## Annex I. Quantitative Illustration of the Interaction and Effectiveness of Macroprudential Tools

## Source details

**Canonical URL:** [Annex I. Quantitative Illustration of the Interaction and Effectiveness of Macroprudential Tools](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023075-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023075-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023075-print-pdf.pdf.json)

---

### Background and problem statement
- Housing market imbalances are a key driver of systemic risk: when house prices exceed sustainable levels, risks for borrowers and creditors—including bank balance sheets—increase.
- Many European economies experienced strong housing price appreciation during the Covid-19 pandemic, prompting monitoring and tightening of macroprudential tools.
- After common legal basis rules in 2014, authorities generally tightened policy levers on rising real estate vulnerabilities, with a temporary relaxation in response to COVID-19.
- The second half of 2022 recorded the fastest policy tightening in the euro area on record: 250 basis points.
- Current environment: unsustainable house prices, rising interest rates, and recession risks intensify trade-offs between financial system resilience and supporting credit flows.

### Macroprudential toolkit, examples, and heterogeneity in Europe
- Tools: capital-based measures (CBMs) and borrower-based measures (BBMs); usage and calibration are heterogeneous across countries.
- CBM examples:
  - Switzerland re-activated a sectoral countercyclical capital buffer (CCyB) for residential real estate at 2.5 percent (September 2022).
  - Sweden extended a minimum level of 25 percent for the average risk weight on Swedish housing loans (November 2021).
- BBM examples:
  - Austria: upper limit LTV = 90 percent; DSTI = 40 percent; maximum maturity = 35 years (August 2022).
  - Czech Republic: DTI limit between 8.5 and 9.5 depending on use and borrower age (April 2022).
- CBMs and BBMs are typically calibrated independently despite important interactions; quantitative evaluation of interactions is provided in this annex.

### Taxonomy, assessment approach, and objective variables
- The assessment measures countries at end-2022 across risk signals and resilience indicators including:
  - mortgage lending standards, household vulnerabilities, banking sector exposure to housing, housing cycle, credit cycle, business cycle, and resilience provided by capital buffers and income-based measures.
- Indicators are bucketed for comparable weighting and cross-country benchmarking.
- To avoid overlap/double counting across tools, proposed calibration derives from calculating bank mortgage portfolio risk accounting for:
  - impact of deployed BBMs, loan loss provisions, and capital measures (sectoral buffers, counter-cyclical buffers, risk weight floors).
- Outcome variable for tool effectiveness: capacity to safeguard banking system resilience by increasing buffers or decreasing build-up of real estate vulnerabilities.

### Framework for interactions and design features
- Novel framework measures relative effectiveness of policy measures, avoids overlaps and pro-cyclicality.
- Design features should be actively used in calibration to minimize unintended effects (e.g., overly restricting credit growth in a downturn).
- Framework accounts for overlap between macroprudential capital buffers and minimum capital requirements which can limit buffer usability during stress.

### Policy guidance and principal findings (summary)
- Under baseline conditions (qualitatively in line with October 2022 IMF projections), authorities should continue addressing medium-term residential real estate vulnerabilities.
- Countries with elevated vulnerabilities could consider tightening macroprudential policies provided procyclical effects are avoided.
- Consider increasing capital buffers if rising interest rates from low levels expose borrowers’ vulnerabilities (push up debt servicing ratios), particularly if:
  - banks’ profitability is boosted facilitating the buildup of buffers, and
  - banks have enough capital headroom to support the real economy.
- When systemic risk is confined to real estate imbalances, a sectoral capital buffer can be more effective than a broad-based cyclical buffer.
- Excessive mortgage flow and rapid house price increases may signal looser underwriting; authorities may consider introducing/tightening BBMs.
- Where housing prices have eased and BBM limits are binding, consider temporarily easing BBM limits (e.g., DSTI ratios) if enough macroprudential policy space exists.
- In a sharp downturn or abrupt house price correction with fire-sale potential:
  - swiftly relax capital-based tools to enable banks to absorb losses and support lending; and
  - ease cyclical BBM limits if conservatively calibrated and overly constraining credit, including by adjusting exemptions.

### Why housing matters at the current juncture in Europe (key statistics and observations)
- Price-to-income (PTI) and price-to-rent ratios: around 30 percent above long-term averages in many countries.
- Real house price growth: resilient in H1 2022, cooling in Finland, Netherlands, UK; more resilient in Iceland, Ireland, Israel, Portugal.
- Mortgages are around one third of total loans as of September 2022.
- Average CCyB rate in Europe reached 91 basis points, above the 67 basis points historical maximum.
- Examples of country actions:
  - UK FPC set UK CCyB at 2% in July 2022 (coming into effect in July 2023).
  - France re-introduced CCyB at 0.5 percent in April 2022 (application starting April 2023), later increased to 1 percent.
  - Four countries activated sectoral SyRB on residential real estate (Belgium, Germany, Lithuania, Slovenia); Switzerland increased sectoral CCyB above pre-pandemic levels.

### Policy tools: selection and calibration
- Two broad policy aims:
  - Contain build-up of vulnerabilities (BBMs, amortization requirements).
  - Strengthen bank resilience (capital-based requirements, provisioning).
- BBMs:
  - Instruments: LTV, DTI, DSTI limits; amortization requirements.
  - Affect new loans directly; appropriate to activate/tighten early in a housing boom.
  - Mechanisms: LTV reduces LGD; DSTI/DTI cap repayment obligations.
- CBMs:
  - Instruments: broad/sectoral CCyB, sectoral SyRB, risk weight measures, regulatory provisions, dynamic provisioning.
  - CCyB properties: increases resilience; activation lead time up to 12 months; release can be immediate in sharp downturns.

### Addressing unintended consequences via design adjustments (examples)
- Constrained credit to creditworthy borrowers:
  - Restrict measures to high-risk loans defined by combined characteristics.
  - Minimum interest-rate fixation for high-LTV loans (Denmark: at least 5y fixation for loans with LTV>60%).
  - Amortization add-ons for high-LTI loans (Sweden: amortization add-on of 1 percent for loans with LTI>4.5).
- Distributional impacts:
  - Less stringent LTV/DSTI for young borrowers (Czech Republic); exemptions to DTI caps for young borrowers (Slovakia); softer rules for first-time buyers (Israel); higher LTV for permanent residence (Portugal).
- Overly restrictive credit growth:
  - Gradual approach: guidance, ex-post evaluation, recalibration, then legally binding measures if needed (Austria).
- Investors crowding out owner-occupiers:
  - Tighter restrictions on buy-to-let/investment segment (Ireland, Latvia, Switzerland).
- Leakages and arbitrage:
  - Broaden regulatory perimeter; consistent application across domestic and foreign banks, including cross-border lending.
- Flexible quotas (“speed limits”):
  - Norway: 10 percent of new mortgages outside Oslo may not meet regulatory DTI/LTV/amortization; Oslo quota = 8 percent.
  - Slovakia: 5 percent of loans exceeding DTI of 8 but not >9 can be granted to borrowers ≤35 years with income <1.3× average wage.

### Quantitative framework (staged approach in Annex 1)
- Stages:
  a) Assess current mortgage portfolio risk: calculate regulatory capital and loan loss provisions built up (micro).
  b) Evaluate system-wide risk environment: vulnerabilities in severe downturn (macro).
  c) Calculate losses in severe downturn: calibrate using historical bank losses or translate macro scenario into PD/LGD for mortgage exposures.
  d) Compute capacity of CBMs to cover future losses: capital buffer nominal size from activating CBMs; compare to future losses net of provisions.
  e) Quantify risk profile of new flows (BBMs): measure credit risk of new issuances absent BBMs; replacement rate assumption used for stock translation.
  f) Compare CBMs’ loss absorption versus BBMs’ systemic risk reduction applying replacement rate, additional loan creation rate, and BBM activation duration.

