Chapter 3. Pension costs in the accumulation phase: Policy options to improve outcomes in funded private pensions

How costs and charges differ in DB and DC funds

There may be additional costs incurred in running DC funds that do not affect DB plans, as Table 3.2 highlights. These are related to the elements of choice and liquidity within some DC arrangements. If either employers or employees are able to choose their DC fund, the provider is likely to spend money on sales and marketing activity. Where DC investment strategies have to accommodate inflows and outflows, there may be layers of cost such as platform or entry and exit costs that are charged to members’ portfolios.

Sales and marketing activity that does not include education or information for members does little to improve outcomes, but the costs are borne by members4. These costs may be indirect, for example commissions paid to advisors that come out of contributions, or they may be direct as in Mexico and Poland, where they are counted as operating costs of the provider. Table 3.3 shows the cost of acquiring new business in Poland from 2008 to 2016 and the impact on operating costs of Pension Societies (PTE) when sales activity was banned from 2012. Notably, PTE did not return the savings from reduced sales activity to members; instead, they enjoyed an increase in operating margins.

Members may pay entry or exit fees to underlying vehicles when DC pension plans execute their investment strategy through investments in underlying vehicles such as mutual funds. This is because the plan has to adjust its holdings in the underlying vehicles when members make contributions or withdraw their assets. Entry and exit costs are implicit – they come out of the value of the member’s portfolio – and can be hard to measure. DB plans may also invest via underlying funds, but they are typically able to buy share classes that do not carry entry/exit fees and they do not have to manage such frequent portfolio adjustments.5 The UK Department for Work and Pensions found that among 14 providers handling 14.4 million pension pots, two were unable to say whether they applied entry costs, six applied entry costs but were unable to say what they amounted to, four did not apply entry costs, and two applied entry costs that amounted to a reduction of between 5 bp and 40 bp per contribution. Providers also said that they found it hard to get clear information on entry and exit costs from the managers of the underlying vehicles (DWP, 2017[9]).

When plan members can decide on the design of the investment strategy, then in addition to any entry or exit fees, the member may also pay fees to the provider for the ability to make changes to the selection of underlying funds and for using the provider’s “platform”. The platform provides a service (choice, customisation and easier implementation) but it may be hard to understand what members are paying for this service and what it is really worth.

As well as having additional layers of cost compared to DB funds, DC arrangements may have higher costs than DB funds for equivalent activities. Administration activity is likely to be more expensive within workplace DC plans than in DB plans. DC administrators handle large volumes of small inflows; they may also incur additional costs such as the need to provide middleware (connecting their IT systems to employers’ payrolls) or to offer transfer services to members when they change employers.

DC providers may also have to invest more in communications designed to build member engagement, to help members make choices about their pension arrangements or to encourage them to contribute more. An alternative to such expenditure that would be more effective and lower cost would be to make auto-escalation compulsory, so that individuals’ contributions increased automatically in line with their earnings.

Complex and opaque charges

The structure of charges is not standardised, making it difficult for employers and employees to compare what different providers charge in total for DC funds. Fees may cover different costs and be expressed in different ways. Deloitte (2014[15]) sampled 525 plans in the United States and found a variety of fee arrangements for similar services. For example, administration costs could be charged directly to the employer, to the participant, or at the level of the plan itself. These fees could be charged on a per-participant basis, per plan, or as a percentage of assets.

Complex charging structures may lead to wide variations in charges within a single jurisdiction. The Deloitte study found that the “all-in” fee (total charge) for the sample universe ranged from 0.28% to 1.38% of assets on a participant-weighted basis. Clearer communication of fees, such as expressing them in cash terms rather than as a percentage of assets, could help employers and members compare plans more easily, although this would not capture charges that are unreported, opaque or not explicit. Taking fees paid by employees directly out of their wages rather than deducting them from member contributions – as is done in Chile – could encourage awareness of fee levels as their impact on disposable income would be felt immediately.

