## 1nldea2024007 ## Source details **Canonical URL:** [1nldea2024007](https://www.imf.org/-/media/files/publications/cr/2024/english/1nldea2024007.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2024/english/1nldea2024007.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2024/english/1nldea2024007.pdf.json) --- ### Overview and key magnitudes - RRE represents nearly 50 percent of total household assets. - Housing loans account for about 85 percent of total household liabilities. - Mortgages comprise more than half of Dutch banks’ domestic loan portfolio. - Insurers: mortgages account for 15 percent of insurers’ assets. - Real estate exposures represent more than a quarter of the combined balance sheet total of banks, insurers, and pension funds. - Residential real estate prices increased by 99 percent between mid-2013 and mid-2022. - Household debt-to-income ratio declined from a peak of 256.7 percent in 2011Q3 to 179.1 percent in 2023Q2. ### Institutional framework, mandates, and powers - DNB has an explicit mandate for financial stability and microprudential supervision (amended Bank Act effective January 2014). - AFM’s mandate clarified since January 2014 to act “also in the interest of financial stability”. - MoF politically responsible for overall financial system functioning and sets limits on LTV and DSTI ratios. - FSC acts as a forum (DNB, AFM, MoF) to identify risks and issue recommendations; FSC established November 2, 2012; legally embedded in Bankwet 1998 as of July 1, 2023. - Current constraint: FSC lacks a “comply or explain” mechanism to influence calibration of borrower-based tools; MoF retains hard powers over LTV/DSTI calibration. - Recommendation 1 (FSAP): Elevate the FSC to a permanent advisory body with semi-hard powers, or vest DNB with hard powers over calibration of borrower-based tools, with guardrails (e.g., limit tightening pace to no more than one percentage point a year; require cost-benefit analyses). ### Surveillance, data, and modelling capabilities - DNB has a dedicated Financial Stability Division (FSD) with Macroprudential Analysis and Policy (MAB) and International Financial Architecture (IFA) departments. - Published outputs and indicators: bi-annual Financial Stability Report (FSR) and annex of early warning indicators; Bulletins, Occasional Studies. - Quantitative models and tools: - Top-down stress test model for banking sector solvency. - Real Estate Vulnerability Assessment for RRE and CRE loan portfolios. - Climate stress test model for flood-related risks. - Firm-level stress test tool for NFCs. - Liquidity risk tool for banking sector. - Tools under development: insurer stress test; structural interconnectedness mapping across banks, pension funds, insurers, and investment funds. - Progress since 2017 FSAP: - CRE monitoring framework at quarterly frequency; granular bank CRE loan data collected. - Corporate monitor distributed quarterly; complemented by granular SME/corporate lending from AnaCredit. - AFM launched multi-year supervisory program on IOLs; mortgage providers contacted 1.68 million customers with an IOL or part-IOL; 370,000 customers tested for future affordability. - Data gaps and LLD suspension: - DNB’s mortgage loan-level data (LLD) collection from banks on hold since 2022Q2 due to NVB concerns about identifiable personal data. - LLD previously covered about 80 percent of all loans of Dutch households and >90 percent of those issued by Dutch banks to residents; biannual non-bank mortgage LLD covers 80 to 90 percent of non-bank mortgage loans. - DNB seeking legal clarification from MoF and working with Dutch Privacy Authority; resumption may take at least two years given legislative timelines. - Potential mitigation: clustered and/or anonymized data with sufficient granularity. - Recommendation 2: Ensure clear legal basis to access granular transaction/loan-level data (residential and commercial) regularly for risk monitoring and stress testing. - Recommendation 3: Develop comprehensive empirical models to regularly monitor drivers of housing prices, including non-credit factors and multiple aggregation levels (micro, regional, macro). ### Macroprudential instruments, calibrations, and recent actions - Capital and liquidity buffers: - CCoB increased four times since June 2013 to 2.5 percent on January 1, 2019. - CCyB introduced in 2016; raised to 1 percent on May 25, 2022 (binding one year later) and to 2 percent on May 31, 2023 (binding one year later) — designated as its “neutral level”. - LR set at 3 percent (announced May 14, 2019; effective June 28, 2021). - Global SII LR surcharge equal to 50 percent of global SII buffer rate since January 1, 2023. - LCR requirement 100 percent effective October 1, 2015; NSFR of at least 100 percent effective June 28, 2021. - Mortgage risk-weight floor (Article 458 CRR) introduced January 1, 2022; calculation: - 12-percent risk weight for mortgage portion with LTV up to 55 percent plus 45 percent risk weight for remaining part; lower bound increases with LTV. - NHG-covered mortgages are exempt. - National borrower-based tools and settings: - Limits on LTV and DSTI ratios introduced January 2013 (Wft, Article 4:34); MoF sets caps by ministerial decree. - LTV limit evolution: 106 percent in January 2013; reduced to 100 percent by 2018. MoF maintained 100-percent LTV limit since 2018. - DSTI stressed interest rate (minimum) set at 5 percent since 2013 and never adjusted by MoF despite rising market rates. - DSTI calculation uses actual mortgage interest rates and redemptions for a 30-year annuity; norms stricter for (partially) IO mortgages; affordability norm supervised by AFM. - Assessment of borrower-based instruments: - LTV limit at 100 percent assessed as inefficient to contain procyclicality and insufficient to protect borrowers in case of price correction. - Recommendation 4: Gradually reduce maximum LTV limit to 90 percent by one percentage point per year. - Recommendation 6: Continue monitoring and addressing IO mortgage fragilities; increase incentives to reduce IO exposure. - Recommendation 5: Gradually remove mortgage interest deductibility (MID); current MID rate decreased from 52 percent in 2013 to 36.93 percent in 2023 but remains “too high”. - DNB suggested guardrails if DNB given hard powers over borrower-based tools: proportionate limits on tightening pace and publication of cost-benefit analyses. ### Housing market dynamics, drivers, and vulnerabilities - Price and debt facts: - Residential property prices nearly doubled from June 2013 trough to July 2022 peak (99 percent). - CRE prices doubled between mid-2014 low and mid-2022 peak. - As of 2023Q2, housing prices declined by 5.7 percent from peak of 2022Q3; signs of stabilization from 2023Q3. - Drivers of the 2013–2022 housing boom: - Higher borrowing capacity: mortgage rates fell by about 2 percentage points (from 3.7 percent to 1.7 percent); average nominal household income rose more than 30 percent from June 2013 to June 2022. - Rising share of non-credit-backed transactions: house sales without an accompanying mortgage rose from 8 percent in 2008 to nearly 17 percent in 2017. - Fundamental supply constraints: MOIKR estimated demand exceeded supply by 279,000 houses in mid-2021. - Persistent tax incentives (MID) and generous borrowing standards (100-percent LTV; IOL availability). - Household leverage and composition: - Cumulative change in level of debt since 2011Q1 contributed +21.3 percentage points to debt-to-income ratio; cumulative change of disposable income contributed −99.0 percent points; net effect: total deleveraging of 77.7 percentage points of disposable income as of 2023Q2. - Less than 3 percent of mortgages had flexible rates in 2023; 25 percent will have flexible rates within next 5 years. - 44 percent of total mortgage debt had an interest-only component as of 2022Q1 (down from almost 59 percent in 2014Q1). - About 2.78 million households (out of ~3.82 million with mortgage debt) have either a partial IO mortgage or a 100-percent IO mortgage. - 29 percent of IO debt matures between 2034-2039. - Mortgage arrears and loss risk: - 2.5 percent of households with an owner-occupied home were in arrears on their mortgage payments (fell quickly afterwards). - High LTV ratios imply large residual debt risk and potential loss given default. - DNB scenario: a 20 percent house price fall would cause 8 percent of homeowners to be in negative equity; LTV ratio of 13 percent of homeowners would rise to over 90 percent. ### Buy-to-Let (BTL) investors and transfer tax measures - BTL prevalence and exposures: - At end-2022, about 15 percent of total housing stock and 33 percent of rental properties were owned by BTL investors. - Kadaster: BTL purchases represent 40 percent of transactions in four major cities; ~20 percent in rest of Netherlands. - BTL loans share in banks’ RRE portfolio (natural persons) <0.5 percent of banks’ total exposures as of 2022Q1. - Assets under management in dwellings by investment funds ≈ 35 billion euros (<4 percent of total assets of all investment funds). - Empirical findings on BTL impact: - Regional panel regressions (12 regions, 2009Q1–2022Q4): a one-standard deviation increase in BTL share (+5.6 percentage points) associated next quarter with a house price growth increase: - >0.1 percentage point (regression (3)); - ~0.2 percentage point (regressions (4) and (6)). - Local projection cumulative effect up to five quarters: peak cumulative increase almost 0.3 percentage points from a one-standard deviation BTL increase. - Reverse causality tests: BTL share responds to lagged rent growth (real rent index growth, lagged: 0.083**) and to lagged changes in mortgage rate (∆Real mortgage rate, lagged: −4.351**), not to short-term house price growth. - Transfer tax differentiation: - From 2021: transfer tax rates differentiated — 0 percent for young first-time homeowners, 2 percent for other main residents, and higher rates for other acquirers (general rate 8.1 percent from January 1, 2021; 10.4 percent from 2023; special exemptions for first-time buyers under 35 and properties under 440,000 euros). - FSAP view: differentiated transfer tax could be integrated into macroprudential toolkit if BTL activity becomes systemically important, but requires careful calibration to avoid excessive reduction in private rental supply. - Recommendation 7: Integrate differentiation of transfer tax in macroprudential toolkit if BTL activity increases further; calibrate size, timing, and perimeter carefully and consider exemptions for long-term private investors contributing to rental supply. ### Supply-side