Mr P. G. Scott, F.I.A. (President): I should like to extend a warm welcome to a number of guests this evening: Barthold Kuipers, who is the principal expert – pensions, European Insurance and Occupational Pensions Authority (EIOPA); Alan Napier, Department for Work and Pensions; John McCallion, Director of Strategy, Private Pensions, at the Department for Work and Pensions; Michael O'Higgins, Chairman of the Pensions Regulator; Bill Galvin, Chief Executive, the Pensions Regulator; Nigel Peaple, Head of Corporate and International Affairs, the Pensions Regulator; Catherine Cunningham, the Pensions Regulator; Brendan Kennedy, Chief Executive, the Pensions Board of Ireland and Chairman of the EIOPA Occupational Pensions Committee; and David Cule, from Punter Southall, representing the ICAEW Pensions Committee. I should also like to extend a very warm welcome to Falco Valkenburg, Chairman of the Pensions Committee of the Groupe Consultatif Actuariel Européan, who will be chairing the discussion of part of tonight's meeting. We also have Huw Evans with us on the top table. Huw was the Chair of the Review Panel that oversaw the project from conception to finish.

Most of you will know that the reliance on employer covenant is a key feature that distinguishes the way in which defined benefit schemes are financed in the UK from the way that such schemes are financed in much of continental Europe. This feature has been part of the UK pensions environment since the dawn of defined benefit pensions, but it is really only since the publication of the Pensions Regulator's Code of Practice on Funding Defined Benefit Schemes that it has become standard practice to take explicit account of employer covenant when deciding how a scheme should be financed. But, there remains no standard way of achieving this end. The Institute and Faculty felt that this was a topic of significant importance to the profession and of especial interest to those working in the pensions area and so, in May 2012, the profession issued a call for a research project on the topic of how employer covenants are best taken into account when assessing the capital supporting pension schemes. The impetus for this project was provided by the review of the European Directive on Institutions of Occupational Retirement Provision (IORPs) and in particular by the Holistic Balance Sheet (HBS) proposal. It would be natural for much of tonight's discussion to be framed in this context, but I do hope that the focus will be on how we can make our methods more rigorous rather than on whether we agree or not with the HBS proposals. The response to the research call was so strong that projects were awarded to two respondents, and we are here tonight to hear presentations from Craig Turnbull, for Barrie and Hibbert, and Richard Setchim, Adam Sutton and Hash Piperdy, for PriceWaterhouseCoopers. Each will give their presentation, and after each presentation you will have an opportunity to ask questions on that particular presentation.

It is now with great pleasure that I invite Falco Valkenburg to open the discussion section of this meeting.

Mr F. R. Valkenburg (Chair, Groupe Consultatif): The vision of the British Actuarial Profession is “We will serve the public interest by ensuring that where there is uncertainty of future financial outcomes, actuaries are trusted and sought after for their varied analysis and authority”.

I think we should keep in mind the objective of “serving public interest” which is easily forgotten when we go into difficult mathematics. In my view, serving the public interest is also the purpose of the holistic balance sheet. It should get all the building blocks of the financial position of the pension fund right, be transparent and allow communication about the position.

In my view, it is to be hoped that the emphasis is not about asking for more money in pension funds: that is just not possible from the figures of which I am aware. It should be more about communication between stakeholders. I should like to hear your perspective on that during the discussion.

Mr C. J. Turnbull, F.I.A.: I have tried to condense my rather long paper in order to try to present some of the key findings to you this evening. I think we will all be aware of what makes this such an interesting topic Clearly, in the UK, we live in interesting economic times where our defined benefit pension fund deficits are at very challenging and high levels, exacerbated by the current exceptionally low interest rate environment. That naturally makes the idea of sponsored commitment to improving the funding position of pension funds and sponsor covenants particularly important for pension fund members and the security of the promised benefits.

You can put that status of defined benefit pension funding alongside some of the developments that we have seen elsewhere in actuarial practice, particularly in the last 10 years. On the insurance side we have seen something of a revolution in terms of how risk, capital and values are calculated and, in particular, the application of quantitative tools and market-based, risk-based methods. We can see those two strands coming together somewhat in some of the ideas behind the holistic balance sheet proposals and the Quantitative Impact Studies (QIS) that have been released by EIOPA.

In the paper I tried to address three broad questions. Firstly, if we think about the idea of a market-consistent valuation for the assets and liabilities of a defined benefit fund, how do we go about calculating those valuations and, in particular, how do we approach the market consistent-valuation of this sponsor covenant asset? Second, assuming that those first calculations are somewhat complex, are there some pragmatic shortcuts that we can take to approximating those full valuations for thousands of pension funds? Third, once we have this idea of an holistic or market-consistent balance sheet, can we use it as a risk-based capital measure in an analogous way to the way in which the global insurance sector have used market-consistent balance sheets, and obtain an idea of value at risk based on that balance sheet to assess prudential capital and funding adequacy for promised benefits?

The idea of the holistic balance sheet, in its simplest form, is trying to capture the fact that over and above the current funding of the assets of the pension fund there is this additional layer of value which can be considered as an asset in the pension fund balance sheet which reflects the sponsor's ability and commitment to make good the current deficit in future years to come.

If we think about each of the items in the balance sheet and the placing of a market value on those items, in turn, first, for the asset portfolio. There is nothing new there in terms of putting a market value on our reasonably liquid set of assets and this not something that we address in the paper. The market-consistent valuation of the liability side of the balance sheet is an interesting topic. In principle, putting a market-based value on a set of fixed cash flows, or even inflation-linked cash flows, is theoretically a straightforward process. But one of the things that we have learnt in Solvency II over the last five or so years is that there is a lot of complexity involved in valuing very illiquid, very long term, fixed cash flows. So, particularly thinking about topics like illiquidity premiums and yield curve extrapolation there are a number of interesting questions, even after you have resolved that a guaranteed cash flow should be valued using a risk-free interest rate. Again, that is not something we have really addressed in the paper but I think it is a natural research strand that overlaps with the interests of insurance actuaries today.

Finally, the main focus of the paper is the third topic which is placing a market consistent value on the sponsor covenant. In the paper we define this idea of the covenant as the sponsor's commitment to making future deficit funding contributions that are required over the lifetime of the pension liabilities. The complexity in the valuation of the covenant primarily arises from the credit risk of the sponsor, and the possibility that the sponsor may essentially go bankrupt prior to funding the deficit in the pension fund. Placing a market value on that risk can be a fairly complex problem in general because the contribution commitment can be very long-term: that is, the sponsor is not committing to making good the deficit tomorrow, he is proposing to do so, potentially, over coming decades. That commitment is credit-risky. The nature of the commitment may also be quite dynamic and path dependent, so the deficit contributions in 2018 will be higher if the deficit is greater in that year than otherwise. The nature of the sponsor credit risk will often be correlated with the size of the deficit, so the same economic exposure that can drive poor asset performance in the pension funds can also increase the probability of default on the sponsor, and observing a market based or market implied cost for bearing that sponsor risk is something that might not be directly observable in financial markets. For those reasons, and others, creating this market-based value valuation for the sponsor covenant is, in general, not a straightforward problem. We describe in the paper how we can go about thinking about that problem and developing a framework to answer that question.

In order to value the sponsor covenant we have to make assumptions about the behaviour of the additional deficit funding sponsor contributions, in particular the timing and the nature of them in terms of how they are calculated and when they are paid into the pension fund. The timing is absolutely crucial because the longer the funding of the deficit is deferred, in theory, the greater the exposure to sponsor credit risk. That is something we will see in the examples: that timing is critical to the valuation of the sponsor covenant, particularly for very credit-risky sponsors. So, we need some sort of algorithm that describes how that contribution will be set and how it will behave under different circumstances in the future as the pension fund runs off. Once we have that description of what the cash flow scheme looks like, we then can go about creating the market-consistent valuation. The key thing in a market-consistent valuation is understanding when the sponsor would not be able to make good on the commitment and what happens in that event. That can be categorised, or represented, by the questions that we have to answer when thinking about this valuation.