In DB funds, there may be information asymmetries between sponsors and boards and their suppliers. RailPen, which manages pension assets of over GBP 20 billion, calculated that the total amount it was paying for external investment management were up to four times more than the GBP 70 million that they were paying annually in direct fees. In response, they started to bring more investment management in house.8 Investment managers might charge different fees for segregated accounts to different clients and impose non-disclosure agreements so that pension providers cannot compare the fees they are paying to those being paid by their peers. This has led to a wide range in the fees paid by institutional investors for similar services (Table 3.4).

An important part of these charging structures is investment expenses, which are likely to be a major cost item and potentially one that can be compressed in both DB and DC funds. Ramsey (2002[16])found that charges to recover the cost of investment management typically were the largest ongoing charges and had the most direct relationship to the performance of retirement funds in Australia. Net investment returns – that is, portfolio performance after deducting investment expenses or costs, are a significant driver of pension outcomes.

However, investment costs warrant special attention because they do not exist in isolation from the investment strategy. Low cost, low risk strategies generally generate lower returns than high risk strategies over time. It may be possible to reduce the transaction costs of an active emerging equity fund by negotiating with external managers and brokers; it is not possible to reduce them to the same level as the transaction costs of a passive bond fund. Most of the investment cost savings made by Dutch pension funds in the wake of their transparency initiative came from changing their investment strategies and implementation styles, although they also made savings by putting pressure on external managers and other intermediaries to offer them lower prices. In particular, pension funds reduced their allocations to high-cost alternative strategies and brought active management in-house.

Focusing only on cost may have unintended consequences. Morkoetter and Wetzer (2016[17])found that the introduction of TER reporting may have led Swiss pension funds to avoid higher-performing asset classes because they are focusing on absolute costs rather than costs in the context of returns. However, Ainsworth et al. (2016[18])found that while higher fees were associated with higher returns, they were also associated with higher risk, so that outcomes were not necessarily better.

Weak governance

Weak governance can result in a failure to identify and align costs and charges. Smaller schemes in particular may lack the resources for effective oversight of costs and be in a weak position to negotiate with their suppliers.

Conflicts of interest may weaken governance: external trustees of small plans might vote against scheme consolidation if it would put their jobs at risk, while the boards of providers that are part of a larger financial institution might be reluctant to reduce costs that are a source of revenue for their parent company.

Governance failings could also arise from agency problems, or from a dilution of fiduciary responsibilities as the intermediary chain gets longer. Employers who are responsible for selecting providers may not focus on the costs borne by employees, while employees may lack representation on fund boards.9 Fund platforms might not use their bargaining power on behalf of members. Investment consultants could be tempted to increase their billings by proposing complex investment strategies or frequent changes to asset allocation. Advisors who benefit from built-in commissions (an implicit cost) are less likely to recommend switching providers. Investment managers who are not required to report indirect costs might not try to control them.

Barriers to entry/switching

Barriers to entry and to switching may exist at the level of the provider or at the level of the plan participant. Competition between plan providers should be strongest when a new plan is being set up or when participants (including employers, if they select the plan for their employees, or plan members if they have a free choice of provider) are thinking of switching plans.

Providers may face barriers to entry due to the high proportion of up-front costs in setting up a new plan - fixed costs such as IT are incurred before significant pools of assets are collected. This may make it especially difficult to attract new providers for automatic enrolment schemes targeting smaller employers and lower paid workers. New plan providers and external asset managers might be at a competitive disadvantage from lower brand recognition or a shorter track record. Vertically-integrated providers might limit access to their platforms by external asset managers or benefit from lower internal management costs thanks to the breadth of their internal trading platforms. On the other hand, if it is too easy for new providers to enter the market there is a risk that they will not achieve sufficient scale and be forced to increase charges or reduce the quality of the plan, or withdraw from the market completely.

When plan participants consider changing plan provider, they may be discouraged from doing so by regulatory barriers, such as taxes, or by commercial barriers such as explicit charges. They may also be unwilling to spend the time and effort researching a new plan, resulting in inertia and the failure to switch out of poorly performing plans. Employees who are enrolled into schemes that are selected by their employer may not have the option of switching provider.

Although the threat of switching can encourage competition among providers, participants should avoid excessive churning of pension portfolios, as this can create additional administration expenses for providers and unnecessary trading costs for members.