policy and housing targets - Government targets: - Plan to build 100,000 houses per year and reach a total increase of almost 1 million by 2030 (later revised to 981,000 units). - Achievement conditional on alleviating zoning constraints and coordinating with policies to reduce nitrogen depositions. - Recommendation 8: Continue addressing supply shortage via looser zoning regulations, rental market liberalization, greater efficiency and speed in building, and coordination with nitrogen deposition reduction policies. ### Effectiveness evidence and research findings - Macroprudential measures improved mortgage quality (lower average LTV at origination and current LTV) and reduced NPL ratios, but evidence mixed on limiting indebtedness growth. - Studies: - DNB (2015): if LTV had been 90 percent instead of 106 percent, only 13 percent (vs. 65 percent actual) of first-time buyers from 2004–2012 would have been underwater by end-2013. - Caloia (2022): LTI limits effective for low-income households; progressive reductions in LTV affected up to 45 percent of highly leveraged borrowers. - Everett et al. (2021): no evidence borrower- or lender-based measures constrained mortgage growth in the Netherlands. - Policy implication: borrower-based tools calibrated at present (100-percent LTV) are insufficient to curb procyclicality and protect borrowers in severe price corrections. ### Climate- and energy-related adjustments - MID reform and energy labeling: - MID reduced from 52 percent in 2013 to 36.93 percent in 2023; authorities presented a long-term vision in February 2024 but do not plan further cuts immediately. - Differentiation of DSTI limits based on energy labels will enter into effect in 2024, allowing higher debt for more energy-efficient properties and for borrowers who invest to increase energy efficiency. - Clarification: increasing DSTI allowances for higher-energy-label properties is not a macroprudential climate measure if it loosens constraints for efficient houses; a macroprudential climate objective would require tighter limits for purchases of low-efficiency properties. - DNB is mapping climate-related exposures before considering activation of climate-specific macroprudential tools. ### FSAP key recommendations (selected and responsible authorities) - Recommendation 1: Elevate FSC to a permanent advisory body and vest semi-hard powers, or vest DNB with hard powers over borrower-based tools; responsible: DNB, MoF — ST. - Recommendation 2: Ensure legal basis to access granular transaction/loan-level data (residential and commercial); responsible: AFM, DNB, MoF — I. - Recommendation 3: Develop comprehensive empirical models to monitor drivers of housing prices (including non-credit factors); responsible: DNB — ST. - Recommendation 4: Gradually reduce maximum LTV limit to 90 percent by one percentage point per year; responsible: DNB, FSC, MoF — ST. - Recommendation 5: Gradually remove mortgage interest deductibility; responsible: MoF — ST. - Recommendation 6: Monitor and address IO mortgage fragilities and increase incentives for borrowers to lower exposure; responsible: AFM, DNB, NIBUD, MoF — I. - Recommendation 7: Integrate transfer tax differentiation into macroprudential toolkit if BTL activity increases, calibrating size/timing/perimeter carefully; responsible: MoF — ST. - Recommendation 8: Address supply shortage via targeted measures and whole-of-government coordination; responsible: MoF, MOIKR — MT. *Source: Technical Note prepared as part of the 2024 FSAP for the Netherlands (1nldea2024007).* ### EXECUTIVE SUMMARY __________________________________________________________________________ 6 ### EXECUTIVE SUMMARY ### Overview - Macroprudential policy in the Netherlands has centered on the residential real estate (RRE) market given the importance of this market for households, banks, and insurers. - Key magnitudes: - RRE represents nearly 50 percent of total household assets. - Housing loans account for about 85 percent of total household liabilities. - Mortgages comprise more than half of Dutch banks’ domestic loan portfolio. - Insurers: mortgages account for 15 percent of insurers’ assets. - Real estate exposures represent more than a quarter of the combined balance sheet total of banks, insurers, and pension funds. ### Institutional framework - The current institutional arrangement is broadly in line with IMF guidance for effective macroprudential policy. - Features: - Clear mandates and well-defined objectives for De Autoriteit Financiële Markten (AFM) and De Nederlandsche Bank (DNB). - DNB, AFM, and the Ministry of Finance (MoF) work together to ensure domestic financial stability. - The Financial Stability Committee (FSC) acts as a forum between these agencies to identify risks and issue recommendations. - DNB communicates macroprudential developments mainly via its bi-annual Financial Stability Report (FSR). - Authorities have broadly adequate powers to collect information for macroprudential policy and to change the level and regulatory perimeter of macroprudential instruments. - Institutional arrangement conducive to effective cooperation and coordination with other institutions. ### Surveillance and systemic risk monitoring - Surveillance relies on comprehensive quantitative information and various property market models and stress tests. - DNB capabilities and outputs: - Dedicated financial stability division. - Publishes a variety of indicators for systemic risk monitoring, including a set of early warning indicators in the FSR annex and analyses in Bulletins, Occasional Studies, and FSR chapters. - Uses several regular and ad-hoc quantitative models, including: - A top-down stress test model to assess banking sector solvency risks. - Quantitative models to quantify credit risks on banks’ RRE and commercial real estate (CRE) loan portfolios. - Models for flood-related climate risks for banks. - Models for risks to the Dutch non-financial corporate (NFC) sector. - A liquidity risk tool for the banking sector. ### Data gaps, monitoring developments, and models - Progress since 2017 FSAP: - DNB closed identified data gaps on CRE and NFCs by developing a monitoring framework and collecting granular information on bank CRE loans. - Developed a corporate monitor complemented by granular information on banks’ SMEs and corporate lending. - AFM launched a multi-year supervisory program to monitor vulnerabilities from interest-only loans (IOLs). - Remaining issues: - Collection of mortgage granular loan-level data from banks has been put on hold since 2022Q2 due to concerns about identifiable personal data raised by the Dutch banking association. - The legal basis to access granular information essential for systemic risk monitoring needs to be clarified. - DNB should develop more comprehensive empirical models to regularly monitor the drivers of housing prices (including non-credit factors). ### Housing market dynamics and vulnerabilities - Price and debt developments: - Household debt-to-income ratio has been declining since 2011. - House prices increased by 99 percent between mid-2013 and mid-2022. - Drivers of the price increase include: - Higher households’ borrowing capacity. - A growing number of transactions not backed by credit. - Fundamental factors: dynamic economy and demography with limited housing supply. - Persistent tax incentives for homeowners with generous borrowing standards. - Effects on borrower risk profiles: - Higher housing prices reduced average LTV of existing loans but led to a growing share of households pushing borrowing limits at origination with higher debt to income (DTI) and loan to income (LTI) ratios. ### Macroprudential instruments and calibration - DNB actions: - Increased the capital conservation buffer (CCoB) four times since implementation in June 2013. - Revised the countercyclical capital buffer (CCyB) framework during the pandemic and raised the CCyB rate twice since May 2022 to bring it to 2 percent (its “neutral level”). - Introduced in January 2022 a floor for the risk weighting of Dutch residential mortgage loans. - Borrower-based tools: - Calibration of borrower-based tools (limits on DSTI and LTV ratios) is controlled by MoF with the aim of protecting consumers and contributing to financial stability; access to homeownership considerations interferes with willingness to act. - The MoF has refrained in recent years from tightening the LTV limit, maintaining a 100-percent LTV limit since 2018 (after a decline from 106 to 100 percent between 2013 and 2018). - At 100 percent, the LTV limit is assessed as inefficient to contain the procyclical effect of increased borrowing capacity during booms and insufficient to protect borrowers in case of a price correction. - Mortgage interest deductibility (MID): - MID has been progressively reduced since 2013: the maximum rate decreased from 52 percent in 2013 to 37 percent in 2023. - The current rate remains assessed as too high and contributes to excessive RRE demand and higher housing prices. - Interest-only mortgages: - Monitoring of IOLs has improved; AFM’s program and supervisory actions target vulnerabilities from IOLs. - Authorities should continue addressing fragilities from IOLs and increase incentives for borrowers to lower exposure. ### Buy-to-Let (BTL) investors and tax measures - BTL investor activity: - Housing price boom associated with a rising share of transactions by BTL investors, often not backed by credit. - FSAP analysis: an increase in the proportion of BTL investor purchases is associated with a subsequent increase in the quarterly regional growth rate of housing prices. - BTL investors can increase fragilities by raising prices and adding volatility. - Transfer tax differentiation: - MoF increased differentiation of the transfer tax from 2021: 0 percent for young first-time homeowners, 2 percent for other main residents, and 10.4 percent for other acquirers. - Differentiated transfer tax could be part of the macroprudential toolkit if BTL activity becomes systemically important, but calibration in size, timing, and perimeter should be cautious to avoid an excessive drop in BTL activity that could hurt private rental supply. ### Supply-side considerations - Government housing targets: - Plan to build 100,000 houses per year and reach a total increase of almost 1 million by 2030. - Achievement is conditional on alleviating current zoning constraints and coordination with policies aimed at reducing nitrogen depositions. - Supply-side measures are critical to limit price pressures and improve access to homeownership. ### FSAP key recommendations (Table 