a new financial assessment framework
(FTK, Financieel Toetsingskader )
in the Netherlands
J.C.D. Chin
Master’s Thesis to obtain the degree in Actuarial Science and Mathematical Finance University of Amsterdam
Faculty of Economics and Business Amsterdam School of Economics
Author: J.C.D. Chin
Student number: 5870186
Email: jcdchin@gmail.com
Date: January 30, 2015
Supervisor: dhr. dr. T.J. Boonen (University of Amsterdam) Second reader: prof. dr. ir. M.H. Vellekoop (University of Amsterdam) Supervisors: drs. J. De Mik (EY) and drs. J. Fischer (EY)
Abstract
In this thesis we examine the intergenerational effects of the shift to a new Financial Assessment Framework (FTK, Financieel Toetsingskader) in the Netherlands. We split the package of measures in separate measures, which are then evaluated with a tool that consists of both a classic Asset-liability Management (ALM) model and a value-based ALM model. The latter is a combination of generational accounting and a value-based approach. With classic ALM we evaluate how the new FTK affects the financial position of the fund. With value-based ALM we evaluate the intergenerational effects. We find that the average financial position of the fund is improved in the sense that the funding ratios are higher. Furthermore, the probability of underfunding and the probability of cuts have decreased. Moreover, when cuts are applied, the cut rates are lower. From the value-based ALM results we find that the middle-aged cohorts, that is, cohorts that are currently of age fifty-five, experience a decrease in their generational accounts due to increased contributions and lower expected benefits. The new FTK is beneficial for retirees who benefit from the delay of nominal cuts in the new FTK. The younger aged cohorts have also improved in their generational accounts. These cohorts face the downside of higher contributions, but are in the long run compensated due to the improved financial position of the pension fund.
Keywords Generational Accounting, Value-based ALM, Financieel Toetsingskader, Intergen-erational Risk Sharing
Contents
1 Introduction 1
1.1 Building a better pension world . . . 1
1.2 Research description . . . 1
I Background 3 2 Dutch pension landscape 4 2.1 Three-pillar system . . . 4
2.2 Second pillar pension . . . 4
2.2.1 Pension types . . . 5
2.2.2 Pension providers . . . 5
2.2.3 Pension plans . . . 6
2.3 Financial Assessment Framework (FTK, Financieel Toetsingskader ) . . 7
3 Policy redesign: new FTK 9 3.1 The road to a new FTK . . . 9
3.2 The new FTK . . . 11
4 Literature review 14 4.1 Social security reforms . . . 14
4.2 Intergenerational risk sharing . . . 14
4.3 Value-based ALM . . . 15 II Method 17 5 Model introduction 18 5.1 Population . . . 18 5.2 Pension fund . . . 19 5.3 Simulation . . . 20 5.4 Value-based ALM . . . 21 6 Policy variants 23 6.1 Benchmark policy . . . 23
6.1.1 Pension fund’s characteristics . . . 23
6.1.2 Policy’s characteristics . . . 24
6.2 Reform I: Increased VEV . . . 26
6.3 Reform II: Policy funding ratio . . . 26
6.4 Reform III: Future-proof indexation . . . 27
6.4.1 Reform IIIa: Future-proof indexation (geometric) . . . 27
6.4.2 Reform IIIb: Future-proof indexation (rule of thumb) . . . 27
6.4.3 Reform IIIc: Spreading catch-up indexation . . . 28
6.5 Reform IV: Exclusion of the short-term recovery plan . . . 28
6.5.1 Reform IVa: Benefit cut after five consecutive years under the
MVEV . . . 28
6.5.2 Reform IVb: Unconditional cut (10-year spread) . . . 29
6.6 Reform V: Combined effects . . . 30
III Results and Analysis 31 7 Results 32 7.1 Benchmark policy . . . 32
7.2 Reform I: Increased VEV . . . 34
7.3 Reform II: Policy funding ratio . . . 35
7.4 Reform III: Future-proof indexation . . . 36
7.4.1 Reform IIIa: Future-proof indexation (geometric) . . . 36
7.4.2 Reform IIIb: Future-proof indexation (rule of thumb) . . . 37
7.4.3 Reform IIIc: Spreading catch-up indexation . . . 38
7.5 Reform IV: Exclusion of the short-term recovery plan . . . 39
7.5.1 Reform IVa: Benefit cut after five consecutive years under the MVEV . . . 39
7.5.2 Reform IVb: Unconditional cut (10-year spread) . . . 40
7.6 Reform V: Combined effects . . . 41
8 Sensitivity analysis 42 8.1 Initial funding ratio . . . 42
8.2 Time horizon of the model . . . 43
8.3 Investment portfolio . . . 45
8.4 Population of the fund . . . 46
9 Conclusions 47 9.1 General conclusions . . . 47
9.2 Comparison of the results . . . 47
9.3 Discussion . . . 48
References 51 IV Appendices 53 A Future-proof indexation 54 B Spreading benefit cuts 56 C Economic scenarios 59 D Model parameter specification 60 D.1 Population of the fund . . . 60
Acknowledgements
First of all, I want to thank my supervisor Tim Boonen for all guidance during the pro-cess and the valuable meetings. I also want to thank EY Actuarissen B.V. for providing me with the opportunity to write this thesis at their office and for their hospitality. Special thanks go to my supervisors Hans de Mik and Jochem Fischer, who helped me with their challenging questions and their constructive feedback. Furthermore, I want to thank Fang Qi Wu and Tim Gunneweg for pre-reading my thesis before handing in my final version.
Last but not least, I want to thank my parents and my brother for their unconditional love and support.
Introduction
1.1
Building a better pension world
The pension system in the Netherlands is currently put under the microscope, which is the result of disappointing pension results in the past, population ageing and a chang-ing economic environment. Since the financial crisis in 2008, the Dutch government is working on structural changes to improve the sustainability of the pension system. The long-lasting preparations for this improved system are still going on, but the government also realizes that some measures cannot be further delayed. Hence, on June 25th, 2014 a proposal for a new Financial Assessment Framework (FTK, Financieel Toetsingskader ) was published, which lays down the statutory financial requirements for pension funds. These changes have effect on the financial position of pension funds, but also also im-ply value transfers between different generations. Both of these potential effects are examined in this thesis.
1.2
Research description
The focus in this thesis is on the intergenerational effects of the new FTK. More specif-ically, we consider the new FTK as a package of measures and subdivide this package in separate policy measures. Hence, allowing us to show how the total of intergenerational effects is composed. The tool that we employ is two-fold. We consider the effects on the financial position of the fund with a classic ALM model. The classic ALM model uses economic scenarios produced by an economic model (Hoevenaars and Ponds, 2008) and this ultimately results in the probability distributions for key variables like the funding ratio. Next, the intergenerational effects are evaluated with a value-based ALM study, which is a combination of generational accounting and a value-based approach. Gener-ational accounting is a concept that is borrowed from public finances (Auerbach et al., 1999). It reveals the zero-sum game in pension funds. Therefore, value-based ALM can show who wins and who loses from a pension policy redesign (Ponds, 2003). The value-based approach can typically be performed with a deflator approach (De Jong, 2004) or risk-neutral scenarios (Hull, 2009). In this thesis we use risk-neutral scenarios. The effects on generations are compared by highlighting three generations, namely gen-erations that are currently thirty-five, fifty-five and seventy-five years old. Our value-based ALM results show that both thirty-five and the seventy-five year old participant gain from the shift to the new FTK. The gain of the thirty-five year old participant is the largest over all participants, despite the fact that he faces a decrease in the economic value of his net benefits. The gain is solely the result of the expectation that the funding ratio will be higher in the future due to the new regulations. The gain of the seventy-five year old participant is mainly due to the exclusion of the short-term recovery plan, which results in a lower frequency of benefit cuts. These two cohorts benefit at the
2 J.C.D. Chin — Intergenerational effects in the FTK
expense of cohort thirty-five, which especially faces a decrease in the value of their net benefits. In a sensitivity analysis which we use to check whether the uncertainty in the input variables causes uncertainty in the output, we find that the loss of the fifty-five year old is the most robust outcome.
