Newsletter
Suscríbete a nuestro Newsletter y entérate de las últimas novedades.
https://centrocompetencia.com/wp-content/themes/Ceco
The new pension law of January 2025 gradually increases the contribution rate for old-age pensions financed from fully funded defined contribution individual accounts.[1] This reform increases political stability by securing the votes of the left, including the Communist Party, for these pensions managed by for-profit firms (the AFP). The reform also introduces an additional contribution that some present as “solidarity-based”, as it funds two DB supplements, one for women and another for those who contribute for more than 20 years (men) or more than 10 years (women).
This reform also provides a new legal framework for the industrial organization of the AFP industry. It was designed by a group led by Senators Juan-Antonio Coloma (UDI), Luciano Cruz-Coke (Evopoli), Rodrigo Galilea (RN), Gastón Saavedra (PS), and Alejandra Sepúlveda (FRVS). The group included advisors to Ministers Mario Marcel and Jeanette Jara, technical staff from the Pension Superintendency, and a Technical Committee of 5 members convened by the aforementioned senators.[2] I was not part of this group, although I was consulted on several occasions, along with other external individuals.
«The decision that the political leadership now faces is between (1) letting time pass and wait for failures associated to 2, 3, 5 and 7 before seeking solutions, exposing participants to extensive delays; and (2) proactively amending now in 2025, before the regulations governing the first auction are issued.«
This group made the right decision: increase measures towards the goal of improving efficiency «for participants” in the pension system. This goal includes both using market competition to align the average return on assets of providers (the AFP) with that of alternative activities and keeping total costs low. The State is ethically obliged to strive towards this goal, that is the counterpart of mandating workers to hire and pay an AFP. The group made significant progress in this direction by securing legislative approval for the following highly positive reforms:
The new law’s architecture is based on the proposals in my recent study «Auctions for Existing AFP participants”.[5] However, it introduces innovations. This note is laser-focused on comments on these innovations, so it does not discuss the overall architecture. There are both desirable and undesirable innovations. I believe the negative ones have technical solutions that are politically acceptable, which I do not present here due to space constraints. The seven most systemically significant innovations—among others not discussed—are, in my opinion, the following:
1. Relief from excessive reserve requirements, that became a barrier to entry over the years: My proposal of 2018 and 2024 was to reduce the reserve amount and change the formula from 1% of AUM to 50% of annual fee income.[6] The law innovated by reducing the reserve amount to 30% of annual fee income. This is still a substantial amount for any AFP. It is also close to doubling the reserve required in 1990.[7] Thus, I evaluate this innovation positively. After paying corporate taxes on capital gains, the excess from current reserves is likely to be returned to AFP shareholders over 2027-30. Specifically, assuming that AFPs Habitat, Provida, Capital, and Cuprum maintain their current fees and that they lose the first two tenders, Habitat would free up approximately USD 504 million, while Provida, Capital, and Cuprum would each free up between USD 330 and 378 million, before taxes. These large distributions weaken potential claims by AFPs in arbitration panels, such as ICSID.[8]
2. Auctioning a single large group in a small lot: My 2024 proposal was to auction multiple small groups (e.g., 5% or 6.67% of total non-pensioned participants) within each lot, which in turn could be large (20 or 25%). With auctions occurring every 24 months (a period required to reduce the risk of collusion among bidders), the transition would reach and benefit all participants in 8 to 10 years. A multiplicity of small groups would not hinder an AFP that prefers to grow quickly or to enter on a larger scale, as it could bid for 2 or more small groups at once. Smaller groups would also help a bidder that prefers slow growth due to the «operational stress» of expanding administrative support.[9]
The law innovates in this regard by auctioning a single large group (10%) per lot, which makes the lot much smaller than what it could be (only 10%). A negative effect is that the first auction for the last group occurs 249 months after the law is published.[10] If the new law is published in April 2025, this group’s auction will take place in January 2047. Half of the worker-participants will wait at least 11 years for some auction to lower their fees, and 10% will have to wait 22 years.[11] Additionally, participants who start receiving pensions in the interim will never access fee reductions. These delays mean that the Chilean State will continues to fail in its duty to protect participants from paying excessive fees and excessive ROAs to many AFP.
