## 1chlea2026002

## Source details

**Canonical URL:** [1chlea2026002](https://www.imf.org/-/media/files/publications/cr/2026/english/1chlea2026002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2026/english/1chlea2026002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2026/english/1chlea2026002.pdf.json)

---

### A. Introduction and overview of the 2025 pension reform
- Reform approved by Congress on January 29, 2025.
- Main reform features:
  - Raises contribution rate by 7 percentage points via gradual employer contributions over a transition period of nine years (or eleven years under an alternative schedule).
  - Employers required to contribute 8.5 percent to individual accounts and the new Social Insurance administered by FAPP (post-reform contribution schedule detailed below).
  - Expands the solidarity pillar: Minimum Guaranteed Pension (PGU) increased from CLP 224,000 to CLP 250,000 depending on beneficiaries’ age.
  - Creates a new Social Insurance (FAPP) financing gender pension gap compensation, a pension bonus based on years of contribution, and contributions to individual accounts with protected returns.
  - Transition from multi-fund system to generational-fund (target-date fund, TDF) system; mandates every two years a reallocation of 10 percent of affiliates to the fund administrator with the lowest fee.
  - Measures to increase contribution densities including coverage during unemployment and a simplified monthly contribution scheme for self-employed workers.
- Estimated fiscal implications (authorities’ estimates):
  - Additional fiscal costs of around 1 percent of GDP over the medium term.
  - Lower tax revenues: around 0.34 percent of GDP by 2035.
  - Higher pension-related expenditures for public sector employees: 0.22 percent of GDP.
  - Fiscal transfers to the FAPP projected to peak at about 0.23 percent of GDP.
  - Increased spending on the PGU projected to add costs of 0.3 percent of GDP under the baseline (CPI-indexation) and accounts for about one third of the total fiscal burden of the reform.
  - Total additional costs projected to peak at around 1.1 percent of GDP by 2035 and gradually decline, stabilizing at approximately 0.9 percent of GDP by 2050.
- FAPP financing notes:
  - FAPP finances disability and survivors’ insurance (SIS), contributed by employers, amounting to 1.5 percent on average of worker’s taxable income.
  - The additional employer contribution is allocated: (i) 4.5 percent to individual accounts; (ii) 1.5 percent to FAPP to finance the Protected Return Contribution (CRP) and the "Benefit for years contributed"; and (iii) 1 percent to FAPP to finance compensation for the gender pension gap.

### B. Distributional findings and challenges in PGU design
- Microdata and sample:
  - Uses Encuesta de Protección Social (EPS) microdata (32,537 individuals; monthly earnings history since 1981; monthly pension savings since 2002).
  - Simulation subsample: individuals aged 40 and above; balanced panel through December 2024 (sample size for key statistics: 18,004 observations).
- Key empirical results on self-funded pensions (PAFE) and PGU interaction:
  - In 2024, 96 percent of retirees aged 65 and above had a self-financed monthly pension below the lower threshold at which the PGU starts phasing out, implying that, if eligible, this share will receive the full PGU.
  - Simulations project more than 92 percent of retirees will be below the lower PGU threshold by 2050.
  - Even if PGU thresholds are kept frozen in nominal terms, by 2050 an extreme scenario projects 78 percent of women and 70 percent of men would remain below the lower PGU threshold.
  - Mass of the PAFE distribution concentrated at very low levels; large disparities in lifetime earnings and contribution densities; substantial share contributing zero or close to zero to individual accounts.
  - Pronounced gender gaps: women have systematically lower earnings and lower contribution densities, amplifying pension outcome gaps.
- PGU eligibility and tapering design (as of February 2026):
  - Full PGU for retirees with self-financed pensions below the lower threshold: CLP 789,139 per month.
  - Tapered benefit between lower threshold and upper threshold.
  - Upper threshold at which PGU is fully eliminated: CLP 1,252,602 per month.
  - Eligibility: residency, age (individuals should be at least 65 years of age), and income criteria (lowest 90 percent of the income distribution).
  - PAFE definition: estimated pension at retirement age calculated as an immediate life annuity considering applicant's age, family group, total balance in individual accounts via mandatory contributions, plus Recognition Bond, bonuses for each live-born child and real interest accrued.
- Simulation assumptions (key):
  - Projections start at end-2024; initial conditions based on observed pension savings as of December 2024.
  - Wage path projected year-by-year by earning decile-age-gender cells; contribution rate schedule applied to project pension savings.
  - Upon reaching statutory retirement age (65 for men, 60 for women), accumulated balances converted into PAFE.
  - Simplifying assumption: only single affiliates with no dependents considered when calculating self-financed pensions (estimates conservative).

### C. Baseline projections and role of indexation
- Projection framework:
  - Projected shares of beneficiaries from micro-simulation applied to UN World Population Prospects (2024) median demographic projections.
  - Wage growth proxied by growth in GDP per working-age population.
  - Macroeconomic assumptions aligned with IMF staff’s baseline macro framework.
- Baseline PGU cost projections (indexation preserved exactly):
  - Under CPI indexation (current legislation), PGU spending projected to increase from about 2.2 percent of GDP to approximately 2.5 percent of GDP by 2040.
  - Under wage rather than inflation indexation, fiscal costs would be about 0.5 percentage points higher by 2040 and about 1 percentage points higher by 2050.
  - Alternative metric: PGU-related spending to total central government revenue would require allocating between 1 to 2 percentage points more of public resources over the next five years.
- Key implication:
  - Despite higher contribution rates under the 2025 reform, most eligible retirees will receive the full benefit in 2050 and PGU expenditure will grow linearly with the elderly population unless parametric changes are made.

### D. Reform scenarios and quantified containment options
- Two main parametric reform approaches quantified:
  1. Income-based targeting:
     - Reduce PGU coverage from 90 percent to the bottom 60 percent of the income distribution, phased in gradually over five years.
     - Rationale: 60 percent corresponds to eligibility threshold established in 2008 under the original solidarity pillar.
     - Fiscal outcomes:
       - Total fiscal costs of the PGU would decline, reaching about 1.6 percent of GDP by 2040.
       - This is 0.8 percentage points below the baseline scenario.
       - If PGU is indexed to wage growth, with narrower targeting PGU spending reaches about 2 percent of GDP by 2040, 1 percentage point below the scenario with 90-percent coverage.
  2. Age-based targeting (increase retirement and PGU eligibility age to 67 for men and women):
     - Simulation results:
       - Total self-funded pension of women (retirement age rising from 60 to 67) would rise by around 30 percent.
       - Total self-funded pension of men (retirement age rising from 65 to 67) would rise by around 13 percent.
       - Number of PGU recipients would drop by about 11 percent by 2040.
       - Total PGU expenditure would be reduced by 0.3 percent of GDP relative to the baseline scenario.
     - Distributional effects:
       - Increasing retirement age increases average pension over the long term; marginal impact on the median total pension.
       - Narrower targeting reduces the average pension but does not affect the median total pension.
- Indexation considerations:
  - Maintaining a prudent indexation rule that avoids catch-up to wage growth helps contain fiscal costs under both baseline and reform scenarios.
  - An automatic wage-indexation mechanism would introduce a permanent and largely irreversible fiscal liability.
  - If benefit adjustment on adequacy grounds is warranted, it should be implemented on a discretionary, one-off basis rather than automatic wage-indexation.

### E. Contribution-rate schedule and calibrated parameters used in simulations
- Policy counterfactual — post-reform mandatory contribution rate to the personal account by period t:
  - Period t = 1 2 3 4 5 6 7 8 > 8
  - Post-reform: 10.1% 10.1% 10.25% 11.0% 11.7% 12.4% 13.1% 13.8% 14.5%
- Key calibrated parameters:
  - Annual real wage growth rate g = 1.4% per year
  - Annual gross real return on savings R = 3.5%
  - Commutation number CNU = 15.41 for males; 17.56 for females
  - Retirement age (baseline): 65 for males; 60 for females
  - Base year = 2024
- Simulation horizon: T = 25 years (2025–2049).
- Outcome variables: cross-sectional mean and median total monthly pension (in UF) for individuals aged 65 or above receiving a positive pension.