The goal of this thesis is to reveal the value transfers between generations resulting from the shift to new FTK and to show how these transfers arise from the different measures of the new FTK. We decompose the package of measures which are proposed in the new FTK, in separate measures. This way we gain in insights how the total gen-erational effects of the shift to a new FTK are constructed. We do not intend to find an optimal solution within the new FTK regarding intergenerational effects.
This thesis starts with a very basic introduction to the Dutch pension system in Chap-ter 2. The new FTK, which is the subject of examination is described in more detail in Chapter 3. Here, we treat the road that preceded the proposal of the new FTK and we discuss the individual aspects of the new FTK. In Chapter 4 we discuss the importance of evaluating intergenerational effects that are caused by policy reforms. Furthermore, we consider how other studies have evaluated generational effects. Moreover, we intro-duce the concepts of intergenerational accounting and of value-based ALM. Next, we introduce the model and its assumptions in Chapter 5. In Chapter 6 we pay attention to the reforms in the new FTK and how these reforms are represented mathematically in our model. After that, we show and discuss in Chapter 7 both the classic ALM and value-based ALM results. In Chapter 8 we check the robustness of the results and finally, in Chapter 9, we draw conclusions and compare our results with similar studies on the proposal of the new FTK.
Background
Chapter 2
Dutch pension landscape
In order to understand the research problem we first describe the environment in which our research is done: the Dutch pension system. We look at the three pillars that the Dutch pension system consists of and pay special attention to the occupational pension in the second pillar, because this is the object supervised by the Dutch Financial Assessment Framework (FTK, Financieel Toetsingskader).
2.1
Three-pillar system
The Dutch pension landscape is shaped by a system consisting of three pillars. The first pillar is the general old-age law (AOW, Algemene Ouderdomswet ) and is managed by the government. The AOW provides an income for elderly to ensure a minimum stan-dard of living. The size of the AOW-payments is linearly dependent on the number of years that someone between age fifteen and age sixty-five has been an inhabitant of the Netherlands. This pension is financed using a pay-as-you-go system (PAYG). A PAYG system means that the AOW-benefits paid to the pensioners are financed directly from AOW-contributions paid by the active working population. There is no fund in which money is accrued and invested for future indexation. The size of the AOW-payments depends on the level of the minimum wage.
The second pillar is the occupational pension. On top of the AOW in the first pil-lar, employees can accrue pension in the second pillar. The second pillar pension is offered by the employer and is partly or fully paid by the employer. In contrast to the AOW, an occupational pension is financed by contributions that have been paid in the past by the employer and/or employee and are governed by a pension fund. Pension funds manage the capital that is accrued from the contributions and invest this capital to be able to provide their participants with a pension that allows the same purchasing power over the years of accrual.
The third pillar is an individual pillar that is not mandatory. The third pillar is not directly relevant for the content of this thesis and therefor not discussed, but mentioned for the sake of completeness.
2.2
Second pillar pension
In the remainder of this thesis we focus on the second pillar: the occupational pension. Whereas the first pillar pension is rather straightforward, the second pillar pension can be shaped in various ways: occupational pension, e.g., can be accompanied with a pension for widows or for orphans which are listed and described in Section 2.2.1. These pensions are provided by several types of institutions which are further explained in Section 2.2.2. These institutions, in turn, offer several types of pension plans, which
differ in the way how pensions are financed and how indexation is handled. Section 2.2.3 gives an overview of possible pension plans.
2.2.1 Pension types
Employees accrue pension by paying contributions during their working life. In return they receive benefit payments after a certain retirement age. This type of pension is called the age pension. In this type of pension benefit payments are only given when the participant reaches the retirement age. However, not all participants reach retirement age. In that case, a pension could be arranged for the relatives that are left behind. In this thesis we only take age pension into account. Nevertheless, for the sake of completeness we briefly discuss the other most commonly found types of pension.
Age pension The age pension provides workers with an income after their retirement. Premiums are paid by the worker, the employer, or both parties. This way, the worker accrues pension rights. Pensions can be seen as deferred wage income. Therefore, a maximum accrual rate and special tax legislation exist for occupational pensions. The retirement age for the age pension is determined in the pension contract and is not necessarily equal to the retirement age of the AOW.
Partner pension and orphan pension Partner pension provides the widow with a pension income to ease the financial burden as a result of the lost income of the deceased. This is predetermined and is often 70% of the pension that would have been received had death not occurred. Depending on the pension contract, pension is granted whenever the main insured dies or only if death occurs before retirement age. A similar provision exists for children. This is called orphan pension and is usually 14% of the original pension. Children often receive this kind of pension until they turn eighteen, although there are exceptions for children who study at an university. They can receive orphan pension up to age twenty-seven.
Disability pension In case of disability there is no or only partial regular wage income. This is combined with benefits paid by the government so that a disabled can retain a certain standard of living. These benefits can be increased with a disability pension, to be arranged via the pension provider. In contrast to the age pension, benefits can be paid before retirement age.
2.2.2 Pension providers
The occupational pensions are managed in pension funds or at an insurer. These financial institutions manage the joint contributions of the participants and invest these to build up capital. This capital is used for benefit payments after retirement. Depending on the risk taken in the investment portfolio, the fund can also use this capital to correct the benefits for inflation, which is called indexation. Indexation ensures that the pension retains its value in real terms. In this thesis we use a stylized model where pension is accrued in an industry-wide pension fund. To understand the features of this kind of fund, we make a comparison with other common pension providers, namely a company pension fund and an insurer.
Industry-wide pension fund An industry-wide pension fund is affiliated to several companies that operate in the same industry. Participation is often mandatory. This type of pension fund is used for our analysis.
6
Company pension fund A company pension fund is a fund typically affiliated with one single employer. Multinationals often have their own company pension fund. Legally, the fund stands apart from the company itself, so that pensions of the workers are not at risk when the company finds itself in financially difficult times.
Insurer Companies which are not required to affiliate to a pension fund can accom-modate their employees’ pension at an insurer. The difference between a pension fund and insurer stems from the way they bear risk. An insurer is required to guarantee nominal pension benefits. It is not possible for an insurer to decrease the pension rights. Due to these guarantees the insurer is subject to stricter buffer requirements than a pension fund. Whenever the insurer sees high returns on their investments they can, like pension funds, give indexation. However, when the insurer has not had enough re-turn on their investments, employers might have to pay extra contributions to finance the indexations. Additionally, the insurer differs in the required contribution rate. A company with relatively old employees has to pay more than a company with relatively young employees, because of higher pension costs.
