## _cr13184

## Source details

**Canonical URL:** [_cr13184](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13184.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13184.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13184.pdf.json)

---

### I. Introduction — key findings and vulnerabilities
- Housing price history and resilience:
  - France’s housing market declined by about 9 percent from its pre-crisis peak during the 2008–09 crisis before rebounding since mid-2009 to surpass its pre-crisis peak.
- Drivers of demand and price resilience:
  - Population growth; historically relaxed lending terms (low interest rates, longer maturities, high LTV ratios); perception of real estate as a safe haven; muted supply response.
  - Tax incentive schemes introduced during the 2008–09 crisis (e.g., the Scellier regime) may have exacerbated imbalances.
- Misalignment and potential correction:
  - Estimates of misalignment at end-2011 suggest prices could be overvalued by 10 percent to 20 percent, with some studies suggesting valuation gaps over 50 percent.
- Risk mitigation factors:
  - Limited availability of home equity loans.
  - Lending criteria mostly based on income with a typical qualifying ratio of one-third.
  - Most mortgages have fixed rates and there are no nonrecourse loans.
  - Household debt levels remain low on average compared to other advanced countries.
- Emerging vulnerabilities:
  - LTV ratios are high compared to other advanced countries; over one-fifth of borrowers have DTI ratios above 35 percent.
  - Residential mortgage loss rates remain structurally low but have risen since 2008.
  - Commercial real estate (CRE) problem loans increased sharply: share of European CRE loans repaid at maturity fell from close to 50 percent in 2009 to 28 percent by November 2011; in late 2011 another 9 percent of loans were repaid late.
- Policy stance:
  - Current dynamics do not call for immediate macroprudential action but require close monitoring, enhanced data collection, and development of macroprudential tools.
  - Macroprudential measures can be targeted by region, borrower category, and/or property value and are preferable to economy-wide monetary or fiscal tightening for addressing real estate booms.

### II. Preventive policy actions and macroprudential recommendations
- Rationale for early intervention:
  - Frictions and externalities: (a) leverage, (b) wealth effects, and (c) supply-side effects.
  - Leverage concerns: first-time borrowers, households in regions with above-average house price inflation (e.g., Ile de France), and rental investors.
- Suggested instruments:
  - Caps on LTV ratios, possibly differentiated by property value and/or whether the loan benefits from state subsidies or guarantees.
  - DTI limits on selected categories.
  - Increased risk weights or provisioning requirements (not recommended as first-line at present).
- Implementation requirements:
  - Strengthen appraisals and reporting on new mortgage origination.
  - Enhance micro-level data disaggregation by region and household category.
  - Prepare to experiment, adjust limits quickly, and coordinate at the European level to limit circumvention.
  - Macroprudential authority should communicate ex ante its analysis and rationale for action or inaction.
- Institutional developments:
  - CORÉFRIS (created October 2010) is responsible for surveillance of macroprudential risks; met first in February 2011 and concluded credit played a significant role in the recent boom (CORÉFRIS, 2012).

### III. Housing market developments, valuation indicators, and scenarios
- Crisis and recovery dynamics:
  - House prices fell by just 9 percent from peak-to-trough in 2008–09; prices recovered and hit new record highs in 2011.
- Valuation metrics at end-2011:
  - House prices-to-income ratio: about 27 percent above its average since 1996.
  - House prices-to-rent ratio: 30 percent above the average since 1996.
  - France ranks alongside Canada, Norway, and Sweden on overvaluation indicators.
- Model-based misalignment estimates:
  - Typical estimates: 10–20 percent.
  - One study (modeling house prices as sum of expected future rents) yields over 50 percent.
  - Cited study ranges:
    - Antipa and Lecat (2012): 15–20
    - Borgy, Clerc, and Renne (2012): very low estimated probability of a housing boom at end-2010
    - Hott and Jokipii (2012): >50
    - Igan and Loungani (forthcoming): 10–20
    - S&P (2012): <15
    - Igan and Loungani (2010): 10–20
- Transactions and construction:
  - Transactions for existing homes: more than 820,000 per year pre-crisis to less than 600,000 in 2009 (a 12 percent decline, after a decrease of 17 percent in 2008).
  - Real housing prices declined by 10.3 percent between end-2007 and late 2009, before recovering by 6.3 percent through March 2012.
  - Housing starts fell from over 460,000 per year pre-crisis to little more than 300,000 in 2009 (a 16 percent decline, after a decrease of 14 percent in 2008); recovered to over 400,000 in 2011 and early 2012 but remained below pre-crisis levels and declined since February 2012.
  - Policy support in 2009 included social housing firms authorized to buy 30,000 dwellings; SFEF issuance amounted to €77 billion at the end of summer 2009.
- FSAP scenarios and projections:
  - FSAP central scenario:
    - Predicts a sharp slowdown in house price growth in 2012–13.
    - Household leverage will likely stabilize or rise at a more moderate clip.
    - Unemployment remains high in 2012–2013 at around 10 percent.
    - Under central scenario a soft house price landing is probable.
  - FSAP adverse scenario:
    - Rising interest rates, sluggish activity, and unemployment rising to 11 percent in 2013–14 could trigger a downturn.
    - House prices projected to correct by up to 5 percent over 2012–13.
  - Alternative expectations:
    - Notaires d’Ile de France expect prices to drop by 5 percent in Ile de France and by up to 10 percent in France as a whole.
    - S&P expects a 15 percent price decline over the next two years as interest rates rise by 50 bps (S&P, 2012).
- Recent market signals:
  - Decreased transaction activity and new construction; BdF’s May 2012 lending survey signals a drop in mortgage demand since late 2011.
  - Sellers cutting prices: Paris apartment prices declined 1.1 percent in 2012: Q1 vs previous quarter; France as a whole declined 0.7 percent (quarter/quarter).

### IV. Financial sector exposures, stress tests, and covered bonds
- Mortgage market features and bank vulnerabilities:
  - Most mortgages have fixed rates; no nonrecourse loans.
  - Almost half of loans originated in 2011 had down payments of 5 percent or less.
  - Residential mortgage loss rates rose since 2008 but remain structurally low.
  - French CRE nonperforming loans: 8 percent in 2011 (down from 8.6 percent in 2010; up from 1.7 percent in 2007).
- Aggregate exposures and bank concentration:
  - Retail mortgages total €832 billion (41 percent of GDP) and represent nearly 41 percent of domestic loans to the private sector.
  - Bank-level mortgage exposures in 2010 (in billions of euros): Crédit Agricole 250; BPCE Group 225; Crédit Mutuel Group 172; Société Générale SA 99; BNP Paribas SA 62; La Banque Postale 37; Crédit Immobilier de France 32; HSBC France 3.
  - Aggregate mortgage exposure shares (2010): BPCE Group 26%; Crédit Agricole 28%; Crédit Mutuel Group 20%; Société Générale SA 11%; BNP Paribas SA 7%; La Banque Postale 4%; Crédit Immobilier de France 4%; HSBC France 0%.
- Covered bond market:
  - Volume outstanding at end-2010: €156 billion (18 percent of the euro area; 9 percent of the worldwide market).
  - Covered bonds accounted for around 20 percent of outstanding residential mortgage loans extended by French banks (ECBC, 2011).
  - Covered bond share in wholesale funding mix rose to 56 percent during the stress period.
  - Spreads rose by 100 bps since mid-May 2011; yield levels 112 bps below their past five-year average as of June 2012.
- Stress-test illustration (aggregate residential mortgage stress):
  - Assumed stress: 100 percent increase in current NPLs of residential mortgages.
  - Exposures considered: €824 billion at end-2011.
  - Current impaired loan coverage: 30 percent.
  - LGD assumption: 33 percent.
  - Implied potential losses under stress: €3-4 billion.
  - Impact on regulatory Tier 1 capital adequacy ratios: decrease by about 0.1 percentage point (to 10.9 percent).
  - Caveat: the assumed downturn is largely confined to a price correction with no indirect income effects or concurrent interest rate hike.