Failure to exploit potential economies of scale

As noted in OECD (2016[19]), there has been a significant decline in the number of private pension funds and plans since 2005 in several OECD member jurisdictions. A number of countries with mature pension systems, including the Netherlands and Australia, are actively encouraging further consolidation of funds. Pension plan size varies considerably across OECD and IOPS members and it is not clear that there is an "optimal" plan size, but the existence of high fixed costs in pension plan administration implies that larger plans will be more efficient. They could also benefit from better governance: the decrease in the number of pension funds in the Netherlands since 2005 from 800 to less than 400 has been accompanied by tougher qualification requirements for trustees.

Administration costs are largely driven by the size of the plan, as shown in Figure 3.3. Cost items such as IT, oversight, communications and collections have a substantial fixed element and there is evidence that scale economies can be captured relatively quickly. Di Gialleonardo and Marè (2015[20]) found that the administrative costs per participant of supplementary closed pension funds in Italy fell from EUR 31.43 for funds with 10 000 - 50 000 members to EUR 19.63 for funds with over 50 000 members. Furthermore, there were no diseconomies of scale for bigger funds. Bikker (2013[21]) reached similar conclusions for Dutch pension funds and found also that the size at which funds could continue to reap scale economies was increasing over time, suggesting that "average administrative costs per participant now decline without limit", in part because of more expensive technology and more complex regulation.

Figure 3.3. Administrative costs of pension management per participant relative to total participant numbers in the Netherlands

Note: Excludes pension funds with > 100 000 participants; all such funds had administrative costs below EUR 120 per year per participant.

Source: Pensioen Federatie (2016[22]).

Both studies found that the evidence for economies of scale in investment costs was less conclusive. Sources of economies of scale in investment costs include negotiating power with asset managers and other suppliers such as custodians, spreading semi-fixed costs such as research on external managers, and bringing investment management in house, especially for complex asset classes. Table 3.5 shows the fees charged by external managers to different types of US institutional investor in a range of asset classes – bigger investors pay less than smaller investors in every case.

Di Gialleonardo and Marè (2015[20]) suggest that the relatively low level of management fees for closed funds in the Italian market might leave fewer opportunities for further cost reduction and that larger funds implement more complex and so more costly investment strategies. Bikker (2013[21]) found that larger funds used more expensive asset classes and that scale economies were exhausted at an asset size of EUR 690 million. However, Broeders, van Oord and Rijsbergen. (2016[24]) found significant and sustained economies of scale for all pension funds with assets over EUR 20 million. Dyck and Pomorski (2011[25]) found that investment economies of scale worth between 43-50 bp per year for the largest DB funds, driven by insourcing asset management and increasing allocations to alternative asset classes.

However, there may be diseconomies of scale arising from concentration risk (investors may have limits on the proportion of an underlying asset that they can own, or of the share of an individual manager’s business that they can represent), from market impact (market delays or adverse price movements from passing large trades) or from the temptation to go into riskier asset classes. Larger DB plans may find it difficult to implement LDI strategies because there are fewer counterparties for large derivatives trades. Reducing the number of plan providers too far could stifle innovation by both providers and external investment managers, although this is unlikely to be a problem in countries such as Australia, the United States or the United Kingdom where there are multiple providers (Axa, 2013[26]).

Ineffective fee structures

Fee structures should be designed to provide the right incentives to managers to achieve the best outcomes for members and plan sponsors. In traditional asset classes, investment manager fees are usually ad valorem, that is they are calculated as a percentage of assets under management. Managers of alternative assets, such as hedge funds or absolute return funds, may charge a combination of an asset-based fee and a performance-based fee. Neither fee structure will necessarily reward managers appropriately.

Ad valorem or asset-based fees provide few incentives for investment managers to become more efficient or to share efficiency gains with clients. They can reward poor performance and penalise good performance. In other words, this fee structure may not provide good value for money. For example, a portfolio manager who generates returns of only 8% when the market rises by 10% will earn more in absolute terms, while one who returns -10% when the market falls by 50% will earn less in absolute terms, even though he has provided a much better service to his clients by limiting their losses. As shall be discussed below, performance-based fees can help avoid this problem but need to be carefully designed so that managers do not keep a disproportionate share of the value they create, and they do not distort incentives.

