6. Sustainable risk sharing in retirement income arrangements

The consequences of an unbalanced distribution of risks within retirement income arrangements have never been more evident than in the current economic environment of low growth and low returns and the demographic context of ageing. Traditional defined benefit (DB) arrangements, which expose the provider to all investment and longevity risks, are facing threats to their solvency due to historically low interest rates, increasing longevity, and a series of financial crises that have left them unable to fulfil their retirement income promises to their members. This has led to a shift to individual defined contribution (DC) arrangements, which put all of these risks squarely on individuals’ shoulders. Individuals are therefore left with no retirement income security, with the only option to obtain a stable income in retirement typically being a traditional annuity offering a rather low guaranteed income due to the current low interest rate environment. Neither of these extremes – either the provider or the individual fully bearing all risks – is sustainable, and both ultimately result in a retirement income that is lower than could be achieved by allowing risks to be shared.

The solution to this problem is to find a way to balance the distribution of risks within the retirement income arrangement in order to maximise the retirement income it can provide while offering the desired level of security of benefits. There is always a trade-off between these two. While risk sharing increases the risk-bearing capacity of members and their retirement income potential, security entails implicit and explicit costs in terms of lower investment returns and the cost of security mechanisms that aim to enforce any guarantees provided. Finding such a balance is therefore a delicate matter of exploiting the positives while minimising the negatives of both risk sharing and risk protection. The desired outcome of a higher and more stable level of retirement income must guard against the negatives of seemingly unfair risk transfers among participants and the reduced income that guaranteed arrangements can pay.

The regulatory framework must promote the sustainability of retirement income arrangements with the clear objectives of risk mitigation to increase potential retirement incomes for all participants, continuity through an even distribution of risks among participants, and security to ensure that promises can be met. It must ensure that the risk sharing among participants is advantageous and fair – for both current and future generations – and that the security mechanisms in place provide a reasonable assurance that the provider will be able to meet any income guarantees offered. If certain participants in the arrangement feel they bear an undue risk burden, the arrangement will ultimately fail. If security mechanisms do not adequately enforce income promises, albeit through lower potential retirement income, those promises will ultimately not be kept. In both cases, the end result it the same: individuals will be back to bearing all risks for financing retirement on their own.

This chapter explores the different ways that retirement income arrangements can distribute risks among participants and providers, and the implications that design and regulatory features have on who bears those risks and the level of retirement income that can be achieved. The first section provides an overview of the different types of retirement income arrangements that exist along the risk-sharing spectrum, from arrangements where participants fully bear all risks individually to those where the sponsor/provider fully guarantees the retirement income benefits. The second section discusses how risk sharing among participants in a collective arrangement can increase comparatively the potential retirement income for participants, and the implications that design has on fairness with respect to how stakeholders share these risks. The third section highlights the risk of continuity for retirement income arrangements that do not evenly distribute risks across participants, and explains how the valuation framework and funding requirements in place can promote sustainability and continuity. The fourth section considers the increased security that external retirement income guarantees can offer, albeit at a cost, and the security mechanisms that need to be in place to back these guarantees. The final section concludes with a discussion on how the regulatory framework can support the objectives of fairness, continuity, and security to promote sustainable risk sharing in retirement income arrangements.

Different types of retirement income arrangements exist along a risk-sharing spectrum, from arrangements where participants fully bear all investment and longevity risks individually to those where the sponsor/provider fully guarantees the retirement income benefits (Figure 6.1).1 Individual account arrangements sit on one end of the spectrum. For these types of arrangements, individuals are fully exposed to all risks related to pension income. At the other end of the spectrum lie arrangements where all risks are fully borne by a sponsor or third party, as is the case for traditional DB arrangements or annuity products that fully guarantee a certain level of retirement income for life.

In between these two extremes lie collective arrangements. These are retirement income arrangements in which risks are shared either collectively among a group of individuals, or between the individuals and the sponsor/provider who offers a minimum guarantee. This latter category can further distinguish between risk sharing in the accumulation phase and in the pay-out phase. The first involves the sharing of primarily investment risk, while the second also shares the longevity risk with members. Collective arrangements can be further broken down within the categories of risk sharing based on some main features that determine the definition of the retirement income benefits received. Table 6.1 provides a brief description of the main characteristics of the different types and a common example of each.2

Arrangements where risks are shared collectively among members include collective target benefit plans or collective pay-out plans. The target retirement income benefits in the former arrangement are accrued during an accumulation phase, and benefits accrued by both pensioners and younger members may be reduced if funding is insufficient. An example of this type of arrangement is the collective defined contribution (CDC) arrangement. For collective pay-out arrangements, risks are only shared during the pay-out phase, with the accumulation phase typically structured as an individual DC plan. The initial benefit depends on the level of assets accumulated at retirement, and it can be adjusted going forward depending on the investment and longevity experience. An example of this type of arrangement is the variable pay-out annuity, or a tontine-type annuity.

Arrangements where investment risks only are shared between members and providers over the accumulation phase include arrangements where retirement income benefits are accumulated either with reference to an index, to a minimum absolute return, or to a salary-based retirement income benefit formula. Indexed accumulation arrangements define the benefit in terms of an accumulated amount of capital that is credited with a rate of return that references some economic variable such as interest rates or wage growth. Normally, the credited return cannot be negative so accumulated benefits cannot decrease from one period to the next. At retirement, members can usually convert the accumulated amount of capital into a guaranteed retirement income for life. An example of this type is a cash balance pension plan.

With minimum return arrangements, the amount of accumulated capital depends on actual underlying investment returns, subject to a floor over the investment horizon. As such, returns over any given period could be negative, but the providers guarantee a minimum level of return for the calculation of the retirement income benefits. As with indexed accumulation arrangements, the balance accumulated in these types of arrangements can usually be converted into a life annuity at retirement. An example of this type of plan is the occupational arrangements common in Switzerland that provide a guaranteed minimum return in accumulation and offer a guaranteed minimum conversion rate to convert the accumulated balance into retirement income payments.

For minimum salary-based benefit arrangements, the retirement income benefits are based on the greater of the guaranteed income that could be purchased with assets accumulated in the account or a DB-like formula based on salary. DB underpin plans are an example of this type.

Arrangements where both investment and longevity risks are shared between the members and the provider during the pay-out phase include arrangements where expected benefits are paid conditionally on funding levels or where discretionary benefits are paid if investment and/or longevity experience has been favourable. Generally, the accumulation phase of these types of arrangements can be structured in a way similar to any other type of arrangement, with the provider bearing all risk or sharing the investment risk with the members. The main difference between conditional benefit arrangements and discretionary benefit arrangements is that for the former the total expected benefit is defined in advance, whereas for the latter this is not necessarily the case. Benefits in both types of arrangements, however, are subject to a minimum level. An example of a conditional benefit arrangement is one that provides conditional indexation. An example of one with discretionary benefits would be participating annuities that share a portion of the provider’s profits with the members.

Arrangements where risks are fully borne by the provider include those where benefits are defined with reference to salary and those where retirement income benefits are defined in an actuarially neutral manner. For the former arrangements, each contribution made gives the member the right to a guaranteed pension benefit that is defined as a percentage of their salary. An example of this type of arrangement, of course, is the traditional DB arrangement. In the latter arrangement, for each contribution the member earns a benefit that is calculated taking discount rates and mortality rates into account, as with a traditional life annuity. For these, contributions can be made regularly over the accumulation phase or all at once at the point of retirement.

Collective retirement income arrangements with risk sharing offer real benefits over individual arrangements in terms of risk mitigation and the level of expected retirement income, even without an external guarantee from a provider. Nevertheless, it is important that the distribution of risks among the participants of a collective arrangement is perceived to be fair.