### V. Household mortgage performance, investor exposures, and vulnerabilities
- Household performance and arrears:
  - Unemployment at a 12-year high; arrears have been low so far.
  - Aggregate NPL ratio for household mortgages: 1.25 percent in 2010–11 (up from 0.9 percent in 2007).
  - NPL ratio for floating rate loans: about 2.6 percent in 2010–11 (up from 1.1 percent in 2007).
- Borrower composition and affordability:
  - Market shifting toward older and more affluent borrowers since end-2009: share of borrowers below 35 declined from 52 percent to 49 percent over two years; share with income above five times minimum wage rose from 20 percent to 22½ percent.
  - After accounting for lower interest rates and longer maturities, housing affordability has decreased by 20 percent in the past 10 years (Friggit, 2010).
  - New fixed-rate housing loans charge about 4.7 percent on average since mid-2011 (up from 4.1 percent at end-2010; below 10-year mean of 5.1 percent and down 260 bps since 1998).
  - Average DTI ratio rose by 5.4 percentage points during 2001–2011; average DTI ratio was about 30.6 percent in March 2012.
- Investor (buy-to-let) exposures:
  - Lending for investment property: around 17 percent of housing loans in 2010 (up from 12 percent in 2001).
  - Investor-owned properties represent around 20 percent of total (INSEE 2009–10 survey).
  - Buy-to-let investors accounted for over 60 percent of properties sold by developers in 2010 (up from 40 percent in 1995).
  - Investors concentrate in Paris, Alps Valley, Lubéron, Côte d’Azur; in Paris investors accounted for 29 percent of transactions in 2012: Q1 (Century 21).
- Distributional heterogeneity and risk concentrations:
  - Average LTV ratio around 80 percent in March 2012; average LTV by borrower type (March 2012): first-time borrowers 83 percent; buy-to-let investors 89.5 percent; other borrowers 73 percent.
  - Share of loans with LTV >100 percent among first-time borrowers: 12.7 percent in March 2012 (down about 4 points since end-2010).
  - Share of borrowers with debt service payments exceeding 35 percent of income: 18.8 percent in 2001; 33.8 percent in 2011; 27.3 percent at end-2011.
  - Mortgage insurance covered 61.6 percent of all loans in 2011 (up from 34.3 percent in 2001).
  - Most loans fixed-rate: 89 percent of loans granted in 2011; flexible-rate share in 2011 originations under 10 percent; hybrid-rate mortgages 1.5 percent of 2011 origination.
- Localized vulnerabilities and higher-risk institutions:
  - Institutions with high shares of floating-rate loans or CRE exposures are more vulnerable.
  - CIF example: floating-rate loans 55 percent of total; potential losses under doubling of NPLs €525 million; CIF Common Equity Tier 1 ratio decrease by 3.2 points to 11.4 percent; CIF NPL ratio 4.1 percent at end-2011; CIF LTV 87 percent.
- Macroeconomic channels and negative equity:
  - CORÉFRIS estimate: a 15 percent correction would result in less than 20 percent of 2006–10 cohorts in negative equity.
  - ACP data: loans with down payments of 5 percent (15 percent) or less represented about one-half of total loans issued in 2011.
  - Negative equity would not necessarily trigger widespread default given full recourse loans and emphasis on income-based underwriting.

### VI. Key statistics (exact figures cited)
- Mortgage lending y/y in May 2012: 5 percent (France); 0.4 percent (euro area); 1.1 percent (Germany).
- Average DTI ratio: 30.6 percent in March 2012.
- Average term to maturity at end-2011: 18.5 years (up 4.4 years since 2001; down 0.8 years from 2008 peak).
- Average loan amount at end-2011: €108,000.
- Average LTV ratio: around 80 percent in March 2012 (lower by 4 bps compared to end-2010).
- LTV ratios by borrower type (March 2012): first-time borrowers 83 percent; buy-to-let investors 89.5 percent; other borrowers 73 percent.
- Share of loans with LTV >100 percent among first-time borrowers: 12.7 percent in March 2012 (down about 4 points since end-2010).
- Share of borrowers with debt service payments exceeding 35 percent of income: 18.8 percent in 2001; 33.8 percent in 2011; 27.3 percent at end-2011.
- Share of mortgage loans with fixed rates: 89 percent of loans granted in 2011.
- Mortgage insurance coverage: 61.6 percent of all loans in 2011.
- Retail mortgages total: €832 billion (41 percent of GDP); nearly 41 percent of domestic loans to the private sector.
- French covered bonds outstanding at end-2010: €156 billion (18 percent of the euro area; 9 percent of the worldwide market).
- Covered bond share in wholesale funding mix of French banks: increased to 56 percent during stress period.
- Covered bond funding accounts for around 20 percent of outstanding residential mortgage loans extended by French banks (ECBC, 2011).
- Covered bond spreads rose by 100 bps since mid-May 2011.
- Yield levels: 112 bps below their past five-year average, as of June 2012.
- Bank mortgage exposure by bank in 2010 (in billions of euros): Crédit Agricole 250; BPCE Group 225; Crédit Mutuel Group 172; Société Générale SA 99; BNP Paribas SA 62; La Banque Postale 37; Crédit Immobilier de France 32; HSBC France 3.
- Aggregate mortgage exposure shares (2010): BPCE Group 26%; Crédit Agricole 28%; Crédit Mutuel Group 20%; Société Générale SA 11%; BNP Paribas SA 7%; La Banque Postale 4%; Crédit Immobilier de France 4%; HSBC France 0%.
- Real losses on French residential mortgages: 0.0116 percent of total loans outstanding at end-2011.
- Aggregate NPL ratio for household mortgages: 1.25 percent in 2010–11.
- NPL ratio for floating rate loans: about 2.6 percent in 2010–11.

### VII. Implementation challenges, design principles, and recommended operational steps
- Implementation challenges and design principles:
  - Need appropriate calibration to avoid threshold effects and circumvention.
  - Differentiated LTV/DTI caps by region or borrower category useful but require enhanced micro-data and frequent appraisals or independent valuation industry.
  - Be ready to adjust limits nimbly to avoid unduly hampering credit availability when interest rate cycle turns.
- Access, affordability, and political economy:
  - Restricting credit could hamper desirable credit deepening; ensure sufficient supply of affordable housing alternatives (e.g., rentals).
  - Effective communication of rationale and decision (action or inaction) is essential to address political economy concerns.
- Circumvention risks and coordination:
  - Main circumvention risk: lending shifting to nonregulated intermediaries or cross-border relocation of mortgage activities.
  - Regional coordination may be needed to prevent circumvention through cross-border banking and to address co-movement in housing cycles.
- Policy recommendations and operational actions:
  - No immediate macroprudential action is warranted given incipient demand downturn since late 2011; continue close monitoring.
  - Prepare a macroprudential framework (e.g., LTV limits) and enhance data reporting requirements at household, transaction, and mortgage origination levels.
  - CORÉFRIS should continue monitoring and communicate more frequently; expand monthly bank reporting to the ACP and build a database of mortgage origination details for ex post evaluation.
  - Consider a uniform LTV ratio of, say, 80 percent as a simple-to-administer first line of defense, complemented by tighter LTV limits for selected categories (e.g., investors) and geographic zones as warranted.
  - Caps on DTI ratios could complement LTV ceilings but may need very low settings to be binding; additional capital/provisioning requirements for real estate may be counter-productive now.

*Source: IMF staff report content as provided in the source PDF.*

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

### _cr13184 - References

### Tables and Appendices (inventory)
- Tables listed:
  - EPRA Global Listed Real Estate Indices: Yield and Market Value Capitalization, May 2012.
  - Concordance between French Housing Cycles and Selected Advanced Countries’ Cycles.
- Appendix Tables listed:
  - International Comparison of Taxation Regimes.
  - Total Return by Asset Class, 1990–2011.
  - Asset Class Correlation Matrix (Q1 1990–Q1 2011—Total Return Series).
  - Use of Macroprudential Tools by Country.
- Appendix Figures listed:
  - Building Investment.
  - Fundamental Factors Influencing House Prices.
  - Mortgage Exposures of French Banks.

### Glossary (selected acronyms and terms)
- ACP: Autorité de Contrôle Prudentiel
- BdF: Banque de France
- Bps: Basis points
- BPCE: Banque Populaire et Caisse d'épargne
- CACES: Collège des autorités de contrôle des entreprises du secteur financier
- CAS: Centre d’Analyse Stratégique
- CRH: Caisse de Refinancement de l’Habitat
- CIF: Crédit Immobilier de France Group
- CGEDD: Conseil Général de l’Environnement et du Développement Durable
- CORÉFRIS: Conseil de la Régulation Financière et du Risque Systémique
- CRE: Commercial real estate
- DTI: Debt service to income
- ECB: European Central Bank
- ECBC: European Covered Bond Council
- EMF: European Mortgage Federation
- EPRA: European Public Real Estate Association
- EU: European Union
- Euribor: Interbank lending rate in the eurozone
- FPI: Federation of French Property Developers
- FSAP: Financial Sector Assessment Program
- GDP: Gross Domestic Product
- IMF: International Monetary Fund
- INSEE: Institut National de la Statistique et des Études Économiques
- IPD: Investment Property Databank
- LCL: Le Crédit Lyonnais
- LGD: Loss-given default
- LTV: Loan-to-value
- MFIs: Monetary and financial institutions
- MI: Mortgage insurance
- NPLs: Nonperforming loans
- PAS: Prêt Accession Sociale
- PD: Probability of default
- PRR: Price-to-rent ratio
- PTZ+: prêt à taux zéro plus
- OECD: Organization of Economic Co-operation and Development
- OF: obligations foncières
- OH: obligations de financement
- PIR: Price-to-income ratio
- SCF: société de crédit foncier
- SFEF: société de financement de l'économie française
- SFH: société de financement de l’habitat
- SGACP: Secrétariat Général de l’ACP
- SGCB: Secrétariat Général de la Commission Bancaire
- S&P: Standard and Poor’s
- UCITS: Undertaking for Collective Investment in Transferable Securities

