## gfsnea2025001

## Source details

**Canonical URL:** [gfsnea2025001](https://www.imf.org/-/media/files/publications/gfs-notes/2025/english/gfsnea2025001.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/gfs-notes/2025/english/gfsnea2025001.pdf.md)
- [Structured JSON version](/-/media/files/publications/gfs-notes/2025/english/gfsnea2025001.pdf.json)

---

### Introduction and context
- Global pension savings at $63.1 trillion at the end of 2023 (OECD data).
- For the OECD’s 38 member countries, pension savings translated into 98 percent of their combined gross domestic product.
- Pension funds have become an important segment of the (nonbank) financial sector, warranting adequate risk surveillance and robust supervision in many jurisdictions.
- Structural shifts accelerated by a prolonged period of low interest rates increased exposure to traditional risks and introduced emerging risks, reflected in growing intra-financial sector interconnectedness and exposure to long-term sovereign bonds.
- Recent transition to higher interest rates is expected to be positive for the pension sector, albeit its pace and abruptness has been associated with liquidity stress and contagion risks in some countries.

### Pension system architecture and market structure
- Three pillars:
  - Pillar 1: Public pensions provided by the government.
  - Pillar 2: Occupational pensions funded by employers and employees; decisions typically taken collectively.
  - Pillar 3: Private pensions (voluntary savings) provided mainly by insurers or pension funds.
- Occupational plans:
  - Mandatory or quasi-mandatory in half the OECD countries.
  - Cover more than 75 percent of the working-age population in 12 OECD countries.
  - Country examples: Finland and Switzerland (employers must operate an occupational plan); mandatory plans in Chile, Colombia, Costa Rica, Mexico; Denmark, the Netherlands, Sweden rely on collective bargaining (quasi-mandatory).
  - Colombia’s participation rate is 55 percent despite mandatory arrangement due to a large informal sector.
- Providers and concentration:
  - Autonomous pension funds manage about 60 percent of the global market.
  - Nearly two-thirds of OECD members have public pension reserve funds.
  - Public pension reserve funds in OECD countries managed $6.4 trillion in assets by end-2022, concentrated as: United States 43 percent, Japan 23 percent, Korea 11 percent, Canada 7 percent.
  - Among OECD pension provider assets, US pension funds held $38.97 trillion by 2023.
  - Other 2023 asset levels: United Kingdom $2.71 trillion; Australia $2.21 trillion; Canada $3.17 trillion; Netherlands $1.74 trillion.
- Cross-jurisdiction asset-to-GDP extremes:
  - Exceeding 150 percent of GDP: Canada, Denmark, Iceland, the Netherlands, Switzerland.
  - As low as 11–12 percent: France, Italy, Spain.
  - Single digits: Germany.

### Pension Fund Sectors: Growth rates and design shifts
- Shift from DB to DC plans:
  - DB predominant in Japan and the United Kingdom; DC increasing in the United States and Italy.
  - Netherlands law adopted in 2023 requiring conversion of DB plans into DC plans by 2028; hybrid plan providing 50 percent of the guaranteed benefits.
- DB prevalence and decline:
  - Over 40 percent of pension fund assets were within DB plans across advanced countries at of the end of 2023.
  - Share of DB plans dropped from more than 40 percent to less than 30 percent in several jurisdictions over the past two decades.
  - Israel: 91 percent to 43 percent.
  - Italy: 40 percent to 2 percent.
- Asset growth since 2013:
  - Advanced economies: nearly 6 percent annual growth.
  - Emerging economies: almost 9 percent annual growth.
- Private pension savings nearly doubled from 2012 to 2021.

### Traditional and emerging risks: empirical evidence
- Guarantees and funding:
  - DB plans guarantee benefit streams; funding ratios deteriorated with prolonged low interest rates and began recovering with higher rates.
- Interest rate, inflation, longevity risks:
  - Netherlands example: most pension funds did not index pensions between 2009 and 2022, leading to a loss in purchasing power of about 25 percent; indexation became possible again in 2022/2023.
  - OECD research: failure to account for further mortality improvements can expose pension funds and annuity providers to an expected shortfall of well over 10 percent of their liabilities.
- Search-for-yield findings (Konradt 2023, sample 2008–2018):
  - Average risky asset share increased by 4.3 percentage points, from 56.3 percent to 60.6 percent.
  - A one percentage point decrease in the risk-free rate was associated with a 0.66 percentage point increase in exposure to risky assets.
  - Private pension savings nearly doubled from 2012 to 2021; sensitivity to bond and equity prices produced significant hits when both markets fell in 2022.
  - Nominal investment rates of return: Netherlands −21.1 percent; United Kingdom −18.5 percent.
  - Dutch interest-rate swaps market values: end-2020 around €100 billion; mid-2022 −€80 billion; mid-2023 around −€80 billion, corresponding to more than 5 percent of total assets.
- Leverage, derivatives, and liquidity:
  - Notional value of derivatives transactions in a sample of pension funds rose to 80 percent of total assets in 2022 from 67 percent in 2016.
  - Financial liabilities through loans and repos typically amount to less than 1 percent of balance sheet.
  - Illiquid assets account for 22 percent of DB assets in EU countries, up from 17 percent in 2021; real estate investment funds constitute 7 percent.
  - For a sample of 12 EU countries, illiquid assets: 22 percent of total assets in DB plans and 6 percent in DC plans as of 2023:Q3.
  - Over the past two years (to 2023:Q3), in DB plans the share of illiquid assets increased by almost 30 percent, while in DC plans this share came down by 25 percent.

