## Annex I. Replacement rates ## Source details **Canonical URL:** [Annex I. Replacement rates](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024053-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024053-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024053-print-pdf.pdf.json) --- ### Key findings and context - The paper assesses outcomes of the 2019 and 2020 pension reforms in Mexico, discusses design inefficiencies, and offers policy options to improve pension system design. - Pension system fragmentation across multiple schemes: IMSS, AFORE, ISSSTE, PEMEX, CFE, ISSFAM, and others. - High informality and low active contribution coverage: - 57 percent of employment in 2021 was outside the formal sector. - ILO estimates active coverage in Mexico was 41 percent, compared to 90 percent for high income countries, 71 percent for other upper-middle income countries, and 47 percent for Latin America and the Caribbean. - On average only 23 percent of the working age population contributed to the IMSS in 2020. - Demographics and spending: - In 2022 there were 12 people over the age of 65 for every 100 people of working age (15-64). - Pension spending: 3.6 percent of GDP on public pensions in 2019 (up from 1.3 percent in 2000); OECD average was 7.7 percent. - Pension spending reached 4.5 percent of GDP in 2022. - Major schemes collectively spent 4.5 percent of GDP in 2022: - IMSS: 2.3 percent of GDP (about 1.7 percentage points are payments under the pre-97 regime) - Pensions for former civil servants: 1.0 percent of GDP - Basic pension: 0.8 percent of GDP - CFE and PEMEX: 0.4 percent of GDP - IMSS projects spending on the old regime to peak around 2045 at 1097 billion Pesos (expressed in 2020 prices). ### 2019 and 2020 reforms — main changes and intended effects - 2019 reform: - The means-tested social pension was replaced by a residency-based basic pension and the benefit level was raised significantly. - 2020 reform: - Contribution rates for the funded defined contribution scheme (DC) were raised. - Minimum pension benefits were increased and made dependent on career length, average earnings and age of retirement. - Minimum years of contribution requirements to receive an annuity were lowered from 25 years to 15 years, before being raised back to 20 years over time. - Intended outcomes: - Increased capacity to redistribute and prevent old-age poverty. - Once higher contribution rates fully take effect, capacity to smooth income should improve. ### Identified design inefficiencies and implications - Coexistence of pre-1997 PAYGO and post-1997 DC regimes creates very different benefits for similar careers; anyone with a contribution before 1997 can retire under the old PAYGO rules. - Long, incomplete transition means the post-1997 DC system has never been fully in place for any retired generation; transition to full effect spans decades. - The new minimum pension mechanism produces volatile and erratic outcomes across successive cohorts, depending on: - Years of contributions, - Past earnings (revalorized using prices), - Age and year of retirement, - Discrete wage-bracket thresholds. - Perverse short- and long-run incentives: - With rising minimum years of contribution requirements, benefits may fall for subsequent cohorts and may even decline if an individual works one extra year in the short run. - In many estimations, the discounted value of future pension benefits from working a year longer (net of contributions paid) is less than the discounted value of early retirement. - Fiscal and policy trade-offs: - Raised benefits close to or exceeding old PAYGO promises substantially increase fiscal cost. - Increased spending limits room for other social, health, and education policy expansions. ### Pattern of replacement-rate outcomes across cohorts (author’s projected sequence) - For average-wage earners, retiring at 65, projected sequence of outcomes: 1) Replacement rates were significantly increased. 2) For any career of less than 20 years (blue line), because minimum years of contributions with the introduction of the new minimum pensions, with some differences depending on career length. This is the results of loosened eligibility criteria and raised benefit levels... requirements rise again, benefits will fall rapidly from cohort to cohort and will eventually not be awarded to later cohorts... 3) For careers of more than 20 years (yellow line), with positive real wage growth, replacement rates will fall because of price indexation of minimum pensions... 4) With rising minimum years of contributions requirements, subsequent generations with less than 25 years of contributions (orange line) will have an occasionally falling minimum pension... 5) Price indexation will again slowly erode replacement rates for all subsequent generations... 6) But because of real wage growth and the price indexation of the wage brackets, over time, subsequent generations might move into higher wage brackets (orange line), which will lead to higher minimum pensions... 7) However, depending on career length, because past earnings are revalorized using prices, for a similar earnings profile, the jump into the higher wage bracket does not happen at the same time... 8) Eventually, the erosion of replacement rates (because of price indexation of minimum pensions) will cause own pension entitlements of those with longer careers to exceed the minimum pension (yellow line). ### Consequences for incentives, fairness, and predictability - Combination of discrete brackets, price revalorization of past earnings, and changing minimum contribution rules creates: - Erratic retirement outcomes for individuals with only small differences in earnings histories or contribution timing. - Reduced fairness across cohorts and similar-career individuals. - Weak or perverse work incentives in many scenarios. - Increased unpredictability of benefits and potential erosion of public trust in the pension system. ### Author’s overarching policy implication - The challenge for Mexico is to iron out design inefficiencies introduced or left by the reforms while making fiscal space for other social spending needs; reforms improved redistribution and poverty prevention but increased fiscal cost and did not address fragmentation. ### Administrative unit governing pension calculations — Unidad de Medida y Actualización (UMA) - UMA replaced the minimum wage to adjust most social benefits, including pensions in 2016. - Prior to 2016 the minimum wage was used to: - index benefits, - determine accrual in the pay-as-you-go pension system, and - determine government subsidies to private accounts and minimum pensions. - Since 2016 social benefits have been decoupled from the minimum wage and adjusted using the UMA. - UMA started at the level of the minimum wage in 2016 and is indexed yearly with inflation of the previous year. - The UMA and the minimum wage diverged quickly after 2016, with the minimum wage increasing significantly compared to inflation, averaging a real increase of around 10 percent per year since 2016. - UMA is used to determine: - the accrual rate in the PAYGO system, - the government contributions to the funded DC system, - the initial minimum pension, and - indexation of both the minimum pension and the basic pension. ### Changes to minimum pension design and levels (2021 onward) - Before 2021 there was only one minimum pension of almost 40 thousand pesos per year (27 percent of the average wage), to