### I. Introduction — Key findings and vulnerabilities
- Housing price history and resilience:
  - France’s housing market declined by about 9 percent from its pre-crisis peak during the 2008–09 crisis before rebounding since mid-2009 to surpass its pre-crisis peak.
- Drivers of demand and price resilience:
  - Population growth; historically relaxed lending terms (low interest rates, longer maturities, high LTV ratios); perception of real estate as a safe haven; muted supply response.
  - Tax incentive schemes introduced during the 2008–09 crisis (e.g., the Scellier regime) may have exacerbated imbalances.
- Misalignment estimates and potential correction:
  - Estimates of misalignment at end-2011 suggest prices could be overvalued by 10 percent to 20 percent, with some studies suggesting valuation gaps over 50 percent.
- Risk mitigation factors:
  - Limited availability of home equity loans.
  - Lending criteria mostly based on income with a typical qualifying ratio of one-third (monthly principal and interest payment cannot exceed one-third of monthly income).
  - Most mortgages have fixed rates and there are no nonrecourse loans.
  - Household debt levels remain low on average compared to other advanced countries.
- Emerging vulnerabilities:
  - Past loosening of lending criteria increased risks: LTV ratios are high compared to other advanced countries; over one-fifth of borrowers have DTI ratios above 35 percent.
  - Residential mortgage loss rates remain structurally low but have risen since 2008.
  - Commercial real estate (CRE) faces a sharp increase in problem loans: share of European CRE loans repaid at maturity fell from close to 50 percent in 2009 to 28 percent by November 2011; in late 2011 another 9 percent of loans were repaid late.
- Policy stance:
  - Current dynamics do not call for immediate macroprudential action but require close monitoring, enhanced data collection, and development of macroprudential tools.
  - Macroprudential measures can be targeted by region, borrower category, and/or property value and are preferable to economy-wide monetary or fiscal tightening for addressing real estate booms.

### II. Preventive policy actions and macroprudential recommendations
- Rationale for early intervention in real estate booms:
  - Several frictions and externalities strengthen the case for early policy intervention: (a) leverage, (b) wealth effects, and (c) supply-side effects.
  - Leverage is a primary concern for: first-time borrowers, households in regions with above-average house price inflation (e.g., Ile de France), and rental investors.
- Suggested macroprudential instruments:
  - Caps on LTV ratios, possibly differentiated by property value and/or whether the loan benefits from state subsidies or guarantees to avoid overly restricting access to credit for first-time borrowers.
  - DTI limits on selected categories.
  - Increased risk weights or provisioning requirements.
- Implementation requirements and challenges:
  - Effective LTV caps require strengthening appraisals (currently infrequent).
  - Designing and calibrating instruments, and evaluating ex post their effectiveness, will require strengthening reporting requirements on new mortgage origination.
  - Maximizing effectiveness requires openness to experimentation, ability to adjust limits quickly, overcoming political economy obstacles, and coordination at the European level to limit circumvention through cross-border banking activities.
  - The designated macroprudential authority should communicate ex ante to the public and affected financial institutions its analysis of risks and rationale for action or inaction.
- Institutional developments:
  - CORÉFRIS, created in October 2010 to replace CACES, is responsible for surveillance of macroprudential risks and met for the first time in February 2011.
  - CORÉFRIS analysis concluded credit played a significant role in the recent boom (CORÉFRIS, 2012).
  - The governor of Banque de France reminded banks of the appropriateness of prudent mortgage lending practices; CORÉFRIS asked ACP to undertake close monitoring of risks including monthly reporting of banks’ lending practices but stopped short of recommending intervention.

### III. Housing market developments and valuation indicators
- Crisis and recovery dynamics:
  - During the 2008–09 crisis, house prices fell by just 9 percent from peak-to-trough compared with more than 30 percent in the United States and Ireland and 18 percent in the United Kingdom.
  - Prices recovered and hit new record highs in 2011.
- Valuation metrics at end-2011:
  - House prices-to-income ratio: about 27 percent above its average since 1996.
  - House prices-to-rent ratio: 30 percent above the average since 1996.
  - France is among the leaders on overvaluation based on both indicators, alongside Canada, Norway, and Sweden.
- Model-based misalignment estimates:
  - Typical model-based estimates of price misalignment range from 10–20 percent.
  - One study modeling house prices as the sum of expected future rents yielded a larger gap (over 50 percent).
  - Cited model results and ranges from various studies include:
    - Antipa and Lecat (2012): 15–20 (percent implied)
    - Borgy, Clerc, and Renne (2012): very low estimated probability of a housing boom at end-2010
    - Hott and Jokipii (2012): >50 (percent implied)
    - Igan and Loungani (forthcoming): 10–20 (percent implied)
    - S&P (2012): <15 (percent implied)
    - Igan and Loungani (2010): 10–20 (percent implied)
- Transactions and construction dynamics (Box 1 highlights):
  - Number of transactions for existing homes: more than 820,000 per year before the crisis to less than 600,000 in 2009 (a 12 percent decline, after a decrease of 17 percent in 2008).
  - Real housing prices declined by 10.3 percent between end-2007 and late 2009, before recovering by 6.3 percent through March 2012.
  - Housing starts fell from over 460,000 per year pre-crisis to little more than 300,000 in 2009 (a 16 percent decline, after a decrease of 14 percent in 2008).
  - Housing starts recovered to over 400,000 in 2011 and early 2012 but remained below pre-crisis levels and declined since February 2012.
  - Policy support measures in 2009 included social housing firms authorized to buy 30,000 dwellings and temporary measures such as doubling PTZ+ ceiling and creation of SFEF; SFEF issuance amounted to €77 billion at the end of summer 2009.
  - Housing loans increased by 3.7 percent in 2009, after 7.5 percent in 2008.
  - Average rate on new fixed-rate housing loans decreased starting April 2009 and reached a low of 4.1 percent at end-2010.

### IV. Financial sector exposures and implications
- Mortgage market features:
  - Most mortgages have fixed rates; no nonrecourse loans.
  - A substantial share of loans originated in 2011 had down payments of 5 percent or less (almost half of loans originated in 2011).
- Bank vulnerabilities and CRE stress:
  - Residential mortgage loss rates rose since 2008 but remain structurally low.
  - French CRE nonperforming loans: decreased slightly to 8 percent in 2011 from 8.6 percent in 2010 and were up from 1.7 percent in 2007.
  - IPD France index composition at end-2011: office market 51.3 percent, retail 22.8 percent, residential 15 percent, logistics 7.4 percent, others 3.5 percent.
- Policy implications:
  - Limits on LTV ratios, combined with DTI limits on selected categories, could be considered.
  - Effective LTV caps will require strengthening appraisals and reporting on new mortgage origination.

*IMF staff note: prepared by Hélène Poirson with research assistance from John De Guzman (both IMF/EUR).*

### 9.      The pre-crisis misalignment detected by these methods has not unwound in

### 9.      The pre-crisis misalignment detected by these methods has not unwound in

### House price misalignment and historical corrections
- A simple affordability price-to-income ratio (PIR) as of end-2006 shows a negative relationship with realized change in house prices since that date across a sample of OECD countries.
- The staff’s model-based estimate, using the estimated price gap as of end-2007, predicts reasonably well the actual path of house prices in 2007–10, though France (and Australia, Canada, Norway, and Sweden) experienced a less severe downturn than predicted.
- Possible explanations for France’s milder correction:
  - A policy-induced rebound (see Box 1 in the source).
  - The correction may still be underway and additional house price declines could be expected over the next few years.
- Timing and length of a correction remain uncertain; several near-term factors continue to support demand.

### Speculative activity, investor share, and market segments
- Investor purchases of new properties have risen since 1995.
- Lending for the purchase of investment property:
  - represented around 17 percent of housing loans in 2010,
  - up from 12 percent in 2001 (ACP, 2012a).
- INSEE 2009–10 household wealth survey suggests investor-owned properties represent around 20 percent of total.
- Among households with real estate wealth (the wealthiest overall), 30 percent own a secondary residence or a property investment in 2010, up 3 points from 2004 (INSEE, 2011).
- In 2010, buy-to-let investors accounted for over 60 percent of properties sold by real estate developers (up from 40 percent in 1995) per FPI data (Moody’s, 2012).
- Investors concentrate in geographic zones: Paris, Alps Valley, Lubéron, and Côte d’Azur; in Paris investors accounted for 29 percent of transactions in 2012: Q1 (Century 21).