### Setup and assumptions for illustrative example (numerical)
- Reference environment: current risk environment as of 2022Q3.
- Outstanding mortgage book (α): 100 billion.
- Annual replacement rate of the portfolio: 10 percent.
- Macroprudential tools in the example:
  - CCyB buffer: 2.5 percent CCyB rate on mortgage exposures.
  - Risk weight floor: 25 percent sectoral risk weight.
  - BBMs: combination of DTI, LTV, DSTI limits anchoring new-mortgage risk to average risk of outstanding portfolio.
- Data source for example: regulatory data reported by Spanish banks in the COREP C9.02 template published by EBA in 2013Q1 and 2022Q3, with adjustments.

### Cycle parameters, provisions, and realized losses (stages a–c: numeric)
- Peak (stage a):
  - Average PD of the mortgage portfolio: 0.9 percent.
  - LGD: 10 percent.
  - Provisions set aside: 0.09 billion (Pr = α · PD · LGD as reported).
- Trough / severe downturn (stage b):
  - Average PD surges to: 2.3 percent.
  - LGD rises to: 30 percent.
- Realized/future losses in adverse scenario (stage c): Future losses reported as 0.68 (billion).

### Capital requirement mapping and risk-weight calculations (stage d: numeric and expressions)
- Basel supervisory mapping used to translate PD, LGD, EAD into capital requirements (99.9 percent confidence mapping).
- Reported/calculated outcomes:
  - Capital requirement ratio: 1.2 percent → applied to portfolio size amounts to 1.2 billion in nominal terms.
  - Risk weight density reported: 0.148 (j RW α = 0.148).
- Interaction with sectoral risk weight floor:
  - Since 0.25 > 0.148, the macroprudential risk weight floor is binding at the peak.
  - Verbatim expressions from source:
    - 25% 0.250.8 floorj CapRWEAD α = Δ =  =
    - 2.5% 0.0250.4 CCyBj CapRWA α = Δ =  =

### Risk profile of new mortgages and BBMs (stage e: modeled parameters)
- Modeled PD and LGD for new mortgages:
  - Baseline (peak) new mortgage PD: 1.3 percent.
  - Baseline (peak) new mortgage LGD: 15 percent.
  - Adverse (trough) new mortgage PD: 4.2 percent.
  - Adverse (trough) new mortgage LGD: 35 percent.
- BBM calibration example: Anchor new-mortgage risk using LTV = 80 percent and DSTI = 40 percent.
- Vintage timing assumption: BBMs introduced 5 years before tail-risk scenario.
- Verbatim reduction expression shown in source:
  - 0.5PrPr0.3 BBMsjiji LossesLossesovisionsLossesovisions ββαα   =       = 

### Comparative effectiveness (key numeric results and table summary)
- Different combinations can yield similar performance:
  - A floor for average risk weights of 25 percent could cover over 85 percent of the losses realized 5 years after introduction.
  - Combining a 20 percent minimum risk weight with a 2.5 percent CCyB could achieve a similar loss-absorption level.
- Reported comparative effectiveness (five years after activation; verbatim figures):
  - RW Floor = 25%: Losses (Billion) = 0.93; Additional capital (Billion) = 0.81; Effectiveness = 87%
  - SCCyB = 2.5%: Losses (Billion) = 0.93; Additional capital (Billion) = 0.37; Effectiveness = 40%
  - BBMs to anchor risk of portfolio: Losses (Billion) = 0.59; Additional capital = N/A; Effectiveness = 58%
  - RW Floor = 20% + SCCyB = 2.5%: Losses (Billion) = 0.93; Additional capital (Billion) = 0.78; Effectiveness = 84%
- Summary comparisons:
  - BBM limits calibrated to stabilize risk of new originations to outstanding portfolio could reduce losses by about 60 percent.
  - A sectoral CCyB/SyRB buffer could absorb 40 percent of loan losses.

### Choice of tools — guidance and trade-offs
- Selection guided by drivers of risk:
  - Imbalances driven by loosening lending standards (higher LTV, DTI, DSTI, longer maturities): BBMs often more effective.
  - Systemic risk affecting the stock of mortgage exposures: capital buffers preferable as they apply to outstanding exposures and operate with fewer lags.
- Timing:
  - Tightening late in the cycle favors capital-based measures.
  - Reversal of cycle tends to tighten underwriting standards, reducing new-vintage riskiness.

### When and how to relax macroprudential policies
- Objectives: give banks flexibility to grant credit to cushion downturns; relaxations should take effect immediately.
- Factors warranting relaxation:
  - Low banks’ balance sheet capacity (losses exceed management buffers).
  - Tight lending conditions hindering creditworthy borrowers.
  - Reduced imbalances; policy space remains; temporary shock.
- Considerations:
  - Cost of relaxation rises with recession length.
  - Complementary measures can lower cost (dividend guidance, monitor underwriting, monetary/fiscal support).
  - Buffer usability issues: stigma and interactions with leverage ratio and MREL; clear communication on objectives/duration of release important.
- Which tools to relax:
  - Capital buffers (Basel III CCyB, sectoral CCyB, sectoral SyRB) can be released countercyclically.
  - Risk weight tools released only as last resort.
  - Borrower-based tools: relax those binding in downturn first; stress type matters (interest-rate shock → DSTI/maturity cap relaxation; income shock → DTI relaxation; stagflation → relax both DSTI and DTI).

### Operationalizing macroprudential policy — indicators and diagnostics
- Key vulnerability indicators:
  - Underwriting standards: average and tail DTI, DSTI, LTV in new loans.
  - Household vulnerabilities: household debt; financial wealth; share of floating rate mortgages.
  - Banks’ exposure: share of mortgage loans to total loans to private non-financial sector.
  - Housing cycle: deviation of PTI from long-term averages; real house price growth accumulated since the pandemic.
  - Credit cycle: credit to GDP gap; credit growth; mortgage growth.
  - Business cycle: output gap.
- Resilience indicators under policy control:
  - Size of capital buffers (CCyB); restrictions associated with BBMs.
- Indicator processing:
  - Use trimmed percentiles to mitigate extremes; transform to 0–1 normalization (low = low risk/high resilience; high = high risk/low resilience).
- Illustrative cross-country diagnostics (examples):
  - Austria, Luxembourg, UK: price-to-income > 40% above long-term averages.
  - Portugal, Slovakia, Switzerland: price-to-income at least 20% above long-term values.
  - Country snapshots: Austria (robust price growth; legally binding BBMs in 2022); Luxembourg (price pressures; LTV limit 2021; CCyB 0.5 percent); UK (high household leverage; CCyB increase); Switzerland (re-introduced sectoral CCyB); Portugal (accelerated price growth; vulnerabilities from euribor-linked mortgages); Slovakia (fast mortgage growth offset by tightened BBMs).

### Timing and transmission through the mortgage cycle (timing diagram summary)
- Risks build up between 0 and T+2; losses materialize between T+2 and T+5.
- BBMs introduced at time T apply to new vintages between T and T+2; they need early activation because they act on flows.
- CBMs become effective at T+2 (implementation lag) and apply to outstanding and new mortgages.
- BBMs binding window and CBMs effective timing imply sequencing and lead-time trade-offs for policy choice.