be taken at the statutory retirement age without differentiation between previous earnings levels. - From 2021 the minimum pension can be taken up to 5 years early, differs by career average earnings (revalorized with prices) and by career length. - The lowest minimum pension increased to 45 thousand pesos per year (28 percent of the average wage) for someone with 20 years of contributions or more in 2021. - For an average wage earner with the same career length (20 years) the minimum pension more than doubled to 85 thousand pesos per year (53 percent of the average wage). - The change led to a large inflow of minimum pension recipients: number of new retirees receiving a minimum pension increased almost 10-fold between 2020 and 2021. - The average benefit went up by 20 percent between 2020 and 2021, compared with 3.4 percent inflation in 2020 that regular indexation would have suggested. - New pension benefit levels went up by 67 percent. ### Fees and contribution requirements - Fees charged by AFOREs: - From 2022, fees cannot exceed average fees in Chile, Colombia and the United States. - OECD: average fees in both Chile and Colombia in 2019 amounted to 0.8 percent of assets. - In 2019 pension funds in Mexico charged 0.9 percent of assets in fees. - CNBC reported that in 2019 401k plans charged fees on average 0.45 percent of assets. - Minimum contribution requirement: - 2020 reform drastically reduced minimum years of contributions needed for an earnings-related pension under the funded regime. - Someone who does not meet the minimum years of contributions requirement will only receive a lump sum, not a permanent benefit. - Reduction to about 15 years would bring Mexico in line with a large group of OECD countries, but the planned increase to 20 years will still leave Mexico as the OECD country with the third highest requirement for earnings-related pensions. - 15 OECD countries have no minimum requirement (or a requirement of less than a year). - Mexico uses the same minimum contribution requirement for minimum pensions; once the transition is completed the minimum requirement in Mexico to receive a minimum pension will be in line with Italy, Hungary and Luxembourg for the basic minimum, while the requirement for the highest minimum pension will be in line with Costa Rica, Poland and Colombia. - Among OECD countries only France and Switzerland have fully pro-rated minimum pension benefits. - Critique of minimum contribution requirements (preserved textual arguments): - Minimum contribution requirements for funded defined-contribution pensions are unnecessary. - They are often claimed to provide incentives to contribute longer, but the effect is ambiguous: a requirement (e.g., 15 years) strongly incentivizes someone with 14 years close to retirement, but provides no incentive for someone with only 5 years of contributions. - For those years removed from retirement, incentives may be limited by myopia; if myopia requires mandating contributions, it is unclear why planning horizons are deemed sufficiently long late in life to ensure meeting minimum requirements. - High informality and options to avoid contributing weaken the effectiveness of minimum requirements. - Fiscal sustainability arguments do not apply to fully funded DC systems (unless government guarantees are given); in PAYGO systems minimum years implicitly redistribute from short careers to long careers, but that rationale is not relevant in funded DC systems. ### Coverage and contribution density implications - Minimum contribution requirement remains much too high, leaving many contributors without permanent pension benefits. - Around 40 percent of both men and women between the ages of 60 and 65 have 5 or fewer years of contributions. - With the exceptions of ages 60 and 61, less than 10 percent has reached between 20-25 years of contributions. - The average contribution density over the period 1997-2020 has been 40 percent. - To reach 20 years of contributions by age 65, the average contributor needs to start working at age 15. - Low contribution density caused by large informal sector and fragmented pension system with limited portability. - Lowering minimum years of contributions required from 25 to 20 years helps increase earnings-related pension benefit receipt but will still leave the majority of contributors without permanent pensions. ### Benefit levels and replacement rate dynamics (post-2019 and post-2020) - Combination of the 2019 and the 2020 reforms dramatically increased expected benefits of the Mexican pension system for post-1997 labor market entrants. - For an average wage earner with a 16-year career: - Under the pre-97 pay-as-you-go system in 2022 the combination of the basic pension, the earnings-related part of the old system and a minimum pension topping the pension up to the minimum wage level results in relatively stable replacement rates of about 50 percent in 2022 and 2042 and 2062. - Projected replacement rates for the pre-2019 DC system were much lower but equally stable, slowly falling because of rising life expectancy and falling government subsidies. - The new post-2020 system leads to highly different replacement rates: in 2022 the replacement rate is made up of the basic pension, entitlements derived from own contributions, employer contributions and government contributions and finally a minimum pension. Over time the basic pension and the rising employer contributions start to dominate. The minimum pension will not be received by those retiring in 2042 and 2062, since by then the minimum requirement will be 20 years of contributions. Replacement rates in 2022 will be similar to the pay-as-you-go system but will fall for later generations, though they will still be well above the pre-2019 DC system. - For longer careers under post-2020 rules (24- or 32-year careers): - An individual with a 24- or 32-year career would be eligible for a minimum pension now and in the future. - Own entitlements would eat into the minimum pension quicker for a longer career. - Since minimum pensions are indexed to prices, replacement rates expressed as share of last earned wage will fall; for higher earnings levels this might lead to not being eligible for a minimum pension anymore. - Sensitivity to real wage growth and real returns: - Higher real wage growth leads to a faster decline in replacement rates for those receiving the basic pension and a minimum pension because both are indexed with prices; the decline is almost entirely made up of a smaller replacement share of the basic pension and of the minimum pension top up. - Own pension part declines less under higher real wage growth because the only negative effect comes from a smaller difference between real wage growth and real returns. - Higher real returns have no effect on replacement rates for those with minimum pensions, but increase the self-financed share of replacement; high enough returns could push the own pension of an average wage earner with a 32-year career over the minimum pension benefit. ### Modeling assumptions (Box 2) - World Economic Outlook numbers are used until 2025, thereafter 2 percent inflation and 2 percent real wage growth. - Nominal returns follow 3 percent plus inflation from 2021 onwards. - The minimum wage follows average wage growth. - UMA is indexed with prices. - AFORE accounts are converted into phased withdrawals based on life expectancy at the age of retirement. - Key parametric mechanics: - Price indexation of most brackets determining government contributions combined with the minimum wage constituting its own bracket causes slow absorption of subsequent brackets into the minimum wage bracket. - Same mechanism gradually pushes people into higher minimum pension brackets because of real wage growth; extent depends on individual wage curve steepness and career length. - Threshold effects can be smoothed using more continuous functions instead of discrete brackets. - If real minimum wage growth keeps its current pace (i.e., 10 percent per year since 2016) it would mean that by the time the reform has been fully implemented the brackets have fully disappeared and only the minimum wage bracket is left. - Example bracket anomaly: in 2025 the government will contribute 10.75 Pesos a day indexed to reflect inflation for someone earning a minimum wage; someone earning one Peso more will receive only 7 Pesos because they fall in the 3.01-3.50 UMA bracket. - Wage and minimum pension bracket dynamics: - Reference wage brackets for minimum pensions drift over time in relative terms due to price indexation and real wage growth. - Example: someone with price uprated career average earnings equal to 75 percent of the average wage falls in the 3-4 UMA minimum pension bracket; assuming a 2 percent real wage increase the same person would fall in the 4-5 UMA minimum pension bracket by 2030. - Lowest minimum pension bracket currently defined as revalorized career average earnings between 1 minimum wage and 2 UMA; if the minimum wage keeps growing in real terms it will soon exceed 2 UMA. - Average revalorized minimum wages fall below the lowest bracket; consequence: those who earned a minimum wage for a full career could technically be ineligible for a minimum pension because no minimum pension bracket is defined for average career earnings less than the current minimum wage. - Early minimum pension — incentives and actuarial neutrality: - Actuarial neutrality for a minimum pension is not well-defined because the minimum pension top-up depends on potential future own contributions. - Example: retiring one year early for someone with a 23-year career at the average wage compared to working and contributing one more year (24-year career) results in a 1 percent lower pension until the generation retiring in 2027; afterward the penalty increases as 23-year careers fall into lower career length brackets. - For a 5-year early retirement the penalty eventually will be very high since no minimum pension eligibility exists for 19-year careers from 2030 onward. - For the first few years after reform, early retirement actually provides a bonus in terms of the minimum pension top up because own contributions and returns can partly pay for minimum pension benefits the individual was already eligible for the year before, creating incentives to retire early or work informally in the last years. - From 2028 onwards early retirement leads to small penalties with spikes due to wage bracket interactions; in the very long run penalties can turn into bonuses once minimum pension top ups shrink and DC savings rise because of contributions, positive real returns and positive real wage growth. - OECD (2017) calculated actuarial neutrality on existing entitlements implies a penalty between 5-6 percent for Mexico at the retirement age and that working one year more leads to a bonus of 7.4 percent. - The penalty on the full minimum pension for retiring 1 year earlier ranges from 1.0 percent to 1.4 percent within each wage and career length bracket. - If retiring one year earlier also leads to ending up in two career length brackets lower (each bracket is 25 weeks) the penalty ranges from 6.0-7.5 percent. - Lifetime benefits and retirement incentives: - Total expected benefits received from the time of retirement (discounted with inflation) for 19-24 year careers: - Start around 10 times the last earned wage in 2020. - Fall gradually to 6 times the last earned wage around 2080. - For almost all generations, individuals retiring at the normal retirement age receive the lowest lifetime benefits net of contributions made during the last 5 years before the normal retirement age. - For most cohorts, retiring 4 years early is the optimal decision since this is the first year it is possible to retire on a minimum pension with 20 years of contributions. - Establishing actuarial neutral penalties for a benefit that negatively depends on entitlements of potential future own contributions is impossible under current design. ### Conclusions and policy options (selected) - High-level conclusions: - The 2019 and 2020 reforms smooth differences between pre- and post-1997 entrants but introduce fiscal, equity, and incentive issues. - Rapid increases in basic pension levels risk crowding out other urgent spending needs. - Current design leads to widely different outcomes for relatively similar careers and weak incentives to work longer due to low penalties on early minimum pensions, limited differentiation by career length, and shifting minimum years of contribution requirements. - Policy options proposed: - Limit the increase of the basic pension while maintaining benefit coverage. Close to 100 percent coverage has only recently been achieved and should be maintained. - Abolish minimum contribution requirement for funded pensions. Implement a minimum balance requirement for phased withdrawals and annuities instead; automate pension payments to reduce administrative costs and lower the minimum balance threshold. - Abolish the minimum wage as a separate government contribution bracket and as a lower bound for the minimum pension. Instead express the minimum wage in terms of UMA like all other earnings levels and have an open-ended lowest minimum pension bracket. - Make redistribution within the pension system transparent and intentional. The system mixes redistribution through the basic pension, government subsidized contributions and elaborate minimum pensions; redistribution through minimum pensions is opaque and leads to unintended outcomes. ### Annex IV — Reform options (high-level findings and simulated options) - Reform rationale: - Minimum pensions often serve to reward long careers of lower earners but career length requirements create arbitrary and inequitable cut offs. - The only fully equitable minimum pension would be fully pro-rated as in France and Switzerland. - In Mexico low earners are already subsidized through government contributions and the basic pension; additional redistribution via minimum pensions is unlikely to add much value and may have negative equity consequences. - Common recommendation across options: - Early retirement on a minimum pension should be abolished and minimum pension benefit levels should remain price indexed. - Offer possibility to defer the minimum pension after the retirement age; determine the minimum pension top up at the normal retirement age and actuarially adjust it irrespective of actual additional contributions. - Option 1: Phase out the minimum pension - Proposal: Phase out the minimum pension (example: fixing the minimum pension levels in nominal terms). - Intended effect: Limit duration of design inefficiencies even if transition inefficiencies remain. - Simulated scenarios: "Half average wage" and "Average wage". - Simulation assumption: keep minimum pension levels constant in nominal terms. - Option 2: Abolish differentiation between career earnings brackets - Proposal: Maintain only the current lowest career earnings bracket and apply this minimum irrespective of career earnings. - Implementation: Gradually let higher earnings brackets converge to the lowest bracket (e.g., index only the lowest bracket while keeping higher brackets constant in nominal terms) or make minimum pension fully proportional to career length and remove discrete steps. - Simulated scenarios: "Half average wage" and "Average wage". - Simulation assumption: let earnings brackets converge to lowest earnings bracket over 10 years. - Option 3: Eliminate threshold effects (express minimum pension as continuous function) - Proposal: Express the minimum pension as a continuous function of average past earnings revalorized with economy-wide average wage growth and career length (or proxied by DC balances/DC annuities). - Calibration example: - New minimum pension is 50 percent of lowest minimum pension, withdrawn at a 50 percent rate against own pension income. - Gradual conversion to new rules over 10 years. - Career length requirements frozen in 2024. - Simulated scenarios: "Half average wage" and "Average wage". - Option 4: Index earnings brackets with wage growth - Proposal: Index earnings brackets with average wage growth after 2024. - Effect: Ensures under stable macro circumstances similar careers receive similar minimum pensions and lowers cost over time; can be combined with revalorizing past earnings with average wage growth (but raises cost). - Simulated scenarios: "Half average wage" and "Average wage". - Simulation assumption: Brackets indexed with average wage growth after 2024; career average earnings gradually converge to average wage uprated earnings over 10 years. - Actuarial neutrality analytical findings (equations preserved in source): - For the full pension to be actuarially neutral the DC part and the minimum pension top up need to both be actuarially neutral independently. - Actuarial neutrality condition examples and decompositions are documented, showing the actuarially neutral level of the minimum pension differs for working and non-working people and that behavioral responses to work incentives complicate neutral calibration. - Simulations and caveats: - Annex simulates reform options for informative purposes only. - Exact calibration should be investigated on a micro level (distributional effects) and macro level (spending effects). - Simulated scenario categories: "Half average wage" and "Average wage". - Time horizons and calibrations used in simulations: - Convergence of brackets or rules over 10 years. - Career length requirements frozen in 2024. - Indexing brackets with average wage growth after 2024. - New minimum pension expressed as 50 percent of lowest minimum pension with a 50 percent withdrawal rate against own pension income. *Source: wpiea2024053-print-pdf — Annex I, Box 1, Box 2, 2031 section, and Annex IV as provided in the source content.* ### Annex I. Replacement rates ............................................................................................. ### Annex I. Replacement rates ### Key findings and context - The paper assesses the likely outcomes of the 2019 and 2020 pension reforms in Mexico, discusses design inefficiencies, and offers policy options to improve pension system design. - Mexico’s pension system is fragmented across multiple schemes (IMSS, AFORE, ISSSTE, PEMEX, CFE, ISSFAM, and others). - High informality and low active contribution coverage: - 57 percent of employment in 2021 was outside the formal sector. - ILO estimates active coverage in Mexico was 41 percent, compared to 90 percent for high income countries, 71 percent for other upper-middle income countries, and 47 percent for Latin America and the Caribbean. - On average only 23 percent of the working age population contributed to the IMSS in 2020. - Demographics and spending: - In 2022 there were 12 people over the age of 65 for every 100 people of working age (15-64). - Pension spending: 3.6 percent of GDP on public pensions in 2019 (up from 1.3 percent in 2000); OECD average was 7.7 percent. - Pension spending reached 4.5 percent of GDP in 2022. - Major schemes collectively spent 4.5 percent of GDP in 2022: - IMSS: 2.3 percent of GDP (about 1.7 percentage points are payments under the pre-97 regime) - Pensions for former civil servants: 1.0 percent of GDP - Basic pension: 0.8 percent of GDP - CFE and PEMEX: 0.4 percent of GDP - IMSS projects spending on the old regime to peak around 2045 at 1097 billion Pesos (expressed in 2020 prices). ### 2019 and 2020 reforms — main changes and intended effects - 2019 reform: - The means-tested social pension was replaced by a residency-based basic pension and the benefit level was raised significantly. - 2020 reform: - Contribution rates for the funded defined contribution scheme (DC) were raised. - Minimum pension benefits were increased and made dependent on career length, average earnings and age of retirement. - Minimum years of contribution requirements to receive an annuity were lowered from 25 years to 15 years, before being raised back to 20 years over time. - Intended outcomes: - The reforms increased the system’s capacity to redistribute and prevent old-age poverty. - Once higher contribution rates fully take effect, the system’s capacity to smooth income should improve. ### Identified design inefficiencies and implications - Coexistence of pre-1997 PAYGO and post-1997 DC regimes creates very different benefits for similar careers; anyone with a contribution before 1997 can retire under the old PAYGO rules. - Long, incomplete transition means the post-1997 DC system has never been fully in place for any retired generation; transition to full effect spans decades. - The new minimum pension mechanism produces volatile and erratic outcomes across successive cohorts, depending on: - Years of contributions, - Past earnings (revalorized using prices), - Age and year of retirement, - Discrete wage-bracket thresholds. - Perverse short- and long-run incentives: - With rising minimum years of contribution requirements, benefits may fall for subsequent cohorts and may even decline if an individual works one extra year in the short run. - In many estimations, the discounted value of future pension benefits from working a year longer (net of contributions paid) is less than the discounted value of early retirement. - Fiscal and policy trade-offs: - Raised benefits close to or exceeding old PAYGO promises substantially increase fiscal cost. - Increased spending limits room for other social, health, and education policy expansions. ### Pattern of replacement-rate outcomes across cohorts (author’s projected sequence) - For average-wage earners, retiring at 65: 1) Replacement rates were significantly increased. 2) For any career of less than 20 years (blue line), because minimum years of contributions with the introduction of the new minimum pensions, with some differences depending on career length. This is the results of loosened eligibility criteria and raised benefit levels... requirements rise again, benefits will fall rapidly from cohort to cohort and will eventually not be awarded to later cohorts... 3) For careers of more than 20 years (yellow line), with positive real wage growth, replacement rates will fall because of price indexation of minimum pensions... 4) With rising minimum years of contributions requirements, subsequent generations with less than 25 years of contributions (orange line) will have an occasionally falling minimum pension... 