### Borrower composition, affordability, and leverage
- Since end-2009 the market is shifting toward older and more affluent borrowers:
  - Share of younger borrowers (below 35) declined from 52 percent to 49 percent over two years (Observatoire du Crédit Logement/CSA).
  - Share of higher-income borrowers (above five times the minimum wage) rose from 20 percent to 22½ percent.
- Affordability trends:
  - After accounting for lower interest rates and lengthening maturities, housing affordability has decreased by 20 percent in the past 10 years (Friggit, 2010).
  - New fixed-rate housing loans charge interest of about 4.7 percent on average since mid-2011 (up from 4.1 percent at end-2010, below 10-year mean of 5.1 percent and down 260 bps since 1998).
  - Average DTI ratio rose by 5.4 percentage points during 2001–2011.
  - Average DTI ratio was about 30.6 percent in March 2012 (banks’ monthly reporting to the ACP).
- Contribution to increased borrowing capacity during 2003–10:
  - Decrease in interest rates estimated to have contributed about 40 percent.
  - Increase in revenues contributed 32 percent.
  - Increase in debt payments to revenue contributed 21 percent (CORÉFRIS, 2012).
- Alternative calculation holding DTI at one-third: lower interest rates (direct plus indirect via longer duration) account for 55 percent to 60 percent of increase in borrowing capacity since 2001 (CAS, 2011).
- Increase in borrowing capacity can explain about 60 percent of the rise in existing home prices through 2011 (CAS, 2011).

### Household debt, asset allocation, and investment incentives
- Household debt to disposable income has risen nearly 30 points in the past decade but remains well below levels in most other developed countries.
- Mortgage debt to GDP is lower than in most other advanced countries, suggesting scope for further mortgage market deepening.
- Deposit rates in real terms turned negative in early 2010, providing incentives for retail and institutional investments into real estate.
- Since 1990, stocks have cumulated a higher return than real estate investment property but with higher volatility; in risk-adjusted terms property investment (IPD indices) has outperformed stocks (Appendix Table 2).
- Listed real estate (EPRA France) has relatively high correlation with stocks but offers high yield and liquidity; France accounts for almost a quarter of the Europe EPRA index.

### Supply constraints and demographics
- Slow supply response has contributed to price increases:
  - No surge in residential construction investment compared with some other advanced countries.
  - Structural constraints: scarcity of buildable land in dense metropolitan areas and regulatory barriers.
- Housing stock barely kept up with population growth over 2005–09:
  - France added around 350,000 new housing units per year according to INSEE.
  - Estimated need was 350,000 to 400,000 units per year to maintain existing occupancy rates (Jacquot, 2007).
- Demographic projections and implications:
  - Population aging and de-cohabitation imply annual increases in number of households by 2030 of 227,000 to 255,000 depending on scenario.
  - Jacquot (2007) estimates 320,000 to 370,000 new dwellings needed each year over the next decade.
- Land prices increased 31 percent (in euros per square meter) during 2006–10.

### Price fundamentals and mean reversion risk
- At the aggregate index level, house prices show low-frequency trends associated with increase in construction costs and population density.
- Since 2000, house prices increased faster than construction costs and caught up with demographics.
- PIR and PRR at national level are well above long-term averages; model-based estimates indicate potential for protracted mean-reversion (Loungani, 2010).
- Cyclical supports such as low interest rates remain supportive but could reverse as banks raise rates to preserve margins in the face of increased financing costs (S&P, 2012).

### FSAP scenarios, near-term prospects, and projected corrections
- FSAP central scenario:
  - Predicts a sharp slowdown in house price growth in 2012–13 consistent with signs of softening demand.
  - Household leverage will likely stabilize or rise at a more moderate clip.
  - Unemployment remains high in 2012–2013 at around 10 percent.
  - Housing market widely seen as at or near its peak.
  - Under central scenario a soft house price landing is probable.
- FSAP adverse scenario:
  - Rising interest rates, sluggish economic activity, and higher unemployment (rising to 11 percent in 2013–14) likely to trigger a downturn.
  - House prices projected to correct by up to 5 percent over 2012–13.
  - Correction could be stronger if accompanied by a shift in investor expectations and asset allocation.
- Alternative external expectations cited:
  - Notaires d’Ile de France expect prices to drop by 5 percent in Ile de France and by up to 10 percent in France as a whole.
  - S&P expects a 15 percent price decline over the next two years as interest rates rise by 50 bps, constraining borrowing capacity (S&P, 2012).

### Market indicators and recent policy changes
- Signs of softening demand:
  - Decreased transaction activity while new construction is also decreasing.
  - BdF’s May 2012 lending survey signals a drop in demand for mortgages since late 2011 (BdF, 2012).
  - Sellers have started cutting prices: in Paris apartment prices declined 1.1 percent in 2012: Q1 vs previous quarter; France as a whole declined 0.7 percent (quarter/quarter comparisons referenced).
- Policy/incentive changes in 2012 reducing real estate incentives:
  - Restriction of the PTZ+ to new home purchases in 2012.
  - Removal of the Scellier incentive for buy-to-let investors.
  - Doubling the holding period for exemption of capital gains tax on secondary properties to 30 years effective February 1, 2012.
- Credit supply and underwriting:
  - Banking regulator (ACP) increased scrutiny of housing due to concerns about NPLs and borrower risk profile.
  - Banks may have tightened lending standards (e.g., decreased mortgage maturity and LTV ratios) in response to more difficult financing conditions.

*Source: IMF staff analysis as presented in the provided PDF content unit.*

### 21.      The discussion above suggests that banks are exposed to a near-term price

### _cr13184 - 21.      The discussion above suggests that banks are exposed to a near-term price

### Central scenario: near-term price correction and bank losses
- Banks are exposed to a near-term price correction that is likely to translate into only limited credit losses under the FSAP’s central scenario.
- Residential mortgage exposures:
  - Banks would be exposed to higher defaults on residential mortgage loans and losses if house prices fall.
  - The 2008–09 experience and institutional features of the mortgage market provide some reassurance that defaults resulting directly from the price decline would be small.
- Commercial real estate (CRE) exposures:
  - Losses on CRE exposures are likely to be higher based on past episodes, although banks are comparatively less exposed to this sector.
- Vulnerabilities to a more adverse scenario:
  - A more adverse scenario of house price declines accompanied by a shock to incomes and a hike in interest rates (as in the FSAP’s adverse scenario) would likely result in greater stress.
  - Variable rate mortgages remain marginal but are predominantly originated by some specialized lenders, making them more vulnerable to higher interest rates while they are already struggling with a higher-than-average increase in problem loans.

### Tax policy–related elements (Box 2)
- Policy changes since end-2010 and end-2011:
  - Mortgage interest deductibility was abolished for loans originated after 2010 and replaced by an updated version of the 2010 zero interest rate loan program (PTZ+).
  - Other measures withdrawn at end-2011 include: cessation of the PTZ+ for existing homes, a reduction in the 2009 Scellier scheme for buy-to-let purchases, and capital gains tax reform.
- Characteristics of the 2011 PTZ+:
  - Unlike the 2010 version, the 2011 PTZ+ has no income restrictions and the amounts borrowed can be higher in higher-priced regions and when the dwelling purchased meets certain energy standards (Poirson, 2011).
- Implications:
  - Abolition of mortgage interest deductibility contributes to reducing incentives for household leverage, but prospective buyers of new homes continue to benefit from the PTZ+.
  - Administered savings plans to facilitate accumulation of down payments play a more limited role in France; similar schemes have been withdrawn elsewhere historically.
- Policy recommendation:
  - Efforts to curb the cost of tax expenditures and reduce the distortions they can cause in the economy should continue.