3. Likely fee increases for a majority of participants: Due to point 2, non-auctioned participants will outnumber auctioned ones until 2036. Therefore, the outcome for non-auctioned participants, served by AFPs that do not participate in or that lose the auctions, will dominate the effects of the new law. Will a losing AFP want to raise its fee in 2025-36, above its current level? Until now, the deterrent to such increases has been the fear of prompting the political system to legislate again to punish the AFP (say by redirecting contributions – which is the base for the fee – to a new state-managed provider). However, one or more AFPs might bet that if only one or a minority of them raise their fee, the political system will not intervene. Another AFP might argue that, as it loses contributors to the auctions, it should compensate for the revenue loss by raising its fee rate. If fees begin to rise gradually, when will a majority of politicians be willing to legislate a punishment?
It depends. The many politicians who have congratulated themselves for approving this new law risk harsh criticism if they legislate a punishment in response to fee hikes, because they would admit failure. In this scenario, it will be attractive for many politicians to downplay the increases and postpone action. This hypothesis is supported by the evidence on past behavior of Chilean politicians during the 2008 reform. In that instance, the AFPs that did not plan to participate in the auctions for new participants that would start in 2010 raised their fees over 2006-2009, harming the majority of participants.[12] Therefore, our prediction is a similar increase in fees in 2025-30. In this scenario, fee increases by losing AFPs, which harm the majority of participants, would create serious political damage to the current reform. The other scenario is that current political leaders act proactively, preventively and promptly. One such measure would be to replicate the punishment that the new law applies to the strategy explained in section 4 (Valdés 2024 proposed other measures). However, so far, the new law omits any preventive measure.
4. The «Planvital strategy» is removed: In the auctions for new participants in 2010-23, one regulation established that, at the end of the period for which the winning bid was committed, the winner would retain that group of participants and could raise its fee freely. A «Planvital strategy» (executed by AFP Planvital in August 2018) is to apply a massive fee increase at that point. This AFP (a subsidiary of the Italian insurer Generali) won the auctions for new participants in 2014 and again in 2016 by reducing its fee so much, that it even suffered a slight operational loss in 2014-18. Then, at the end of the second service period (2016-18), AFP Planvital almost tripled its fee (from 0.41% to 1.16% of earnings, a rate maintained to this day). Thanks to the inertia of participants, Planvital retained almost all of them (it lost only 3% of its aggregate fee base). This strategy is scandalous because it limits the effectiveness of the auction to a transient period of 2 years, contrary to the spirit of the 2008 law that created the auctions.[13], [14] Valdes (2024) proposed one way to block this strategy: to re-auction the AFP service for the group at the end of the period for which the winning bid is committed.
The new law removes the Planvital strategy in a different way: first, it reduces the period for which the winning bid must be committed (5 years), far below the duration of the service period after which that group is re-auctioned (20 years). Second, it establishes a disincentive against raising the fee at the end of the 5-year period. This incentive is to bring forward the re-auction schedule for the participants in the group subject to an increase, in partial installments with a schedule set by law.[15] This “early re-auction” provision may be effective in hindering the «Planvital strategy” for the first few years after the 5-year commitment period.[16]
5. Preserves a regulation that cuts the number of bidders and reduces competition: An auction with a single bidder is a monopoly.[17] Such an auction allows an average return on assets (ROA) for the winning AFP that is way above the ROA observed in alternative activities.[18] So, what influences the number of bidders? One factor is a regulation that requires the winner to extend its offer to all the other participants it serves, despite not being members of the group for which the auction is organized and despite having different rights.[19] Such a mandatory extension of the offered bid to all other participants previously served by the bidder is a substantial economic burden on an incumbent AFP that dares to bid.[20] Indeed, the evidence from the 2010-2023 auctions for new participants is that most incumbent AFPs chose not to participate or to make soft offers, in response to this biased burden. The main explanation of why auctions for new participants in 2010-23 failed to attain a larger reduction in fees, despite ample space for it, is this regulation.
The new law preserves this regulation, contrary to the proposal in Valdés (2024). By encouraging monopoly auctions, or auctions with limited rivalry, this aspect of the new law supports future fees that are substantially larger than what competitive auction could attain.