5) Price indexation will again slowly erode replacement rates for all subsequent generations... 6) But because of real wage growth and the price indexation of the wage brackets, over time, subsequent generations might move into higher wage brackets (orange line), which will lead to higher minimum pensions... 7) However, depending on career length, because past earnings are revalorized using prices, for a similar earnings profile, the jump into the higher wage bracket does not happen at the same time... 8) Eventually, the erosion of replacement rates (because of price indexation of minimum pensions) will cause own pension entitlements of those with longer careers to exceed the minimum pension (yellow line). ### Consequences for incentives, fairness, and predictability - The combination of discrete brackets, price revalorization of past earnings, and changing minimum contribution rules creates: - Erratic retirement outcomes for individuals with only small differences in earnings histories or contribution timing. - Reduced fairness across cohorts and similar-career individuals. - Weak or perverse work incentives in many scenarios. - Increased unpredictability of benefits and potential erosion of public trust in the pension system. ### Author’s overarching policy implication - The challenge for Mexico is to iron out design inefficiencies introduced or left by the reforms while making fiscal space for other social spending needs; reforms improved redistribution and poverty prevention but increased fiscal cost and did not address fragmentation. *Source: wpiea2024053-print-pdf - Annex I. Replacement rates* ### Box 1. Administrative unit governing pension calculations ### Box 1. Administrative unit governing pension calculations ### Role and reform of the Unidad de Medida y Actualización (UMA) - The Unidad de Medida y Actualización (UMA) replaced the minimum wage to adjust most social benefits, including pensions in 2016. - Prior to 2016, the minimum wage was used to: - index benefits, - determine accrual in the pay-as-you-go pension system, and - determine government subsidies to private accounts and minimum pensions. - Before 2016, minimum wage increases followed inflation closely, yielding close to price indexation. - Since 2016 social benefits have been decoupled from the minimum wage and adjusted using the UMA. - The UMA started out at the level of the minimum wage in 2016 and is indexed yearly with inflation of the previous year. - The UMA and the minimum wage diverged quickly after 2016, with the minimum wage increasing significantly compared to inflation, averaging a real increase of around 10 percent per year since 2016. - Source for figure referenced: WEO, Mexican authorities. ### Importance of the UMA for pension calculations - The UMA is used to determine: - the accrual rate in the PAYGO system, - the government contributions to the funded DC system, - the initial minimum pension, and - indexation of both the minimum pension and the basic pension. _Italic: Source — content from wpiea2024053-print-pdf (Box 1)._ ### 2031. This both holds for the regular pension derived from the DC system as well as the minimum pension. ### wpiea2024053-print-pdf - 2031. This both holds for the regular pension derived from the DC system as well as the minimum pension. ### Changes to minimum pension design and levels (2021 onward) - Before 2021 there was only one minimum pension of almost 40 thousand pesos per year (27 percent of the average wage), to be taken at the statutory retirement age without differentiation between previous earnings levels. - From 2021 the minimum pension can be taken up to 5 years early, differs by career average earnings (revalorized with prices) and by career length. - The lowest minimum pension increased to 45 thousand pesos per year (28 percent of the average wage) for someone with 20 years of contributions or more in 2021. - For an average wage earner with the same career length (20 years) the minimum pension more than doubled to 85 thousand pesos per year (53 percent of the average wage). - The change in minimum pension rules led to a large inflow of minimum pension recipients: the number of new retirees receiving a minimum pension increased almost 10-fold between 2020 and 2021. - The average benefit went up by 20 percent between 2020 and 2021, compared with 3.4 percent inflation in 2020 that regular indexation would have suggested. - New pension benefit levels went up by 67 percent. ### Fees charged by AFOREs (fee cap introduced 2022) - From 2022, fees cannot exceed average fees in Chile, Colombia and the United States. - According to the OECD, average fees in both Chile and Colombia in 2019 amounted to 0.8 percent of assets. - In 2019 pension funds in Mexico charged 0.9 percent of assets in fees. - CNBC reported that in 2019 401k plans charged fees on average 0.45 percent of assets.18 ### Minimum contribution requirement: reform, international comparisons, and critique - The 2020 reform drastically reduced minimum years of contributions needed for an earnings-related pension under the funded regime. - Someone who does not meet the minimum years of contributions requirement will only receive a lump sum, not a permanent benefit. - The reduction to about 15 years would bring Mexico in line with a large group of OECD countries, but the planned increase to 20 years will still leave Mexico as the OECD country with the third highest requirement for earnings-related pensions. - 15 OECD countries have no minimum requirement (or a requirement of less than a year). - Mexico uses the same minimum contribution requirement for minimum pensions; once the transition is completed the minimum requirement in Mexico to receive a minimum pension will be in line with Italy, Hungary and Luxembourg for the basic minimum, while the requirement for the highest minimum pension will be in line with Costa Rica, Poland and Colombia. - Among OECD countries only France and Switzerland have fully pro-rated minimum pension benefits. - Many countries have higher minimum contribution requirements for minimum pensions than for regular earnings-related pensions (example: Belgium has the highest requirement for a full minimum pension but no minimum contribution requirement for a regular earnings-related pension at all). Critique of minimum contribution requirements (textual arguments preserved): - Minimum contribution requirements for funded defined-contribution pensions are unnecessary. - They are often claimed to provide incentives to contribute longer, but the effect is ambiguous: a requirement (e.g., 15 years) strongly incentivizes someone with 14 years close to retirement, but provides no incentive for someone with only 5 years of contributions. - For those years removed from retirement, incentives may be limited by myopia; if myopia requires mandating contributions, it is unclear why planning horizons are deemed sufficiently long late in life to ensure meeting minimum requirements. - High informality and options to avoid contributing weaken the effectiveness of minimum requirements. - Fiscal sustainability arguments do not apply to fully funded DC systems (unless government guarantees are given); in PAYGO systems minimum years implicitly redistribute from short careers to long careers, but that rationale is not relevant in funded DC systems. ### Coverage and contribution density implications - The minimum