### Interconnectedness, concentration, and herding
- Interconnectedness channels:
  - Asset-based linkages, derivatives, repurchase agreements, common trustees/actuaries/asset managers.
  - Pension sector can be a dominant investor in domestic equity and government bond markets in some jurisdictions.
  - Pension risk transfer market growth (Ireland, the Netherlands, United Kingdom, United States) increases cross-border and cross-sector interconnectedness.
- Concentration and herding:
  - Common service providers and concentrated holdings can amplify market stress.
  - Appendix Table 1.1 research highlights country-specific herding and mechanisms (e.g., Raddatz and Schmukler 2011; Blake and others 2016; Broeders and others 2016; Bauer and others 2018; Papaioannou and others 2013; Han and others 2021).
  - Empirical pattern: offset of passive changes through active within-month rebalancing — equity allocation: more than 20 percent; bond allocation: 25 percent.

### Dutch pension funds: currency and liquidity mismatches (selected findings)
- Currency mismatch and hedging:
  - Dutch pension sector outstanding notional exposure in foreign exchange swaps almost €400 billion at the end of 2022, down from almost €460 billion two years earlier.
  - Hedge positions typically rolled over monthly or quarterly, creating maturity mismatch.
- Liquidity mismatch and systemic examples:
  - March 2020 “dash-for-cash”: variation margins of euro area insurers and pension funds moved with interest and exchange rates; MMF net flows correlated with margins of Dutch insurers and pension funds.
  - Canada, week of March 16, 2020: margin calls of the eight largest Canadian pension funds peaked at around CAD 30 billion; pension funds met demands via collateralized cash-raising strategies; Bank of Canada interventions helped repo market functioning.
- Illiquid asset exposures in EU DB plans and cross-country variation (see above).

### Data, valuation, and reporting gaps
- Pension fund data patchy and difficult to compare across jurisdictions despite OECD Global Pension Statistics expansion.
- Reporting on derivatives and leverage incomplete: Oh and Stańko (2023) — out of 22 jurisdictions where pension funds are allowed derivatives, only 12 require periodic reporting (three semi-annually or annually).
- Valuation heterogeneity:
  - Assets accounted at fair value, historic cost, or amortized cost.
  - DB liabilities valued with market-based discount rates or constant reference rates; small differences in discount rates materially affect valuations.
  - Iceland’s framework complicated by parallel accounting and actuarial regimes.

### Quantitative risk-analysis tools and stress-test findings
- Toolkit evolution: solvency stress tests for DB plans, projections for DC plans, liquidity analysis, interconnectedness and contagion analysis, reverse stress testing.
- Netherlands 2024 top-down stress test (ten largest funds, covering 70 percent of sector assets):
  - Higher interest rates lowered the value of pension fund liabilities by 27 percent on average, compensating for asset value declines.
  - Asset values declined by 23 percent (−€235 billion for the sample), overcompensated by a decline in liabilities of €245 billion.
  - Average funding ratio increased by 7 percentage points, to 122 percent (from 115 percent in the sample).
  - Funds with larger duration gaps experienced improvements greater than 10 percentage points.
- Liquidity stress test (Netherlands FSAP 2024):
  - The five largest pension funds faced a cash collateral call of €18.4 billion; cash collateral calls close to €20 billion could be met by tapping different sources even with limited repo market access.
- EIOPA 2019 solvency stress test (DB sector, 99 DB plans in 19 European countries):
  - Aggregate shortfall in the adverse scenario: €180 billion (national methodologies) and €216 billion (common methodology).
  - Under common methodology, shortfalls would have triggered aggregate benefit reductions of €173 billion and financial support from sponsors of €49 billion.
- Fire-sale systemic stress test (Netherlands FSAP 2024):
  - Losses from the bank solvency stress-test exercise led to fire-sale induced losses of about 5 percent of initial equity.
  - Early-round losses concentrated in banking, spreading to insurance and pension sectors in later rounds.
  - No single agent’s default triggered defaults of other agents in the exercise; banks suffer the highest total loss as a share of initial equity following defaults of securities-issuing agents.

### Change in the value of assets and liabilities — scenario results
- Representative member with 10 years prior to retirement: pension value in the median fund declines by 13 percent.
- Iceland FSAP projected future pension value declines between 9 and 15 percent for a member with 10 years prior to retirement; almost unaffected for a member with 30 years prior to retirement (maximum of −2 percent).
- Scenario comparison notes:
  - Sovereign spread shocks: EIOPA +81 bps and +140 bps for EU and US sovereigns; FSAP used +60 bps for both the Euro Area and the US.
  - Property shocks: EIOPA −38 percent for EU real estate; Netherlands FSAP used −15 percent for domestic commercial real estate and −12 percent for foreign commercial real estate.
  - Shock example: About 36 bps increase of EUR interest rates combined with a 4.4 percent EUR appreciation against USD.