### F. Generational fund system (TDFs): design, glide-path construction, and market impact
- Institutional changes:
  - AFPs will provide target-date funds (TDFs) distinguished by target retirement year and a glide path (age-based schedule of portfolio allocations).
  - Pension Supervisor required to publish the reference glide path by September 2026.
  - Pension reform law stipulates at least ten generational funds; chapter considers ten age cohorts for analysis.
  - Proposed more granular cohort system (February 2026) suggests a 15-fund structure with three-year brackets for most cohorts.
- Three example glide-path construction approaches and estimated impacts:
  1. Status quo approach — prioritize mitigation of market impact by keeping aggregate allocations unchanged relative to 2024 average allocations.
     - Example outcome: with youngest cohort allocation equal to Type A allocation, oldest cohort allocation resembles Type D: about 65 percent domestic fixed income, 15 percent foreign risky assets, 12 percent foreign fixed income.
  2. Multi-fund-based approach — build glide paths from current allocations of the five multi-funds.
     - Estimated impact: weight on foreign risky assets will drop by about 12 percentage points of total assets; weight on domestic fixed income will increase by about 14 percentage points.
  3. Model-based approach — life-cycle optimal portfolio model (Cocco, Gomes, Maenhout 2005, CGM).
     - Baseline model-mapped glide path implies a mild shift from domestic fixed income to foreign risky assets: domestic fixed income decreases by 2.1 percent of total pension assets; foreign risky assets increase by 2.5 percent of total pension assets.
- Life-cycle model baseline and sensitivity (selected):
  - Baseline Sharpe ratio: 0.183; alternative Sharpe ratio: 0.304 (S&P 500 partially hedged).
  - Baseline median worker fully invests financial wealth in the risky asset up to age 31; risky asset weight declines to 32 percent at retirement (age 65).
  - Relative risk aversion parameter γ: baseline γ = 5; alternative γ = 3 yields much higher risky allocations (e.g., fully invested in risky asset up to age 46).
  - Reasonable range for change in domestic fixed income asset weight in aggregate: within ten percentage points of total assets in absolute terms (acknowledging wider model-implied ranges up to ~27.3 percent in scenarios).
- Transition timing and mitigation:
  - Transition scheduled to start April 2027 and end March 2030 (three-year transition period).
  - Mitigating channels:
    - Diversification of asset sales over time reduces immediate pressure.
    - Net non-investment revenue channel: AFPs currently earn annual net non-investment revenue of around 1.5 percent of total assets that can be allocated to target asset classes without selling other assets.
    - Weight-drift channel: higher returns on risky assets can mechanically increase risky asset weights over time if funds do not rebalance.
  - Illustrative example:
    - Adjusting initial 50-50 risky/fixed to 60-40 (ten percentage-point reduction in fixed income weight) with net non-investment revenue = 1.5% of initial total asset and equity risk premium = 4% → pension funds only need to sell 2.5 percent of fixed income holdings each year to attain 60-40 in three years.
- Announcement and communication risks:
  - Predictable trading patterns can be exploited; consistent and transparent communication is crucial but overly granular disclosure could exacerbate anticipatory trading behavior.
- Interaction with competition-enhancing mechanisms:
  - Auction mechanism: randomly chosen 10 percent of affiliates reassigned to AFP with the lowest commission every two years.
  - Investment performance-based reward-penalty mechanism: performance measured relative to reference portfolio; track records count from April 2028; application starts April 2029 based on past 12 months and extends up to 36 months.
  - Coincidence of auctions/reward-penalty mechanism with transition may incentivize AFPs to accelerate portfolio adjustments, potentially amplifying market impact.

### G. Asset reallocation magnitudes and financial-stability considerations
- Historical context:
  - Pandemic pension withdrawals decreased pension assets-to-GDP ratio by about 20 percentage points.
  - Past withdrawals reduced domestic fixed income holdings by more than 10 percent of GDP without major market disruptions due to offshore issuances and central bank liquidity support.
- Potential market effects from reallocations:
  - Outflows from domestic fixed income assets could affect funding for issuers (government, banks, non-financial corporates) and face liquidity/pricing challenges given AFPs’ large presence in local fixed income markets.
  - If reference glide path tilts toward risky assets, domestic pension asset allocations-to-GDP ratio will be lower and take more time to recover to pre-pandemic levels.
  - Long-run pension asset accumulation expected to help recover to pre-pandemic level regardless of glide-path design, but timing and intermediate market impacts differ.
- Supply-side considerations:
  - Encouraging supply of high-quality long-term fixed income assets is important to match pension asset growth; failure to do so risks yield compression and a search-for-yield dynamic with financial-stability implications.

### H. Derivatives, recent regulatory revisions, and risk controls (Annex III)
- Market context:
  - Chile’s derivatives market among the deepest peers per BIS Triennial Central Bank Survey (2025); FX and interest rate derivatives market depths (turnover-to-GDP) largest among Latin American peers with rapid increase since 2022.
  - Pension funds are major participants in FX derivatives and interest rate swaps markets; use FX derivatives to hedge FX risks and meet regulatory minimum FX hedge ratios.
- Rapid growth in interest rate swap positions:
  - Compared to 2023, notional amount of USD interest rate swap positions taken by pension funds are more than 30 times larger, surpassing 200 percent of total pension assets.
  - Prior regulation (pre-April 2026) limited aggregate notional size of investment-purpose derivatives within 3 percent of total pension assets (netting provision allowed much larger gross notional positions).
- Valuation and liquidity risks (staff estimates as of September 2025):
  - DV01 estimates:
    - DV01 of receiver USD interest rate swaps: around 18 million USD.
    - DV01 of payer USD interest rate swaps: around -4 million USD.
    - Net effect: one basis point parallel upward shift of US swap yield curve → USD 14 million valuation loss.
  - Standardized DV01:
    - USD interest rate swaps: about USD 6.4 thousand per USD 100 million AUM.
    - CLP interest rate swaps: about USD 0.8 thousand per USD 100 million AUM.
  - Example valuation impact: when US long-term swap rate rose from around 3.5 percent to 4.5 percent between September 2024 and January 2025, valuation of USD interest rate swaps decreased from around 1.0 percent of total assets to -1.5 percent of total assets.
- April 2026 regulatory revisions (post-public consultation in February 2026):
  - Shift from notional-based regulation to two metrics—limit on risk exposure and liquidity coverage ratio.
  - Limit on risk exposure defined as maximum potential valuation loss across six stress scenarios (most relevant: upward parallel shift of the yield curve by 40bps).
  - Limits on risk exposure (as percent of total assets):
    - Multi-fund Types A and B: 3 percent.
    - Type C: 2.5 percent.
    - Types D and E: 2 percent.
  - Liquidity coverage ratio: required high-quality liquid assets to cover current collateral needs plus potential liquidity needs under stress scenarios.
- Policy guidance:
  - Calibrate derivative regulations consistently within overall risk management framework of the generational fund system, aligned with the reference glide path and acceptable risk allowances.
  - Place appropriate limits on derivatives and alternative assets to avoid excessive risk taking and ensure liquidity risk management.

### I. Conclusions and policy recommendations (summarized)
- PGU and fiscal containment:
  - PGU’s tapered design is largely ineffective in practice: most eligible retirees receive the full PGU, implying substantial fiscal burden that will grow with the elderly population absent parametric changes.
  - Maintain prudent indexation rule (CPI indexation) and avoid automatic wage-indexation to prevent a permanent, largely irreversible fiscal liability.
  - Consider parametric adjustments to contain fiscal costs, acknowledging distributional trade-offs:
    - Narrower income targeting (90% → 60%) reduces PGU spending to about 1.6 percent of GDP by 2040 (0.8 percentage points below baseline).
    - Increasing retirement and PGU eligibility age to 67 reduces PGU expenditure by 0.3 percent of GDP relative to baseline and raises self-funded pensions (women ~30%, men ~13%).
- Generational fund transition and market stability:
  - Design the reference glide path to balance life-cycle optimality and mitigation of market impact; consider heterogeneity across affiliates (income, risk aversion).
  - Favor gradual implementation and use regulatory flexibility for auctions and reward-penalty mechanisms to smooth transition and reduce concentrated trading pressures.
  - Calibrate communication strategy to mitigate announcement effects while avoiding overly granular disclosure that enables anticipatory trading.
  - Encourage supply-side development of high-quality long-term fixed income assets to absorb pension asset growth without adverse yield compression.
- Derivatives and investment risk controls:
  - Regulate non-traditional assets and derivatives consistently with the generational fund investment regime; maintain limits and liquidity coverage requirements introduced in April 2026.
  - Embed derivative risk regulations in review cycles: Pension Supervisor and Technical Investment Committee to review reference portfolio and investment regime at least every seven years.