2.2.3 Pension plans
A pension plan defines what guarantees are given with respect to the benefits. Further-more, this plan specifies how these payments are financed with regards to the contri-bution that has to be paid by the participants. Pension plans can differ in how they handle under- and overfunding. Different steering instruments can be used depending on the type of pension plan. Traditionally, most pensions in the Netherlands were fol-lowing a Defined Benefit plan (DB). However, in an attempt to address a sequence of problems at pension funds during the 2000s, pension funds increasingly shifted towards a Defined Contribution plan (DC). This has ultimately resulted in most pension plans in the Netherlands being a hybrid plan, which is a combination of a DB-plan and a DC-plan.
Defined Benefit plan In a Defined Benefit (DB) plan the only steering instrument that pension funds can utilize is the contribution rate. Benefits are guaranteed. A low funding ratio leads to a high contribution rate and a high funding ratio leads to contri-bution cuts. The guaranteed benefit level is defined in the contract with the two most common forms in the Netherlands being final pay and average wage. Final pay means that the level of benefits after retirement is based on the final wage of a participant. This can lead to a sudden increase in provision for the pension fund when the salary of the participant is raised. During the funding crisis in the early 2000s most Dutch pension funds reacted by switching from DB-final pay to the other type of DB-plan, namely DB-average wage (Ponds and Riel, 2009). Typical for this latter type of plan is that the level of the benefits is based on the average wage of the participant during his/her career. In a DB-average wage plan, indexation of accrued benefits is made dependent on the solvency position of the pension fund.
Defined Contribution plan In a Defined Contribution plan (DC) the contribution rate is fixed. Participants pay premiums and these premiums are invested. The pension that is ultimately received depends on the returns gained via these investments. In other words, the pension benefits are not guaranteed and depend on the number of years premiums are paid and on the returns earned from these premiums. In addition, at retirement the amount that is saved plus the amount acquired from the investments is mandatorily to be transformed to an annuity. The price of the annuity depends on the interest rates at that moment. Therefore, in addition to the investment risk the participant also has to bear transformation risk.
Hybrid plan Official statistics in the Netherlands classify the current average wage plans as DB-plans, but since indexation is made solvency-contingent this type of plan is more accurately described as a hybrid plan (Ponds and Riel, 2009). A hybrid plan is a combination of solvency-contingent indexation and flexible contributions. It is therefore partly a DB-plan because pension rights are accrued in a similar way, and it is partly a DC-plan because indexation is dependent on the financial position of the fund.
2.3
Financial Assessment Framework (FTK, Financieel
Toetsingskader )
Important for the sustainability and valuation of the Dutch pension system is the Finan-cial Assessment Framework (FTK, Financieel Toetsingskader ). It sets finanFinan-cial require-ments for pension funds to protect nominal rights of participants. These requirerequire-ments are based on the size of the risk taken by a pension fund. The FTK sets requirements on the level of capital that must be present. Furthermore, it states how assets and liabilities must be valued, and it prescribes what measures have to be taken when the financial position of the fund is at risk. This section treats the most relevant features of the FTK. Valuation The valuation of both the assets and the liabilities are done on a fair value basis, which means that the assets are valued using the daily market prices and the liabilities are determined by discounting the expected cash flows with the risk-free rate. This risk-free rate is also derived from market data. The FTK’s valuation method results in dependency on the financial market. Therefore, the capital of the fund has to be protected from the volatility of financial markets. The capital requirements set by the FTK are an attempt to protect the financial position of a fund against this financial market volatility.
Capital requirement (VEV, Vereist Eigen Vermogen ) To ensure that pension funds have enough assets to pay their liabilities in the future, the FTK sets two capital requirements. The first one is the VEV (Vereist Eigen Vermogen), which is defined on a risk basis. This means that the level of the VEV depends on the amount of risk taken by the fund. The level of the VEV is such that with a certainty of 97.5% the fund does not become underfunded within one year.
Minimal capital requirement (MVEV, Minimaal Vereist Eigen Vermogen ) The second capital requirement is a minimum capital requirement (MVEV, Minimaal Vereist Eigen Vermogen). This requirement stems from European guidelines and is set at a mimimum capital of 105% of the value of the liabilities at all times.
Recovery plans When a fund does not meet one of the capital requirements it has to compose a recovery plan. The FTK distinguishes two recovery plans, namely a long-term recovery plan and a short-long-term recovery plan. A long-long-term recovery plan has to be composed whenever the VEV-requirement is not met. In this long-term recovery plan the pension fund states how it can regain the VEV within fifteen years. The short-term recovery plan is composed whenever the minimal capital requirement MVEV is not met. This short-term recovery plan is stricter. It must describe how the pension fund’s funding ratio will be back at the MVEV within three years. Furthermore, it adds the restrictions that the pension fund cannot take more risk in the investment policy and that the investment policy does not influence the probability of indexation negatively. If the pension fund cannot accomplish the recovery in three years, then the fund must recover in the one year thereafter. It can then use benefit cuts, but only as a last resort.
8
Indexation To be able to compensate for inflation, pension funds have the ambition to increase the benefits with the same rate as the inflation. Pension funds can choose whether they have the ambition to grow their pensions with the price inflation or with the wage inflation. In the Netherlands, indexation is often conditional on the funding ratio. Pension funds can finance indexation through the capital buffer above the capital requirement VEV. It can also raise the premium, to cover for the cost of indexation. Cost-effective premium A pension fund is required to demand a cost-effective pre-mium. This means that the premium covers the costs for that particular year. The cost-effective premium consists of four parts. First, it requires a premium that covers the increase in liabilities due to accrual. Second, a premium is needed to maintain the capital requirement VEV. Third, a mark-up can be demanded for future indexation costs. And finally, execution costs need to be covered in the premium. These four com-ponents together lead to a so-called cost-effective premium that could change from year to year.
Since approximately 91% of the employees in the Netherlands accrues pension in the second pillar and because the capital in pension funds is higher than ever, the impor-tance of the FTK cannot be underestimated. The problems as a result of the financial crisis in 2008 therefore gave rise to extensive evaluation of the FTK. This ultimately led to a new FTK which is the research object of this thesis. In the next chapter we discuss how the proposal of the new FTK was established and we describe the altered features of the FTK.
Summary
The Dutch pension landscape consists of three pillars. We focus on the second pillar, i.e., the occupational pillar. Workers can accrue their pension in this pillar at a pension fund or an insurer. We focus on an industry-wide pension fund in particular. Such a fund has the characteristic that employers from the same industry are gathered in the same fund. Participation is quasi-mandatory. Within the pension fund several pension plans are possible which define how pension is accrued and how it is financed. The pen-sion funds are supervised through the financial assessment framework (FTK, Financieel Toetsingskader ). The FTK subjects the pension funds to certain capital requirements and prescribes what has to be done whenever these requirements are not met. Further-more, the FTK defines how a fund must value its assets and liabilities. Finally, the FTK demands a cost-effective premium.
Policy redesign: new FTK
On January 1, 2015 a new financial assessment framework will come into force: the new FTK. It is an adjusted version of the current financial assessment framework namely: the Financieel Toetsingskader (FTK). Years of discussion about how to strengthen the Dutch pension system have led to the proposal of the new FTK. In this chapter we give a short overview of how years of discussion ultimately led to the new FTK, what the intention of the changes is and we discuss the different changes in the FTK separately.