1) - Recommendations, Responsible Authority, Timeframe: 1. Elevate the FSC to a permanent advisory body and vest it with semi-hard powers, or vest DNB with hard powers over the calibration of the borrower-based tools. (¶ 17). — DNB, MoF — ST 2. Ensure that authorities have clear legal basis to access granular transaction/loan-level data on a regular basis for risk monitoring and analysis, including residential and commercial real estate loans (¶24). — AFM, DNB, MoF — I 3. Develop comprehensive empirical models to regularly monitor the drivers of housing prices (including non-credit factors) (¶24). — DNB — ST 4. Gradually reduce the maximum limit of the LTV ratio to 90 percent by one percentage point per year (¶61). — DNB, FSC, MoF — ST 5. Gradually remove the mortgage interest deductibility (¶61). — MoF — ST 6. Keep monitoring and addressing fragilities from IO mortgages, including by increasing incentives for borrowers to lower their exposure to these mortgages (¶61). — AFM, DNB, NIBUD, MoF — I 7. Integrate the differentiation of the transfer tax on housing transactions in the macroprudential toolkit, should buy-to-let investors’ activity further increase, and in the meantime, carefully calibrate the transfer tax for buy-to-let investors in terms of size, timing, and perimeter (¶61). — MoF — ST 8. Continue to address a supply shortage of residential rental and owner-occupied housing by targeted measures, such as looser zoning regulations and liberalization of rental markets, and greater efficiency and speed in the building process, while ensuring coordination and consistency with the policy measures aimed at reducing nitrogen depositions (¶61). — MoF, MOIKR — MT *Source: Technical Note prepared as part of the 2024 FSAP for the Netherlands.* ### 5. The institutional settings of the AFM and DNB for macroprudential policymaking ### 5. The institutional settings of the AFM and DNB for macroprudential policymaking ### Institutional mandates and roles - DNB has an explicit mandate for financial stability in addition to its responsibility for microprudential supervision under the amended Bank Act that became effective in January 2014. - Since January 2014, the macroprudential policy role of the AFM is clarified in its mandate that its activities are conducted “also in the interest of financial stability”. - The MoF is politically responsible for the overall functioning of the financial system and for financial market regulation, making rules to ensure a stable financial system, including limits on LTV and DSTI ratios. - The ECB can apply more stringent measures to the Dutch banking system, including higher capital buffers (the so-called “topping-up power”). - The European Systemic Risk Board (ESRB) can issue recommendations on a “comply or explain” basis to member states. ### Policy trade-offs and political constraints - The willingness to act of the MoF in relation to financial stability may be hampered by the pursuit of objectives not related to the safeguard of financial stability. - Limits on the DSTI and especially on the LTV ratio are used by the MoF as tools to protect consumers and to manage accessibility to homeownership and not exclusively to address risks to financial stability. - Example: Following the FSC’s recommendation in May 2015 and the 2017 FSAP recommendation to continue to gradually reduce the maximum LTV ratio after 2018 (from 100 to 90 percent), as well as DNB calls from the ESRB for a lower LTV limit since then (see, e.g., DNB’s Financial Stability Report, fall 2019), the MoF has refrained from tightening the LTV ratio limit further, arguing that the current limit was adequately mitigating the systemic risks of the housing market and that “in the interest of ensuring accessibility of the residential real estate market for first time buyers, the Dutch government [had] no plans of lowering the legally binding loan-to-value (LTV) limit beyond 100%" (see ESRB, 2020). ### FSC role, composition, and transparency - The FSC acts as a forum to identify risks and issue recommendations but is not responsible for the use of macroprudential tools. - The FSC was established effective November 2, 2012, by means of a Ministerial Decree to expedite its creation. - The FSC is a forum in which DNB, the AFM, and the MoF meet to identify and discuss potential risks to financial stability and ways to mitigate them. - The FSC does not take policy decisions; instruments under CRR II and CRD V are assigned to DNB, and limits on LTV and DSTI ratios are assigned to the MoF. - Transparency: a summary of each meeting is published on the FSC website to make warnings and recommendations public. The FSC is accountable to the Parliament via the MoF and produces an annual report for the MoF, which sends a copy to Parliament. - Composition: The FSC is composed of seven representatives: three from DNB, and two each from the MoF and the AFM, and is chaired by the President of DNB. The MoF does not have any voting rights. ### DNB responsibilities and communications - DNB is responsible for both microprudential policy (as the National Designated Authority, NDA) and for macroprudential policy (as the National Competent Authority, NCA) for the purpose of the CRR II and CRD V. - DNB collaborates with the ECB under the Single Supervisory Mechanism (SSM), the ESRB, and the European Banking Authority (EBA) for macroprudential policy for banks. - For non-banks, DNB and the AFM conduct a joint risk assessment and collaborate with the ESRB, EIOPA, and ESMA. - DNB communicates macroprudential developments mainly via the bi-annual Financial Stability Report (FSR). - Other communications: DNB publishes documents on its financial stability task and a framework to set the Countercyclical Capital Buffer (CCyB) rate; the AFM publishes the Rapportage Financiële Stabiliteit (in Dutch); the CPB produces an annual financial stability report per request from Parliament. - The 2017 FSAP recommended publishing a summary of DNB Governing Board meetings on macroprudential policy decisions; this recommendation was not implemented. ### Legal and information powers - DNB has direct access to information from regulated entities and legal powers to collect data from non-regulated entities (excluding households). - DNB has been calibrating and publishing the countercyclical capital buffer rate for the Netherlands since 2016 and imposed a minimum risk weight on banks’ mortgage portfolios effective January 2022. - DNB can designate individual financial institutions as systemically important and require additional capital buffers. - The MoF can recalibrate limits on LTV and DSTI ratios and adjust their perimeter in collaboration with the Ministry of the Interior and Kingdom Relations (MOIKR). ### Institutional strengthening and limitations - The FSC’s legal status was strengthened by establishing it in primary legislation: Parliament approved a bill after an amended legislative proposal and the FSC is legally embedded in the Bank law 1998 (Bankwet 1998) as of July 1, 2023. - The current institutional arrangements do not provide sufficient powers to ensure the ability to act of the FSC over the calibration of macroprudential tools due to the absence of a “comply or explain” mechanism. - The consideration by the MoF of access to homeownership when managing borrower-based tools (limits on LTV and DSTI ratios) can create an inaction bias detrimental to financial stability. - The authorities explained that an “act-or-explain” mechanism vesting FSC powers over the government would not fit within the Dutch constitutional system because it would alter government accountability to Parliament. ### Coordination and information-sharing - The FSC meets three times a year to discuss current financial stability risks, providing a formal coordination mechanism across DNB, the AFM, and the MoF. - An MoU supports information sharing regarding financial stability between DNB and the AFM and the MoF, respectively. - Recent strengthened collaboration on non-bank issues: during 2018 and 2019 DNB, AFM, and MoF cooperated on Brexit preparedness; in 2022 DNB and AFM jointly worked on an initial analysis of liquidity risks for pension funds leading the FSC to recommend further analysis and possible measures. - DNB and the AFM actively cooperate at the European level (ESRB, FSB, EBA, ECB) and participate in ESRB General Board, Advisory Technical Committee (ATC), and technical working groups. - The ESRB role: issues “comply-or-explain” recommendations to Member States; has taken a hub-function to streamline notification and information sharing, improving timeliness of national measure adoption. ### Ability to act summary - National authorities have broadly adequate powers to collect information for macroprudential policy and to change the level and regulatory perimeter of macroprudential instruments. - Weakness: FSC’s practical ability to influence calibration of macroprudential instruments remains limited without a “comply or explain” mechanism; MoF’s political considerations about homeownership can impede timely tightening of borrower-based tools. ### Recommendation (explicit) - Recommendation 1: The authorities should consider elevating the FSC to a permanent advisory body and vesting it with semi-hard powers, or vesting DNB with hard powers over the calibration of the borrower-based tools, with some guardrails for DNB to use these tools in a proportionate way. - Rationale: Current institutional arrangement does not guarantee that calibration of the LTV limit is in line with the pursuit of the financial stability objective; consideration of access to homeownership can interfere with MoF willingness to act for financial stability. - Options: - Elevate the FSC to a permanent advisory body and vest the FSC with semi-hard powers over macroprudential tools. - Transfer hard powers over calibration of the LTV and DSTI limits from the MoF to DNB (consistency suggests both DSTI and LTV limits should be under the same authority). - Guardrails suggested if DNB is vested with hard powers: - Use proportionate limits on tightening (example: LTV limit should not be tightened too fast – e.g., by no more than one percentage point a year, as suggested by the 2015 FSC recommendation). - DNB could use cost-benefit analyses to inform choices and guide optimal calibration of borrower-based measures; these analyses could be published for transparency. ### Systemic risk monitoring — Assessment (selected findings) - The Dutch financial system is generally healthy, but some pockets of vulnerabilities deserve close monitoring. - Banks: - The banking sector appears resilient to adverse macrofinancial