## wpiea2021232-print-pdf

## Source details

**Canonical URL:** [wpiea2021232-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021232-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021232-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021232-print-pdf.pdf.json)

---

### I. Introduction — context and objectives
- Chile replaced a traditional PAYG defined-benefit system with a fully funded defined-contribution system financing individual capital accounts managed by private fund managers (AFPs).
- Motivations: efficiency, fiscal concerns, and “a desire to reduce the role of the government in economic affairs” (OECD, 1998).
- Early benefits: growing private savings and development of local financial markets (Roldos, 2007).
- Limitations:
  - Mandatory contribution rates set at relatively low levels, yielding low replacement rates.
  - Informality, self-employment, and low job tenure produced low contribution densities and coverage.
- Policy responses and shocks:
  - Solidarity pillar introduced in 2008; 50 percent increase in the minimum pension introduced in December 2019.
  - COVID-related extraordinary measures in 2020 allowed pension-account withdrawals; three rounds amounted to 19 percent of GDP (by early-May 2021).
  - First two withdrawals: about 30 percent of individuals who withdrew funds depleted their pension accounts (as of early-May 2021).
- Paper objectives:
  - Project system outcomes under different scenarios for current contributors.
  - Assess impact of COVID-related withdrawals on expected replacement rates and fiscal costs.
  - Explore reform options affecting future retirees.

### II. Key findings on replacement rates (pre-withdrawals and comparisons)
- Pre-withdrawals: Chile’s replacement rates compared unfavorably to OECD peers with substantial heterogeneity across cohorts and income groups.
- Benchmarks:
  - Altamirano et al. 2018: expected replacement rate for an average Chilean worker retiring in 2015 was 38 percent.
  - OECD projection: a Chilean retiring around 2060 would have a replacement rate of about 30 (percent), about 20 percentage points below the OECD average.
- Cohort patterns:
  - Male workers aged 60–65 today: expected pension about 45 percent of final wage.
  - Male workers aged 20–25 in 2020: expected replacement rates roughly 40 percent.
  - Females 60–65: roughly 40 percent; females 20–25: about 30 percent.
- Solidarity pillar effects:
  - Increases replacement rates for low-income pensioners (bottom 60 percent); without it replacement rates would be lower.
  - Before withdrawals, solidarity benefits projected to account for ~30 percent of total expected average pension at retirement for men and 50 percent for women.
  - For contributors in 2020, solidarity pillar raises expected average replacement rate by almost 15 percentage points for men and close to 20 percentage points for women.

### III. Causes of low and declining replacement rates
- Low contribution rate: effective contribution rate lower than most OECD countries; initially set low for PAYG transition and not increased.
- Low contribution density:
  - Average probability a male worker contributes in a given month: 60 percent; for women: 50 percent.
  - Average contribution density for males retiring between 2017 and 2020: 60 percent; for females: 46 percent.
- Demographics: younger generations face longer life horizons—workers retiring in 40 years will spread savings over an additional five years (four years for women) compared to those retiring today.
- Lower future expected returns: real interest rates have declined and are expected to remain low.
- Historical illustrative exercise (1981 vs 2020 assumptions):
  - Implicit retiree life expectancy in 1981 about 78 years; with real return 6 percent replacement rate ~90 percent.
  - With real return 8 percent (1981–2019 average) replacement rates reach 160 percent.
  - Increase in life expectancy to 2020 lowers replacement rate from 90 percent to 62 percent (real return 6 percent).
  - Lowering real return to 4.15 percent with 1981 life expectancy falls from 90 percent to 55 percent.
  - Combined higher life expectancy and lower returns lowers replacement rate to 38 percent (current young male entering workforce).

### IV. Description and distribution of COVID-related withdrawals
- Three rounds of withdrawals reached US$48 billion (about 19 percent of GDP) by early-May 2021.
- Allowed withdrawal: generally 10 percent per round; minimum and maximum set at 35 UF and 150 UF respectively.
- First and third withdrawals tax-exempt; second exempt only for lower earners under threshold.
- Official data to February 2021: close to 10.5 million people withdrew in first or second withdrawals; 30 percent depleted their accounts.
- Participation:
  - Roughly 95 percent of people with positive balances in June 2020 used at least one of the first two withdrawals; over 7 million withdrew twice.
  - Close to 3 million depleted accounts from the first two withdrawals; close to 3.8 million exhausted accounts through the three withdrawals.
- Average amounts and shares:
  - Average amount withdrawn in each round about US$2,000.
  - Average individual withdrew 40 percent of account in first withdrawal, slightly over 30 percent in second.
  - Of US$48 billion withdrawn in the two rounds, about 55 percent came from individuals 46 and older.
- Detailed first-withdrawal distribution (examples):
  - Less than $1,459 balance: mean amount withdrawn $606; 806,055 people; 14.3%.
  - $1,459–$14,594: mean amount withdrawn $1,461; 2,530,247 people; 45.0%.
  - $14,594–$62,547: mean amount withdrawn $2,822; 1,980,438 people; 35.2%.
  - More than $62,547: mean amount withdrawn $6,226; 275,292 people; 4.9%.
  - Total: 5,624,149 people; mean amount withdrawn $2,057; mean % of balance withdrawn 32.8%.
- Withdrawal behavior by age (as of April 27):
  - Below 25: 260,346 people; 4.6%; average amount $649; mean share 88.4%.
  - 26–35: 1,452,607; 25.8%; avg $1,369; mean share 44.6%.
  - 36–45: 1,439,173; 25.6%; avg $1,935; mean share 28.4%.
  - 46–55: 1,375,072; 24.4%; avg $2,572; mean share 23.3%.
  - 56–65: 915,760; 16.3%; avg $2,913; mean share 21.1%.
  - 66+: 180,574; 3.2%; avg $2,330; mean share 25.2%.
  - Total: 5,624,149; average amount $2,057; mean share 32.8%.