### Policy recommendations and supervisory considerations (preserved wording and focus)
- Define supervisory objectives and represent pension supervisors in inter-institutional financial stability and macroprudential committees where pension funds are significant.
- Enhance reporting requirements proportional to major risks and complexities:
  - Granular reporting of material on- and off-balance sheet exposures, ideally detailed asset-by-asset and derivatives data where the sector is an important capital market player.
  - Include cash flow data enabling indicators on counterparty/sectoral concentration, liquidity, and leverage (including through derivatives and securities financing transactions).
  - Ensure data are available at adequately high frequency and authorities have legal powers to increase reporting frequency in times of crisis.
  - Apply the principle of proportionality when designing new reporting requirements.
- Establish close coordination between pension supervisors, central banks, and other financial authorities for data sharing and addressing financial stability concerns.
  - Pension fund supervisors should be able to share detailed data with central banks and other financial authorities to conduct system-wide analysis jointly.
- Enhance corporate governance, internal controls, and risk management requirements:
  - Improve the IOPS Principles by providing more detail on board independence and objectivity, regular assessments of board effectiveness, remuneration, establishment of risk management or compliance functions, definition of risk appetite, and use of stress testing and scenario analyses.
  - Align these requirements with other financial sector international standards (Basel Core Principles, Insurance Core Principles, IOSCO Principles of Securities Regulation).
- Develop a risk-based capital regime for supervisors with large DB plans:
  - Risk-based capital regimes tailored to DB plans would better reflect complex risks than simple metrics such as the funding ratio.
  - Consider analogous regimes in banking and insurance to mitigate regulatory arbitrage across financial sectors.
- Monitor and mitigate liquidity risk:
  - Address liquidity mismatches from higher illiquid asset investments and increased liquidity needs from derivatives, short-term securities financing transactions, margin calls, and portfolio redemption/switching flexibilities.
  - Improve liquidity risk management practices and supervisory monitoring; act when high liquidity risks are identified.
- Enhance toolkits for quantitative risk analysis:
  - Include solvency stress tests for DB plans, projections of future pension values for DC plans, liquidity analysis, interconnectedness and contagion analysis with the wider financial sector.
  - Use reverse stress testing based on individual risk factors (e.g., equity prices, interest rates).
  - Coordinate these exercises closely with central banks and other financial sector authorities to analyze interconnectedness and contagion.

### Appendix — research on herding behavior (selected empirical findings)
- Raddatz and Schmukler (2011) — Chile: herding observed in asset classes with scarce market information and when risk increases.
- Blake and others (2016) — UK: DB funds with similar size and sponsor type tend to move in and out of different asset classes; systematic switch from equities to bonds as liabilities mature noted.
- Broeders and others (2016) — The Netherlands: evidence for semi-strong and strong herding for large pension funds; weak herding showed countercyclical offsets exceeding 20 percent for equities and 25 percent for bonds within the month.
- Bauer and others (2018) — The Netherlands: strategic allocation changes align when funds share actuaries or dominant asset managers, especially into alternative assets.
- Papaioannou and others (2013) — Multiple jurisdictions: DB decisions driven by regulatory constraints; DC decisions influenced by similar benchmarks, contributing to herding.
- Han and others (2021) — Multiple jurisdictions: procyclical behavior in Poland and Italy; countercyclical in Chile (less than statistically significant).
- Empirical within-month offset metrics:
  - Equity allocation: more than 20 percent.
  - Bond allocation: 25 percent.

*International Monetary Fund — Global Financial Stability Notes: Pension Funds and Financial Stability (extracted content).*

### Introduction ...........................................................................................................

### Introduction

### Overview and context
- Global pension savings at $63.1 trillion at the end of 2023 (OECD data).
- For the OECD’s 38 member countries, pension savings translated into 98 percent of their combined gross domestic product.
- Pension funds have become an important segment of the (nonbank) financial sector, warranting adequate risk surveillance and robust supervision in many jurisdictions.
- The sector has undergone structural shifts accelerated by a prolonged period of low interest rates, increasing exposure to traditional risks and introducing emerging risks, reflected in growing intra-financial sector interconnectedness and exposure to long-term sovereign bonds.
- The recent transition to higher interest rates should be positive for the pension sector, albeit its pace and abruptness has been associated with liquidity stress and contagion risks in some countries.

### Pension system architecture
- Pension systems comprise three pillars:
  - Pillar 1: Public pensions provided by the government to support basic income needs.
  - Pillar 2: Occupational pensions funded by contributions from both employers and employees; decisions typically taken collectively.
  - Pillar 3: Private pensions (voluntary savings) provided mainly by financial institutions such as insurers or pension funds.
- Relative importance of pillars varies substantially across countries (examples in source: Netherlands, Iceland).

### Structure and coverage of occupational pensions
- Occupational pension plans are a major component of pension assets.
- Occupational plans are mandatory or quasi-mandatory in half the OECD countries and cover more than 75 percent of the working-age population in 12 OECD countries.
- Country-specific arrangements cited:
  - Finland and Switzerland: employers must operate an occupational pension plan and contribution rates are set by law.
  - Mandatory plans prevalent in Chile, Colombia, Costa Rica, and Mexico.
  - Denmark, the Netherlands, and Sweden: obligations are often determined through collective bargaining (quasi-mandatory) with participation rates similar to mandatory systems.
- Colombia’s participation rate is a modest 55 percent despite a mandatory arrangement because of its large informal sector.
- Exemptions can exist for self-employed and seasonal workers.

### Providers and market structure
- Autonomous pension funds manage about 60 percent of the global market.
- Nearly two-thirds of OECD members have set up public pension reserve funds.
- Public pension reserve funds in OECD countries managed $6.4 trillion in assets by the end of 2022, largely concentrated as follows:
  - United States: 43 percent of global public pension reserve fund assets
  - Japan: 23 percent
  - Korea: 11 percent
  - Canada: 7 percent
- Other types of pension savings include book reserves, pension insurance contracts, and funds managed by investment companies and banks.
- Notable cross-country differences: in Denmark (largest pension savings market by assets-to-GDP) and Sweden, most pension savings are through insurance companies, not autonomous funds.
- Autonomous pension funds can be single-employer or multi-employer; single-employer funds tend to experience greater fluctuations in membership, contributions, and non-retirement payouts.