First, we have to determine in what circumstances the sponsor will be assumed to be unable to make good on the commitment. In our example, we assume that this is simply equal to the event of the corporate sponsor defaulting so then, and only then, do we assume the sponsor does not make good on their commitment. Then we have to assume what the size of the pension fund deficit is when that sponsor default occurs. In our approach, we model that stochastically. Finally, we have to make an assumption about what happens when default does occur, and how much can be recovered, if anything, from the sponsor in that circumstance?

In our work we make some assumptions, one of the effects of which is that we treat the pension fund as an unsecured creditor. But that is something which can be generalised. Once we have made all those assumptions, we then have a framework to value the cash flows implied by that model. The key point here is the valuation exercise is intended to produce something that we think of as market-consistent. The main factor that we use to establish that market-consistent basis is corporate bond prices of the sponsor. What we are trying to assess here is the market-based estimate of the cost of being exposed to the default risk the pension fund bears currently at any point in time.

An important point to note there is that, because we are interested in a market-consistent valuation, in our modelling we do not need to make any assumptions about our own best estimates of default probabilities of the sponsor. All we need to observe is what the market charges, if you like, for being exposed to that risk. That is a general result that we see again in insurance when it comes to creating market consistent balance sheets there. Because we are interested in a market-based estimate, our own estimates of some of these probabilities are not important. What is important is that we can observe what the market is charging for the exposures with which we are trying to be consistent.

In order to catch all the variables, we have developed a stochastic modelling approach which can allow us to project forward a number of economic variables, such as interest rates and equity markets, in order to capture how the pension fund's assets and liabilities will behave over time, what that means for the contributions stream that is produced and for sponsor default behaviour. Once we have developed and calibrated that model to market prices, we can then essentially carry out the sponsor covenant market-consistent valuation. Figure 1 shows you an example that we developed in the paper.

Figure 1

Case study

We start with a set of liability-related promised cash flows that you can see on the left-hand chart on the screen which will roll off over the next 60 years. When those are valued using a government bond yield at the end of 2011 their present values sum to exactly £1 bn. We then assume that the asset portfolio of the scheme today is currently £800 million. We consider two different contribution strategies for funding the deficit. The first one goes to the theoretical limit of how long we can defer making contributions by saying we will only make contributions into the scheme once the asset portfolio has been exhausted by the liability cash flow outflows. The second one which looks a reasonably extreme position in the other direction. It says we will make an annual contribution that is equal to the market-consistent deficit of the scheme divided by five.

So we can then to run our model and understand what these values look like. We can start with the very simple case where we assume the sponsor is risk-free. In this case if we have a risk-free sponsor we generally expect the value, on a market-consistent basis, of these contributions should be whatever the deficit between assets and liabilities is today. In this example it would be 200 million. That is what we find here when we adopt the first contribution strategy where we defer our contributions for as long as possible with a risk-free asset portfolio. Interestingly, as we increase asset risk, moving to a very risky portfolio, even when it is 50-50 risk-free and risky, we find that the covenant valuation increases; and similarly with the second contribution strategy we see that result again to an even greater extent. That might look a puzzling result, producing covenant valuations well in excess of the size of the deficit today. This is basically reflecting an optionality that is in these investment and contribution strategies, and the possibility of creating some terminal surplus assets in the scheme. Under these asset strategies, we might find we have made more contributions than we needed to. At the end of the day we could have a surplus pool of assets. Depending on where we put those assets, that will either increase the covenant valuation or not. So if we assume, for example, that the sponsor could recover those surplus assets back in the event that they rise, that reduces these valuations. In all these cases, we get a £200 million result. I have not done that in this analysis but have assumed that those are orphan assets and so that creates an option or a surplus valuation.

Now we can consider the more interesting case, where we turn on the sponsor default risk. In general what we find is the sponsor covenant valuation has three components. The current size of the deficit on a market-consistent basis plus the terminal surplus asset value, that I have just described, less the sponsor default risk that we are valuing using the simulation model that captures the market-based default risk and its correlation with the assets of the portfolio. We can then use that to produce a number of different results for the covenant valuation.

A key thing to bring out in figure 2 is, firstly, as you would expect, as we reduce the sponsor's credit rating, we get quite a significant drop in the value of the covenant. Because we are assuming the sum recovery can be made even in the case of an immediate default, we find that the covenant valuation is never zero in these examples but falls to 35% in this case of the current deficit.

Figure 2

Case study with results

As we move from contribution strategy 1, to strategy 2, we are basically assuming that much more of the deficit is funded much sooner, that generally increases the value of the covenant because it reduces the exposure to sponsor default risk.

Finally, I should like to touch on the idea of using a market-consistent balance sheet as a risk-based solvency measure. Again, certainly in the insurance context, and similarly in the EIOPA proposals, the market-consistent balance sheet, or the holistic balance sheet, is really only a means to an end. That end is ultimately understanding the security of benefits by using a market-consistent balance sheet to assess capital adequacy. In that context, the capital adequacy is measured using the idea of a one-year value at risk which is looking at how the net assets of a market-consistent balance sheet can behave over a one year horizon. In particular, considering a 99.5th percentile tail of that distribution of net assets and what level of capital would be needed to absorb that loss.

If we take that as an example, we can go about implementing, in a fairly simple way, a one-year value risk. As a side note, there are all sorts of much more sophisticated ways of implementing value at risk in a rigorous way which are being exposed in insurance. For this example, we simply do a series of stress tests of the balance sheet for each of the key risks that drive the balance sheet, and understand the losses created in the balance sheet at that 99.5th percentile level and then basically sum them using a correlation matrix. We make a number of different assumptions about what those stress tests look like for the key four different financial market risks that drive our example balance sheets. The details of this approach are in the paper. When we do that, we can then establish the capital levels that are created by each of these risks for different examples. In this case we are looking at A rated and DD rated sponsors. Finally, we can create a total balance sheet that includes the solvency capital requirement.

If we run through the numbers in figure 3, we are looking at three different sponsors here, risk-free, A rated and BB rated sponsors. In all cases, the starting asset portfolio is £800 million and the promised liability is equal to £1 bn. In the previous results on the valuation, we found we obtained the results you can see in figure 3 for the value of the sponsor covenant. This is under contribution strategy 2, and for a risk-free sponsor it is £345 million. That falls significantly as we move the sponsor down the credit spectrum. That knocks on to the current value of the assets in the balance sheet. Hence, starting net assets in the balance sheet for a risk-free sponsor, amount to £145 million. The A rated sponsor is £7 million and minus £45 million in the BB case.

Figure 3

Case study: balance sheet summaries

By stressing these net asset values, and understanding how each of these items in the balance sheet changes on each of the stress tests, and then summing those requirements, we then obtain the solvency capital requirement number. For the risk-free sponsor, this number is zero because basically any changes in the assets or liabilities under stress are offset by an exact offsetting impact of the sponsor covenant. As we move down the credit spectrum, the ability of the sponsor covenant to act as a loss absorber or shock absorber under stress is diminished and that ultimately produces higher and higher solvency capital requirements.

This allows you to produce a capital requirement for the pension fund that incorporates the sponsor's ongoing commitments and reflects the credit risk of the sponsor in doing so. In the cases with credit risk in this stylised example the solvency capital requirement is 12% to 18% of the market-consistent liability value so it is a fairly significant margin.

There are a number of interesting ways in which this analysis can be extended. As I mentioned, producing techniques to approximate the market-consistent valuation so we do not have to use a full blown Monte-Carlo simulation technique is a natural area, and one on which we briefly touch in the paper. The analysis here has not considered the role that pension protection plans play on the balance sheet and its behaviour. That is an interesting, but potentially complex, area into which this research could be extended. As I mentioned, even outwith the covenant, and just looking at the actual valuation of the core promised cash flows on the liability side is an area which is causing a huge amount of discussion and research on the insurance side and this clearly has significant implications for the numbers produced in the type of analysis shown here.