The ability for a collective retirement income arrangement to pool risks and smooth funding shocks over time can significantly mitigate the risks that individuals would otherwise bear on their own. This increases the certainty that they will be able to receive a reasonable level of retirement income for life. The mitigation of the risk at the individual level allows higher retirement incomes to be paid, and ultimately increases the collective capacity of the arrangement to invest in higher risk assets that will provide an even higher expected retirement income overall.

Collective risk sharing can be limited to within a specific cohort or shared across cohorts or generations. Risk sharing within cohorts functions as a regular insurance contract through the pooling of a large number of individuals. Idiosyncratic longevity risk, or the risk that any individual will live longer than the average life expectancy, is easily mitigated by pooling risks within a given cohort. With this type of risk sharing, people who die earlier subsidise those dying later. This means that all participants can increase their retirement income because they do not need to plan to have additional savings to cover the risk of living beyond the average life expectancy. Arrangements that spread risks across several cohorts or even generations can share both investment and systemic longevity risks – and in some arrangements even wage and inflation risks – among participating members. Table 6.2 describes the nature of the risks shared across different groups of participants and the objective and mechanisms of this risk sharing.

Sharing risks across cohorts and generations allows for intertemporal smoothing of shocks that cannot be mitigated periodically through risk pooling. Investment shocks in particular can only be smoothed over time and cannot be diversified through pooling a larger number of participants. Systemic longevity risk (i.e. the risk that all members of a cohort may live longer than expected) can also be smoothed over time and shared across cohorts rather than be borne by individual cohorts.

The main objective for risk sharing across cohorts is to provide retirement income stability. This stability is achieved through smoothing features incorporated into the design of the retirement income arrangement that aim to avoid frequent and/or large retirement income benefit adjustments due to changes in funding levels. Examples of such mechanisms include corridors and amortization periods. With these mechanisms, younger generations effectively provide a subsidy to retirees, which can be more or less temporary.

The main objective of risk sharing across generations is to improve welfare and maximise the expected utility of all participants. Given their long duration and long-term outlook, retirement income arrangements are one of the few types of arrangements that can allow for sharing risks across generations, and even across non-overlapping generations. In an ideal setting, this can allow for ex-ante welfare gains, as it allows for participants to trade risks that they would normally not be able to share due to incomplete markets, as well as to share funding mismatches due to investment and longevity shocks over a long time horizon. These arrangements theoretically allow young participants to borrow against their future human capital and thus take advantage of the equity premium earlier in life as compensation for securing the retirement income of current retirees (Bovenberg et al., 2007[1]). Such risk sharing relies upon intergenerational solidarity and the participation of future generations who will inherit any funding mismatch, whether positive or negative, and therefore can be viewed as a social contract.

Intergenerational risk sharing increases the risk bearing capacity of the retirement income arrangement and the demand for higher risk investments (Bonenkamp and Westerhout, 2014[2]; Cui, de Jong and Ponds, 2011[3]). Welfare gains from arrangements that allow for intergenerational risk sharing largely come from the ability for the arrangement to take on more investment risk without increasing the risk borne individually. Even though total investment risk exposure for the arrangement is higher, this is partially offset by better intertemporal diversification, so individuals can have the same level of risk exposure while benefiting from higher expected returns (Gollier, 2008[4]). Nevertheless, the conditions required to achieve optimal outcomes from intergenerational risk sharing do not always materialise, and the design of the arrangement also needs to take a realistic view of the context in which it operates.

The assessment and measurement of risk and value transfers is necessary to determine whether the design of the retirement income arrangement is seen as fair vis-a-vis different cohorts of participants. The design and features of the arrangement will determine whether the allocation of risks across the participants in the arrangement is efficient and fair ex-ante, and that value transfers are acceptable ex-post. However, transfers that occur in retirement income arrangements often lack transparency and adequate assessment.

The definition of fairness used for assessing risk and value transfers depends on the objective of the retirement income arrangement. Fairness could be defined as having no inter-cohort transfers, with each cohort bearing their own risk. Alternatively, fairness could be viewed through a lens of expected welfare improvements for all cohorts. In this case, fairness is best assessed from the inception of the arrangement (ex-ante), as after a funding shock (ex-post) there will always be value transfers that will affect cohorts differently depending on the source, direction and magnitude of the shock.

Once fairness has been defined, the potential risk transfers need to be assessed in order to establish whether they are fair. The magnitude of transfers can vary depending on the specific features of the plan and the source of the funding shock. Assessment of fair design will need to consider both investment and longevity shocks, as these two risks do not have the same implications for every cohort. Additionally, the demographic structure of the retirement plan can influence how much risk is borne by the various cohorts.

Assessing fairness is only one angle, however. The long-term continuity and sustainability of the retirement income arrangement given the size of the potential transfers of value must also be a consideration. As such, other criteria need to be considered in addition to fairness.

Retirement income arrangements face a risk of sustainability and continuity if value transfers within the arrangement are too large, as the intergenerational solidarity required for the arrangement to operate could break down. Any retirement income arrangement that shares risks will have value transfers that make some groups worse off ex-post after a funding shock, even if it is fair and welfare improving for all participants ex-ante. Risk sharing in retirement income arrangements need minimum funding requirements to limit the size of risk transfers and ensure the continuity of the arrangement.

There is a tension between maximising welfare ex-ante and ensuring the sustainability and continuity of a retirement income arrangement ex-post. Optimising outcomes via expected welfare improvements tends to favour heavy investment in equities, with the level of expected welfare gains increasing with the level of equity premium (Cui, de Jong and Ponds, 2011[3]). Without constraints, the asset allocation into equities to maximise expected utility can go up to 100% (Gollier, 2008[4]). Obviously, such a strategy would present a high risk of insolvency and threaten sustainability, therefore additional constraints such as a minimum funding level need to be considered.

Continuity risk is a particular concern when risks are transferred primarily from the older generations to the younger ones. If the funding mismatch is too negative, the younger generations may come in knowing that they will lose from the retirement income arrangement and prefer to default on their obligations to the older generations and not participate in the arrangement. Even if participation is mandatory, they may exert political pressure to change the arrangement or alternatively adjust their labour supply by changing their employer, their industry or even their country in order to avoid participating (Bovenberg and Mehlkopf, 2014[5]). Indeed, the more flexible the labour supply is, the less risk-bearing capacity society has and therefore the lower the demand for risky assets is. Flexible labour markets therefore reduce the potential gains from risk sharing (Bovenberg and Mehlkopf, 2014[5]).

There is also a risk that the current rules of the system will be changed in a way that would aggravate the continuity risk and increase the probability and magnitude of future shortfalls. When deficits exist, there may be a reluctance to improve funding through reductions in retirement income benefits, which could be delayed or modified due to political pressure to avoid penalising the current generations. However, even large positive transfers can put the continuity of the arrangement at risk. Older generations will be tempted to consume large positive buffers rather than leaving them to reduce the risks for the younger generations, and can exert pressure to release those buffers. This increases the probability that future generations will not see an advantage to participate in the plan and that the arrangement will not be sustainable (Bovenberg and Mehlkopf, 2014[5]).

In order to avoid the breakdown of the social contract and disincentives for future generations to participate in the retirement income arrangement, constraints need to be put in place to limit the risk of a significant funding mismatch. Minimum funding levels can be imposed to this effect. Nevertheless, this will reduce the risk-taking capacity of the retirement income arrangement, and thereby the potential welfare gains from risk sharing (Gollier, 2008[4]).

Defining a minimum funding requirement must also involve defining a valuation methodology with which to calculate the funding requirement. Since the funding requirement is normally defined as the value of assets over the value of liabilities, the way in which these values are calculated will directly determine the calculation of the funding level and the assessment of whether there are sufficient assets to finance the promised retirement income.