### STEPwise operational guidance for authorities (summary of Section VI / Steps 1–6)
- STEP 1: Initial assessment — evaluate real estate market, mortgage market, banks’ balance sheets, households’ balance sheets.
- STEP 2: Identify drivers — loose credit standards, search-for-yield, lax monetary policy, accommodative macroprudential policy, fiscal incentives, supply constraints.
- STEP 3: Quantify housing risk exposure — forward-looking stress projections or historical loss approaches.
- STEP 4: Calibrate tools — assess risk profile of recent flows (LTV, DTI, DSTI, maturity, amortization, repricing), conduct vintage analysis, convert origination parameters to PiT, project loss absorption across tools.
- STEP 5: Cost-benefit criteria — financial soundness, allocative efficiency, administrative costs; consider data availability and remedies (speed limits, exemptions) for BBMs.
- STEP 6: Timing and cycle-neutral options — give lead time during build-up; plan immediate release during crisis; prospective CCyB lead time up to 12 months; consider positive neutral CCyB buffer where appropriate.

### Design equations, vintage methods, and annex summaries (technical notes)
- Annex II: vintage analysis approaches — simplified and modelling approaches to calculate new-mortgage risk profiles; replacement rate and vintage life-cycle considerations.
- Annex III: procedure to convert origination risk parameters to point-in-time (PiT) risk parameters using supervisory data on LTV, DTI, DSTI, maturity, amortization, and time series on income, interest rates, and prices.
- Verbatim PiT and DSTI/D TI equation fragments and lending-rate notation provided in Annex III for implementation.

### Concluding remarks and policy implications
- Macroprudential policy in Europe is at a critical juncture; instrument design and usage should learn from recent years and the Covid-19 episode.
- When housing vulnerabilities rise but cyclical position is uncertain, cost-benefit considerations suggest favoring targeted capital-based measures over tightening BBMs.
- A sharp downturn causing widespread credit disruptions warrants immediate relaxation of capital-based tools; cyclical BBMs could also be eased if overly constraining and policy space exists.
- Key challenges: no established method to assess normal BBM levels; calibrating a neutral CCyB rate is challenging across diverse country conditions; authorities must explain how multiple tools combine into a neutral long-term setting that supports stability and limits social costs, especially on first-home buyers.

*International Monetary Fund — Annex I. Quantitative Illustration of the Interaction and Effectiveness of Macroprudential Tools (wpiea2023075-print-pdf).*

### Annex I. Quantitative Illustration of the Interaction and Effectiveness of Macroprudential Tools ........... 37

### Annex I. Quantitative Illustration of the Interaction and Effectiveness of Macroprudential Tools

### Background and problem statement
- Imbalances in housing markets are an important driver of systemic risk; when house prices exceed sustainable levels, risks of a housing market downturn increase, and so do risks for borrower and creditor—including bank—balance sheets.
- Many European economies experienced strong housing price appreciation during the Covid-19 pandemic, prompting closer monitoring and tightening of a wide range of macroprudential policy tools.
- After European rules establishing a common legal basis for macroprudential instruments came into effect in 2014, authorities mostly tightened policy levers on the back of rising real estate vulnerabilities, except for a temporary relaxation in response to COVID-19.
- The second half of 2022 recorded the fastest policy tightening in the euro area on record: 250 basis points.
- The current environment combines unsustainable house prices, rising interest rates, and recession risks, intensifying trade-offs between financial system resilience and supporting the flow of credit to the economy.

### Macroprudential toolkit, examples, and heterogeneity in Europe
- Policy instruments include capital-based measures (CBMs) and borrower-based measures (BBMs); tools are widely used and heterogeneous across countries.
- CBM examples:
  - Switzerland re-activated a sectoral countercyclical capital buffer (CCyB) for residential real estate at 2.5 percent (September 2022).
  - Sweden extended a minimum level of 25 percent for the average risk weight on Swedish housing loans (November 2021).
- BBM examples:
  - Austria set an upper limit of 90 percent for LTV ratios, a 40 percent upper limit for DSTI ratios, and an upper limit of 35 years for the maturity of housing loans (August 2022).
  - Czech Republic set a DTI limit ranging between 8.5 and 9.5 depending on the use of the property and the age of the borrower (April 2022).
- CBMs and BBMs are typically calibrated independently despite important interactions; a quantitative evaluation of interactions is lacking.

### Taxonomy, assessment approach, and objective variables
- The paper provides a taxonomy measuring countries along key risk signals and resilience indicators evaluated at end-2022; the macroprudential policy stance measures balance between:
  - systemic risk related to mortgage lending standards,
  - household vulnerabilities,
  - banking sector exposure to the housing market,
  - the housing cycle,
  - the credit cycle,
  - the business cycle,
  - resilience provided by capital buffers and income-based measures.
- Indicators are bucketed for comparable weighting and cross-country benchmarking to guide macroprudential assessment.
- To avoid overlap and double/triple counting across tools, the proposed calibration of any measure is derived by calculating the risk of the bank mortgage portfolio taking into account:
  - the impact of any borrower-based measure already deployed,
  - loan loss provisions,
  - capital measures (e.g., sectoral buffers, counter-cyclical buffers, or risk weight floors).
- The outcome variable to assess tool effectiveness is its capacity to safeguard banking system resilience (absorb potential future losses) by increasing buffers or decreasing the build-up of vulnerabilities in real estate.

### Framework for interactions and design features
- The paper presents a novel framework aimed at measuring the relative effectiveness of policy measures, avoiding overlaps, and avoiding pro-cyclicality.
- Design features should be actively used in calibration to minimize unintended effects (for example, overly restricting credit growth in a downturn).
- The framework accounts for overlap between macroprudential capital buffers and minimum capital requirements which can limit buffer usability during stress.

### Policy guidance and principal findings
- Under baseline conditions (qualitatively in line with October 2022 IMF projections for most European economies), authorities should continue to address existing medium-term vulnerabilities in the residential real estate sector.
- Countries with elevated vulnerabilities could consider tightening macroprudential policies provided that procyclical effects are avoided.
- An increase in capital buffers may be considered in case of rising interest rates from low levels that may expose (outstanding) borrowers’ vulnerabilities (pushing up debt servicing ratios), particularly if:
  - banks’ profitability is boosted facilitating the buildup of buffers, and
  - banks have enough capital headroom to support the real economy.
- When systemic risk is confined to imbalances in real estate, a sectoral capital buffer can be more effective than a broad-based cyclical buffer because it helps build resilience in a targeted manner while minimizing unintended side effects on broad credit supply.
- An excessive build-up flow of mortgage debt, associated with rapid increases in house prices, may signal a loosening of mortgage underwriting standards that could lead to a material increase in the number of highly indebted households; authorities may consider introducing or tightening BBMs to guard against this risk.
- In countries where housing prices have eased, BBMs are calibrated as a cyclical tool, concerns over credit supply are emerging, and BBM limits are binding, authorities may consider temporarily easing some of these limits (e.g., DSTI ratios)—provided there is enough macroprudential policy space—to avoid unduly tightened credit conditions and exacerbating a contraction in the financial system and the economy.
- If facing a sharp downturn in the financial cycle causing widespread disruptions in credit or an abrupt house price correction with potential for fire sales:
  - authorities should swiftly relax capital-based tools to enable banks to absorb losses and support lending activity; and
  - cyclical BBM limits could also be eased, if they are conservatively calibrated and overly constrain the flow of credit, including by adjusting exemptions.

