Real Pension Returns in Latin America: AFORE, AFP and Private Funds Compared (2016-2025)

For a decade, four Latin American countries managed more than 100 million retirement savings accounts using four different models, four different regulators and four different fee structures. The result was almost identical.

Between 2016 and 2025, the default funds of Mexico, Peru, Chile and Colombia delivered real returns - that is, after inflation - of between 3.25% and 3.50% per year. The gap between the best and the worst system over ten years was just 3.4 percentage points cumulative. To put that in perspective: the difference between countries is smaller than what any one of these systems lost in a single year, 2022.

There is a second finding, more useful for anyone actually paying in: the difference between funds within a single country was eight to ten times larger than the difference between countries. A Chilean in the most aggressive fund finished the decade with 25% more capital than one in the most conservative - same system, same fees. The decision that mattered most was not where you contribute, but which fund you chose, and, in 2022, discovering that "conservative" does not mean the same thing in Santiago as it does in Bogota.

This analysis brings together, for the first time in a comparable format, the official series of the four regulators: CONSAR (Mexico), SBS (Peru), Superintendencia de Pensiones (Chile) and Superintendencia Financiera (Colombia). The full methodology, including its limitations, is at the end.

The table below shows what happened to 100 units of purchasing power invested in each country's default fund on 1 January 2016, through 31 December 2025. All figures are in real terms (adjusted for each country's own inflation) and before administration fees (see the fees section, where this is explained and corrected).

What this means in practice: over ten years, a Latin American worker's retirement savings grew by roughly 40% above inflation in all four countries. That is not a bad outcome. It is also not the outcome most people believe they are getting - in none of the four countries did the money double.

The annualised figure of roughly 3.4% is a calm average that conceals a very volatile decade. The chart below shows the real annual return of each country's default fund, year by year.

In 2022 all four systems lost money in real terms, and not by a little: between -9.2% (Chile) and -15.7% (Colombia). It was the worst figure of the decade in all four countries, without exception. The cause was the same everywhere - the synchronised rise in interest rates hit bonds and equities at the same time, the combination that normally offsets. In Colombia the blow was larger because the collapse in local government bonds (TES) came on top of 13.12% inflation that year: the nominal return was -4.6%, but inflation turned that into -15.7% in real terms.

The number that matters: a worker checking their statement at the end of 2022 saw the equivalent of three years of gains evaporate. Anyone who did nothing got it back. 2023, 2024 and 2025 were positive in all four countries, and 2025 was one of the best years of the decade (+12.5% real in Mexico, +12.2% in Chile). The two best years were 2019 (between +10.8% and +15.0% real) and 2025 (between +5.8% and +12.5%), in both cases for the same reason: rising global equity markets, with Latin American funds increasingly exposed abroad.

So far we have compared each country's default fund. But workers in Chile, Peru and Colombia choose between several funds with different risk levels. That is where the most important finding of this study appears. The chart below ranks every fund by its annualised real return over 2016-2025.

The central number: the difference between countries, comparing each one's default fund, is 0.25 percentage points per year. The difference within a single country is 2.35 pp in Chile (A vs E) and 1.90 pp in Colombia (Mayor Riesgo vs Conservador). The decision of which fund to choose mattered roughly eight to ten times more than the decision of which country you contribute in. A Chilean in Fondo A ended the decade with 25% more capital than one in Fondo E - same system, same regulator, same fee, same ten years. A Colombian in Mayor Riesgo, with 20% more than one in Conservador.

This is the part usually left out. The risk ladder worked cleanly in two of the four systems, and did not in the other two. In Chile the order was perfect: A (4.48%) > B (4.17%) > C (3.25%) > D (2.34%) > E (2.13%), without exception. In Colombia it also held: Mayor Riesgo (4.04%) > Moderado (3.50%) > Conservador (2.14%).

In Peru it flattened at the top. Fondo 3 (growth) returned 3.40% against 3.35% for Fondo 2 - five hundredths of a point of premium, in exchange for clearly higher volatility (a range of -14.8% to +16.5%, versus -14.4% to +10.4%). At the lower end it did work: Fondo 1 (2.47%) beat Fondo 0 (1.27%). Mexico's gradient turned out hump-shaped, explained further below.

Ten years is a short window on which to judge a risk ladder designed for forty. But the result is clear and worth stating: in half of these systems, the most aggressive fund did not compensate for its additional risk during this decade.

The bad year exposed the fact that the label "conservative" means very different things depending on the country. The chart below shows the real 2022 return of the aggressive, default and conservative fund in each of the three multifondo systems.

Chile's conservative fund protected: it gained 7.8% in real terms while the aggressive fund lost 20.8% - a gap of almost 29 points in a single year. Colombia's conservative fund protected nothing: it lost 15.3%, essentially the same as the highest-risk fund. The reason is that "conservative" in Colombia means local government debt (TES), which collapsed as rates rose, and 13.12% inflation took the rest. A Colombian who switched to the conservative fund looking for shelter found none.