6. Repairs another norm that previously raised fees: In the 2010-23 tenders for new participants, the combination of two other norms also reduced the number of bidders and raised fees. One required the winning fee to be lower than the lowest fee in effect across all AFP, not just lower than the other bids. The other regulation established that, in the absence of bidders, the flow of new participants would be assigned to the AFP with the lowest fee among all AFPs at that moment. Consider now a low-fee strategy, whose goal is to make it difficult for new providers to enter and learn the operational know-how, to be followed by an increase in the fee to just below the second-lowest fee among all AFPs, aiming to receive the flow of new participants. These two regulations combined with this strategy ensure (i) the absence of bidders, (ii) to receive all new participants, and (iii) to raise the fee to just below the second lowest among all AFPs. In fact, this combination was in effect for 14 months in 2018-19 and forced those new participants to pay a fee of 0.77%, substantially more than the fees that won in previous and subsequent auctions for new participants.[21]
The new law innovates by modifying the first norm, as it only requires that each offer be lower than or equal to the fee that each particular bidder has applied up to that point.[22] This flexibility allows a fee to win an auction despite being higher than another fee being charged by a non-bidder, if the new winner is a different AFP. This innovation reduces the likelihood of zero bidders and increases competition.[23]
7. Awards and penalties for financial performance are subject to a rigid and vulnerable scheme, despite the existence of superior schemes. An AFP manages pension funds by delegation from participants and the authorities who mandate contributions. In all delegated financial management, a benchmark return is essential to measure the delegate’s performance. The benchmark also greatly influences the overall allocation of the actual portfolio, which determines the average return and risk provided to participants.[24] In the U.S.A., competitive delegated management markets use the returns obtained by similar managers (“peer benchmarks”) as a reference. The other alternative is an “external benchmark”, given by some weighted average of certain external indices to be selected, such as the S&P500, one of Barclays’ bond indices, and indices for domestic securities. Of course, hybrid benchmarks are possible and are used in the U.S.A.
The «peer» reference has been defended with evidence that the effort of rivals is incorporated more quickly into the reference and that it is more demanding as the exact reference portfolio is known with a one-month lag.[25] Although the peer reference has been efficient when the challenge is limited to select investments within a predefined scope, there is no evidence yet that it is efficient when the main task is to choose the overall allocation of a balanced (multi-asset) portfolio, despite the fact that Chile has used peer benchmarks since 1990 (the peer return is the average of returns obtained by pension funds managed by other AFPs).
An external benchmark in the AFP context requires trusting public officials of the Superintendency of Pensions (SP) and members of the Consejo Técnico de Inversiones (CTI) to determine and to update the external indices and their weights. There is evidence on this for the Chilean bureaucratic and political context, from the experience of the Unemployment Fund Administrator (AFC). The evidence is that officials take too many years to incorporate novelties to the reference return, lack practical experience in portfolio management, and may use their regulatory powers for their own bureaucratic purposes. Specifically, the AFC benchmark has been set by the SP and CTI since 2009 as a weighted average of certain public indices they chose. This benchmark gives zero weight to peer returns. The resulting benchmark has been so outdated (it was updated for the first time 13 years after the first benchmark) that it allowed AFC II to beat it almost every year for a decade, by about 15 basis points per year, despite its modest investment in management teams.
Moreover, the potential payment of million-dollar awards to the AFC was perceived by officials as exposing them to «headline risk» (and removal risk). The SP and the CTI used their regulatory powers to create a range of values for performance that pay zero awards and charges zero penalties. They exaggeratedly widened this range, ensuring that AFC II would never receive awards (or pay penalties) and that “headline risk” never materialized. Thus, the AFC has had zero incentive to invest in quality financial management, harming its participants while SP and CTI officials became protected.
How does the new pension law respond to this? The new Article 50 bis takes the extreme position of requiring pure external benchmarks. It denies the possibility of a relevant weighting for peer benchmark returns. This choice forces 100% reliance on SP officials and CTI members. The new law also allows regulators to widen without limit a range with zero awards or penalties. These aspects of the new law are undesirable.