### V. Impact of withdrawals on pensions and fiscal costs
- Aggregate withdrawal magnitude: three rounds amounted to 19 percent of 2020 GDP (by early-May 2021).
- Expected impact on current affiliates (without compensating government support):
  - Average withdrawal projected to cause a 7 percent decline in pension at retirement.
  - Three rounds reduce the self-financed portion of pensions of current affiliates by 21 percent on average.
  - Buffering by the solidarity pillar increases government-funded supplements, yielding a reduction in total pensions of about 7 percent and decline in average expected replacement rate from 37 percent to 35 percent.
- Distributional impacts on self-funded pensions:
  - Projected average decline in self-funded portion: 19 percent for men and almost 23 percent for women; larger effects among older cohorts.
  - Males in their 20s: reductions in self-funded pensions of 5 to 12 percent after withdrawals.
  - Older cohorts: reductions can exceed 60 percent for elderly with lower balances.
- Replacement rates:
  - Projected decline about 3 percentage points after withdrawals for average male worker, and 1.5 percentage points for female workers.
  - In absence of additional government support, falls would be over 4 percentage points for men and over 2 percentage points for women.
  - APS dampening: for men APS mitigates most for 50–55 age group by 1.5 percentage points; for women largest additional impact for 40–45 age group.
- Fiscal impacts of solidarity pillar:
  - Withdrawals increase number of APS recipients and amounts received.
  - Additional fiscal cost estimated at a net present value of about 6 percent of 2020 GDP.
  - Additional fiscal costs peak in 2060 between 0.09 and 0.17 percent of GDP depending on assumptions.
  - Increasing profile of additional fiscal costs equivalent, in net present value, to a one-off fiscal cost between 3 and 6 percent of GDP in 2020.
  - If withdrawals do not affect income distribution, fiscal cost would peak at about 0.09 percent of GDP (baseline) and amount to a net present value of 3 percent of GDP in 2020.
  - Fiscal cost from withdrawals by current pensioners expected to peak at 0.045 percent of 2020 GDP in next few years (0.025 percent if support only to pensioners in lower 60 percent).
- Tax revenue effects:
  - Under constant income tax structure, government would lose over USD 1.6 billion over 40 years (net present value 2020 terms).
  - Foregone revenue would peak around 2060 at approximately 0.008 percent of GDP.
  - If three withdrawals had been fully taxable, tax collection would have increased by over USD 1.8 billion, or 0.7 percent of GDP.

### VI. Short-term macroeconomic and financial effects of withdrawals
- Withdrawals mitigated falling household income and boosted retail sales and consumption of durables.
- Measure was not well targeted—reached upper quintiles who saw labor income gains early in the pandemic.
- Many people transferred withdrawn money into bank accounts rather than immediate spending.
- Potential adverse effects: inflationary pressure, deterioration of local financial conditions, greater exchange rate volatility, and decoupling of long-term interest rates.

### VII. Reform experiments and policy implications
- Reform types examined: increases in contribution rates, increases in retirement age, improvements in contribution density, and combinations.
- Key assumptions: reforms implemented immediately after the three withdrawals; results are upper bounds and do not model behavioral responses.
- Isoquant insights (targeting a 40 percent average replacement rate):
  - Female retirement age to 65 and contribution rate to 14.5 percent; or
  - Current contribution rate and retirement age to 69.5 for all.
  - Increasing contribution density to 70 percent reduces required contribution rate and retirement age increases.
- Example multi-parameter reform (relative to post-withdrawal baseline):
  - Increase contribution rate to 16 percent.
  - Increase retirement age for men and women to 67.
  - Improve contribution density to 70 percent.
  - Projected outcomes:
    - Population average expected replacement rate rises to 50 percent from 35 percent.
    - For males aged 20–25: expected replacement rate rises to 70 percent in one presentation; other passages report 66 percent for young males and 59 percent for young females under combined measures — preserve each quoted figure exactly as in source contexts.
    - For females aged 20–25: expected replacement rate rises to close to 60 percent in one presentation; other passages report 59 percent under combined measures.
  - Example combination reduces solidarity support needed; fiscal cost of system expected to be 0.8 percent of GDP lower in 2060.
- Single-parameter impacts:
  - Increase contribution rate from 10 percent to 13 percent:
    - Average expected replacement rate from 35 to 37 percent.
    - Young affiliates (20–25): males 37 to 45 percent; females 29 to 33 percent.
  - Increase mandatory contribution rate from 10 percent to 16 percent (Annex IV):
    - For cohort currently 20–25: self-financed balance at retirement increases by about 59 percent.
    - Expected replacement rates for aged 20–25: overall 46 percent; men 53 percent; females 36 percent.
    - Fiscal impact: reduces fiscal cost by 0.3 percentage points of GDP by 2060.
  - Increase contribution density to 70 percent:
    - Young males: expected replacement 37 to 41 percent.
    - Young females: expected replacement 29 to 34 percent.
    - Fiscal impact: lowers government fiscal cost by 0.10 percent in 2060.
  - Increase retirement age to 67 for both sexes:
    - Males: expected replacement increases by 2 percentage points.
    - Females: expected replacement increases by 3 percentage points.
    - Real wage growth assumption (1.25 percent) dampens increases.
- Caveats and behavioral considerations:
  - Framework omits behavioral responses; higher contribution rates may increase informality.
  - Evidence cited: a 5 percentage point increase in contribution rates increases informality by 12.5 percent for men and 9.3 percent for women in Chilean estimates.
  - Some results should be interpreted as upper bounds; reforms may need phasing to address political economy and labor market considerations.
- Policy recommendations:
  - Update system parameters periodically (e.g., automated revisions every five or 10 years) to reflect life expectancy and global financial conditions.
  - Task specific institutions with preparing analysis and proposals.
  - Increase contribution density via labor market and structural reforms encouraging formal-sector employment.

### VIII. Universal Basic Pension (UBP) — scenarios and fiscal quantification
- Two UBP scenarios:
  - UBP set today at half real minimum wage and remains constant in real terms.
  - UBP set at 50 percent of the real minimum wage and grows at same rate as wages (real increase 1.25 percent).
- A UBP of half the minimum wage approximates paying today’s PBS to every retiree and falls slightly below the poverty line.
- Fiscal cost estimates for UBP of half the minimum wage:
  - 5.6 UF or about US$ 230 (Using exchange rate at end-February 2021).
  - Approximately 2.5 percent of GDP each year (projected over time).
  - This would amount to close to 5 percent of GDP in 2020.
  - If basic pension increases with wages, cost would increase to over 3.5 percent of GDP by 2050 (close to 8 percent of today’s GDP).