Mexico cannot be placed in the same table, for two reasons. First, its axis is not risk, it is age. Since December 2019 the country has operated ten generational SIEFOREs assigned by year of birth, with a profile that becomes automatically more conservative as the cohort ages. One frequent misunderstanding is worth correcting: Mexico did not eliminate freedom of choice. Before 2017 there was none - savings were transferred by force from one SIEFORE to another on reaching a given age. CONSAR introduced free choice in January 2017, and the 2019 reform kept it (savers can switch every three years). What was eliminated were the forced transfers.

Second, the series is shorter. The generational SIEFOREs have existed only since December 2019, so there is no comparable ten-year series. CONSAR publishes the cumulative return since that date, and it does not rise in a straight line with youth.

The gradient is not linear - it is hump-shaped. Returns rise from the oldest cohorts, peak at the generations born between 1980 and 1989 (workers aged roughly 37 to 46) and fall back in the two youngest funds. The gap between the peak (3.73%) and the pensioners' fund (2.51%) is 1.22 points. Two documented factors explain it: the return contribution from international instruments also peaks at the 80-84 cohort and declines towards the younger funds; and SIEFORE 95-99 only began operating on 26 August 2024, so much of its series is derived from prices inherited from SB Inicial. The honest formulation: the asset allocation of Mexico's glidepath is indeed monotonic (the younger the cohort, the more equity), but the realised return over this period was not.

Here is the uncomfortable finding of this study, and the reason the earlier figures carry the label "before fees". None of the four regulators publishes a series that is simultaneously real and net of administration fees. Each publishes a different piece of the puzzle. CONSAR (Mexico) publishes real returns, but before fees; its only net-of-fee indicator, the IRN, is nominal. SBS (Peru) publishes real returns from the unit value, but the fee is charged outside the fund, so the figure is gross. Chile's Superintendencia publishes real returns deflated by the UF, but the fee is deducted from salary and never enters the fund. Colombia's Superintendencia Financiera publishes nominal returns, and its fee is deducted before the contribution enters the fund. In other words, the return figures the public sees in all four countries do not include what the administrator charges.

The difference is structural, not cosmetic. A fee on the balance (Mexico, Peru) is charged every year on everything accumulated: it grows with the savings and compounds against the saver across an entire working life. A fee on salary or on the contribution (Chile, Colombia, and Peru's legacy 17% of members) is charged once, on the way in. Its relative weight falls as the balance grows - but it takes a large bite out of the earliest capital, which is precisely the capital with the most time to benefit from compound interest.

There is one historical exception worth recovering, because it shows the true size of the problem. In its fourth-quarter report for 2016, CONSAR published Table IV-17, "Real returns, net of fees, of the SIEFOREs" - exactly the figure that today exists in none of the four countries. It covered 2009 to June 2016.

That table never appeared again. We reviewed every fourth-quarter report from 2016 to 2025: the real, net-of-fee measurement by fund was published that year and never again. That is itself a finding about transparency.

Applying the fees in force in 2026 to the decade's gross return, the ranking changes. Colombia charges no fee on the balance and none on returns during the accumulation phase - a genuine structural difference. In exchange, it takes 3% of the contribution base up front: of every 16% contributed, only 11.5% reaches the individual account, so 18.75% of the contribution is never invested. Chile's fee is on salary, so its equivalent in asset terms depends on the ratio of balance to salary - around 0.5% a year early in a working life, falling to roughly 0.1% at the end.

Once fees are incorporated, the gap between the cheapest and the most expensive system is roughly one percentage point a year. Over a 40-year working life, one percentage point a year amounts to roughly a third less final capital. Fees matter far more than the differences in returns between countries.

Here the four countries genuinely diverge - not on what they earned, but on whether the money stayed in. Peru and Chile allowed mass withdrawals from their pension funds. Mexico and Colombia did not.

Between 2020 and 2025 Peru's Congress approved eight extraordinary withdrawals. The managed fund peaked in January 2020 at around S/175 billion; by 2026 the central bank reported it had fallen by roughly 34%, close to S/60 billion drained out. Around 4 million members - close to 38% of the 10.55 million registered - had a zero balance in December 2025, according to the SBS. Peru's central bank quantified the cost precisely: a 25-year-old member who took part in the withdrawals will see their expected pension at 65 reduced by 23%; a 40-year-old, by 48%.

Between 2020 and 2021 Chile authorised three 10% withdrawals. According to the Superintendencia de Pensiones, around US$50 billion was disbursed; approximately 11 million people withdrew at least once, and 57% used all three opportunities. Some 3.8 million people - 34.5% of those who withdrew - were left with a zero balance in their mandatory individual account. The total was equivalent to 25% of the funds' assets as of June 2020 and 18% of 2020 GDP.