Greater transparency to influence the level of costs and charges

Making costs and charges more transparent has been a key objective of policy in all jurisdictions, with measures to improve reporting, communication and benchmarking of investment costs and plan charges. Such measures target market failures arising from lack of engagement by plan participants, complex and opaque charging structures, and weak governance.

Cost transparency has been successful in encouraging providers to monitor and control their costs in both DB and DC occupational pension markets. In the Netherlands, pension funds are required to provide granular information on administration and investment costs and this has led to greater cost awareness among sponsors and fund boards, resulting in better outcomes (see Box 3.1). Funds must report on their costs to a level of detail similar to that in Table 3.2, and the Dutch National Bank benchmarks the costs and investment performance of plans against their peer group and calls “expensive” plans to account.

Disclosure requirements are accelerating in many jurisdictions (Box 3.2). However, cost transparency can be cumbersome. The potential list of total direct and indirect investment costs is extremely long – the FCA Institutional Disclosure Working Group (IDWG) came up with over 300 discrete cost items – but many indirect costs are difficult to capture and may not be meaningful. Governing bodies, sponsors, regulators and members may be overwhelmed by too much data. In addition, some costs are relatively small, so may not be worth measuring; others are relatively large but cannot easily be compressed even once they are revealed. For example, stamp duty is a large part of trading costs, but cannot be avoided. An analysis of Local Government Pension Scheme costs in 2014 estimated that direct equity trading costs on a portfolio with turnover of 140% per annum were at least 75 bp. Of this, only around 10 bp was commission (which can potentially be squeezed through negotiation with brokers) while 65 bp was stamp duty and taxes. This figure was considerably higher than the direct investment management costs of 25.4 bp and administration costs of 12.6 bp (Sier, 2014[27]).

Regulators therefore need to decide which costs are relevant and whether and how they should be made transparent. Transparency does not have to create complexity. Meaningful cost reductions can be achieved even before detailed information about indirect costs is available. From 2011 to 2017, pension providers in the Netherlands were able to use standard, proxy spreads to measure transaction costs in fixed income portfolios and to use entry and exit charges as a proxy for look-through transaction costs. Despite the lack of detailed information about indirect costs, cost awareness increased and total costs fell.

A number of jurisdictions have introduced measures to improve transparency about DC costs and performance, in order to raise member engagement. The Danish government-backed site www.pensionsinfo.dk provides members with information on their individual accounts including direct and indirect administration and investment costs and past returns. The 2015 Communications Act in the Netherlands requires schemes to provide standardised information to members.

This type of information can enable members to make more straightforward comparisons between funds. In Mexico, individual pension statements include information on net-of-fees returns across AFORES (providers). Both Hong Kong (China) and Italy provide comparison websites and therefore impose a degree of standardisation on the way information about fund charges is collected and displayed.

Demands from the public or press for better disclosure of costs have spurred transparency initiatives in several countries, including Israel and the Netherlands. However, a clear signal from the regulator has been the decisive factor in ensuring that providers sign up to disclosure standards. The Netherlands reporting framework was developed by the pensions industry in response to demands from the regulator. The United Kingdom relies on a combination of regulation and industry co-operation, while binding transparency requirements have been introduced in Australia and the European Union. The United States approach is different: the law requires fiduciaries to act prudently and in the interests of participants, and implicitly relies on market forces, supervisory activity including disclosure and the right of legal redress by members to contain costs.

Disclosure will only be effective in bringing down costs and fees if providers, sponsors and members understand and act on the information. Most OECD and IOPS member countries have found that disclosure-based initiatives, giving members and sponsors the information they need to apply competitive pressure to pension providers, have had some effect but not enough to align charges with costs. Therefore, they have supplemented these initiatives with measures to regulate fees or influence the structure of the market or of the providers themselves.

Benchmarking to help improve outcomes for members

Improved transparency can help align costs and charges across funded pension systems but will not necessarily reveal whether there is further room to improve outcomes at the level of the individual provider. Would it be possible for a given pension provider to deliver more speedy and accurate administration services at the current cost, or is the provider offering expensive services that members did not value? Could investment costs be reduced to increase the net returns generated by the investment strategy?