Arguably, the most important methodological decision is which discount rate to use to value the retirement income liabilities. A lower discount rate will result in a higher liability value and lower funding ratio, all else equal. At one extreme, the valuation of retirement income liabilities could require the use of a risk-free rate. This matching view provides the value of assets that would be needed today to finance all future retirement income promises. At the other extreme, the discount rate could be the expected return on the asset portfolio backing the liabilities. This takes a budgeting perspective that takes into account how the liability is expected to be financed going forward, i.e. through investment returns and future contributions (The Institute and Faculty of Actuaries, 2012[6]). In practice, several options exist between these two extremes that aim to account for the long-term nature of retirement income liabilities, the liability driven investment strategies that are normally adopted to match future expected cash flows, and the nature of the retirement income guarantees.

The logic of using the risk-free rate to discount liability cash flows under a market consistent view is that the liability value should reflect the value of an asset portfolio that perfectly replicates the future liability cash flows, thereby allowing for certainty in the ability to meet future retirement income payments. This value would in theory be the value at which the liabilities could be traded or transferred to a third party, which is consistent with the matching valuation perspective. Nevertheless, the long-term and illiquid nature of retirement income liabilities may justify adjustments to the risk-free rate for a market consistent valuation to the extent that the assets backing the liabilities are invested in a way that aims to match the long-term cash flows of future retirement income payments. Furthermore, market consistent valuation does not impose the use of the risk-free rate. If benefits are not fully guaranteed, as with target benefit arrangements, the market consistent value of the liabilities should reflect the uncertainty in retirement income paid.

Which perspective to take is strongly linked to the objective of the funding calculation. Valuation using the risk-free rate is useful to have a transparent assessment of the underlying risks of the retirement income arrangement, and is more in line with risk management strategies. Valuation using the expected return on assets is helpful to align risk assessment with the long-term objectives of the arrangement, but disconnects from what is happening in the financial markets (Farr, Koursaris and Mennemeyer, 2016[7]).

The discount rate used to calculate the liability value will not only affect the funding calculation, but will also affect the extent to which risks are transferred to future generations and therefore has implications for fairness. Assessing the funding position of a retirement income arrangement based on the expected return on assets effectively allows the risk premiums that are expected to be earned in the future to be spent upfront.

Using the risk-free rate, on the other hand, only allows the risk premium to be spent once it has been earned. In this way, a lower discount rate will gradually release any excess return to the plan participants once the risk premium materialises.

Using a funding ratio based on the expected return on assets shifts value to current pensioners at the expense of future cohorts. As the expected risk premium is fully consumed in advance, the probability that adjustments will be needed to contributions and/or retirement income benefits increases because this risk premium may not be realised (Sanders, 2016[8]). To the extent that the funding mismatch is positive, the current retirees could immediately consume this expected surplus, reducing the value of the future generations’ contingent claim on the surplus and increasing the risk that adjustments will be needed to recover from a deficit (Yi, 2018[9]).

Relying solely on risk sharing among members and allowing for flexibility in benefits – both those being paid and those being accrued – can spread the risks and benefits of participating in the retirement income arrangement more evenly across all members. Nevertheless, participants may desire some additional certainty as to the level of the benefit that they will receive. Guarantees can provide this certainty, albeit at an additional cost.

The provider or sponsor of the retirement income arrangement may offer a certain level of guaranteed retirement income to provide retirement income certainty and stability, but securing these guarantees will come at a cost. The added certainty reduces the risk-bearing capacity of the arrangement, and involves an additional cost from any security mechanism that the regulatory framework requires to ensure that there will be sufficient financing to pay for the retirement income guarantee. Who bears this cost will depend on the design features of the arrangement and how the security mechanism is financed.

While guarantees can certainly be valuable and more than pay off ex-post in market downturns, credible promises to guarantee retirement income must come at a cost ex-ante. These costs take the form of opportunity costs stemming from a reduced risk-bearing capacity of the retirement income arrangement as well as explicit and implicit costs to support any security mechanisms in place to secure the guarantees provided.

First, there is an opportunity cost of lost investment returns when arrangements offer retirement income guarantees. Providing retirement income guarantees will reduce the risk-bearing capacity that the retirement income arrangement has to invest in assets generating higher expected returns, thereby also reducing the level of expected retirement income that the arrangement can deliver. Nevertheless, this also reduces the risk that returns will be significantly lower than expected. As discussed in the previous section with respect to valuation, the only investment strategy that will generate the level of assets needed to finance any guaranteed retirement income with certainty will be a strategy investing in risk-free assets. Any other strategy will present a risk that the actual returns will be lower and that there will not be sufficient assets to meet the promised retirement income.

Additional costs will be incurred depending on the security mechanism that the regulatory framework requires in order to ensure that the sponsor or provider will be able to pay the retirement income guaranteed. The security mechanism can generally either take the ex-ante approach of requiring that potential deficits be funded upfront or require that these deficits will primarily be funded ex-post through additional contributions. Ex-ante requirements rely on a capital buffer to finance any future adverse deviation. Ex-post requirements rely on the value of sponsor support to make any additional contributions in case the level of assets backing the liabilities becomes insufficient to meet the promised retirement income payments. In addition, some jurisdictions rely on a pension protection fund (e.g. the United Kingdom and the United States) to fulfil the promised retirement income payments, at least in part, in the event that the sponsor/provider faces insolvency and is unable to finance the additional contribution required.

Whether the costs are explicit or implied will depend on the security mechanism in place. Imposing a capital buffer makes the cost of financing a potential funding shortfall explicit, as the sponsor/provider needs to come up with the additional capital to establish the buffer. Sponsor support imposes an implicit cost to secure retirement income guarantees that should be reflected in the sponsor’s market value, as it represents an implicit liability for the sponsor. Pension protection funds offer an additional layer of protection to members against the risk of sponsor default at the explicit cost of an additional premium paid by the sponsor.

The regulatory framework must ensure that the security mechanisms in place are effective to reduce the consequences of insolvency for participants. For all security mechanisms, the cost of the additional security will be borne primarily by the sponsor/provider of the arrangement and/or any shareholders. They will bear the cost of coming up with additional capital related to the explicit costs and any implicit liability reflecting the value of sponsor support. Nevertheless, in some cases the additional costs could be passed to participants directly through lower retirement income levels (e.g. if passed on through the pricing of an insurance product) or indirectly through pressure on current wages, which would reduce contribution levels. If the security mechanism fails, participants will also bear this cost to the extent that they will not receive their retirement income benefits at the guaranteed level.

Numerous factors can affect the value of security mechanisms and their ability to protect the long-term retirement income benefits promised by retirement income arrangements. Table 6.3 summarises the key drivers of the value of these alternative security mechanisms. The way in which these mechanisms are designed can also influence incentives regarding the investment strategy, and as such, the magnitude of the opportunity cost of taking a lower risk investment strategy to match the guarantees.

The purpose of a capital buffer is to ensure that the sponsor/provider of retirement income arrangements will have sufficient capital set aside to meet promised retirement income obligations with a high probability, even in the event of a significant adverse financial, demographic, or business shock. The choice of methodology for the calculation of the capital buffer normally involves decisions as to whether the capital buffer will be risk-based, the time horizon of risk assessment and the desired level of confidence. Each of these elements contributes to the strength of the buffer in securing the guaranteed benefits for participants.

Risk-based capital requirements call for the provider to have more capital for greater risk exposures. Risk-based requirements specify the level of the required capital buffer based on the underlying asset and liability risk exposure. Dynamic risk-based requirements, as opposed to static approaches, have the advantage of being able to account for the interaction between assets and liabilities and being more sensitive to changes in the underlying risk.

A higher confidence level will result in a larger capital buffer. The confidence level sets the probability at which the available capital will be sufficient to pay liabilities following an adverse shock.