All returns in this article are calculated per unit (unit value): they are unaffected by withdrawals. A Peruvian and a Chilean earned the same percentage return as if nothing had happened. But a percentage return does not pay pensions - the balance does. A real return of 3.35% a year on an emptied account produces nothing. This is the central asymmetry of the study: on returns, the four systems are almost indistinguishable; on accumulated capital, two of them suffered damage that returns do not reveal.

All four countries are reforming their pension systems at the same time, in different directions.

The 2020 reform gradually raises the total mandatory contribution from 6.5% to 15% of the contribution salary between 2023 and 2030, almost entirely at the employer's expense. Mexico is also the only one of the four with a legal cap on fees: since 2021 the maximum is set each year as the average of the implicit fees in the defined-contribution systems of the United States, Colombia and Chile. For 2026 that average came to 0.54%, and the system's average fee has fallen from 1.887% in 2008 to 0.538% in 2026. The Supreme Court upheld the mechanism as constitutional in October 2025. Ten AFOREs manage around 8.3 trillion pesos as of March 2026 across roughly 70 million accounts.

Ley 21.735, published on 26 March 2025, introduces the largest change since 1981: a new employer contribution rising gradually to 8.5%, a Social Insurance component with elements of intergenerational solidarity, an increase in the Universal Guaranteed Pension to $250,000 phased between 2025 and 2027, and biennial auctioning of 10% of non-retired members to the administrator offering the lowest fee. Above all, it marks the end of the multifondos: from 1 April 2027, funds A/B/C/D/E will be replaced by generational funds assigned by age. The regulator's stated rationale is that a Superintendencia study found between 72% and 78% of members who switched funds ended up worse off than if they had stayed put. 2026 is therefore the last full year of Chile's A-E series.

Ley 32123 (September 2024) promised to modernise the system: a minimum pension of S/600, a unified retirement age of 65, mandatory contributions for self-employed workers and, explicitly, a ban on further withdrawals. A year later, on 20 September 2025, the same Congress passed Ley 32445, which authorised an eighth withdrawal against the ban it had approved the previous year, eliminated the contribution obligation for self-employed workers, and restored the option to withdraw 95.5% of the fund on retirement. Supreme Decree 061-2026-EF formalised the reversal in April 2026. During 2026 nine further withdrawal bills were introduced; the finance ministry and the SBS opposed them and Congress closed its session in July without debating them.

Ley 2381 of 2024 would reorganise the Colombian system into four pillars and restrict private funds to income above 2.3 minimum wages - a drastic change, given that roughly 93% of contributors earn up to 4 minimum wages. It was due to take effect on 1 July 2025. It did not. On 17 June 2025, through Auto 841 of 2025, the Constitutional Court suspended the law's entry into force over a procedural defect and ordered a vote to be repeated, which the Chamber did on 28 June 2025. But the Court has still not ruled on the merits. To avoid the most common error: Ley 2381 was not repealed, is not in force, and was not declared unconstitutional - it is suspended, pending a ruling. Ley 100 of 1993 remains the applicable law.

The data show several things robustly. The country is not what differentiates these systems: four different models converged on roughly 3.4% real a year over a decade, so any debate focused solely on "which system performs better" is arguing about the wrong variable. The fund chosen does differentiate, and substantially - eight to ten times more than the country, and it is the only one of these variables the worker controls directly.

What this analysis does not say is equally important. It does not compare final pensions, which depend on contribution density, salary, contribution periods and each system's rules. It does not assess which system is "better" - a system with slightly lower returns and much wider coverage may be preferable, and that is a public-policy discussion, not a return comparison. And it does not project future returns: ten years is a short window for a retirement horizon of forty.

The period is the ten full calendar years 2016-2025; 2026 is excluded as incomplete. The exception is Mexico by generation, whose window is December 2019 to December 2025, because the generational SIEFOREs did not exist before then - those figures are not comparable with the ten-year columns and are presented separately. For the comparison between countries we use each country's default or balanced-profile fund; for the comparison between funds we use every fund type available in each system.

Each regulator uses its own deflator, and figures are presented exactly as published: Mexico deflated by the UDI, Peru by the Lima Metropolitan CPI, Chile by the UF, and Colombia's annual real series calculated by el-fondo.com from the official daily unit values (SFC dataset uawh-cjvi on datos.gov.co), asset-weighted and deflated by DANE's CPI. The Colombian calculation was validated against the SFC's own official figures at two cut-off dates, with a maximum deviation of 0.01 percentage points. All return figures are gross of administration fees, because that is how all four regulators publish them; the net estimates are el-fondo.com calculations applying 2026 fees and slightly overstate the net outcome in Mexico and Peru, where fees were higher at the start of the period. The figures are comparable as gross real fund returns, not as final outcomes for the member.