Brief 009Financial fragility

Where does resilience become financial stress?

Rates, prices and leverage transmit the same shock differently. Aggregate balance sheets can look sound while recent borrowers, liquidity-poor households, rate-sensitive budgets or particular institutions carry the risk.

Evidence vintage 26 July 2026World → Europe → NetherlandsDefinitions first
  1. 01 Where now?
  2. 02 Which direction?
  3. 03 Who carries it?
  4. 04 Why?
  5. 05 What changes our mind?

01 Where are we now?

Four balance sheets. Four different failure modes.

Household leverage cannot stand in for sovereign leverage, and neither establishes bank or pension resilience. Start with the relevant stock, denominator and shock channel.

Households

93.8%household & NPISH credit / GDPNL · 2025-Q4 · gross stock, not debt service

Government

44.4%gross debt / GDPNL · 2025 · not net debt or interest burden

Banks

15.99%CET1 / risk-weighted assetsEuropean significant institutions · 2026-Q1 · capital, not liquidity

Pensions

122.9%assets / technical provisionsDutch funds · 2025-Q4 · coverage, not adequacy

Stock ≠ service. Debt outstanding says how leveraged a sector is; debt service says how much income meets interest and amortisation.

Flow ≠ buffer. The saving rate is a flow. It does not locate accessible cash or show how long a household can cope.

Safeguard ≠ outcome. Capital, liquidity and pension funding are distinct buffers—not household welfare measures.

Sources: BIS WS_TC household credit ↗, Eurostat government debt ↗, ECB supervisory statistics ↗, DNB Annual Report 2025 ↗.

02 What direction are we moving?

Leverage fell. Repayment relief slowed. New money stayed expensive.

The three series below deliberately keep unlike denominators and units in separate panels. A lower debt/GDP stock can coexist with a higher payment burden for the borrowers who reset today.

Netherlands · quarterly

Household & NPISH credit

93.8% of GDP
Dutch household credit as a percentage of GDPThe ratio fell from 119 percent in late 2015 to 93.8 percent in late 2025.
Gross credit stock from all sectors. The GDP denominator can move faster than debt.
View data and exact scope
Selected household-credit observations
PeriodCredit / GDP
2015-Q4119.0%
2020-Q4114.6%
2023-Q497.0%
2024-Q494.3%
2025-Q493.8%

Source: BIS WS_TC 2.0 · Q.NL.H.A.M.770.A ↗.

Netherlands · modelled quarterly

Household debt service

13.0% of income
Dutch household debt-service ratioThe BIS modelled share of income used for interest and amortisation declined after 2015 and edged higher from its 2024 low.
Interest plus amortisation under the BIS unified aggregate model. Prefer within-country movement to absolute country ranking.
View data and exact scope
Selected debt-service observations
PeriodDebt service / income
2015-Q418.8%
2020-Q415.3%
2023-Q413.0%
2024-Q212.8%
2025-Q413.0%

Source: BIS WS_DSR 1.0 · Q.NL.H ↗.

Netherlands · monthly

Rate on new mortgage business

3.67% per year
Rate on new Dutch mortgagesNew mortgage pricing rose sharply after 2022 and remained above the ultra-low-rate period in May 2026.
Marginal new-business pricing. This is not the rate on the outstanding stock or a schedule of future resets.
View data and exact scope
Selected new-mortgage-rate observations
PeriodAnnualised rate
2015-013.05%
2021-011.72%
2023-013.71%
2024-014.12%
2026-053.67%

Source: ECB MIR.M.NL.B.A2C.A.R.A.2250.EUR.N ↗.

Saving flow

17.3%of adjusted gross disposable income · 2025

National saving recovered after the pandemic spike eased. It does not tell us which household holds cash.

Source: Eurostat nasa_10_ki · SRG_S14_S15 ↗

Data gap · not a zero

Arrears are not defaults.

No whitelisted series here joins mortgage and consumer-credit arrears with a stable days-past-due clock, balance denominator and cure policy. Bank NPLs stay in the institutional panel below.

03 Who is benefiting—or carrying the cost?

Debt and buffers do not belong to the same representative household.

Tenure, income, borrower age, rate fixation and asset liquidity determine who feels a shock. National wealth cannot be netted against national debt to infer household safety.

Distributional household exposure: populations and channels remain separate
GroupExposureShock channelEvidence boundary
Recent, leveraged ownersLarge mortgage relative to income or home valueIncome loss, refinancing, house-price declineUnder-35 owner LTV averaged 0.64 in 2024; LTV is not payment-to-income.
Low-income and renter householdsFew liquid or housing assets; high essential spendingInflation and income lossMedian wealth: €1.5k in the lowest income decile; wealth is not liquidity.
Older ownersOften lower mortgage leverage and more housing equityValuation, pension and care costsGroup averages hide renters and households without accessible savings.
Pension participantsLarge deferred but inaccessible claimsFunding, valuation and transition outcomesFunding coverage cannot pay today's emergency bill.