### Size and concentration of pension assets
- Among OECD pension provider assets, more than half of global assets are with US pension funds.
- By 2023, US pension funds had accumulated $38.97 trillion.
- Other notable markets in absolute terms (2023):
  - United Kingdom: $2.71 trillion
  - Australia: $2.21 trillion
  - Canada: $3.17 trillion
  - Netherlands: $1.74 trillion
- Since 2013, pension assets growth:
  - Advanced economies: nearly 6 percent annual growth
  - Emerging economies: almost 9 percent annual growth
- Cross-jurisdiction asset-to-GDP extremes cited:
  - Exceeding 150 percent of GDP: Canada, Denmark, Iceland, the Netherlands, Switzerland
  - Sizable: United States and Australia
  - As low as 11–12 percent: France, Italy, Spain
  - Single digits: Germany

### Contribution to financial stability and emerging vulnerabilities
- Pension funds have been seen historically as contributors to financial stability due to:
  - Long-term investment horizon
  - Well-diversified liabilities
  - Predictable cashflows supported by diversified member contributions and limited redemption options
- Evidence exists that pension funds have acted countercyclically during past financial stress episodes.
- Potential additional returns from countercyclical investment and illiquidity premiums may be available but can be attenuated without appropriate liquidity risk management.
- Structural shift and increased exposures (traditional and emerging risks) raise supervisory and systemic risk surveillance priorities, given increased intra-financial sector interconnectedness and exposure to long-term sovereign bonds.

*Source: Introduction chapter of the IMF Global Financial Stability Note "Pension Funds and Financial Stability" (extracted content).*

### 2. Pension Fund Sectors: Growth Rates

### 2. Pension Fund Sectors: Growth Rates

### Shift in pension system design and implications
- Pension systems shifting from defined benefit (DB) to defined contribution (DC) plans to address interest rate, inflation, and longevity risks.
- DB plans (guarantee specific benefits) predominant in countries such as Japan and the United Kingdom; DC plans (transfer investment risk to the individual) increasingly popular in the United States and Italy.
- Many systems transitioning toward a greater share of DC plans because of cessation of existing DB plans and new contributions made predominantly toward newly created DC plans; transitions often supported by hybrid arrangements where sponsor and members share investment risks.
- Even after shifts, pension funds often continue to manage DB plans corresponding to existing members’ accumulated benefits.
- The Netherlands is an exception wherein, during the ongoing transition, even accumulated benefits have been shifted into the new DC plan.
- Netherlands: law adopted in 2023 requiring conversion of DB plans into DC plans by 2028; the hybrid plan is providing 50 percent of the guaranteed benefits.
- Shift increases importance of strengthening conduct oversight, members’ financial literacy, and disclosures.

### Traditional risks (persisting and amplified)
- Guarantees
  - DB plans guarantee a stream of benefits; actuarially determined by salary and tenure.
  - Globally, over 40 percent of pension fund assets were within DB plans across advanced countries at of the end of 2023.
  - The share of DB plans has dropped from more than 40 percent to less than 30 percent in several jurisdictions over the past two decades.
  - Examples of steep declines: Israel from 91 percent to 43 percent; Italy from 40 percent to 2 percent.
- Interest rate risk and duration mismatches
  - Low interest rates increase present/economic/accounting value of long-duration pension liabilities.
  - Asset–liability mismatches are common; liabilities typically have longer duration than assets.
  - Funding ratios (assets divided by liabilities) deteriorated with prolonged low interest rates and began recovering only recently because of higher interest rates.
- Inflation
  - DB benefits often linked to wages or consumer prices; higher inflation poses risk if investments do not yield necessary returns.
  - Limited availability of long-duration inflation-linked bonds exacerbates this challenge.
  - Netherlands example: most pension funds did not index pensions between 2009 and 2022, leading to a loss in purchasing power of about 25 percent; indexation became possible again in 2022/2023 as funding ratios improved.
- Longevity
  - Use of outdated or improperly adapted mortality tables can create material risk.
  - OECD research: failure to account for further improvements in mortality can expose pension funds and annuity providers to an expected shortfall of well over 10 percent of their liabilities.
- Operational and conduct risks
  - Cyber-attacks can (1) target personal information of members; (2) cause system outages delaying benefit payments; (3) retrieve insider information about upcoming investment deals.
  - DC funds face market conduct risk related to inappropriate actions and conflicts among managers, advisors, and administrators.
  - Example: Netherlands’ transition from DB to DC requires extraordinary efforts on data quality for each individual pension contract.

### Search-for-yield and empirical evidence
- In a low-interest-rate environment, pension funds have been incentivized to search-for-yield by increasing leverage and exposure to illiquid assets.
- Lower interest rates increase present value of DB liabilities; some large funds increased allocations to illiquid investments while actively using derivatives and other leverage.
- Konradt (2023) empirical findings (sample: 105 pension funds from 14 countries, 2008–2018):
  - Average risky asset share increased by 4.3 percentage points, from 56.3 percent to 60.6 percent.
  - A one percentage point decrease in the risk-free rate was associated with a 0.66 percentage point increase in exposure to risky assets (after accounting for valuation effects).
  - European funds mainly increased equity exposure; funds outside Europe favored alternative assets.
  - Search-for-yield more pronounced for funds with greater capacity to take risks (higher funding ratios or lower incumbent holdings of risky assets).
- Private pension savings nearly doubled from 2012 to 2021; sensitivity to bond and equity prices led to significant hits when both markets fell in 2022.
- Examples of investment returns and losses:
  - Netherlands: nominal investment rate of return −21.1 percent.
  - United Kingdom: nominal investment rate of return −18.5 percent.
- Dutch interest-rate swaps (used to reduce duration gap):
  - End of 2020: market value around €100 billion.
  - Mid-2022: market value −€80 billion.
  - As of mid-2023: market value amounted to around −€80 billion, corresponding to more than 5 percent of total assets.