*International Monetary Fund — Annex I. Quantitative Illustration of the Interaction and Effectiveness of Macroprudential Tools*

### Section IV presents a quantitative framework to assess the relative effectiveness of macroprudential tools to

### Section IV presents a quantitative framework to assess the relative effectiveness of macroprudential tools to 

### II. Why Housing Matters at the Current Juncture in Europe
- Housing prices: price-to-income (PTI) and price-to-rent ratios are around 30 percent above long-term averages in many countries.
- Real house price growth: house price growth was resilient in the first half of 2022, with cooling signs emerging in Finland, Netherlands, and the UK; comparatively more resilient in Iceland, Ireland, Israel, and Portugal.
- Transactions have eased in most jurisdictions and price declines have occurred in some cases.
- Macrofinancial risks:
  - An abrupt correction of house prices could trigger or amplify an economic slowdown in Europe.
  - Financially stretched households could cut consumption to service mortgage debt (cash-flow channel) and reduce spending via perceived wealth declines (wealth effect).
  - Banks’ exposures: mortgages are around one third of total loans as of September 2022.
  - US contrast: US banks keep most securitized mortgages off their balance sheets.
- Policy context:
  - Monetary policy should focus on inflation target and internal balance; macroprudential policy should mitigate systemic risk (Integrated Policy Framework, IMF, 2020).
  - ESRB flagged increased probability of tail risk scenarios since the beginning of 2022.
  - UK Financial Policy Committee set the UK CCyB rate at 2% in July 2022 (coming into effect in July 2023) and remains ready to vary it.
  - France re-introduced the CCyB at 0.5 percent in April 2022 (application starting April 2023) and later increased it to 1 percent.

### How Have Macroprudential Policies Evolved Most Recently in Europe?
- Many authorities tightened macroprudential policy during 2021-23 to address real estate and mortgage market vulnerabilities.
- Interventions focused on broad-based capital buffers (CCyB) and sectoral buffers (SyRB):
  - Average CCyB rate in Europe reached 91 basis points, above the 67 basis points historical maximum.
  - Around two thirds of countries with a CCyB measure announced a positive rate higher than the pre-pandemic level; only three countries implemented a rate at a lower level; remainder set CCyB at historical maximum.
  - Four countries activated a sectoral SyRB on residential real estate (Belgium, Germany, Lithuania, and Slovenia); Switzerland increased sectoral CCyB above pre-pandemic levels.
- Examples of specific measures:
  - Switzerland re-activated sectoral CCyB at the maximum 2.5 percent in January 2022.
  - Germany introduced a sectoral SyRB of 2 percent on housing loans and increased CCyB to 0.75 percent.
  - Belgium activated a 9 percent sectoral SyRB to replace minimum risk weight add-ons on residential real estate exposures.
  - Netherlands implemented risk weight floors increasing with mortgage risk profile.
  - Czech National Bank re-instated a DSTI cap at 45 percent in November 2021.
  - National Bank of Slovakia tightened DTI limit with age-differentiated calibration (start at 8 with a floor at 3 at age 60).

### III. Macroprudential Tool Selection and Calibration During the Tightening Phase
- Macroprudential authorities’ objective: limit systemic risk and strengthen financial system resilience preventively.
- Two broad policy aims:
  - Contain build-up of vulnerabilities (borrower-based measures and amortization requirements).
  - Strengthen bank resilience (capital-based requirements and provisioning).

### Policy Tools to Contain the Build-up of Vulnerabilities
- Instruments: LTV, DTI, DSTI limits; amortization requirements (mandatory amortization; maximum amortization period).
- Characteristics:
  - Affect new loans directly; take time to influence existing loan portfolios.
  - Appropriate to activate/tighten early in a housing boom.
- Mechanisms:
  - LTV limits reduce loss given default and can reduce default probability under the ‘double trigger of default’ hypothesis.
  - DSTI/DTI limits cap borrowers’ mortgage repayment obligations, limiting extreme lending.
- Calibration drivers: estimated house price overvaluation, household indebtedness, projected interest rate paths, loan maturity profile.
- Complementarity: BBMs can be complementary if well-designed; authorities can choose different calibrations to achieve similar resilience.

### Policy Tools to Strengthen Bank Resilience
- Instruments: broad/sectoral CCyB, sectoral SyRB, risk weight measures, regulatory provisions, dynamic provisioning.
- CCyB properties:
  - Increases bank resilience during periods of excessive credit growth.
  - May dampen credit growth via price effect (higher capital cost) and quantity effect (reducing risk-weighted assets).
  - Activation lead time up to 12 months; release can be immediate in sharp downturns or gradual as systemic risk dissipates.
- Dynamic provisioning:
  - Built from retained profits during booms to cover realized losses during stress.
  - Activation linked to bank prudential behavior rather than financial cycle.

### Broad-Based Versus Sectoral-Based Countercyclical Tools: the European Experience
- Sectoral tools can be more efficient by targeting specific risks and exposures, minimizing side effects and improving cost-effectiveness.
- Regulatory changes:
  - Basel introduced broad-based CCyB in January 2016; guiding principles for sectoral CCyB followed three years later.
  - CRD V (transposed 2020) allowed targeted use of the SyRB.
- Interaction considerations:
  - CCyB and sectoral CCyB/SyRB may act as substitutes or complements depending on macrofinancial environment.
  - Example scenarios:
    - Exuberant real estate with strong overall lending: activate sectoral CCyB and broad-based CCyB.
    - Low growth with overvalued real estate and high uncertainty: activate sectoral CCyB while keeping CCyB buffer at zero.
- Cross-country treatment: mostly substitutes, except Germany, Lithuania, and Slovenia where both were activated.
- Figure/indicator usage: CCyB uses credit growth to GDP gap and others; sectoral tools use credit gap, credit growth, PTI, PTR, credit conditions, spreads.

### Addressing Unintended Consequences of Tools by Adjusting Design Settings
- Cost-benefit analysis is essential though quantifying unintended impacts is challenging.
- Main concerns with BBMs:
  - Allocative efficiency distortions and distributional impacts (e.g., limiting market access of low-income or young borrowers).
  - Legal/regulatory agility constraints: BBMs may require legislation and be less agile than CBMs.
- Design solutions and country examples (from Table 5):
  - Constrained credit to creditworthy borrowers:
    - Restrict measures to high-risk loans defined by combined characteristics.
    - Minimum interest-rate fixation period for high LTV loans (Denmark: at least 5y fixation for loans with LTV>60%).
    - Amortization rate add-ons for high DTI loans (Sweden: amortization add-on of 1 percent for loans with LTI>4.5).
  - Distributional impacts (delay age of home purchase, limit first-time buyers):
    - Less stringent LTV and DSTI for young borrowers (Czech Republic).
    - Exemptions to DTI caps for young borrowers (Slovakia).
    - Softer LTV and DTI rules for first time buyers (Israel); higher LTV caps for permanent residence (Portugal).
  - Overly restrictive credit growth:
    - Gradual approach: guidance/recommendations, ex-post evaluation, recalibration, then legally binding measures if needed (Austria).
  - Investors crowding out owner-occupiers:
    - Tighter restrictions on investment-led / buy-to-let segment (Ireland, Latvia, Switzerland).
  - Tools weakening banks’ incentives to conduct risk assessment:
    - Use exemptions to regulatory limits (Norway).
  - Leakages to other asset classes:
    - Introduce macroprudential caps on other retail lending segments; carefully monitor consumer loans (Slovakia).
  - Minimum risk weights making risky mortgages attractive:
    - Make risk-weight floor risk-sensitive (Netherlands).
  - Administrative costs for banks:
    - Lower implementation/enforcement costs by imposing restrictions on loan characteristics included in regulatory reporting and give lead time (Belgium).
  - Regulatory arbitrage:
    - Broaden regulatory perimeter; consistent application across domestic and foreign banks, including cross-border lending.

### Design Solutions During the Tightening Phase (summary)
- Use differentiated calibrations and exemptions to mitigate distributional and allocative side effects.
- Apply targeted instruments to improve cost-efficiency (e.g., age- or purpose-differentiated DTI/LTV settings).
- Sequence measures: start with guidance, evaluate ex-post, recalibrate, then legislate if necessary.
- Monitor and address leakages and potential regulatory arbitrage through broader perimeter and consistent application.

