Volatility in pensions

Volatility is seen as the great evil of pension schemes. Be they defined benefit (DB) or defined contribution (DC), it is volatility that is the greatest cause of concerns. But should it be?

To answer this we need to look at the causes of volatility and the impact of it.

There are two types of volatility people worry about, the first of which is the volatility of asset values. It is easy for trustees and companies to overly focus on this volatility as asset values are very visible and stock market swings are publicised in the news. But is this volatility important? Not really.

If I have 500 euros it would be easy to get concerned about their value going up and down with the exchange rate. But if I’d agreed to pay 500 euros for my hotel when I go on holiday next month then it really doesn’t matter to me how much they are currently worth in sterling. My transaction has no volatility at all.

The key focus of pension scheme volatility therefore needs to consider both assets and liabilities and hence, for DB schemes, the deficit or funding level. And this can be very volatile!

I did some analysis for a recent conference talk looking at history and put together the chart below. This shows how the funding level of a scheme would change over time assuming market based (gilts+) type valuations and 10 year spreading of surpluses/deficits. The funding level routinely changes by about 10% each year.

Funding volatility

This chart also shows the recent crisis in pensions caused by a perfect storm of:

  • Not great investment returns
  • Increased mortality expectations
  • Falls in bond yields

The current ultra-low bond yields actually being the second wave of large falls in bond yields/interest rates. The first starting at the end of the 90s.

This perhaps also brings into focus what we really mean when we say we dislike volatility, and that is falls in the funding level. Upside volatility is fine. For example some people talk about volatility from mortality but actually this has generally been pretty smooth (albeit with some step changes to reflect late application) upward only (in terms of liabilities) adjustments. So, not volatile, just more costly! The fact that it is relatively smooth reflects the fact that data only arrives each year and 1 year’s data is not used to determine the future expectations. It is accepted that 1 year mortality figures can be volatile and not necessarily the best guide of the future. A useful concept to bear in mind when we look at the causes of volatility.

So what are the causes of funding volatility?

There are many specific causes but fundamentally it’s because liability values and asset values don’t change in the same way.

The biggest specific cause of volatility in UK pensions? Long dated bond yields. And this is primarily because liabilities are generally pegged in some way to these yields via the discount rate used. For some measures, such as accounting bases like FRS17 or IAS19, this is prescribed. For others, such as buy-out costs, it reflects reality in how insurers are required to reserve for the benefits. But, for pension scheme funding, it is often a modelling choice made by the actuary, led on in no small way by the Pensions Regulator and a bunch of “economists” that don’t really understand what pension scheme funding is all about (and who struggle so much with 2 definitions of the word valuation that actuarial standards now require us to attempt to explain it to them in every report we do!). You can perhaps see what side of this debate I sit on.

Essentially the biggest cause of volatility in UK DB pensions is our choice of tape measure.

Real volatility

When it comes down to it, the fact that pensions are funded at all is a choice, and it would actually be more efficient not to fund them. A pension scheme itself is merely a series of future cashflows. We fund the scheme by investing in assets to enable us to pay those cashflows. So what really matters is the volatility of cashflows that can be obtained from our assets.

Bonds are therefore of course a great asset for this as you can build a cashflow stream in line with that expected for your liability payments. Investing in bonds can therefore remove a large amount of volatility through matching – and in this case there is no debate that the yield on those bonds should be the discount rate used.

These days things are even easier as synthetic bonds can be used to provide better and/or more cost effective matches for liability payment streams.

The downside of bonds though is their return – or lack thereof – particularly gilts in the current climate.

So what about growth assets or, in particular, equities. Equities prices are extremely volatile. They also have almost no correlation to gilt yields so if your liability valuation is pegged to gilt yields your volatility goes up further still. But in the long-term (and let’s face it pension scheme are long term) the vast majority of the return you get from holding equities comes from the dividend payments, not the change in capital value. And dividend payments aren’t that volatile. So, if we look at things from a cashflow matching perspective, a low volatility dividend payment combined with greater expected long term returns looks very attractive. If you make allowance for dividends when determining your discount rate this reduced volatility can effectively filter through to your funding position.