The time horizon specifies the horizon over which the risk is measured. A short-term assessment of the risk looks at whether the assets will be sufficient to finance the liabilities following an extreme adverse experience at the desired confidence level at the end of the time period for assessment. This view does not ignore the long-term nature of the liabilities per se, but it aims to ensure that assets will continue to be sufficient to finance the liabilities at each future assessment point in time. A long-term assessment is performed by projecting the cash flows for each future period under a certain risk scenario to see whether all obligations can be satisfied with the current level of assets (Kapel, Antioch and Tsui, 2013[10]). Compared to the short-term perspective, a long-term perspective can be more forgiving with respect to the risk of short-term deficits in funding. However, it remains more disconnected from the financial markets and is less aligned with risk mitigating actions that could be taken such as modifying the investment strategy or transferring the liabilities to a third party.

Risk-based capital requirements provide incentives for reducing investment risk exposure. They provide an incentive to control the underlying risk exposure and optimise the investment risk taken accounting for the additional cost of capital needed for higher risk strategies. Short-term time horizons also provide incentives to reduce investment risk, as they decrease the tolerance for short-term funding deficits that could result from negative investment shocks.

The sponsor covenant refers to the obligation of the sponsor of the retirement income arrangement, typically the employer, to make additional contributions to the plan if assets are not sufficient to finance the retirement income liabilities. The strength of the sponsor covenant is highly dependent on the strength of the sponsor and its correlation with the assets backing the retirement income liabilities, as well as the legal framework in place to enforce additional contributions (Broeders and Chen, 2013[11]).

Factors that reduce the probability that the sponsor would be able to make additional payments deteriorate the value of the sponsor covenant. As such, the financial strength of the sponsor is positively correlated with the value of support (Jaegers, 2013[12]). It then also follows that a positive correlation of sponsor strength with the assets held by the pension fund is negative for the value of the sponsor covenant. This is because a higher correlation between the sponsor strength and the pension fund’s assets implies that it would be more likely that the sponsor would not be able to make additional payments at the time they are needed, i.e. when funding levels drop (Broeders and Chen, 2013[11]). The correlation of the sponsor’s financial strength with the broader market also drives the value, with higher correlation implying a higher value during good economic times and vice versa (Jaegers, 2013[12]).

The legal framework with respect to the sponsor covenant also impacts its value. In some jurisdictions, sponsor support can be legally binding up to a certain limit (e.g. Norway). In others it can be unlimited. Under some legal frameworks, sponsor support is not legally enforceable (e.g. Ireland). The fact that sponsor support would be limited or not legally enforceable would reduce its covenant value and its loss-absorbing capacity under an adverse scenario.

Sponsor covenants may also provide an incentive for the pension fund to invest in higher risk assets. Technically the sponsor covenant can be viewed as an embedded put option that the pension fund owns, as it can receive the difference between the value of the assets and liabilities if the assets fall below the strike price (the value of the liabilities).3 This option is an implicit buffer for the pension fund, even if it is an implicit liability for the sponsor. Higher risk investment strategies increase the probability that there will be a funding shortfall, resulting in a higher option value. As such, the pension fund has an incentive to invest in a higher risk strategy that would increase the value of the implicit buffer. This incentive is enforced by the fact that the sponsor and its shareholders would not have to finance any deficit in the case that the sponsor defaults.

The third type of security mechanism available is the pension protection fund, which guarantees the payment of retirement income promises, either in full or partially, in the event that the sponsor becomes insolvent. Covered retirement income arrangements usually pay premiums to finance these funds, which effectively function as an insurance pool against the risk of sponsor default. How these premiums are set, the mechanisms the fund uses to cover the pension liabilities, and the extent to which the fund is backed by the government drive the strength of the protection fund.

For the premiums financing the protection fund to be fair and sufficient to cover the risk of insolvency that they insure, they would need to account for the three key factors that drive the ability of sponsors to pay additional contributions to the plan: sponsor strength, underfunding, and investment strategy. Despite their function as an insurance arrangement, pension protection funds do not typically base their premiums on all of these risk factors, meaning that the premiums charged do not fully reflect the risk that the protection fund will have to cover retirement income promises.4

The lack of risk-based premiums leads to increased risks to the solvency of the protection fund, diminishing the value of the security it can offer. Two key solvency risk factors faced by protection funds are adverse selection and moral hazard (Stewart, 2007[13]; Blake, Cotter and Dowd, 2006[14]). Adverse selection relates to sponsor strength and funding levels, and refers to the risk that the sponsor with a lower risk of insolvency will not want to subsidize the plans with a weak sponsor, and will close their plans to avoid paying the premiums. Moral hazard relates to a plan’s investment strategy, and refers to the risk that the pension plan or sponsor will engage in more risky behaviors knowing that the protection fund will cover the benefits in case of insolvency. This could occur, for example, if the plan pursues a very risky investment strategy in order to try to recover a deficit in an underfunded plan.

A third key solvency risk to the protection fund is systemic risk, which results because insolvencies tend to follow the business cycle. The intervening action has implications for the systemic risk faced. Funds which take over the plan assets rather than insuring them through an external annuity may face increased systemic risk (Stewart, 2007[13]).

Government backing of the protection fund will increase the value of the security it offers. However, in this case the taxpayers will ultimately bear the cost of insolvency. This could also imply generational redistribution to the extent that current workers have to bail out poorly managed and overly generous benefits for current pensioners.

The extent to which premiums are risk-based determines whether the protection fund provides an incentive for lower- or higher-risk investment. Premiums that reflect the investment risk taken by the pension fund would provide an incentive to reduce investment risk, and thereby reduce the premium owed. However, as the premiums owed to most protection funds do not reflect the investment strategy taken, the retirement income arrangement may instead have the opposite incentive to invest in high-risk assets because it does not bear any additional cost and the protection fund takes on the downside risk by insuring the retirement income liabilities. Making the sponsor support legally enforceable can mitigate this risk, however, as the sponsor could then only escape its obligations through default.

In order to balance the desire for higher retirement income with the need for stability and security, the design of retirement income arrangements and the regulatory framework supporting them must strive to achieve a balanced distribution of risks. Collective risk sharing allows for higher expected retirement incomes for all participants compared to what they could achieve on their own. However, ensuring that retirement incomes are also secure will come at a cost and will reduce the expected level of income that any arrangement can deliver. Sustainable designs will distribute the benefits of collective risk sharing among participants in a fair manner, ensure that the arrangement remains sufficiently funded, and provide additional security around any retirement income guarantees. They should also aim to be transparent about the conditions for the guarantees to be met. The design of the regulatory framework should promote these objectives.

The distribution of risks among current and future participants in a retirement income arrangement should be perceived as fair. While risk sharing among participants and across cohorts can improve expected outcomes for all participants, no group of participants should bear an undue burden to secure the benefit of another group. Otherwise, the arrangement will not be socially or politically sustainable. To ensure fairness, fairness first needs to be defined. The arrangement should then be designed to achieve this objective, and the risk transfers that could result from any shocks should be assessed. The rules of the arrangement then need to be applied in a way that enforces the continued fair treatment of members.

It is not enough to simply require fair treatment of members. Fairness can be defined in numerous ways. While it is common for jurisdictions to have requirements regarding the fair treatment of members, these requirements tend to refrain from specifying what is meant by fairness. For example, in Australia, plans are required to treat each member group fairly between and within groups. In Europe, institutions for occupational retirement provision (IORPs) should aim to have an equitable spread of risks and benefits between generations in their activities.5 In the Netherlands, the board is required to weigh the interest of all plan members in a balanced manner.

The definition of fairness must consider the context in which the retirement income arrangement operates, as different objectives for fairness can result in different optimal designs depending on the context. If public pension arrangements – which tend to rely heavily on intergenerational solidarity – play an important role in the provision of retirement income, additional intergenerational risk sharing within the funded arrangement may overburden the younger generations. Significant inter-cohort risk sharing will also not be desirable in settings that prioritise portability and the transfer of assets between different arrangements.

