2015 - are we there yet?

Tweet
Defined benefit pension funds that are financially unsound will have less room to manoeuvre following recent proposals ( FSB Draft Notice on Financial Soundness of SA Pension Funds ) by the South African pension fund regulator, the Financial Services Board (FSB).  The key proposals by the regulator will see:

• Greater prescription of calculations used to determine a pension fund's financial soundness. In particular, an obligation on actuaries, who carry out these financial soundness assessments, to estimate a value for  the pension fund's future obligations using a "market-consistent" approach. In other words, forcing a change from a "long-term" actuarial value based on an actuary's assumptions about the future to the use of a more transparent and readily observable "market-related" value for calculating the value of future obligations of the pension fund.  

• Any pension fund with a funding level (or ratio of assets to liabilities) that is less than 95% being required to submit, to the FSB, a clear plan for restoring its funding level to at least 100% within a period not exceeding 3 years.

Pension funds should ensure they understand the implications of these proposals and the clear impact on their investment strategies. Not to do so could lead to the adoption of inappropriate measures to counter the adverse impact of the proposed regulations. This, in turn, could condemn their pensioners to a lifetime of pension payments that are lower than they might otherwise have been.

1. Calculating the financial soundness of a pension fund requires an assessment of both asset and liability values.

Assessments of financial soundness of South African pension funds are typically carried out by a qualified actuary who will also be a Fellow member of the Actuarial Society of South Africa and be recognised as having appropriate experience to carry out such valuations. 

Assessing a pension fund's financial status requires two key inputs:

1. an assessment of the value of the pension fund's assets and

2. an assessment of the value of the pension fund's liabilities.

By comparing the value of assets to the value of the liabilities we can quickly determine whether the pension fund is financial sound or not. If the value of assets exceeds the value of the liabilities then the pension fund may be considered to be financially sound.

2. Asset values are relatively easy to calculate....

Placing a value on the assets is relatively easy - a good starting point is the market value of the assets. Market values are readily available for assets that are listed on a recognised exchange, such as the Johannesburg Stock Exchange (JSE). Placing a value on assets that are not listed on exchange, for example property, is a little more difficult but even here qualified valuation agents exist and these agents are able to estimate the value using, for example, data for recent purchase (or sale) transactions carried out in similar assets. This is similar to a home-buyer looking at recent property sales for comparable properties to gauge the price for a property she may be interested in acquiring.

3. Liability values are an estimate of how much money we need today to pay all future obligations to pensioners...

The liabilities of a pension fund are its future obligations to pay pension benefits. It is, therefore, the amount of money that the pension fund must set aside today to be certain of meeting all of its future obligations to pensioners. 

4. ...and this amount varies depending on what return can be earned on the money that has been set aside.

If we agree that the value of the pension fund's liability is the amount of money it must set aside today to meet its future obligations to pensioners then this amount (and hence our estimate of the pension fund's liability) will vary depending on what we estimate we may be able to earn on these assets. Intuitively this can be seen as follows: if future returns that can be earned on monies set aside to pay future pension are lower, then more money must be set aside today to be sure of being able to meet all future obligations. To calculate the value of the liabilities we therefore need to estimate a future long-term rate of investment return. 

5. Actuaries have used a specific method to derive their estimate for future investment returns....

In estimating future investment returns in the past, actuaries have relied on long-term historical returns to guide them on what an appropriate assumption may be for future returns from different asset classes. So, an actuary would:

a) consider the asset classes (for example, equities, bonds, property etc) that the pension fund intends to invest the monies it has set aside to pay pensions;

b) derive an average rate of investment return on that portfolio from looking at historical returns (potentially with some adjustment for mean reversion of returns) and

c) use this average rate to estimate how much of money the pension fund should set aside to pay future pensions. 

6. ...and this method means that two actuaries can place a different value on the same set of pensioner liabilities...

Under this approach, the value of the pension fund's liability varies depending on the investment mix of the pension fund. For example, if equities are assumed to return 15% p.a. over the long-term and property 12% p.a. over the long-term then a pension fund that invests 75% of its assets in equities and 25% in property will expect to earn a higher investment return than a pension fund that invests 75% in property and 25% in equities. The former will therefore believe that it can set aside a smaller amount of money (in other words there is a smaller liability value) to meet its future obligations than  the latter.

Also under this approach, the the value of the pension fund's liability varies depending on the actuaries assumption for future investment returns. For example, if an actuary assumes that equities will return 15% p.a. over the long-term while another assumes that equities will earn 20% p.a. then the latter will believe that he can set aside a smaller amount of money (in other words there is a smaller liability value) to meet its future obligations than  the former. 

In summary, an actuary's estimate of the liability value varies depending on: 

• their assumed mix of assets and
• their assumed rate of return for each asset type in that mix.