### Emerging and amplifying risks
- Liquidity imbalances and leverage
  - Search-for-yield into illiquid assets with leverage changed liquidity profiles; margin calls since 2020 in Canada, the Netherlands, and the United Kingdom triggered contagion to money-market funds, repo borrowing, and equity markets.
  - Financial liabilities of pension funds through loans and repos typically amount to less than 1 percent of their balance sheet.
  - Leverage can amplify stress episodes via asset fire sales and abrupt margin/collateral calls; 2022 gilt crisis example involved off-balance-sheet leverage (liability-driven investment vehicles).
- Interconnectedness and concentration
  - Interconnectedness via asset-based linkages, derivatives, and repurchase agreements; use of common trustees, actuaries, and asset managers increases interconnectedness.
  - Notional value of derivatives transactions in a sample of pension funds rose to 80 percent of total assets in 2022 from 67 percent in 2016.
  - Pension sector can be a dominant investor in domestic equity and government bond markets in some jurisdictions, increasing counterparty and liquidation risks and potential destabilizing effects from rapid portfolio rebalancing.
  - Pension risk transfer market is growing (Ireland, the Netherlands, United Kingdom, and the United States), increasing cross-border and cross-sector interconnectedness when obligations are reinsured offshore.
- Concentration and herding
  - Common service providers and concentrated holdings can amplify market stress when a small number of institutional investors hold concentrated positions (e.g., long-dated gilt markets).
  - Herding behavior observed in rebalancing strategies and reactions to shocks can worsen systemic vulnerabilities; cyclicality of investment behavior can contribute to systemic risk build-up.
- Operational amplification
  - Delegated management, pooled investments, or cross-border interconnectedness can create operational bottlenecks that delay responses to exogenous shocks and exaggerate systemic implications.

### Key figures and metrics (preserved exactly as reported)
- Over 40 percent of pension fund assets were within DB plans across advanced countries at of the end of 2023.
- Share of DB plans dropped from more than 40 percent to less than 30 percent in several jurisdictions over the past two decades.
- Israel: 91 percent to 43 percent.
- Italy: 40 percent to 2 percent.
- Netherlands law adopted in 2023 requiring conversion by 2028; hybrid plan providing 50 percent of the guaranteed benefits.
- Loss in purchasing power in Netherlands between 2009 and 2022: about 25 percent.
- Konradt (2008–2018 sample): risky asset share increased by 4.3 percentage points (56.3 to 60.6 percent).
- Reach-for-yield elasticity: a one percentage point decrease in the risk-free rate associated with a 0.66 percentage point increase in exposure to risky assets.
- Private pension savings nearly doubled from 2012 to 2021.
- Nominal investment rates of return: Netherlands −21.1 percent; United Kingdom −18.5 percent.
- Dutch swaps market values: end-2020 around €100 billion; mid-2022 −€80 billion; mid-2023 around −€80 billion, corresponding to more than 5 percent of total assets.
- Mortality assumption shortfall: well over 10 percent of liabilities (OECD 2014 finding).
- Financial liabilities through loans and repos typically amount to less than 1 percent of balance sheet.
- Notional derivatives transactions rose to 80 percent of total assets in 2022 from 67 percent in 2016.

### Policy and oversight implications emphasized in the text
- Importance of robust liquidity management and systemic risk oversight to address margin/collateral-call contagion risks.
- Strengthen conduct oversight, disclosures, and members’ financial literacy as risks shift toward beneficiaries under DC and hybrid arrangements.
- Need for careful assessment and management of leverage, liquidity imbalances, currency mismatches, concentrated investments, and interconnectedness.
- Monitor and mitigate operational and cyber risks, especially during large structural transitions (for example, DB-to-DC transitions requiring high-quality individual data).
- Evaluate pension fund interconnectedness and concentration exposures in domestic markets and cross-border pension risk transfer activities.

*International Monetary Fund — Global Financial Stability Notes: Pension Funds and Financial Stability (section 2).*

### 3. Dutch Pension Funds: Net Asset Transactions

### 3. Dutch Pension Funds: Net Asset Transactions

### Currency Mismatch
- Pension fund liabilities are typically denominated entirely in domestic currency, and DB funds have no significant incentive to invest in foreign currency denominated assets.
- In jurisdictions with small onshore financial markets, pension funds may not be able to find an adequate amount of domestic long-term assets with stable values to invest into.
- Prolonged periods of low interest rates may generate substantial search-for-yield incentives for DB funds with large, guaranteed portfolios of liabilities, which can be addressed in part by increasing overseas investments.
- Feasibility and cost of hedging currency risks is critical; in many emerging economies there is no liquid market for swaps and other hedging instruments.
- Even if currency risk is hedged, rollover risk can be significant when pension funds use short-term foreign exchange swaps:
  - The Dutch pension sector had an outstanding notional exposure in foreign exchange swaps of almost €400 billion at the end of 2022, down from almost €460 billion two years earlier.
  - Hedge positions are typically rolled over on a monthly or quarterly basis, creating maturity mismatch.
- Iceland example:
  - Foreign-denominated investments reached 38 percent at end-2021, up from 26 percent at end-2017, with most invested in US dollars.
  - A stress test in the 2023 FSAP assumed an ISK depreciation of −30.6 percent in 2023; the assumed depreciation would largely offset the effect from other asset-side shocks.

