ALM PRODUCTS COMPARED Alphen, Joeri van; As, Jelles van; Brings, Hans F.; Heerdt van, Wout; Steenkarnp, Tom; Valkenburg, Falco R.;Wenting, Dick VBA Committee Portfolio Management Dutch Society of Investment Analysts (VBA) C/OA ctuarieel Bureau Hammer Postbus 1364 3000 BJ ROTTERDAM The Netherlands Telephone: +31 10 207 55 00 Facsimile: +31 10 456 07 57 e-mail: [email protected] Abstract In June 1996 a product presentation of five ALM consultants was organised by the VBA Committee Portfolio Management. To our knowledge this has been the first time in the world that five providers of ALM models presented their products on the same day, on the same floor, using the same model pension fund. The five providers were: A Frank Russel Company A ORTEC Consultants A PVF Nederland A Towers Perrin A W.M. Mercer Klein Haneveld This paper summarizes the findings of this most interesting event. The objectives were getting to know better the different approaches to ALM, offering pension funds and insurance companies the possibility to see five ALM organisations on one day and learning where the differences in the approaches are and why. It has not been the objective to come to a kind of ranking of the providers of ALM models. As an introduction a short overview of the Dutch history in ALM is given. The case the working group has put together for the providers of ALM models is introduced in the second part of the paper. All providers of ALM models have received the same case, scenario's and questions. This in order to be able to compare the results and, more important, the approach to come to the results as good as possible. In the third and most important section the main results and the differences in approach are discussed. The highlighted topics are: A Organization A ALM approach A Planning horizon A Scenario's A Economical assumptions A Management devices A Recommended investment policy A Findings case-study We, and with us all participants of this very practical ALM event, have enjoyed this day extremely and have found it very valuable. We hope the findings are also of interest for colleagues outside the Netherlands. We strongly recommend to organize similar events in your own country. CONTENTS Abstract 1. Preface 1.1. VBA Committee Portfolio Management 2. Pensions and Investments in the Netherlands 3. ALM in the Netherlands 3.1 Introduction 3.2 ALM is integral risk 4. Description of the case "A Dutch pension fund" 5. Analysis of the results of the ALM studies 5.1 Introduction 5.2 Organisation 5.3 ALM approach 5.4 Planning Horizon 5.5 Scenarios 5.6 Economic assumptions and assumptions about the financial market 5.7 Instruments 5.8 Advice on investment policy 5.9 Findings from the case study 5.10 Conclusions 6. Bibliography 7. Endnotes 1. Preface 1.1 VBA Committee Portfolio Management The Committee Portfolio Management was founded at the end of 1988 initiated by several members of the Dutch Society of Financial Analysts (VBA). The Committee's objectives are as follows: to gather, analyse, discuss and present theoretical and practical aspects of portfolio management in a changing world. The Committee has about fourty members. They come from banks, insurance companies, consultancy pratices (including actuarial), pension funds and universities. The expertise of the individual members is fully brought to bear in the selection of study topics, which are then pursued in further detail by small working groups. The results are published in reports and announced at presentations. At the end of 1990 an "Asset & Liability Management" working group was set up within the Committee Portfolio Management. End 1992 this group presented its first findings in a detailed report in Dutch. A translation of an abridged version was presented at the third AFIR Colloquium in Rome (1993). In 1994 the results of a survey on assets, liabilities and ALM amongst Dutch pension funds were presented. In 1995 a seminar on the same topics was organised. 2. Pensions and Investments in the Netherlands The Dutch pension scene is undergoing significant changes. The ageing of the population (baby boomers) places pressure on the future costs of provision for old age in general and early retirement in particular. Not only the pension funds are confronted by this, but it is above all the public sector which has to struggle with increasing costs within the social insurance system. Another aspect is that the citizens are becoming aware of the need to cater for their individual needs and demand opportunities to do so. These circumstances lead to the following picture: In their traditional form pension schemes were created in the Netherlands on the basis of the concept of caring for people, paying attention to financial security in old age and the covering of risks related to death and disability. In addition to the provision of pensions, early retirement schemes were created, in which making employment available to others was the primary aspect. In recent years the view has arisen that the traditional schemes are too rigid and that in particular the Voluntary Early Retirement Schemes (VUT) offer little security. By definition VUT schemes are temporary and in many cases there is concern about their cost. Naturally conversion to a pension also results in costs but these are easier to manage. For the employees it is important that a pension offers more security in the long term and that the entitlements are accrued during employment, which reduces the likelihood of a shortfall in pension. Social developments lead to the demand for more flexible schemes in order to offer employees more options in order to be able to cater for individual needs. Family circumstances can be an important factor in this. The Dutch government also recognises the altered circumstances which have resulted in a loss of social security. Thus the WAO [Dutch law on invalidity insurance] and AWW [Dutch Widows and Orphans Act] have been considerably restricted. In many cases the social partners adapt to this by making substitute provisions within the pension schemes. In fiscal terms the increasing desire for individualisation has led to a study to determine adjustments in the fiscal treatment of supplementary pensions and the associated fiscal regulations are desirable and possible. A report was issued on this in 1995'. Far-reaching possibilities for increasing flexibility were proposed. There can thus be said to be an enormous shift in the system of provision. The Dutch government is taking a step back and at the same time the funds are adapting their schemes in the private sector. A great deal is at stake: Within Continental Europe the Netherlands holds first place in terms of pension assets. Internationally the Netherlands holds fifth place behind countries such as the USA, the United Kingdom, Canada and Japan. The Netherlands also score well in terms of pension coverage: a good 85% of employees are covered by a pension scheme, with invested assets of some F1. 