### IX. Fiscal footprint and projections (baseline and alternative scenario)
- By 2020, fiscal costs associated with the pension system were approximately 2.2 percent of GDP, of which 1.1 percent stem from the solidarity pillar.
- Baseline projection: total fiscal costs projected to increase gradually to 1.6 percent of GDP by 2060 (assuming PBS and APS parameters unchanged).
- Alternative scenario (PBS and APS grow at wages, 1.25 percent real):
  - Fiscal costs can reach 3 percent of GDP in 2060.
  - Alternative scenario: young males projected replacement rate at retirement increases by 10 percentage points from about 39 to 50 percent; young females increase by 10 percentage points from about 30 to 42 percent (driven by government supplement increases).
- Additional fiscal costs from withdrawals:
  - Peak at about 0.17 percent of GDP (0.21 percent in alternative scenario).
  - Net present value of additional fiscal costs about 6 percent of 2020 GDP.
  - If withdrawals do not affect income distribution, peak at about 0.09 percent of GDP and net present value 3 percent of GDP.
  - Fiscal cost derived from withdrawals by current pensioners peaks at 0.045 percent of 2020 GDP (or 0.025 percent if support limited to pensioners in lower 60 percent).
- Tax revenue effects:
  - Government would lose over USD 1.6 billion over 40 years (net present value 2020 terms) under constant tax structure.
  - Foregone revenue peak ~0.008 percent of GDP around 2060.
  - If withdrawals had been fully taxable, tax collection would have increased by over USD 1.8 billion, or 0.7 percent of GDP.

### X. Modeling and key methodological assumptions
- Projection start: June 2020 data (one month before 1st withdrawals) to build counterfactual.
- Key parameter assumptions:
  - Real returns on pension accounts: 4.15% per year.
  - Life annuity rate: 3.36%.
  - Real wage growth: 1.25% per year.
  - Mandatory contribution rate: 10% of gross salary (baseline).
  - Contribution density assumed constant over time in baseline exercises.
- Withdrawals assumption: individuals in each age-gender-account balance cell withdraw the maximum allowed amount per withdrawal (formula specified in source).
- Solidarity pillar parameters:
  - PBS and PMAS values up to 2022 set per 2019 announcement; from 2022 results shown under two assumptions: inflation indexation (2008 rule) and alternative with real growth 1.25 percent.

### XI. Annex highlights — Alternative scenario and individual reform scenarios
- Alternative scenario (Annex III):
  - PBS and PMAS increase in real terms by 1.25% per year.
  - Outcome: young males’ projected replacement rate increases by 10 percentage points (39 to 50 percent); young females increase by 10 percentage points (30 to 42 percent).
  - Mechanism: larger government supplement increases fiscal cost medium-term.
- Individual reform scenarios (Annex IV):
  - Increase mandatory contribution rate from 10% to 16%:
    - Self-financed balance for current 20–25 cohort rises ~59%.
    - Expected replacement rates for 20–25: overall 46 percent; men 53 percent; females 36 percent.
    - Fiscal impact: reduces fiscal cost by 0.3 percentage points of GDP by 2060.
  - Increase contribution density to 70%:
    - Young males: 37 to 41 percent; young females: 29 to 34 percent.
    - Fiscal impact: lowers government fiscal cost by 0.10 percent in 2060.
  - Increase retirement age to 67 for both sexes:
    - Males: +2 percentage points expected replacement.
    - Females: +3 percentage points expected replacement.
    - Noted that equalizing retirement ages and contribution density does not eliminate gender gaps due to life expectancy and accumulated assets differences.

*Source: wpiea2021232-print-pdf - References..............................................................................................................*

### References..............................................................................................................

### wpiea2021232-print-pdf - References..............................................................................................................

### I. Introduction — context and objectives
- Chile replaced a traditional PAYG defined-benefit system with a fully funded defined-contribution system financing individual capital accounts managed by private fund managers (AFPs).
- Motivations for the reform included efficiency, fiscal concerns, and “a desire to reduce the role of the government in economic affairs” (OECD, 1998).
- Early benefits linked to the reform included growing private savings and development of local financial markets (Roldos, 2007).
- Limitations emerged over time:
  - Mandatory contribution rates were set at relatively low levels, resulting in low replacement rates relative to initial expectations and international standards (Barr and Diamond, 2016).
  - Informality, self-employment, and low job tenure produced relatively low contribution densities and coverage.
- Policy responses and shocks:
  - Introduction of the solidarity pillar in 2008 to improve fairness and functioning; a 50 percent increase in the minimum pension introduced in December 2019.
  - COVID-related extraordinary measures in 2020 allowed pension-account withdrawals; three rounds of withdrawals amounted to 19 percent of GDP (by early-May 2021).
  - The first two withdrawals resulted in about 30 percent of individuals who withdrew funds depleting their pension accounts (as of early-May 2021).
- Paper objectives:
  - Project key outcomes of the system under different scenarios, focusing mainly on current contributors who will retire in the future.
  - Assess impact of COVID-related withdrawals on expected replacement rates and fiscal costs.
  - Explore reform options affecting future retirees.

### II. Key findings on replacement rates (pre-withdrawals and comparisons)
- Prior to COVID-related withdrawals, Chile’s pension system yielded replacement rates that compared unfavorably to OECD peers, with substantial heterogeneity across cohorts and income groups.
- Replacement rates were larger compared to what a PAYG would have produced for many groups, particularly men.
- Drivers of relatively low average replacement rates:
  - Policy parameters (contribution rates and retirement age) originally set to yield higher replacement rates but not adjusted for higher life expectancy, declining interest rates, and low contribution density.
  - Chile has a lower effective contribution rate than most OECD countries.
  - The solidarity pillar introduced in 2008 increases replacement rates for low-income pensioners (bottom 60 percent), without which replacement rates would be even lower.
- Benchmarking evidence:
  - Altamirano et al. 2018: expected replacement rate for an average Chilean worker retiring in 2015 was 38 percent—lower than the average LAC country.
  - OECD projection: a Chilean retiring around 2060 would have a replacement rate of about 30 (percent), about 20 percentage points below the OECD average.
- Cohort patterns:
  - Male workers aged 60–65 today are expected to receive a pension of about 45 percent of their final wage (Figure 3, Panel A).
  - Male workers aged 20–25 in 2020 are expected to have replacement rates of roughly 40 percent.
  - For women: the 60–65 age group have average projected replacement rates of roughly 40 percent, while females aged 20–25 are expected to have replacement rates of 30 percent at retirement.