contribution requirement remains much too high, leaving many contributors without permanent pension benefits. - Around 40 percent of both men and women between the ages of 60 and 65 have 5 or fewer years of contributions. - With the exceptions of ages 60 and 61, less than 10 percent has reached between 20-25 years of contributions. - The average contribution density over the period 1997-2020 has been 40 percent. - To reach 20 years of contributions by age 65, the average contributor needs to start working at age 15. - One reason behind low contribution density is a large informal sector and fragmented pension system with limited portability, leaving many people with insufficient entitlements in multiple schemes. - Lowering the minimum years of contributions required from 25 to 20 years helps increase earnings-related pension benefit receipt but will still leave the majority of contributors without permanent pensions. ### Benefit levels and replacement rate dynamics (post-2019 and post-2020 reforms) - The combination of the 2019 and the 2020 reforms dramatically increased expected benefits of the Mexican pension system for post-1997 labor market entrants. - For an average wage earner with a 16-year career (Figure 17, first row): - Under the pre-97 pay-as-you-go system in 2022 the combination of the basic pension, the earnings-related part of the old system and a minimum pension topping the pension up to the minimum wage level results in relatively stable replacement rates of about 50 percent in 2022 and 2042 and 2062.19 - Projected replacement rates for the pre-2019 DC system were much lower but equally stable, slowly falling because of rising life expectancy and falling government subsidies.21 - The new post-2020 system leads to highly different replacement rates: in 2022 the replacement rate is made up of the basic pension, entitlements derived from own contributions, employer contributions and government contributions and finally a minimum pension. Over time the basic pension and the rising employer contributions start to dominate. The minimum pension will not be received by those retiring in 2042 and 2062, since by then the minimum requirement will be 20 years of contributions. Replacement rates in 2022 will be similar to the pay-as-you-go system but will fall for later generations, though they will still be well above the pre-2019 DC system. - For longer careers under post-2020 rules (24- or 32-year careers; Figure 17, second and third row): - An individual with a 24- or 32-year career would be eligible for a minimum pension now and in the future. - Own entitlements would eat into the minimum pension quicker for a longer career. - Since minimum pensions are indexed to prices, replacement rates expressed as share of last earned wage will fall; for higher earnings levels this might lead to not being eligible for a minimum pension anymore (see Annex I). - Sensitivity to real wage growth and real returns: - Higher real wage growth leads to a faster decline in replacement rates for those receiving the basic pension and a minimum pension because both are indexed with prices; the decline is almost entirely made up of a smaller replacement share of the basic pension and of the minimum pension top up. - Own pension part declines less under higher real wage growth because the only negative effect comes from a smaller difference between real wage growth and real returns. - Higher real returns have no effect on replacement rates for those with minimum pensions, but increase the self-financed share of replacement; high enough returns could push the own pension of an average wage earner with a 32-year career over the minimum pension benefit. *IMF WORKING PAPERS — An assessment of the 2019 and 2020 pension reforms in Mexico* ### Box 2. Modeling assumptions ### Box 2. Modeling assumptions ### Modeling assumptions - World Economic Outlook numbers are used until 2025, thereafter 2 percent inflation and 2 percent real wage growth. - Nominal returns follow 3 percent plus inflation from 2021 onwards. - The minimum wage follows average wage growth. - UMA is indexed with prices. - AFORE accounts are converted into phased withdrawals based on life expectancy at the age of retirement. ### Parametric choices — key findings and mechanics - Price indexation of most brackets determining government contributions combined with the minimum wage constituting its own bracket causes slow absorption of subsequent brackets into the minimum wage bracket. - The same mechanism gradually pushes people into higher minimum pension brackets because of real wage growth; extent depends on individual wage curve steepness and career length. - Threshold effects present in the Mexican pension design can be smoothed out using more continuous functions instead of current discrete brackets. - Since delinking pensions from the minimum wage there is a disconnect: brackets determining government contributions are linked to inflation while the minimum wage adjusts yearly in real terms. - Current practice: minimum wage increases in real terms. If real minimum wage growth keeps its current pace (i.e., 10 percent per year since 2016) it would mean that by the time the reform has been fully implemented the brackets have fully disappeared and only the minimum wage bracket is left, meaning only minimum wage earners receive a government contribution. - Example of bracket anomaly: in 2025 the government will contribute 10.75 Pesos a day indexed to reflect inflation for someone earning a minimum wage; someone earning one Peso more will receive only 7 Pesos because they fall in the 3.01-3.50 UMA bracket. - Note on sensitivity: Faster real minimum wage growth would speed up bracket disappearance; slower real minimum wage growth would slow the process. However, any real minimum wage growth would eventually lead to the disappearance of other brackets. ### Wage and minimum pension bracket dynamics - Reference wage brackets for minimum pensions drift over time in relative terms due to price indexation and real wage growth. - Current instance: someone with price uprated career average earnings equal to 75 percent of the average wage falls in the 3-4 UMA minimum pension bracket; assuming a 2 percent real wage increase in the future the same person would fall in the 4-5 UMA minimum pension bracket by 2030. - The lowest minimum pension bracket is currently defined as revalorized career average earnings between 1 minimum wage and 2 UMA. If the minimum wage keeps growing in real terms it will soon exceed 2 UMA. - Average revalorized minimum wages fall below the lowest bracket (i.e., price revalorization does not compensate past earnings for real minimum wage growth). Consequence: those who earned a minimum wage for a full career could technically be ineligible for a minimum pension because no minimum pension bracket is defined for average career earnings less than the current minimum wage. - Minimum wage earners will only become eligible for a minimum pension once their career average earnings exceed 2 UMA in the future. - Minimum pension changed from a single benefit to a benefit variable by wage level, career length and retirement age: - Before 2021: one minimum pension of almost 40 thousand pesos per year, taken at statutory retirement age without differentiation by previous earnings. - From 2021: minimum pension can be taken 5 years early, differs by career average earnings (uprated with prices) and by career length. Minimum pensions