Benchmarking providers could give regulators, members and employers and sponsors information about the relationship between costs and outcomes of their pension fund that would complement other policy efforts to bolster market mechanisms.

Finding a relevant point of comparison for pension funds is not straightforward. Cost and quality can vary widely, depending on the nature of the pension arrangement, the complexity of its administration and investment operations, its membership structure and the design and implementation of its investment strategy. Benchmarking can lead to herding behaviour in investment and less innovation in administration, resulting in weaker competition among providers, and might encourage short-termism in investment. However, peer groups, proxies and other reference points can be constructed.

It should be relatively straightforward to assess whether there is scope to improve outcomes being delivered by DB arrangements. DB providers have a clear target – to grow assets so that they meet future liabilities – and often have a strategic asset allocation to guide long-term investment strategy. Net portfolio returns give an indication of the quality and cost of the investment strategy and can be compared to market returns for equivalent asset classes, or to the investment gains made by other DB funds with similar portfolios. Administration costs can be identified and the quality of administration services can be judged using largely quantitative criteria. Absolute levels of cost and quality will vary across plans, but comparisons can be made across providers with similar activities.

Table 3.7 shows the information revealed in the financial statements of two large occupational DB plans in the Netherlands and two in the United Kingdom. Table 3.7 does not show that one fund offers better value than another, but it highlights areas where there may be scope for funds to reduce costs. Administration costs of the four funds range from 4 bp to 10 bp of assets under management and direct investment costs from 10 bp to 19 bp. The United Kingdom does not require funds to report total investment expenses.

If DB providers understand their cost structures and they are able (or required) to benchmark the cost and performance of their administration and investment activities against relevant peers, they will be able to see where there is scope to improve outcomes. If there is an incentive to improve relative performance – or a sanction for being at the bottom of the range – then it is even more likely that pension providers will deliver good and improving value.

The supervisor enforces both transparency and benchmarking in the Netherlands. Pension plans are required to explain deviations from the cost structure of the peer group. The average investment costs for all pension funds were 58.5 bp in 2015; administration services for members (i.e. excluding the costs of governance and oversight of the plan) cost a further 7.5 bp; total costs across all funds ranged from 15 bp to 200 bp.11

In Switzerland, pension funds have been required to report their Total Expense Ratio (TER) in their annual reports since 2013;12 in that year they were also obliged to collect TER data from underlying vehicles in which they were invested and publish a blacklist of those who did not comply. The Swiss supervisor, OAK, is expected to publish comparative data on more than 2 000 pension funds in the near future.

It is more difficult to assess and compare across DC funds. They have a wide range of administration service levels, investment strategies and liquidity constraints, especially in DC systems where there is an element of choice. Employers or employees may be able to choose their provider, they may select bespoke investment strategies or require different levels of administrative support. There may be intermediaries between the member and the provider. This means that a wide range of DC outcomes is possible, making it complicated to construct peer groups.

Unlike DB funds, which have a unique long-term investment objective (meeting liabilities), DC plans can have different goals. PPI (2016[30]) noted that while high charges erode returns, members may prefer a higher priced, lower volatility investment strategy to either a lower cost strategy or a higher risk strategy, depending on their risk tolerances and other sources of retirement income. Where members are allowed to choose between different strategies offered by their provider or to build their own asset allocation from a range of underlying investment options, they could have widely differing investment outcomes in terms of returns, risks and costs (Table 3.8). Absolute investment gains and performance relative to market benchmarks or peers could also vary markedly over different time horizons.

Given the difficulty of creating DC peer groups of funds with similar administration and investment activities, a simpler method of assessing the relationship between provider costs and member outcomes could be to use default funds as a reference point for DC plans. Default funds are intended to provide an investment strategy that is suitable for the majority of DC members, which limits the design options. They are usually required to be low cost. Harrison, Blake and Dowd (2014[31])suggest that 50 bp is a reasonable TER for a default fund operating at scale. The OECD Roadmap for the Good Design of Defined Contribution Pension Plans recommends establishing appropriate default investment strategies, in particular lifecycle strategies.