7. ...but can the same commitment to pay the same set of pension cashflows have different values simply based on how assets are invested or assumed rates of return for those assets?

Under the above approach a pension fund could reduce the value of its liabilities by switching its investments into higher yielding assets. For example, if a pension fund actuary valued its liabilities at R1m on a specific day based on an investment strategy that was 75% in property and 25% in equities, he could simply change the investment strategy to a higher yielding one (say 75% equities and 25% property) and reduce that liability to, say, R0.8m. Alternatively, the actuary could assume that returns on his existing portfolio were going to be higher in the future and similarly he could decrease the liability value.

What this illustrates is that estimates of pension fund liabilities can vary dramatically depending on assumptions made about investment strategy and future investment returns on a wide range of asset classes. But is this correct?

8. This method of valuing liabilities ignores the fact that investment markets already offer us an estimate of the cost of meeting this liability... 

At first glance the above approach sounds sensible. It is only when one is told that an existing investment market, the bond market, already provides an estimate for the cost of meeting this liability that we begin to question whether it is fair for actuaries to place their own value on this liability -an "actuarial" value that is different to the one that the market places on this liability.

This leads to the nub of the trickery that the actuarial profession has for a long time being deluding itself with. The very same liability or commitment to pay pensions can have two very different values depending on how the assets are invested or what the actuary assumes those assets will yield. This "actuarial" view of the world was, for the most part, kicked into touch by a seminal paper ( The Financial Theory of Defined Benefit Pension Schemes - Jon Exley, Shyam Mehta and Andrew Smith ) in 1997 by three British actuaries Jon Exley, Shyam Mehta and Andrew Smith who dared to challenge the conventional view at the time.

In essence, Exley, Mehta and Smith recognised that a pension fund's commitment to pay pensions is very similar to an investment that already exists and is traded and priced by investors, namely bonds. Like our payments to a pensioner which are for a specific amount, at a specific point in time, a bond is a promise to pay defined cashflows at specific points in time. By purchasing a bond an investor will receive his investment in that bond back with interest. That interest is the return on the bond.

The payments to the bond investor will take the form of a future stream of defined cash amounts to be paid at specific points time. Included in these amounts will be both the return of the initial capital plus an amount of interest (the return earned on the capital invested).  The price that an investor in this bond is prepared to pay to secure the cashflows from that bond reflects an estimate, in today's money terms, of all those future cashflows. This price (or value) provides us with information about the price (or value) that we too could use in order to place a value on our future stream of cashflows to be paid to pensioners. 

To price (or value) our pensioner cashflows in a way that is inconsistent with the way in which the market prices cashflows to be paid on a bond would be to assume that we have more information and greater insight into the course of future investment returns than investment markets themselves. We should be loathe to ignore this information as it captures the expectations of millions of investors and (much like the value placed by the markets on our assets) it is a value placed by the markets on a stream of cashflows, a stream of cashflows that is very similar to our pensioner obligations.

In summary then, when working out how much money to set aside in order to meet a pension fund's commitments to its pensioners to pay a future stream of cashflows, the basic building block for an actuary's assumption about future investment returns should be the return that could be earned on a bond whose cashflows were similar to our pensioner cashflow commitments.

9. ...and by ensuring that pension fund's use this information about investment returns, from the bond market, to estimate their liabilities would not only ensure consistency amongst actuaries but also ensure that pension fund's liability estimates were consistent with the market's price for those liabilities.

Following the publication and debate of Exley, Mehta and Smith's paper in 1997 the UK pension fund industry, led by the actuarial profession, rapidly moved away from valuing their liabilities using investment returns that were inconsistent with those implied by the bond market. By doing so the actuarial profession would ensure that two different actuaries, valuing the same set of liabilities, would achieve consistency in their valuation and would no longer arrive at very different valuations because each was assuming a different rate of investment return based on either different investment policies or different assumptions for future investment returns. 

Furthermore, that the estimate of those liability cashflows would also be continuously updated as new information about future bond market investment returns was received. This occurs because as bond prices change so too does the implied future investment return on those bonds. And as this investment return changes, our pension fund's estimate of its liability value should also change to reflect the market's revised view of the value of that set of cashflows.

10. But what about the fact that we could earn a higher return if we invested our pension fund money in assets which yielded more than bonds?

When we value our pensioner cashflows by looking at the value that the bond market would place on those cashflows we are also implicitly assuming that any monies set aside by our pension fund to meet its future obligations to pensioners will only ever earn a rate of return implied by the bond market. Historically, over the longer-term, bonds have tended to deliver lower yields than equities. 

