## sipea2026055

## Source details

**Canonical URL:** [sipea2026055](https://www.imf.org/-/media/files/publications/selected-issues-papers/2026/english/sipea2026055.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/selected-issues-papers/2026/english/sipea2026055.pdf.md)
- [Structured JSON version](/-/media/files/publications/selected-issues-papers/2026/english/sipea2026055.pdf.json)

---

### Trends in mortgage share and LTI/LTV measures
- Mortgage share fell from 58.6 percent in 2021Q4 to 43.7 percent in 2024Q1 and increased to over 50 percent in 2025Q2.
- In 2025Q2—“the latest available quarter for this paper”—one‑third of new issuances had an LTI ratio above 5.
- High‑risk mortgages (LTV > 80 percent) rose from 32.2 percent in 2018 to 34.7 percent in 2025Q2.
- LTV ratios declined steadily over the previous decade up to the text break; separately, average LTV fell from about 80 percent in 2012 to 64 percent in 2023 and rose to 68 percent by 2025Q2.
- High‑LTV originations (LTV > 80 percent) declined from over 40 percent to just 9 percent by 2023 and increased to 16 percent by 2025Q2.
- Before the 2021 hiking cycle, over 70 percent of mortgages had repricing schedules of less than one year; this share has fallen to around half for the most recent issuances.

### Debt‑service metrics (DSTI and LSTI) and estimation approach
- Average DSTI ratios across all mortgages rose from 30 percent in 2018 to 33 percent in 2025Q2.
- DSTI definition: borrower’s total debt repayments—including the new mortgage and existing obligations—as a share of disposable income.
- DSTI data are available only for 2018 and 2024–25; analysis combines DSTI with LSTI (available since 2012).
- LSTI captures servicing costs of the new mortgage relative to income, rather than total debt payments.
- Econometric approach: estimate a long time series of DSTI by LTV bucket (starting in 2012) based on historical relationship with LSTI.

### Mortgage portfolio evolution and borrower vulnerability
- Vintage and product features:
  - Legacy exposures include a material portion of outstanding loans originated during the 2005–06 housing boom.
  - Mortgages were historically predominantly floating‑rate; share of fixed‑rate loans increased following the 2022 ECB hiking cycle.
  - Recent vintages show rising high‑LTV and high‑LTI segments and smaller equity buffers.
- Borrower repayment capacity:
  - Borrowers from recent vintages have repaid less principal and accumulated smaller housing equity buffers, increasing vulnerability to rising interest rates and declining house prices.

### Model structure, calibration, and key parameters
- Default framework:
  - “Double trigger” approach: default only when (i) financial distress (liquidity shock; inability to service the loan) and (ii) economic default (negative home equity) coincide.
- Financial distress probability:
  - Modeled as a non‑linear function of affordability risk (DSTI and its change), labor market conditions (unemployment level and change), and demographic factors.
  - Parameters (퐷푖, 훼, 훽1, 훽2, 훽3, 훾) set to match observed financial distress probabilities.
- Economic default:
  - Occurs when property sale proceeds net of constant transaction cost C plus liquid financial assets do not cover loan NPV.
  - Liquid financial assets estimated from the 2023 HFCS as 1.15 percent of average property value.
  - NPV includes outstanding loan amount and early prepayment penalty; average mortgage rate as of 2025Q2 (푟푡푓) used to discount future interest payments.
- Calibration and validation:
  - Calibrated using Spanish 2013–14 real estate crisis, 1990s UK crisis, and Spanish IRB banks' 2024Q4 reported default rates.
  - Model replicates IRB banks' reported default risk under baseline and the 2014 real estate crisis adverse environment.
- Monte Carlo implementation:
  - 660 vintage‑risk buckets (12 LTV categories × 55 quarterly vintages).
  - For each bucket, N = 10,000 house price draws centered on scenario house price path; repeated 1,000 times.
  - Bucket PD = average of economic default outcomes × probability of financial distress.
- Recovery and LGD:
  - LGD = NPV(L) − (1 − δ) × discounted recovery value.
  - Average time to sell collateral n = 6.17 years (Bank of Spain estimate for 2022–2024).
  - Foreclosure discount δ, and spread applied to discount recovery.
- Key calibrated parameter values (Annex II highlights):
  - 27.26% Foreclosure discount
  - 6.17 year Time to sell collateral after foreclosure
  - 2.7% Discount rate of real estate collateral
  - 1.0% Spread on discount rate on foreclosed collateral
  - Liquid financial assets: 1.15% of average property value