_Italic: Source — wpiea2023075-print-pdf (IMF Working Paper excerpt)._

### introduction of risk-weight floors which have been increasingly differentiated across types of exposures. For

### Introduction of risk-weight floors which have been increasingly differentiated across types of exposures

### Examples of differentiated risk-weights
- Netherlands: the macroprudential authority introduced differentiated minimum risk-weights for the calculation of capital requirements by LTV (i.e., 45 percent risk weight for the portion of a loan exceeding 55 percent LTV and 12 percent otherwise) which entered into force by end 2020.

### Design considerations to minimize macroprudential tools’ unintended effects
- Quantitative guidance versus hard limits
  - Quantitative guidance can be implemented as a first step, before considering hard limits in a second step, when there is large uncertainty over macroprudential policy impact.
  - Austria (2018): stipulated quantitative criteria for sustainable lending in real estate financing on LTV, DSTI, and loan maturity to collect additional data on the risk profile of new mortgage flows and help calibrate instruments.
  - Financial Market Authority (Austria) — June 2022 (effective in August 2022): set an upper limit of 90 percent for LTV, 40 percent for DSTI and 35 years for the maturity of loans with some exemption buckets, following an ex-post policy evaluation analysis on a sufficiently granular database.
- Introduction of flexible quotas (“speed limits”)
  - Flexible quotas on new loans set the maximum volume of high-risk lending that can exceed the limits set by the macroprudential authority, giving banks flexibility in customer allocation and increasing credit market efficiency.
  - Norway: 10 percent of new mortgages outside Oslo is allowed not to meet regulatory DTI, LTV, and amortization rate requirements; the quota is stricter in Oslo at 8 percent.
  - Slovakia: 5 percent of loans exceeding a DTI ratio of 8 but not greater than 9 can be granted to borrowers not older than 35 years with income below 1.3 times the average wage.
- Temporary versus permanent restrictions
  - Cyclical buffers are temporary tools by design, to be adjusted throughout the financial cycle.
  - Borrower-based limits (BBMs) may be relaxed when systemic risk eases and imbalances recede (example: New Zealand in 2017) or in the face of extreme uncertainty posed by a recessionary environment to support lending to the real economy (example: Czech Republic in 2020).
  - Some countries consider BBMs mainly as structural caps and favor adjusting ‘speed limits’ over the cycle (example: Norway).
  - Considerations for adjusting BBMs’ upper limits or exemption buckets across the cycle include: volatility of the business cycle, fluctuations in underwriting standards, and uncertainty of transmission channels. Countries with larger and more frequent shocks and larger fluctuations in underwriting standards may benefit more from resetting BBMs limits; countries with more uncertain transmission may prefer adjusting BBM flow limits (the number of mortgages that can be extended at or above the upper threshold).

### How macroprudential tools address procyclicality and enhance resilience
- Purpose and mechanism
  - Macroprudential tools are designed to address the procyclical bias of capital requirements and loan loss provisions due to the mismatch between cyclical risks and the reserves built in regulatory capital and provisions.
  - During a boom, loan portfolio quality generally improves, resulting in lower capital reserves and a drop in provisions; cyclical underestimation of expected losses and a lower risk premium leads to higher credit growth that sows the seeds of the downturn.
  - IFRS 9’s expected credit loss (ECL) approach should limit the procyclicality of provisions, but benefits depend on banks’ ability to forecast future materialization of credit risk; some banks underestimated future credit losses during the COVID-19 pandemic, relying mainly on days past due as a trigger for a significant increase in credit risk.
- Comparing CBMs and BBMs
  - BBMs act through quantity restrictions on new flows of lending while capital measures (CBMs) work through higher capital requirements on the stock of exposures.
  - BBMs enhance resilience with a lag depending on the replacement rate of the portfolio; a common measure is needed to quantify BBMs’ reduction of systemic risk against CBMs’ loss absorption capacity.
  - In the tightening phase, BBMs tend to have more impact on credit growth than CBMs as they operate through quantities.
  - In the loosening phase, the impact of relaxing tools depends on: (i) nature of the shock (supply or demand); (ii) banks’ capital headroom; and (iii) interaction with other capital requirements.
    - Evidence from euro area loan data: banks with less capital headroom tended to lend less during the pandemic, indicating higher effectiveness of capital buffer releases for capital constrained banks and for banks that did not breach other capital measures (such as leverage ratios, or MREL).
  - If policymakers tighten too late in the cycle, credit growth may retrench further; impact is expected to be more severe if tightening BBMs (as the impact feeds directly into credit restrictions rather than pricing).
  - Raising the CCyB buffer is likely to yield greater benefits than tightening BBMs because it increases banks’ reserves to absorb losses and will not reduce credit as much, particularly if banks have enough capital headroom.
  - Tightening BBMs will have no impact on realization of immediate losses (it applies only to new flows) and may constrain credit growth further.
- Operationalization
  - The paper outlines practical stages to operationalize assessment of macroprudential tool effectiveness, taking into account interactions with minimum capital requirements and bank provisions.

*Source: wpiea2023075-print-pdf*

### Annex 1 provides a quantitative example based on mortgage risk parameters of a major European banking

### wpiea2023075-print-pdf - Annex 1 provides a quantitative example based on mortgage risk parameters of a major European banking

### Quantitative framework (staged approach in Annex 1)
- a) Assessing the current risk of the mortgage portfolio:
  - Use risk characteristics of the stock of mortgages to calculate regulatory capital and loan loss provisions already built up to absorb future losses (micro perspective).
- b) Evaluating the system-wide risk environment:
  - Consider vulnerabilities that could crystallize in a severe downturn at the macro level, including broader systemic risk from spillovers, which may differ from bank-level risk reserves.
- c) Calculating losses in a severe downturn:
  - Two approaches:
    - Calibrate future losses using historical bank losses in periods of elevated risk.
    - Translate a macro severe scenario into micro parameters (e.g., PD, LGD) and apply to banks’ mortgage exposures; benchmark scenario to historical recession-like episodes (e.g., Federal Reserve) or current market exuberance (e.g., Bank of England).
- d) Computing the capacity of CBMs to cover future losses:
  - Calculate the capital buffer in nominal terms corresponding to activating CBMs, considering scope and calibration.
  - Compare buffer size against future losses from stage (c) net of provisions from stage (a).
- e) Quantifying the risk profile of new flows relative to the outstanding portfolio:
  - Since BBMs apply to the flow of new mortgages, measure credit risk of new issuances absent BBM deployment.
  - Difference between loss of the loan portfolio with and without BBMs determines reduction in systemic risk achieved by BBMs.
  - A simplified or modeling approach can estimate credit losses at the vintage level (Annex 2).
  - Annex 1 assumption: replacement rate of 10 percent based on supervisory information from a major European country.
- f) Comparing CBMs’ loss absorption capacity against BBMs’ systemic risk reduction:
  - Apply replacement rate, rate of additional loan creation, and duration of BBM activation to translate flow impacts into expected effect on stocks.
  - A higher replacement rate and earlier BBM activation relative to systemic risk materialization enhance capacity to absorb future losses.

### Capital-Based Tools (CBMs) — illustrative example findings
- In the illustrative example (outstanding mortgages size = 100):
  - If tail risk (the trough) materializes, mortgage losses would amount to 0.68, exceeding the 0.09 booked as loan loss provisions.
  - If the build-up of risk were to continue over 5 years, losses net of provisions would reach 0.93.
- Results indicate:
  - A sectoral CCyB (or SyRB) buffer (activated during run up to the peak and released gradually or at the trough) and minimum risk weights on banks’ IRB models can build resilience during expansion when housing risks are underestimated.
  - Both capital-based tools become binding during the upturn and are not binding during the trough.