*Source: IMF staff chapter titled "CHILE’S PENSION REFORM—FISCAL IMPLICATIONS OF THE MINIMUM GUARANTEED PENSION" and related chapters/annexes (IMF staff chapter PDF).*

### 1. Distribution of Pension Contributions and Savings at Retirement ____________________ 7

### 1. Distribution of Pension Contributions and Savings at Retirement

### A. Introduction
- Chile approved a comprehensive pension reform in 2025 that raises the contribution rate by 7 percentage points and expands coverage to lift pension adequacy.
- The reform expands the solidarity pillar, the Minimum Guaranteed Pension (PGU), which accounts for roughly one-third of the estimated cost of the reform.
- Microdata analysis indicates the PGU’s tapered design does not effectively target beneficiaries in practice, implying a substantial fiscal burden.
- Measures that could contain PGU fiscal costs (noting distributional trade-offs) include:
  - narrower targeting of PGU;
  - maintaining a prudent indexation rule that precludes catch-up to wage growth; and
  - increasing the statutory retirement age and PGU eligibility age in line with life expectancy gains.

### B. Overview of the Chilean Pension System and the 2025 Pension Reform
- System structure:
  - First pillar: non-contributory Minimum Guaranteed Pension (PGU), ages 65 and above, lowest 90 percent of income distribution.
  - Second pillar: mandatory contributory individual accounts with private AFPs; workers contribute 10 percent of taxable wages; following the 2025 reform, employers required to contribute 8.5 percent to individual accounts and the new Social Insurance administered by FAPP.
  - Third pillar: voluntary supplemental savings.
- Key reform elements:
  - Reform approved by Congress on January 29, 2025.
  - Gradual increase in employer contributions by 7 percentage points to reach 8.5 percent of workers’ taxable income over a transition period of nine years (or eleven years under an alternative schedule).
  - PGU increased gradually from CLP 224,000 to CLP 250,000 depending on beneficiaries’ age.
  - New Social Insurance (FAPP) finances gender pension gap compensation, pension bonus based on years of contribution, and contributions to individual accounts with protected returns.
  - Transition from multi-fund to generational-fund system; mandates every two years a reallocation of 10 percent of affiliates to the fund administrator with the lowest fee.
  - Measures to increase contribution densities including coverage during unemployment and a simplified monthly contribution scheme for self-employed workers.
- Estimated fiscal implications:
  - Authorities estimate additional fiscal costs of around 1 percent of GDP over the medium term from the reform.
  - Projections include: lower tax revenues at around 0.34 percent of GDP by 2035; higher pension-related expenditures for public sector employees estimated at 0.22 percent of GDP; fiscal transfers to the FAPP projected to peak at about 0.23 percent of GDP.
  - Increased spending on the PGU is projected to add costs of 0.3 percent of GDP under the baseline (CPI-indexation) and accounts for about one third of the total fiscal burden of the reform.
  - Total additional costs projected to peak at around 1.1 percent of GDP by 2035 and gradually decline, stabilizing at approximately 0.9 percent of GDP by 2050.
- FAPP financing notes:
  - FAPP also finances disability and survivors’ insurance (SIS), contributed by employers, amounting to 1.5 percent on average of worker’s taxable income.
  - The additional employer contribution is allocated: (i) 4.5 percent to individual accounts; (ii) 1.5 percent to FAPP to finance the Protected Return Contribution (CRP) and the "Benefit for years contributed"; and (iii) 1 percent to FAPP to finance compensation for the gender pension gap.

### C. Challenges in the Current Design of the PGU
- Microdata and distributional findings:
  - Analysis uses Encuesta de Protección Social (EPS) microdata (32,537 individuals; monthly earnings history since 1981; monthly pension savings since 2002).
  - The mass of the distribution is highly concentrated at very low levels of the reference self-funded pension (PAFE), indicating a vast majority of retirees enter retirement with limited self-funded pensions and high reliance on the PGU.
  - Large disparities in lifetime earnings and contribution densities; substantial share contributing zero or close to zero to individual accounts.
  - Pronounced gender gaps: women systematically have lower earnings and lower contribution densities, amplifying pension outcome gaps.
- PGU eligibility and tapering design (as of February 2026):
  - Full PGU for retirees with self-financed pensions below the lower threshold: CLP 789,139 per month.
  - Tapered benefit between lower threshold and upper threshold.
  - Upper threshold at which PGU is fully eliminated: CLP 1,252,602 per month.
  - Eligibility criteria include residency, age (individuals should be at least 65 years of age), and income criteria (lowest 90 percent of the income distribution).
  - PAFE definition: estimated pension at retirement age used to determine PGU amount; calculated as an immediate life annuity considering applicant's age, family group, total balance in individual accounts via mandatory contributions, plus Recognition Bond, bonuses for each live-born child and real interest accrued.
- Simulation approach:
  - Projections start at end-2024 using subsample aged 40 and above, following Cabezon (2024) methodology.
  - Initial conditions based on observed pension savings as of December 2024.
  - Wage path projected year-by-year by earning decile-age-gender cells; contribution rate schedule applied to project pension savings.
  - Upon reaching statutory retirement age (65 for men, 60 for women), accumulated balances converted into PAFE.
  - Simplifying assumption: only single affiliates with no dependents considered when calculating self-financed pensions; estimates conservative because affiliates with dependents would have lower monthly PAFE from the same balance.
- Key empirical results:
  - In 2024, 96 percent of retirees aged 65 and above had a self-financed monthly pension below the lower threshold at which the PGU starts phasing out, implying that, if eligible, this share will receive the full PGU.
  - Simulations project more than 92 percent of retirees will be below the lower PGU threshold by 2050.
  - Even if PGU thresholds are kept frozen in nominal terms, by 2050 an extreme scenario projects 78 percent of women and 70 percent of men would remain below the lower PGU threshold.
  - Implication: despite higher contribution rates under the 2025 reform, most eligible retirees will receive the full benefit in 2050 and PGU expenditure will grow linearly with the elderly population.

### D. Baseline Scenario and the Role of Indexation
- Projection methodology:
  - Projected shares of beneficiaries (full and partially phased-out PGU) from micro-simulation applied to UN World Population Prospects (2024) median demographic projections.
  - Wage growth proxied by growth in GDP per working-age population.
  - Macroeconomic assumptions aligned with IMF staff’s baseline macro framework.
- Baseline cost projections (indexation assumptions preserved exactly):
  - Under CPI indexation (current legislation), PGU spending projected to increase from about 2.2 percent of GDP to approximately 2.5 percent of GDP by 2040.
  - Under wage rather than inflation indexation, fiscal costs would be about 0.5 percentage points higher by 2040 and about 1 percentage points higher by 2050.
  - Alternative metric: PGU-related spending to total central government revenue would require allocating between 1 to 2 percentage points more of public resources over the next five years.

### E. Reform Scenarios
- The chapter simulates the long-term fiscal implications of the current PGU design under different indexation regimes and reform scenarios to evaluate containment options (details and full scenario analysis are presented in subsequent sections and annexes).

*Source: IMF staff chapter titled "CHILE’S PENSION REFORM—FISCAL IMPLICATIONS OF THE MINIMUM GUARANTEED PENSION" (June 18, 2026).*

### 13.      With additional savings generated

### 13.      With additional savings generated

### Issue and reform options
- The nominal value of the PGU should be preserved at least by CPI indexation; if so, additional parametric changes are needed to contain its fiscal cost.
- Two main reform approaches are quantified:
  - Income-based targeting: reduce PGU coverage from 90 percent to the bottom 60 percent of the income distribution.
  - Age-based targeting: increase retirement and PGU eligibility age to 67 for men and women.
- If the PGU were indexed to wage developments, much larger offsetting parametric changes would be required.

### PGU costs under narrower income targeting
- Reform: gradually reduce eligibility from current coverage of 90 percent to 60 percent over a five-year period.
- Rationale: the 60 percent threshold corresponds to the eligibility threshold established in 2008 under the original solidarity pillar.
- Fiscal outcomes:
  - Total fiscal costs of the PGU would decline, reaching about 1.6 percent of GDP by 2040.
  - This is 0.8 percentage points below the baseline scenario.
  - If targeting is applied when PGU is indexed to wage growth, fiscal gains are larger:
    - With narrower targeting, PGU spending reaches about 2 percent of GDP by 2040, 1 percentage point below the scenario with 90-percent coverage.