Not all pension systems include a unique default fund. Countries may have a number of competing default funds offered by different providers, or no default option. Other, similar types of pension plan could serve as a benchmark, such as the lifecycle funds that are offered by the Thrift Savings Plan in the United States. Alternatively, a proxy portfolio that follows the principles of a default lifecycle option could be constructed as the basis for comparing the investment design and cost of actual DC arrangements. It would not, however, give information about administration activity.

DC plans could be assessed against the fees (to members and employers) and net returns of the default strategy or proxy. Members of other DC plans could reasonably ask their provider whether they were on track to achieve equivalent outcomes to the default for a similar cost, and if not, why not.13

Default funds with very similar investment designs can deliver different investment returns and charge a wide range of fees. Table 3.9 shows the investment strategy and fees for selected default funds or proxies in different jurisdictions. Return targets and investment performance differ between the funds, but the investment designs are similar: all include a form of de-risking as the member approaches retirement age. Fees vary from a minimum of 3.8 bp (United States) to a maximum of 95 bp (Hong Kong, China).

Benchmarking DC plans against a default could help to identify outliers in terms of outcomes, but it would remain complicated to draw conclusions about the potential to improve outcomes at individual funds in jurisdictions with a large number of providers and a wide variety of plan designs. Therefore, benchmarking should complement and not replace measures to align costs and charges at competitive levels.

Investment strategy reporting

Investment costs do not exist in isolation from the investment strategy. They should therefore be reported together. Indicators that take account of investment performance, manager skill and asset class are needed in addition to cost information to assess whether a fund’s investment expenses are reasonable. This assessment of reasonableness can be made both in absolute and in comparison to the costs incurred by other portfolios.

These indicators would also make it possible to tie rewards more directly to outcomes. Investment managers and strategies can be analysed in terms of the returns generated per unit of risk taken, fees per unit of return and the amount of return that is retained in fees.

This analysis is robust across different asset classes and investment styles. For example, passive management will score highly in terms of cost versus risk because passive portfolios track the reference benchmark closely and have low management fees and transaction costs. However, passive management scores poorly in terms of cost versus alpha because passive funds are not designed to outperform the market. This type of analysis makes it easier to improve the value for money of investment activities by tying rewards for investment managers to outcomes.

Pricing regulation

Pricing regulation can contribute to disclosure efforts by simplifying fee structures, making it easier for participants to understand what services they are paying for and exactly how much they are paying. Alternatively, pricing regulation can force providers to take action on costs, by imposing limits on what they can charge.

DC arrangements are subject to this type of price regulation in several countries. Australia, Estonia and Hong Kong (China) have introduced low-cost plans. Chile, Sweden, Turkey and the United Kingdom have imposed charge caps. Mexico simplified fee structures in 2008 by stopping the use of fees on contributions, so that AFORES can only levy a fee calculated on assets under management, which was also capped. Costa Rica replaced its mixed fee structure with a single, asset-based fee in 2011. In Denmark and Poland, fees are levied on both contributions and assets under management, whereas in Spain only asset-based fees are permitted.

Different fee structures may be appropriate at different points in the development of the pensions system or the growth of a pension fund. Assets grow over time, so asset-based fees reward established providers who have had time to collect and grow contributions. Contribution-based fees enable new DC schemes to raise revenues more quickly, but may not cover costs as plans mature if members leave or defer. Pricing regulation may therefore need to evolve as pension systems mature.

However, unless both asset-based and contribution-based fees cover a significant part of total costs, including indirect and implicit costs, this may not be sufficient to deliver better pension outcomes. Therefore, more direct controls over pricing, such as charge caps, may be introduced. The United Kingdom introduced a charge cap of 0.75% of assets under management on workplace default funds from April 2015. The cap applies to all direct and indirect administration and investment costs, but does not include transaction costs. As well as putting an upper limit on default fees, the cap also helped to raise awareness of high charges in other DC arrangements. The charge cap appears to have been effective in reducing fees for DC funds: all qualifying schemes (those that are eligible to be used as defaults) are now priced below 75 bp and the prices of other schemes have also fallen.