Current equity prices, rather like last year’s mortality experience as I mentioned above, are not a great guide to their long-term future worth in many cases.

Other growth vehicles

The growth vehicle of choice at the moment is the Diversified Growth Fund (DGF) of some form or other. These are all very different in strategy but have similar aims of long-term equity like returns but with lower volatility. In actual fact their “target” is often stated as cash plus.

The reduced volatility of assets that these bring is attractive. However, they offer no form of cashflow matching at all so in some ways could be viewed as more risky than holding equities over the long term. Just a thought…

Where cash plus does come into its own though is when using derivative products to generate long term cashflow matching in exchange for cash outgo. If your DGF gives you cash plus with vastly reduced capital volatility then maybe the biggest form of volatility in scheme funding is solved without losing out on the longer-term return benefits.

I’m more traditional and holistic in view point and favour long term equity biased growth strategies with more appropriate valuation techniques. However, in the current legislative environment, a DGF + swaps type strategy looks very attractive.

Advertisements

How to assess a pension fund’s health

Originally posted on IPE: How to assess a pension fund’s health

I read Jeroen Wilbrink and Jelle Beenen’s article on determining the health of a pension fund with interest, as this is an ongoing debate Wilbrink and I have had for the last two years. I therefore couldn’t resist writing a response to their arguments. They make the claim that Cools and van Nunen are confusing issues. I hope to show that it is actually Wilbrink and Beenen that suffer from this confusion.

Their article makes many references to arbitrage-free pricing and risk-free valuation, quoting many academics in this area. Of course, quoting many names does not make the theory any more appropriate to use. Risk free, or more appropriately, risk-neutral valuation is indeed the foundation of a huge market of derivatives. However, the clue as to why this is not relevant is in the name, ‘derivatives’.

Derivatives are products whose value is derived from other assets. A simple example is a forward contract and a more complicated one an option over an asset. Derivatives are valued using risk neutral, or arbitrage free, pricing because it is possible to create a perfectly replicating portfolio that will have the same payoff as the derivative from a portfolio of the underlying asset and cash. This means that no assumptions need to be made about the valuation of the underlying asset. Derivatives are about relative pricing. You wouldn’t, for example, use risk-neutral valuation to determine the value of an equity or property.

The theories quoted are powerful but not relevant for valuing pension schemes. A pension benefit is an asset in its own right, not a derivative of another asset.
Wilbrink and Beenen would perhaps like to suggest that pension benefits are a derivative of risk-free investments. Indeed, they quote the example of a member having no risk tolerance and that the pension benefit should be treated as being risk free. They have ignored a big part of what a pension benefit is, though, when making this assertion.

Let’s say there is a 20% chance of me dying over the next 10 years. You have a choice as to whether you buy Asset 1 that is a payment of £1,250 in 10 years’ time, provided I am alive, or Asset 2 that pays £1,000 in 10 years’ time with no risk attached – secured by the UK government, for example. The expected cash flow in both cases is £1,000 – i.e. 80% x 1,250 for Asset 1 and 100% x 1,000 for Asset 2. Which asset do you prefer? Which asset, therefore, has the highest value?

Asset 1 is, of course, very similar to a pension scheme benefit where the payment is dependent on the member being alive. To suggest that the ‘value’ of that benefit is the same as a risk-free bond of the same term is just plain wrong.

Wilbrink and Beenen give the example of a scheme changing its investment strategy and having better coverage. They write: “We would argue that the coverage ratio is still 100% for both funds; nothing changed in the value of the investments or the liabilities.” I completely agree with the second part of their sentence, that nothing changed in the “value” of the investments or liabilities. However, I disagree with the first part (assuming it really is in the pension context such that the cash flow was dependent on the member being alive). In line with my comments above, the coverage under both investment strategies is more than 100%, as the pension benefit is not worth as much as a zero-coupon bond.