### III. Causes of low and declining replacement rates
- Low contribution rate:
  - Effective contribution rate is lower than most OECD countries. Contribution rates were initially low to encourage transition from PAYG and have not been increased.
- Low contribution density:
  - Self-employment and worker turnover led to low contribution densities, especially among women.
  - Average probability a male worker contributes in a given month: 60 percent; for women: 50 percent.
  - Average contribution density for males retiring between 2017 and 2020 was 60 percent and 46 percent for females.
- Demographics:
  - Younger generations face longer life horizons: workers who will retire in 40 years will need to spread savings over an additional five years (four years for women) compared to those retiring today.
- Lower future expected returns:
  - Real interest rates have gradually declined since adoption of the defined-contribution system and are expected to remain low over the medium term.

### IV. Impact of COVID-related withdrawals on pensions and fiscal costs
- Aggregate withdrawal magnitude:
  - Three rounds of withdrawals amounted to 19 percent of 2020 GDP (by early-May 2021).
- Expected impact on current affiliates (in absence of compensating government support):
  - Average withdrawal is projected to result in a 7 percent decline in pension at retirement.
  - Three rounds of withdrawals are expected to reduce the self-financed portion of pensions of current affiliates by 21 percent, on average.
  - Buffering by the solidarity pillar increases government-funded pension supplements, yielding a reduction in total pensions of about 7 percent and a decline in the average expected replacement rate from 37 percent to 35 percent.
- Fiscal impacts of the solidarity pillar:
  - Withdrawals increase (i) the number of recipients of the pension supplement (as some pensioners fall into the lower 60 percent of the income distribution) and (ii) the amount received by each recipient.
  - Additional fiscal cost estimated at a net present value of about 6 percent of GDP in 2020.
  - Additional fiscal costs are expected to peak in 2060 between 0.09 and 0.17 percent of GDP depending on assumptions.
  - The increasing profile of additional fiscal costs would be equivalent, in net present value, to a one-off fiscal cost of between 3 and 6 percent of GDP in 2020.
  - Future increases in solidarity contributions would increase such a cost.

### V. Reform experiments and policy implications
- Types of reforms examined: increases in contribution rates, increases in retirement age, and improvements in contribution density; also combinations of reforms.
- Example reform package and projected outcomes (relative to post-withdrawal baseline):
  - Increase contribution rate to 16 percent.
  - Increase retirement age for men and women to 67.
  - Improve contribution density to 70 percent.
  - Projected result: raise the average expected replacement rate to 50 percent from 35 percent for the average worker.
  - For young people (who have more time to benefit from changes):
    - Expected replacement rate would increase to 70 percent for males aged 20–25.
    - Expected replacement rate would increase to close to 60 percent for females aged 20–25.
- Caveats and behavioral considerations:
  - Framework does not model behavioral responses. For example, increasing the contribution rate may induce lower contribution density via higher informality.
  - Some results should be interpreted as upper bounds on potential benefits.
  - Reforms may need to be phased-in to address political economy and labor market considerations.
- Distributional and cohort effects:
  - Changes in contribution rates and policies that increase contribution density have large positive effects on expected replacement rates of younger cohorts, leaving older cohorts virtually unchanged.
  - Changes to retirement age generate non-negligible improvements in expected replacement rates for all cohorts.
  - Improvements in system resiliency can be achieved by allowing periodic revisions to key parameters to reflect secular changes in life expectancy and global financial conditions.

### VI. Structure of the paper and annexes (content outline)
- Section II: state of the pension system under current legislation prior to withdrawals; benchmarks against Latin American and OECD peers; projections pre-withdrawals using supervisory agency data.
- Section III: description of withdrawals and quantification of impact on expected replacement rates and expected fiscal costs.
- Section IV: impact of different pension reform avenues on replacement rates and fiscal costs accounting for withdrawals.
- Section V: conclusions.
- Annexes:
  - Annex I. The Structure of the Chilean Pension System.
  - Annex II. Data and Methodology (A. International data; B. Chile specific data; C. Withdrawals; D. Projecting Pensions: Methodology and Assumptions).
  - Annex III. Expected Replacement rates under the alternative scenario.
  - Annex IV. Individual reform scenarios.

*Source: wpiea2021232-print-pdf - References..............................................................................................................*

### 1.25 percent). Let us first consider the implicit life expectancy of a retiree in 1981 (which

### wpiea2021232-print-pdf - 1.25 percent). Let us first consider the implicit life expectancy of a retiree in 1981 (which

### Replacement rates: historical exercises and drivers
- Assuming an implicit life expectancy of a retiree in 1981 (about 78 years old) and a conservative real return on assets of 6 percent, the replacement rate would be about 90 percent.
- A higher return of 8 percent (average over 1981-2019) would deliver replacement rates reaching 160 percent.
- Increase in life expectancy from the one expected in 1981 to the one expected in 2020 lowers the replacement rate from 90 percent to 62 percent (holding real return at 6 percent).
- Keeping life expectancy at its 1981 level but lowering the real return on assets to 4.15 percent causes the replacement rate to fall from 90 percent to 55 percent.
- Combining the increase in life expectancy with the fall in global safe interest rates lowers the replacement rate further, to 38 percent (this corresponds to the current young male entering the workforce today and contributing in the current pension system).
- The exercise illustrates the system at inception was well positioned to deliver adequate pension levels and replacement rates; changes in life expectancy, real interest rates, and unchanged system parameters (contribution rates and retirement age) contributed to worse outcomes over time.
- A hypothetical PAYG system, fully financed via contributions of current workers and with parameters consistent with the current system, would produce replacement rates for a hypothetical male (female) worker retiring today that are 5 (2) percentage points lower compared to those of the current system.