### Structural factors limiting mortgage market risks
- Mortgage credit growth and market positioning:
  - Mortgage lending rose 5 percent year-on-year in May 2012 compared to 0.4 percent in the euro area as a whole and 1.1 percent in Germany.
  - Mortgage lending growth has generally kept up with or outpaced overall domestic credit growth by a significant margin since 2006.
  - Home-owning rates and mortgage debt-to-GDP ratio remain lower in France than in many other advanced countries.
- Underwriting and borrower sustainability:
  - Underwriting criteria emphasize sustainability of borrower income rather than collateral value.
  - Banks tend to set a debt service (including mortgage service payment) to income ratio equal to one-third.
  - The average DTI ratio declined since its June 2011 peak to reach 30.6 percent in March 2012, but remains higher than at end-2010 (ACP, 2012a).
  - The average term to maturity at end-2011 was 18.5 years, up 4.4 years since 2001 but 0.8 years down from its 2008 peak; the average loan amount was €108,000 (ACP, 2012b).
  - The average LTV ratio is around 80 percent in March 2012, lower by 4 bps compared to end-2010 but higher than the advanced country mean.
  - Leverage is decreasing in the CRE market; some market participants reduced leverage ratio targets in late 2011, achieved mainly via asset disposals.
- Distributional heterogeneity:
  - Average LTV ratio in Ile de France is 74 percent versus 81 percent in the provinces.
  - Loans to first-time borrowers have average LTV of 83 percent; buy-to-let investors 89.5 percent; other borrowers 73 percent.
  - LTV ratios exceeding 100 percent represented 12.7 percent of loans to first-time borrowers in March 2012, down about 4 points since end-2010 (ACP, 2012a).
  - DTI ratios: Ile de France 31.5 percent; provinces 30.3 percent.
  - Share of borrowers with debt service payments exceeding 35 percent of income rose from 18.8 percent to 33.8 percent during 2001–11 but declined to 27.3 percent at end-2011 (ACP, 2012b).
- Interest rate risk and guarantees:
  - Most mortgage loans (89 percent of loans granted in 2011) carry fixed rates and nearly all flexible rate loans are capped (ACP, 2012b).
  - The share of flexible rate loans in successive vintages has been divided by more than three since 2005 to under 10 percent of total loans granted in 2011.
  - Hybrid-rate mortgages accounted for 1.5 percent of origination in 2011.
  - Loans are full recourse. Bank penalties for early repayment are capped at 3 percent of the remaining principal.
  - Mortgage insurance (MI) covered 61.6 percent of all loans in 2011 (up from 34.3 percent in 2001).
  - Around 5 percent of loans do not have any form of guarantee.
  - Insurance against death and disability is mandatory; unemployment insurance is optional and less than 10 percent of borrowers subscribe.

### Covered bond market (Box 3) — structure, role, and risks
- Role of the mortgage market for banks:
  - Mortgage market enables banks to issue mortgage covered bonds, supporting funding when senior unsecured debt markets were impaired.
  - The share of covered bonds in the wholesale funding mix of French banks increased to 56 percent during the stress period.
  - Covered bond funding has accounted for around 20 percent of outstanding residential mortgage loans extended by French banks (ECBC, 2011).
  - Volume of French covered bonds outstanding at end-2010 amounted to €156 billion (18 percent of the euro area and 9 percent of the worldwide market).
- Characteristics and legal frameworks:
  - France has covered mortgage bonds: obligations foncières (OFs), obligations de financement de l’habitat (OHs), billets hypothécaires de la Caisse de Refinancement de l’Habitat (CRH).
  - OFs and OHs are unsubordinated senior secured obligations issued by specialized subsidiaries; SCFs are “bankruptcy remote.”
  - Minimum coverage ratio for OFs is 102 percent.
  - 2010 law created sociétés de financement de l’habitat (SFHs) with refinancing up to 80 percent limit for residential loans and 60 percent for commercial loans; minimum coverage ratio must exceed 102 percent.
  - CRH requirements: only residential mortgages with LTV capped at 80 percent or guaranteed loans are eligible; coverage ratio must exceed 125 percent.
- Market dynamics and spreads:
  - Covered bonds are perceived as safe and offer relatively cheap funding; spreads have risen by 100 bps since mid-May 2011, more than for German banks, due to concerns about sponsor bank credit quality.
  - Yield levels remain historically low (112 bps below their past five-year average, as of June 2012).
  - Spread volatility rose since mid-2008 and post-crisis issuance tends to pay higher spreads.

### Mortgage loans: exposure detail and bank-level concentration
- Aggregate figures and portfolio weights:
  - Retail mortgages total €832 billion (41 percent of GDP) and represent nearly 41 percent of domestic loans to the private sector.
  - The weight of real-estate related loans is at historically high levels in March 2012.
- Bank-level exposure (eight banks with data as of end-2010):
  - Mortgages represent on average 55 percent of retail and corporate loans.
  - Mortgages represent 42 percent when considering French banks’ aggregated accounting data at end-2011.
  - Aggregate mortgage loan exposure shares (2010): BPCE Group 26 percent; Crédit Agricole 28 percent; Crédit Mutuel Group 20 percent; Société Générale SA 11 percent; BNP Paribas SA 7 percent; La Banque Postale 4 percent; Crédit Immobilier de France 4 percent; HSBC France 0 percent.
  - Mortgage loan exposure by bank in 2010 (in billions of euros): Crédit Agricole 250; BPCE Group 225; Crédit Mutuel Group 172; Société Générale SA 99; BNP Paribas SA 62; La Banque Postale 37; Crédit Immobilier de France 32; HSBC France 3.
- Variation across banks:
  - Importance of mortgage loan exposures ranges from 5 percent of total for HSBC France to over one-half for Crédit Agricole, Crédit Mutuel Group, La Banque Postale, and Crédit Immobilier de France.
  - BNP Paribas, BPCE Group, and Société Générale SA have mortgage loans representing less than one-half of exposures.
  - Exposures to commercial real estate are less prominent in aggregate than exposures to residential real estate.

### Key statistics and exact figures cited
- Mortgage lending y/y in May 2012: 5 percent (France); 0.4 percent (euro area); 1.1 percent (Germany).
- Average DTI ratio: 30.6 percent in March 2012.
- Average term to maturity at end-2011: 18.5 years (up 4.4 years since 2001; down 0.8 years from 2008 peak).
- Average loan amount at end-2011: €108,000.
- Average LTV ratio: around 80 percent in March 2012 (lower by 4 bps compared to end-2010).
- LTV ratios by borrower type (March 2012): first-time borrowers 83 percent; buy-to-let investors 89.5 percent; other borrowers 73 percent.
- Share of loans with LTV >100 percent among first-time borrowers: 12.7 percent in March 2012 (down about 4 points since end-2010).
- Share of borrowers with debt service payments exceeding 35 percent of income: 18.8 percent in 2001; 33.8 percent in 2011; 27.3 percent at end-2011.
- Share of mortgage loans with fixed rates: 89 percent of loans granted in 2011.
- Share of flexible rate loans in 2011 originations: under 10 percent.
- Share of hybrid-rate mortgages in 2011 origination: 1.5 percent.
- Mortgage insurance coverage: 61.6 percent of all loans in 2011 (up from 34.3 percent in 2001).
- Retail mortgages total: €832 billion (41 percent of GDP); nearly 41 percent of domestic loans to the private sector.
- French covered bonds outstanding at end-2010: €156 billion (18 percent of the euro area; 9 percent of the worldwide market).
- Covered bond share in wholesale funding mix of French banks: increased to 56 percent during stress period.
- Covered bond funding accounts for around 20 percent of outstanding residential mortgage loans extended by French banks (ECBC, 2011).
- Covered bond spreads rose by 100 bps since mid-May 2011.
- Yield levels: 112 bps below their past five-year average, as of June 2012.
- Bank mortgage exposure by bank in 2010 (in billions of euros): Crédit Agricole 250; BPCE Group 225; Crédit Mutuel Group 172; Société Générale SA 99; BNP Paribas SA 62; La Banque Postale 37; Crédit Immobilier de France 32; HSBC France 3.
- Aggregate mortgage exposure shares (2010): BPCE Group 26%; Crédit Agricole 28%; Crédit Mutuel Group 20%; Société Générale SA 11%; BNP Paribas SA 7%; La Banque Postale 4%; Crédit Immobilier de France 4%; HSBC France 0%.

*IMF staff report content as provided in the source PDF.*

### 31.      Despite the unemployment rate at a 12-year high, arrears have been low so far.

### _cr13184 - 31.      Despite the unemployment rate at a 12-year high, arrears have been low so far.

### Household mortgage performance and arrears
- Unemployment rate at a 12-year high; arrears have been low so far.
- Aggregate NPL ratio for household mortgages in France: 1.25 percent in 2010–11, up from 0.9 percent in 2007.
- NPL ratio for floating rate loans: about 2.6 percent in 2010–11, up from 1.1 percent in 2007.
- Key factors underpinning household resilience: sound bank lending practices, limited financial innovation, predominance of fixed-rate loans, an active refinancing market, reliance on MI, and generous unemployment insurance.

### Commercial real estate (CRE) and provisioning dynamics
- Problem CRE loans: reached 8.5 percent of total in 2010, declining only by 50 bps in 2011.
- Provisioning rates (system-wide): peaked at 42 percent in 2005, and was slightly below 30 percent in 2010.
- Provisioning for CRE exposures: around 60 percent prior to the subprime crisis (2001–07) to slightly above 30 percent since 2008.
- Real losses on French residential mortgages: 0.0116 percent of total loans outstanding at end-2011.
- Policy concern: currently low provisioning rates could be the most important channel through which banks might be hit by lower real estate values, accompanied by lower household disposable incomes impacting ability to pay.