### Stress test scenarios and macro assumptions
- Scenarios:
  - Baseline follows the IMF's January 2026 WEO projections.
  - Adverse scenario: three‑year period (2027Q2–2030Q2) combining an inflationary shock with a domestic recession (based on 2025 EU‑wide stress test exercise for Spain).
- Key adverse scenario assumptions and baseline comparators (reported as cumulative or changes):
  - Household disposable income: Baseline YoY growth = 12.9%; Adverse YoY growth = 0%.
  - Real estate price: Baseline N [0, standard deviation = 15%]; Adverse N [-17.2%, standard deviation = 15%].
  - Unemployment: Baseline -0.1 ppts; Adverse 5.6 ppts.
  - Benchmark rate: Baseline 0.19 ppts; Adverse 1.62 ppts.
  - Mortgage rates widen by an average of 160 basis points in adverse scenario (varying by mortgage type and repricing schedule).
  - House prices assumed to fall 17.2 percent cumulatively in the adverse scenario, with stochastic variation ±15 percent.
  - Household disposable income remains broadly flat in nominal terms in the adverse scenario.
- Annex III baseline and adverse annual paths (selected values by year):
  - Years covered: 2025, 2026, 2027, 2028, 2029, 2030.
  - Baseline:
    - Unemployment: 10.8, 10.7, 10.7, 10.7, 10.7, 10.7
    - Interest rates: 2.6, 2.5, 2.5, 2.5, 2.5, 2.5
    - Income growth: 6.0, 5.1, 4.2, 4.4, 4.0, 4.0
    - House prices: 8.5, 7.6, 6.7, 4.9, 4.9, 4.9
  - Adverse:
    - Unemployment: 10.8, 10.7, 10.7, 14.0, 17.2, 16.4
    - Interest rates: 2.6, 2.5, 2.5, 5.1, 4.5, 4.1
    - Income growth: 6.0, 5.1, 4.2, 4.4, 4.0, 4.0
    - House prices: 8.5, 7.6, 6.7, -4.3, -11.0, -5.6

### Stress test results and concentration of losses
- Outstanding portfolio (2027Q2 horizon):
  - Annualized mortgage loss rates:
    - Baseline: about 30 basis points.
    - Adverse: about 150 basis points.
  - Cumulative losses over three years in the adverse scenario:
    - Roughly 4.6 percent of the total mortgage portfolio, about €23 billion (gross losses, before netting out credit provisions).
- New vintages (2025Q3–2027Q2):
  - Loss rates for these cohorts could surge to around 520 basis points under the adverse scenario (nearly 3.5 times the portfolio average).
- Distribution of losses:
  - Losses heavily concentrated in recent vintages with smaller equity buffers and higher exposure to rising rates and falling house prices.
  - Under stress, banks' credit losses are mainly concentrated in non‑collateralized retail and corporate loans rather than residential mortgages.
  - The 1 percent countercyclical capital buffer (CCyB), scheduled for October 2026 and estimated at around EUR 16 billion, would mainly help absorb non‑mortgage losses.

### Policy interventions: borrower‑based measures (BBMs) and impact
- Timing assumption:
  - Regulatory limits on new vintage flows assumed introduced 8 quarters before the 3‑year stress materializes.
- BBM types tested:
  - LTV caps: 80 percent, 85 percent, 90 percent.
  - LTI caps (e.g., LTI = 5).
  - DSTI caps: DSTI = 30 percent, DSTI = 40 percent.
  - Combinations with speed limits (e.g., speed limit 10 percent, 20 percent).
- Quantified impacts:
  - Well‑calibrated BBMs could reduce portfolio losses under severe stress by up to 25 percent after two years of implementation—equivalent to roughly €6 billion.
  - An LTV cap of 80 percent (either as a hard limit or combined with an LTI limit of 5 under a speed‑limit framework) is found to be most effective.
  - DSTI caps calibrated at 30 or 40 percent yield minimal benefits because system‑wide average DSTI ratios already fall well below tested thresholds.
  - Looser LTV thresholds deliver substantially smaller gains:
    - 85 percent cap reduces adverse‑scenario losses by about 6 percent relative to the no‑BBM baseline.
    - 90 percent cap would be almost non‑binding.
- Caveats and interpretation:
  - Benefits are based on LTV ratios rather than loan‑to‑price (LTP); LTP calibration would likely yield larger resilience gains for the same ratio level.
  - CIRBE data cover only mortgages secured on households' main residences; excluding second homes and buy‑to‑let mortgages likely understates BBM benefits.
  - Analysis assumes “bunching” of new lending just below regulatory limits and incorporates speed limits adjustments proportional to initial issuance distributions.