shocks, but some vulnerabilities exist. - A solvency stress test on significant institutions reveals that the sector would remain sufficiently capitalized even if all the main risk factors materialize simultaneously. - Some banks might need additional resources to maintain a comfortable buffer position after consecutive years of weak earnings and upcoming increase in the countercyclical capital buffer. - Liquidity buffers appear generally sufficient, though close monitoring of banks’ ability to handle sudden and severe runoffs in foreign currencies would be useful. - Sensitivity analysis incorporating potential deposit outflows and solvency-liquidity interactions generates additional losses that require banks to further draw down their capital buffers. - Insurance: - The Dutch insurance sector, particularly the property and casualty and health insurers, appear broadly resilient to macrofinancial shocks. - Vulnerabilities exist for some life insurers in an environment of further rising interest rates. - Liquidity risks from margin calls appear largely contained. - Pension funds: - Pension funds benefit from further rising interest rates, on top of their already-strong funding ratios. - A bottom-up analysis shows pension funds are resilient to liquidity risks from margin calls, even when access to the repo market is restricted. - Repo markets remain important sources of liquidity, especially for the largest pension funds; close monitoring of market conditions and liquidity risk management practices remains crucial. - Households: - Under a severe stress test scenario, the probability of default of mortgages would in general increase only moderately. - Lower-income households (with monthly income in the first quartile) and younger borrowers (under 35) with a mortgage service-to-income ratio above the NIBUD DSTI limit would see their proportion increasing the most. - The proportion of households with high joint LTV and DTI ratios would almost triple (although from a low level). - Corporate: - The adverse scenario results in a marked increase in the proportion of firms facing debt repayment difficulties or higher borrowing needs. *Source: 1nldea2024007 - 5. The institutional settings of the AFM and DNB for macroprudential policymaking* ### 19. DNB has a dedicated division and uses a variety of indicators and models for systemic ### 19. DNB has a dedicated division and uses a variety of indicators and models for systemic risk monitoring. ### Institutional setup and monitoring framework - Financial Stability Division (FSD) comprises two departments: Macroprudential Analysis and Policy (MAB) Department and International Financial Architecture (IFA) Department. Both are "staffed adequately." - MAB responsibilities: - Produces the semi-annual DNB FSR. - Conducts real estate market analyses and stress testing. - Coordinates the Dutch Financial Stability Committee. - IFA responsibilities: - Coordinates DNB’s contributions in international fora, such as the IMF. - Provides input into regulatory discussions. - Strong focus on non-bank financial intermediation. - DNB monitors indicators for: - Structural dimensions: concentration risk and system relevance and interconnectedness. - Temporal dimensions: credit growth, leverage, real estate market, and bank liquidity. - International risks (as outlined in its financial stability task document). - Publication channels: - Macroprudential indicators published as an annex of the bi-annual FSR. - Analyses also appear in Bulletins, Occasional Studies, and FSR chapters. - Netherlands-specific Financial Stress Index (FSI) developed as an application of the ECB systemic stress indicator. ### Quantitative models and tools used or under development - Regular and ad-hoc quantitative models include: - Top-down stress test model to assess solvency risks for the banking sector (outputs regularly published in the FSR). - Real Estate Vulnerability Assessment model to quantify credit risks on banks’ RRE and CRE loan portfolios. - Climate stress test model to quantify flood-related climate risks for banks. - Firm-level stress test tool to quantify risks for the Dutch NFC sector. - Liquidity risk tool for the banking sector. - Recent and pending analytical work: - Study on interconnectedness in the Dutch interest rate swap market (DNB, 2021). - Tools under development: - Stress test tool for insurers. - Structural "interconnectedness" tool to map financial system exposure linkages across banks, pension funds, insurers, and investment funds, and asset types (loans, derivatives, securities, and money market instruments). ### Progress closing data gaps since 2017 FSAP - CRE data: - DNB developed a monitoring framework at a quarterly frequency and collects granular information on bank CRE loans. - Collected data include transaction price and valuations, supply and demand, vacancy rates, investments, exposures of banks and non-banks, credit risk of bank CRE loans (LTV, IFRS stage allocations, NPLs), financial market developments (Real Estate Investment Trusts), and investor survey information. - Netherlands graded "fully compliant" (see June 2021 Summary Compliance Report) with respect to ESRB recommendation on closing real estate data gaps (ESRB/2019/3). - Corporate sector data: - DNB developed a corporate monitor distributed internally at a quarterly frequency (based largely on supervisory and publicly available data). - Monitoring is being complemented by more granular SME and corporate lending data from Anacredit. - DNB performed ad-hoc firm-level data analysis to assess corporate sector risks. ### AFM action on Interest-Only Loans (IOLs) - AFM launched a multi-year supervisory program to monitor vulnerabilities from IOLs: - Issue raised in 2016; approach rolled out in 2019. - Mortgage providers have contacted 1.68 million customers with an IOL or part of IOL. - A total of 370,000 customers have been tested for future affordability of the IOL. - AFM conducted a 2021 detailed analysis on vulnerabilities from maturing IOLs (Rapport over financiële kwetsbaarheid huishoudens met aflossingsvrije hypotheek). ### Credit Registration Office (BKR) reforms and data access - BKR background: - Established in 1965 to share information about borrowers' overall debt burden. - Currently a private institute only allowed to share micro data with suppliers of information for checking excess credit. - Legal and institutional reform: - MoF decided to make the registration of credit a statutory task and assign it to BKR, which will remain a private institute but become a "legal person with statutory task" (rechtspersoon met wettelijk taak, or RWT). - Consultation for the new system of credit registration closed in August 2023. - According to the new law scheduled for approval in 2025: - AFM can on request receive data on an aggregated level. - Data will include, if the law is approved without modification, positive registrations (information on all outstanding loans of an individual, such as personal loans, consumer credit, credit card debt, etc.). - Mortgages will be excluded from the positive registration. - Conditions of credit (e.g., the interest rate) are not registered. ### Loan-level data (LLD) collection challenges and status - LLD coverage and history: - DNB started LLD collection in 2012. - LLD covers about 80 percent of all loans of Dutch households, and more than 90 percent of those issued by Dutch banks to residents. - DNB also collects mortgage loan-level data from non-banks (insurers, pension funds, investment funds) on a biannual basis, covering 80 to 90 percent of non-bank mortgage loans. - Suspension and causes: - Since 2022Q2, collection of mortgage granular loan-level data from banks has been put on hold due to concerns raised by the Nederlandse Vereniging van Banken (NVB) that the LLD may contain identifiable personal data. - DNB assessing extent of identifiable personal data; discussions ongoing with the banking sector and the Dutch Privacy Authority and working with the MoF to resolve the issue. - DNB has asked the MoF to clarify the legal basis for collection of granular data; talks on required legal clarifications are ongoing. - Outlook: - Once clarity on privacy and legislative amendments (including a time path) are available, DNB plans to discuss resumption with banks. - It might take at least two years for DNB to resume collecting the LLD, given the length of the legislative process. - Potential mitigation: - Option to use clustered and/or anonymized data with granularity sufficient to monitor macroeconomic risks. ### Housing price analysis and models - Empirical work to date: - DNB contributed to a 2020 BIS study on property price dynamics and drivers. - Published analysis on role of borrowing capacity for house price developments (Eijsink and van Dijk, 2022). - Bulletin on effect of market liquidity on property prices in commercial real estate markets (April 2021). - Gaps identified: - No comprehensive, regular tool to monitor overvaluation risks in the property market. - No comprehensive model to quantify contributions of various drivers of housing prices (including non-credit factors). - DNB has not actively collected and systematically used non-credit transaction information for regular monitoring, viewing its mandate as monitoring housing market developments to the extent they provide forward-looking information on mortgage risk. - Contextual facts: - Housing prices almost doubled from 2013 to 2022. - Higher housing prices due to non-credit funded transactions can have indirect consequences for mortgage riskiness. ### Recommendations (from the text) - Recommendation 2: - Ensure that authorities have clear legal basis to access granular transaction/loan-level data on a regular basis for risk monitoring and analysis, including residential and commercial real estate loans. - Rationale: Loan-level data is critical for DNB to monitor developments of the Dutch mortgage market based on granular information, and conduct stress test analysis (see Technical Note "Systemic Risk Analysis"). - Recommendation 3: - DNB should develop comprehensive empirical models to regularly monitor the drivers of housing prices. - Models should consider all possible drivers of housing prices, including non-credit factors, and use multiple aggregation levels of data (microeconomic, regional, and macroeconomic). - Rationale: Quantifying contributions of drivers will allow better targeting of macroprudential policy given the importance of housing price developments for financial stability in the Netherlands. ### Macroprudential instruments and recent DNB actions - Regulatory frameworks and