### PGU costs under increased retirement age
- Reform: increase retirement and PGU eligibility age to 67 for both men and women.
- Context: OECD data show health-adjusted life expectancy (HALE) at birth in Chile rose from 76.8 years in 2001 to 79 years in 2021 — an increase of approximately two years over the past two decades.
- Channels affecting fiscal costs:
  - Direct reduction in the number of PGU recipients.
  - Increased individual savings from longer contribution periods, raising self-funded pensions.
- Simulation results:
  - Total self-funded pension of women (retirement age rising from 60 to 67) would rise by around 30 percent.
  - Total self-funded pension of men (retirement age rising from 65 to 67) would rise by around 13 percent.
  - Number of PGU recipients would drop by about 11 percent by 2040.
  - Total PGU expenditure would be reduced by 0.3 percent of GDP relative to the baseline scenario.
- Distributional effects:
  - Increasing the retirement age would increase the average pension over the long term (longer contribution period, higher accumulated investment returns, shorter PGU payout time).
  - Changing retirement and PGU eligibility age would have a marginal impact on the median total pension.
  - Narrower targeting would reduce the average pension (fewer public resources devoted to the solidarity pillar) but would not affect the median total pension.

### Comparative effects and indexation considerations
- Narrower targeting vs higher retirement age:
  - Narrower targeting: fiscal savings but lower average pension (Figure 4, left chart, dotted line).
  - Higher retirement age: fiscal savings and a higher average pension in the long term (Figure 4, left chart, blue line).
- Indexation regime importance:
  - Maintaining a prudent indexation rule that avoids catch-up to wage growth helps contain fiscal costs under both the baseline and reform scenarios.
  - An automatic wage-indexation mechanism would introduce a permanent and largely irreversible fiscal liability.
  - If benefit adjustment on adequacy grounds is warranted, it should be implemented on a discretionary, one-off basis rather than automatic wage-indexation.

### Model, data, and key parameters used for simulations
- Data: Administrative records from the Encuesta de Protección Social (EPS), balanced panel through December 2024; monetary variables expressed in UF using contemporaneous UF deflator.
- Simulation horizon: T = 25 years (2025–2049).
- Earnings schedule:
  - 10 × 5 matrix W of mean annual earnings by decile and five-year age bin.
  - W updated each year by uniform real wage growth factor g.
- Survival: annual Bernoulli draw with success probability qₓ(a) from CB-2020 male and RV-M-2020 female mortality tables.
- Savings accumulation (while working, age < 65 for males, < 60 for females):
  - Sᵢ(t+1) = Sᵢ(t) × R + cₜ × Wₜ[dᵢ, b(aᵢ(t))]
- Retirement conversion (age = 65 in baseline):
  - Pᵢ = Sᵢ / (CNU × 12)
- Post-retirement: pensions remain constant in real terms.
- Policy counterfactual — post-reform contribution-rate schedule (mandatory contribution rate to the personal account):
  - Period t = 1 2 3 4 5 6 7 8 > 8
  - Post-reform: 10.1% 10.1% 10.25% 11.0% 11.7% 12.4% 13.1% 13.8% 14.5%
- Key calibrated parameters:
  - Annual real wage growth rate g = 1.4% per year
  - Annual gross real return on savings R = 3.5%
  - Commutation number CNU = 15.41 for males; 17.56 for females
  - Retirement age (baseline): 65 for males; 60 for females
  - Base year = 2024
- Outcome variables: cross-sectional mean and median total monthly pension (in UF) for individuals aged 65 or above receiving a positive pension.

### Sample summary statistics (EPS; selected)
- Sample size: 18,004 observations
- Age (years): Mean 62.25; Median 61; Std. Dev. 13.63; Min 40; Max 101
- Savings (UF): Mean 420.12; Median 87.31; Std. Dev. 829.56; Min 0; Max 19767.81
- Annual earnings (UF, age ≤ 60): Obs 8,927; Mean 215.78; Median 97.18; Std. Dev. 276.19; Min 0; Max 968.24
- PAFE (UF/month, age ≥ 65): Obs 7,043; Mean 3.11; Median 0; Std. Dev. 6.52; Min 0; Max 66
- PAFE | PAFE > 0 (age ≥ 65): Obs 2,834; Mean 7.74; Median 5; Std. Dev. 8.36; Min 1; Max 66

### Conclusion and policy implications
- The 2025 pension reform improved pension adequacy, but reliance on the PGU remains high and fiscal costs are rising.
- Microdata analysis shows substantial heterogeneity in pension savings at retirement; the PGU’s tapered benefit structure is largely ineffective in practice, with most eligible retirees receiving the full PGU benefit.
- Higher contribution rates under the 2025 reform are unlikely to translate into a lower fiscal burden associated with the PGU without further parametric adjustments.
- Two explored parametric reforms (narrowing PGU targeting to poorest 60 percent; increasing retirement and PGU eligibility age to 67) would both generate sizable fiscal savings relative to the baseline, though with important distributional implications outside the scope of this chapter.
- Policy recommendation: maintain a prudent indexation rule (avoid automatic wage-indexation); implement any pension adequacy adjustment as a discretionary, one-off measure rather than through an automatic wage-indexation mechanism.

*Source: 13. With additional savings generated (IMF chapter PDF).*

### 6.      The relative efficiency of Chile's public expenditure in health and education can be

### 6.      The relative efficiency of Chile's public expenditure in health and education can be

### Methodology and scope
- Relative efficiency is assessed using Data Envelopment Analysis (DEA), a non-parametric statistical method that constructs an "efficiency frontier" representing the maximum attainable output for a given level of input in a sample of comparator countries.
- Comparator group: OECD economies (used as aspirational peers); analysis also references peer OECD and LA5 countries.
- Primary input: public spending per capita (in purchasing power terms).
- Primary outputs: sector-specific outcome indicators (e.g., students’ performance scores, health-adjusted life expectancy, age-adjusted mortality rates) rather than intermediate outputs (e.g., number of teachers, medical visits).
- Caveat: DEA does not control for country-specific structural characteristics (for example, geographic dispersion or prevalence of remote communities) that may necessitate higher inputs to achieve similar outputs.
- The analysis focuses on healthcare and education; exploring specific policy measures to address identified efficiency gaps is beyond the chapter’s scope.

### Health sector: system, capacity, spending, and efficiency
- System characteristics:
  - Mixed public and private provision, financing, and insurance.
  - National Health Fund (FONASA) covers around 80 percent of Chile’s population.
  - Financing of FONASA: a 7 percent mandatory contribution of taxable income by employed individuals and pensioners and fiscal transfers.
  - Private insurer (ISAPRES) covers around 13 percent of the population; relies on the same mandatory contribution supplemented by voluntary top-ups and co-payments.
  - Approximately 60 percent of total health spending in Chile is covered via publicly mandated schemes (OECD 2025).
- Capacity constraints and utilization:
  - ICU occupancy rate in Chile: 84 percent in 2023.
  - Gap with OECD average ICU occupancy: 16 pps (percentage points).
  - Public hospital bed density: 1½ beds per 1,000 inhabitants.
  - Chile ranks at the lower end of OECD countries in overall hospital bed density (public and private combined).
  - Longer waiting times for different types of healthcare compared to OECD average, potentially linked to bed capacity deficits.
- Health outcomes:
  - Health-adjusted life expectancy (HALE) at birth increased from 76.8 years in 2001 to 79 years in 2021.
  - LA7 average HALE declined from 74.5 to 73.4 years over the past two decades.
  - In 2021, perceived good health: around 45 percent for same-age individuals at the bottom quintile of the income distribution vs 60.5 percent at the top quintile.
  - Gender gap in perceived health in Chile exceeds that observed across the OECD.
  - Infant mortality rates remain higher than in the OECD; age-adjusted overall mortality and preventable and treatable mortality rates broadly in line with OECD average.
- Public spending trends and composition:
  - Public healthcare spending as a share of GDP: increased from 3.7 percent in 2013 to 5.5 percent in 2023.
  - Real per-capita public health spending (PPP-adjusted) in 2023 exceeded its pre-pandemic level only slightly.
  - At the central government level in 2024:
    - Employees’ compensation: 35.6 percent of total spending (up from 34.1 percent in 2019).
    - Subsidies and grants: 25.4 percent of total spending.
    - Goods and services: 22.9 percent of total spending.
  - Public spending (including via mandatory health insurance contribution schemes) represents approximately 60 percent of total health spending in Chile, compared with a weighted OECD average of 80 percent.
- Efficiency assessment:
  - Using government healthcare spending per capita (PPP terms) as input and HALE and age-adjusted mortality rate as outputs, Chile is positioned at or close to the efficiency frontier relative to comparator countries.
  - Interpretation: Chile manages its resources relatively well in achieving health outcomes compared with peers with similar spending allocation.
  - Nonetheless, there remains a need for absolute efficiency gains to contain rising spending pressures driven by demographic change and upward healthcare spending trends.