### Aggregate stress-test illustration and system resilience
- Assumed stress: 100 percent increase in current NPLs of residential mortgages.
- Exposures considered: €824 billion at end-2011.
- Current impaired loan coverage: 30 percent.
- Loss-given default (LGD) assumption: 33 percent.
- Implied potential losses under stress: €3-4 billion.
- Impact on regulatory Tier 1 capital adequacy ratios: decrease by about 0.1 percentage point (to 10.9 percent).
- System resilience underpinned by:
  - Overall sound underwriting and risk management producing a prudent DTI ratio on average.
  - Absence of non-recourse loans, which suppresses risk of “strategic default” by borrowers in situation of negative equity.
  - Reliance on third-party guarantees.
  - Cautious lending criteria for CRE.
- Note on the assumed downturn: the downturn is assumed to be largely confined to a price correction (with no indirect effects on income and no concomitant hike in interest rates).

### Localized vulnerabilities and higher-risk institutions
- Institutions with higher-than-average shares of flexible rate loans or CRE loans are more vulnerable to sharper downturns.
- CIF example:
  - Share of floating rate loans: 55 percent of total.
  - Potential losses under the same stress (doubling of NPLs): €525 million (CIF, 2012).
  - Impact on CIF’s Common Equity Tier 1 ratio at end-2011: decrease by 3.2 points to 11.4 percent.
  - CIF’s loan-book characteristics: higher-than-average proportion of low-to-middle income borrowers; subsidized loans (PTZ and Prêt Accession Sociale (PAS)) represented 17 percent of the overall portfolio at end-November 2011.
  - CIF NPL ratio: 4.1 percent at end-2011 (Fitch, 2012).
  - CIF LTV ratio: 87 percent compared to the national average of 81 percent.
  - Only 17 percent of CIF’s loan book secured by a third-party guarantee at end-September 2011.
- Risk concentration: specialized lenders and portfolios with high shares of floating-rate loans, high LTVs, or limited third-party guarantees are more vulnerable.
- Mitigating factors: relatively low share of low-income borrowers overall; evidence suggesting highly leveraged households are often also higher-income.

### Investor (buy-to-let) exposures and market reversal risk
- Buy-to-let investors have a high gearing ratio.
- Share of such borrowers with DTI ratios above 35 percent: almost 24 percent, compared to 24.5 percent for primary residence buyers (ACP, 2012a).
- Investor risk mitigants: investors are primarily high-income households who can likely access other financial assets.
- Countervailing risk: investor financing is more subject to the investment cycle and may be a source of future price instability if investors reallocate assets.

### Macroeconomic channels and scenario for house-price correction
- Broader consequences of a real estate bust include direct effects on LGD, confidence, wealth, and employment (especially construction); second-round effects could raise delinquencies via higher unemployment.
- CORÉFRIS estimate: a 15 percent correction would result in less than 20 percent of the recent cohorts of borrowers (those that incurred loans during 2006–10) in negative equity (CORÉFRIS, 2012).
- ACP data: loans with down payments of 5 percent (15 percent) or less represented about one-half of total loans issued in 2011.
- French lending standards emphasize income stability and loan serviceability rather than collateral value; negative equity would not necessarily lead to widespread default because banks have recourse to all of a borrower’s assets.
- Direct wealth effects likely small based on existing estimates for France; however, large corrections could affect confidence and trigger precautionary saving or asset reallocation, potentially causing further price declines.

### Vulnerability of household balance sheets (Box 4 highlights)
- Contribution of residential investment to GDP growth is limited; consumption remains central to the French economy.
- Jobless claims: climbed 5.6 percent in 2011.
- GDP growth forecast (scenario context): slow to 0.5 percent (official forecast: 0.7 percent) this year from 1.7 percent in 2011.
- Private consumption projected to contribute 0.3 percent (52 percent) of growth this year.
- Household wealth composition (2005): non-financial assets accounted for 73 percent of households’ net wealth; housing 31 percent; land 38 percent.
- Households’ net wealth in 2005: over seven years of gross disposable income (up from five years in 1995–97).
- 2008 lending cohort: about 40 percent of borrowers had LTV ratios above 95 percent; 11 percent had LTV ratios above 100 percent (SGCB, 2009).
- Non-insured loans for lower-quality borrowers: 43 percent of the total at end-2011.
- France is relatively less vulnerable to an equity price shock: stocks and other securities accounted for only 11 percent of household net wealth in 2005; insurance products 13 percent.

### Policy conclusions and macroprudential recommendations
- Measures already taken to dampen price rises: phasing-out of the Scellier tax incentive and limiting access to the PTZ+ from purchases of new dwellings; banks tightening lending criteria (shortening loan duration, requiring higher down payments).
- Effects observed: younger households increasingly priced-out; expected loan demand has fallen sharply.
- Changing price conditions remain a significant vulnerability; perception of real estate as “riskless” persists and may fuel expectations of long-run price appreciation.
- Macroprudential measures can help alter expectations and prevent prices from diverging from fundamentals; they are preferable to higher interest rates when overheating is localized or when broader macro conditions (high unemployment, large output gap, fiscal consolidation) make rate hikes inappropriate.
- Evidence from other countries: limits on LTV and DTI (e.g., Korea) dampened house price appreciation and transaction activity and altered expectations (Igan, 2011).
- Recommended first-line macroprudential tool for France:
  - Consider a uniform LTV ratio of, say, 80 percent as a simple-to-administer first line of defense.
  - Complement with tighter LTV limits for selected categories (e.g., investors) and geographic zones as warranted.
  - Caveats: no case for immediate introduction given incipient downturn in demand since late 2011; LTV ceilings can limit transaction activity and may reduce efficiency of price discovery.
- Operational recommendations:
  - CORÉFRIS should continue monitoring and prepare to implement a macroprudential framework to act if price dynamics show “one-way” expectations or indicators (e.g., real estate risk premium) point to excessive risk-taking.
  - Enhance data reporting requirements at household, transaction, and new mortgage origination levels.
  - Expand monthly bank reporting to the ACP and build a database of mortgage origination details for ex post evaluation of effectiveness.
- On complementary measures:
  - Caps on debt service-to-income (DTI) ratios could complement LTV ceilings but may need to be set very low to be binding; current industry practice commonly limits DTI ratios to 33 percent on average.
  - Measures imposing additional capital or provisioning requirements for real estate (higher risk weights, higher provisioning) may be counter-productive now: CRE exposures already carry higher risk weights; French banks should comply with Basel 3 requirements beginning of 2013; further real estate-specific requirements could accelerate deleveraging and harm activity.
  - If the objective is to target pockets of speculative activity, LTV or DTI ceilings are likely to be more effective in obtaining a counter-cyclical effect in the short term.

*Source: _cr13184 - 31.      Despite the unemployment rate at a 12-year high, arrears have been low so far.*

### 44.      The measures need to be carefully designed and communicated to ensure their

### _cr13184 - 44.      The measures need to be carefully designed and communicated to ensure their

### Implementation challenges and design principles
- Measures need careful design and communication to ensure effectiveness and avoid key implementation challenges, including:
  - the need for appropriate calibration;
  - the risk of hampering desirable financial development;
  - political economy considerations (the perception that affected socioeconomic demographic groups are unjustly targeted);
  - the risks of circumvention.
- While few advanced countries currently have formal LTV limits, Canada, Germany, and Sweden are exceptions. In other countries (e.g., Australia and the United States), there are no formal ceilings; but high LTV ratios may trigger higher capital adequacy ratings and private mortgage insurance requirements.

### Calibration
- Calibration is a key implementation challenge:
  - Avoid threshold effects (e.g., limits on LTVs that are effective only for certain loan categories) because they can be easily circumvented.
  - Differentiating LTV or DTI caps by geographic zones and/or borrower category can be useful to target “speculative” areas or income groups, but:
    - Such fine-tuning requires enhanced data—micro-level data disaggregated by region and household category.
    - Calibration depends on the extent regulatory authorities judge prices to be misaligned and on the macroeconomic environment.
  - Example considerations:
    - If interest rates are very low and/or banks are already conservative in assessing debt payment capacity, limits on DTI ratios may need to be set at unusually tight levels to be binding.
    - Unusually tight limits could cause an overly strong drop in transaction volumes, hampering price discovery and market efficiency.
  - Authorities should be ready to nimbly adjust limits as warranted by macroeconomic developments to avoid overly restraining credit availability and housing demand when the interest rate cycle turns.