To conclude, the innovations presented in sections 1, 4 and 6 are desirable. Regarding the weaknesses identified in sections 2, 3, 5 and 7, my view is that they have technical solutions that are politically defensible. Legally, it is feasible to repair them through miscellaneous articles in some future social security law to be presented in 2025.[26] The decision that the political leadership now faces is between (1) letting time pass and wait for failures associated to 2, 3, 5 and 7 before seeking solutions, exposing participants to extensive delays; and (2) proactively amending now in 2025, before the regulations governing the first auction are issued.
*A spanish version of this publication is available in the following link.
[1] The reference is to the bill of January 29, 2025, sent by the National Congress to the Executive and to the standard revision by the Constitutional Court.
[2] This Committee was coordinated by Cristóbal Huneeus and included Paula Benavides, Cecilia Cifuentes, Hermes Gutiérrez, Soledad Hormazábal, and Juan-Pablo Letelier.
[3] Other entry requirements remain, such as experience, capital and proper previous behavior. Previously, these institutions could not create AFP subsidiaries because their sectoral laws required specialization in their designated activity. The new law also authorizes the government agency IPS to provide administrative support services to AFPs that may wish to hire it. Due to concerns about predatory pricing by an IPS managed by leftist civil servants, the IPS’s fees were required to be set by the Ministries of Finance and Labor, based on an external cost study and reported to Congress (Articles 23 and 55 bis). It is unclear whether an AFP would want to hire the IPS.
[4] The new law raises the minimum capital to create an AFP to about USD 2 million. It also allows the Tender Bases to require much larger guarantees (Article 162). In the 2011 and 2021 tenders for the Unemployment Fund Administration (AFC), the Superintendency required guarantees of approximately USD 16 million. Sources: DS 86 Min. Trab. D. Of. October 27, 2011, and DS 11 Min. Trab. D. Of. August 28, 2021.
[5] Chapter 4 in Valdés-Prieto, S. (2024) «Auctions for Existing Participants in AFP «, Diálogos CeCo, Competence Center of the Adolfo Ibáñez University, July 3, 2024, available in <https://centrocompetencia.com/licitacion-afiliados-antiguos-propuesta-aumentar-competencia-afps/>. The discussion in that paper can be traced to Valdés-Prieto, S. (2005) «To Increase Competition among AFPs», Estudios Públicos N° 98, Autumn, Santiago, p. 87-142.
[6] Valdés-Prieto, S. (2018) «AFP Costs and Entry: The Role of the Reserve Requirement. Diagnosis and Reform Proposal», Clapes UC, Working Paper No. 47, Santiago, Chile.
[7] On December 31, 1990, the pension system’s reserve was USD 66 Million (1% of the fund). Adjusted for U.S. inflation (CPI), this amount is equivalent to USD 157 Million as of September 2024. The new law requires, as of September 2024, USD 282 Million, 80% more than in 1990.
[8] ICSID is the International Centre for Settlement of Investment Disputes, operated by the World Bank in Washington D.C. Its acronym in Spanish is CIADI. The reduction of the reserve will also make the financial statements of these AFPs show a more realistic (higher) return on assets (ROA) and return on equity (ROE), increasing transparency on the industrial organization’s outcomes.
[9] See more in Valdés, S. «AFP Auctions: Stressful?», column in Diario Financiero, August 20, 2024.
[10] The first 10 auctions (first fill round) are separated by 24 months, and the first occurs 33 months after the law is published. Therefore, the calculation is: 33 + (9 x 24) = 249 months.
[11] A 20-year duration of the service period, after which that group is re-auctioned, exposes bidders to large risks of general cost increases and decreases in the fee base. A widely used solution in private long-term supply contracts is indexation. Similarly, the fee rate that won the bid could be indexed to total costs and to the fee base (see a specific proposal in section 4.3.F in Valdés 2024, p. 56).