### Education sector: structure, outcomes, spending, and efficiency
- System characteristics and reforms:
  - Mix of public, fully private, and state-subsidized private schools financed through voucher-based mechanisms.
  - Higher education includes public and private universities, technical institutes, and training centers.
  - Key reforms:
    - School Inclusion Law (2016): phased out co-payments in subsidized private schools, mandated non-selective admissions and conversion into non-profit entities, expanded Preferential School Subsidy (SEP).
    - Gratuidad reform (2016): substantially expanded tuition-free access for eligible higher-education students.
    - Local Public Education Services (SLEP) reform: gradually transferring public school management from municipalities to the central government.
  - Based on OECD 2022 data:
    - Public financing accounts for roughly 80 percent of total resources for primary to post-secondary non-tertiary education.
    - Public financing accounts for about 38 percent for tertiary education.
  - Note: education accounts for around 7 percent in recent years of “below-the-line” operations, mainly via the State-Guaranteed Student Loan Program (CAE).
- Educational performance and inequality:
  - Chile’s PISA scores: higher than LA7 comparators but below OECD averages across all domains, with the largest divergence in mathematics.
  - Socio-economic disparities: Chile recorded lower socio-economic disparities in academic performance relative to both the OECD and LA7 averages (measured by PISA indicators), but higher gender gaps in math performance.
  - Adult skills: Chile ranked last among OECD countries in the 2023 Survey of Adult Skills (OECD, 2024) for proficiency in literacy, numeracy, and problem solving; evidence points to persistent qualification and skill mismatches.
- Spending levels and composition:
  - Government education expenditure as a share of GDP: increased from 3.8 percent to 4.5 percent in 2023; OECD average in 2023: 4.9 percent.
  - Chile ranks at the lower end of OECD distribution across different education levels when assessed in spending per student (PPP-adjusted); this largely reflects differences in income levels.
  - Spending-to-GDP ratios: Chile’s public spending on primary to post-secondary education was close to the OECD average in 2022; tertiary education spending exceeded the OECD average.
  - Central government education expenditure composition in 2024:
    - Subsidies and grants: roughly 85 percent of total spending.
    - Employee compensation share rose from 4.8 percent to 9.3 percent of total central government education expenditure (linked to SLEP reform).
    - Capital spending contracted from 3.3 percent to 0.5 percent of total central government education spending.
  - A breakdown by economic classification is only available at the central government level and not for the general government.
- Efficiency assessment:
  - Mapping PISA performance against inputs (government spending as share of GDP and PPP-adjusted dollars per student) shows Chile operates below the education spending efficiency frontier and lags behind peer countries at the frontier.
  - Distance from the efficiency frontier indicates significant scope to achieve the same performance outcomes at lower fiscal cost.
  - Challenges also affect tertiary education: despite relatively high public spending, adult proficiency outcomes remain low and skill mismatches persist.

### Conclusion and implications for fiscal space
- Overall public spending context:
  - Chile’s total public spending remains relatively small compared with OECD countries across most categories, despite a gradual upward trend.
  - Composition remains heavily concentrated in social sectors: health, education, and social protection.
  - Public investment has declined over time and currently stands at historically low levels (note: this may be understated if not all regional-level investment is captured).
- Efficiency findings:
  - Healthcare: Chile performs comparatively well relative to peers.
  - Education: significant potential exists to enhance the efficiency of public education spending.
  - Even in sectors with relatively good performance, expert assessments indicate scope for absolute efficiency gains.
- Policy implication:
  - Improving the efficiency of public spending is an important avenue for creating additional fiscal space.
  - Realizing relative and absolute efficiency gains would strengthen value for money of public spending, alleviating fiscal pressures while preserving quality of public services amid growing structural demands.

*Source: IMF staff chapter on the relative efficiency of Chile's public expenditure in health and education.*

### 2.      The 2025 pension reform introduces several key changes to the pension system and

### 1chlea2026002 - 2.      The 2025 pension reform introduces several key changes to the pension system and

### Key changes introduced by the 2025 pension reform
- Increase of contribution rate to the mandatory individual pension account from 10 percent of taxable income to 14.5 percent by 2033.
- There will also be a 2.5 percent contribution to fund the newly created social security benefits. As a result, the total contribution rate will increase to 17 percent (18.5 percent including contributions to Disability and Survival Insurance) by 2033.
- Transition of the pension fund system from the multi-fund system to a generational fund system.
- Introduction of competition-enhancing mechanisms among pension fund administrators (AFPs).
- Creation of a public autonomous fund, FAPP (Fondo Autónomo de Protección Previsional), that administers, among others, newly created pension benefits.

### Short-term challenges and financial-stability implications
- Pension withdrawals during the pandemic decreased the pension assets-to-GDP ratio by about 20 percentage point, resulting in a shallower capital market that is susceptible to external shocks.
- While the reform will deepen capital markets in the long run, its implementation could entail various short-term challenges:
  - Depending on the design of the transition to the generational fund system, it may induce large portfolio reallocations across major asset classes, with financial stability implications if abrupt adjustment in asset prices occurred and impacted funding of local bond issuers.
  - Introduction of competition-enhancing mechanisms among AFPs should lower commission fees and discipline AFPs’ investment performance, but it could have unintended consequences on AFPs’ risk-taking behavior.
  - The timing of the introduction of the new competition-enhancing mechanisms could complicate the adjustment process of asset allocations toward the generational fund system as both occur simultaneously.

### Institutional overview — Multi-fund system (current)
- Structure and operation:
  - Currently, there are seven AFPs, among which pension affiliates can freely choose.
  - Since 2002, the contributory pension fund system has operated under a multi-fund system. Each AFP provides five types of funds ranging from “Most Risky” Type A to “Most Conservative” Type E.
  - Affiliates can allocate their pension balance to up to two types of funds.
  - Investment limits by asset classes are differentiated by multi-fund type (e.g., the limit of the risky asset weight is 80 percent for Type A, while it is 5 percent for Type E).
  - Affiliates can switch between multi-funds within the same AFP without fees at least twice a year.
- Behavioral and market issues:
  - Excessive switching problem: In 2020, the cumulative volume of multi-fund type switches was equivalent to about 80 percent of the total pension fund assets.
  - Most switches occurred between types A and E; evidence suggests about three quarters of affiliates who switched the fund experienced worse performance compared to the benchmark.
  - Abrupt fund outflows from Type E funds could lead to dislocations in bank and corporate bond funding, and associated foreign exchange flows affected FX market frictions.
- Age-patterns and concentrations:
  - The most popular fund type evolves from Type B for affiliates under 40, to Type C for affiliates in their 40s and 50s, and to Type D for those in their 60s and older, with significant heterogeneity across cohorts.
  - Pension assets under management are the largest for the age cohort between the ages of 55 and 59.
  - The legal retirement age in Chile is 65 for men and 60 for women.

### Institutional overview — Generational fund system (TDFs)
- Design and features:
  - AFPs will provide target-date funds (TDFs) distinguished by target retirement year and a glide path (age-based schedule of portfolio allocations).
  - The pension reform law stipulates that there must be at least ten generational funds.
  - This chapter considers ten age cohorts for analysis: affiliates at or younger than ages 24, 25-29, 30-34, ..., 60-64, affiliates at or older than 65.
  - The Pension Supervisor is required to publish the reference glide path by September 2026.
- International and regional context:
  - TDFs have grown globally; in the US, AUM in TDFs reached USD 1.8 trillion in 2021.
  - Mexico transitioned to a TDF-style generational fund system in 2019; other Latin American countries are at various stages of adopting similar models.
- Expected benefits:
  - Automatic adjustment feature addresses excessive switching in the multi-fund system.
  - TDF glide paths designed with life-cycle optimal portfolio allocations are expected to enhance long-term investment performance. FIAP (2025) notes transition enabled pension funds to increase risky asset weights by three percentage points and preliminary results suggest investment return improved more than 100 basis points, contributing to improving the replacement rate.
- Operational design note:
  - The Pension Supervisor is currently designing the reference glide path and an investment regime for generational funds.
  - A proposed more granular cohort system (February 2026) would use a 15-fund structure: one fund for younger generations (34 years old or younger at the end of 2027), 13 funds with a three-year bracket range (age 35-37, 38-40, ..., 71-73), and one fund for the oldest generation (age at or above 74). Every three years, affiliates between 35 and 37 would be separated from the youngest generational fund to form a new fund, while the affiliates in the second to the oldest generational fund will be merged into the oldest generational fund. Note that the youngest cohort covers a somewhat wide range of ages.