Context can also influence the extent to which the benefits of inter-cohort risk sharing can be expected to be realised and therefore whether the arrangement will ultimately be fair. For example, a defined-benefit accrual formula, where each contribution accrues a certain percentage of salary replacement at retirement, involves a subsidy by younger cohorts to finance the retirement income benefits accrued by older cohorts. This can be viewed as fair to the extent that the younger cohorts will eventually benefit from that subsidy when they are older, and everyone should have the same expected relative benefit. However, participants may be increasingly viewing this accrual formula as unfair to younger generations given the growing mobility of the workforce and the fact that many will not remain in the same retirement income arrangement over their entire career. Such mobility challenges the fairness of these types of retirement income arrangements, as it has in the Netherlands.

Some jurisdictions may therefore consider that sharing risks across cohorts and generations is unfair, and design the arrangement to limit these transfers. The New Brunswick target benefit plan in Canada requires that no single cohort should unduly subsidise another, implying a desire to limit inter-cohort risk sharing to the extent possible. Responses to the FCA’s consultation on how CDC schemes should be designed in the United Kingdom tend to also favour such a definition of fairness.

The definition of fairness should also consider the desired level of risk sharing within cohorts. Factors such as differences in life expectancy between genders or across socioeconomic groups have implications for the cross-subsidisation across members within the arrangement. Definitions of fairness may account for such considerations to inform the design of the retirement income arrangement to enforce the desired level of risk sharing.

In any case, fairness needs to be more precisely defined in terms of the objectives that it should achieve, even if inter-cohort risk sharing is viewed as desirable and beneficial. Different approaches have both advantages and drawbacks. For example, fairness could be defined as everyone having the same expected benefits relative to the contributions they have made, or having the same probability that retirement income will not fall below a certain level. While simple to understand, this approach has the drawback of not taking into account fairness in terms of value. Fairness could also be that all participants have the same expected improvement in welfare from participating in the retirement income arrangement. However, this ignores who bears the risk in the arrangement, and can grant equivalent benefits to cohorts who have not shared any risk. Another alternative is taking a fair value approach, which accounts for how those who bear the risk are compensated in terms of expected benefits. However, this approach can also be more complex to understand and value.

Given these considerations, defining fairness will not always be easy nor agreed upon by all parties. This highlights the importance of transparency around how the design of retirement income arrangements achieves the stated fairness objective and independent oversight to enforce the stated rules.

The design of retirement income arrangements should aim to achieve the defined fairness objectives. Design architects should understand the implications that various features of the arrangement impose with respect to the direction and magnitude of risk transfers.

The rules that determine how retirement income benefits and/or contributions accrue and are adjusted following changes in funding levels will drive the extent to which risks are transferred across cohorts. These rules include the accrual formula as well as the form, timing, and magnitude of the adjustment and how the needed adjustments are valued. These features may also have different implications for specific arrangements given their particular demographic profiles.

Defining retirement income benefits through a return on contributions rather than a percentage of salary will reduce the subsidies of younger participants to older participants and make the retirement income arrangement more actuarially fair for all.

Allowing for the adjustment of retirement income to reduce any funding mismatch that materialises will reduce intergenerational risk transfers towards future participants. This will prevent significant negative funding deficits from being pushed indefinitely to future generations. Further allowing these benefit adjustments to be shared across both active and retired participants will make this more efficient, and help to reduce the size of the necessary adjustments for individual participants.

Introducing a buffer can reduce the probability of a reduction in retirement income, but involves a value transfer to future generations. Buffers require that some benefits be withheld today to cushion the risk in the future. Where positive buffers are required to be maintained, future generations will benefit the most. As buffers may take some time to build up, the economic context matters for their introduction.

No-action corridors, which allow funding levels to vary within defined thresholds without requiring benefit adjustments, limit inter-cohort risk sharing compared to other mechanisms that aim to provide benefit stability, such as buffers. Corridors allow for some intertemporal smoothing of investment risk and can significantly reduce the volatility of adjustment and the probability of large benefit cuts. While such features do involve subsidies by working age cohorts participating in the arrangement, these are largely temporary and the corridor does not seem to result in significant long-term value transfers (Sanders, 2016[8]).

Shortening the period of amortization of adjustments will reduce the value transfers across cohorts. While amortization periods to implement the necessary adjustments are a way to reduce the magnitude of immediate adjustments, they can also result in persistent value transfers to younger generations that lack transparency. Asymmetric amortization periods for workers and pensioners can further obscure the amount and direction of value transfers.

Market consistent adjustments to retirement income benefits and contributions will minimise any value transfers related to the measurement of funding and the needed corrections. For retirement income arrangements that allow for the adjustment of benefits, the market consistent discount rate will fall somewhere between the risk-free rate and the expected return on assets. Basing contributions solely on expected returns will allow current workers to accrue all benefits of the expected risk premium before it materialises, whereas basing the contributions on fair value allows those who bear the risk – those with the largest benefit liabilities – to also benefit from potential gains. Measuring benefit adjustments on a market consistent basis will also reduce any opaque residual mismatch that could remain and be passed on to future generations.

To ensure that the retirement income arrangement is meeting its objectives in terms of fairness, it can be assessed from the inception of the arrangement (ex-ante) or after a funding shock (ex-post). Considering fairness ex-ante will show whether the design of the pension itself achieves the objectives of fairness. Ex-post assessments can highlight how different types of shocks impact different cohorts and what this implies given the objective of the retirement income arrangement, particularly where the objective is to limit inter-cohort transfers.

Assessment of the fair design of a retirement income arrangement will need to consider both investment and longevity shocks. These two risks do not have the same implications for every cohort.

To ensure that retirement income arrangements remain fair, the rules for adjustments should be objective and clearly defined in advance. Canada, for example, requires that arrangements define a clear sequence of actions in advance to take when the funding position changes. This limits potential bias from human judgement that could favour certain groups of participants over others.

The rules in place should be tested against a variety of funding shocks to assess their outcomes with respect to fairness. As such, there should then be limits to the extent that the rules for needed adjustments can be modified following a funding shock, particularly when this is done to favour certain groups. Such changes will likely benefit the current pensioner generation at the expense of future generations. In the Netherlands, for example, previously defined benefit cuts have been delayed or modified in light of the protracted low interest rate environment and reduced funding levels. Such delays shift risk to future generations, who will be more likely to bear a higher cost of the needed benefit reductions themselves. These types of decisions may reduce confidence and trust in the pension system, and deteriorate public support for these types of pension institutions, threatening their viability and sustainability going forward.

A strong governance framework is essential to make sure retirement income arrangements achieve their fairness objectives. Governing bodies are commonly responsible for determining these objectives as well as deciding how adjustments to the arrangement will be made in line with these objectives. This role is particularly crucial for arrangements where risk transfers may be less transparent to members or complex to assess, as is the case with target benefit type arrangements. Any governance framework should implement requirements and processes to establish and assess the impact that any changes to the arrangement have on fairness.

The regulatory framework needs to ensure that retirement income arrangements will continue to operate fairly and effectively over their long-term horizon. As such, the regulatory framework needs to limit the potential for the costs of any funding deficit to be pushed indefinitely to future generations. Regulatory requirements for funding need to be coherent with the regulatory objectives and the nature of the retirement income guarantees offered by the arrangement, and provide incentives for providers to effectively manage the risk exposures.

To limit the potential for the costs of any funding deficit to be pushed indefinitely to future generations, it is essential to limit the size of inter-cohort and intergenerational transfers. Large risk transfers across cohorts and generations will put the continuity of the retirement income arrangement at risk. While such risk sharing can improve the welfare for all participants, fairness alone is not a sufficient criterion to ensure the long-term continuity of a retirement income arrangement. Even if an arrangement is fair ex-ante, large risk transfers can mean that after a funding shock, certain cohorts could be worse off by participating in the retirement income arrangement.