available instruments: - Macroprudential instruments include tools under CRR II and CRD V, tools under AIFMD, and borrower-based tools under national law. - Instruments available include the countercyclical capital buffer (CCyB), property market-related tools (LTV and DSTI limits), sectoral systemic risk buffers, and liquidity measures (LCR and NSFR). - CRR II / CRD V specifics: - Five capital buffers: Capital Conservation Buffer (CCoB), Countercyclical Capital Buffer (CCyB), G-SII/OSII buffers, Systemic Risk Buffer (SRB), and sectoral buffers (Article 133 paragraph 5 letter b (i) of CRD). - Leverage ratio (LR) limit and measures on risk-weights and loss given default for RRE and CRE exposures. - Liquidity Coverage Ratio (LCR) and Net Stable Funding Ratio (NSFR). - DNB actions and numeric policy settings: - CCoB increased four times since June 2013 to reach 2.5 percent on January 1, 2019 (ECB and DNB allowed banks under direct supervision to operate temporarily below this level in response to the COVID-19 crisis). - LR set at 3 percent, announced May 14, 2019 (effective June 28, 2021). - Global SII LR surcharge: global SIIs must satisfy an LR surcharge equal to 50 percent of the global SII buffer rate since January 1, 2023 (initially planned for January 1, 2022; delayed due to COVID-19 crisis). - CCyB timeline and settings: - Introduced in 2016 and at 0 percent until raised to 1 percent on May 25, 2022 (effective one year later). - Raised from 1 to 2 percent on May 31, 2023 (considered its target "positive neutral rate"). - SRB: SRB imposed on three systemically important banks (ING Bank, Rabobank, and ABN AMRO Bank) at 3 percent from 2019 (phased in between 2016 and 2019) has been abolished. - O-SII rate: introduced April 2014; cut on May 31, 2023 (to come into force May 31, 2024) by 0.25 to 0.75 percentage points depending on institution. - Liquidity tools: - LCR requirement set immediately at 100 percent effective October 1, 2015. - NSFR of at least 100 percent announced May 20, 2019 (effective June 28, 2021). - Mortgage risk-weight floor under Article 458 of the CRR: - Introduced January 1, 2022; extended in October 2022 until December 1, 2024. - Minimum risk weight calculation: - 12-percent risk weight for the relative part of the mortgage with an LTV up to 55 percent plus a 45 percent risk weight for the remaining part of the loan. - Hence, the lower bound increases steadily with the LTV ratio. - Mortgages covered by the national mortgage guarantee (NHG) are exempted from the measure. *Source: IMF staff summary of chapter 19 from 1nldea2024007.* ### 30. There are no other measures among the flexibility measures of Article 458 of the CRR ### 30. There are no other measures among the flexibility measures of Article 458 of the CRR ### Flexibility measures under Article 458 of the CRR - The flexibility package under CRR Article 458 includes macroprudential instruments available for DNB and the ECB to address potential systemic and macroprudential risks at the member state level. - Apart from minimum risk weights for targeting asset bubbles in the residential property sector, there are no other measures among the flexibility measures of Article 458 that DNB has used or is considering using in the near future. - These tools are not currently active because authorities consider that corporate exposures do not pose systemic concerns at this moment. - DNB closely monitors potential vulnerabilities regarding banks’ exposures to the corporate sector and will consider using such tools if systemic risks materialize. ### Tools under the AIFMD - Article 25(3) of the AIFMD: DNB can “impose limits to the level of leverage that Alternative Investment Fund Managers (AIFM) are entitled to employ or other restrictions on the management of the Alternative Investment Fund (AIF) with respect to the AIFs under its management to limit the extent to which the use of leverage contributes to the buildup of systemic risk in the financial system or risks of disorderly markets.” - Article 46 of the AIFMD: AFM can “require the suspension of the issue, repurchase or redemption of units in the interest of the unit-holders or of the public.” - This ex-post macroprudential tool is the only one available to the AFM so far, has never been used, and the AFM is not planning to introduce macroprudential tools in the near future if these are not part of the EU law. - MiFIR offers a third (ex-ante) tool for ESMA or the NCA “to temporarily restrict or prohibit the marketing, distribution, or sale of financial products if this poses a threat to the stability of the whole or part of the financial system.” - DNB has never used the leverage cap instrument so far. - Authorities monitor Dutch real estate funds’ leverage and redemption terms and concluded that, despite the depressed outlook of the real estate sector, risks from those funds related to excessive leverage or to liquidity mismatches are limited in the current juncture (FSR, Autumn 2023). - Authorities contribute to international discussions on a macroprudential framework for investment funds: - AFM position paper (AFM, 2022). - Contributions to ESRB efforts; AFM chairs the policy task force (PTF) of the ESRB expert group on non-bank financial intermediation (NBEG). - Contribution to an ESRB “issues note” on risks and vulnerabilities related to investment funds investing in assets that are inherently illiquid or might become illiquid in times of stress; discussed at ESRB General Board meeting of June 2023. ### Macroprudential measures (high-level summary from Table 2) - Countercyclical capital buffer (CCyB): Wft; DNB, ECB; All banks and some investment firms; January 2016; Last change May 31, 2023 (2) - Capital conservation buffer (CCoB): Wft; DNB, ECB; All banks; January 2013; Last change January 1, 2019 - Limit on leverage ratio (LR) Article 92 CRR: DNB, ECB; All banks & Dutch G-SIB; May 2019; Last change January 1, 2023 - Global-Systemically Important Institution buffer (G-SII): Wft; DNB, ECB; ING Bank; January 2016; Last change December 1, 2022 - Other-Systemically Important Institution buffer (O-SII): Wft; DNB, ECB; ABN AMRO, ING Bank, Rabobank, BNG Bank, and Volksbank; January 2016; Last change May 31, 2023 (2) - Systemic risk buffer (SRB): Wft; DNB, ECB; Abolished since December 2020; January 2014; Last change December 2020 - Risk weight for RRE and CRE exposures Article 124.2 of the CRR: DNB, ECB; All banks and some investment firms; January 2014 - Loss given default for RRE and CRE exposures Article 164.5 of the CRR: DNB, ECB; All banks and some investment firms; January 2014 - Flexibility measures Article 458 of the CRR: DNB, ECB; Banks using IRB models to calibrate risk weights for their Dutch mortgage exposures; January 2014; Last change January 2022 - Liquidity Coverage Ratio (LCR) Article 460 of the CRR: DNB, ECB; All banks and some investment firms; October 2015; Last change October 2015 - Net Stable Funding Ratio (NSFR): DNB, ECB; All banks and some investment firms; January 2018; Last change June 2021 - Leverage limit on AIFM Art. 25 of AIFMD: DNB; Investment firms; July 2011 - Suspension of redemption and subscription Art. 46 of AIFMD: AFM; Investment firms; July 2011 - Limits on LTV ratios Wft: MoF; All mortgage providers; January 2013; Last change January 2018 - Limits on DSTI ratios Wft: MoF; All mortgage providers; January 2013; Last change January 2024 - Mortgage tax deductibility Wib (3): MoF; Amortized mortgages within 30 years; January 2013; Last change January 2023 ### The Revised CCyB Framework (key elements and recent actions) - Revised CCyB framework published March 2022 implementing a 2 percent neutral rate. - Framework clarifies identification of cyclical systemic risk phases to guide CCyB calibration. - Four phases of cyclical systemic risk and CCyB responses: - Phase 1 (recovery): CCyB fully or partly released; can remain at 0 percent. - Phase 2 (normality): CCyB built up at a pace of 1 percentage point per year or reaches neutral level of 2 percent. - Phase 3 (increased risk): CCyB raised above neutral but not higher than 2.5 percent if at least one dimension poses high cyclical risk; a CCyB above 2.5 percent considered if multiple dimensions pose high or very high cyclical risk. - Phase 4 (materialization): CCyB fully or partly released. - DNB raised the CCyB rate from 0 to 1 percent in May 2022, and from 1 to 2 percent in May 2023 (both binding one year later). - The revised framework aligns with IMF recommendations and implements a “positive neutral rate.” ### Tools under national law: LTV and DSTI limits, affordability norms, and supervision - Limits on LTV and DSTI ratios introduced in January 2013, implemented by the MoF by ministerial decree, based on Article 4:34 Wft. - These caps apply to a comprehensive set of credit providers and loans, limiting regulatory arbitrage. - MoF sets caps on DSTI and LTV ratios by ministerial decree; DNB (prudential supervisor) and AFM (conduct supervisor) participate in formal consultation and in the FSC, which can issue non-binding recommendations. - For DSTI limits, MoF uses input from NIBUD (National Institute for Budget Education), which develops an annual table with maximum allowed DSTI ratios, taking a microprudential perspective based on available income for individual households. - DSTI calculation: based on actual mortgage interest rates and redemptions for a 30-year annuity mortgage; norms stricter for clients opting for (partially) IO mortgages (no mortgage interest tax deductibility). - Mortgage providers must apply a stressed DSTI for mortgages with interest rates fixed for less than 10 years, using a stressed interest rate set at a minimum 5 percent since its introduction, determined and published by the AFM. - Affordability norm applies to new loans and refinancings and there is no age limit. AFM supervises application of affordability norms as conduct supervisor. - LTV limit evolution: - Initially set at 106 percent in January 2013, expected to be reduced by 1 percentage point each year to reach 100 percent in 2018. - FSC issued a recommendation in 2015 to continue tightening by 1 percentage point every year after 2018 to reach 90 percent. - DNB has called for a lower LTV-limit; 2017 FSAP recommended “continue gradually reducing maximum limits on LTV ratio to no more than 90 percent after 2018 (...).” - MoF refrained from further tightening, citing mitigation of systemic risks and concerns about accessibility for first-time buyers. - DSTI system and procyclicality: - In the NIBUD system, maximum DSTI ratio increases with disposable income, contributing to procyclicality. - 2017 FSAP recommended introducing prudential ceilings beyond which DSTI caps by income cannot be relaxed across the credit cycle. - NIBUD (2020) concluded it would be feasible to impose maximum DSTI limits on high incomes; an alternative cap on net mortgage expenditures of 50 percent of disposable income was considered but estimated to have limited impact. - Government decided not to introduce such a cap; instead, changes in DSTI limits are averaged out over four years since 2017. - Despite strong household disposable income growth before COVID-19, DSTI limits have been tightened over time and remain well below the 40-percent maximum limit suggested by the literature. *Source: IMF staff summary of the PDF chapter content.