increase with career length and average earnings but fall for each year of early retirement. ### Minimum pension outcomes and replacement rate patterns - For average wage earners with a 24-year career, replacement rates: - Jump between 2020 and 2021 because of the introduction of the minimum pension for that career length. - Rise further for next generations because of the increase in the basic pension. - Slowly fall after 2024 because basic pension and minimum pension benefits are price-indexed (reducing their value relative to average wage) and minimum requirements for a minimum pension slowly rise. - Jump between 2037 and 2038 caused by average wage earners’ reference wage moving into the highest wage bracket due to bracket drift. - For careers of 26 years or more (always in highest career length bracket), generations earning half the average wage for all career lengths are projected to receive a minimum pension. After an initial jump, relative benefits of the minimum pension decline because of price indexation. - For the average wage, own entitlements of individuals with longer careers eventually will exceed the minimum pension level (desired outcome for fiscal containment), but sudden jumps into higher wage brackets are the result of flawed bracket design. - Shorter careers produce more volatile replacement rates: - Those with 14 years of service will not receive a minimum pension. - Only generations retiring between 2021 and 2030 with more than 15 years but less than 20 might receive a minimum pension. - As minimum years of contribution requirements rise, eligibility for minimum pensions shifts and becomes more transitory (e.g., 16-year career sufficient in 2021 but not after 2024). - Price revalorization of past earnings benefits individuals with steeper wage curves and makes minimum pension dynamics sensitive to real wage growth. Two individuals with the same lifetime average wage but differently sloped wage curves will have different price revalorized career average earnings and therefore different minimum pensions. ### Early minimum pension — incentives and actuarial neutrality - Actuarial neutrality for a minimum pension is not well-defined because the minimum pension top-up depends on potential future own contributions, contradicting actuarial neutrality. - Example effects: - Retiring one year early for someone with a 23-year career at the average wage compared to working and contributing one more year (24-year career) results in a 1 percent lower pension until the generation retiring in 2027; afterward the penalty increases as 23-year careers fall into lower career length brackets. - For a 5-year early retirement the penalty eventually will be very high since no minimum pension eligibility exists for 19-year careers from 2030 onward. - For the first few years after reform, early retirement actually provides a bonus in terms of the minimum pension top up because: - Working one year longer increases funds in the DC account and increases own-funded pension benefits more than the corresponding small (1-1.5 percent real) increase in minimum pension benefits. - Own contributions and returns can therefore partly pay for minimum pension benefits the individual was already eligible for the year before, creating incentives to retire early or work informally in the last years. - From 2028 onwards early retirement leads to small penalties with spikes due to wage bracket interactions; in the very long run penalties can turn into bonuses once minimum pension top ups shrink and DC savings rise because of contributions, positive real returns and positive real wage growth. - Penalties and international benchmarks: - OECD (2017) calculated actuarial neutrality on existing entitlements implies a penalty between 5-6 percent for Mexico at the retirement age and that working one year more leads to a bonus of 7.4 percent. - The penalty on the full minimum pension for retiring 1 year earlier ranges from 1.0 percent to 1.4 percent within each wage and career length bracket. - If retiring one year earlier also leads to ending up in two career length brackets lower (each bracket is 25 weeks) the penalty ranges from 6.0-7.5 percent, with the upper end closer to the implied actuarial neutral penalty. ### Lifetime benefits and retirement incentives - Total expected benefits received from the time of retirement (discounted with inflation) for 19-24 year careers: - Start around 10 times the last earned wage in 2020. - Fall gradually to 6 times the last earned wage around 2080. - For almost all generations, individuals retiring at the normal retirement age receive the lowest lifetime benefits net of contributions made during the last 5 years before the normal retirement age. - For most cohorts, retiring 4 years early is the optimal decision since this is the first year it is possible to retire on a minimum pension with 20 years of contributions. - Establishing actuarial neutral penalties for a benefit that negatively depends on entitlements of potential future own contributions is impossible under current design. One way to make minimum pensions actuarially neutral would be to fix the top up at the earliest age of retirement and adjust in an actuarial way independent of any potential future additional entitlements, but this would defeat the intended purpose of the minimum pension. ### Conclusions and policy options (selected) - High-level conclusions: - The 2019 and 2020 reforms smooth differences between pre- and post-1997 entrants but introduce fiscal, equity, and incentive issues. - Rapid increases in basic pension levels risk crowding out other urgent spending needs. - Current design leads to widely different outcomes for relatively similar careers and weak incentives to work longer due to low penalties on early minimum pensions, limited differentiation by career length, and shifting minimum years of contribution requirements. - Policy options proposed: - Limit the increase of the basic pension while maintaining benefit coverage. Close to 100 percent coverage has only recently been achieved and should be maintained. - Abolish minimum contribution requirement for funded pensions. Implement a minimum balance requirement for phased withdrawals and annuities instead; automate pension payments to reduce administrative costs and lower the minimum balance threshold. - Abolish the minimum wage as a separate government contribution bracket and as a lower bound for the minimum pension. Instead express the minimum wage in terms of UMA like all other earnings levels and have an open-ended lowest minimum pension bracket. - Make redistribution within the pension system transparent and intentional. The system mixes redistribution through the basic pension, government subsidized contributions and elaborate minimum pensions; redistribution through minimum pensions is opaque and leads to unintended outcomes. Several options exist to eliminate or limit these inefficiencies and some can be combined. *Source: Box 2. Modeling assumptions (IMF Working Paper).