The Chairman: Are there any points of clarification needed in respect of what Mr Turnbull has presented?

Mr J. R. Manion, F.I.A.: I have a question relating to your example. When we have an insurer with a large diversified portfolio, the portfolio loss given default is not necessarily a jump. But when you are dealing with a corporate sponsor, let us say, rather than a financial institution, you can have a jump to default. When you take your BB case, or a lower grade case, you have had a probability distribution to look at the solvency margin, but then you have taken at the top end on an assumption of, say, a 3% default. But the reality is that the sponsor will jump to default. So you do not have this probability distribution. The 99.5% outcome is default and you need to reflect that, do you not?

Mr Turnbull: You are absolutely right. There are two levels to this in terms of modelling response to default. One is the market-consistent valuation itself will have a model which gives the probability of defaulting in each year and may be correlated to equity markets, and so on. Those default probabilities will actually be derived for market prices and will be risk-neutral. But we will be capturing that jump in the sense that it will either have defaulted or not. We will model it using our ratings-based approach so more often than not it will move ratings and then default, it will not normally jump to default immediately. There will be a rating transition matrix that is derived from corporate bond spreads.

Solvency was the aspect you mentioned. We have to make an estimate of the 99.5^th percentile one year change in the rating. We can use a weighting matrix to make an estimate of that. As an example for a sponsor which is A-rated a 99.5^th percentile change in the rating over one year will be down to BB. So for the A-rated sponsor we are assuming that it is a BB sponsor at the end of the year. Whereas for a BB sponsor today, the 99.5th level would be to default. In that case we are actually assuming that they do default at the end of the year. So solvency capital has to reflect that.

Mr Manion: If you look at the BB-rated sponsor, you have £800 million assets. The covenant there, £155 million in the scenario, could jump to £0 in default. Let us say that there is no recovery.

Mr Turnbull: Unless we have a 35% recovery rate.

Mr Manion: They would have £800 million assets and a defaulting employer. The £155 million and £175 million on a 99.5 percentile would fall to, say, £100 million. They would be £100 million short. So the members of that pension fund would be not protected by that solvency capital requirement in a 99.5% scenario?

Mr Turnbull: But this is saying that the fund is not adequately funded today. So that £175 million is not in the fund today. What this is saying is that the fund needs total assets of £1175 million. It has £955 million. So this analysis is saying that the fund has to go and get £120 million of additional assets in order to be funded to that 99.5% value at risk level.

Mr P. V. Kelly, F.F.A.: In your paper you say you have not dealt with longevity and inflation in the Solvency Capital Requirement (SCR) calculation. But you note that it is likely to have a significant impact. Do you have any rough idea of this, given that those are in the QIS?

Mr Turnbull: That is a very good question. I have not done any explicit quantitative numbers on that element. Given that those material risks are not in the example, we can expect there to be another margin to include all of those risks. There will be lots of interesting real-life features to incorporate into these sorts of calculations.

Mr G. D. Clay, F.I.A.: Do you consider the relative strengths of the sponsor covenant and the covenant on the bond issued by the sponsor?

Mr Turnbull: I have made the simplifying assumption here that essentially the covenants are comparable in the sense that I have assumed a loss default for the pension fund to the sponsor covenant which is similar to that on an unsecured corporate bond.

Again, where scheme specifics dictate that there is clearly a different type of hierarchy then there would be nothing stopping you making different assumptions.

Mr J. G. Dewey, F.I.A.: On that specific point of default probabilities, just to check my understanding, are you using the actual credit spread on the corporate bond of the sponsor and assuming that all of that premium is in respect of default probability, or are you using the credit rating which you subsequently referred to, and using a transition matrix based on that credit rating, or is it a combination of both?

Mr Turnbull: In the valuation we are using probabilities that are derived from credit spreads. Those credit spreads could be for a liquid sponsor, if you like, a FTSE 100 company. Those spreads would be directly observed from the sponsor's corporate bonds.

For smaller businesses we could simply allocate a rating to them and then look at what our ratings curve looks like and derive default probabilities from there. In the valuation we are using a ratings-based approach but it is calibrated to those credit spreads. We are not using any real world transition matrix. The only place where we do use a real world matrix is in the solvency piece where we are identifying the size of the credit downgrade stress.

Mr A. O'Mahony: I spent a lot of time in the credit rating business. I know a lot of the data is actually calibrated from US ratings which have the greatest universe of data. The European data, and especially the UK data, is very limited. Is there a danger of contagion, in that if we do calibrate to credit ratings, we are getting a US probability of default and US transition data imposed on a UK system?

Mr Turnbull: In the valuation work we were trying to directly use the UK relevant bond prices, so it is not reliant on weightings to the extent that if we are doing this for, say, Marks & Spencer, we look at what the Marks & Spencer credit spreads. In that sense, it is not directly a function of how the rating agencies have treated UK sponsors and how it differs from the US.

Mr O'Mahony: I think that is a good answer. However, I would point out that, in general, it is not listed companies that are the sponsors of the pension schemes, it is mainly subsidiaries that are the sponsors.

Mr Turnbull: Absolutely. That is not a point I have yet addressed.

Mr C. Keating (Non-member, Brighton Rock): Could you just clarify one point? You are looking at credit spreads. Are you using the naked spread or are you decomposing the spread into liquidity default and default volatility terms?

Mr Turnbull: For the market-consistent valuation it is the former. It is the naked spread because it is intended to be a market-consistent valuation. So all we care about is what is the market price of that debt.

Mr Keating: Then you are pricing all sorts of wrong things in there.

Mr Turnbull: We are not taking a view on whether the market price is wrong. We are taking a view on what is consistent.

Mr Keating: I do not see why you are not following standard practice, as, for example, regularly illustrated by the Bank of England on decompositions.

Mr Turnbull: To create a value which aims to be consistent with that market value, we do not have that decomposition.

Mr Keating: The naked spread has a very small amount to do with default. You are calibrating this to something which has two other terms which are more important than default terms. You are massively exaggerating the likelihood of default.

Mr Turnbull: I am not trying to estimate the probability of default. I am trying to estimate the market-based cost of there being an exposure to default.

Mr Keating: The problem is that the market-based measure is actually exposure to rather a lot more than just default. So, you are not being market-consistent in estimating market default risk.

Mr Turnbull: Then that is something we can discuss.

The Chairman: There will be much more time later for further discussion on this difficult matter.

Mr R. Setchim (Non-member, PwC): I lead the PwC team here. I have been a PwC partner for the last 25 years, a chartered accountant and an insolvency practitioner. I could perhaps call myself a company doctor. I was a trustee of the PwC staff scheme for 10 years and I enjoyed the interaction with members of your profession so much that my client work now is full-time pensions advisory, specifically helping trustees understand their sponsors’ covenants. PwC has a team of 50 people nationally who do that over a broad range of market sectors. My own client work includes financial institutions, a couple of clearing banks and a life insurer, so the world of financial regulation and Solvency II stress testing is a familiar one to me.

I am joined on the platform by Adam Sutton and also Hash Piperdy. They will introduce themselves more fully before they speak. I have another specialist valuations colleague sitting in the front row, Charles Sword, who may join us for questions.

My task is to set our work in context. When the EIOPA consultation hit my desk in October 2011 it was of particular interest to me as a covenant assessor because of the holistic balance sheet. You need to understand what the holistic balance sheet is about, and what it is trying to achieve. Figure 4 sets out how it is described.

Figure 4

Holistic balance sheet (HBS)

The key points are quantifying the obligations and the resources available. We have reproduced a schematic of the holistic balance sheet in our paper. You will find it as figure 1 in the PwC part of the paper.

What I can say, as an accountant, is that it is not like any other balance sheet. EIOPA describes it as follows: an assessment tool rather than the usual balance sheet based on generally agreed accounting standards.