### Access to credit and housing affordability concerns
- Restricting access to credit can hamper desirable credit deepening; this concern is important in France where low-income borrowers face difficulty accessing credit.
- Policy measures should ensure a sufficient supply of affordable housing alternatives (e.g., rentals), because tighter LTV or DTI norms could further restrict vulnerable groups (younger, lower-income) from access to credit and home ownership if these norms do not trigger near-term house price adjustments but mainly affect activity.
- More generally, policies should ensure a sufficient supply response to accommodate the projected rise in number of households due to both demographic and socioeconomic changes.

### Political economy and communication
- Political economy considerations can be addressed through an effective communication policy:
  - Macroprudential measures tend to have differentiated and clearly identified impacts on different categories, unlike uniform interest rate hikes.
  - The rationale and long-run benefits of the measures need to be clearly communicated to the public and affected MFIs.
  - The decision to take no action also needs to be communicated, with explanations of the pros and cons.
  - Higher market transparency, frequent communication, and investor education can support realistic assessments of the real estate risk premium and the need to diversify long-term wealth portfolios across asset classes.

### Circumvention risks and regional coordination
- In France, the main circumvention risk is lending shifting to nonregulated intermediaries; foreign lenders currently have only limited access to retail customers.
- Given extensive foreign retail operations of some large French banks, there is a risk of circumvention through shifting mortgage loan portfolios to other countries in the euro area.
- Measures may need regional coordination to prevent circumvention through cross-border banking activities and to address evidence of co-movement across countries in real estate cycles.
- Empirical evidence and IMF staff analysis:
  - Strong empirical support for a relationship between the French and Spanish housing cycles; German cycle possesses its own dynamics (Ferrara, et al., 2010).
  - IMF staff analysis finds strong co-movement between the French and U.K., Dutch, Italian, and Spanish housing cycles, and weaker concordance with Germany’s housing cycle.

### Conclusion and policy recommendations
- The current price adjustment does not call for immediate macroprudential action.
- Recommendations for authorities:
  - Continue to monitor closely the building of risk and pursue data collection efforts.
  - Continue to develop macroprudential tools (such as LTV limits) and prepare for implementation in line with European Union (EU)-level requirements.
  - Consider pros and cons of uniform vs. finely tuned LTV limits:
    - A uniform LTV limit would be straightforward to administer.
    - Fine-tuning LTV ceilings by region could be more effective but would require more disaggregated data on loan origination and practices, and support from an independent valuation industry or more frequent appraisals.
  - Whether implementing tools or not, communicate the decision and rationale to the public and affected MFIs.
  - Increase frequency of communication by the macroprudential authority to improve housing market transparency and risk assessment by market participants. The CORÉFRIS could consider strengthening its communication, including engagement with private sector analysts and industry participants.

### Select numerical and institutional details cited in the text
- Country examples and specific measures mentioned:
  - Canada, Germany, Sweden — exceptions with formal LTV limits.
  - Australia and the United States — no formal ceilings; high LTV may trigger higher capital adequacy ratings and private mortgage insurance requirements.
- Examples of country-level LTV/DTI and related measures (as reported in Appendix Table 4 excerpts):
  - Canada: 80 percent for privately insured mortgages and 90 percent for government insured mortgages; 85 percent when refinancing; lower limits for commercial investors.
    - 45 percent DTI for all existing debts.
    - Maximum term for mortgages reduced to 35 years (2008); maximum amortization period for new government mortgages with LTV above 80 percent reduced to 30 years.
  - China: lowered from 80 percetn to 70 percent for first homes and 40 percent for second homes (2004-11). Guidelines: 40 percent for third homes.
    - 50 percent DTI (2004).
    - Accompanying measures include caps on credit growth; higher provision ratio; folding back of tax incentives; higher mortgage rates on 2d and 3rd homes; tighter eligibility criteria for land development projects; tighter monetary policy; supply of affordable housing (2010-11).
  - Croatia: 75 percent LTV (2006); 100 percent on total repayment obligations (2006).
  - Greece: indicative DTI 40 percent (2005).
- Other numerical references:
  - The report notes the importance of micro-level data disaggregated by region and household category.
  - References to empirical concordance and co-movement findings are drawn from period analyses and underlying sources cited in the Appendix.

*Source: IMF Staff Report content (text provided from PDF content unit).*

### Appendix Table 4. France: Use of

### _cr13184 - Appendix Table 4. France: Use of

### Overview
- Dataset: Appendix Table 4 (continued and concluded) listing macroprudential tools by country (As of April 2012).
- Focus: main policy objectives, LTV ceilings (year introduced), DTI ceilings (year introduced), real-estate-specific loan loss provisioning, real-estate-specific risk weights, accompanying measures, effectiveness score (0-4), impact on house price appreciation and transaction activity, and impact on credit growth.