* ### 36. The stressed interest rate of the DSTI system has been fixed since its introduction in ### 36. The stressed interest rate of the DSTI system has been fixed since its introduction in ### Stressed interest rate of the DSTI affordability test - The minimum key interest rate applied in the affordability test (for loans with interest rates fixed for less than 10 years) was calibrated in 2013 by averaging the 10-year fixed interest rate quoted by the six largest mortgage providers, which was approximately 5 percent in 2013. - Since 2013, the Ministry of Finance (MoF), responsible for setting this minimum key interest rate, has never adjusted the rate despite the recent increase in interest rates. - International examples of rule-based calibration are cited (Canada, Estonia): the stressed interest rate is defined as the maximum between a flat rate and the current mortgage rate augmented by a certain margin. - A possible rule for the Netherlands is proposed: define the stressed interest rate as the maximum between 5 percent (the flat rate imposed in the last decade) and the current mortgage rate augmented by a given margin. - Authorities’ concern: increasing the stressed interest rate in times of historically high mortgage rates would deter borrowers from using flexible-rate mortgages because credit conditions for those loans would become much tighter; many households would instead opt for a fixed-rate mortgage and not benefit from a lower debt service burden if mortgage rates decrease in the future. - Practical note: the proportion of households opting for a mortgage rate fixed for less than 10 years has been small in recent years, so the affordability test applies to a small share of borrowers. ### Tax treatment of owner-occupied housing (MID) and related reforms - Since 2013, tax deductibility of interest payments (MID) for new loans has been restricted to annuity or linear mortgages that are fully repaid in at maximum 30 years to reduce incentives for interest-only loans and household debt buildup. - The maximum rate of the MID has been gradually decreased from 52 percent in 2013 to 36.93 percent in 2023. - Assessment: despite the reduction, the current MID rate "remains too high" and the tax treatment of owner-occupied housing remains favorable compared to other forms of investment. - The 2017 FSAP recommendation: “reduce the ultimate tax deductibility rate to a tax neutral level.” - Authorities do not plan to further cut the MID rate but presented a long-term vision on the tax treatment of owner-occupied dwellings in February 2024, recognizing that tax neutrality depends on tax rates and the definition of the tax base of income from owner-occupied dwellings and income on investments in housing. - Additional fiscal distortion: homeowners face lower taxation on total wealth because equity built up in their own home remains largely untaxed, unlike tenants, whose entire assets are taxed in Box 3. ### Macroprudential policy and climate-related measures - DNB does not plan to introduce macroprudential tools so far specifically to address physical and transition risks from climate change but is working on climate-related risk assessment. - A differentiation of maximum DSTI limits based on energy labels will enter into effect in 2024: - Expected effects: (i) borrowers can take up higher debt amounts when purchasing higher-energy label properties, relative to equivalent borrowers; and (ii) borrowers can take up higher debt amounts when investments are made to increase the energy efficiency (and label) of the purchased property. - Clarification: the differentiation of DSTI limits based on the energy label cannot be considered a macroprudential measure to address climate risks if it loosens limits for more energy-efficient houses; a macroprudential objective would require tighter DSTI limits on loans financing purchase of lower energy-efficient houses. - DNB is investing resources and capacity to map out and quantify exposures and the size of main climate-related risks before considering activation of any macroprudential tool. ### Developments of the property market and household debt — residential real estate (RRE) - Residential real estate prices increased by 99 percent from their latest trough of June 2013 to their peak of July 2022. - The Dutch housing price boom was almost twice as large as the average housing price boom experienced in the euro area. - The boom was heterogeneous across regions, with large cities recording the largest increases. - Price-to-rent and price-to-income ratios are high by international standards and have hit historical highs, suggesting possible overvaluation. - ESRB risk dashboard indicates overvaluation of Dutch housing prices ranging from about 10 to more than 20 percent as of 2023Q2, depending on the approach. ### Commercial real estate (CRE) - CRE prices doubled between their low of mid-2014 and their peak of mid-2022, with large heterogeneity across segments: - Retail: depressed demand due to structural changes, worsened by COVID-19. - Offices: impacted by home working. - Industrial: benefited from the rise of online shopping. - Since mid-2022, price pressures appeared across all segments due to higher financing costs. - Authorities’ view: limited financial risks for banks because banks significantly reduced exposure to CRE; CRE now represents 7 percent of total bank assets. - Pension funds and insurers: CRE represents 7 and 8 percent of their assets, respectively, and much of their exposure is via investment funds. - DNB considers risks due to liquidity mismatches or excessive leverage limited among Dutch real estate investment funds. ### Household debt dynamics and leverage - Household debt to disposable income declined from a peak of 256.7 percent in 2011Q3 to 179.1 percent in 2023Q2. - Deleveraging drivers: - Active deleveraging (numerator effect): nominal stock of debt decreased in absolute terms during 2011-2015 due to credit tightening, borrower-based tools, and higher voluntary debt repayments (e.g., lower proportion of IOLs, tax-exempt gifts). - Passive deleveraging (denominator effect): from 2016, stock of debt increased again, but strong nominal disposable income growth (especially given higher inflation in recent years) reduced the debt-to-income ratio. - As of 2023Q2: - Cumulative change in level of debt since 2011Q1 contributed to increasing the debt-to-income ratio by 21.3 percentage points. - Cumulative change of disposable income contributed to decreasing the ratio by 99.0 percent points. - Net effect: total deleveraging of 77.7 percentage points of disposable income. ### Interest-rate fixation, interest-only loans (IOLs), and refinancing risks - Fixed-rate status (2023): - Less than 3 percent of Dutch mortgages have interest rates which are flexible in 2023. - 25 percent of mortgages will have flexible rates within the next 5 years. - On average, households that took out a mortgage at the beginning of 2022 had an average fixed-rate period on new mortgages of more than 15 years in 2022Q1. - Mortgage risk from higher interest rates is assessed as relatively limited for the next three years by household stress test analysis. - Interest-only loans: - 44 percent of total Dutch mortgage debt had an interest-only component as of 2022Q1, down from almost 59 percent in 2014Q1. - About 2.78 million households (out of about 3.82 million households with mortgage debt) have either a partial IO mortgage or a 100-percent IO mortgage. - 29 percent of the stock of IO debt matures between 2034-2039. - Regulatory and market measures reducing IOL use: - Since 2011, the IOL part of a mortgage loan has been capped at origination to 50 percent of the value of the dwelling according to the Dutch Code of Conduct for Mortgage Loans (with limited exceptions). - Since 2013, tax deductibility for new lending restricted to loans amortized according to 30-year annuity, thereby excluding IOLs for new lending. - NHG (guarantee up to 435,000 euros) is restricted to amortizing loans. - AFM supervisory program encouraging mortgage providers to address vulnerable customers with IO mortgages and inform customers on risks. ### Refinancing risk scenarios for IOLs - AFM (2021) assessment of refinancing risks of maturing IO loans: - In the basic scenario, 44,000 households might be left with some residual debt. - In a higher interest rates or housing market crisis scenario, 44,000 households. - In a very deep and protracted crisis, 147,000 households (about 5 percent of households with IOLs) might be left with residual debt. - Note: AFM analysis does not consider simultaneous risks of higher interest rates and lower housing prices. ### Institutional features of the Dutch mortgage market (summary) - Nationale Hypotheek Garantie (NHG): - NHG is a public voluntary insurance scheme for loans under 435,000 euros. - End-2021, NHG guaranteed approximately 196 billion euros of mortgage debt, about 25 percent of the total mortgage market, and 67 percent of the 405.000 home-pricing class. - NHG reimbursement: takes over 90 percent of residual debt after foreclosure if rules met; mortgage providers offer 0.2 to 0.3 percentage points lower interest rate to NHG borrowers. - NHG provided by a government-backed foundation (Waarborgfonds Eigen Woningen, WEW); DNB considers NHG a government guarantee. - Full recourse and foreclosure practices: - Mortgage providers have automatic right of recourse to borrowers’ income and right to sell the house when arrears occur without a court order; redemption period of two months. - Remaining debt after sale must be paid by the consumer. - Strong recourse laws and practices contribute to high repayment incentives. - Outcomes: - Annual losses on Dutch residential mortgages have not exceeded 0.2 percent since 1970. - During the GFC, 34 percent of owner-occupied households had underwater mortgages without causing material extra defaults or arrears. *Source: 1nldea2024007 - 36. The stressed interest rate of the DSTI system has been fixed since its introduction in (IMF PDF chapter).