It is interesting in that it includes all the economic exposures to which the Institutions for Occupational Retirement Provision (IORP) is exposed, whether they would be on a conventional balance sheet or not. The driving principle is that the balance sheet views security from the perspective of the member rather than the financing vehicle, and consequently the asset side should capture all the resources that can be used to meet the pension promise. These resources, of course, include sponsor support. If the balance sheet is ever to balance, the method for valuing the sponsor support has to allow sufficient cover for the obligations, including those risk reserves on the obligations side. That is a very important point. In EIOPA's own words, this is a major difference from the Solvency II directive.

Despite this major difference from Solvency II, in terms of valuing those resources, EIOPA noted that they want the values to be on market-consistent, sound economic principles, that is, made according to fair value.

Back in October 2011, I wondered how on earth that was going to be achieved. When the draft QIS was produced we worked on holistic balance sheets for 20 clients. It quickly became apparent that the methodology derived a value for sponsor covenants which did not reflect the commercial reality that we see every day. That is what led me to engage with specialist colleagues who spend their working lives valuing businesses. It is these practitioner insights that we bring as our contribution to the debate. As some will know, the debate remains live because EIOPA has acknowledged that this area of the advice to the Commission needs more work, and a working party has been set up to make further recommendations.

Having outlined the context of our challenge, what we are now going to do is as follows. Mr Piperdy is going to demonstrate how the QIS values “sponsor support”. He is going to do it by looking closely at the example in our paper. Mr Sutton will then step into the world of business valuations. Having introduced methods which, we hope, will be familiar to many, he is going to look at the same example that Mr Piperdy first looked at through a valuer's lens, and is then going to go on to describe the methodology that we have put forward. We believe the methodology is flexible enough to allow for sponsor-specific valuations on consistent principles which are well-understood in the market and one which views those assets from the perspective of the member while having transparency, consistency and verifiability.

Mr H. E. Piperdy, F.I.A.: I am an actuary in PwC's pensions practice and also represent PwC on the Pensions Regulators QIS Expert Group. I feel intimately acquainted with the EIOPA specifications, having carried out holistic balance sheets calculations for the 20 or so examples at which we have looked. What I am going to do over the next few minutes is give an overview of the formulae and the methodology in the technical specification for the QIS, and bring that to life with the example that is in section 4.3 of our paper. That will show what the numbers look like and I will share some observations with you.

It was always was going to be a challenge to come up with a mathematical formula that values 6000 or so sponsors. EIOPA introduced two concepts. There is “maximum sponsor support” and “market value of Sponsor Support.” I will explain the principles behind the formulae. In principle, the “market value of Sponsor Support” is based on the deficit in the scheme today. So, if you look at the technical provisions measured on a risk-free basis, less the financial assets, and project that into the future over the next 15 or 20 years, and allow for the probability of the employer defaulting on those contributions, introducing an element of credit risk, then there is a terminal value. If the employer does default, how much of that deficit would be recovered? These calculations are focused purely on the funding position of the pension scheme and how it varies over time and how much you need to plug that deficit.

The focus of our research is the “maximum value” which looks at the wider resources of the business backing this particular pension scheme. We are not just looking at the funding position, but thinking in terms of the holistic balance sheet when you are considering the uncertainty around the solvency capital requirement to determine what are the wider resources of the business that are available now and in the future? That is conceptually made up of two components: the future wealth and the current wealth.

Figure 5

Maximum value of Sponsor Support per EIOPA definition

(Source: Quantitative Impact Study (QIS) on Institutions for Occupational Retirement Provision (IORPs), Technical Specifications , European Insurance and Occupational Pensions Authority, dated 8 October 2012)

How does EIOPA define those terms? Starting on the right-hand side of the slide on the screen, current wealth is based on 50% of the shareholder funds added to the pension liability recorded in the sponsor's balance sheet. Note that by “liability” we have interpreted this to be the deficit under IAS 19. Then, looking forward, in terms of future wealth we have two components. We have the recovery plan contributions under the schedule of contributions at the moment. You put a present value on those. On top of this element, the key driver is projected future cash flows or future earnings, and you can see 33% of the projected cash flows of the business projected over the average duration of the scheme.

I want to share three key observations on this formula. Firstly, there is, potentially, an element of double counting. When doing a discounted cash flow valuation for business valuation purposes, typically you would only consider the future earnings generated by the assets of the business. By adding back the assets that are used in generating those earnings there is an element of double counting.

The second observation is typical of the 20 or so examples at which we have looked: the term “cash flow” is not actually defined. If you pick up a set of company accounts there are lots of different lines giving different values for the cash flows. Which one do you take? You could take operating cash flows, financing, and so on, and, clearly, the maximum sponsor support value is going to be very sensitive to which ones you use.

The third observation is around the discount rate. In a business valuation, it probably does not really make sense to use the same discount rate that you use to discount pension liabilities.

If we consider the example in section 4.3 of our paper, the situation is you have a single company sponsoring the scheme with assets of £2.5 billion and technical provisions of £4 billion. So immediately you have a deficit of £1.5 billion to try to cover with the sponsor support (before one considers any additional solvency capital requirement). Applying the methodology, the current wealth of the sponsor is £400 million from the shareholder funds and another £400 million you add back for the pension liability on the balance sheet. So that is £800 million. In terms of future earnings, the value of the recovery plan contributions is just under £600 million.

The key driver of value for this particular example is one third of future projected cash flows. We have interpreted cash flows as equity cash flows, given the way the formula is constructed, and that gives you only around £300 million.

Putting that together, we get a maximum sponsor support value of around £1.7 billion. Remember, you have £1.5 billion in terms of deficit. That was before you even consider the solvency capital requirement to go on top.

To give this some context, figure 6 shows the holistic balance sheet for this particular example?

Figure 6

Sponsor Support value in the HBS

You can clearly see that having a maximum sponsor support value of only £1.7 billion actually leaves a £1 billion gap in the holistic balance sheet and a pretty unhappy client.

Mr A. Sutton (Non-member, PwC): Mr Piperdy has left me, an accountant, with a balance sheet that does not balance, a quite horrifying concept.

My background is that I have been a business valuer now for around 12 years. I have valued over 500 companies in that time for a variety of purposes including disputes, acquisitions, disposals and financial reporting. Most recently, I have been working a lot with the trustees of the Nortel pension scheme, doing a lot of valuation work.

Initially, I will discuss the basics of some of these business valuation techniques.

On Figure 7 there are three approaches: the income approach, or discounted cash flow (DCF); the market approach, which is typically measured using either listed companies or transactions as a guide; and, on the far right, cost approaches. The most commonly used valuation techniques in business valuation are the first two, DCF and market approach. The cost approach is sometimes a relevant approach to use in a distress scenario where you might be looking at a liquidation valuation.

Figure 7

Traditional business and asset valuation approaches

Taking the income approach first, there are two key inputs. One is the cash flow forecast or business plan for the sponsor. The other is the discount rate. There are, potentially, quite a significant number of variables and considerations that need to go into those cash flows. I have listed just a few: macro economic; market and industry positioning; and the specific market positioning and competitive positioning of the sponsor.

The other input is one with which you are all very familiar – the discount rate. We would typically use a capital asset pricing model to come up with a weighted average cost of capital for the business or the sponsor. That allows the calculation of the DCF.

Turning to the market approach, there are two options. I can try to find listed companies which are similar to my sponsor. Alternatively I can go and try to find some transactions where people have bought and sold businesses, to give me a guide as to what values have been paid for businesses which are similar to my sponsor. In terms of the actual calculations, it is pretty simple to apply a market approach. The tricky bit is in first of all finding the right comparables and the analysis that is used to assess their comparability to your sponsor. Should you, for example, be adjusting the multiple derived from those comparisons when you get your valuation as sponsor?

So, Mr Piperdy left us with a problem: a balance sheet that does not balance. I am going to take you through some simple numbers, using both a market approach and a DCF or income approach, to explain how I view this valuation question.