### Borrower-Based Measures (BBMs) — illustrative estimation approach
- Four-step estimation:
  1. Use Basel III IRB supervisory formula to back out effective maturity consistent with banks’ reported PDs and LGDs and observed risk-weights.
  2. Use a semi-structural model to project PDs and LGDs of new vintages (no policy) and plug into IRB formula to compute capital requirements.
  3. Calibrate regulatory limits on BBMs (LTV, DTI, DSTI, amortization requirements) to match risk of new vintages to average risk of outstanding portfolio.
  4. Use replacement rate of the portfolio to assess BBMs’ capacity to reduce tail risk of the portfolio over time.

### Quantifying relative effectiveness (key numeric results)
- Different tool combinations can yield similar performance:
  - A floor for average risk weights of 25 percent could cover over 85 percent of the losses realized 5 years after introduction.
  - Combining a 20 percent minimum risk weight with a 2.5 percent CCyB buffer could achieve a similar level of loss absorption.
- Comparative effectiveness reported (Table 6, five years after activation):
  - RW Floor = 25%: Losses (Billion) = 0.93; Additional capital (Billion) = 0.81; Effectiveness = 87%
  - SCCyB = 2.5%: Losses (Billion) = 0.93; Additional capital (Billion) = 0.37; Effectiveness = 40%
  - BBMs to anchor risk of portfolio: Losses (Billion) = 0.59; Additional capital = N/A; Effectiveness = 58%
  - RW Floor = 20% + SCCyB = 2.5%: Losses (Billion) = 0.93; Additional capital (Billion) = 0.78; Effectiveness = 84%
- Summary comparisons:
  - BBM limits calibrated to stabilize risk of new originations to outstanding portfolio could reduce losses by about 60 percent.
  - A sectoral CCyB/SyRB buffer could absorb 40 percent of loan losses.

### Choice of tools — guidance and trade-offs
- Tool selection guided by predominant drivers of risk:
  - If imbalances are driven by loosening lending standards (increases in LTV, DTI, DSTI, longer maturities): BBMs are often more effective.
  - If systemic risk affects the stock of mortgage exposures (excessive credit growth, unsustainable house prices, higher repricing rates): capital buffers may be preferable as they apply to outstanding exposures and operate with fewer lags.
- Timing considerations:
  - Tightening late in the cycle favors capital-based measures.
  - Reversal of cycle tends to tighten underwriting standards, reducing riskiness of new vintages.

### When and how to relax macroprudential policies
- Objectives of relaxation: give banks more flexibility to grant credit to cushion downturns; relaxations should take effect immediately to support real economy.
- Factors warranting relaxation:
  - Low banks’ balance sheet capacity (losses exceed management buffers).
  - Tight lending conditions hindering creditworthy borrowers.
  - Reduced imbalances (mortgage and real estate markets cooled off).
  - Policy space remains (tools calibrated tightly so relaxation still offers protection).
  - Temporary shock (predictably short-lived recession).
- Considerations and constraints:
  - Cost of relaxation rises with recession length as future credit losses escalate.
  - Complementary measures can lower cost of relaxation (conservative dividend guidance, monitoring underwriting standards, monetary/fiscal support).
  - Buffer usability issues: banks may be reluctant to use released buffers due to stigma or interactions with leverage ratio and MREL; clear communication on objectives and duration of buffer release is important.
- Which tools to relax:
  - Capital buffers (Basel III CCyB, sectoral CCyB, sectoral SyRB) can be released countercyclically.
  - Risk weight tools could be released only as last resort and provided systemic risk has dissipated.
  - Borrower-based tools: relax those becoming binding in a downturn first; the stress scenario matters (interest-rate shock favors DSTI/maturity cap relaxation; income shock favors DTI relaxation; stagflation may require relaxing both DSTI and DTI).

### Operationalizing macroprudential policy — indicators and diagnostics
- Key vulnerability indicators:
  - Underwriting standards: average and tail DTI, DSTI, LTV in new loans.
  - Household vulnerabilities: household debt; financial wealth; share of floating rate mortgages.
  - Banks’ exposure to real estate: share of mortgage loans to total loans to private non-financial sector.
  - Housing cycle: deviation of PTI from long-term averages; real house price growth accumulated since the pandemic.
  - Credit cycle: credit to GDP gap; credit growth; mortgage growth.
  - Business cycle: output gap.
- Resilience indicators under policymaker control:
  - Size of capital buffers (CCyB).
  - Restrictions associated with BBMs.
- Indicator bucketing and normalization:
  - Use trimmed percentiles to mitigate extreme values.
  - Transform indicators to normalize between 0 and 1 (low = low risk/high resilience; high = high risk/low resilience).
- Illustrative cross-country diagnostics:
  - Example groups: Austria, Luxembourg, UK (price-to-income > 40% above long-term averages); Portugal, Slovakia, Switzerland (price-to-income at least 20% above long-term values).
  - Country-specific observations (illustrative):
    - Austria: robust real estate price growth; resilience built via legally binding BBMs in 2022.
    - Luxembourg: price pressures and deterioration in lending standards; LTV limit in 2021; CCyB at 0.5 percent.
    - UK: high household leverage and bank mortgage exposure; housing cycle cooling and CCyB increase strengthened resilience.
    - Switzerland: systemic risk from banks’ real estate exposure and elevated household indebtedness; re-introduction and increase of sectoral CCyB.
    - Portugal: accelerated real house price growth since pandemic; limits in LTV, DSTI, and loan maturity exist but household vulnerability to euribor-linked mortgages and lack of activated capital buffers noted.
    - Slovakia: fast mortgage growth offset by tightened BBM limits and eased household indebtedness growth.

*Italic line: Source: Annex 1 and related sections of the IMF working paper (wpiea2023075-print-pdf).*

### Section VI has presented a set of indicators to quantify current vulnerabilities in residential real estate markets

### wpiea2023075-print-pdf - Section VI has presented a set of indicators to quantify current vulnerabilities in residential real estate markets

### Overview
- Section describes granular steps to develop a macroprudential stance assessment tailored to specific country circumstances and to evaluate potential need to adjust policy settings.
- Notes: footnote numbering appears in-text as "34", "35", "36", "37", "38" referencing specific remarks and examples in the source.

### STEP 1. Initial assessment of systemic residential real estate-related vulnerabilities
- Objective: Assess whether current housing market imbalances are of systemic importance from a financial stability perspective across these segments:
  - Real estate market:
    - Look at house price developments: growth rate; growth rate relative to fundamentals (income growth; working age population growth).
    - Indicators of overvaluation: deviation from long-term averages in PTI, PTR; model-based indicators such as user-cost model.
  - Mortgage market:
    - Look at mortgage credit growth; change in mortgage-to-GDP ratio.
    - Important to strip out refinanced loans from mortgage issuance.
    - Example: In many countries an increase in new originations has been driven by increasing shares of refinanced loans as borrowers lock-in lower lending rates (e.g., Belgium, Slovakia, Switzerland), although the stock should not be affected.
  - Banks’ balance sheets:
    - Look at the share of mortgages in the loan book relative to assets, relative to capital, and identify possible concentration risk (e.g., lack of diversification of business models, geographical concentration of mortgages).
  - Households’ balance sheets:
    - Look at level/trends in debt to GDP ratio; if possible, look at the most vulnerable households (low income), and account for mitigating factors (e.g., financial wealth).