### Heterogeneity in projected pensions and replacement rates
- Roughly two thirds of the 55–60 cohort (600,000 people) were projected to retire with a pension below 10 UF prior to withdrawals.
- Approximately 50 percent of those 40–45 (approximately 600,000) were projected to retire with a pension below 10 UF prior to withdrawals.
- Mode of the distribution of projected pensions for currently young cohorts is between 10 and 15 UF, with a significant number in the 5–10 UF category.
- The distribution of expected replacement rates is bimodal for all cohorts, with replacement rates ranging from the low 20s to 100 percent; women have lower expected replacement rates than men, and there is a large difference between low and high wage earners.

### Solidarity pillar: role and magnitude
- Chile introduced a solidarity pillar in 2008 to supplement pensions of individuals with low self-financed pensions.
- Benefits from the solidarity pillar were (before the 2020 pension withdrawals) projected to account for approximately 30 percent of total expected average pension at retirement for men and 50 percent for women prior to withdrawals.
- The solidarity pillar increases the expected average replacement rate for those who contributed to the pension system in 2020 by almost 15 percentage points for men, and by close to 20 percentage points for women.
- For some retirees in the lower 60 percent of the income distribution, the solidarity pillar can represent close to 100 percent of their pension.

### Fiscal footprint and projections
- By 2020, fiscal costs associated with the pension system were approximately 2.2 percent of GDP, of which 1.1 percent stem from the solidarity pillar.
- Under baseline projections (assuming solidity-pillar parameters for PBS and APS remain unchanged), total fiscal costs were projected to increase gradually to 1.6 percent of GDP by 2060.
- In an alternative scenario where PBS and APS grow at the same rate as wages (1.25 percent per year in real terms), fiscal costs can be expected to reach 3 percent of GDP in 2060.
- Fiscal costs reflect authorities’ projections of future costs associated with the old PAYG system up to 2050.

### Description of the withdrawals (COVID-19 policy response)
- Two rounds of withdrawals were approved in July 2020 and December 2020; a third was approved in April 2021. As of August 2021, Congress was discussing a fourth round.
- Total withdrawals reached US$48 billion (or about 19 percent of GDP) by early-May 2021.
- Allowed amount for each withdrawal was generally 10 percent, but minimum and maximum withdrawals set at 35 UF and 150 UF respectively meant the share of assets withdrawn varied with balances.
- First and third withdrawals were tax-exempt; the second was exempt only for those who earned on average below a certain threshold.
- Official data up to February 2021: close to 10.5 million people withdrew money using the first or second withdrawals; of those, 30 percent depleted their accounts.
- Roughly 95 percent of all people with positive pension balances in June 2020 made use of at least one of the first two withdrawals; over 7 million people withdrew twice.
- The average amount withdrawn in each round was about US$2,000.
- Average individual took 40 percent of their account balance in the first withdrawal and slightly over 30 percent in the second withdrawal.
- Close to 3 million people depleted their account balance from the first two withdrawals; close to 3.8 million people exhausted their accounts through the three withdrawals.
- Of the US$48 billion withdrawn in the two rounds, about 55 percent came from individuals 46 and older.

### Distributional details of withdrawals
- Panel data (as of April/May 2021) on the first withdrawal (examples from Table 2, Panel A):
  - Less than $1,459 balance: 100% of balance allowed withdrawal; mean amount withdrawn $606; 806,055 people; 14.3%.
  - Between $1,459 - $14,594: allowed $1,459 or >10% of balance; mean amount withdrawn $1,461; 2,530,247 people; 45.0%.
  - Between $14,594 - $62,547: allowed 10% of balance; mean amount withdrawn $2,822; 1,980,438 people; 35.2%.
  - More than $62,547: allowed $6,255 (<10% of balance); mean amount withdrawn $6,226; 275,292 people; 4.9%.
  - No information: 32,117 people; 0.6%.
  - Total: 5,624,149 people; mean amount withdrawn $2,057; mean % of balance withdrawn 32.8%.
- Median person who withdrew was in the 36-45 year-old range; median age of those who withdrew tracked the median age of the overall system (~40 years).
- Withdrawal behavior by age (Table 3, as of April 27):
  - Below 25: 260,346 people; 4.6% of withdrawers; average amount $649; mean share of balance requested 88.4%.
  - 26-35: 1,452,607 people; 25.8%; average amount $1,369; mean share 44.6%.
  - 36-45: 1,439,173 people; 25.6%; average amount $1,935; mean share 28.4%.
  - 46-55: 1,375,072 people; 24.4%; average amount $2,572; mean share 23.3%.
  - 56-65: 915,760 people; 16.3%; average amount $2,913; mean share 21.1%.
  - 66+: 180,574 people; 3.2%; average amount $2,330; mean share 25.2%.
  - Total: 5,624,149 people; average amount $2,057; mean share of balance requested 32.8%.

### Short-term macroeconomic and financial effects of withdrawals
- Withdrawals mitigated falling household income and were associated with boosts in retail sales and consumption of durables.
- The measure was not well targeted: it reached the upper quintiles of the income distribution, who saw labor income gains in the first nine months of the pandemic.
- Transfers and withdrawals more than offset any income losses due to the prolonged pandemic into the first half of 2021.
- Many people transferred the withdrawn money into bank accounts rather than supporting spending.
- Pension withdrawals were a funding source across the income distribution, but stimulus can lead to inflationary pressure and deterioration of local financial conditions, generating greater exchange rate volatility and decoupling of long-term interest rates.

### Projected long-term impacts on pensions and replacement rates
- The self-funded portion of pensions is projected to decline on average by 19 percent for men and by almost 23 percent for women, with larger effects among older cohorts.
- Males in their 20s are projected to experience average reductions in self-funded pensions of 5 to 12 percent after withdrawals.
- Older cohorts experience reductions that can go up to over 60 percent, with higher numbers for elderly with lower balances who withdrew proportionally more.
- Reductions in self-funded pensions are smaller when weighted by assets (because larger balances faced proportionally smaller withdrawal shares).
- Replacement rates are projected to decline by about 3 percentage points after withdrawals for the average male worker, and by 1.5 percentage points for female workers.
- In absence of additional government support, replacement rates would fall by over 4 percentage points for men and by over 2 percentage points for women.
- For men, APS mitigates the adverse effect of withdrawals on expected replacement rate most for the 50-55 age group—APS dampens the adverse effect by 1.5 percentage points.
- For women, the additional impact of APS is largest for the 40-45 age group.