* ### 2.5 percent of households with an owner-occupied home were in arrears on their mortgage payments, ### 1nldea2024007 - 2.5 percent of households with an owner-occupied home were in arrears on their mortgage payments, ### Mortgage arrears, debt risk, and loss-given-default - 2.5 percent of households with an owner-occupied home were in arrears on their mortgage payments, and this fell quickly afterwards. - However, the relatively high loan-to -value ratios for Dutch residential mortgages make the residual debt risk and the potential loss given default large. ### Factors driving the 2013-2022 housing price boom - A higher borrowing capacity - Mortgage interest rates fell by about 2 percentage points (from 3.7 percent to 1.7 percent). - Dutch households’ average nominal income increased by more than 30 percent from June 2013 to June 2022. - DNB analysis: a 3.5-percent increase in borrowing capacity is associated with a 3.9-percent rise in house prices after around 20 quarters (model caveats: supply-side variables not significant; potential endogeneity). - A rising proportion of transactions not backed by credit, including by BTL investors - House sales without an accompanying mortgage more than doubled from 8 percent in 2008 to nearly 17 percent in 2017 (Land Registry figures). - BTL investors reallocated funds toward real estate in a low interest rate environment and search for yield. - Fundamental factors: dynamic economy, demography, and limited housing supply - MOIKR estimate: demand exceeded supply by 279,000 houses in mid-2021. - Persistent tax incentives for homeowners with generous borrowing standards - MID rate decreased since 2013 but remained financially attractive to borrow. - High LTV ratio limit at 100 percent since 2018. - Possibility to use IOLs (up to 50 percent of the total value of the mortgage). ### Buy-to-let (BTL) investors: growth, market share, and systemic implications - BTL investor presence increased significantly during the boom. - At end-2022, about 15 percent of the total housing stock and 33 percent of rental properties were owned by BTL investors. - According to Kadaster data, BTL investors’ purchases represent 40 percent of all housing transactions in the four major cities and fluctuate around 20 percent for the rest of the Netherlands. - Market composition and effects - The private rental sector represents approximately 13 percent of the total housing stock; owner-occupied homes represent more than 57 percent; social housing about 30 percent. - Many BTL transactions are not credit funded: in 2016, 61 percent of houses of owners of multiple properties were purchased without any mortgage (Dutch Land Registry). - Share of BTL loans in the RRE portfolio (natural person borrowers) of Dutch banks is less than 0.5 percent of banks’ total exposures as of 2022Q1. - Total assets under management invested in dwellings by investment funds equal about 35 billion euros, representing less than 4 percent of total assets of all investment funds. - Price and stability impacts - Research indicates BTL investors contribute to higher prices and may increase housing market instability. - Cross-region panel regressions: a one-standard deviation increase in the proportion of BTL investors is associated with an increase in the quarterly real growth rate of house prices the next quarter of 0.1 to 0.2 percentage points (effect persistent up to five quarters). - BTL investors’ demand is significantly explained by the growth rate of the rent index and by the change in the mortgage rate, but not by housing price growth of the previous quarter. - If interest rates remain high for a prolonged period, some investors may reallocate funds away from real estate, potentially creating selling pressures and price drops in areas with large prior BTL activity. ### Non-bank lending, banks’ exposure, and recent price developments - Non-bank mortgage lending trends - Mortgage lending by insurers, investment funds, and pension funds increased over the past decade; OFIs experienced a decline. - Overall, the share of non-banks in total mortgages has been stable. - Banks remain the largest mortgage providers, with more than 550 billion euros of outstanding loans (representing about 20 percent of their total assets), and more than 69 percent of the total volume of mortgages as of 2023Q2. - Residential property price trajectory - Residential property prices began to decrease by the end of 2022 as interest rates increased. - As of 2023Q2, housing prices declined by 5.7 percent from their peak of 2022Q3. - Starting from 2023Q3, the housing market showed signs of stabilization and the price index slightly increased, supported by a tight labor market, the large housing shortage, and stabilizing mortgage rates. ### Effectiveness of macroprudential instruments for RRE - General findings - Studies find macroprudential measures in the Netherlands have been effective in improving the quality of loans, but less so in avoiding a build-up of indebtedness. - Evidence that stricter borrower-based measures are associated with a lower probability that borrowers fall under negative equity or default on their mortgage. - Mixed evidence on indebtedness and scope of measures - Literature offers mixed findings on effectiveness in limiting households’ indebtedness, possibly because the 100 percent LTV limit is not tight enough and measures have focused on resilience rather than constraining debt growth. - Caloia (2022): LTI limits efficient in limiting indebtedness of low-income households but not of high-income households; progressive reductions of the LTV limit impacted up to 45 percent of highly leveraged borrowers. - Everett et al. (2021): no evidence that borrower-based tools or lender-based measures constrained mortgage growth in the Netherlands. - Research-based quantitative indications - DNB (2015): 65 percent of households that took out a mortgage as first-time buyers during 2004-2012 were underwater with their mortgage as of end-2013; if the LTV limit had been 90 percent (instead of 106 percent by end-2012), only 13 percent of these households would have been underwater. - De Haan and Mastrogiacomo (2020): probability of mortgage non-performance over 1996-2015 is positively explained by originating loan-to -value ratio (OLTV) and by current DSTI; to avoid acceleration of non-performance probabilities, OLTV-limit should be about 70-80 percent for uninsured mortgages and about 90 percent for those with mortgage insurance. - Some studies assess limited negative side effects of borrower-based tools on access to homeownership. *Source: KINGDOM OF THE NETHERLANDS—THE NETHERLANDS, INTERNATIONAL MONETARY FUND (excerpts).* ### 54. Macroprudential measures have been associated with a better quality of mortgages on ### 54. Macroprudential measures have been associated with a better quality of mortgages on ### Mortgage quality and borrower behavior - Average LTV ratio at origination and average LTV ratio of current loans have declined in the past decade, although the decrease of the second ratio partly reflects the housing price boom and voluntary repayments of mortgage debt, and not exclusively the tightening of the LTV limit over 2013-2018. - The NPL ratio on mortgages has also declined from already low levels for RRE loans. - Averages hide large differences across borrowers: more recent first-time (often young) buyers have to repay a larger mortgage debt than older homeowners who built equity during a rising market (see Technical Note “Systemic Risk Analysis”). - Given booming house prices, households have increasingly pushed borrowing limits in recent years, implying an increase in debt-to-income ratios: - The LTI on the production of new loans increased from about 400 percent in 2017-19 to now more than 460 percent. - Around 60 percent of borrowers under the age of 36 have a debt-to-income ratio above 450 percent. - 45 percent of older borrowers have a debt-to-income ratio above 450 percent. ### Risks from house price dynamics and loan structures - The almost doubling of housing prices could pose risks to financial stability going forward. - Both capital-based and borrower-based measures have been introduced and tightened in the past decade, but these measures could not counter effects from very low interest rates on house prices. - A DNB (2022b) scenario analysis indicates: - If house prices were to fall by 20 percent, the LTV ratio of 13 percent of homeowners would rise to over 90 percent. - Such a price fall would cause 8 percent of homeowners to go into negative equity. - Risks from the refinancing of IOLs remain: - Large amount of IOLs maturing around 2036 could leave households with residual debt if higher interest rates and a drop in disposable income of retired households coincide with a drop in housing prices. - AFM and DNB analyses suggest such risks are unlikely to be of a systemic nature (see Technical Note “Systemic Risk Analysis”). - The amortization requirement (mandatory for at least 50 percent of total debt) contributed to reduce the LTV ratio, e.g., from 100 to 90 percent in the first 5 years of a loan. ### Assessment of the LTV limit and transfer tax measures - The LTV limit at 100 percent is inefficient in containing procyclicality during booming markets and does not provide protection to borrowers in case of a price correction: - LTV limit reduced from 106 percent in 2012 to 100 percent in 2018. - With an LTV of 100 percent at origination, households do not need equity to purchase a house and are exposed to higher probability of negative equity in a price correction. - Part of this risk is covered for some mortgages by the NHG, transferring risk to the government. - The inefficiency in limiting debt buildup could reflect loose calibration of the LTV limit. - Transfer tax differentiation introduced to improve the position of owner-occupiers relative to BTL investors: - From January 1, 2021 the acquisition of real estate not used as a principal residence was subject to a general tax rate of 8.1 percent, and to the rate of 10.4 percent from 2023 (before, the general tax rate was 2 percent). - First-time homeowners under 35 and for a property value under 440,000 euros are exempted from the tax. - Main residents pay the reduced rate of 2 percent, while other acquirers pay the general rate. - The measure will be first evaluated by authorities in 2024 and the results will be published in 2025. - The differentiation of the transfer tax was not introduced with a financial