### Estimating the size of asset reallocations due to the transition — approach overview
- Trade-off in glide path design:
  - The design faces a trade-off between life-cycle optimality and market impact due to asset reallocations. If the theoretically optimal glide path results in aggregate allocations quite different from current allocations, large reallocations could cause market disruptions.
  - The Pension Supervisor stated it would carefully consider both objectives when developing the reference glide path.
- Three approaches employed to produce example glide paths:
  1. Status quo approach — prioritize mitigation of market impact by keeping aggregate allocations unchanged relative to 2024 average allocations.
     - Construction assumptions: (i) linear glide path for intermediate age cohorts, and (ii) asset allocation of the youngest cohorts (those under age 30) given as either the current Type A or Type B multi-fund. Pin down oldest 65+ cohort allocation accordingly.
     - Example outcome: When the youngest cohort allocation equals the Type A allocation, the oldest cohort should allocate about 65 percent of their pension balance on domestic fixed income assets, 15 percent on foreign risky assets, and 12 percent on foreign fixed income assets. This allocation resembles the allocation of the Type D fund.
  2. Multi-fund-based approach — construct glide paths based on current allocations of the five multi-funds.
     - Default multi-fund option path: built from the default fund choices by age and gender (Type B for men and women below age 35; Type C for men 35-54 and women 35-49; Type D for men at or above 55 and women at or above 50).
     - Estimated impact: the weight on foreign risky assets will drop by about 12 percentage points of total assets, while the weight on domestic fixed income assets will increase by about 14 percentage points.
     - An alternative glide path evolving from Type A to Type E yields a qualitatively similar result but with a much steeper slope.
  3. Model-based approach — standard life-cycle optimal portfolio model (Cocco, Gomes, and Maenhout 2005, CGM).
     - Model setup: two financial assets (riskless bond and risky asset), labor income subject to income shocks, optimization of expected lifetime utility to derive optimal age-dependent risky asset weight.
     - Incorporates key parameters: risk aversion, correlation between income risk and risky asset returns, and social security benefits after retirement.

### Analytical utility and limitations
- Uses model-based exercises to:
  - Understand implications of transition for financial stability and pension affiliates’ welfare.
  - Provide inputs on sensitivity to key parameters and specifics of the Chilean pension system.
- Not a prescriptive recommendation:
  - The exercise does not intend to recommend any specific glide path as optimal. Designating one reference glide path for heterogeneous affiliates requires holistic review of affiliate characteristics and professional judgement and is beyond the scope of the model-based analysis.

*Prepared by Kazuhiro Hiraki (MCM) and IMF staff; content based on the provided chapter text.*

### 22.      Life-cycle optimal portfolio models tend to recommend large allocations weights on

### 22.      Life-cycle optimal portfolio models tend to recommend large allocations weights on 

### Model intuition and calibration
- In a simple mean-variance portfolio model for a constant relative risk-aversion investor, the optimal risky asset allocation is (휇−푟f)/(훾휎2).
- The optimal risky asset weight is proportional to the expected excess risky asset return (휇−푟f) and inversely proportional to the variance of the risky asset return (휎2) and the relative risk aversion parameter (훾).
- Life-cycle portfolio models incorporate human capital (labor income as a stream of cashflows) into total wealth; when there is no labor income risk the ratio of the optimal risky asset allocation to total wealth equals the optimal mean-variance weight (Merton, 1971; Jagannathan and Kocherlakota, 1996; CGM).
- The model is calibrated to reflect the institutional setting of the Chilean pension system and Chile-specific financial and labor market data. The baseline calibration is deliberately kept simple and supplemented by alternative specifications. (See Annex II in source for calibration details.)

### Baseline estimation results and mapped glide path
- Median worker: fully invests financial wealth in the risky asset up to age 31 (Figure 1, left column).
- Risky asset weight declines with age, reaching 32 percent at retirement (age 65).
- Risky asset weight remains above the “complete market case” (simple mean-variance solution) even in retirement because retirees still receive social security benefits.
- Mapping the estimated optimal risky asset weight to the four major asset classes assumes domestic-to-foreign allocation ratios within risky and fixed income categories remain as in 2024 (domestic:foreign = 76:24 for fixed income; 18:82 for risky assets).
- Baseline glide-path implication: mild shift from domestic fixed income assets to mainly foreign risky assets:
  - Domestic fixed income assets weight decreases by 2.1 percent of the total pension asset.
  - Foreign risky assets weight increases by 2.5 percent of the total pension asset.

### Results under alternative specifications
- Social security income after retirement (PGU-like retirement income):
  - Replacing retirement income with a constant function (reflecting Minimum Guaranteed Pension, PGU) yields an optimal risky weight path that is above baseline during younger periods and below baseline during later stages.
  - Interpretation: for younger workers the insurance effect of the PGU is larger because longer remaining working periods make negative income shocks more likely; higher expected replacement rate encourages more investment in risky assets early in life.
  - Aggregate-level effect: slightly larger outflows from domestic fixed income assets at 4.3 percent of the total pension asset.
- Risk-return profile of the risky asset (S&P 500 with partially hedged FX risk):
  - Baseline Sharpe ratio: 0.183. Alternative Sharpe ratio: 0.304.
  - Result: more allocations to the risky asset throughout life due to improved risk-return profile even after partially hedging exchange rate risk.
  - Shift implication: larger outflow from domestic fixed income assets estimated at -12.4 percent of the total pension asset.
- Relative risk aversion parameter (훾):
  - Baseline 훾 = 5. Alternative 훾 = 3 (less risk-averse) yields dramatic shift:
    - Median worker fully invests in risky asset up to age 46.
    - Risky asset weight stays above 50 percent throughout the life cycle.
    - Induced shift from fixed income to risky assets reaches around 30 percent of total assets.
  - Empirical note: Henríquez and Barrero (2026) document Chilean household risky allocations are much lower than implied by 훾=3, suggesting 훾 may be larger than three in aggregate.
- Correlation between income risk and risky asset return (휌):
  - Baseline sets persistent income-risks vs. risky return correlation = 0.
  - If correlation = 0.2, risky asset allocation during working age is significantly dampened.
  - Combined high risk aversion and positive correlation could reverse baseline reallocations—possible shift from risky assets to fixed income assets.

### Caveats and model limitations
- Model is stylized and does not capture full complexity of Chilean pension system and labor market. Key caveats include:
  - Unemployment spells and contribution density:
    - Calibration assumes full contribution density, but actual average density ≈ 60 percent for males and ≈ 50 percent for females due to unemployment and informality.
    - Accounting for unemployment risk and lower contribution densities would make human capital riskier and reduce optimal risky asset weight, especially early in life (Bagliano et al., 2019).
  - Segregation of mandatory pension account from private savings:
    - Model assumes affiliates manage pension and private savings jointly; in reality pension contributions are compulsory and pension balances cannot be withdrawn before retirement and are reimbursed under specific rules.
    - Literature (Gomes, Michaelides, and Polkovnichenko, 2009) suggests segregated tax-deferred accounts promote wealth accumulation and increase retirement consumption.
  - Habit formation:
    - Habit formation preferences (utility depends on excess consumption over habit level 퐻t) may lower risky asset allocations, especially for younger and low-income workers whose income is near habit/subsistence levels.
  - Choice of pension benefit modality:
    - Chilean retirees may choose life annuity or programmed withdrawal; model implicitly assumes programmed withdrawal and treats retirees as managing asset allocations post-retirement—this has implications for pensioner welfare and macro-level allocations between pension funds and life insurers.
  - Mapping single risky asset weight to actual asset allocations:
    - Estimation assumed (i) domestic-to-foreign allocation ratio stays as in 2024 and (ii) ignored the “other” asset class (alternatives and derivatives).
    - Exact rebalancing depends on investment limits, hedging requirements, and relative performance; inclusion of “other” assets could visibly affect aggregate allocations.

### Heterogeneity and distributional implications
- Optimal risky asset weight varies significantly by investor characteristics:
  - Degree of risk-aversion, income level, and income-asset risk correlation materially affect the optimal path.
  - Within the same age group, investors with less accumulated wealth and/or lower income allocate a higher weight to the risky asset.
- Dispersion under baseline:
  - Except very early life when investors fully invest in risky asset, interquartile range of optimal risky weight typically hovers around 10 to 20 percentage points.
  - Implication: a single system-wide reference glide path may have heterogeneous welfare effects across population groups; targeting the average worker may not suit many subgroups.