600,000 million (USD 315,000 million) It is not only in terms of absolute value that what is at stake is increasing. This is emphasised by looking at the trend of pension fund assets in relation to the GDP: So far we spoke about the content of the pension schemes was discussed. However, also on the yield side a lot is happening. The Netherlands is not alone in this. The stock markets in Europe and America have performed magnificently in recent years. Interest rates are now historically low and share prices regularly break records. Thus in the years 1993-1995 the Dutch pension funds achieved a performance of 22%, - 3% and 15% respectively2. It is clear that among the pension funds there is an increasing inclination to invest in commercial securities. The private sector is also making a considerable contribution to shaping share prices on the stock markets. The investment funds benefit from this too. There is also a clear case of a self-hlfilling prophecy, the investors themselves supporting the shaping of share prices and maintaining them. The pension funds want to reduce contributions by means of higher investment yields and if possible also use these to finance the conversion of voluntary early retirement schemes into early retirement pension schemes. All this can lead to dangerous situations. The risks increase as share profits are actually used to finance backservice and profits are thus distributed. Pension funds are increasingly seen as a profit centre by the sponsor. The fund is capable of making a contribution to the profit of the sponsor. Pension costs form 5 to 15% of wage costs. Pressure on the fund to perform is increasing. That has its good and its bad sides. Another trend that is important to avoid is that the beneficiaries become accustomed to the annual index-linking of their pension and come to regard inflation-proofing as a matter of course. In most cases index-linking is conditional. The question is whether the beneficiaries recognise that it is conditional. It becomes an unwritten law. This increases the risks for the funds. ALM studies can demonstrate this. The importance of this is emphasised by the following. n 70% of the supplementary schemes there is a provision for index-linked wage-related ion provisions. This applies to 75% of the members of supplementary pension es. About 95% of these members have a final pay scheme at the level of 70% or Until recently pension schemes were characterised by collectivity and solidarity. This is decreasing. In many schemes collectivity is being decreased in favour of individual options. This too means an increase in the risk which can be managed by ALM. In connection with the calculation of correct performance many funds are switching to valuation at market value instead of at face value. The Anglo-Saxon valuation methods are being used more (SAP24 and FAS87). ALM studies can show what this can mean for the risks run by the pension fund and the sponsor. The altered attitude to risk makes the responsible use of ALM methods essential. This is therefore the trend which we see arising and which we expect to intensify. Fortunately in the Netherlands ALM applications are increasingly being developed by pension funds and sufficient help is being offered to achieve adequate ALM studies. This paper therefore discusses this. 3. ALM in the Netherlands 3.1. Introduction The organisation of a pension fund or insurer can be represented by a triangle: Assets Liabilities The left- and right-hand halves represent the investments andtor the liabilities. Here two different disciplines dominate at different levels. At the base is administration and enrolment, the operational level. Enrolment can be used to make predictions for the medium term, the tactical level. The policy makers who are situated at the top outline the long-term strategic policy. At the operational and tactical level the two disciplines can operate separately. Contact between the two is not necessary for their work. In the diagram this is represented by a wall between the two disciplines. However, at the strategic level this wall can no longer exist. This is the level at which ALM is used and the investments and liabilities are connected to one another. Here there must be open communication. There can be no ALM without focus. Or, as a Chinese proverb teaches us: "If you don't know where you're going, it doesn't matter which direction you go" The first stage in the process of ALM is to determine the aims that are being sought. On the basis of this an ALM study can be created. The main aim is always as follows: A at all times to be able to meet the obligations undertaken In addition, aims such as the following are often formulated: A at cost as low as possible; A at cost as stable as possible . It is clear that the terms yield and risk play an important role here. The term risk, in particular, is interpreted in different ways. Here some choose standard deviation or downside risk, others choose concepts such as the likelihood of loss or the likelihood of ruin (over what period?) The degree of fluctuation in the contributions can also be seen as a risk factor. It is of the greatest importance at the beginning of an ALM study to take ample time to define the risk. This can be different in each individual situation and can to a large degree influence the outcome of the study. The best definition is the one with which the client is most confortable. The client's attitude to risk, the profile of both the sponsor and the fund, its financial position, etc., all play an important role. ALM has only been actively used for about ten years. But even in this short period there has already been clear development. It began with a deterministic approach to both investments and liabilities. On the basis of a limited number of scenarios a number of possible views of the future were examined. In case of a deterministic appraoch we advise the use of an even number of scenarios. ALM is all about a process of learning or becoming aware. We have seen that part of the learning effect threatens to be lost if a uneven number of scenarios is used and that number is usually three. In the case of three scenarios everyone is inclined to leave the two outer ones alone and only to look at the middle one, the "most probable". In the case of an even number of scenarios there are fortunately two middle scenarios, so that there has to be further consideration of at least two scenarios. The purely deterministic models were followed by stochastic models. Initially a stochastic approach was primarily applied