However, I couldn’t tell you how much above 100% the coverage was, as I don’t know what the value of a pension benefit is. There is no market to tell me this, nor would it be desirable to have one. In fact, in the UK, we have specific legislation to prevent it. I could construct models to come up with a ‘value’ in the same way as I could value a company share or a property. It would only be a model, though, and highly subjective to my personal views. A market price would reflect everyone’s view. For these reasons, talking about ‘market value’ of pension liabilities is a pretty pointless and unnecessary exercise.

This brings me to the other two approaches to looking at the health of a pension fund. Moving away from ‘value’, we can think about the concepts of ‘budgeting’ and ‘reserving’. In many ways, these are the same thing and just reflect the level of prudence allowed for. Let’s look at budgeting first.

Budgeting is about answering the question ‘How much money do I need today to pay for things tomorrow?’ This is fundamental to what pensions actuaries call a ‘valuation’. In many ways, it is merely this confusion of terminology that has led to so much debate.

If we are to budget for something on a best-estimate basis, then we arrive back at the position Wilbrink and Beenen dislike, where different investment strategies suggest that less money is needed. There are two key points they make. First, that this is not appropriate when monitoring progress, and second, that this is extremely risky.

Dealing with the first, I’d like to return to the football analogy they used to dismiss the idea of allowing for expectations. They gave the example of a football team fooling themselves that it was OK to be a goal down, as they expected to score three goals. If you are the manager of a football team and expect to win a game but then go a goal down in the first 10 minutes, what is your reaction? I would suggest the reaction is that there is no need to panic, as there is plenty of time left to turn things around. The manager would be unlikely to change his tactics. He certainly wouldn’t give in and accept that 1-0 was the result.

If it got to half-time and it was still 1-0, then the manager might be a little bit more anxious and change things around a bit. Once it got to the last 10 minutes, the super sub would be on, and all-out attack might be on the cards. The manager of the team monitors the game against his expectations of what the team might achieve. Monitoring is, therefore, more appropriate allowing for expectations than without it. Without the expectations, the manager might make bad decisions early in the game.

The second point they make relates to the risk or variance of outcomes not being allowed for. They use an expected equity return over 200 years to demonstrate this. All they are really showing, though, is the power of compound interest. A similar picture could be shown with bond yields and large changes in interest rates. This is a completely unrealistic variance to allow for, though, as it effectively assumes that no monitoring or adjustments would be made throughout the entire 200-year period.

It is also suggested that interest rate risks can be hedged. This is true and, with legislation where it is, certainly worthy of consideration. However, if you are investing in other assets, all you are really hedging against is the way we choose to ‘value’ the liabilities.

Finally, this second point brings me to a third possible approach in considering the health of a pension fund – ‘reserving’. Reserving is merely budgeting with prudence. It is the backbone of how insurance companies operate. When Wilbrink and Beenen talk about a member’s expectations of benefits being risk free, it is reserving, rather than valuing, that makes sense. A large reserve would provide such certainty, whereas a small one wouldn’t. In this scenario, bond yields could be used as part of the measure. This is a choice, though, rather than a necessity and is certainly not the only answer. The level of any such reserve is open for substantial debate – the larger it is, the more certainty it provides. However, the larger it is, the less efficient the provision of benefits becomes.

The solution to this protection vs efficiency equation comes in the form of insurance, but that is a topic for another day.

I hope I have shown why risk-free is not always the answer and why pension benefits should not be confused with bonds. Ultimately, the decision on how schemes are funded is a political one. If good, efficient occupational pension schemes that share risks are to thrive, then the answer should lie at the budgeting end of the spectrum. I therefore hope politicians will take note and regulate pensions in a way that provides the best outcomes for all of us.