### Size and direction of aggregate asset reallocations and reasonable range
- Estimated results vary by glide-path design:
  - Default multi-fund option path could induce a large shift toward domestic fixed income assets worth 13.9 percent of total assets.
  - Model-based path with low risk-aversion parameter could induce a massive outflow from domestic fixed income assets of 27.3 percent of total assets.
- Considering full estimation range and additional non-modeled factors, a reasonable range for the size of asset reallocations (change in domestic fixed income asset weight) could be within ten percentage points of total assets in absolute terms.

### Transition to the generational fund system and market impact mitigation
- Outflows from domestic fixed income assets could affect funding for issuers (government, banks, non-financial corporates) and may face market liquidity/pricing challenges given AFPs’ large presence in local fixed income markets.
- Historical precedent: pension withdrawals during 2021-22 decreased domestic fixed income holdings by more than 10 percent of GDP but did not cause major market disruptions because bond issuers shifted issuances offshore and the central bank provided temporary liquidity support programs.
- Pension reform law transition period:
  - Transition scheduled to start April 2027 and end March 2030 (three-year transition period).
- Channels by which a transition period mitigates market impact:
  - Diversification of asset sales over time reduces immediate pressure to sell.
  - Net non-investment revenue channel: AFPs currently earn annual net non-investment revenue of around 1.5 percent of the total assets, which can be allocated to asset classes targeted for increases without selling assets targeted for decreases.
  - Weight drift channel: because risky assets typically have higher returns than fixed income assets under normal circumstances, risky asset weights can mechanically increase over time if pension funds do not rebalance.

*Sources: SP and IMF staff calculations.*

### 32.      These mitigating factors combined would dramatically alleviate the annual size of

### 1chlea2026002 - 32.      These mitigating factors combined would dramatically alleviate the annual size of

### Mitigating effect of a transition period and illustrative numerical example
- Scenario: pension fund adjusts initial 50-50 allocations on risky and fixed income assets to 60-40 allocations (a ten percentage-point reduction in the fixed income asset weight).
- Baseline assumptions used in the illustration:
  - Net non-investment revenue is set to 1.5 percent of the initial total asset.
  - The equity risk premium (which determines the impact of the weight drift channel) is set to four percent.
- Resulting illustrative finding:
  - Pension funds only need to sell 2.5 percent of their fixed income asset holdings each year to attain the 60-40 allocations in three years.
- Notes and definitions from the example:
  - “Net contributions” is calculated as contribution revenue less commission and SIS premium expenses.
  - “Free disposal” includes cash out due to the pension withdrawals.
  - The dark bar series shows the net revenue of pension funds except investment returns. The light bar series further excludes the free disposal item, which includes cashout due to the pension withdrawals during the pandemic.

### Announcement effects and communication
- Key risk:
  - Even with a sufficiently long transition period, asset prices may sharply react upon the announcement of new regulatory designs due to predictable trading patterns by institutional investors.
- Evidence and literature:
  - Predictable trading patterns (e.g., due to regulatory reasons, routine rebalancing) can be exploited by sophisticated investors (see e.g., Harvey, Mazzoleni, and Melone, 2025 and references therein).
- Policy implication on communication:
  - Consistent and transparent communication is crucial, but overly granular disclosure could exacerbate anticipatory trading behavior.
  - Communications should be mindful that regulatory requirements and implementation timelines that allow limited flexibility increase the predictability of trading patterns during the transition period.

### Competition-enhancing designs and implications for AFPs during transition
- Reform mechanisms intended to enhance competition among AFPs:
  - (i) An auction of randomly chosen ten percent of affiliates to reassign them to the AFP with the lowest commission fee in every two years.
  - (ii) An investment performance-based reward-penalty mechanism.
- Interaction with transition:
  - These mechanisms coincide with the three-year transition period to the generational fund system and may complicate asset management and portfolio adjustment by AFPs.
  - AFPs may be incentivized to complete portfolio adjustments well before the end of the transition period if the reward-penalty mechanism is tightly designed.
    - Performance is measured based on deviations from the reference portfolio.
    - Track records from April 2028 (one year after the start of the transition period) are counted for the performance measurement.
    - The application of the reward-penalty mechanism starts in April 2029 initially based on the performance over the past 12 months period. The window of the performance measurement period extends until it reaches 36 months.
- Financial stability concern:
  - The design of the performance-based reward-penalty mechanism may influence herding behavior.
  - Under the current multi-fund system, AFPs face a minimum guaranteed return: AFPs need to compensate for shortfalls from the minimum guaranteed return at the cost of their capital. The minimum guaranteed return is time-varying and calculated separately for each multi-fund type as the average return of seven AFPs over last 36 months.
  - The relative benchmarking design strongly encourages AFPs to mimic other AFPs to avoid underperformance relative to industry average (see e.g., Raddatz and Schmukler, 2011).
  - The new reward-penalty mechanism would mitigate herding as it is no longer based on relative performance within the industry; however, if tightly designed it may mechanically result in similar portfolio allocations around the reference portfolio.

### Long-term implications for contributions, asset accumulation, and market development
- Contribution changes and immediate effects:
  - Prior to the reform, employees contributed 10 percent of their taxable income to the mandatory pension account.
  - The reform requires contributions by employers totaling 7 percent of their employees’ taxable income, of which 4.5 percent goes to the individual pension account.
    - Footnote detail: Of seven percent of the additional employer contribution, 1.5 percent is tied to the protected-return contribution component. Between 2045 to 2055, this 1.5 percent contribution gradually migrates to the contribution to self-funded pension account.
  - At the aggregate level, increased contributions help accumulate pension assets and eventually strengthen pension benefits (see e.g., SP, 2024).
  - Employer contribution: It will increase to 6 percent in and after 2055. (Annex I referenced for details.)
- Long-run asset allocation and recovery:
  - Investment in domestic assets will recover to the pre-pandemic level regardless of the design of the glide path.
  - If the reference glide path is tilted more toward risky assets:
    - The domestic pension asset allocations-to-GDP ratio will be lower and take more time to recover to the pre-pandemic level.
  - The long-run effect of aggregate pension asset accumulation will eventually help recover to the pre-pandemic level.
- Supply-side considerations and financial stability:
  - Encouraging supply of high-quality long-term fixed income assets is important for capital market development.
  - If supply of high-quality long-term assets does not keep up with pension asset growth:
    - Excessive demand could lower yields of these assets, negatively affecting investment returns for pension funds and life insurance companies, ultimately affecting affiliates’ welfare.
    - Low-return environments may encourage a search for yield, potentially increasing financial stability risk.
  - Short-term assessment: this downside scenario seems unlikely in the short term as Chilean pension fund sector is in the process of recovering its pre-pandemic depth, but it may be an important long-term agenda as pension fund assets grow over coming decades.

### Regulatory design and risk controls
- Non-traditional assets and derivatives:
  - Regulations on the “other” asset category (which includes alternative assets and derivatives) should be consistently designed to safeguard the risk-return profile of generational funds.
  - Derivative investments warrant attention:
    - AFPs use derivatives for hedging exchange rate risks and for investment purposes, especially in recent years.
    - Derivatives may have significant financial stability concerns; regulations should place appropriate limits to avoid excessive risk taking and have necessary arrangements to manage liquidity risks.
    - Annex III referenced for recent developments in derivative investments by pension funds.
- Review mechanism:
  - The reform law mandates the Pension Supervisor and the Technical Investment Committee to review the reference portfolio and investment regime at least every seven years and adjust these regulations if necessary.

### Conclusions and policy recommendations (summarized)
- Balance objectives in designing the reference glide path:
  - The reference glide path should balance the optimality of the investment strategy and mitigation of market impact during the transition.
  - Consider heterogeneity: the optimal life cycle portfolio may significantly differ by income level and risk aversion; design should consider heterogeneous implications across groups.
  - Mitigating market impact benefits affiliates, especially older generations, by safeguarding investment performance.
- Favor gradual implementation and use regulatory flexibility:
  - Gradual implementation of reform elements (e.g., auctions, reward-penalty mechanism) can help achieve a smoother transition to the generational fund system.
  - Employ available regulatory flexibility in designing competition-enhancing mechanisms so AFPs can prioritize a smooth transition to the generational fund system.
- Calibrate communication strategy:
  - Gradual and well-calibrated dissemination of information on new regulatory designs can help mitigate sharp announcement effects on asset prices.
  - Avoid excessively detailed disclosures that would allow non-constrained investors to precisely predict AFPs’ future trading patterns.
  - Embedding appropriate flexibility in the investment regime and competition-enhancing mechanisms can reduce excessive predictability of AFP trading patterns.
- Regulate non-traditional assets and derivatives:
  - Place appropriate limits on derivatives and alternative assets to avoid excessive risk taking and ensure liquidity risk management.
- Encourage supply of high-quality long-term fixed income:
  - Support development of domestic supply to match pension asset growth to avoid adverse yield compression and a search-for-yield dynamic.