### Projected fiscal implications from increased APS and PBS take-up
- Under the baseline scenario, close to 270,000 additional people are projected to receive self-funded pensions below PMAS at retirement, making them eligible to APS after the withdrawals if they fall into the lower 60 percent of the income distribution (240,000 under the alternative scenario).
- Current APS recipients are expected to see an increase in APS due to the adverse effect of withdrawals on the self-funded portion of pensions, leading to:
  - An expected increase of 13 percent in the average supplement received by males (10 percent in the alternative scenario).
  - An expected increase of 7 percent in the average supplement received by females (5 percent in the alternative scenario).
- These effects lead to a gradual increase in fiscal costs as new cohorts with lower pension account balances retire and access additional APS payments.
- Additional fiscal costs stemming from the solidarity pillar peak around 2060 (Figure 12 panel C).

*Source: Authors’ calculations based on data from Superintendencia de Pensiones, OECD, Central Bank of Chile, and related in-text figures and tables contained in the source content.*

### 0.17 percent of GDP, (0.21 percent in the alternative scenario). The net present value of the

### wpiea2021232-print-pdf - 0.17 percent of GDP, (0.21 percent in the alternative scenario). The net present value of the

### Fiscal impact of withdrawals
- Additional fiscal costs from withdrawals peak at about 0.17 percent of GDP (0.21 percent in the alternative scenario).
- The net present value of the additional fiscal costs stands at about 6 percent of 2020 GDP.
- These fiscal costs represent an upper bound because some individuals falling below the PMAS line will not fall into the lower 60 percent of the income distribution and thus will not be eligible for solidarity pillar benefits.
- If withdrawals do not affect the income distribution (beneficiaries of the solidarity pillar remain unchanged), fiscal cost would:
  - Peak at about 0.09 percent of GDP under the baseline.
  - Amount to a net present value of 3 percent of GDP in 2020.
- Fiscal cost derived from withdrawals by current pensioners is expected to:
  - Peak at 0.045 percent of 2020 GDP in the next few years.
  - Be 0.025 percent of 2020 GDP if government support is only provided to pensioners in the lower 60 percent of the income distribution.
- Notes on scenarios and exclusions:
  - In the alternative scenario the number of additional people is lower than in the baseline because a higher PMAS gives more people a government supplement pre-withdrawal, making the post-withdrawal change smaller; a higher PBS implies larger post-withdrawal amounts, producing a slight increase in fiscal cost relative to baseline.
  - Pensioners with life annuities were excluded from the 1st and 2nd pension withdrawals; pensioners with programmed withdrawals were allowed to access all three withdrawals.
- Figures referenced: Figure 12 panels A–D summarize people affected and additional APS costs (Panel C: current affiliates; Panel D: pensioners), in thousands and percent of GDP respectively.  
- Source for calculations: Authors’ calculations.

### Tax revenue effects of withdrawals
- Self-funded pensions in Chile are taxable; reductions in self-funded pension balances reduce future tax collection.
- Under the assumption that Chile’s income tax structure remains constant over the next 40 years, the government would lose over USD 1.6 billion over 40 years (expressed in net present value 2020 terms).
- Foregone revenue would peak around 2060 at approximately 0.008 percent of GDP.
- Alternative quantification: computing tax losses from tax exemptions in the law gives similar results.
- If the three withdrawals had been fully taxable, tax collection would have increased by over USD 1.8 billion, or 0.7 percent of GDP.

### The March towards reform — policy context and recent changes
- System shortcomings: infrequent contributions, falling global returns on safe assets, relatively low mandatory contribution rate, and retirement age not catching up with demographics have dampened pensions.
- Solidarity pillar:
  - Introduced in the 2008 reform; largely increased in 2019.
  - At end-2008 the basic pension was 65,470 CLP, about 40 percent of the minimum wage at the time.
  - Extraordinary real increase in PBS of 10% in January 2017; increase by 50% announced at end-2019 will bring minimum guaranteed pension to 50% of the minimum wage when finalized in January 2022 for all pensioners.
- Grant per Child (Bono por Hijo):
  - Supplement equivalent to 10% of 18 times the minimum wage at time of birth plus the average net rate of return from date of birth until claim.
  - Paid 360,000 beneficiaries in October 2020, supplementing monthly pension on average by 9,796 CLP (about US$ 12.70).
- Demographic pressures:
  - Population eligible for pension expected to double from about 3 million in 2021 to 6 million in 2050, adding pressure on the solidarity pillar.
  - The solidarity pillar ensures almost everyone over age 65 receives a pension, close to 95%.
- Benefits of preserving and improving system:
  - AFP-managed financial assets provide a large stock of savings; about half of assets under management invested domestically.
  - Individual link between contributions and pension incentivizes employment and savings.

### Recent reform proposals and alternatives
- President Piñera’s March 2021 proposal (revamp of January 2020 proposal):
  - Increase contribution rate to 16 percent.
  - Expand coverage of the solidarity pillar from 60 percent to 80 percent of the population.
  - Additional 6 percent contribution paid by employer and managed by a public autonomous body:
    - 3 percent to employees’ individual pension savings.
    - 3 percent to a collective saving fund; earnings used to incentivize contributions via additional payments as years of contribution rise.
- Alternative prominent proposal: attribute all additional 6 percent to the solidarity pillar (no additional increase to individual pension savings) — would weaken ability to deliver adequate pensions absent large fiscal costs.
- Other proposals: introduce contributions for informal workers (a share of VAT payments or permit fees routed to individual pension accounts) to increase contribution density.