3.1
The road to a new FTK
The new financial assessment framework (new FTK) is the end of years of pension de-bates in the Netherlands. The route to this proposal started in 2008 during the financial crisis. Low interest rates and low investment returns had put pension funds under pres-sure. In addition, an increased life expectancy had put pension funds under even more pressure. These developments resulted in a low funding ratio at pension funds, which is the measure of the financial position of the fund.
Committee Frijns and Committee Goudswaard In reaction to the problems caused by the financial crisis, the Dutch government appointed two committees in 2009, namely the committee Frijns and the committee Goudswaard. These committees had to investigate what the underlying problems were that resulted in the troubles at pen-sion funds and how these problems could be addressed. They found that adjustments of the occupational pension pillar and of the financial assessment framework were nec-essary. First, they observed that the problems were caused by the low interest rates during the financial crisis. Since pension funds have to compute their liabilities on a fair value basis, low interest rates led to high liabilities and thus to low funding ratios. An increased life expectancy put the funding ratios under even more pressure. Second, they observed that the premium has become insufficient as a steering instrument. This is caused by the aging of the population of participants in a fund. There are relatively many retirees with respect to active participants. This means that increasing the contri-bution will hurt the paying active participants but will have relatively little effect on the funding ratio. Third, they found that since especially the downward risks have material-ized, participants had to face indexation cuts and in some cases even benefit cuts. This has led to deterioration of the trust of the participants in the pension system as a whole. The committees recommended making structural changes to the pension system. Among other things they recommended a pension contract that more explicitly distributes risks among the participants, a better communication to participants and a new balance between ambition and costs. At last, they recommended increasing the capital require-ment (VEV, Vereist Eigen Vermogen) with approximately 5%. This increase is needed
10
to meet the requirement that with a certainty of 97.5%, the fund will not find itself in a position of underfunding in the coming year.
Hoofdlijnennota The recommendations of the two committees were used as input for the discussion between the government and the social partners to restructure the pension system. The debate resulted in the agreement to make pension rights fully conditional. Both before and after retirement pension claims can be raised or lowered depending on the financial position of the fund, which requires a revision of the pension law. The main issues with respect to this revision were worked out in a note by the government called the Hoofdlijnennota in 2012. Starting point in this note was the option to choose between two types of contracts: a nominal contract and a real contract. In the nominal contract, the technical provision of the fund only consists of nominal rights. The conditional indexation is financed via buffers. In the real contract, indexation is already incorporated in the technical provision of the fund. Therefore, indexation is financed through this technical provision.
Septemberpakket In September 2012, the Dutch government decided to address the problems of the short-term recovery plans. The fixed length of the recovery period caused problems when additional aversive shocks occurred at the end the recovery period. The government made it possible to spread benefit cuts over time. This measure had to contribute to more stability. Additionally, the Ultimate Forward Rate (UFR) was introduced. This UFR-method is used to determine the long-term risk-free interest rate. Likewise, the introduction of this methodology has to contribute to a more stable pension system. In return, pension funds were allowed to give indexation only when the nominal funding ratio is above 110%.
Internet consultation Together with the Hoofdlijnennota the government published a preliminary draft of the proposed changes in legislation for an online consultation. This has led to a significant number of reactions by parties from the pension industry. A conclusion that can be drawn from these reactions is that instead of the two types of contracts, a more unequivocal supervisory framework is preferred. An unequivocal framework contributes to more clarity which helps to recover the confidence and trust of participants in the pension system.
Final proposal The reactions received from the internet consultation were used as final input for the current proposed reforms of the FTK. The framework with two types of contracts has been dropped. Instead only one contract will exist with nominal rights and conditional indexation. Additionally, the most important advantages of the earlier proposed real contract will be incorporated in the new FTK. The new FTK makes it possible to spread financial shocks over time so that pensions are less dependent on the vagaries of the financial markets. This contributes to a more stable pension. The recovery plans are also reformed such that they give room for an investment policy that better suits a conditional indexed pension. Furthermore, the contract will be more complete and the required capital will be increased.
In the remainder of this thesis we release the thought that we are dealing with a proposal and assume that the actual measures taken in the FTK are the ones that are stated in the proposal. Essentially, this assumption is based on the expectation that a large number of the measures mentioned in the proposal will be implemented.
3.2
The new FTK
We now treat the new FTK in more detail. We discuss the most relevant features in the new FTK separately and compare these with their counterparts in the old FTK. In general, the new FTK intends to make the pension contract more complete. Further-more, it makes pensions less dependent on shocks in the financial markets. Finally, it stabilizes the cost-effective premium.
Policy funding ratio The new FTK uses an average of the funding ratio to evaluate the financial position of the fund. This average is called the policy funding ratio. The funding ratio is averaged over twelve months. The averaging obviously results in a more stable funding ratio. In contrast, the old FTK used a three month average of the term structure of interest rates in combination with the UFR method to compute the liabilities.
UFR (Ultimate Forward Rate) The convergence level of the Ultimate Forward Rate (UFR) is changed from a fixed 4.2% to a 10-year moving average of forward rates. The UFR was introduced in the Dutch pension system in September 2012 with a convergence level of 4.2%. It is a method to compute long-term risk-free interest rates. VEV (Vereist Eigen Vermogen) The capital requirement VEV is increased with approximately 5% compared to the old FTK. This increase is needed to meet the 97.5% certainty requirement. This means that with 97.5% certainty the pension fund will not find itself in a position of underfunding next year. Upon evaluation of the FTK it came to light that under the current market conditions, the old model for computing the VEV produced a VEV that was too low. For an average fund this was approximately 21.7%. In the new FTK this becomes approximately 26.6% (CPB, 2014). The reason for this underestimation is that markets have become more volatile since the introduction of this model.
Recovery plans In the new FTK pension funds are required to compose a recovery plan when the policy funding ratio is below the funding ratio needed for the VEV-requirement. In this recovery plan the pension fund has to show that it can regain the capital of the VEV within a period of ten years. Moreover, pension funds can choose a shorter recovery period than ten years. For every year the VEV-requirement not being met, a new recovery plan has to be composed. This means that the actual recovery can take longer than ten years. Essentially, it comes down to an elimination of one tenth of the shortage every year. Whenever the fund regains the VEV, the recovery plan is dropped. In their recovery plans, pension funds can use the recovery capacity that is implied in their investment policy. Whenever this capacity is insufficient, additional measures have to be taken by the fund, e.g., indexation cuts or raising the contribution. Only if all measures do not lead to an expected regain of the VEV, nominal benefit cuts can be used.
The short-term recovery plan is excluded in the new FTK. This recovery plan had to be composed whenever the fund did not meet the MVEV. However, in the new FTK immediate actions have to be taken whenever the MVEV-requirement is not met for five consecutive years. The new FTK states that this may be achieved by one single benefit cut, but it is also possible to spread this cut unconditionally over a period of maximum ten years.
The moving recovery period in the recovery plan is an improvement with respect to the old FTK. In the old FTK, there was a fixed length of the recovery period. Shocks
12
occurred at the end of a recovery period were typically hard to recover from. This has led to significant and abrupt cuts of nominal rights. Moreover, the old recovery plans made it hard for pension funds to adapt their investment policy to their indexation ambition. The new FTK is an attempt to address these problems.
