A new AFP crisis and the need for a system change
By Gonzalo Martner, October 20, 2026
In 1981, the dictatorship in Chile replaced the pay-as-you-go pension system (where current mandatory contributions fund current pensions) with individual accounts managed by private administrators (where current mandatory and voluntary contributions fund future pensions based on their performance over time in financial markets). It reduced the contribution to 10% of taxable salary (plus commissions and disability and survivor insurance) to encourage workers to join the new system, but at the cost of the future, as in the long term, this contribution is insufficient to fund pensions with a good replacement rate of the previous salary. Furthermore, there was a large fiscal transition cost, as existing pensions had to be funded without the income from contributors, which contracted public spending on health, education, and infrastructure, in addition to not adjusting pensions for inflation in 1985.
Since then, capitalization has been the central pillar of the pension contribution system rather than a supplement, as it is in many countries, including the United States. This is why Chile is often cited as the "purest" example of individual capitalization.
A first serious problem with the imposed system is that average pensions maintain a low ratio to pre-retirement salaries, especially for women, and accumulate gaps over time due to the structure of the labor market. Since 2002, the system has yielded between 3% and 5% real annual return, depending on the fund and the internal rates of return in each individual case, with a downward trend. The average profitability since 1981, which is usually presented, is heavily influenced by the exceptional returns of the 80s and 90s, when real interest rates in Chile were very high, while the average unit value overestimates what a typical affiliate earns because they had few contributions when returns were higher, and that average does not discount commissions.
The fact is that the State has had to increasingly take charge of retirements, especially through the guaranteed pension for those over 65, which currently has a value of 232,000 to 250,000 pesos per month depending on the case. With around 3 million people aged 65 or older in Chile, the 2.3 million people currently benefiting are equivalent to approximately three out of every four older adults. Those belonging to the top 10% of earners do not have access. Let us remember that in 1989, the welfare pension covered 8% with miserable amounts.
This mechanism is, in any case, a pay-as-you-go system (current taxpayers pay current pensions) and has a high cost—fully justifiable—that is financed by taxes. The issue is that these taxes are globally regressive (the wealthiest pay less in proportion to their income, especially due to the weight of VAT and exemptions on capital income) and will be even more so with the current government's reform, which lowered taxes for the wealthiest segments.
A second serious problem is that for decades, workers have had to not only allocate 10% of their taxable income to an individual account over which they have no control, except to switch between 5 funds without any guarantee of results, but also pay substantial profits to the private administrators of their savings. In a 2023 study, economist Salvador Valdés concluded that the high commissions charged by the AFPs allow them to obtain a higher profitability than almost any other private activity. The return reached 33.3% in 2017-2022, compared to the 12.5% average for large and medium-sized companies. Only in Chile does it happen that one can recover an investment in a social security activity in three years, based on mandatory contributions imposed by the State on the income of those who work. The inertia of affiliates—who usually do not switch to the AFP with the lowest commission, in a case where competition does not work with price signals as it theoretically should according to conventional economics textbooks—is also a barrier to entry for new competitors. If a portion of the contributor portfolio were periodically auctioned, as provided for by the reform approved by the previous government but not yet implemented, the average commission could be reduced by half. In other words, today, double what should be charged is being collected for the benefit of the AFPs.
AFPs are parasitic institutions. This is a form of rentier capitalism imposed in Chile by the economic oligarchy that allows for unjustified profits and the construction of hyper-concentrated dominant business positions, all at the expense of the system's contributors. It is argued that they receive in exchange returns on their funds that increase their pensions, but these returns have not been higher than average long-term financial investment indices. Nothing prevents this profitability from being obtained, for example, by a public fund that carries out capitalization investments with low risk and low intermediation costs to complement pay-as-you-go pensions. In the reform approved in 2025, up to 4% will go to a public-sector Autonomous Pension Protection Fund.
Incidentally, the argument that pay-as-you-go systems are bankrupt is simply false. This is not the case in the United States, Spain, or Germany, for example. What does have an impact, and this is valid for all systems, is a demographic change that forces societies to spend more on pensions if they want to maintain them (and also on health), for which contribution rate and duration parameters and retirement age are adjusted with full rights. Societies can address this change in the logic of "every man for himself" or in a structured way with solidarity components between generations and within each generation to allow for decent ends of life, without prejudice to private supplements.