[12] Flores (2012, p. 47) investigated the evidence on politician behavior in the 2006-2009 reform. Initially in 2006, the «Marcel Commission» supported previous proposals to auction AFP services, although only for new participants. These proposals were made into law. Between December 2005 and August 2009, all AFPs gradually raised their so-called variable fee (the main fee), citing compensation for the elimination of the fixed fee mandated by the same 2008 reform. That law also separated the variable fee from the premium for Disability and Survivorship Insurance (SIS), requiring Flores (2012) to estimate and subtract this premium for each AFP. Flores finds that the compensation was far from balanced: AFPs applied a net increase in fees. Updating Flores’ figure for the net aggregate increase in fee (in 2008 pesos) by the CPI variation to 2024, the increase was CLP$93.367 million per year (in 2024 Chilean pesos), about USD 100 million per year. In sum, the evidence from that episode is that many politicians chose to hide the fee increase and preferred to congratulate themselves for this aspect of the 2008 reform. Source: Flores, Yarela (2012), “Essays on Dynamic Models Applied to the Pension Fund Management and Banking Industry in Chile”, Tesis de Doctorado, “Anexo 10: Detalles sobre demanda y costos”, p. 58-61, I. de Economía, P. Universidad Católica de Chile. https://repositorio.uc.cl/items/61a121c6-0197-492d-935f-aabf336be3a1
[13] This strategy was foreseen a decade earlier. See model by Farrell, J. and C. Shapiro (1988).
[14] Two responses to this episode by the Chilean political system are revealing. On the left, several parties proposed nationalizing the service provided by AFPs. On the right, no political party denounced this abuse. Overall, the political system postponed taking measures until this new law, whose first auction will occur only in early 2028, 10 years after the Planvital scandal. This evidence on the past behavior of politicians also supports the hypothesis on their future behavior that was presented in section 3.
[15] See the early re-auction schedule in new Article 163, third paragraph, final sentence.
[16] The early re-auction provision in the new law has weaknesses: (1) the AFP may raise its fee even more to compensate; (2) the early re-auction for this group is rigidly defined in the law with rigid quotas, which reduce its power as a punishment in certain cases; and (3) starting from the 12th year of service, approximately, the proximity of the end of service (year 20) weakens the punishment.
[17] More precisely, it is a «bilateral» monopoly, because there is only one buyer, the Superintendency.
[18] An example: in the 2011 and 2021 auctions for the Unemployment Fund Administration (AFC), the fee agreed with the sole bidder allowed its shareholders to obtain a ROA of 34% per year (for AFC II) and of at least 30% per year (for AFC III). Both are far above the Chilean average ROA, which various methods set between 11 and 12.5% real annually (Valdés 2024, section 3.4 p. 39-42).
[19] There are 2 standard defenses for this old regulation (in effect since 1981). First, the belief that participants would protest if there were different fees for contracts with a fixed commission for 5 years and for contracts with a fee modifiable with 90 days’ notice. In fact, few would protest because participants who negotiate their net salary expect their employer to pay the fee. Second, the myth that the uniform fee rate (a % of salary) would create progressive redistribution, from participants with a high fee base (high salary) towards others with a low salary and towards participants that are not currently contributing. In reality this has not occurred, because this regulation redistributes towards the owners of those AFPs with clienteles with a higher average fee base. In addition, this regulation reduces the number of bidders and competition. There have been proposals to partially relax this regulation that forces the extension of the winning bid to others that are not members of the auctioned group (Valdés 2024, section 4.4).
[20] Some authors have defended this as an incentive for entry, but that forgets that in successful bidding contests many bidders are incumbents.
[21] This combination impaired first-time holders of jobs covered by social security who joined from August 2018 to September 2019. This delay in calling the replacement auction has been questioned.
[22] See the new Article 160, second paragraph.
[23] The new law also transfers to the Tender Bases the responsibility of defining who serves the flow of new participants in the absence of bidders. There is a need to improve coordination with the long service period of 20 years, as this period allows up to 9 different AFPs to have won and become obliged to maintain their fees, preventing them from bidding. This would reduce the number of aggressive bidders in auctions in the last part of the cycle.
[24] Buffa, A.M., D. Vayanos, P. Woolley (2022) “Asset Management Contracts and Equilibrium Prices”, Journal of Political Economy, 130, 12, December, p.3146-3201.
[25] Evans, R.B., J.P. Gomez, L.L. Ma, Y. Tang (2023) «Peer versus Pure Benchmarks in the Compensation of Mutual Fund Managers» Journal of Financial and Quantitative Analysis, p.1-38.
[26] It is doubtful that a good vehicle for this would be the bill committed for a reform of Disability and Survivorship Insurance, as the topic requires lengthy studies and arduous negotiations to cut fraud.
Regístrate de forma gratuita para seguir leyendo este contenido
Contenido exclusivo para los usuarios registrados de CeCo