To limit the magnitude of risk transfers to younger cohorts in particular, the size of the potential funding deficit needs to be constrained. The mismatch between assets and liabilities represents the value of potential transfers to future generations, barring any adjustment to current contributions and/or benefits. Limiting the size of this mismatch, and any funding deficit, will limit the potential size of these transfers to future retirees and ensure that they are not so large as to threaten the sustainability of the arrangement.

The regulatory framework therefore needs to have minimum funding requirements in place to ensure that there will be sufficient assets to finance future retirement income promises with a reasonable level of certainty. This will prevent future retirees from being at a significant disadvantage from the outset, and preserve the collective benefits of their participation in the retirement income arrangement. The level of the minimum funding requirement, however, should depend on the nature of the arrangement and the types of guarantees that it offers.

The regulatory framework needs to establish funding requirements by defining how funding levels should be measured as well as the required level of funding. These requirements should be consistent with the regulatory objective as well as with the nature of the arrangement and the strength of the retirement income promises that it offers.

The methodology used to calculate the funding level of a retirement income arrangement should be consistent with the goal of the calculation. A funding ratio calculation using the risk-free rate to value the liabilities will show whether the assets are currently sufficient to finance the expected pension benefits with certainty. A funding ratio calculation based on the expected return on the asset portfolio backing the pension liabilities will show whether the assets will be sufficient to finance the expected pension benefits in a central scenario based on an actuarial projection.

The methodology used to calculate the funding level should also be consistent with the regulatory objectives. Calculating funding levels based on liabilities valued at the risk-free rate has the advantage of being more transparent with respect to the uncertainty that future benefits will be able to be paid, as this approach does not rely upon the realisation of uncertain investment returns. It also allows for better comparability, as different retirement income arrangements would be valued at the same rate, rather than a rate specific to the investment strategy of the particular arrangement.

Other methodological considerations for the funding assessment, such as what future cash flows to include and whether to include a risk margin, should also be decided in line with the objectives. For example, future benefit accruals could be considered if assessing the continued viability of the arrangement. Otherwise, if assessing the funding adequacy of current liabilities, valuation could assume immediate termination and wind-up.

Regulators and supervisors may therefore require multiple calculations to have a better view on the funding status of the arrangement. Given the significant impact of the different valuation approaches on the funding ratio calculation, relying on a funding ratio based on a single methodology will not provide a full picture of how likely it is that the assets will be sufficient to finance the liabilities. Canada takes this approach, and requires that DB pension funds assess their funding levels based on a going concern basis, using the expected asset return, and on a solvency basis using long-term bond yields. For the first calculation, the following year’s benefit accruals are accounted for, whereas for the second the plans are assumed to terminate immediately. Stochastic analysis of potential funding scenarios would also be helpful to assess the likelihood that funding levels will be sufficient to cover benefits.

The minimum required level of funding should be at a threshold that is consistent with the nature of the retirement income promises that the arrangement provides. An arrangement that aims to fully guarantee the promised retirement income should logically be fully funded at the risk-free rate, as this is the level of funding required to provide full certainty that benefits will be met. For arrangements that are able to adjust the benefits payable, lower funding levels are justifiable because the strength of the benefit promise that these types of arrangements offer is lower.

The regulatory framework should require stronger funding for arrangements that offer firm guarantees. Aligning the funding requirements with the nature of the retirement income arrangement could be done in principle by either adjusting the discount rate used to calculate the funding level or adjusting the required funding level itself. Using a higher discount rate to value retirement income liabilities will reflect any uncertain nature as to the future retirement income that would be paid. To be consistent with the nature of the retirement income guarantees, the discount rate should be set in a market consistent manner. Alternatively, the valuation could still be based on the risk-free rate, but the required funding level would have to be adjusted downward to reflect the weaker retirement income promise. This approach would have the strong advantage of increased transparency and comparability with regard to the strength of funding across different retirement income arrangements.

The existence of a protection fund should not be used to justify softer funding requirements. The likelihood that premiums are under-priced on average means that pension protection funds themselves face increased risk of insolvency. This risk implies that adequate funding requirements for retirement income arrangements are still needed to reduce the risk of insolvency for protection funds (Stewart, 2007[13]).

The regulatory framework should ensure that providers have incentives to manage their risk appropriately. The valuation and funding requirements that the regulatory framework imposes will also have implications for the incentives that the retirement income arrangements will have in order to employ certain risk management strategies.

Funding requirements below those implied by a market consistent valuation of assets and liabilities could discourage arrangements from using risk management strategies that involve the sale or transfer of assets and/or liabilities. Adapting the investment strategy to changing market environments may not be attractive where assets are valued at historical cost, as a change could force the realisation of investment losses if these assets are sold.

Similarly, if funding requirements are below the level that a third party would be willing to accept to take over the retirement income liabilities, the transfer of these liabilities may not be a feasible option for relatively underfunded arrangements. This relates not only to funding requirements and the discount rate assumption, but also to all assumptions used to value the liabilities, including mortality assumptions. If mortality assumptions are not in line with reasonable expectations and they do not account for future expected improvements in longevity, the pension liabilities will be underestimated compared to their actual market value.

The regulatory framework should also be careful to not to create incentives to increase risk taking. If valuation and funding requirements are based on the expected return of the asset portfolio, increasing the riskiness of the investment strategy would automatically improve the funding position in the short term. However, the risk that assets would not be able to finance the retirement income obligations would increase. Public pension funds in the United States provide an example of how such incentives could be detrimental. These plans have been shown to take more investment risk to offset deteriorating funding levels, which has resulted in worse performance and funding in the long run (Andonov, Bauer and Cremers, 2017[15]). This example also highlights the importance of assessing multiple measures in order to have a more complete picture of the risk exposure of the retirement income arrangement.

The regulatory framework needs to provide for effective retirement income security in a cost-efficient manner when a retirement income arrangement provides guarantees. This will involve requiring effective security mechanisms to secure retirement income guarantees and ensuring that the governance framework in place enforces their effectiveness. The regulatory framework must also promote increased transparency with respect to the strength of the guarantees that arrangements offer in order to align their design with member preferences.

The regulatory framework relies primarily upon two types of security mechanisms, either alone or in combination. The first is a capital buffer, which requires the sponsor/provider to set aside additional capital to cover the risk of adverse deviations in funding levels. The second is a sponsor covenant, which requires the sponsor of the retirement income arrangement to finance any funding shortfall once it materialises. In addition, jurisdictions may also have a pension protection fund that will at least partially insure the pension benefits in case of sponsor default.

The value of the protection that the different mechanisms offer and their ability to protect promised retirement income benefits depend on numerous underlying drivers. For capital buffers, risk-based calculations, shorter time horizons, and higher confidence levels provide more security. For sponsor covenants, stronger sponsors whose strength is not correlated with the assets backing the retirement income liabilities and a legal framework that requires the sponsor to contribute additional capital needed will offer higher security. Finally, the value of protection from a pension protection fund will be higher if premiums are risk-based, if the transferred arrangement is insured through a third party, and if the government provides its backing.

The security mechanism should be designed to offer the level of security required by the strength of the retirement income guarantees that the arrangement offers and the regulatory objectives in place. The level of benefit security provided should not depend on the security mechanism relied upon. Where sponsor support is the primary mechanism relied upon, the regulatory requirements need to account for the additional risk factor of sponsor insolvency. Compared to a solvency buffer, reliance on sponsor support introduces an additional risk variable due to the correlation of sponsor solvency and the assets backing the liabilities (Broeders and Chen, 2013[11]). One way to account for this additional risk would be to require a certain level of solvency capital in addition to the sponsor covenant. Canada is taking this approach with new requirements for a provision for adverse deviation (PfAD), which requires higher funding levels depending on certain risk factors for DB arrangements.