### Policy recommendations and implementation considerations
- Anchor framework on a collateral‑based cap (LTV or LTP), possibly complemented by an LSTI cap to target loans that are both highly leveraged and have stretched debt‑service burdens.
- View BBMs and capital‑based measures as complements; stronger lender capitalization only partially offsets default risk embedded in high‑LTV loans.
- Introduce BBMs pre‑emptively—before they become binding—to minimize economic and social costs and reduce public resistance.
- Implement a gradual, well‑communicated transition to ease adjustment.
- Use flexibility features (speed limits, differentiation by borrower type) to mitigate distributional concerns while preserving access to home ownership.

*Source: IMF sipea2026055 (excerpt).*

### 58.6 percent in 2021Q4 to 43.7 percent in 2024Q1 but has since increased to over 50 percent in

### sipea2026055 - 58.6 percent in 2021Q4 to 43.7 percent in 2024Q1 but has since increased to over 50 percent in

### Trends in mortgage share and LTI/LTV measures
- A key share fell from 58.6 percent in 2021Q4 to 43.7 percent in 2024Q1 but has since increased to over 50 percent in 2025Q2.
- In 2025Q2—“the latest available quarter for this paper”—one-third of new issuances had a ratio above 5, high by international standards where a common LTI cap is 4.5.
- High-risk mortgages (those with an LTV ratio above 80 percent) rose from 32.2 to 34.7 percent between 2018 and 2025Q2.
- LTV ratios had declined steadily over the previous decade up to [text ends in source].

### Debt-service metrics (DSTI and LSTI) and estimation approach
- Average DSTI ratios across all mortgages rose from 30 to 33 percent between 2018 and 2025Q2.
- The DSTI ratio measures a borrower’s total debt repayments—including the new mortgage as well as existing obligations such as car loans and credit card debt—as a share of disposable income.
- Because DSTI data are available only for a short period (2018 and 2024–25), the analysis combines DSTI data with LSTI information, which is available since 2012.
- The LSTI captures the servicing costs of the new mortgage relative to income, rather than total debt payments.
- Using econometric techniques, a long time series of DSTI by LTV bucket (starting in 2012) is estimated based on its historical relationship with LSTI in the available data.

### Repricing schedules and interest-rate sensitivity
- Before the 2021 hiking cycle, over 70 percent of mortgages had repricing schedules of less than one year.
- This share has fallen to around half for the most recent issuances.

*Source: IMF sipea2026055 (excerpt).*

### 2023. The average LTV ratio fell from about 80 percent in 2012 to 64 percent in 2023, and the share

### sipea2026055 - 2023. The average LTV ratio fell from about 80 percent in 2012 to 64 percent in 2023, and the share

### Mortgage portfolio evolution and recent trends
- Average LTV:
  - Fell from about 80 percent in 2012 to 64 percent in 2023.
  - Rose to 68 percent by 2025Q2.
- High‑LTV originations (LTV > 80 percent):
  - Declined from over 40 percent to just 9 percent by 2023.
  - Increased to 16 percent by 2025Q2.
- Vintage and product features:
  - Legacy exposures significant, with a material portion of outstanding loans originated during the 2005–06 housing boom.
  - Mortgages historically predominantly floating‑rate; share of fixed‑rate loans increased following the 2022 ECB hiking cycle.
  - Recent vintages show rising high‑LTV and high‑LTI segments and smaller equity buffers.
- Borrower repayment capacity:
  - Borrowers originating loans in recent years have repaid less principal and accumulated smaller housing equity buffers, making them more vulnerable to rising interest rates and declining house prices.