### STEP 2. Identify key drivers of real estate imbalances (to inform policy options)
- Loose credit standards:
  - Higher demand than fundamentals may be driven by loose credit standards due to intense competition on the mortgage loan market.
  - Look at trends in banks’ net interest margins, pass-through from policy rates to mortgage rates, activity by new entrants (including insurance firms, pension funds, and fintech), and changes in housing finance (e.g., securitization, role of non-bank financial intermediaries).
- Search-for-yield:
  - Low interest rate environments may drive investors to build up residential property portfolios.
  - Look at signs of overheating in income-producing residential real estate (IPRRE) relative to the owner-occupied segment.
- Lax monetary policy:
  - Construct a simple indicator of the financial and business cycle (for instance credit gap and output gap).
  - If the financial and business cycles are out of sync, monetary policy could be too accommodative from a financial stability perspective (too low interest rates).
- Accommodative macroprudential policy:
  - Macroprudential policy loosened in most countries during the pandemic by easing capital requirements, removing regulatory limits, and directly encouraging lending.
  - Look at policy changes and the impact on lending rates and credit volumes.
- Fiscal incentives:
  - Look at changes in housing taxation (e.g., tax mortgage interest tax relief, tax rate on transfer of property, imputed rent taxation, recurrent property tax in dwellings, capital gains tax on selling property); and government subsidies (e.g., for low-income households).
- Supply constraints:
  - Consider tighter housing supply (lower construction volumes); rising construction costs (e.g., due to tighter building standards; commodity price shock); and tougher regulation (zoning restrictions).

### STEP 3. Quantify level of housing risk exposure to the financial system
- Two complementary approaches:
  - Forward-looking approach:
    - Project loan loss rates on mortgage loans (using own model, or authorities’ stress testing results) under an adverse scenario with a real estate correction calibrated to past crises in the country (or in peer countries) or to the estimated size of overvaluation.
  - Historical approach:
    - Use historical losses in the mortgage book during a real estate crisis in the country or in peer economies, and adjust as needed for structural changes, and policy interventions.

### STEP 4. Calibrate the most suitable macroprudential tool
- Assess risk profile of recent mortgage flows; main risk parameters include:
  - LTV, DTI, DSTI, maturity, amortization rate, and repricing schedule.
  - Identify increases in affordability risks measured by lower downpayment (higher LTV), higher DTI, higher DSTI (evaluated at long-term mortgage rates), or longer amortization rates.
  - Importance of tail analysis: look at tails of distribution to unveil pockets of vulnerabilities (e.g., at the 75th, 90th percentile of the distribution).
  - Note: In countries where BBMs have been applied, typically there is bunching just below regulatory limits.
- Conduct a vintage assessment:
  - Analyze effects of loan amortization and external factors (changes in real estate prices, interest rates, disposable income) on credit risk.
  - Vintage analysis tends to show dispersion of delinquency between more recent vintages originated during a run-up in housing prices and older vintages originated before the expansionary phase.
  - Identify outstanding mortgages by vintage in banks’ portfolios using regulatory data provided by authorities.
- Convert origination risk parameters to Point-in-time (PiT):
  - Capture loan life-cycle and impact of housing/business cycle.
  - Newer borrowers are generally more vulnerable (repaid less principal, smaller home equity gains, debt service assessed at lower rates).
  - Credit risk of the loan portfolio depends on PiT risk parameters of mortgage vintages.
  - A procedure is described in Annex 3 (referenced in source).
- Project loss absorption / reduction of risk capacity across tools:
  - Use the framework laid out in the source to map tool effects to quantified risks.

*Source: wpiea2023075-print-pdf - Section VI has presented a set of indicators to quantify current vulnerabilities in residential real estate markets*

### Annex 1 to compare the effectiveness of policies through the cycle. Policymakers need to be forward-looking

### Annex 1 to compare the effectiveness of policies through the cycle. Policymakers need to be forward-looking

### Timing and transmission of tools through the mortgage cycle
- Risks build up between 0 and T+2; losses materialize between T+2 and T+5.
- BBMs (behavioral/borrower-based measures) are introduced at time T and apply only to new vintages issued between T and T+2 under baseline conditions (green); they take time to build resilience because they apply to new flows only and therefore need to be activated well before tail risk materializes.
- CBMs (capital-based measures) become effective at T+2 (given lags in implementation) and apply to all mortgages including the outstanding portfolio and newly granted mortgages; capital-based instruments strengthen resilience as soon as they become effective.
- Figure 8: timeline of the mortgage cycle (risks between 0 and T+2; BBMs binding for new vintages between T and T+2; CBMs effective at T+2; losses between T+2 and T+5).

### STEP 5 — Cost-benefit analysis criteria for choosing tools
- Potential criteria to assess tools:
  - (i) financial soundness (need to build capacity to improve resilience);
  - (ii) allocative efficiency (minimize credit restrictions to otherwise creditworthy borrowers);
  - (iii) administrative costs (assess whether the proposed tools are easy to implement in terms of reporting and compliance).
- Example finding from Section IV referenced in the annex:
  - The most effective tool to strengthen resilience is to set a risk weight floor of 25 percent, assuming banks using IRB models are underestimating current systemic risk.
  - If mortgage portfolio risk profiles are very heterogeneous across banks, a risk weight floor could penalize high-rated portfolio banks relative to low-rated peers, increasing suitability of activating a sectoral CCyB instead.
- Practical considerations:
  - When selecting macroprudential measures, authorities should consider whether the requested data to compute buffers are already available from current sources (e.g., regulatory data) to reduce costs and encourage reciprocity.
  - For BBMs, unintended consequences of unduly restricting market access can be addressed by using ‘speed limits’ (allowing a certain proportion of the volume of new loans to be exempt; limits could be unconditional or targeted to specific loan types, e.g., first-time home buyers, green mortgage loans).

### STEP 6 — Timing of application and cycle-neutral options
- Give lead time during build-up and plan for immediate release during a crisis.
- Prospective CCyB requires a lead time of up to 12 months, according to Basel rules, to allow banks time to meet additional requirements before they take effect.
- Trade-offs in calibration:
  - A gradualist approach to set the CCyB rate may make it difficult to reach a level needed to absorb losses before risks crystallize.
  - To address uncertainty of BBM impact, a gradual calibration approach may be preferable; authorities could re-assess the risk profile of new mortgages between t and t+5 and adjust BBM limits both up and down in line with evolving conditions, observed pass-through to bank lending, and expected loan losses.
- Release and neutral buffer design:
  - When there is a decrease in capital buffers or regulatory limits, banks can apply revised rules immediately to absorb losses and reduce the risk that credit supply is constrained during a downturn.
  - When sectoral cyclical risks do not materialize but recede slowly, a gradual release may be appropriate.
  - Authorities could consider a positive neutral CCyB buffer when risks are judged to be neither subdued nor elevated; this would allow release of the capital buffer in times of acute stress (BIS, 2022b) and help address time lags and uncertainty in measuring cyclical systemic risks.
  - The UK and other seven countries in the EEA have introduced a positive cycle neutral CCyB.

### Short-term context: housing buying power and distributional consequences
- In the current cost-of-living crisis, housing buying power in main cities across Europe has already decreased as mortgage rates have risen sharply (Figure 9).
- Tightening DSTI ratios to level off the risk of new vintages to the risk of the portfolio can be overly restrictive in this environment because:
  - It is unlikely that prices will drop at the same pace as buying power.
  - The cost of further tightening would fall disproportionately on low income or young borrowers with social/distributional consequences.
- Policy design implication:
  - Authorities should weigh social distributional consequences when tightening borrower-based measures and consider targeted exemptions or ‘speed limits’ to mitigate adverse effects on groups such as first-time home buyers.

### Concluding remarks and policy implications
- Macroprudential policy in Europe is at a critical juncture as authorities refine instrument design and usage, informed by lessons from recent years and the Covid-19 crisis.
- When housing market vulnerabilities rise but uncertainty about the cyclical position of the economy persists, cost-benefit considerations suggest increasing targeted capital-based measures may be preferable to tightening BBM tools.
- The materialization of a sharp downturn causing widespread credit disruptions would warrant immediate relaxation of capital-based tools; cyclical BBMs could also be eased if they become overly constraining and there is enough macroprudential policy space.
- Key challenges going forward:
  - No established method exists for assessing the normal level of BBMs; the Covid experience may shed light on interactions between policy relaxation and the housing cycle.
  - Calibration of a neutral CCyB rate and its adjustments through the cycle will be challenging given cross-country differences in macroeconomic conditions and the menu of available tools.
  - Authorities need to explain how multiple macroprudential tools combine into a neutral long-term setting that supports financial stability through the cycle.
- Selection principle for combinations of tools:
  - Authorities should favor tools that are highly effective and have a low social cost, and design them to minimize negative impacts, particularly on first-home buyers.