The design of security mechanisms should also aim to balance the need for security with the cost of providing that security, particularly with respect to the opportunity cost of lower expected returns on investment. The risk-bearing capacity of arrangements with retirement income guarantees should be lower than for those without, and as such, the security mechanisms in place should provide incentives for an appropriate investment strategy. Nevertheless, they should not diminish the remaining benefits of collectivity. While risk-based capital requirements will increase the required capital for higher-risk investment strategies, the risk charges need to be defined with sufficient granularity and in a consistent manner across risk types to encourage a diversified long-term investment strategy. For arrangements relying on sponsor covenants, the existence of a protection fund should not encourage higher risk investment strategies. The Pension Protection Fund in the United Kingdom avoids such incentives by reflecting investment risk, sponsor strength and funding levels in the levies it charges to providers. In addition, trustees are not allowed to take the existence of the Pension Protection Fund into account when making decisions in the interest of their members.

The governance of a retirement income arrangement will also be a factor in the protection value of the security mechanisms in place. A strong governance framework will improve the value of all security mechanisms, albeit at an additional cost. Nevertheless, having a strong governance framework in place will help to enforce appropriate risk management and investment strategies.

Participants in retirement income arrangements need to better understand the nature of the retirement income guarantees that these arrangements provide. As discussed earlier, funding and valuation requirements must be designed to reflect the strength of a guarantee. Weaker funding requirements also imply a weaker promise. The extent to which certain arrangements have weaker guarantees needs to be made more transparent for participants.

Indeed, the difference in the regulatory framework between insurance-provided arrangements and employer/pension fund-provided arrangements is often justified by the differences in the strength of these promises and the contractual terms of these arrangements. One argument commonly cited is the difference in the nature of the contract. Employer/pension fund-provided retirement arrangements are more commonly viewed as a social contract for which the various bodies that oversee the arrangement may take decisions that allocate wealth differently across members (Broeders, De Jong and Schotman, 2016[16]). As such, actions that are taken to adjust plan rules are discretionary and not necessarily defined in advance. Another common argument against equal regulatory treatment is the difference in the nature of the retirement income promise, and that promises made by occupational arrangements can be more easily changed than the retirement income guarantees promised in an insurance contract (Van Hulle, 2016[17]). This argues that employers finance the promise on a best-effort basis, and that employees should understand that there is a risk that these promises will not be fulfilled (International Actuarial Association, 2018[18]).

Yet participants are arguably not fully aware of the risk that they may not receive their promised retirement income benefits, particularly for DB arrangements, and view them as future guaranteed retirement income in exchange for service during employment. Societal reactions to sponsor failures support this view, where pressure for increased regulation of retirement income arrangements follows incidences of funding difficulties or sponsor insolvency (for example, following the collapse of BHS in the United Kingdom (Cumbo, 2017[19])). Furthermore, in practice, retirement income benefits are not regularly adjusted, even when the contract specifies the mechanism to do so in advance (EIOPA, 2014[20]). The existence of pension protection funds further supports the expectation that retirement income promises are guaranteed, and shifts the nature of the retirement income promise towards one that is more of a hard guarantee (Blake, Cotter and Dowd, 2006[14]). This observation also supports the better alignment of valuation and funding requirements with the true nature of the retirement income guarantee.

Communication of retirement income guarantees and risks to participants needs to be clear, straightforward and transparent. To the extent that the regulatory framework shares the view that employer-provided retirement income promises by nature provide softer guarantees, providers need to make this clearer to the participants, who are not likely to understand this if it is only implicit in the nature of the arrangement. To this end, participants need to be told under what conditions their retirement income benefits will be reduced, and to have more transparency around the probability that this could happen. The need for more transparency would support the communication of a funding ratio based on the risk-free rate. Communicating a funding ratio of 80% based on the risk-free rate would be a more transparent way of indicating the potential that the assets backing the retirement income promise may not be sufficient to finance that promise compared to communicating a funding ratio of 100% for the same arrangement based on expected returns.

An even more transparent approach would be to explicitly design the arrangement to adjust retirement income benefits from the outset, and clearly communicate the rules to participants. This will better enable designs to optimise the risk sharing in the arrangement in a sustainable way. Several examples of these types of plans exist in practice, for example conditional indexation arrangements or target benefit arrangements.

However, even if retirement income benefit adjustments are explicit in plan design, this still needs to be better communicated to participants. The Netherlands, for example, has faced a large challenge with respect to communicating to members that their benefits from the CDC arrangements could be reduced. These plans are still widely communicated as being ‘DB’, which participants understand as providing the same level of promise as the previously offered fully guaranteed DB arrangements. Participants need to be regularly told how their benefits could be adjusted and under what conditions this will occur.

Increased transparency with respect to the nature of pension promises would also allow participants – and society more broadly – to decide the strength of the retirement income guarantee they prefer given the additional costs of providing that guarantee. Participants may very well prefer a lower retirement income with a higher level of certainty compared to a higher expected retirement income with less certainty. Without transparency, however, participants will not be able to express these preferences.

Risk sharing in the design of retirement income arrangements offers benefits in terms of risk mitigation and the level of expected income in retirement compared to individual retirement arrangements. The ability for a collective retirement income arrangement to pool risks and smooth funding shocks over time can significantly mitigate the risks that individuals would otherwise bear on their own. This allows for higher retirement incomes to be paid, and ultimately increases the collective capacity of the arrangement to invest in higher risk assets that can provide a higher expected retirement income overall. However, providing retirement income guarantees in such arrangements can offset some of these expected benefits, as the security mechanisms needed to enforce these guarantees involve both implicit and explicit costs. Designing the arrangements to share risks either between the provider and the participants or solely among the participants themselves allows for a more efficient and sustainable distribution of risks by allowing the risks to be shared among more stakeholders.

The design of retirement income arrangements should keep in mind their long-term sustainability. For this, the design of these arrangements must be such that they promote fairness among the participants and long-term continuity. Retirement income promises, or lack thereof, must be transparent and any retirement income guarantees that the arrangement provides must be secured.

The regulatory framework needs to ensure fair risk sharing among participants. This starts with defining fairness, designing the arrangements to achieve that objective through the mechanisms to adjust contributions and benefits, and enforcing these rules when these mechanisms are triggered. Risk sharing among participants can be limited to within a specific cohort or shared across cohorts or generations. Risk sharing across cohorts (to provide income stability) and generations (to improve welfare and maximise the expected utility of all participants) allows for intertemporal smoothing of shocks that cannot be mitigated periodically through risk pooling. Any retirement income arrangement that shares risks will have value transfers that make some groups worse off ex-post after a funding shock, even if it is seen ex-ante to be fair and welfare improving for all participants. It is therefore necessary to assess and measure the risk and value transfers to make sure that risk sharing across cohorts is likely to be perceived as fair.

The regulatory framework needs to ensure the continuity and sustainability of risk-sharing arrangements. It can do so by limiting the size of inter-cohort and intergenerational transfers, making funding requirements consistent with regulatory objectives and the nature of the retirement income arrangement, and providing incentives to promote effective risk management. Retirement income arrangements face risks to sustainability and continuity if value transfers are too large, since they could break the intergenerational solidarity. Funding requirements limit the size of risk transfers and help to ensure the continuity of the arrangements; however, this also comes at the cost of reduced risk-bearing capacity.

Defining a minimum funding requirement also involves defining a valuation methodology with which to calculate it. One of the most important factors is which discount rate to use to value the retirement income liabilities. Lower discount rates result in a higher liability value and a lower funding ratio, all else equal. The choice of the discount rate lies between the two extremes of the risk-free rate and the expected return on the asset portfolio. The former provides a more transparent assessment of the underlying risk of underfunding and is more in line with typical risk management strategies. The latter aligns risk assessment more with the long-term objectives, but has less connection with what is happening in the financial market. The choice also affects the extent to which risks are transferred to future generations and therefore the perception of fairness. Funding requirements based on the expected asset returns effectively allow the risk premiums that are expected to be earned in the future to be spent upfront, and shifts value to current pensioners at the expense of future cohorts. Using the risk-free rate only allows the risk premium to be spent once it has been earned, releasing any excess return to plan participants once the risk premium materialises.