Starting with the market approach shown on the left of Figure 8. I managed to find some transactions and some listed companies. I have looked at business value as a measure over EBITDA, which is earnings before interest, tax, depreciation and amortisation, which is a very commonly used measure of earnings in calculating a multiple. I have a number of eight to nine times EBITDA for my comparable transaction. On the listed companies, they were showing seven to eight times EBITDA, which resulted in the valuation ranges for the business that you can see underneath. So, an overall range of £3.7 to £4.8 billion in this case. I applied my judgement as a valuer. I could choose one end or other of that range. I could plump for something in the middle. I could decide that I am more happy with one approach and the quality of the data and analysis that is coming from transactions versus listed companies or vice-versa. In this simple example, I have taken a number that is broadly in the middle. So £4.2 billion is my business valuation conclusion. I have taken off the net debt, which does not include the pension liability. That leaves me with a value of sponsor support of £4 billion.

Figure 8

Business valuation methods applied to calculate the maximum value of Sponsor Support

I have conducted a similar analysis using my DCF approach on the right-hand side. So I get down to a measure of free cash flows, which are the cash flows to all of the providers of capital to the business. I have not subtracted any debt-related cash flows at that stage. I have a forecast period of three-years, and then I have applied a terminal value.

I have made an assumption about future long-term growth for this business. I have calculated the present value of those cash flows using the weighted average cost of capital that I have calculated for this business. That has given me a business valuation conclusion of £4.5 billion. I take off the same net debt number. That results in a value of sponsor support of £4.3 billion.

So my valuation conclusion is a range of £4 billion to £4.3 billion which you will note is significantly higher than Mr Piperdy's answer of £1.7 billion. Interestingly, the market cap of this particular business was around the £4 billion mark.

Figure 9 illustrates my position, in contrast to Figure 6.

Figure 9

Sponsor Support per EIOPA compared to Sponsor Support using Business valuation

That brings me on to the approach that we feel is a sensible and reasonable approach to apply. Essentially, this is a staged approach. The reason that we thought it should be a staged or phased approach is because we realise that there are probably a lot of companies out there that can fairly easily support their pension schemes. Therefore we wanted to give them the option of doing something relatively high level, a little bit quick and dirty, in order to tick the box while recognising that there are companies who similarly like to be very close to the line or even below the line in terms of filling that gap and therefore need a much more robust and rigorous approach to valuation.

So stage one is to value the sponsor support using market multiples. That will give me a value of sponsor support and I then say is that a certain multiple of the holistic balance sheet deficit? I have put two times, but the quantum is clearly a point for discussion. What is sufficient? What is a level above which you would not need to do any more work, and with which you would be satisfied? Clearly, that will depend on the type of industry, the volatility of the earnings of companies within that industry, and also the particular sponsor.

If sponsor support is clearly sufficient then that is the end of the process. If it is not clear that sponsor support is sufficient, then we go to stage two, which is a more rigorous application of the valuation methods that I have just described. We would look at both a refinement of the market multiples approach and a DCF analysis for that business. Using that analysis, we would apply our judgement to reach what we consider to be a reasonable maximum value of sponsor support. If the business is in distress it may be relevant to apply different types of valuation methodology. This is covered in our paper in the piece on the demise curve. If you are a highly distressed business, the maximum value of your sponsor support may be related to what you can realise for those assets in an insolvency.

To summarise our approach and our method, it is grounded in business valuation theory so will be recognisable to anyone who has done a valuation in an accounting or deals context. For example, in the accounting world, you have annual impairment tests. Companies are required, essentially, to carry out a business valuation to test the carrying value of the assets on the balance sheet. Our approach, we believe, allows more flexibility and the greater application of judgement which is more suited to the variety that you see in the UK pensions landscape. It is proportionate, meaning that for some businesses the approach will be relatively quick and simple to apply, while for others there will need to be a robust process to determine whether or not headroom exists and, if so, how much. Lastly, it is consistent with approaches that we currently see in the market in terms of assessing the strength of the covenant and the sponsor.

Mr J. G. Spain, F.I.A.: On Figure 6 in Mr Piperdy's presentation, he had assets of £2.5 billion and liabilities of £4 billion, leaving £1.5 billion deficit. Tell me, please, how do you get to £400 million deficit on the balance sheet under something like IAS 19? I know that the numbers might be massaged a bit, but not that much, surely.

Mr Piperdy: The £400 million deficit was taken from the accounts. It is not the liability: it is the net deficit.

Mr Spain: The net deficit is £1.5 billion not £400 million. Four, minus two and a half is one and a half, not 0.4.

The following was clarified in a subsequent email exchange between Mr Piperdy and Mr Spain: Mr Piperdy explained that the large change in basis (i.e. accounting liabilities of £2.9bn switching to £4bn on a risk free measure) is due to the fact that the accounting basis is significantly weaker than the basis used for the holistic balance sheet which uses a negative net discount rate. Mr Spain shared his own independent calculations with Mr Piperdy and acknowledged that the figures presented by PwC were reasonable.

A speaker: Question one is on the discounted cash flow. I notice the terminal value is a massive component of the present value. What kind of assumptions do you use to come up with the terminal value?

Question two is around the approaches taken to come up with a value for the business, if you are using a discounted cash flow and a multiples approach, should you not be able to reconcile the two approaches? That is, should you not be able to back out market implied assumptions from using multiples by essentially using the same model for discounted cash flow?

Mr Sutton: To answer your first question, we would either use a multiple to calculate the terminal value or, more commonly, a Gordon Growth model, to look at what we consider to be a reasonable terminal year cash flow. That might involve projecting out the business plan for the sponsor further than the example that I have shown there, and then considering what an applicable long-term growth rate might be, taking into account that industry and the macro economic context in which that business is operating.

You are right, the market and DCF approaches should be consistent. They are not. You do not always find inconsistency between the two approaches. There are some reasons and some circumstances in which they would not be consistent. That is where the application of judgement comes in.

Mr T. W. Keogh, F.I.A.: I should say that I work for the Pensions Regulator, but this is a personal question.

It seems that when trying to reconcile the £1.7 billion number, which is clearly a strange EIOPA number, and your £4 billion mark to market, one of the things to consider is that you would surely want to take a haircut to the current market implied value of the sponsor, or increase your solvency capital to allow for the fact that there is uncertainty in that number. I wondered whether you had looked at either of those things.

Mr Sutton: I think good valuation practice would involve looking at the sensitivities to your numbers. They could be upward or downward sensitivities. The valuation that we calculated is an expected value rather than, if you like, a prudent value, although there are elements of prudence in it in the sense that the only debt added back is the pension debt. So the scheme would rank equally with some other creditors within the business. Arguably, therefore, the number might be a little bit bigger.

Mr Keogh: Where assets are held in the pension scheme, you allow for the uncertainty around their value on the liabilities side of the balance sheet, through the thing called SCR, so would it be consistent to have some sort of equivalent allowance for uncertainty in the covenant? Whether you would actually have a lower covenant number or a higher SCR does not seem to matter, but do you want to do one or the other?

Mr Setchim: I will give you the non-technical answer. We thought the liability side was really stacked quite high enough. The deficit, the £1.5 billion, is calculated on a risk-free rate. So, although they are called technical provisions, it looks awfully like the buy-out level of funding in the current world. Then you add on what I call the Rumsfeld layers for the unknown unknowns. Then the challenge is, what are the assets available to cover that? As I say, our starting point was the method which is proposed currently does not seem to make enough allowance for the cover for all these levels.

Our methodology is a contribution to the debate rather than settled methodology. There is a bit of judgement there. I think the obligation side of this balance sheet is a very comprehensive piece of work which we did not want to add to. It is the asset side on which we are focusing.

Mr Piperdy: Just to add to Mr Setchim's comments, in the examples I have looked at, before you start even thinking about sponsor support value, you have a solvency capital requirement of some 20% to 30% of the liabilities, depending on the proportion of your assets and equities. You are already thinking about a considerable increase above the risk-free measure of the liabilities.