### Liquidity Mismatch and Systemic Risks
- Younger pension systems: contributions tend to exceed benefit payouts, allowing buy-and-hold behavior and acting as stabilizers in capital markets.
- Mature pension systems: net cash flows can turn negative as benefit payouts exceed contributions, creating potential liquidity pressures that can persist for years or decades.
- Specific vulnerabilities:
  - Closed pension funds, single-employer or industry-wide funds in shrinking sectors, and funds in aging countries with high old-age dependency ratios face heightened liquidity management needs.
  - Political or social pressures can prompt sudden changes in redemption rules, leading to liquidity stress (examples noted: Chile, Iceland, Namibia, Malawi, Botswana).
  - Namibia: since 2018 contributions fell below benefits and expenses for a sample of pension funds; assets with derivatives increased share of illiquid investments (figure-based evidence).
- Sources of liquidity shocks:
  - Extraordinary withdrawals allowed for crisis relief (temporary or permanent), mortgage repayments, medical bills, dismissals, or other household needs.
  - Examples:
    - Malawi: legislative amendment allowed pensioners within five years before retirement to access up to 50 percent of pension entitlements; pension funds were permitted to use repurchase agreements to meet surge in claims.
    - Botswana: expanded scope for members to withdraw funds (dismissal, mortgage repayment, medical bills) and larger lump sums, transferring liquidity from pension funds to households.
- Illiquid asset exposures:
  - Illiquid assets account for 22 percent of DB assets in EU countries, up from 17 percent in 2021.
  - Real estate investment funds constitute 7 percent and are the most relevant illiquid class.
  - The share of illiquid assets varies between 2 and 27 percent among different EU member states.
  - For a sample of 12 EU countries, the share of illiquid assets accounts for 22 percent of total assets in DB plans, but only 6 percent in DC plans as of 2023:Q3.
  - Over the past two years (to 2023:Q3), in DB plans the share of illiquid assets has increased by almost 30 percent, while in DC plans this share came down by 25 percent.
  - Some directly held real estate investments exceed 10 percent of total assets in a few countries according to EIOPA data.
- March 2020 “dash-for-cash” episode lessons:
  - Variation margins of euro area insurers and pension funds moved with interest and exchange rates.
  - Net flows of Irish and Luxembourgish MMFs were highly correlated with variation margins of Dutch insurers and pension funds.
  - Acute liquidity needs can arise from margin calls, dry-up of market segments, or political/regulatory drivers allowing extraordinary withdrawals.
- Canada, March 2020:
  - In the week of March 16, 2020, margin calls of the eight largest Canadian pension funds peaked at around CAD 30 billion, coinciding with the peak of the volatility index (VIX).
  - Liquidity in the commercial paper market dried up, removing a source of funding for Canadian pension funds.
  - Pension funds met liquidity demands via raising cash using equities as collateral, increasing term repos, withdrawing purchases, outright sales of bankers’ acceptances, increasing use of long-term repos and total return swaps.
  - Bank of Canada interventions (including the Contingent Term Repo Facility) helped support repo market functioning.

### Data and Tools for Systemic Risk Analysis
- Pension fund data is patchy and difficult to compare across jurisdictions.
- OECD Global Pension Statistics expanded to include more than 50 non-OECD countries, following a joint initiative with IOPS and the World Bank, but data lacks full comparability and critical data for financial stability analysis is often not publicly available or even available to supervisory authorities.
- Reporting gaps on derivatives and leverage:
  - Oh and Stańko (2023) report that out of 22 jurisdictions where pension funds are allowed to hold derivatives, only 12 require periodic reporting on these positions (three of which report semi-annually or annually).
- Valuation heterogeneity:
  - Assets can be accounted for at fair value, historic cost, or amortized cost.
  - DB plan liabilities can be valued with market-based discount rates or with a constant reference rate; some countries use fixed discount rates while others use market rates.
  - Due to long durations of pension liabilities, small differences in discount rates have significant impacts on liability valuations and derived statistics.
  - Iceland: framework complicated by parallel regimes of accounting valuation and actuarial valuation.

### Tools for Quantitative Risk Analysis and Solvency Stress Tests
- Toolkit evolution: recent FSAPs have included solvency stress tests for DB plans, projections for DC plans, liquidity analysis, interconnectedness and contagion analysis, and reverse stress testing based on individual risk factors.
- Netherlands 2024 FSAP top-down stress test (ten largest pension funds, covering 70 percent of sector assets):
  - Higher interest rates lowered the value of pension fund liabilities by 27 percent on average, compensating for asset value declines.
  - Asset values declined by 23 percent (−€235 billion for the sample), overcompensated by a decline in liabilities of €245 billion.
  - Average funding ratio increased by 7 percentage points, to 122 percent.
  - For funds with larger duration gaps between assets and liabilities, improvements were greater than 10 percentage points.
  - Average funding ratio increased from 115 to 122 percent in the sample.
- EIOPA 2019 solvency stress test (DB sector, 99 DB plans in 19 European countries):
  - Aggregate shortfall in the adverse scenario of €180 billion according to national methodologies and €216 billion according to the stress test’s common methodology.
  - Under the common methodology assumptions, shortfalls would have triggered aggregate benefit reductions of €173 billion and financial support from pension plan sponsors of €49 billion.
- Other observations:
  - Some of the world’s largest pension funds, with assets in excess of $7 trillion, have significantly increased allocations to illiquid investments while actively using derivatives and other forms of leverage.
  - Supervisory practices: few pension supervisors have identified liquidity issues related to margin calls per an IOPS survey (37 members responded); 34 authorities had not seen liquidity issues associated with margin calls by April/May 2023; three reported liquidity issues but no liquidity shortages.
  - Jurisdictions such as Australia and Colombia use supervisory liquidity stress tests for pensions; Colombia uses reverse stress testing.