Sources: CBS household wealth ↗, CBS financial sustainability ↗, DNB Annual Report 2025 ↗.

Fiscal channel Government interest burden

One interest bill. Two denominators.

Interest relative to GDP speaks to the economy's scale. Interest relative to government revenue speaks more directly to budget capacity. Neither is gross debt.

Interest / GDP0.7%Eurostat PC_GDP
Interest / revenue1.7%D41PAY ÷ TR × 100
Gross debt / GDP44.4%Maastricht debt · separate stock
Dutch general-government interest payable with separate denominators
YearInterest, €mInterest / GDPRevenue, €mInterest / revenue
20159,3801.3%303,5583.1%
20196,4550.8%365,1871.8%
20237,1250.7%458,5571.6%
20258,4960.7%510,1951.7%
Derived formula: D.41 payable interest in million euros ÷ total general-government revenue in million euros × 100. The GDP ratio is Eurostat's published ratio, not a recomputation from rounded values.

Sources: Eurostat gov_10a_main · S13 · D41PAY and TR ↗; gov_10dd_edpt1 · S13 · GD · PC_GDP ↗.

Institutional channel Bank and pension safeguards

Buffers matter—but they are not interchangeable.

Capital absorbs losses, liquidity meets cash outflows, NPLs record recognised problem credit, and pension funding compares assets with technical provisions.

Banks · Europe

Strong starting point, wide dispersion

CET1 ratio
15.99%
risk-weighted capital · 2026-Q1
NPL ratio
2.18%
ECB significant institutions · 2026-Q1
EBA NPL ratio
1.82%
different sample/denominator · 2025-Q4
Capital headroom
≈500 bp
above OCR + P2G · aggregate

Do not average these into resilience. EBA reports low aggregate NPLs alongside concentration in SMEs, consumer credit and commercial real estate.

Sources: ECB Q1 2026 ↗ · EBA Q1 2026 ↗

Pensions · Netherlands

Coverage is not adequacy

122.9%assets / technical provisions

The aggregate ratio improved at end-2025. It does not say whether benefits are adequate, indexed, equally funded across schemes or accessible to a household under stress.

Source: DNB Annual Report 2025 ↗

04 What are the competing explanations?

Similar stress can arrive through different mechanisms.

Leverage, income, rate reset, inflation, scarcity and repricing can reinforce or offset one another. Their labels below are part of the evidence, not decoration.

Mechanism

Leverage

Large debts amplify income and valuation shocks even when collateral initially exceeds the loan.

Mechanism

Income shock

Job loss or real-income compression can produce stress without any interest-rate reset.

Exposure

Rate reset

New rates are visible; the missing join is the stock by remaining fixation and borrower capacity.

Mechanism

Inflation

Inflation can erode real fixed debt while compressing cash flow and lifting rates.

Association

Housing scarcity

Scarcity supports collateral values but can force new buyers toward larger loans.

Scenario

Market repricing

Fast yield moves can expose duration and liquidity mismatch before defaults rise.

05 What would change the conclusion?

Watch realised transmission, not a synthetic alarm.

There is no financial-fragility composite or traffic-light total. These independent observations would change the current conditional conclusion.

  1. 01

    Persistent household defaults

    A broad rise under consistent default and cure definitions would overturn the aggregate-resilience read.

  2. 02

    Depleted liquid buffers

    Falling accessible assets among mortgagors and lower-income households would contradict comfort from national saving.

  3. 03

    Fiscal crowd-out

    Persistent interest/revenue pressure that displaces protective or productive spending would change the fiscal read.

  4. 04

    Bank losses

    Rising NPLs, Stage 2 exposures and losses alongside falling capital and liquidity would show safeguards being consumed.

  5. 05

    Restored affordability

    Lower new-borrower payments relative to income without looser risk layering would strengthen household resilience.

  6. 06

    Pension resilience

    Stable benefits and funding through transition and valuation shocks would strengthen the deferred-buffer case.

Method How to read this brief

Definitions travel with every number.

Observation

A measured stock, flow, ratio or supervisory aggregate under the cited scope.

Exposure

A channel through which a shock could transmit; not a realised loss.

Mechanism

A causal pathway to test, not proof that it dominates this period.

Scenario

A conditional possibility—not a forecast.

The updater owns only stable, no-key official series and validates identity, geography, unit, denominator and chronology before an atomic deterministic write. EBA annexes, ECB supervisory aggregates, CBS distributions and DNB pensions stay manually review-gated.

Sources Inspectable ledger

Series, tables, methods, vintages and revisions.

Each live evidence block resolves to one or more entries here. Derived ratios identify both inputs and the formula.