### Data, taxonomy, and analytical contributions
- The paper provides:
  - A taxonomy of relevant risk and resilience indicators for a broad range of European countries evaluated at end-2022.
  - A novel approach to calibrate macroprudential instruments and measure their effectiveness in building resilience taking into account banks’ provisioning practices.
  - Evidence that different combinations of tools can yield similar levels of resilience; therefore, choice should weigh effectiveness and social cost and seek to reduce overlap with other macroprudential tools and microprudential minimum capital requirements.

*Source: Annex 1, "to compare the effectiveness of policies through the cycle. Policymakers need to be forward-looking", wpiea2023075-print-pdf.*

### Annex I. Quantitative Illustration of the

### Annex I. Quantitative Illustration of the Interaction and Effectiveness of Macroprudential Tools

### Setup and assumptions
- Objective: assess relative effectiveness of macroprudential tools to absorb future mortgage losses under a housing market correction.
- Reference environment: current risk environment as of 2022Q3.
- Portfolio and flows:
  - Outstanding mortgage book (α): 100 billion.
  - Annual replacement rate of the portfolio (used later): 10 percent.
- Macroprudential tools considered:
  - CCyB buffer: 2.5 percent CCyB rate on mortgage exposures.
  - Risk weight floor: 25 percent sectoral risk weight.
  - BBMs: combination of DTI, LTV, and DSTI limits that anchor the risk of new mortgages to the average risk of the outstanding portfolio.
- Data source for example: regulatory data reported by Spanish banks in the COREP C9.02 template published by EBA in 2013Q1 (crisis) and 2022Q3 (current), with some adjustments.

### Cycle parameters, provisions, and realized losses (stages a–c)
- Peak (stage a):
  - Average PD of the mortgage portfolio: 0.9 percent.
  - LGD: 10 percent.
  - Provisions set aside: 0.09 billion (Pr = α · PD · LGD as reported: Provisions = 0.09 billion).
- Trough / severe downturn (stage b):
  - Average PD of the portfolio surges to: 2.3 percent.
  - LGD rises to: 30 percent.
- Realized / future losses in adverse scenario (stage c):
  - Future losses reported as: 0.68 (billion).

### Capital requirement mapping and risk-weight calculations (stage d)
- Basel supervisory mapping function used to translate PD, LGD, and EAD into capital requirements (99.9 percent confidence level mapping).
- Reported/calculated outcomes:
  - Capital requirement ratio: 1.2 percent, which applied to the size of the portfolio amounts to 1.2 billion in nominal terms.
  - To derive risk weighted assets, the capital ratio is multiplied by EAD and the reciprocal of the minimum capital ratio of 8 percent (factor of 12.5): 12.5Re14.8 jj RWACapqEAD αα =⋅⋅ =
  - Risk weight density: 0.148 (j RW α = 0.148).
- Interaction with sectoral risk weight floor:
  - Since 0.25 j RW α <, the macroprudential risk weight floor is binding at the peak.
  - Expressions shown in the text (verbatim):
    - 25% 0.250.8 floorj CapRWEAD α = Δ =  = 
    - 2.5% 0.0250.4 CCyBj CapRWA α = Δ =  = 
  - Activation of a sectoral CCyB of 2.5 percent is reported to yield a capital reserve of the magnitude implied by the expression above.

### Risk profile of new mortgages and BBMs (stage e)
- Modeled PD and LGD for new mortgages under baseline (peak) and adverse (trough) conditions:
  - Baseline (peak) new mortgage PD: 1.3 percent.
  - Baseline (peak) new mortgage LGD: 15 percent.
  - Adverse (trough) new mortgage PD: 4.2 percent.
  - Adverse (trough) new mortgage LGD: 35 percent.
- BBM calibration example:
  - Anchor the risk profile of new mortgages to the portfolio using: LTV = 80 percent, DSTI = 40 percent on new mortgages.
- Example vintage timing assumption:
  - BBMs are introduced 5 years before the tail-risk scenario materializes.
- Reduction of losses achieved by the activation of BBMs (verbatim expression shown):
  - 0.5PrPr0.3 BBMsjiji LossesLossesovisionsLossesovisions ββαα   =       = 

### Comparison of tools and effectiveness (stage f)
- Approach to compare tools:
  - Use the annual replacement rate of the portfolio (assumed at 10 percent).
  - Compare expected losses on new mortgages (those subject to macroprudential limits) with those of older mortgage vintages granted before limits.
- Timing assumption reiterated:
  - BBMs introduced 5 years prior to tail-risk scenario.

### Annex II — Approaches to calculate the risk profile of new mortgages (summary)
- Need for vintage analysis: BBM limits apply only to new vintages; risk differs between vintages for three reasons:
  - (i) change in underwriting standards;
  - (ii) life cycle of the loan (current holders have amortized principal relative to new borrowers);
  - (iii) cyclical fluctuations in house prices, interest rates, and household income.
- Two approaches to calculate new-mortgage risk profile:
  - Simplified approach:
    - Use origination risk characteristics (DTI, DSTI, LTV, maturity) from mortgage surveys.
    - Convert risk parameters at origination to point-in-time (PiT) using procedure in Annex III.
    - Calculate an adjustment factor between outstanding portfolio (PiT) and new issuances to obtain a single indicator I by weighting.
    - Compute PD of new loans (β) relative to outstanding mortgages (α) as: I·PDβ = PDα·Iβ/ Iα (verbatim from text: I PDPD I β βα α = · ).
    - Note: due to life cycle, PDPD β α > and LGDLGD β α > (verbatim inequalities).
  - Modelling approach:
    - Semi-structural model of default risk as function of loan/borrower characteristics, macrofinancial drivers, and regulatory environment.
    - Calibrate at risk bucket and vintage level; simulate default rates and expected losses.
    - Behavioral model links borrower distress probability to debt servicing capacity and negative home equity; LGD linked to outstanding commitments net of sale proceeds.

### Annex III — From origination to PiT risk parameters of housing loans (summary and equations)
- Use supervisory data on LTV, DTI, DSTI, maturity, and amortization rates at origination; match with time series on household disposable income, interest rates, and real estate prices.
- Notation: issuance time s, current period t, maturity T.
- PiT LTV computation (equation A.1, verbatim):
  -  , Orig ss PiT st t Tt LTVP Ts LTV P       = 
- Income at origination and quarterly income growth (equations A.2 and A.3, verbatim):
  -  1 OrigOrig sss s t iLTVP T Income DSTI      = 
  -  1 ts ts IncomeIncomeg  =  + 
- PiT DSTI and DTI formulas (equations A.4 and A.5, verbatim):
  -  , , 1 Origs k ssst PiT st t LTVPi Ts DSTI Income        = 
  -  , Orig ss PiT st t Tt LTVP Ts DTI Income       = 
- Stress-testing shock to DSTI at horizon t+j (equation A.6, verbatim):
  -   , , 1 1 Origs ssstj PiT st  j tj LTVPi Ts DSTI Incomeshock            =  + 
- Lending-rate notation in equations:
  - , sk st i  is the lending rate as of t of a mortgage issued in s with last re-setting period in s+k.
  - During stress horizon at t+j, , s stj i   is the lending rate as of t+j of a mortgage issued in s with last re-setting period in s+λ.

*Annex I, II and III as provided in the source PDF.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023075-print-pdf.pdf_