stability objective but could be integrated into the macroprudential toolkit should BTL investors’ activity gain systemic importance. - Stamp duties targeting BTL investors directly address the source of residential price pressures and can be less distortive than broader macroprudential credit measures, given many BTL transactions are not backed by credit. - However, deterring BTL investors can negatively affect supply of private rental housing; calibration, timing, and perimeter must be carefully designed. - The cumulative increase in transfer tax applied to BTL investors might have been too large; a more gradual tightening could have been preferable. - Exceptions could be granted for long-term private investors who contribute to affordable rental supply. ### Supply-side constraints and housing policy - Supply-side measures are critical to limit house price pressures and improve access to homeownership: - Government plan to build 100,000 houses per year and reach a total increase of 900,000 houses by 2030 (MOIKR, 2022), later revised to 981,000 units to accommodate higher than anticipated household growth, is welcome. - Without alleviating current zoning constraints, the plan is unlikely to be fully achieved. - Climate-related objectives (e.g., nitrogen deposition constraints and permit challenges after the 2019 Council of State ruling on PAS) affect location and speed of building, requiring whole-of-government coordination. - DNB notes increasing housing supply alone may be insufficient to address high prices and rents in the most popular areas. ### Policy recommendations (FSAP) - Authorities should fine tune a multipronged and “whole-of-government” approach to manage risks from the RRE market through the FSC, employing demand- and supply-side tools (credit-based, fiscal-based, administrative). Specific FSAP recommendations: - Recommendation 4: The LTV limit should be further tightened below 100 percent. The maximum limit on the LTV ratio should decline to 90 percent through a gradual reduction of 1 percentage point per year. - Recommendation 5: Continue efforts to monitor and address fragilities from IO mortgages, including by increasing incentives for borrowers to lower their exposure to these mortgages. Pursue incentives to switch IOLs to amortizing mortgages or to prepay loans voluntarily; keep monitoring vulnerabilities from IOLs maturing around 2036. - Recommendation 6: The MoF should gradually remove the MID. The MoF has reduced the MID rate to 37 percent, but further cuts are required until the tax incentive is fully phased out. - Recommendation 7: The differentiation of the transfer tax on housing transactions could be integrated into the macroprudential toolkit should BTL investors’ demand reach systemic levels, but would need careful calibration in size, timing, and perimeter; avoid excessively deterring BTL activity and consider exemptions when BTL investors contribute to provision of new housing. - Recommendation 8: Continue to address supply shortage of residential rental and owner-occupied housing by targeted measures, such as looser zoning regulations, liberalization of rental markets, and greater efficiency and speed in the building process, while ensuring coordination and consistency with policies aimed at reducing nitrogen depositions. *Source: International Monetary Fund (excerpt).* ### 2. Several studies find evidence of a significant impact of BTL investors on house prices ### 2. Several studies find evidence of a significant impact of BTL investors on house prices ### Evidence from micro-level studies - Jellesma (2020), using more than 500,000 property-level observations between 2015 and 2017, finds: - No significant national price effect of BTL purchases for the whole country. - Significant city-level effects in regressions run separately for the four largest cities: - If a property is bought by an investor, the transaction price is 9.6 percent higher in the case of Amsterdam. - If a property is bought by an investor, the transaction price is 8.9 percent higher in The Hague. - Hans et al. (2019) find regional differences; in Amsterdam, private investors with ten or more homes have paid significantly more for the same home in more recent years of their sample (2014-2018) than first-time buyers. ### Econometric approach at the regional level - Sample and frequency: - 12 regions with quarterly series over 2009Q1-2022Q4. - Main dependent variable: - Growth Prices = quarterly growth rate of the real residential price index (deflated by the CPI) in region r and quarter t. - Key explanatory variables (lagged in the baseline specification): - BTL = proportion of transactions by BTL investors (in total transactions). - Owner Occupiers = proportion of transactions by owner-occupiers (in total transactions). - Growth Housing Sales = quarterly growth rate of total housing sales. - Growth Rents = growth rate of the rent index (in real terms). - Growth Unfilled Vacancies = quarterly growth rate of unfilled vacancies (proxy for labor market activity). - Growth Stock = growth of the stock of housing. - Controls and estimation: - Time fixed effects (λt) and region fixed effects (λr). - Robust standard errors clustered by region. - Additional regressions include the change in the real mortgage rate (without time fixed effects). ### Main regression findings (Appendix Table 1 and text) - BTL investors and house price growth: - The growth rate of residential real estate prices is significantly related to BTL investors transactions. - A one-standard deviation increase in the proportion of BTL investors (+5.6 percentage points) is associated the next quarter with: - an increase in the quarterly growth rate of the house price index of more than 0.1 percentage point based on regression (3); - around 0.2 percentage point according to regressions (4) and (6). - Comparison benchmark: standard deviation of the quarterly growth rate over the estimation sample is less than 2 percentage points. - Owner-occupiers and flippers: - Purchases by owner-occupiers are not significant. - Purchases by flippers (buyers reselling in the next 6, 12, or 24 months) are not significant. - Real mortgage rate: - A one-standard deviation increase in the real mortgage rate is associated with a drop of the quarterly housing price growth of almost 0.2 percentage points. - Robustness: - Results are robust to various specifications, including incorporating the change in the real mortgage rate and dropping time fixed effects. ### Dynamic and cumulative effects (local projections; Appendix Table 2) - Persistence: - The effect of BTL investors’ purchases is persistent, affecting housing price growth over several quarters. - Local projection results: - Using Jordà (2005) local projection approach with controls from regression (3). - The effect of BTL transactions is significant in explaining cumulative growth of real housing prices for up to five quarters ahead. - At the peak, a one-standard deviation increase in BTL investors’ purchases is associated with a cumulative increase of real housing prices of almost 0.3 percentage points. - Appendix Table 2 coefficients for BTL purchases share, lagged (t to t+5): - t: 0.021** (t-statistic (2.25)) - t+1: 0.034** (2.42) - t+2: 0.037*** (4.05) - t+3: 0.046*** (3.29) - t+4: 0.051** (3.02) - t+5: 0.042 (1.59) ### Tests for reverse causality (Appendix Table 3 and text) - Regression estimated: BTL share on lagged BTL, lagged real house price growth, and controls. - Key results (Appendix Table 3): - The proportion of BTL investors' purchases in total transactions is never related to house price growth (at least in the short term). - BTL share is significantly and positively explained by the lagged growth rate of rents (regression (2)): - Real rent index growth, lagged: 0.083** (t-statistic (2.34)) - BTL share is negatively related to changes in the mortgage rate (regression (3)): - ∆Real mortgage rate, lagged: -4.351** (t-statistic (-2.54)) - Interpretation: - Results indicate no short-term reverse causality from housing price growth to BTL activity; instead, BTL demand responds to rents and mortgage rates. ### Selected regression statistics and sample details - Appendix Table 1 (selected coefficients): - Real house price index growth, lagged: coefficients include -0.273***, -0.195*, -0.210**, 0.597***, -0.269***, 0.596*** (t-statistics shown in parentheses). - BTL purchases share, lagged: 0.016*, 0.043**, 0.021**, 0.058***, 0.057*** (t-statistics shown). - Total housing sales growth: 0.012**, 0.006, 0.009*, 0.007**, 0.013**, 0.005. - ∆Real mortgage rate, lagged (when included): -1.532*** (t-statistic (-3.88)). - Sample size and fit (Appendix Table 1): - Nb. of regions: 12 in each specification. - Nb. of observations: 602, 266, 470, 470, 626, 470 (by column). - Adjusted R-squared: 0.862, 0.745, 0.869, 0.438, 0.867, 0.445 (by column). - Appendix Table 2 model fit: - Nb. of regions: 12. - Nb. of observations: 470 (for each horizon). - Adjusted R-squared by horizon (t to t+5): 0.869, 0.928, 0.938, 0.941, 0.942, 0.938. ### Variable definitions and data sources (Appendix Table 4) - House price index: Quarterly house price indices for 12 and 40 (COROP) regions and the four largest cities. Source: CBS. - Consumer prices: Quarterly Consumer Price Index (2015 value = 100). Source: CBS. - Buy-to-let investors’ demand: Quarterly share of BTL (corporate and individual) investors’ purchases in total transactions by region. Sources: Kadaster and IMF staff calculations. - Owner-occupiers’ demand: Quarterly share of owner-occupiers purchases in total transactions by region. Sources: Kadaster and IMF staff calculations. - Flippers’ demand: Quarterly share of flippers’ purchases (resold within 6, 12, or 24 months) in total transactions by region. Sources: Kadaster and IMF staff calculations. - Number of sold dwellings: Quarterly series of total number of dwellings sold for 12 regions and the four largest cities. Source: CBS. - Stock of housing: End-of-quarter stock of dwellings for housing function, by region. Source: CBS. - Total housing sales: Quarterly volume of total housing sales by region. Source: CBS. - Mortgage rate: Quarterly average mortgage rate on total loans for house purchase. Source: DNB. - Rent index: Quarterly index of rents by region. Sources: Pariarius and IMF staff calculations. - Unfilled vacancies: Quarterly volume of unfilled vacancies by region. Source: CBS. *Source: 1nldea2024007 - 2. Several studies find evidence of a significant impact of BTL investors on house prices (IMF PDF).* --- _Source: https://www.imf.org/-/media/files/publications/cr/2024/english/1nldea2024007.pdf_