to the investments, subsequently it was also applied to the liabilities. Starting from the scenario outlines of many scenarios (e.g. 500 or 1000) an examination is made of which current investment mix displays the most robust behaviour in achieving the aims. The latest development, only used very recently, is the dynamic approach. In this stochastic approach the search is not for a single fixed investment mix which provides the most robust result of all the scenarios, but the model takes into account the fact that in real life managers can adapt the investment mix along the way. The dynamic approach is powerful when decision rules can be formulated and not only with respect of investment strategies, but also for allowing dynamic contribution policies and the problem of raising pensions in line with inflation but with a maximum of the available profit on investment. It is of course not the case that managers sit with their arms crossed if in practice it becomes apparent that an adverse undesired scenario seems to develop. In that case an intervention will be made and the management will adapt their policy to the economic circumstances. In all these cases a dynamic appraoch is very powerfull in modeling the reality. In the case of the dynamic ALM models, decision-making rulesare used to adapt the investment mix continuously to the economic circumstances in the scenario that is being examined. Naturally only the knowledge that is available at the time of the decision is used. The future belongs to the dynamic method. We have already heard many managers say that the current ALM models are not very realistic, because no account is taken of the actions of the management. We think that the models in which these actions can be modelled have a great future. 3.2 ALM is integral risk analysis In essence ALM pictures incoming and outgoing cashflows in the light of many economic and actuarial scenarios, taking into account the dynamic (re)actions of the managers. In order to make an integral risk analysis of both investments and liabilities, models are required. Many models are now available that are marketed by various professional suppliers. In this paper the approaches and products of five professional suppliers are examined more closely. Insofar as we know, nowhere in the world has a day yet been organised in which five professional suppliers have placed themselves in such a vulnerable position. Even in the United States and the United Kingdom, countries where professional ALM has been practised for much longer, it has not yet happened that a number of suppliers have been willing to take the stage one after another before a hall of ALM-experts and potential customers in order to explain their model using one and the same case. How should one compare the results of ALM studies? How should one compare results from different scenarios within a single study? Indices are a useful aid in this. It is of course very important that the indices give an accurate reflection of the results, taking into account the formulated aims.Indices reproduce the expected result in relation to relevant aspects. In addition, it is important to gain insight into the distribution of the results. Sensitivity analysis is possibly the most important element of an ALM study. Too much importance must not be attached to the score in indices in itself. Such an enormous number of assumptions has been made in order to reach a result that the absolute value must be qualified as unreliable. However, what is of great importance is the alteration in the result if an assumption changes. This gives information about how sensitive the pension fund or the insurer is to the various factors. When several indices are used it is usually not immediately clear which result is the best. In one instance one index appears better than in another case and that differs again from index to index. In order to reduce indices to a common denominator one can use multi-criteria analysis. This paper is not an appropriate place to discuss this in detail. But the method must at least be mentioned. It is known of PGGM, the second largest pension fund in the Netherlands, that its ALM model compares results on the basis of multi-criteria analysis. Multi-criteria analysis amounts in brief to the setting of relevant criteria. Then each criterion is weighted. The weighting clearly depends on the point of view from which the results are being seen. Thus a manager may use different weightings from a member or beneficiary, a sponsoring company, the Dutch Chamber of Insurance (Verzekeringskamer), etc. Multi-criteria analysis has methods of reducing the scores for various criteria, taking their weightings into account, reducing them to a single denominator and reaching a single figure. Multi-criteria analysis is a very powerful and practical method which in our opinion receives too little attention within the context of ALM. 4. Description of the case "A Dutch pension fund" A case was created for the purpose of presenting ALM products at the ALM seminar. This case serves as a guideline for an ALM study which is performed using the models of the various suppliers. The aim of the case study is: 1) To gain insight into: A the structure of the model used in ALM products; A the operation of ALM models, including the importance of model assumptions and input; A the connections between the investments and liabilities of pension funds. 2) To compare ALM products and ALM studies on the basis of: A user-friendliness (clarity); A transfer of ALM knowledge; A product characteristics: approach, functionality. 3) To learn to assess the value of ALM studies. In order to achieve these aims a number of simplifications were made to some parts of the case. In this case a fictitious Dutch pension fund was used which had the following characteristics in its initial situation: A aim of the pension fund (general): little likelihood of inadequate cover of liabilities with a low contribution and a stable contribution trend; A structure of the liabilities of the pension fund. The membership consists of three sub-memberships with only male (former) employees: A active members 2000 A paid-up former members 1470 A pensioners 950 The following are the ways in which the members of the three sub-memberships join and leave the fund: A active members death, retirement, joining and leaving A paid-up members death and retirement A pensioners death and retirement For the active members the number of people joining over a period is the same as the number of retirements plus the number of deaths over the same period.The number of people joining is evenly distributed over the ages 25 to 34. The likelihood of leaving for active members is 0.9 to the power x, where x stands for the age of the member.