*Source: IMF staff calculations and analysis drawn from De Nederlandsche Bank, Central Bank of Iceland, NAMFISA, EIOPA, IO PS survey results, FSAPs, and related IMF staff work as presented in the provided chapter.*

### 1. Change in the Value of Assets and Liabilities

### 1. Change in the Value of Assets and Liabilities

### Quantitative findings and scenario results
- Effect of an interest rate shock on liability values compensates almost all asset-side shock effects, especially those on stocks and fixed-income assets in the Netherlands.
- For a representative member with 10 years prior to retirement, the pension value in the median fund declines by 13 percent.
- In Iceland, future pension value results vary between pension funds from −9 to −15 percent for a member with 10 years prior to retirement.
- The Iceland FSAP projected future pension values under an adverse scenario:
  - Declines under the scenario by between 9 and 15 percent for a member with 10 years prior to retirement.
  - Almost unaffected for a member with 30 years prior to retirement (a maximum of −2 percent).
- Scenario attribute comparisons between EIOPA stress test and FSAP for the Netherlands:
  - Sovereign spread shocks: +81 bps and +140 bps for EU and US sovereigns in EIOPA stress test; FSAP used +60 bps for both the Euro Area and the US.
  - Property valuation shocks: –38 percent for EU real estate in EIOPA stress test; NLD FSAP used –15 percent for domestic commercial real estate and –12 percent for foreign commercial real estate.
- A shock assumption noted: About 36 bps increase of EUR interest rates, combined with a 4.4 percent EUR appreciation against USD.

### DC plans — projection methodology and impacts
- DC plan risk analysis is from the member perspective via projection of future pension values.
- The 2023 Iceland FSAP:
  - Projected future pension value for representative members with 10 or 30 years to retirement, both in baseline and adverse FSAP scenarios.
  - Accrued pension benefit is shocked with market risk stresses in each of the first three years of the projection horizon; afterward annual investment returns align with the baseline.
  - In the adverse scenario, pension fund assets decline considerably in the first two years, materially reducing future pension values.
- Members face conversion risk from accumulated pension savings at retirement age to the income stream after retirement.

### Liquidity stress test results (Netherlands FSAP 2024)
- The five largest pension funds faced a cash collateral call of €18.4 billion in the scenario; repo markets remained an important source of liquidity for them.
- Cash collateral calls close to €20 billion could be met by tapping different sources, even when assuming limited repo market access.
- Smaller pension funds exempted from clearing obligations use bilateral swaps which allow for settlement-in-kind that lowered liquidity risks.

### Interconnectedness and contagion analysis (Netherlands FSAP 2024)
- Fire-sale systemic stress test design:
  - Starts with one or more agents experiencing a balance sheet loss; agents keep leverage ratios constant, translating losses into proportionate sales of securities.
  - Two types of initial losses: (1) losses of the six significant banks in the adverse bank solvency scenario (considered jointly); (2) losses generated when each securities-issuing agent defaults in turn.
- Results:
  - Losses from the bank solvency stress-test exercise led to fire-sale induced losses of about 5 percent of initial equity.
  - In early rounds losses are concentrated in the banking sector, spreading to insurance and pension sectors in later rounds.
  - Individual-agent defaults can cause substantial losses amplified through fire-sale channels, but no single agent’s default triggered defaults of other agents.
  - Banks suffer the highest total loss as a share of initial equity following the defaults of securities-issuing agents.
  - Contagion can be caused by the default of relatively small agents.

### Regulatory and funding statistics cited
- Funding-rule examples:
  - Some DB plans subject to funding requirements where funding ratio should remain above 100 percent.
  - In Iceland pension funds must permanently maintain a funding ratio between 90 and 110 percent and not outside 95 and 105 percent for five consecutive years.
  - In the Netherlands, a 12-month average of the “market funding ratio” is used for regulatory purposes.
- International supervisory membership and standards:
  - Pension supervisors from 79 jurisdictions are IOPS members.
  - IOPS Principles of Private Pension Supervision were issued in 2006, reviewed in 2010; a recently revised version (2024) is under public consultation.

### Selected historical and comparative references
- A 2022 exercise covered 187 DB and DC plans in 18 European countries (market coverage of 65 percent) testing a climate transition risk scenario (EIOPA 2022).
- The 2019 Canada FSAP reported that large Canadian pension funds conducted liquidity stress tests based on a modified liquidity coverage ratio framework; eight largest Canadian public pension funds apply the liquidity coverage ratio framework with adjustments.
- The Finland FSAP discussed liquidity risks and recommended developing a liquidity regulation for pension funds.
- A staff analytical note by the Bank of Canada documented liquidity frameworks applied by large Canadian public pension funds.

### Source data and calculation notes
- Source: IMF staff calculations; IMF staff calculations based on the data from the Central Bank of Iceland.
- In the Iceland FSAP’s analysis, most of the valuation impact stemmed from lower stock prices. In the first year, the depreciation of the Krona assumed by the scenario counterbalanced the decline in domestic equity value by the increase in the value of foreign exchange–denominated investments.

### Fire-sale and liquidity visualization references (figures described)
- Figure 10 panels referenced:
  - Panel 1: Collateral calls (EUR billions) and ability to meet cash collateral calls.
  - Panel 2: Sources of liquidity (EUR billions).
  - Panel 3: Sectoral losses from bank solvency shock (cumulative by round).
  - Panel 4: Losses from defaulting agents.