Indexation The new FTK puts requirements on when and to which extent indexa-tion may be given. It therefore introduces an indexaindexa-tion threshold in terms of the policy funding ratio. This threshold is set at 110%. Whenever the funding ratio is above this threshold, indexation is allowed. In the old FTK, indexation was essentially possible whenever the MVEV was met. However, indexation that was given in case the fund had a funding ratio was just over 105% could easily lead to a situation of underfunding in the year thereafter.
The new FTK also prescribes what size of the indexation that may be given. The degree of indexation should be such that this degree of indexation can also be given permanently in the future, which means that the fund must be able to increase the pension rights in the future every year with this same degree of indexation. The costs of this future indexation are determined by discounting the future indexation cash flow with a discount rate, which has a maximum that is defined by the expected return on stocks. In practice, this comes down to approximately 1% indexation for every 10% of the funds funding ratio in excess of the indexation threshold.
Furthermore, catch-up indexation and compensation for benefit cuts from the past are allowed when the funding ratio is higher than the VEV and higher than the funding ratio for which full indexation is possible. One tenth of the difference between this lower bound and the actual capital can be used for catch-up indexation or compensation for past benefit cuts. To prevent redistribution between generations, indexation is only pro-vided to those groups that have actually experienced the cuts. These regulations make it possible to spread the capital surplus over time. This spread is in favor of the younger participants. In contrast, the new regulations concerning the recovery plans are in favor of the older participants. The combination of the new recovery plan and indexation rules is expected to result in a balanced effect on the generations of an average fund (Klijnsma, 2014). In this thesis, we check whether this expectation is accurate.
Cost-effective premium In the new FTK, pension funds can base their premium on a 10-year moving average of the risk-free interest rate or on the real expected return. Funds that base their premium on the risk-free interest rate, have to take into account a mark-up in order to maintain the capital requirement VEV. Funds that base their premium on the expected return have to take into account an increase of the discount rate for the risk in their investment portfolio. For a 50% stocks and 50% bonds portfolio this comes down to an increase of 1.5%. Furthermore, they have to take into account a decrease of the discount rate for the costs of indexation. For wage-linked indexation this comes down to a decrease of 2.5%.
Additionally, the requirement that the premium has to contribute to recovery is dis-missed. The contributions are already at such a high level that increasing the contribu-tion rate could have undesirable effects on a macro-economic level. Furthermore, since pension funds in the Netherlands are rather mature (high number of older participants compared to number of younger participants), premium as a steering instrument has become insufficient. This is because an increase of the premium would have relatively little effect on the financial position of the fund.
Summary
The new FTK is an attempt to obtain a more stable financial assessment framework that is less dependent on the vagaries of the financial markets. The new FTK also makes pension contracts more complete by letting pension funds describe explicitly what will be done in financially good and bad situations. More specifically, the problems of the recovery plans are addressed. Instead of a fixed recovery period, the recovery plans now have a moving recovery period. The short-term recovery plan is dropped. Furthermore, the new FTK puts more regulations on indexation. There is an indexation threshold above which indexation is allowed (110%) and the size of the indexation is such that this indexation can be permanently provided in the future. Moreover, the capital requirement VEV is increased and a new, more stable funding ratio is introduced: the policy funding ratio. The separate features of the new FTK together are expected to have a more stable and balanced effect on pensions and they are expected to lead to little intergenerational redistribution (CPB, 2014).
Chapter 4
Literature review
In the previous chapter we discussed the changes of the FTK. Such changes to the pension system imply intergenerational value transfers due to the intergenerational risk sharing feature in the pension contracts. In this chapter we make clear why it is im-portant to gain insights in intergenerational effects due to policy changes and how in-tergenerational effects can be evaluated using generational accounting and a value-based approach.
4.1
Social security reforms
The change to a new Financial Assessment Framework (FTK) is part of a much larger trend to improve the sustainability of pension systems. The constant flow of changes already started in the beginning of the 2000s. The value of assets dropped and the li-abilities increased as a result of the drop in interest rates (Ponds and Riel, 2009). To make the underfunding problems more visible the Dutch pension funds adopted a new accounting method using fair-value principles, which made the funding ratio more de-pendent on the daily financial markets. As a result, the funding ratios have decreased heavily, which was amplified by the fact that pension funds adopted more risky invest-ment portfolios in search of higher returns.
These problems have led to a series of changes to improve the sustainability of the pension system. In the United States and in the United Kingdom the trend is to re-place defined benefit plans by defined contribution plans (Munnell, 2006). Moreover, a shift to more stand-alone pension funds can be observed (Ambachtsheer, 2007). In the Netherlands, the pension funds shifted from a DB-final pay to a DB-average wage plan, in which all accrued liabilities were made dependent on the solvency position of the fund (Ponds and Riel, 2009). Hence, the indexation policy and the contribution rates are both depending on the funding ratio. This is also an important assumption in the model that we discuss in Chapter 5.
4.2
Intergenerational risk sharing
Additionally, population ageing and the maturity of pension funds have put even more pressure on pension funds. Given the maturity of pension funds, benefit cuts are an effective way to improve the financial position of the fund (Ponds and Riel, 2009). How-ever, because these cuts were poorly communicated and because of the fuzzy definition of property rights on the pension funds wealth, these cuts were hard to implement (Teul-ings and De Vries, 2006). Questions were raised about the sustainability of the pension fund and in particular about the intergenerational risk sharing feature. However, studies have shown that intergenerational risk sharing within pension funds is welfare improving (Bonenkamp and Westerhout, 2014; Beetsma and Bovenberg, 2009). A well-structured
intergenerational risk sharing plan can be welfare enhancing compared to an individual plan. Furthermore, the expected welfare gain of the current cohort is not at the cost of the older and future cohorts (Cui et al., 2011). Gollier (2008) adds to this that the welfare gain is not because risk-sharing reduces the risk born by each generation but because it increases the expected return to the contribution. However, Boender et al. (2000) find that intergenerational risk sharing feature does reduce risks for individual participants and this effect is even stronger for funds that keep a buffer.
Due to the intergenerational risk sharing feature pension plan redesign leads to in-tergenerational transfers. The magnitude and the direction of the transfers are decisive for the sustainability of the reform. In general, the sustainability of a fund depends on the willingness of various stakeholders to participate. To ensure this willingness Ponds (2003) puts forward two criteria to evaluate alternative policies. First, the funding and allocation rules must give an ex ante fair compensation for the risks borne. Second, ex post redistributive effects between generations must be evaluated to ensure the sustain-ability of a pension plan.
The effects of policy reforms on the risk-sharing feature of generations are widely stud-ied. One way to evaluate these is the use of an overlapping generations model. This model originates from Fisher et al. (1965) and has also been applied in a pension envi-ronment (Chen and Ponds, 2012). This method is mainly used to show welfare effects due to policy reforms (Cui et al., 2011; Beetsma and Bovenberg, 2009). Another way to gain insights in the intergenerational effects is with generational accounting.
4.3
Value-based ALM
To evaluate the intergenerational effects in this thesis, we use the concept of genera-tional accounting. This is a method to record cash flows during the lifetime of different cohorts, which can be used to compute the present value of the changes in cash flows due to a policy change. The method originates from public finances and is based on an intertemporal budget constraint, which means that either current or future generations pay for government spending via taxation (Hoevenaars and Ponds, 2008). Generational accounting reveals that a pension system can be seen as a zero-sum game, which means that the increase in net lifetime income from one generation must be paid by a decrease in net lifetime income from another generation.