Mr Setchim: I should perhaps say, in case we have given the impression that this was an exercise to make the asset side as high as possible, there were examples in the 20 that we looked at where applying the method produced a higher value for sponsor covenant than we would regard as a normal valuation. This is not just about keeping clients happy. This is really trying to propose methodology that is appropriate to the circumstances.

Mr O'Mahony: This question is in relation to the market value multiple of the HBS deficit. Is an appropriate threshold two times or three times? I think two times intuitively feels very low when you consider all the variables, and the fact that no dividends had been coming out of the business, which is slightly unrealistic. It could be two, it could be three, it could be four.

The danger, of course, is that when we start making assumptions we can get any number that we want if we just double it or halve it. It is quite an imprecise science and we must not be hoodwinked into believing we are giving absolute or precise values. This is a very theoretical approach.

Mr Sutton: Valuation does involve judgement and there are ranges, but there are reasonable ranges, and robust analysis will get you to as narrow a reasonable range as you can. Those multiples are very industry–dependent.

It will be interesting over time to see whether you can build up a set of statistics which allows you more accurately to assess, based on the data that you have collected, what is a reasonable number for that particular sector or type of business.

Mr O'Mahony: The approach is do a quick calculation and see whether it is two times coverage at least and then, if it is, to put it to one side. As you say, maybe if you had polled people in advance, three times may be the appropriate multiple. I do not know whether you can calibrate that in any way.

Mr Sutton: We actually had three times in a number of drafts of the paper. In fact, in the last few iterations we were trying to get rid of the inconsistencies. We had two in some places and three in others. We ended up with two. I take your point.

Miss C. P. Hildebrand, F.I.A.: Can you comment on the availability and objectivity of the projected cash flows, and also the multiples?

Mr Sutton: In terms of multiples, there are a number of databases that are readily accessible to find transactions which have occurred. For a listed company, similarly, there are a lot of information sources that will give you that data.

On the discounted cash flow, I think that is a very good question. What sort of access to information would people have to be able to do the calculation? I think our view is sponsors should be able either to do the calculations or provide whoever is doing the calculations with a business plan as a basis for valuation.

Miss Hildebrand: But that sort of information is not routinely available to trustees right now.

Mr Sutton: It is available to some trustees. It is something that may need to change if this becomes law. May I add one more general point? If you are in a situation where you are close to the edge, then I think that there is every incentive for a sponsor to share that information with the trustee and make sure an appropriate answer is reached.

Mr Setchim: As a more general point, although we are looking at the components of the balance sheet, my hope is that this approach, if we are able to agree on how a holistic balance sheet should be constructed, should serve to improve joint working on these issues between sponsor and trustees. At present, I spend quite a lot of my time playing a game of tennis on funding issues, hitting the funding ball as far as possible over the net, and it comes back as hard as possible and that is just not the right way to do it.

I am less concerned about the availability of information. I think it would be in the sponsor's interest to help with the information required to reach the right numbers.

A speaker: You are favouring an approach that involves much more judgement than the approach which is proposed, which is much more parameterised and formulaic, and I think that is right. In bringing your practitioner's experience, how do you think it will work with a valuation where you have a lot of uncertainty? You give ranges for these sorts of things. Moreover any valuation of a business is potentially subject to some sort of bias where you have to agree a covenant valuation between trustees, sponsors and potentially the regulator. How do you see that working in practice?

Mr Sutton: I think a robust debate probably is the answer. I think where you could extend the methodologies would be, perhaps, to try to give some idea of what are the probabilities of various ranges in the valuation just so that people can see what the extent of the probability of the range might be. That I think would help to illustrate how variable is the number.

Mr Setchim: The perspective is intended to be that of the member. As a member, I am interested in my scheme and my sponsor. I think conversations between my sponsor and my trustees will go rather more smoothly if the valuation methodology at least enables there to be sufficient cover for the obligations rather than starting off with a built in gap. It is almost a political question. You have to choose your axis. Comparability is something which IAS 19 tries to achieve. You can read across financial statements and that is not scheme-specific.

What we are looking at is a method. People start off by saying, “It is very difficult to value sponsor covenant sponsor by sponsor.” Yes, it is difficult, but it can be done. At the moment we do that, but in a qualitative way. This is a way of doing it on a sponsor-specific basis which will produce a number. People are going to have to take their choices on the starting points.

Mr M. G. White, F.I.A.: How realistically do you think the market reflects the very different levels of threat or liability that the pension represents from one company to another?

Mr Sutton: I think my observation is that sometimes the market does not seem to reflect, for some companies, the full extent of the issue. Certainly, I have seen in less heavily traded businesses significant differences between what I would regard as being an equity value versus what the business seems to be trading at, given its pension situation.

Mr S. Willes (Non-member, Gazelle Group): I want to introduce into the discussion some work that we have been doing which is related to valuing sponsor covenant risk rather than sponsor covenant support. Both papers demonstrated that it is not easy to develop a practical quantitative approach to valuing the sponsor covenant itself which is required by the EIOPA insurance-based approach. It is also rather unclear what to do with the result when the holistic balance sheet does not balance.

Some time before the EIOPA proposals came into focus, we rather independently set about developing a quantitative methodology for valuing covenant risk itself. We felt that the current UK pension regulatory framework was more in tune with characterising pension deficits as debts, with a debt-like obligation from an employer to the pension scheme, where it is not certain whether the employer can pay the contributions required to fund the scheme benefits. This is not an insurance-based approach but more of a debt-based approach.

Quantification of default risk as a monetary value is a cornerstone of banking regulation and determining bank capital requirements. That is where we started, and we set about trying to adapt that to pension obligations which are rather different from loans, being rather longer term, having no fixed term structure and being a rather open-ended obligation, allowing affordability constraints, and things like that, to play out. We have done that by integrating the actuarial scheme funding model with a stochastic model of employer financial resources alongside the default risk quantification used in debt markets. It works rather neatly. One by-product of integrating the actuarial model, is that you obtain a quantitative linkage between the covenant risk, the liability risk and the investment risk, which is something for which quite a lot of people have been looking. A description of the methodology is on our website under the heading Gazelle MT (or Mousetrap), which is what we call the model.

I think being able to quantify covenant risk solves quite a few problems. You can use the value of covenant risk, to adjust the technical provisions discount rate in scheme valuations. It is obviously good for transactions because it gives a very precise result for before and after covenant risk. I think, in conjunction with asset liability modelling, it could give funding targets which reflect both uncertain contributions and uncertain investment performance, and it should give the Pension Protection Fund (PPF) a very accurate idea of its financial exposure to schemes.

It gives the current existing regulatory approach we have in the UK some quantitative teeth which perhaps it lacks at the moment. We wonder whether the insurance approach to quantification would really be necessary if the current approach had quantitative teeth. If the market could access quantified covenant risk values for defined benefit schemes and use those to manage pension risks, do we need to value sponsor support as a balance sheet item?

Mr Setchim: Our starting point was in response to the IORP consultation, so we were faced with the insurance-based approach to liabilities. We turned our minds to what methodology would be well-understood in the market (well-understood not just in financial services but also across other market sectors) for deriving values that will achieve EIOPA's objective of showing all the assets on which the pensions obligation can rely.

Whether, in another world, or if the IORP did not come to pass, there are lessons which can be learnt either from here or indeed the Mousetrap, is a question beyond our remit. My challenge to our valuers was to use methods that are not too new because they have to be understood and capable of verification and challenge to achieve what EIOPA is setting out to do. There were plenty of commentators who were challenging whether EIOPA should be there at all. That is not the problem we were addressing.

The Chairman: One of the remarks made was what to do if there is no balance? That is a very interesting question. First of all, it is of course how do we value everything, the liabilities, all the add-ons to the liabilities, the SCR and then, on the other hand, the sponsor covenant.

If it does not balance, just talking from my perspective, which is a Dutch perspective, I would say: “let us sit round the table with all the stakeholders and discuss the situation”. This could be a call for real discussion on the nature of the pension promise in the first place. I would say it is a social contract between an employer and employees. These two parties can decide what it is and what it is not.