### Quantifying reform proposals — methodology and key results
- Analysis assumes immediate implementation and that reforms are implemented after the three withdrawals; results are upper bounds of potential benefits.
- Key levers to increase expected replacement rates: contribution rates, retirement age, and contribution density.
- Isoquant insights (Figure 14):
  - Population average 40 percent expected replacement rate achievable by:
    - Increasing female retirement age to 65 and contribution rate to 14.5 percent; or
    - Keeping current contribution rate and increasing retirement age to 69.5 for all.
  - Increasing contribution density from current levels (60 for men, 50 for women) to 70 percent reduces required increases in contribution rate and retirement age to reach target replacement rates.
  - For younger cohorts (aged 20–25), reaching expected replacement rate of 70 percent possible with contribution rate 18.5 percent and retirement age 70; with contribution density 70 percent the same 70 percent replacement reachable with retirement age 66.5.
- Impact of increasing mandatory contribution from 10 percent to 13 percent:
  - Increases average expected replacement rate from 35 to 37 percent.
  - For young affiliates (aged 20–25):
    - Males: expected replacement increases from 37 to 45 percent.
    - Females: expected replacement increases from 29 to 33 percent.
  - Even with contribution rate at 13 percent, expected replacement rates for youngest cohorts remain below 50 percent (OECD average), motivating larger increases.
- Combined reform example: contribution rate to 16 percent, retirement age to 67 (from 65 for men and 60 for women), contribution density to 70 percent (from 60 men, 50 women):
  - Expected replacement rate for young cohorts: 59 percent for women and 66 percent for men.
  - Population average expected replacement rate for current contributors reaches about 50 percent.
  - Combination reduces solidarity support needed; fiscal cost of the pension system expected to be 0.8 percent of GDP lower in 2060.
  - Even after equalizing retirement age and contribution density, gender inequality in expected replacement rates persists due to differences in life expectancy and accumulated assets.
- Isolated impacts:
  - Increasing contribution rate from 10 to 16 percent is the single most powerful independent measure for raising expected replacement rates (e.g., increases expected replacement for ages 20–25 from 34 to 45 percent; population average from 35 to 40 percent).
  - Increasing contribution density to 70 percent raises expected replacement for current affiliates from 35 to 37 percent (and for ages 20–25 from 34 to 38 percent).
  - Increasing retirement age to 67 increases expected average replacement rate by 2 percentage points (from 35 to 37 percent) in the model; effects dampened by assumed real wage growth of 1.25 percent.
- Caveats:
  - Policy analysis abstracts from potential unintended consequences (e.g., higher contribution requirements could increase informality); evidence suggests a 5 percentage point increase in contribution rates increases informality by 12.5 percent for men and 9.3 percent for women in Chilean estimates.

### Universal Basic Pension (UBP) — fiscal quantification
- Exercise studies UBP tied to the minimum wage and abstracts from eligibility criteria (thus likely overstating true cost).
- A UBP providing a fraction of the minimum wage to everyone of eligible pension age (65+ for males, 60+ for females) would cost anywhere between 2.5 and 4 percent of GDP by 2050 (5 to 8 percent of today’s GDP), depending on parameter choices.

*Source: Authors’ calculations.*

### introduction of a universal pension tied to the minimum wage. To quantify the cost of this

### introduction of a universal pension tied to the minimum wage. To quantify the cost of this reform, independently of its source of financing, we consider two scenarios:

### Universal Basic Pension (UBP) scenarios and cost estimates
- Two scenarios considered:
  - UBP set today at half real minimum wage and remains constant in real terms.
  - UBP set at 50 percent the real minimum wage and then grows at the same rate as overall wages (assuming a real increase of 1.25 percent).
- A UBP of half the minimum wage is roughly equivalent to paying today’s PBS value to every retiree; this pension level falls slightly below the poverty line.
- Fiscal cost estimates for a UBP of half the minimum wage:
  - 5.6 UF or about US$ 230 (Using the exchange rate at end-February 2021).
  - Approximately 2.5 percent of GDP each year (projected over time, as ageing increases old-age population but the minimum pension declines with respect to GDP per capita).
  - This would amount to close to 5 percent of GDP in 2020.
  - If the basic pension is assumed to increase with wages, the cost of a UBP of half the minimum wage would increase to over 3.5 percent of GDP by 2050 (close to 8 percent of today’s GDP).

### Key pension system findings and projections
- Replacement rates and adequacy:
  - Replacement rates in Chile are low by international standards and are expected to fall further, especially after the three rounds of withdrawals in response to the pandemic.
  - The system’s low replacement rates are mostly explained by low contribution rates.
- Impact of pension withdrawals (three rounds through August 2021):
  - Self-funded pensions are projected to fall, on average, by 21 percent.
  - Total pensions (self-funded plus solidarity supplement) decline by 7 percent due to increased solidarity supplements cushioning the fall.
  - Fiscal costs: withdrawals will gradually increase government supplement payments relative to current levels.
    - At the peak (near 2060), withdrawals are expected to lead to an increase of 10 percent in fiscal costs relative to pre-withdrawal levels (or an annual 0.17 percent of GDP).
    - The net present value of the flow of additional costs stands at roughly 3 to 6 percent of 2020 GDP (depending on assumptions), but could be much more with increases in the public solidarity contribution.
  - Further rounds of withdrawals would accentuate these numbers.

### Reform simulations and fiscal implications
- Single-parameter reform:
  - Increasing the contribution rate by 6 percentage points devoted to the self-funded pension:
    - Raises the expected replacement rate for the average of all current affiliates to 40 percent from 35 percent.
    - Raises the expected replacement rate for 20-25 years old to 45 percent from 34 percent.
- Multi-parameter reform (example package):
  - Contribution rates to 16 percent, retirement age increased to 67, and contribution density to 70 percent:
    - Expected replacement rates for young people increase to 59 percent for females and 66 percent for males.
  - Such combined measures can achieve similar adequacy outcomes with more gradual changes and broader cohort impacts, easing the political economy of reform.
- Fiscal space implications:
  - Strengthening the self-funded portion reduces reliance on the solidarity pillar.
  - The example combination of measures noted above would entail a reduction in the fiscal cost of the system by 0.8 percent of GDP in 2060.
  - This fiscal space could be used to strengthen the solidarity component in a targeted way.