Charge caps set a clear and simple standard for member charges but can have unintended consequences. If the cap is set too high, charges tend to rise to the level of the cap, as was seen earlier in the UK market when stakeholder pensions were introduced with a maximum charge of 1%, which quickly became the market price for all similar retirement savings products. In a similar manner, the German government’s estimated total charge for Riester products of 10% has become the de facto standard charge (Better Finance, 2016[32]). If the cap is set too low, plan providers might try to cut costs by offering lower-quality plan designs or by reducing the number of transactions they undertake, even when the trades would be in the best interests of members.

If the cap does not include all direct and indirect costs, then providers might have an incentive to exaggerate uncapped costs in order to compensate for any lost profits in areas that do fall within the scope of the cap. Turkey reduced its charge cap quite significantly in 2013 and introduced a new charging structure consisting of a capped ad valorem fee that varies by asset class (ranging between 1.09% and 2.28%), a fee on contributions and a small fixed on-boarding fee. The cap takes into account all fees that pension providers can earn from participants and total fees are capped from the sixth year of a contract.

Establishing the correct level of the cap is especially complicated in markets where providers have different cost structures. Setting the cap in line with the cost structure of large, vertically integrated providers might squeeze out smaller providers. A low cap can also discourage new entrants. However, a relatively high cap could enable lower-cost providers to generate excessive profits if they chose to price at the level of the cap, unless there was pressure from participants or other stakeholders to reduce charges.

In order to capture economies of scale on behalf of members, charge caps should not be static. Costa Rica initially capped fees at 1.1% and this will reduce to 0.35% by 2020. In Estonia, fees must be reduced by 10% for each EUR 100 million increase in assets.

Charge caps focus on fees rather than design, putting the onus on providers to come up with a pension design that is profitable for them within the cap. An alternative form of pricing regulation is to introduce standardised pension designs with regulated fees. MySuper products in Australia have simplified product specifications and a limited range of fees, although this has not led to a high degree of standardisation in terms of product attributes, costs and outcomes. The Default Investment Strategy introduced by Mandatory Provident Fund (MPF) schemes in Hong Kong (China) in 2017 has a fixed asset allocation and maximum fees. MPF schemes were already required to offer a low-fee fund with no restrictions on design but with maximum management fees of 1% of assets.

Another form of price control is to change the basis for charges. Member-borne commissions have been banned in UK DC schemes; similar measures have been introduced in Australia and the Netherlands. Mexico changed the incentive structure of sales agents to limit negative-yielding switches.

Introducing performance fees to achieve better investment outcomes

Asset-based fees can fail to provide incentives to investment managers to become more efficient or to share efficiency gains with clients. Performance-based fees are one solution. Unlike price caps which specifically limit charges, performance-based fees aim to better align the interests of investment managers and pension fund members and sponsors. Hamdani et al. (2017[33]) use evidence from a regulatory experiment in Israel to demonstrate that funds with performance fees exhibit higher risk-adjusted returns than funds with asset-based fees operating in different competitive environments.

Performance-based fees can complement other pricing regulation such as price caps. This may be useful since ad valorem fees can limit the effectiveness of price caps. The average net profit margins of Mexican Pension Fund Providers (AFORES) rose from 33.5% in 2013 to 37.4% in 2017, despite pressure from the regulator, CONSAR, that pushed average fees down from 1.29% to 1.03% of assets under management over the same period (Table 3.10). Assets under management grew rapidly, implying that the AFORES were reaping economies of scale, and the cost of acquiring new contributions fell (Table 3.10). CONSAR will therefore start considering cost indicators, such as net profit, return on equity and return on assets, when assessing fee proposals from the AFORES. It also intends to introduce performance-based assessments, by taking into account historical investment performance, and investment performance relative to a benchmark portfolio.

Performance-based fees must be structured in such a way as to give the right incentives to pension funds and their investment managers. Rewards should be paid for delivering high returns per unit of risk taken, and a fair share of returns should stay in the portfolio rather than being paid out in fees. The basic parameters of a performance fee design (there are many more complex structures in operation) are:

• The fee base (e.g. is the fee calculated on investment gains or on assets under management (AUM)) and the fee rate

• The performance reference point, i.e. the minimum return the fund must earn before a performance fee is due

• The measurement period, i.e. the length of time over which performance is calculated

• The high watermark (HWM), if any, which is the last highest value that the fund has reached