The concept was originally used to evaluate the fiscal policy of the US government and the effects for various generations (Auerbach et al., 1999; Kotlikoff, 2002). After that, studies followed in which generational accounting was used for fiscal policies in other countries (Van Ewijk and Donders, 2006; Bettendorf et al., 2011). Moreover, the concept proved to be also applicable to pension systems because of two similarities be-tween pension systems and public finances (Hoevenaars and Ponds, 2008). First, the intertemporal budget constraint is present in both pension systems and public finances. For pension systems this is reflected by the promised pension rights that have to be paid by current and future contributions. Second, both the government and the board of a pension fund have specific instruments to close the budget over time. The govern-ment uses the tax instrugovern-ment. The board of a pension fund uses the contribution and indexation instrument.
The concept of generational accounting was first applied to the Dutch pension system by Ponds (2003). He combined generational accounting with a value-based approach to price the stakes of different generations in economic value terms. This combination is known as value-based generational accounting or value-based ALM and it implies that
16
the benefit and premium payments that are recorded, are seen as contingent claims since they depend on the solvency position of the fund. The payments are then priced using either the deflator approach (De Jong, 2004) or risk-neutral scenarios (Hull, 2009). Both methods lead to the same result. In this thesis we use risk-neutral scenarios.
Value-based ALM is in fact an addition to classic ALM. Classic ALM gives insights in the distribution of future outcomes. It uses an economic model to produce stochastic simulations, which after a scenario analysis result in probability distributions for key variables (Hoevenaars and Ponds, 2008). Classic ALM is widely used by pension funds to manage risks and to gain insight in the sustainability of the fund. However, the method is also criticized because it only shows that the more risk you take, the more volatile the results become (Chapman et al., 2001). Value-based ALM adds new information relative to classic ALM (Kortleve et al., 2006). It outputs the economic value of future cash flows and therefore it can be used to evaluate value transfers between different cohorts due to a policy reform. Since a pension fund can be seen as a zero-sum game, value-based ALM shows who wins and who loses from a policy reform. Hence, it leads to extra insights, which can be crucial for the sustainability of a pension deal regarding the willingness to participate. A more extensive comparison between classic ALM and value-based ALM can be found in Kortleve et al. (2006).
Value-based ALM is also implemented in several other studies. Hoevenaars and Ponds (2008) rewrites the pension fund as a collection of embedded generational options and evaluates the intergenerational value transfers that arise from a plan redesign. Lekniute (2011) uses value-based ALM to evaluate several pension schemes like DB plans, col-lective DC and hybrid plans. In Lekniute et al. (2014), the same technique is employed to evaluated pension redesign in the United States. Finally, Ponds and Lekniute (2011) uses value-based ALM to evaluate possible intergenerational redistribution due to the implementation of the Pensioenakkoord, which is a reform of the Dutch pension system that was in fact the precursor of the new FTK. Some elements of this are visible in the proposal of the new FTK.
Summary
The pressure on the pension systems has raised questions about the sustainability of the system and in particular the intergenerational risk sharing feature. Several studies show that this risk sharing feature is, in fact, valuable and may be welfare enhancing. However, the risk sharing feature causes intergenerational value transfers as a consequence of policy reforms. Therefore, it is important for the sustainability of the fund to check these value transfers. A way to do this is with the use of generational accounting in combination with a value-based approach. Generational accounting records the lifetime payments of different cohorts and value-based approach prices the lifetime payments. Because generational accounting reveals that a pension fund is a zero-sum game, we can use value-based ALM to evaluate changes in terms of which generations win and which generations lose. This visualizes the value transfers between generations.
Method
Chapter 5
Model introduction
To evaluate the intergenerational effects of the shift to the new FTK we use a fictive pension fund that is based on real life characteristics. In this chapter we introduce this pension fund and the assumptions on the participants. Furthermore, we describe the tool used for evaluation, which consists of a classic ALM model and a value-based ALM model. For the latter, we use the concepts of generational accounting in combination with a value-based approach. Both are described in this chapter. Finally, for both the classic ALM model and the value-based ALM model, we use economic scenarios that are generated from an economic scenario generator. For this, we use the Vasicek model, which is also discussed in this chapter.
5.1
Population
The population of our fund is subdivided in cohorts numbered from −5 to 122. The number of the cohort indicates the age at the start of the model. In this sense, partic-ipants in cohort −5 are born 5 years from now. The cohort sizes are determined using data from Statistics Netherlands (CBS, 2014) and are scaled in such a way that the ratios between the cohort sizes represent those of the actual Dutch population in 2014 (see Appendix D). The sizes of those cohorts that are not born yet, i.e., cohorts −5 to −1, are estimated using the data from the Dutch population forecast 2013-2060 (CBS, 2013).
Mortality The cohort sizes are only adjusted for death. The mortality rates are ob-tained from a mortality table called the AG Prognosetafel 2012-2062, which is composed by the association of Dutch actuaries (AG-AI, 2012). This mortality table takes into account that life expectancy will increase in the future and therefore the survival prob-abilities will typically differ from year to year. In our model the survival probprob-abilities are used as deterministic input so that there is no longevity risk for the pension fund. Next, we formalize the sizes of the cohorts in our model. We denote the number of males of cohort x at time t as Mtx, with x ∈ X = {−5, . . . , 122}. In similar fashion we use Ftx for the number of females. The mortality rates of males qmx,t and females qx,tf is the probability that a person with age x dies between time t and t + 1. The initial cohort sizes are determined from the data, thereafter the cohort sizes develop as follows:
Mt+1x+1 = (1 − qx,tm) ∗ Mtx, (5.1)
Ft+1x+1 = (1 − qx,tf ) ∗ Ftx, (5.2)
for x ∈ X and t ≥ 0.
Wages We define an uniform wage level within each cohort, i.e., we neglect intragen-erational effects. The wage level is adjusted for both inflation and wage growth as a consequence of age dependent promotion careerx−1t−1, which is specified in Appendix D and stylizes the average Dutch career path. All cohorts follow the same career path. Also, there is no difference between price and wage inflation. We assume one uniform and constant inflation rate π = 2%. This gives the following wage development:
Wtx = Wt−1x−1∗ (1 + π) ∗ (1 + careerx−1t−1), (5.3) for x ∈ X and t ≥ 0.
Accrued benefits The wage levels are used to compute the cohort-specific accrued benefit levels Bt,sx . Like the wages, we assume that the accrued benefit levels do not depend on gender. Every participant which is older than 25 and younger than the retirement age accrues benefits at an accrual rate of 2%. The starting values of the benefits at t = 0 are equal to initial wage levels at t = 0 times the accrual rate times the number of accrual years. This gives the following accrued benefits development:
Bt,sx = 0 if x < 25, Bt−1,sx−1 + 0.02 ∗ Wt−1,sx−1 if 25 ≤ x ≤ 67, Bt−1,sx−1 if 67 < x, (5.4) for t ≥ 0 and s ∈ S.