If a pension promise is like an insurance product, and if it is a fully guaranteed promise, you will always get your pension no matter what happens. Then, I think, it is very difficult to argue that a different method from insurance valuation should be applied. In most cases, pensions, given the social contract element, are not a full guarantee; but that is difficult to discuss because the members normally experience it as a full guarantee and want to keep experiencing it as a full guarantee. A lot of people do not want to break the dream or to make it different. A full guarantee is, in my view, impossible. To do this theory says you would need an infinite amount of money. That is not possible, so we already allow for some uncertainties.

Also on the insurance side we have 99.5% certainty. But perhaps for pensions it is lower, why not discuss that more clearly? Why should we not invite social partners to become more clear on what is the pension contract?

Mr F. N. Fernandes, F.I.A.: I would like to echo that point.

I am pleased to see that in the paper, in the final paragraph of Section 1.1 of the PwC part of the paper, there is a reference to what happens if you fail the holistic balance sheet test. Is this not the key question? At the moment, until it is clear what happens in this scenario, who really cares how it is calculated? From my point of view, we need to take a step back. I think that some of the areas you have highlighted in the paper raise many questions.

Firstly, having worked on merger and acquisition (M & A) transactions at investment banks, the paper highlights many of the problems and varied approaches to addressing those problems. For example, in the discussion so far today, some good points have been made about the basis underlying an approach using terminal values and multiples in valuing businesses. Again, you have to wonder if it is right to create over-complicated stochastic models referencing available corporate bond spreads when, in reality, the entity you are referencing is not the entity sponsoring the pension scheme.

The Chairman: Would it be possible to discuss that upfront? What would happen if the balance sheet does not balance? That is the debate in the Netherlands at the moment: to make the pension contract more transparent and to discuss up front with social partners what will happen. In the end there will always be a situation that you have not been able to think of beforehand. It is valuable to at least try to describe a lot of possible scenarios that can occur and the actions that you can do or are not willing to do.

Pensions are part of a total remuneration package. You can ask for additional capital for pensions. But that should come from somewhere. For example, do we need lower salaries, or for people to be fired? The pensions question also balances with other elements. We can answer the pensions problem only to then have a problem elsewhere. That makes the discussion even more difficult.

Mr K. R. Wesbroom, F.I.A.: I would like to say something about market consistency. The methods to arrive at assets and liabilities are wholly inconsistent. On the liability side we have a risk-free rate. So far as I am aware, nobody, when they are valuing businesses, uses risk-free rates. So why are we bothering with trying to get consistency with a mark to market asset value? Perhaps what we ought to be doing is using judgement to put an equivalent value on either the asset or the liability side of the balance sheet. You can see where this one is going: “Join me going back to the 1970s, we can get there – let's use actuarial values for assets and liabilities rather than mark to market values”.

The other point that struck me was that Mr Piperdy and Mr Sutton seemed slightly surprised that they came up with different values. But Mr Piperdy is flying with one hand tied behind his back because his method takes off half the shareholders’ value and a third of the income. I think that if Mr Sutton did that, he would probably come up with the same number.

Mr Sutton: In terms of the numbers, the EIOPA approach does have elements built into it, such as taking 33% of the cash flows, which my approach does not. Another important difference is that I have normalised the cash flows rather than just taking the actual ones.

Mr Piperdy: If we consider the example we look at in section 4.3 of our paper, the reason the company is penalised is that it had negative cash flows in two of the last three years which it actually used to pay down debt in the business which is normally considered a covenant enhancing measure. Under the QIS Approach, this action actually destroyed value in the sponsor support calculation.

The Chairman: I must say I like the idea of judgement in the valuation of the liabilities. But it should be based on the contract. I am going back to what I said earlier. It is the contract between social partners.

A while ago I was referring to 99.5% certainty. Perhaps that is one of the elements that we can change and make lower, probably, not higher. Another way of doing it that could also be market-consistent is saying we accept that the pension promise is not fully guaranteed so a risk-free rate is not the appropriate discount rate because that belongs to a fully guaranteed liability. If it is not fully guaranteed, tell us, social partners, what the discount rate shall be. How uncertain do you want your pension promise to be?

That could then be the risk-free rate plus a risk premium as is the case on the asset side. Then we actuaries can put it into our calculators and come up with the valuation of the liabilities. That could be much lower than on a risk-free rate and not because we want to have it lower but because it reflects the risk appetite of the social partners.

Mr J. M. Lowes, F.I.A.: Mr Sutton is looking at the maximum value of sponsor support. You can go about doing that in different ways. The EIOPA way is slightly different but I understand what he is trying to do. The next step that the QIS considers, and I think Mr Turnbull was taking the same approach, is to look at trying to put a value on actual sponsor support rather than the maximum sponsor support. They both go about it by putting a market-consistent value on a set of cash flows.

Actually, the set of cash flows that the QIS looks at, and which Mr Turnbull looked at, come from absolutely nowhere. They take an assumption that you fund to buyout, make technical provisions, or whatever, over a period. That is not what is actually going to happen unless you suddenly change the whole regime so you are funding pension schemes to buyout level, so it does not reflect the cash flows that are going to happen. We put an awful lot of work into putting a market-consistent valuation onto a completely random set of numbers. That does not sound to me like something that you would sensibly do.

It also does not look very holistic because it is looking at the support from the sponsor as if it was identical to a set of payments from a bond that the pension scheme happens to hold. It is not treating pension schemes differently from insurers by looking at the value of support from the sponsor and treating it differently. It is looking at a notional bond as if it were held by an insurance company and treating it exactly the same. It is no more holistic than saying an insurance company can hold a bond. It is not treating the sponsor balance sheet as an integral part of the support available to the pension scheme.

It is like looking at a set of insurance business and saying that we are only going to value the cash flows in and out of the asset share in respect of that block of business, and ignoring the support available (but not expected to be paid into the asset share) from the wider insurance company balance sheet. If you looked at insurance companies like that, every insurance company would look insolvent. Mathematically, the way the QIS looks at it and the way Mr Turnbull's calculations do it, the actual value of sponsor support, or the values placed on that, will never be enough except for a risk-free business. It does seem nonsense to put a lot of calculations in to derive something which, by definition, will never be enough.

Mr Turnbull: With the method applied, it can be enough and, in some cases it is more than enough. With a risk-free sponsor that can come through in the calculation. I think the more interesting point is the point you make about assuming away risk, if you like, by how we assume these cash flows arise. It is absolutely right. It is completely contingent on when we assume this cash will be paid and in what circumstances it will be paid.

A sponsor covenant, however we value and analyse it, must be very sensitive to that because at the point at which the funds go into the pension fund if it is no longer exposed to that sponsor risk, if that cash is paid in the next year or in 10 years’ time the riskiness of the pension fund would be materially different. Whether that is encapsulated by a mathematical formula or some other piece of judgement, it has to be considered somewhere.

More generally, there is an interesting parallel with what has been done with with-profits where you have a very similar situation where there is a temptation to be able to assume away risk by assuming that the company will take certain management action which will sort the problem out in the future and therefore the risk disappears and the need for capital disappears. In that instance, there has been a lot of governance put around that process whereby the assumptions that go into the valuation and capital models about the point the entity does certain actions is documented and the company is held to account over time by the regulator about whether or not their actual actions are consistent with the assumptions that they had made. I absolutely agree with your point that it is fundamental to the analysis that these assumptions are made. But actuaries have gained some credibility and experience in managing that type of sensitivity and uncertainty in looking at the risk in long-term liabilities in other fields.

Mr Lowes: I think I would go along with that if the cash flows were arguably paid a bit earlier or a bit later, but they are not. They are just wrong numbers.

Mr Turnbull: But you are assuming what your cash flow policy is.

Mr Lowes: That is not what the QIS is looking at.

Mr Turnbull: I agree with that.

The Chairman: The QIS, in my view, is just an exercise to have discussions like this and to learn from them. I am sure that EIOPA and the European Commission is willing to listen if there is a good story to tell.