The regulatory framework also needs to ensure benefit security for guaranteed retirement income in order to reduce the risk of insolvency for participants and enforce the payment of the guarantees. Guarantees can provide additional certainty as to the level of benefits that individuals will receive. However, this certainty comes at an additional cost. Opportunity costs result from lower investment returns due to the lower risk-bearing capacity for the retirement income arrangement to invest in assets generating higher expected returns. The security mechanisms to enforce the guarantees also come at a cost.

The regulatory framework should require effective security mechanisms and ensure that the governance framework enforces their effectiveness. The security mechanisms relied upon are typically a capital buffer, a sponsor covenant and a pension protection fund, or a combination of these. The regulatory framework should design the specific mechanisms relied upon to offer the desired level of security, which should not depend on the mechanism selected. The regulatory framework also needs to promote transparency and communication to participants regarding the trade-offs between costs and the security of benefits, informing them regularly about the potential for adjustments to their benefits and the circumstances in which this could happen.

The design of sustainable retirement income arrangements must find a way to balance the distribution of risks in order to maximise the retirement income it can provide while offering the desired level of security of benefits. While collective risk sharing increases the risk-bearing capacity of members and their retirement income potential, security entails implicit and explicit costs in terms of lower investment returns and the cost of security mechanisms that aim to enforce any guarantees provided. A sustainable arrangement will have a balanced distribution of risks across the stakeholders.

References

[15] Andonov, A., R. Bauer and K. Cremers (2017), “Pension Fund Asset Allocation and Liability Discount Rates”, The Review of Financial Studies, Vol. 30/8.

[14] Blake, D., J. Cotter and K. Dowd (2006), Financial Risks and the Pension Protection Fund: Can it Survive Them?, https://arxiv.org/ftp/arxiv/papers/1103/1103.5978.pdf.

[2] Bonenkamp, J. and E. Westerhout (2014), “Intergenerational Risk Sharing and Endogenous Labour Supply within Funded Pension Schemes”, Economica, Vol. 81/323, pp. 566-592, http://dx.doi.org/10.1111/ecca.12092.

[1] Bovenberg, L. et al. (2007), “Saving and Investing Over the Life Cycle and the Role of Collective Pension Funds”, De Economist, Vol. 155/4, pp. 347-415, http://dx.doi.org/10.1007/s10645-007-9070-1.

[5] Bovenberg, L. and R. Mehlkopf (2014), “Optimal Design of Funded Pension Schemes”, Annual Review of Economics, Vol. 6, pp. 445-74, http://dx.doi.org/10.1146/annurev-economics-080213-040918.

[11] Broeders, D. and A. Chen (2013), “Pension Benefit Security: A Comparison of Solvency Requirements, a Pension Guarantee Fund, and Sponsor Support”, The Journal of Risk and Insurance, Vol. 80/2, pp. 239-272, http://dx.doi.org/10.1111/j.l539-6975.2012.01465.x.

[16] Broeders, D., F. De Jong and P. Schotman (2016), Interest rate models for pension and insurance regulation, Netspar, https://www.netspar.nl/assets/uploads/Netspar_Design_56_WEB.pdf (accessed on 27 February 2019).

[3] Cui, J., F. de Jong and E. Ponds (2011), “Intergenerational risk sharing within funded pension schemes”, Journal of Pension Economics and Finance, Vol. 10, pp. 1-29, http://dx.doi.org/doi:10.1017/S1474747210000065.

[19] Cumbo, J. (2017), Green paper warns on extra pensions regulation, https://www.ft.com/content/a6bcdc12-f769-11e6-9516-2d969e0d3b65.

[20] EIOPA (2014), Mapping Exercise for Further Work on Solvency of IORPs, https://eiopa.europa.eu/Publications/Consultations/EIOPA-CP-14-040_Mapping_Exercise_for_Further_Work_on_Solvency_of_IORPs.pdf.

[7] Farr, I., A. Koursaris and M. Mennemeyer (2016), Economic Capital for Life Insurance Companies | SOA, SOA, https://www.soa.org/research-reports/2016/research-econ-cap-life-ins-co/.

[4] Gollier, C. (2008), “Intergenerational risk-sharing and risk-taking of a pension fund”, Journal of Public Economics, Vol. 92, pp. 1463-1485, http://dx.doi.org/10.1016/j.jpubeco.2007.07.008.

[18] International Actuarial Association (2018), Defined Benefit Pension Plan Funding and the Role of Actuaries, https://www.actuaries.org/IAA/Documents/Publications/Papers/PEBC_Pension_Funding_Monograph_May2018.pdf (accessed on 25 February 2019).

[12] Jaegers, J. (2013), Valuation of the sponsor support option. A practical tool for Risk Management, Tilburg University, https://www.netspar.nl/assets/uploads/011_MSc_Joeri_Jaegers.pdf (accessed on 27 February 2019).

[10] Kapel, A., D. Antioch and E. Tsui (2013), Economic capital for life insurers, Towers Watson, https://www.actuaries.asn.au/Library/Events/SUM/2013/Sum2013PaperKapelTsui.pdf (accessed on 14 March 2019).

[8] Sanders, B. (2016), Analysis of Target Benefit Plan Design Options, Society of Actuaries, https://www.soa.org/globalassets/assets/files/research/research-2016-analysis-tbp-plan-design.pdf.

[13] Stewart, F. (2007), “Benefit Security Pension Fund Guarantee Schemes”, OECD Working Papers on Insurance and Private Pensions, No. 5, OECD Publishing, Paris, https://dx.doi.org/10.1787/260604113335.

[6] The Institute and Faculty of Actuaries (2012), A Framework for the use of Discount Rates in Actuarial Work, http://www.actuaries.org.uk/research-and-resources/documents/final-.

[17] Van Hulle, K. (2016), “Risk Disclosure in the European Insurance Industry: Implications for Occupational Pension Funds”, in Mitchell, O., R. Maurer and J. Orszag (eds.), Retirement System Risk Management Implications of the New Regulatory Order, Oxford University Press, Oxford, https://pensionresearchcouncil.wharton.upenn.edu/wp-content/uploads/2018/02/05-Corrected-Proof.pdf.

[9] Yi, L. (2018), Quantifying Inter-generational Equity under Different Target Benefit Plan Designs, Simon Fraser University, https://www.netspar.nl/assets/uploads/P20180620_MSc19_Lu-yi.pdf (accessed on 25 September 2019).

Notes

← 1. The sponsor/provider refers to the entity ultimately responsible for ensuring that the required payments are made to members, for example an employer or an insurance company, and this chapter uses these terms interchangeably.

← 2. The names chosen for different types of arrangements try to convey their main features and to avoid common nomenclature, which can carry different connotations in different jurisdictions.

← 3. Valuing the sponsor support as a put option assumes that the sponsor will not benefit from any surpluses, which may not be the case where pension arrangements are closed. In this case, the value of the sponsor covenant would resemble more a futures contract.

← 4. The main argument as to why premiums are not risk-based is that forcing weaker sponsors to pay higher premiums would increase their risk of insolvency, and given that insolvencies are also pro-cyclical this would increase the systemic risk faced by the protection fund. Fully risk-based premiums would also likely be too expensive for most firms, so could not be a solution in practice (Stewart, 2007[13]).

← 5. Article 7 of the Directive (EU) 2016/2341 of the European Parliament and of the Council of 14 December 2016 on the activities and supervision of institutions for occupational retirement provision (IORPs) states that “As a general principle, IORPs should, where relevant, take into account the objective of ensuring the intergenerational balance of occupational pension schemes, by aiming to have an equitable spread of risks and benefits between generations in occupational retirement provision.”

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

© OECD 2020

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at http://www.oecd.org/termsandconditions.