### Model structure, calibration, and key parameters
- Default approach:
  - Uses a “double trigger” approach where default occurs only when both (i) financial distress (liquidity shock; inability to service the loan) and (ii) economic default (negative home equity) are present.
- Financial distress probability (equation (1)):
  - Modeled as a non-linear function of affordability risk (DSTI ratio and its change), labor market conditions (unemployment level and change), and demographic factors.
  - Parameters (퐷푖, 훼, 훽1, 훽2, 훽3, 훾) set to match observed financial distress probabilities.
- Economic default (equation (2)):
  - Occurs when property sale proceeds net of constant transaction cost C plus liquid financial assets do not cover the loan’s net present value (NPV).
  - Liquid financial assets estimated from the 2023 HFCS as 1.15 percent of average property value.
  - NPV includes outstanding loan amount and early prepayment penalty; the average mortgage rate as of 2025Q2 (푟푡푓) is used to discount future interest payments.
- Calibration and validation:
  - Calibrated using historical crisis episodes (Spanish 2013–14 real estate crisis, 1990s UK crisis) and Spanish IRB banks' 2024Q4 reported default rates.
  - Model replicates IRB banks' reported default risk under baseline and the 2014 real estate crisis adverse environment.
- Monte Carlo implementation:
  - 660 vintage‑risk buckets (12 LTV categories × 55 quarterly vintages).
  - For each bucket, N = 10,000 house price draws centered on scenario house price path; process repeated 1,000 times for stability.
  - Bucket PD computed as average of economic default outcomes times probability of financial distress.
- Recovery and LGD:
  - LGD = NPV(L) − (1 − δ) × discounted recovery value; average time to sell collateral n = 6.17 years (Bank of Spain estimate for 2022–2024).
  - Foreclosure discount δ and spread applied to discount recovery.

### Stress test scenarios and macro assumptions
- Scenarios:
  - Baseline follows the IMF's January 2026 WEO projections.
  - Adverse scenario: three‑year period (2027Q2–2030Q2) combining an inflationary shock with a domestic recession (based on 2025 EU‑wide stress test exercise for Spain).
- Key adverse scenario assumptions and baseline comparators (reported as cumulative or changes):
  - Household disposable income: Baseline YoY growth = 12.9%; Adverse YoY growth = 0%.
  - Real estate price: Baseline N [0, standard deviation = 15%]; Adverse N [-17.2%, standard deviation = 15%].
  - Unemployment: Baseline -0.1 ppts; Adverse 5.6 ppts.
  - Benchmark rate: Baseline 0.19 ppts; Adverse 1.62 ppts.
  - Mortgage rates widen by an average of 160 basis points in adverse scenario (varying by mortgage type and repricing schedule).
  - House prices assumed to fall 17.2 percent cumulatively in the adverse scenario, with stochastic variation allowing fluctuations of ±15 percent.
  - Household disposable income remains broadly flat in nominal terms in the adverse scenario.

### Stress test results and concentration of losses
- Outstanding portfolio (2027Q2 horizon) results:
  - Annualized mortgage loss rates:
    - Baseline: about 30 basis points.
    - Adverse: about 150 basis points.
  - Cumulative losses over three years in the adverse scenario:
    - Roughly 4.6 percent of the total mortgage portfolio, about €23 billion (gross losses, before netting out credit provisions).
- New vintages (2025Q3–2027Q2) results:
  - Loss rates for these cohorts could surge to around 520 basis points under the adverse scenario (nearly 3.5 times the portfolio average).
- Distribution of losses:
  - Losses heavily concentrated in recent vintages—these borrowers have smaller equity buffers and higher exposure to rising rates and falling house prices.
  - Under stress, banks' credit losses are mainly concentrated in non‑collateralized retail and corporate loans rather than residential mortgages.
  - The 1 percent countercyclical capital buffer (CCyB), scheduled for October 2026 and estimated at around EUR 16 billion, would mainly help absorb non‑mortgage losses.