### Selected country measures and attributes
- Hong Kong SAR
  - Main policy objective (Macroprudential authority): moderate mortgage growth and property price inflation; limit risks posed by the housing market boom (HKMA)
  - LTV ceiling (year introduced): up to HK$8 million property value, 70 percent with loan cap of HK$4.8 million ; HK$ 8-12 million, 60 percent, with loan cap of HK$6 million; 50 percent for the rest and for commercial properties (2010)
  - DTI ceiling (year introduced): 50 percent including payments on all loans (early 2000s)
  - Real-estate-specific loan loss provisioning: no
  - Real-estate-specific risk weights: no
  - Accompanied by other measures: ceiling of 15 percent p.a. on mortgage loan growth (1994); exposure to property limited to 40 percent (1994-98); stamp duty increase for transactions above HK$20 million and increased land supply (2010)
  - Effectiveness score (0-4): 3
  - Impact on house price appreciation and transaction activity: inconclusive evidence on effectiveness, as boom driven by external buyers from Mainland China rather than domestic buyers
  - Impact on credit growth: ?
- Hungary
  - Main policy objective: address excessive FX lending to households
  - LTV ceiling: 75 percent (2010); lower limits for FX loans ; 70 percent for loans funded by covered bonds
  - DTI ceiling: yes
  - Real-estate-specific loan loss provisioning: yes
  - Real-estate-specific risk weights: no
  - Accompanied by other measures: reduction in housing subsidies (2003-04); efforts to promote financial literacy (2006); modification in banks' scoring systems and ban on FX mortgage lending (2010)
  - Impact on house prices: House prices stabilized in 2005
  - Impact on credit growth: significant reduction in mortgage lending growth following cuts in subsidies but expansion of foreign-currency loans
- India
  - Main policy objective: reduce procyclicality
  - LTV ceiling: 80 percent for residential real estate loans (2010)
  - DTI ceiling: no
  - Real-estate-specific loan loss provisioning: housing loans with 'teaser' rates (2010)
  - Real-estate-specific risk weights: housing and commercial real estate (2004-07 and 2010)
  - Accompanied by other measures: higher reserve requirements and general provisions (2004-07)
  - Effectiveness score (0-4): 2  yes
  - Impact on house prices and credit growth: ?
- Ireland
  - Main policy objective: dampen credit growth; strengthen banks
  - LTV ceiling: on portion of mortgages exceeding 80 percent of the value of the property (2006)
  - Effectiveness score (0-4): 1
  - Impact on house prices: yes (but could be due to the own dynamics of the market)
  - Impact on credit growth: yes (but could be due to the own dynamics of the market)
- Israel
  - Main policy authority: Central bank
  - LTV ceiling: increased risk weight for floating-interest-rate mortgages greater than NIS 800,000 with LTV above 60 percent (2010)
  - DTI ceiling: hike in the benchmark interest rate (2010)
  - Real-estate-specific loan loss provisioning: none so far
  - Effectiveness score / impacts: ?
- Latvia
  - LTV ceiling: 90 percent for loans exceeding 1,000 times the minimum salary (2007); abolished in 2008
  - DTI ceiling: for loans with LTV above 70 percent
  - Accompanied by other measures: raise in refinancing rates and reserve requirements (2004-05); increase in land and mortgage registration fees; taxes on speculative transactions; income certification requirements
  - Real-estate-specific loan loss provisioning: no
  - Effectiveness score (0-4): moderate (2 months)
- Malaysia
  - Main policy objective: moderate excessive investment and speculative activity in residential market
  - LTV ceiling: 70 percent for third residential property purchases (2010); otherwise 90 percent
  - Real-estate-specific loan loss provisioning: cyclically-based as provisions increase with the period in default regardless of collateral value
  - Real-estate-specific risk weights: for non-performing residential mortgage loans(2005)
  - Accompanied by other measures: tax measures (2009) and hike in the price floor for foreign buyers (2010)
  - Effectiveness score (0-4): 2  yes
  - Impact on house prices and credit growth: yes
- Netherlands
  - LTV ceiling: 112 percent with the portion exceeding 100 percent being redeemed in 7 years
  - DTI ceiling: yes
  - Real-estate-specific loan loss provisioning: no
  - Real-estate-specific risk weights: no
- Norway
  - Main policy objective: curb credit growth and property price inflation; address high housing debt
  - LTV ceiling (guidelines): 90 percent for housing loans and 75 percent for home equity loans (2010)
  - DTI ceiling (guidelines): about 30 percent (2010)
  - Real-estate-specific loan loss provisioning: no
  - Real-estate-specific risk weights: for loans with LTV above 60 percent  (1998); abolished in 2001
  - Effectiveness score (0-4): 2  significant (1 year)
  - Impact on credit growth: yes
- Poland
  - Main policy objective: mitigate credit and FX risk; strengthen buffers (Commission for Financial Supervision)
  - LTV ceiling: tighter LTV for FX mortgage lending based on loan maturity (2010)
  - DTI ceiling: 50-65 percent ; 42 percent for FX loans to households (2010)
  - Real-estate-specific loan loss provisioning: no
  - Real-estate-specific risk weights: for FX residential loans (2008)
  - Accompanied by other measures: abolished mortgage interest rate deductibility (2007); limits on FX lending (2002, 2006, and 2010); restriction on profit distribution (2009)
  - Effectiveness score (0-4): 2  no
  - Impact on credit growth: yes, FX lending
- Portugal
  - DTI ceiling: for loans with LTV above 75 percent (1999)
  - Accompanied by other measures: formation of a national council of supervisors (2000); interest rate rise
  - Effectiveness score (0-4): 2  ?
  - Impact on credit growth: significant
- Russia
  - LTV ceilings: 60 percent for loans purchased by state agency; 45 percent for loans purchased by state agency
  - Real-estate-specific loan loss provisioning: yes
  - Real-estate-specific risk weights: no
- Singapore
  - Main policy objective: ensure a stable and sustainable property market; reduce further speculative demand
  - LTV ceiling: caps reduced to 80 percent (2010); 60 percent on second and subsequent mortgages ; 50 percent for non-natural individuals (2011)
  - Real-estate-specific loan loss provisioning: no
  - Real-estate-specific risk weights: no
  - Accompanied by other measures: caps on exposures; stimulus measures for property developers discontinued (2009); sellers' stamp duty on residential land and properties sold within a year (2010-11); supply measures; curbs on foreign ownership (2010)
  - Effectiveness score (0-4): 3
  - Impact on house prices and transaction activity: yes, slowdown in price appreciation for apartments; drop in sales
- South Korea
  - Main policy objective: maintain house price appreciation rates within the zero and nominal GDP growth rate band (Financial Supervisory Service, FSS)
  - LTV ceiling: 50-60 percent in non-speculative areas; 40 percent in speculative areas (2002-09)
  - DTI ceiling: 40 percent for loans used to buy houses in the speculative zones for single borrowers under 30 or married borrowers and the spouse has debt (2005-09)
  - Real-estate-specific loan loss provisioning: no
  - Real-estate-specific risk weights: no
  - Accompanied by other measures: tax measures
  - Effectiveness score (0-4): 4
  - Impact on house price appreciation and transaction activity (measure, horizon of the effect):
    - decrease in monthly percent change in real price level: 0.5 percent (LTV, six months)
    - decrease in monthly percent change in number of transactions: 16 percent (LTV, three months); 21 percent (DTI, three months)
  - Impact on credit growth (measure, horizon of the effect): decrease in monthly percent change in household debt levels: 0.2 percent (DTI, three months)
- Spain
  - Main policy objective: stem credit growth and build a buffer (Banco de Espana)
  - Real-estate-specific loan loss provisioning: based on the 'latent loss' in loan portfolios (on average 10 percent of net operating income) (2000), differentiated by high/low LTV (2004)
  - Real-estate-specific risk weights: for loans with LTV above 95 percent for residential property and 80 percent for others (2008)
  - Effectiveness score (0-4): 2
  - Impact on house prices: only a brief decline in 2000
- Sweden
  - Authority: Riskbank
  - LTV ceiling: 85 percent (2010)
  - DTI ceiling: tighter monetary policy (2006)
  - Real-estate-specific loan loss provisioning: no (monetary policy); ? (LTV)
  - Real-estate-specific risk weights: no (monetary policy); ? (LTV)
- Thailand
  - Main policy objective: reduce the cyclicality of the real estate sector; reduce currency risk
  - LTV ceiling: 70 percent (2003); 80 percent for high-value loans (2009); 90 percent for condos (2011); 95 percent for low-rise housing units (2012)
  - DTI ceiling: yes (2004)
  - Real-estate-specific loan loss provisioning: no
  - Real-estate-specific risk weights: for high value loans with LTV above 80 percent (2009) and lower value residential loans with LTV above 90 percent (2011)
  - Accompanied by other measures: limits on net open currency positions (2002)
  - Effectiveness score (0-4): 3  ?
- Ukraine
  - LTV ceiling: 100 percent
  - DTI ceiling: yes
  - Accompanied by other measures: increase in minimum CAR and monetary tightening (2004); higher reserve requirement ratios (2008)
  - Effectiveness score (0-4): 1  no
  - Impact on credit growth: no
- United States
  - Main policy attribute: mortgage insurance required for LTV >80 percent
  - Real-estate-specific risk weights: yes
  - Accompanied by other measures: anti-predatory lending laws in some states
  - Effectiveness score (0-4): no
  - Impact on credit growth: lower subprime loan origination in states with an anti-predatory lending law in place
- Uruguay
  - Main policy objective: build up buffers and limit cyclicality; reduce currency risk
  - LTV ceiling: 90 percent for 'special loans'
  - DTI ceiling: 20-30 percent
  - Real-estate-specific loan loss provisioning: subject to a dynamic setting that applies to all loans (2001)
  - Real-estate-specific risk weights: for FX loans (2005)

### Effectiveness scores and notable impacts (summary)
- Effectiveness scores reported for multiple jurisdictions range from 1 to 4.
  - Score 4: South Korea (reported quantitative impacts on price change, transactions, household debt)
  - Score 3: Hong Kong SAR; Singapore; Thailand (score entries include 3 and 3  ?)
  - Score 2: India; Malaysia; Norway; Poland; Portugal; Spain
  - Score 1: Ireland; Ukraine
- Measured impacts vary in type and horizon; South Korea provides the most explicit quantitative impacts:
  - decrease in monthly percent change in real price level: 0.5 percent (LTV, six months)
  - decrease in monthly percent change in number of transactions: 16 percent (LTV, three months); 21 percent (DTI, three months)
  - decrease in monthly percent change in household debt levels: 0.2 percent (DTI, three months)
- Several countries reported inconclusive or context-dependent effects (e.g., Hong Kong SAR, Ireland).

### Appendix figures (selected metrics reported)
- Appendix Figure 1. France: Building Investment
  - Time series contributions to growth (percent) shown for 2001–2010 with series labeled: Household consumption; Fixed investment; Net exports; Residential Construction; Other Construction; Machinery and Equipment; Cultivated Assets.
  - Employment in Construction (percent of total employment) for 2010/2011 shown for countries: USA GBR DEU IRE SWE CAN ITA ESP FRA AUS (axis 0–10 and extended bars).
  - Housing Investment (percent of GDP) series across countries labeled: USA GBR DEU ESP CAN IRE SWE FRA (axis 0–26).
- Appendix Figure 2. France: Fundamental Factors Influencing House Prices
  - Home Prices in the Long Run, 1961-2010 (Index 2000=1) plotted alongside Population (in millions), Long-Term Interest Rates (percent), Construction costs.
  - Population axis values include 47, 49, 51, …, 67 (years 1961–2009 labeled); interest rates and construction costs plotted on same chart.
- Appendix Figure 3. France: Mortgage Exposures of French Banks (retail mortgage loans; other retail loans; corporate loans)
  - Banque breakdowns (percentage shares as shown):
    - BNP Paribas: Retail mortgage loans 47%; Other retail loans 7%; Corporate loans 46%
    - BPCE Group: Retail mortgage loans 39%; Other retail loans 12%; Corporate loans 49%
    - Credit Mutuel Group: Retail mortgage loans 53%; Other retail loans 10%; Corporate loans 37%
    - Societe Generale SA: Retail mortgage loans 29%; Other retail loans 12%; Corporate loans 59%
    - HSBC France: Retail mortgage loans 5%; Other retail loans 12%; Corporate loans 83%
    - Credit Agricole: Retail mortgage loans 60%; Other retail loans 6%; Corporate loans 34%
    - La Banque Postale: Retail mortgage loans 95%; Other retail loans 5%; Corporate loans 0%
    - Credit Immobilier de France: Retail mortgage loans 98%; Other retail loans 2%; Corporate loans 0%
  - Sources for these figures: SNL and banks’ public information.

*Sources: Crowe et al. (2011); IMF (2011); Igan and Kang (2011); Lim et al. (2011).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13184.pdf_