* ### Annex IV for numerical examples): ### Annex IV for numerical examples): ### Reform options — high-level findings and rationale - Minimum pensions often serve to reward long careers of lower earners but career length requirements create arbitrary and inequitable cut offs between those that just reach the requirement and those that almost reach it. - The only fully equitable minimum pension would be fully pro-rated as is the case in France and Switzerland. - Example: in the Netherlands pension entitlements that would lead to a pension of less than EUR 500 per year can be paid out as lump sum. This is equivalent to less than 1 percent of the average wage. However, the retiree can refuse the lump sum and request an annuity. Only pensions of less than EUR 2 a year are automatically paid out as lump sum. - In Mexico low earners are already subsidized through government contributions and redistribution and adequacy is enhanced by the introduction of the basic pension; an additional form of redistribution is unlikely to add much value, especially if it has negative equity consequences. - Across all reform options: early retirement on a minimum pension should be abolished and minimum pension benefit levels should remain price indexed. - Rationale: There is no way to make early retirement on minimum pensions equitable and to maintain work incentives. Allow minimum pension receipt only from the retirement age. - Option: offer possibility to defer the minimum pension after the retirement age; the minimum pension top up should be determined at the normal retirement age and actuarially adjusted (i.e., irrespective of actual additional contributions to the funded pension system). - Price indexation of minimum pensions is necessary to ensure the fiscal cost of the minimum pension doesn’t perpetuate, causing a slow phase out of the minimum pension (i.e., when even the lowest DC balance eligible for a minimum pension, results in an annuity exceeding the minimum pension level). - Timing of when this perpetuation will happen depends on real wage growth, real returns and annuity factors. ### Option 1: Phase out the minimum pension - Proposal: Phase out the minimum pension (example policy: fixing the minimum pension levels in nominal terms). - Intended effect: Phase out would limit the time design inefficiencies exist even if it would not completely eliminate transition inefficiencies. - Simulated scenario headings in the annex: - Half average wage - Average wage - Assumption stated for simulations: keep minimum pension levels constant in nominal terms. ### Option 2: Simplify by abolishing differentiation between career earnings brackets - Proposal: Abolish differentiation between career earnings for the minimum pension; maintain only the current lowest career earnings bracket and apply this minimum irrespective of career earnings. - Implementation pathways: - Gradually let higher earnings brackets converge to the lowest bracket, for instance by only indexing the lowest bracket while keeping higher brackets constant in nominal terms. - Alternatively, make the minimum pension fully proportional to career length only (i.e., truly rewarding longer careers) and remove discrete steps. - Simulated scenario headings in the annex: - Half average wage - Average wage - Assumption stated for simulations: let earnings brackets converge to lowest earnings bracket over a period of 10 years. ### Option 3: Eliminate threshold effects (express minimum pension as continuous function) - Proposal: If the link with career earnings is not abolished, express the minimum pension as a continuous function of average past earnings revalorized with economy-wide average wage growth and career length (alternatively proxied by DC balances or DC annuities). - Note: Using wage revalorized past earnings would increase average wages, minimum pension levels and therefore spending; combine this with a downward adjustment of benefit levels. - Example reform calibration used in simulations: - New minimum pension is 50 percent of lowest minimum pension, withdrawn at a 50 percent rate against own pension income. - Gradual conversion to new rules over 10 years’ time. - Career length requirements frozen in 2024. - Simulated scenario headings in the annex: - Half average wage - Average wage ### Option 4: Index earnings brackets with wage growth - Proposal: Index earnings brackets with average wage growth after 2024. - Effect: While not eliminating all discontinuities, this ensures under stable macro-economic circumstances similar careers receive similar minimum pensions and lowers the cost of the minimum pension over time. - Could be combined with revalorizing past earnings with average wage growth to enhance stability but this raises cost; careful calibration required to balance equity, adequacy and fiscal sustainability. - Simulated scenario headings in the annex: - Half average wage - Average wage - Assumption stated for simulations: Brackets are indexed with average wage growth after 2024. Career average earnings gradually converge to average wage uprated earnings over a period of 10 years. ### Actuarial neutrality and minimum pension top-up — analytical findings - For the full pension to be actuarially neutral the DC part and the minimum pension top up need to both be actuarially neutral independently. - Definition: Actuarial neutrality means that the discounted value of accrued benefits does not change by working an additional year: - 푃푉퐴,푡 = 푃푉퐴+1,푡 - Present value decomposition (as in source): - 푃푉퐴,푡 = 푃푉퐴,푡퐷퐶 + 푃푉퐴,푡푀 - 푃푉퐴,푡푀 = ∑ 푏푖푠푖 (1+푟)푖−푡 푖=푡 - With indexation: - 푃푉퐴,푡푀 = 푏퐴,푡 ∑ (1+푖)푖−푡 s푖 (1+푟)푖−푡 푖=푡 - 푃푉퐴+1,푡푀 = 푏퐴+1,푡 [∑ (1+푖)푖−푡 s푖 (1+푟)푖−푡 푖=푡 −1] - Top-up definition: - 푏퐴,푡 = 푀퐴,푡 − 푏퐴,푡퐷퐶 - 푏퐴,푡퐷퐶 = 퐷퐶푡 ∑ (1+푖)푖−푡 s푖 (1+푟)푖−푡 푖=푡 - For a non-working person the actuarial neutrality condition implies: - 푀퐴+1,푡 = [∑ (1+푖)푖−푡 s푖 (1+푟)푖−푡 푖=푡] / [∑ (1+푖)푖−푡 s푖 (1+푟)푖−푡 푖=푡 −1] × 푀퐴,푡 - For a working person the actuarial neutrality condition becomes: - 푀퐴+1,푡 = [∑ (1+푖)푖−푡 s푖 (1+푟)푖−푡 푖=푡] / [∑ (1+푖)푖−푡 s푖 (1+푟)푖−푡 푖=푡 −1] × 푀퐴,푡 + 푐푤푡 / [∑ (1+푖)푖−푡 s푖 (1+푟)푖−푡 푖=푡 −1] - Implication: The actuarially neutral level of the minimum pension is not the same for working and non-working people; the second element reflects the added benefit in the DC system of working and contributing one more period. - Behavioral issue highlighted: - At time t it is unclear if someone will work or not; taking an expected value of 푐푤푡 would likely involve the employment rate of a subgroup (e.g., 푐푒퐴,푡푤퐴,푡). - However, the decision to work will be influenced by the payoff of that work. If the minimum pension level remains the same and the payoff for the next period is zero in most occasions, past observations of work are likely an unreliable indicator for the probability of work. - If people only take into account pension payoff, the expected level of employment will fall to zero; this means 푀퐴+1,푡 converges to the non-working value and there is a strong disincentive to work. ### Simulations and caveats - The annex simulates a few reform options for informative purposes only. - The exact calibration of parameters should be carefully investigated on a micro level (i.e., distributional effects) and a macro level (i.e., spending effects). - Simulated scenario categories used across options: "Half average wage" and "Average wage". - Time horizons and calibration parameters referenced in simulations: - Convergence of brackets or rules over a period of 10 years. - Career length requirements frozen in 2024. - Indexing brackets with average wage growth after 2024. - New minimum pension expressed as 50 percent of lowest minimum pension with a 50 percent withdrawal rate against own pension income. *IMF Working Paper — Annex IV for numerical examples: An assessment of the 2019 and 2020 pension reforms in Mexico* --- _Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024053-print-pdf.pdf_