### Policy interventions: borrower‑based measures (BBMs) and impact
- Timing assumption:
  - Regulatory limits on new vintage flows assumed introduced 8 quarters before the 3‑year stress materializes.
- BBM types tested:
  - LTV caps (80 percent, 85 percent, 90 percent), LTI caps (e.g., LTI = 5), DSTI caps (DSTI = 30 percent, DSTI = 40 percent), and combinations with speed limits (e.g., speed limit 10 percent, 20 percent).
- Quantified impacts:
  - Well‑calibrated borrower‑based measures could reduce portfolio losses under severe stress by up to 25 percent after two years of implementation—equivalent to roughly €6 billion.
  - An LTV cap of 80 percent (either as a hard limit or combined with an LTI limit of 5 under a speed‑limit framework) is found to be most effective.
  - DSTI caps calibrated at standard ratios (30 or 40 percent) yield minimal benefits because system‑wide average DSTI ratios already fall well below tested thresholds.
  - Looser LTV thresholds deliver substantially smaller gains: an 85 percent cap would reduce adverse‑scenario losses by only about 6 percent relative to the no‑BBM baseline; a 90 percent cap would be almost non‑binding.
- Caveats and lower‑bound interpretation:
  - Benefits based on LTV ratios rather than loan‑to‑price (LTP); LTP calibration would likely yield larger resilience gains for the same ratio level.
  - CIRBE data cover only mortgages secured on households' main residences; excluding second homes and buy‑to‑let mortgages likely understates benefits of BBMs.
  - Analysis assumes “bunching” of new lending just below regulatory limits and incorporates speed limits adjustments proportional to initial issuance distributions.

### Policy recommendations and implementation considerations
- Preferred BBM anchoring:
  - Anchor framework on a collateral‑based cap (such as LTV or LTP), possibly complemented by an LSTI cap to target loans that are both highly leveraged and have stretched debt‑service burdens.
- Complementarity with capital measures:
  - BBMs and capital‑based measures should be viewed as complements rather than alternatives; stronger lender capitalization only partially offsets default risk embedded in high‑LTV loans.
- Timing and communication:
  - Introduce BBMs pre‑emptively—before they become binding—to minimize economic and social cost and reduce public resistance.
  - A gradual, well‑communicated implementation path would ease the transition.
- Design features to mitigate distributional concerns:
  - Flexibility features such as speed limits and differentiation by borrower type can help preserve access to home ownership while strengthening resilience.

*Source: IMF staff analysis and stress‑test results as presented in the provided content unit.*

### References

### sipea2026055 - References

### Annex I — Full Regression Results for the Loan-Level Analysis
- Annex I. Table 1. Spain: Effects of High-Risk Loans on Mortgage Defaults
  - LTV>80% coefficients by column: 0.0117***, 0.0116***, 0.0115***, 0.0100***, 0.0093***
  - LTI>6 coefficients by column: 0.0025***, 0.0012, 0.0000
  - LSTI>40% coefficients by column: 0.0037***, 0.0026**, 0.0003
  - Interaction terms:
    - LTV>80% # LTI>6: 0.0030 (standard error 0.0019)
    - LTV>80% # LSTI>40%: 0.0054*** (standard error 0.0020)
  - Observations: 10155091030201101260210155099936981015509993698
  - R-squared by column: 0.0453, 0.0448, 0.0435, 0.0454, 0.0441, 0.0454, 0.0441
  - Fixed effects: Time-Bank FE Yes; Time-Province FE Yes
  - Notes: Dependent variable is mortgage delinquency. Lending standards are indicator variables equal to one when the loan exceeds thresholds: LTV > 80% , LTI > 6 and LSTI>40. All specifications control for loan, borrower risk and property characteristics. Regressions include bank-by-time and province-by-time fixed effects. Robust standard errors in parentheses. ***, **, * denotes coefficient statistically distinct from 0 at 1%, 5% and 10% levels, respectively. Columns (1)-(3) show single-measure specifications; columns (4)-(5) joint specifications entering LTV together with either LTI or LSTI; columns (6)-(7) augmented with the interaction between the LTV dummy and the income-based dummy.