There is a third serious problem. AFP funds are invested in financial instruments with random results, which is very risky in general and especially for those who are close to retiring. It was supposed that Fund E (and D) should invest with greater security the funds that come from the mandatory contributions of salaried workers (or voluntary contributions with tax incentives) and past returns. It happened that Fund E accumulated a 10.4% drop between January and September 2026, which reduces the pension to be received by a similar figure. It is not just any loss.
There were AFP investments in sovereign bonds, especially United States Treasury bonds, and in what are called "dollar interest rate derivatives." These are "interest rate swaps" that constitute a contract between two parties that exchange interest flows on a reference amount, without exchanging that amount. For example, an AFP agrees with a bank in the United States to receive a 4% fixed annual rate on US$100 million for 10 years, and in exchange pays the variable market rate for each period. If market rates fall, the AFP continues to receive 4% but pays less, and the contract gains value. If rates rise, the AFP pays more than it receives, and the contract loses value.
In practice, that position behaves similarly to holding a long-term bond with a fixed interest rate issued to finance a government or a company, but without buying it. It should not be forgotten that once they are issued, bonds are traded in financial markets at a value lower or higher than the issue value depending on their supply and demand. For fund managers, these operations are part of their fixed-income investment strategies.
In the last two years, the seven AFPs, most of which are owned by foreign insurance companies, invested in derivatives that would give them profits if US interest rates fell, which worked at one stage. But later, US Treasury bond yields rose in a context of global inflation linked to the war in Iran and the expansion of investment in Artificial Intelligence, which substantially increased the demand for funds, while the Federal Reserve chose to raise short-term interest rates for the first time in three years, with some impacts on long-term instruments. As a result, the market value of those positions went from a gain of nearly US$1.4 billion a year ago to a loss of around US$7.7 billion in September, according to the Superintendence of Pensions. The positions as of September were taken with 15 financial entities, the largest of which are US and European banks such as Bank of America (-US$1.376 billion), Société Générale (-US$1.270 billion), BNP Paribas (-US$1.137 billion), Barclays (-US$941 million), Deutsche Bank (-US$543 million), and Standard Chartered (-US$459 million). The inverse bet is held by another investor.
Fund E concentrated most of the receiving positions (those that receive a fixed rate), which left it especially exposed to US rates. Added to this was the fall in the bonds it already had in its portfolio.
As interest rates decrease, bonds increase in price due to higher demand, given that future bonds will be placed at lower rates. In a context of low interest rates, the AFPs increased their exposure to rates through derivatives. They swapped floating rates on instruments whose yields are dictated by the market for a fixed rate. In the process, they lengthened the duration of the portfolios and therefore the risk of exposure to rate variations. If the fund had had a risk limit at the portfolio level, the 2025 loss would not have occurred. Between 2002 and 2025, the risk of Fund E increased steadily due to a regulatory omission: the regulatory framework sets investment limits by asset class and credit quality for fixed-income instruments, but not on their duration. In 2009, the five pension funds had similar durations of 4.5 to 5 years, which were increasing. In 2025, the duration of Fund E went from 7.2 to 10.2 years, meaning the fund that was supposed to be the most conservative ended up being the most exposed to interest rate variation risk.
There were warnings from the previous Superintendent of AFPs that were not considered by the fund managers, who are autonomous in their investment decisions within certain parameters and tend to have herd behavior, while not being responsible for the eventual losses from their decisions. The Central Bank, in its 2025 and 2026 Financial Stability Reports, had also warned about the risks of these exposures.
The US$7.7 billion corresponds to the negative market value of the positions. The cost of exiting each contract increased as rates moved against the funds. In principle, not all of that amount is necessarily a definitive loss: if rates fall again, a portion can be recovered. The problem is that in April, the Superintendence set new limits on risk in derivatives. The rules will be in effect from November 4 and will force the AFPs to liquidate a good part of their positions.
It is thus confirmed that the necessary reform to the pension system in Chile maintains a great challenge: to substantially reduce the part of mandatory contributions that go to the AFPs, without prejudice to limiting exposure to risk. It will be inevitable to move forward, although it will hardly happen under the current government, to a mixed system whose center of gravity is the recently created social security fund, which was reduced to 4% of the contribution. The mandatory contribution that goes to the AFPs was increased without any justification from 10% to 14.5% of the taxable salary in the reform approved by a parliament with a right-wing majority in 2025. This is what must be reversed.