5.2
Pension fund
The pension fund in our model is a stand-alone industry-wide pension fund. There is no risk-absorbing sponsor, thus all risk is borne by the participants of the fund. All management decisions of the fund are based on the funding ratio. We define both a nominal FRt,s and a real funding ratio FRRt,s. For convenience, we omit the superscript
in case of nominal terms and we denote a variable in real terms with a superscript R. Furthermore, the funding ratios are dependent on the scenario s ∈ S = {1; . . . ; 1, 000}. We get:
FRt,s= At,s/Lt,s, (5.5)
FRRt,s= At,s/LRt,s, (5.6)
for t ≥ 0 and s ∈ S.
Assets The development of the assets depends on the previous level of the assets, contributions Ct,s paid by the participants, the benefits Bt,s paid to the pensioners
the and return Rinvestments
t,s on the investment portfolio. The total benefits paid is the
sum product of the cohort-specific accrued benefit levels Bt,sx and the total number of participants of cohort x, where the sum is only taken over the participants that are older than retirement age. This investment portfolio consists of 50% bonds and 50% stocks and this ratio does not change throughout the course of the model, that is, the fund adjusts its value of stocks and bonds such that this ratio is maintained. This gives the following development of the assets:
At+1,s= (At,s+ Ct,s− Bt,s)(1 + Rinvestmentst,s ), (5.7)
20
Liabilities The determination of the development of the liabilities starts with the calculation of a deferred annuity m|ax,ζt,s, that is, the present value of all future benefit payments after retirement resulting from one unit of accrued benefits at time t. This present value depends on the survival probabilitiesipxt, i.e., the probability that at time t a person aged x will survive i years in the future. Again, the AG Prognosetafel 2012-2062 is used. Furthermore, it depends on the nominal term structure rt,si (Ponds and Riel, 2009). Multiplication of the deferred annuity with the actual benefits Bt,x until
time t results in the age- and cohort-specific liabilities Lx,ζt,s. After accounting for the actual cohort sizes Mtx and Ftx, we get the total liabilities Lt,s:
m|a x,ζ t,s = 122−x X i=m ip x,ζ t (1 + rt,si ) −i, (5.8) Lx,ζt,s = Bt,sx ∗max(67−x,0)|ax,ζt,s, (5.9) Lxt,s= Mtx∗ Lt,sx,m+ Ftx∗ Lx,ft,s, (5.10) Lt,s= 122 X x=25 Lxt,s, (5.11) for s ∈ S, t ≥ 0, x ∈ X and ζ ∈ {m, f }.
5.3
Simulation
To evaluate the effects of a policy reform, we run our model through time. First, we set an initial wage level and an accrued benefit level based on age. From the accrued benefit level and the cohort sizes we calculate the initial level of liabilities, which multiplied with the predetermined initial funding ratio leads to the initial level of assets. Based on the funding ratio a management decision on indexation, nominal cuts and premium is derived for the next year.
After initialization the management decision on indexation and cuts is executed, that is, the benefit levels are increased or decreased. Directly thereafter, all benefit payments are made and all contributions are received, i.e., all cash flows occur at the beginning of the year. This results in the value of the assets next year (see Equation 5.7). To calculate the liabilities next year we start by correcting the cohort sizes for mortality (see Equation 5.1 and 5.2) and by correcting the wage for inflation and promotions (see Equation 5.3). The benefits are then updated for new accrual, which is based on the obtained sizes of the cohorts and the wage levels. From the age and gender specific ben-efits, the age and gender specific liabilities can be derived (see Equation 5.9) and this ultimately leads to the total liabilities (see Equation 5.11). Next, the funding ratio is computed (see Equation 5.5), which again, is the input for a policy-specific management decision on indexation, nominal cuts and premium. With this management decision the procedure starts all over again. This procedure is repeated until the time horizon is reached.
Real world scenarios The simulation procedure produces one run of the model under one specific scenario. In classic ALM we utilize a whole range of scenarios that are generated from the same process. In this thesis, we use a Vasicek model to obtain 1,000 real world scenarios (Vasicek, 1977). Since an economic scenario generator deserves a thesis on its own, e.g., Plomp (2013), we only briefly discuss it. The Vasicek model is a one-factor model, which has a mean reverting feature and is easily implemented. We use it to model the short rate and the stock prices, which are related to each other with a correlation factor ρ. A more detailed description can be found in Appendix C.
We get:
drt= ar(µr− rt)dt + σrdWtr,
dSt= (rt+ µSt)Stdt + σSStdWtS
with Cov[Wtr, WtS] = ρt.
5.4
Value-based ALM
With value-based ALM we can express a pension policy in economic value terms and therefore, we can express the change in economic value as a result of policy changes. In general, value-based ALM shows which stakeholders will gain and which stakeholders will lose from a policy change. These stakeholders are in this case the different gen-erations. Therefore, we rewrite the pension fund in generational accounts and apply a value-based approach (Hoevenaars and Ponds, 2008).
Generational accounting A generational account for a particular cohort comprises the cash flows and the claim on the residue. The cash flows are recorded in a matrix Gs. In this matrix, the for cohort x specific contributions cxt,s and benefits bxt,s in each
time step t are recorded as long as the model runs. The rows of this matrix represent the generational account of the different cohorts and the columns account for the time steps. Therefore, the first element of the matrix is always zero since cohort −5 starts contributing thirty years from now. We get a new matrix Gs for each scenario s:
Gs= 0 0 · · · c25,30,s .. . ... ... ... c250,s c261,s · · · c5530,s c260,s c271,s · · · c5630,s .. . ... · · · · c660,s b671,s · · · b9630,s b67 0,s b681,s · · · b9730,s .. . ... . .. ... b1220,s 0 · · · 0 , (5.12) for s ∈ S.
The pension fund’s residue is the difference between assets and nominal liabilities. The residue is used to account for the financial position of the fund. Like Hoevenaars and Ponds (2008), we allocate the residue among the cohorts proportionately to each co-hort’s stake of nominal liabilities, i.e., the share lxt,s of the residue is the total cohort specific liabilities Lxt,s (see Equation 5.10) divided by the total liabilities for all cohorts Lt,s. The cohort-specific residue value Rxt,s is the multiplication of this share and the
value of the total residue Rt,s. We get:
lxt,s= Lxt,s/Lt,s, (5.13)
Rt,s= At,s− Lt,s, (5.14)
Rxt,s= lxt,s∗ Rt,s (5.15)
for s ∈ S and t ≥ 0.
The generational account is the sum of a net benefit part and a residue part (Ho-evenaars and Ponds, 2008). The net benefit part is the collection of the present value
22
of all benefit payments, all contributions occurred in the past and the change in liabil-ities Lx
t,s during the course of the model, which is thirty years. The residue part is the
difference between the residue value at the start and in the end of the model. We use the notation V0[. . . ] to denote the economic value at t = 0, which is calculated by
risk-adjusted discounting of the possible outcomes, that is, either with a deflator technique or with a risk-neutral valuation. However, both methods result in the same outcome for V0[. . . ] (Ponds, 2003). Here, we use risk-neutral valuation. For comparison of the
different policies we are especially interested in the generational account at t = 30. For this we get: GAx30= (V0[Lx30,s] − Lx0,s) + 30 X i=0 V0[bxi,s] − 30 X i=0 V0[cxi,s] + V0[Rx30,s] − V[Rx0,s], (5.16) for t ≥ 0.