- Annex I. Table 2. Spain: Effects of High-Risk Loans on Mortgage Defaults Contingent on Lenders’ Capital Ratio
  - Main coefficients:
    - LTV>80%: 0.0106*** (0.0018)
    - LTI>6: 0.0023* (0.0013)
    - LSTI>40%: 0.0042*** (0.0014)
  - Interactions with RWA Capital Ratio:
    - LTV>80% # RWA Capital Ratio: -0.0039*** (0.0012)
    - LTI>6 # RWA Capital Ratio: -0.0018** (0.0007)
    - LSTI>40% # RWA Capital Ratio: 0.0004 (0.0011)
  - Observations: 463933470862462178
  - R-squared: 0.0451, 0.0445, 0.0433
  - Fixed effects: Time-Bank FE Yes; Time-Province FE Yes
  - Notes: Dependent variable is mortgage delinquency. Lending standards are indicator variables equal to one when the loan exceeds thresholds: LTV > 80%, LTI > 6 and LSTI>40. The risk weighted assets (RWA) capital ratio is winsorized at the 1st and 99th percentiles and standardized. Regressions include bank-by-time and province-by-time fixed effects. Robust standard errors reported in parentheses. ***, **, * denotes coefficients statistically different from 0 at 1%, 5% and 10% levels, respectively.

### Annex II — Scenario-Based Default Model Calibration
- Annex II. Table 1. Spain: Key Parameters of the Mortgage Model
  - Source: Bank of Spain and IMF staff calculations.
  - Note: Calibration and parameterization based on loan loss experience of the burst of the real estate bubble in Spain in 2012-2014. Risk characteristics of vintages adjusted to reflect current conditions as of 2025Q2.
- Annex II. Table 2. Spain: Other Parameters of the Mortgage Model
  - Exact parameter entries as presented:
    - Calibrate sensitivity of FD to shifts in DSTI: 2
    - Calibrate sensitivity of FD to initial DSTI by bucket: 0.06 β = 1
    - 0.0016 β = 2.5 γ = 0.02 D = , 17.6% FDu FD u ξ ∂ = = ∂ (as shown)
    - 15% 15%30%30% 15% DSTI bucket formulation (as shown)
    - 1 DSTI DSTI DSTI otherwise φ = − (expression as shown)
    - 3 0.727 β = 1 α = 3.65% C = 1.15% P· , 14.8% FDi FD i ξ ∂ = = ∂ (as shown)
  - Value / Parameter explicit entries:
    - 27.26% Foreclosure discount
    - 6.17 year Time to sell collateral after foreclosure
    - 2.7% Discount rate of real estate collateral
    - 1.0% Spread on discount rate on foreclosed collateral
  - Note: Calibration and parameterization based on the 2012-2014 Spanish real estate crisis loan loss experience.

### Annex III — Macrofinancial Variables Path Assumptions
- Context and methodology:
  - Time paths of key macrofinancial variables used in scenario-based analysis: unemployment, interest rates (weighted average of fixed/floating mortgages across all maturities whose counterparty is a household), income growth (year-on-year), and house prices (nominal year-on-year growth).
  - Baseline paths drawn from the January 2026 WEO (except house price projections).
  - House price projections and adverse paths follow the European Banking Authority's (EBA) 2025 EU-wide stress test exercise.
  - Adverse paths generated by applying deviations from the baseline assumed in the EBA 2026 EU-wide stress test scenario to the January 2026 WEO projections.
  - These trajectories feed into model simulations for mortgage portfolio loss estimates—with and without BBMs—reported in Section E.

- Annex III. Table 1. Spain: Baseline and Adverse Scenario (values by year)
  - Years: 2025, 2026, 2027, 2028, 2029, 2030
  - Baseline (January 2026 WEO / EBA 2025 stress test for house prices):
    - Unemployment: 10.8, 10.7, 10.7, 10.7, 10.7, 10.7
    - Interest rates: 2.6, 2.5, 2.5, 2.5, 2.5, 2.5
    - Income growth: 6.0, 5.1, 4.2, 4.4, 4.0, 4.0
    - House prices: 8.5, 7.6, 6.7, 4.9, 4.9, 4.9
  - Adverse (EBA 2025 stress test deviations applied to January 2026 WEO):
    - Unemployment: 10.8, 10.7, 10.7, 14.0, 17.2, 16.4
    - Interest rates: 2.6, 2.5, 2.5, 5.1, 4.5, 4.1
    - Income growth: 6.0, 5.1, 4.2, 4.4, 4.0, 4.0
    - House prices: 8.5, 7.6, 6.7, -4.3, -11.0, -5.6

*Source: sipea2026055 - References.*

---


_Source: https://www.imf.org/-/media/files/publications/selected-issues-papers/2026/english/sipea2026055.pdf_