The performance reference point can be fixed (a “hurdle rate”) or variable (e.g. a market index). It is important that the reference point reflects the investment strategy of the fund. An equity manager should not be paid a performance fee for outperforming a cash benchmark, since the bulk of any outperformance is likely to be due to the investment characteristics of equities relative to cash rather than to manager skill. Pension fund management companies operating in Latvia’s mandatory funded scheme are benchmarked against a composite debt and equity index. In Poland, the reference point is competitors’ performance. Performance fees are calculated as 0.06% of assets under management multiplied by the “percentage premium ratio”, or the relative performance of the fund versus its peers. The ratio is 1 for the best performing fund – which receives a performance fee of 0.06% of AUM – and 0 for the worst performer – which receives no performance fee. Other funds receive an amount below 0.06% of AUM that reflects their relative position.

High watermarks (HWM) are intended to ensure that managers are not paid a performance fee for good performance in one period unless they have caught up any underperformance in previous periods. The fund value must always be above the previous HWM for a performance fee to be earned. The Czech Republic and the Slovak Republic both use HWM in the calculation of performance fees; the Czech Republic sets the HWM as the historic maximum value of the fund while the Slovak Republic looks back over a shorter period.

High watermarks can be effective in preventing a “double fee”. Such double charging is a feature of ad valorem fees, where the fee basis is the full AUM. An individual, who joins a DC scheme at the age of 25 and retires at the age of 65, will pay annual fees on his first year’s contributions 40 times. Fees will also be paid each year on the prior years’ investment gains. Box 3.3 illustrates the application of a performance fee with a fee base of investment gains, a fee rate of 20%, a hurdle rate of 5%, a measurement period of one year and a HWM. Box 3.3 also shows that a high watermark can be effective in preventing a “double fee” (i.e. the fund manager is not rewarded for the Year 1 performance in both Year 1 and in Year 4). Conversely, if an asset class suffers a sharp fall, a manager might take several periods to get back to the HWM even if he outperforms the market consistently. This outperformance would not be rewarded. This can be resolved through the use of “relative high watermarks”, relative to different asset classes.

However, this may not be enough to create alignment of interests. The upside potential of a strongly performing manager is potentially unlimited, while the downside risk is limited – especially since most performance fees are payable on top of an asset-based fee. Fee caps can help to create more symmetrical payouts; so can reserve systems, whereby a portion of any performance fees earned are “banked” in a reserve pool, and used to top up the portfolio if the manager subsequently underperforms. This cushions investors against a run of poor returns and leaves the possibility for the manager to earn back lost reserves and fees once performance improves.

There are practical difficulties in implementing performance fees. Parameters need to be negotiated and calculating fee accrual can be complex; performance fees may be higher than asset-based fees for strong managers and so increase investment expenses. They may also pose stability risks, as performance fees are more volatile than asset-based fees. Given the importance of investment manager fees and performance to outcomes, they are nonetheless a potentially powerful means of improving the value for money of the investment activities of pension funds.14

Box 3.3. Simple performance fee

The chart shows the path of a pension fund portfolio, with a starting value of 80 rising to 100 at the end of Year 1, falling to 82 at the end of Year 2, climbing back to 90 at the end of Year 3 and finishing Year 4 with a value of 110.

The fund manager receives a performance fee of 20% for performance above a hurdle rate of 5%, subject to a high watermark, HWM. The fee is calculated on the absolute increase in the value of the portfolio.

At the end of Year 1, the fund manager has earned a performance fee of 4. This is equal to 5% of AUM and 20% of investment gains. Absolute performance is 20 (100-80), equating to an investment return of 25% (20/80) and the hurdle rate is 5%. The performance fee is therefore calculated as [20 × (25% - 5%)]. The high watermark rises from 80 to 100.

The fund ends Year 2 with a negative performance and fails to meet the hurdle rate, so no performance fee is due. At the end of Year 3, the fund value has risen from 82 to 90, a return of 9.75% that is almost double the hurdle rate. However, the value of the fund remains below the HWM so no performance fee is due.

At the end of Year 4, the fund value has risen by 20, or 22%. The fund manager earns a performance fee of [20 × (22% - 5%) = 3.4]. This is equivalent to 3% of AUM and 17% of investment returns. The new HWM is 110.