### Policy recommendations and governance
- Update system parameters periodically to adapt to changing demographics, global returns, and the labor market.
  - Suggestion to develop an automated system of updating key parameters (e.g., contribution rate and retirement age) at regular intervals, such as five or 10 years.
  - Specific institutions could be tasked with preparing analysis and proposals.
- Recent reform proposal noted as a positive step:
  - Proposes that the Social Security Advisory Council reviews demographic, economic and labor market trends every three years to suggest amendments to the system.
- Note on contribution density:
  - Contribution density is not a direct policy parameter; increasing it requires labor market, structural, and fiscal policies that encourage labor market participation and formal job creation.

### Modeling and key methodological assumptions relevant to projections
- UBP scenario framing:
  - Two UBP assumptions explicitly modeled: constant-real half minimum wage; half minimum wage growing with wages (real wage growth assumed 1.25 percent).
- Projection methodology highlights:
  - Projections begin in June 2020 data (one month before the 1st withdrawals) to construct a theoretical counterfactual.
  - Real returns on pension accounts assumed at 4.15% per year.
  - Life annuity rate assumed at 3.36%.
  - Real wage growth assumed at 1.25% per year.
  - Mandatory contribution rate is 10% of gross salary.
  - Contribution density assumed constant over time in baseline exercises.
- Withdrawals assumption for projection:
  - Individuals in each age-gender-account balance cell withdraw the maximum amount allowed for each withdrawal, implemented by the formula given in the source.
- Solidarity pillar parameters:
  - PBS and PMAS values up to 2022 set according to 2019 announcement; from 2022 onwards results presented under two assumptions: inflation indexation (2008 rule) and alternative with real growth of 1.25 percent.

*Source: Authors’ calculations and analysis as presented in the supplied content unit.*

### ANNEX III. EXPECTED REPLACEMENT RATES UNDER THE ALTERNATIVE SCENARIO

### ANNEX III. EXPECTED REPLACEMENT RATES UNDER THE ALTERNATIVE SCENARIO

### Alternative scenario: assumptions and aggregate outcome
- Assumption: the basic solidarity pension (PBS) and the threshold that defines the cutoff for receiving support (PMAS) increase in real terms by 1.25% per year — the same as the assumption for real wages.
- Effect: the fall in expected replacement rates due to an aging population and real wage increases above the rate of increase for the PBS and PMAS is neutralized.
- Outcome for young cohorts:
  - For young males in the system: projected replacement rate at retirement is expected to increase by 10 percentage points, from about 39 to 50 percent.
  - For young females in the system: projected replacement rate at retirement is expected to increase by 10 percentage points, from about 30 to 42 percent.
- Mechanism: the rise is driven by the increase in the government supplement received by affiliates, leading to a higher fiscal cost of the system in the medium-term.

### Figures referenced
- Figure A3.1 (Panel A): Male expected replacement rate at retirement by current age, pre and post withdrawals.
- Figure A3.1 (Panel B): Female expected replacement rate at retirement by current age, pre and post withdrawals.
- Source for figures: Authors’ calculations.

### ANNEX IV. INDIVIDUAL REFORM SCENARIOS

### Reform: Increase mandatory contribution rate from 10% to 16%
- Effect on self-financed balance: increases the self-financed balance of workers at retirement by about 59% for the cohort currently between the ages of 20 and 25 in the system.
- Effect on expected replacement rates for those aged 20–25:
  - Overall expected replacement rate: 46 percent.
  - Men: 53 percent.
  - Females: 36 percent.
- Fiscal impact: the higher mandatory contribution rate is expected to gradually lower the fiscal cost of the pension system, reducing the cost by 0.3 percentage points of GDP by 2060.
- Distributional note: benefits younger workers more (they have more time to contribute); impact smaller for affiliates closer to retirement but still increases their individual balances at retirement.
- Mechanism: greater share of wages deposited in individual pension accounts increases accumulated savings and reduces the share subsidized by the government under the complementary supplement formula.

### Reform: Increase average contribution density to 70%
- Baseline contribution density: current average contribution density of males about 60 percent, and females slightly below 50 percent.
- Effect on expected replacement rates for young cohorts:
  - Young males: increase from about 37 percent to 41 percent.
  - Young females: increase from about 29 percent to 34 percent.
- Fiscal impact: would lower the fiscal cost for the government by 0.10 percent in 2060.
- Caveat: further improvements occur if contribution density increases above 70 percent.
- Policy note: increasing contribution density is intrinsically linked to informality and is not a standard pension reform — better addressed through structural reforms that lead to sustained formal-sector employment.

### Reform: Increase retirement age to 67 for both sexes
- Current retirement ages: 60 for females and 65 for males.
- Effect on expected replacement rates:
  - Males: increase by 2 percentage points on average.
  - Females: increase by 3 percentage points on average.
- Mechanisms:
  - More working years increase time to contribute, boosting self-financed pensions.
  - Fewer years in retirement reduce the period over which savings are spread.
- Distributional and behavioral notes:
  - Self-financed component for women makes up only 35 percent of their total pension payments, limiting the overall boost to their expected replacement rates.
  - In the model, real wage growth assumed at 1.25 percent dampens the increase in expected replacement rates because pensions and final wages both rise.
  - Even with equalized retirement ages, expected replacement rates remain unequal due to assumed lower contribution density (50 for women and 60 for men) and longer female life expectancy (stretching pensions further).
  - A 7-year increase in women’s retirement age has a particularly significant impact on reducing the fiscal cost, since the majority of solidarity-supported pensioners are women.
- Figures referenced:
  - Figure A4.3 (Panel A): Increase in self-financed pension due to increasing retirement age to 67.
  - Figure A4.3 (Panel B): Expected replacement rate at retirement by age cohorts, increasing retirement age to 67.
  - Figure A4.3 (Panel C): Fiscal cost in % of GDP of increasing retirement age to 67.
- Source for figures: Authors’ calculations.

*Source: wpiea2021232-print-pdf - ANNEX III. EXPECTED REPLACEMENT RATES UNDER THE ALTERNATIVE SCENARIO*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021232-print-pdf.pdf_