Risk neutral scenarios The economic value of a generational account is determined with risk-neutral valuation. We use risk-neutral scenarios, which are transformed sce-narios from the real world scesce-narios. The use of risk-neutral simulation is essentially the expectation under the Q-measure of the cash flows discounted with the risk-free rate (Lekniute et al., 2014). The risk-neutral scenarios assign more weight to unfavorable scenarios than to favorable scenarios. This way, the risk-averse property of participants is included.
Policy variants
The reform of the FTK can be divided in multiple individual measures, e.g., measures on indexation and measures on the recovery plan. To evaluate the individual effect of a particular measure we first examine the measures separately. We do this by selecting a benchmark policy and modifying its features one at the time. In this chapter we describe our benchmark policy and variants of this policy that reflect measures of the new FTK.
6.1
Benchmark policy
The benchmark policy represents a policy in the old FTK setting, hence enabling us to compare the results of policy changes in the new FTK relative to the old FTK. We subdivide the characteristics of this policy into two parts. First, we consider the pension fund’s characteristics, which hold for all policies considered and represent the characteristics of an average pension fund in the Netherlands. Second, we consider pol-icy characteristics, which represent the old FTK and are altered in the new FTK setting. We emphasize that many different policies are possible within the boundaries of the financial assessment framework FTK. Therefore, we choose a pension fund that has the characteristics of an average Dutch pension fund. The benchmark policy we consider is chosen such that that an altered version of its characteristics can represent the shift from the old to the new financial assessment framework.
6.1.1 Pension fund’s characteristics
We consider a pension scheme that is partly defined benefit (DB) and partly defined contribution (DC) (i). The DB part is reflected in a contribution rate, which is contin-gent on the funding ratio. The DC part is reflected in a yearly indexation rate that is related to the financial position of the fund (Ponds and Riel, 2009). Participants enter this hybrid scheme at the age of twenty-five, retire at age sixty-seven (ii) and in between participants work and accrue 2% of their wage level each year (iii). The pension fund has the ambition to index the pension rights for inflation π and the indexation rate given is uniform across all generations. We assume that the initial pension rights are fully indexed (vi).
The pension fund has an initial funding ratio of 110% (viii), which represents the av-erage funding ratio of Dutch pension funds in the first quarter of 2014 (DNB, 2014). Furthermore, the fund’s investment portfolio consists of 50% stocks and 50% bonds (ix). We assume that the pension fund rebalances at the end of the year to this asset mix, hence the asset mix remains constant over time. For convenience the pension fund’s characteristics are put in an overview.
24
Overview of the pension fund’s characteristics
i. The pension scheme is an average wage hybrid DB-DC scheme; ii. Participants enter at age 25 and retire at age 67;
iii. The accrual rate is 2%;
iv. Benefits are indexed for inflation π;
v. The indexation rate yt is uniform across generations;
vi. At the start of the model pension rights are fully indexed; vii. The contribution rateect is uniform across generations; viii. The initial funding ratio is 110%;
ix. The investment portfolio consists of 50% bonds and 50% stocks.
6.1.2 Policy’s characteristics
The pension fund is subject to two capital requirements. First, the minimal capital requirement MVEV is set at 105%. Second, the capital requirement VEV depends on the amount of the risk that a fund takes. For simplicity, we assume that the VEV is constant over time and use the level of the VEV for an average Dutch pension fund, i.e., 121%. At the end of this section, we show an overview of the policy’s characteristics. Long-term recovery plans Whenever the capital requirements are not met, the pension fund has to compose a recovery plan. For a funding ratio below the capital requirement VEV, a long-term recovery plan is required. The intention of this recovery plan is to restore the funding ratio to the VEV by increasing premiums and cutting indexation. Premiums are increased by demanding an additional premium. The index-ation rate is decreased proportionally to the distance of the funding ratio to the VEV. More specifically, the indexation rate ytis linearly interpolated between the MVEV and
the VEV. We get:
yt(FRt) = π if FRt≥ VEV, π ∗ FRt−MVEV
VEV−MVEV if MVEV < FRt< VEV,
0 if FRt≤ MVEV,
(6.1)
for t ≥ 0.
Short-term recovery plan If the funding ratio is below the MVEV additional mea-sures are needed, which are described in a short-term recovery plan. The long-term recovery plan has in this case already ensured that no indexation is given and premiums are increased. In addition, the short-term recovery plan demands that one-third of the deficit is eliminated by nominal cuts. If the MVEV is not met for three consecutive years, it is even demanded that the deficit is fully eliminated with a benefit cut. We get:
cutt= 1 − At MVEV∗Lt if years MVEV t,s ≥ 3, 1 3(1 − At MVEV∗Lt) if 0 < years MVEV t,s < 3, 0 if years<MVEVt,s = 3, (6.2)
for t ≥ 0 and yearsMVEVt,s is the number of consecutive years that the funding ratio is under the MVEV.
Contributions The demanded contribution consists of three components, namely a base contribution Cbase
t,s , an additional contribution Ct,sadd and a mark-up for the capital
requirement (see Equation 6.3). The base contribution is equal to the fair value of new accrued liabilities. The additional contribution is part of the recovery plans and increases the contribution rate whenever the funding ratio is below the VEV. The elimination of the deficit below the capital requirement is smoothed over thirty-five years. In case the funding ratio is higher than 135%, the additional premium can become negative, which means that premiums are cut. Likewise, negative additional premiums are determined from a surplus that is smoothed over thirty-five years. At last, the sum of the base and the additional contribution is multiplied by the VEV to obtain a mark-up for the capital requirement.
To compute the uniform contribution rate, we divide the total contributions Ct by
the total wage level. To control for extremely unrealistic premiums we set boundaries to the contribution rate, that is, the contribution rate cannot be higher than 35% of the wage level and it cannot be lower than 5% of the wage level. We get:
Ct,s= (Ct,sbase+ Ct,sadd) ∗ VEV, (6.3)
e Ct,s= max 0.05, min Ct,s P x(Mtx+ Ftx) ∗ Wtx , 0.35 , (6.4) for s ∈ S and t ≤ 0.
Catch-up indexation Catch-up indexation is extra indexation that is given to com-pensate for indexation cuts or benefit cuts from the past. This form of indexation is given when the funding ratio is above the VEV. We use 30% of the surplus in capital with respect to the VEV for catch-up indexation (see Equation 6.5). Next, we determine how much indexation is missed missedxt,sby a particular cohort. This is the difference be-tween the actual cohort-specific liabilities Lxt and the cohort-specific liabilities Lx,full-indt that would have been reached if full indexation was given every year. We get:
surplust,s= 3
10(At,s− Lt,s∗ VEV), (6.5)
missedxt,s= Lx,full-indt − Lxt, (6.6)
for s ∈ S and t ≥ 0.
We assume that all generations are equally compensated. Hence, we define a uniform compensation rate compt,s across all generations, that is, the percentage of the missed indexation that will be compensated. We get:
compt,s= 1 if surplust,s≥ missedt,s, surplust,s/P
xmissedxt,s 0 < if surplust,s< missedt,s,
0 if surplust,s≤ 0,
(6.7)
for s ∈ S and t ≥ 0.
The actual value of the cohort-specific catch-up indexation is computed by multiplying the uniform compensation rate with the cohort-specific missed indexation. This gives:
catch-upxt,s = compensationt,s∗ missedxt,s, (6.8) for s ∈ S and t ≥ 0.