_ CHILE  INTERNATIONAL MONETARY FUND_

### 11.      Model parameters are calibrated to the Chilean data and institutional setup. To follow

### 11.      Model parameters are calibrated to the Chilean data and institutional setup. To follow

### Model calibration and population/income profile
- Calibration target: an average formal male worker with full contribution density.
- Base year: 2024.
- Life cycle:
  - Enters labor force at age 20.
  - Retires at age 65 (legal retirement age for men is 65).
  - Lives up to age 110 (mortality table terminal age).
- Conditional survival rate 푝푡: taken from the Pension Supervisor’s mortality table for 2020.
- Deterministic income curve f(t):
  - Constructed from ESI (Encuesta Suplementaria de Ingresos) 2024 average formal male worker data.
  - Smoothed income curve by age from age cohort data.
  - Peak of income curve: CLP 15.3 million per year (around USD 16.3 thousand) at age 43.
- State variables:
  - In baseline constant replacement rate retirement income setting, value function is homogeneous with respect to current permanent labor income enabling dimension reduction by scaling; however, vt is explicitly included as a state variable because the PGU-style retirement income specification prevents that reduction.

### Preference and retirement income parameters
- Relative risk-aversion parameter γ:
  - Baseline: γ = 5.
  - Alternative scenario: γ = 3.
  - Note: γ = 5 is lower than CGM for the US (γ = 10) but used in other studies.
- Replacement rate λ:
  - Baseline: λ = 0.38.
  - Rationale: average replacement rate for men retired between 2015 and 2022, excluding self-financed component (SP, 2024).
  - Model treatment: only the non-self-financed pension component is considered for retirement period income because pension account balance is treated as part of accumulated wealth.
- PGU-like retirement income specification (alternative):
  - Annual retirement income = CLP 2.57 million if permanent income at retirement < CLP 26.3 million.
  - Annual retirement income = 0 if permanent income at retirement > CLP 41.7 million.
  - For permanent income at retirement between CLP 26.3 million and CLP 41.7 million, retirement income is linearly interpolated between CLP 2.57 million and zero.
  - Note: The baseline CLP 2.57 million is based on the PGU benefit for 2024. The thresholds are mapped from pension account balance–based thresholds by assuming the self-financing replacement rate of 33 percent (SP, 2024).

### Income risk and return parameters
- Income shock standard deviations (from Huneeus and Repetto (2005)):
  - Permanent income shock standard deviation: 0.1449.
  - Temporary income shock standard deviation: 0.2049.
  - Relative to CGM: permanent income risk is about 1.5 times larger than CGM; temporary income risk is about three quarters of CGM.
- Real risk-free rate: 2 percent (same as in CGM; consistent with historical short-term real interest rate level in Chile).
- Risky asset calibration (IPSA, 2016–2025 historical average):
  - Risk premium: 3.2 percent.
  - Volatility: 17.5 percent.
  - Excess return calculation: IPSA return minus central bank’s policy rate.
- Alternative risky asset profile (S&P 500 with partially hedged exchange rate risk):
  - Risk premium: 6.2 percent.
  - Volatility: 20.4 percent.
- Correlation between permanent income shock and stock return ρ:
  - Baseline: ρ = 0.
  - Alternative positive correlation scenario: ρ = 0.2.

### Annex III — Pension funds’ derivative investment: market context and recent developments
- Market depth:
  - Chile’s derivatives market among the deepest peers per BIS Triennial Central Bank Survey (2025); FX and interest rate derivatives market depths (annualized turnover-to-GDP ratio) largest among Latin American peers with rapid increase since 2022.
  - Annualized turnover calculated by multiplying daily turnover by 252; GDP converted to USD by average exchange rate in respective years.
- Pension funds as market participants:
  - Major participants in FX derivatives and interest rate swaps markets.
  - Use FX derivatives to hedge FX risks and meet regulatory minimum FX hedge ratios.
  - Active trading in derivatives exchanging pesos and UFs (Unidad de Fomento).
  - Outstanding FX derivative contract size stable in recent years.

### Rapid increase in interest rate swap positions and regulatory context
- Notional expansion:
  - Compared to 2023, notional amount of USD interest rate swap positions taken by pension funds are more than 30 times larger, surpassing 200 percent of total pension assets.
  - CLP interest rate swap positions also increased but more moderately.
- Prior regulation (pre-April 2026):
  - Regulatory metric: aggregate notional size of investment-purpose derivatives within 3 percent of total pension assets (netting provision between payer and receiver positions).
  - Netting provision allowed pension funds to take gross notional positions much larger than 3 percent by offsetting payer and receiver positions with similar notionals.
  - Net notional size was close to zero and below the 3-percent limit for both CLP and USD interest rate swaps.
- Valuation and liquidity risks from swap positions:
  - Valuation risk: swap contract valuation fluctuates with yield curve changes; example—pension funds paying floating rates and receiving fixed rates incur valuation losses when yield curve shifts up.
  - Liquidity risk: negative valuation exceeding pledged collateral triggers margin calls; margin calls occur daily and collateral must typically be posted within two business days, potentially forcing asset sales or position closures if eligible collateral is insufficient.
  - Duration mismatch risk: pension funds hold longer-maturity receiver swaps and shorter-maturity payer swaps, producing positive duration risk (receiver duration > payer duration) and valuation losses when yield curve shifts up.
- Quantified exposure and recent valuation impacts (staff estimates as of September 2025):
  - DV01 (sensitivity metric) estimates:
    - DV01 of receiver USD interest rate swaps: around 18 million USD.
    - DV01 of payer USD interest rate swaps: around -4 million USD.
    - Net effect: one basis point parallel upward shift of US swap yield curve → USD 14 million valuation loss.
  - Standardized DV01:
    - USD interest rate swaps: about USD 6.4 thousand per USD 100 million AUM.
    - CLP interest rate swaps: about USD 0.8 thousand per USD 100 million AUM.
  - Example valuation impact: when US long-term swap rate rose from around 3.5 percent to 4.5 percent between September 2024 and January 2025, valuation of USD interest rate swaps decreased from around 1.0 percent of total assets to -1.5 percent of total assets (a marked swing).
  - Pension funds increased cash and deposit holdings when required to pledge more collaterals around end-2024.

### April 2026 regulatory revisions on derivative investments
- Timing: revised regulations published in April 2026 after public consultation in February 2026.
- Key changes: shift from notional-based regulation to two metrics—limit on risk exposure and liquidity coverage ratio.
- Limit on risk exposure:
  - Potential valuation loss amount calculated under six stress scenarios; the maximum loss across scenarios defines risk exposure.
  - Most relevant stress scenario given the maturity structure: upward parallel shift of the yield curve by 40bps.
  - Limits on risk exposure (as percent of total assets):
    - Multi-fund Types A and B: 3 percent.
    - Type C: 2.5 percent.
    - Types D and E: 2 percent.
  - Other five scenarios: parallel downward shift of the yield curve; decrease and increase in the shorter-end of the yield curve; steepening (decrease in the shorter-end and increase in the longer-end); flattening (increase in the shorter-end and decrease in the longer-end) of the yield curve.
- Liquidity coverage ratio:
  - Requirement that pension funds hold sufficient high-quality liquid assets to cover required liquidity.
  - Required liquidity = current collateral needs + potential liquidity needs under stress scenarios (latter tied to the risk exposure amount).
  - High-quality liquid assets defined by eligible asset classes and haircut rates, if applicable.

### Policy guidance and system integration
- Recommendation: calibrate regulations on derivative investments consistently within the overall risk management framework of the generational fund system.
- Rationale:
  - Rapid build-up of USD interest rate swap positions in recent years prompted Pension Supervisor’s prompt regulatory amendment prior to transition to the generational fund system.
  - As transition of the pension fund system starts in 2027, derivative risk regulations should be embedded in the investment regime of the generational fund system consistently with:
    - The reference glide path.
    - Acceptable risk allowance for pension funds under the new system.

*Source: Chapter 11 and Annex III of the IMF staff chapter on Chile (1chlea2026002).*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2026/english/1chlea2026002.pdf_