spreads (Livret A and PEL relative to policy or government bond rates) enter long-run relations with varying signs and statistical significance across loan types and maturities. *Source: IMF staff calculations, _cr05185 - 4.25 percent to 3.25 percent on August 1, 2004.* --- _Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2005/_cr05185.pdf_