---

### Policy recommendations and supervisory considerations
- Define supervisory objectives and represent pension supervisors in inter-institutional financial stability and macroprudential committees where pension funds are significant.
- Enhance reporting requirements proportional to major risks and complexities:
  - Granular reporting of material on- and off-balance sheet exposures, ideally detailed asset-by-asset and derivatives data where the sector is an important capital market player.
  - Include cash flow data enabling indicators on counterparty/sectoral concentration, liquidity, and leverage (including through derivatives and securities financing transactions).
  - Ensure data are available at adequately high frequency and authorities have legal powers to increase reporting frequency in times of crisis.
  - Apply the principle of proportionality when designing new reporting requirements.
- Establish close coordination between pension supervisors, central banks, and other financial authorities for data sharing and addressing financial stability concerns.
  - Pension fund supervisors should be able to share detailed data with central banks and other financial authorities to conduct system-wide analysis jointly.
- Enhance corporate governance, internal controls, and risk management requirements:
  - Improve the IOPS Principles by providing more detail on board independence and objectivity, regular assessments of board effectiveness, remuneration, establishment of risk management or compliance functions, definition of risk appetite, and use of stress testing and scenario analyses.
  - Align these requirements with other financial sector international standards (Basel Core Principles, Insurance Core Principles, IOSCO Principles of Securities Regulation).
- Develop a risk-based capital regime for supervisors with large DB plans:
  - Risk-based capital regimes tailored to DB plans would better reflect complex risks than simple metrics such as the funding ratio.
  - Consider analogous regimes in banking and insurance to mitigate regulatory arbitrage across financial sectors.
- Monitor and mitigate liquidity risk:
  - Address liquidity mismatches from higher illiquid asset investments and increased liquidity needs from derivatives, short-term securities financing transactions, margin calls, and portfolio redemption/switching flexibilities.
  - Improve liquidity risk management practices and supervisory monitoring; act when high liquidity risks are identified.
- Enhance toolkits for quantitative risk analysis:
  - Include solvency stress tests for DB plans, projections of future pension values for DC plans, liquidity analysis, interconnectedness and contagion analysis with the wider financial sector.
  - Use reverse stress testing based on individual risk factors (e.g., equity prices, interest rates).
  - Coordinate these exercises closely with central banks and other financial sector authorities to analyze interconnectedness and contagion.

*Source: IMF staff calculations and IMF staff analysis as presented in the referenced chapter.*

### Appendix Table 1.1  Research on Herding Behavior of Pension Sector

### Appendix Table 1.1  Research on Herding Behavior of Pension Sector

### Key findings by study and jurisdiction
- Raddatz and Schmukler (2011) — Chile
  - Herding investment behaviors were observed among Chilean pension funds in asset classes where market information is scarce and when risk increases and often among funds that narrowly compete.
- Blake and others (2016) — UK
  - DB funds with similar size and sponsor type tend to move in and out of different asset classes.
  - A systematic switch was observed from equities to bonds as liabilities mature and the mechanical rebalancing of portfolios occurs in the short term.
- Broeders and others (2016) — The Netherlands
  - Distinguished three types of herding: weak, semi-strong, and strong (definitions below).
  - For a sample of large pension funds in the Netherlands, evidence was found for semi-strong and strong herding behaviors.
  - In the weak herding category, pension funds acted countercyclically, where the pension funds offset more than 20 percent of the passive changes in their equity allocation and 25 percent for their bond allocation through active changes within the month.
- Bauer and others (2018) — The Netherlands
  - Pension funds tend to change their strategic asset allocation in the same direction if they are connected through actuaries or dominant asset managers, particularly prevalent regarding allocations into alternative asset classes.
- Papaioannou and others (2013) — Multiple jurisdictions
  - Pension funds’ investment decisions were largely driven by regulatory constraints for DB plans.
  - For DC plans, investment decisions were influenced by the use of similar benchmarks, which could contribute to herding behavior.
- Han and others (2021) — Multiple jurisdictions
  - Pension funds in different countries behave procyclically (in Poland and Italy) or countercyclically (in Chile, although it is less than statistically significant).
  - Procyclicality could potentially be driven by the regulatory or institutional framework.

### Types and mechanisms of herding (as characterized)
- Weak herding
  - Described as similar rebalancing strategies.
  - Example finding: countercyclical action offsetting more than 20 percent of passive equity allocation changes and 25 percent of passive bond allocation changes through active within-month adjustments.
- Semi-strong herding
  - Similar reaction to external shocks or regulatory changes.
- Strong herding
  - Intentional replication of other pension funds’ strategic asset allocation.

### Drivers and channels identified
- Information scarcity in certain asset classes and increases in risk can trigger herding (Raddatz and Schmukler (2011)).
- Similar size and sponsor type among DB funds can produce synchronized moves across asset classes (Blake and others (2016)).
- Connections through shared actuaries or dominant asset managers can align strategic allocation changes, especially into alternative asset classes (Bauer and others (2018)).
- Regulatory constraints influence DB plans’ investment decisions and can drive common behavior across funds (Papaioannou and others (2013)).
- Use of similar benchmarks in DC plans can contribute to herding (Papaioannou and others (2013)).
- The regulatory or institutional framework can influence whether funds behave procyclically or countercyclically (Han and others (2021)).

### Empirical patterns and notable quantitative findings
- Offset of passive changes through active rebalancing within the month:
  - Equity allocation: more than 20 percent.
  - Bond allocation: 25 percent.
- Country-specific procyclical behavior documented: Poland and Italy.
- Country-specific countercyclical behavior documented: Chile (less than statistically significant in Han and others (2021)).

*Source: Appendix Table 1.1, Research on Herding Behavior of Pension Sector — GLOBAL FINANCIAL STABILITY NOTES: Pension Funds and Financial Stability (INTERNATIONAL MONETARY FUND).*

---


_Source: https://www.imf.org/-/media/files/publications/gfs-notes/2025/english/gfsnea2025001.pdf_