Mr Turnbull: Absolutely. The approach I envisage working would be one where at a scheme-specific level the contribution policy is stated and specifically captured in the valuation. That is what I am trying to capture in the framework. Then you would understand if you had enough to meet the capital requirement.

Mr D. O. Cule, F.I.A.: I sit on a subcommittee of the ICAEW which advises 6800 clients, who are not quite as big, I suspect, as the 20 clients that we have been discussing. The members of the subcommittee will worry about who would sign off balance sheets such as we have been discussing.

From my personal point of view, like many others, I find a balance sheet that does not balance a bit odd. I also find an impact statement that has no impact in it also equally odd. I do go with the Chairman's view about the need for public interest not just meaning an individual public interest but the public interest from a wider viewpoint. Here we already have the PPF in play. Politically, that seems to give an acceptable reduction in benefits that occurs when companies fail.

The only time in fact we can let benefits fail is when companies fail. That is perhaps the first very interesting point: trustees should never let a company fail because it is absolutely not in the trustees’ interests and the members’ interests for a company to fail. That is a great challenge, I know, to the regulator. He is a policeman who should be looking after that world and not necessarily the trustees. The trustees have a great interest in keeping the thing going.

We should have a social debate about who benefits from pensions. I suspect our world is a bit different to the Dutch world. Our world of social debate takes place in the Daily Mail, where it is now quite clear that pension pots, which are simply a multiple of pension rights, are unacceptable and they are unacceptably big. That is an interesting one.

We also then have the other side that transfer enhancement offers are now unacceptable, although I did not see that being the Daily Mail's issue. I did see that being the adviser's issue, that they were not being paid enough for giving advice that you should not transfer. But that is another issue. It would be interesting if we could trade pension rights. The DC people with pots think that it is too expensive to buy pensions. All these people with pensions would love to trade them for cash. Could we not work out some sort of market where they trade with each other and where they would get the appropriate risk to be shared across each other?

I do not know whether that helps the debate a great deal other than echoing the Chairman's lead that, actually, liabilities are not guaranteed. That is what makes the balance sheet balance. You cannot pay £1.7 trillion of liabilities with only £1 trillion of assets put aside. You cannot get society to do that. You have to move the liabilities downwards. Whatever system you look around for measuring it and observing it, they, ultimately, have to be reduced.

My final bit of evidence as to that was a letter that my wife and I received this week. We are the same age. My letter said that my State retirement age has gone up to 66, and so did the one my wife received. My State retirement age went up only one year, whereas her State retirement age went up six years. You cannot at this stage do that with private sector schemes but it is only a matter of time.

Mr A. G. Sharp, F.F.A.: One question for the panel generally at a more prosaic level: I know that you have all tried to look at practical outcomes for measurement. I just wonder how practical it will be for a small charity with a 10 million DB scheme or another non-profit organisation with a 25 million DB scheme to use any of the methods that you are suggesting in a practical way.

Mr Chairman, I wonder if you have any views or comments that you could make with your Groupe Consultatif (GC) hat on as to the likelihood of the HBS actually coming into force in a new IORP directive.

The Chairman: The Groupe Consultatif is very much in favour of a risk-based approach. That is not necessarily the same as using an HBS. One of the comments that we have made is that it is unclear so far how the HBS will be used, what supervisory measures will be taken and when. That can give a completely different picture on how you value such an approach.

The basic idea of trying to identify all the building blocks of the financial position of a pension fund, and how that evolves through time, is something that we do support. There is more knowledge about the various risks that are involved in pensions. That is a good thing. How it works out in the end is a political matter.

We can wait for the European Commission or a supervisor. We can also start debating this among ourselves, the stakeholders of a certain pension scheme sitting together and discussing these issues. There is not one answer to it, I think. There are different consequences, depending on the measure that you will take.

I think, perhaps this is also very Dutch, that you should sit around a table together, discuss it and come to a solution. It can take a long time. In the end, everybody needs to give a little bit and come up with something that makes sense. Then I think the regulators have to follow.

All of a sudden there is a different agreement, or perhaps I should say more clarity about the agreement, than was before. Then we have to supervise this agreement. That is how the Groupe thinks of it supplemented by some of my own remarks.

Mr C. Patel, F.I.A.: I speak as an actuary, not in my capacity as a regulator helping to develop the GC approach in this area.

I prefer to look at these two pieces of research on an enterprise risk basis, in other words, on a joined up basis. When I am given a business value which is supposed to provide some support against a liability valuation then I need to join those together and I look for consistency. The point that Mr Wesbroom made earlier was very well made and worth emphasising.

There are apparent inconsistencies in the discount rates used on the liability side by the actuary and that used on the asset side by the covenant valuer. Does this, I wonder, open up the whole debate about what is fair value in this context? What is a market consistent value of the sponsor covenant? Should this be a value based on your valuation or mine? In a commercial market there is room for both, as the PwC approach highlights. In the context of the holistic balance sheet there is a third one from the members’ perspective. That, perhaps, is a subject on its own. So we look forward to another paper from somebody focussed specifically on the purpose and uses of such a balance sheet.

But there are other aspects of the valuation basis which, as a user, upon which I can comment. When it comes to business value the growth rate of the company is, I believe, a critical component of the calculation, as is the duration of that growth. Two points arise. One is that the growth rate itself must in some ways be linked to some of the macroeconomic parameters, for example the inflation rate. Those are the very same parameters that the actuary needs to take into account in assessing the liabilities. So there may be a disconnect there unless the two are made consistent. On the duration aspect, similarly, the business valuation might just take a short-term view of three years, five years or 10 years at most. The actuary's view is very different – he or she is concerned with running off a liability flow that might span 70 or 80 years.

Therefore, when looking at capital values under such duration mismatches, there is a danger that users begin to believe that the holistic balance sheet actually balances because the arithmetic says so, when in reality it may not if you look at it on a time-related basis. This may not matter to the traditional users of the business valuation approach because their purpose is different, but it is important to us actuaries who need to address a very different question – that of solvency against a liability for which the covenant value is being offered as support.

One final comment is on the subject of whether a two times or three times multiple is sufficient. I would look at it differently. What is the purpose? It is to give you a buffer against the liabilities. Given this why not look at the liabilities, because they give you some useful information? For example there is a lot of volatility in the liability cash flows and some of this may have been measured. Investment risk is usually a significant factor in a solvency balance sheet, and nowadays easily quantifiable. If you have used Mr Turnbull's light method, then there might be a lot of uncertainty in the measurement, so how about bringing that in? That might actually help to put some science behind what might be the rule of thumb for a sufficient multiple.

Mr T. R. Webb, F.I.A.: I think that the point that you raised earlier was that we need to look at what happens if you fail the holistic balance sheet test. If we look at what has happened in the UK: once upon a time pensions used to be a best endeavours promise, then someone came along and changed that down to “It is better to make the company insolvent than risk the members not getting their pension.” We are worrying about market-consistent valuations. We have a fundamental inconsistency in that the stick (the Section 75 debt) that we use to beat up the company is requiring absolute certainty that the pension will be paid otherwise you can go into insolvency, but we do not have a compensation scheme which does anything more than pay about 80% overall and then only if we (the PPF) can afford it in the long term.

It would be lovely if we could get round the table, but I think the sceptics here will assume that we are going to be beaten up with another stick.

The Chairman: We have a lot of new ingredients, I think, to the discussion. I really would like to thank the authors for the work that they have done, and also I should like to thank Mr. Evans who was in charge of the review of all this work.

The President: It remains for me to formally express my own thanks to all those who contributed to tonight's discussion and, in particular Mr Valkenburg for chairing the discussion which has given us an additional perspective to the debate.

I want to pose three questions which have been running through my mind, just to get them on the record. Firstly, how sound is market-consistency for valuing deficits and covenants of defined benefit schemes? Secondly, if there is a weak covenant, so what, where does the extra money come from? Thirdly, should we allow funded DB schemes to be further killed off without challenging the significant quantity of unfunded pensions?