So it could still be argued that as our pension funds do not invest all their assets in bonds then the the future investment returns that our pension fund's can expect to earn will be higher than those implied by bond market returns. If this is the case then by solely relying on the bond markets for placing a value on how much of money to set aside to meet our future pensioner cashflows we could be adopting an unnecessarily cautious view of the amount of money we need to set aside.

That is true, however, Exley, Mehta and Smith argued that this should not change the fact that the starting point for future investment returns to be assumed by actuaries should be those implied by the bond market. It is then up to each actuary to determine how much additional return they believe that their pension fund's investment strategy could earn over and above bond returns and this could then be used to justify setting aside a smaller amount of money in anticipation of higher returns than those offered by bonds. 

Again, at first glance, this may seem like no change at all.

11. We can still allow for higher returns if we take care to separate this allowance from the "market price" of our liabilities so that we are transparent about the quantum with which we are reducing our estimate of the liability due to the expectation of higher returns (than bonds) in the future.

Under the proposed approach by Exley, Mehta and Smith (and now the FSB) our assumption for future investment returns is made up of:

• the return on bonds as provided by the market plus
• an additional return from pursuing a riskier investment strategy as estimated by the actuary.

The difference to the past approach adopted by actuaries is that this process creates transparency and a clear attribution of the liability calculation - in particular, that part of the liability value that is a market related assessment and that part that is attributable to actuarial assumptions about future expected returns and that cannot be quantified with any certainty. The first component provides us with a way of directly linking the value of our pension fund liabilities to the investment markets price for a similar stream of cashflows. The second component explicitly quantifies by how much the trustees wish to reduce the quantum of assets they wish to set aside to meet their future pension obligations because they believe that their investment strategy offers them the scope to earn greater returns than the return on a portfolio of bonds - for the first time there is a separation of the market price for the liability value and that part of the calculation which is down to "actuarial judgement".

Now, unlike before, as bond prices (and hence expected returns in bond markets) change so too would our assumed investment return. This would mean that our liability value was being continously updated for new market information about the price of a future stream of defined cashflows.

Importantly, in deriving the second component using a carefully thought through assessment of the additional return above bonds that the actuary felt the investment strategy could earn over the longer-term, it would be very difficult for the actuary to justify a change in his assumption without sound justification and reasoning.

We can illustrate the above point with an example, Let us assume that an actuary decides that as a result of pursuing a higher yielding investment strategy he can use an additional return of 4% above the return on bonds being used to value the pension fund's liabilities. This figure of 4% would be arrived at after careful consideration of the long-term additional returns offered by other asset classes relative to the bonds being used to value of the pension fund's liabilities. This is an important point and deserves further consideration.

By studying long-term returns on bonds being used to value the liabilities we will naturally consider a variety of market cycles; both when bond yields are low and and high.

By expressing our "additional return" assumption relative to bond returns achieved over different market cycles we are recognising that whilst the absolute level of yields on different investments can vary depending on the market cycle, the return of different asset types relative to each other should display far greater stability. We may well expect that in times when interest rates are high for bond yields to be high and vice versa but the spread of returns of other assets (equities, property etc) relative to those higher (or lower) bond yields can be expected to be more stable.

We would therefore not expect this spread to change as a result of movements in bond yields, except of course if bond yields changed very dramatically so as to call into question the long-term analysis previously carried out.

Returning to our example of using a 4% additional return to assess our liabilities, then

• If, at a point in time, the bond returns we were using to value our liabilities were 9% p.a., the investment return assumption used to value the liabilities would be 9% + 4% = 13%.
• If, one year later, those bond yields fall to 8%, then mechanically the new process, for setting the investment assumption to be used to value the pensioner liabilities, would result in the return assumption falling to 8% + 4% = 12% from 13%. 
• The assumption that we can now only earn 12% (rather than 13%) would see the value that we place on our liabilities increase. 
• Importantly, we can now clearly see that the increase in liability value is due to a change in the market's assessment of the present value of those cashflows but our assessment of the additional return we expect our adopted strategy to deliver has not changed. 
• The actuary could of course change his assumption of the additional return to be earned from 4% to 5% to compensate for this change but it is debatable to what extent he could justify that long-term analysis that he previously did was not invalidated by a fall in yields of just 1% p.a.

In summary:

• Liability values are split in two parts - i) the value placed on them by the bond market (in our example above, this is assessed through the 9% return assumption) and ii) the amount that the actuary assumes will be delivered by higher investment returns than the bond market can deliver ( this component of the liability is assessed through the 4% additional return assumption).
• The assumed return used in the first component is transparent and can be quickly and easily assessed by looking at bond market returns which are freely and readily available.
• The assumed additional return used in the second component of the liability valuation is based on the pension fund's asset mix and assumed returns for different asset types. But this assumption should be rigourously set and would not be expected to change frequently.
• Liability values will now mechanically change in response to changes in bond market returns and also changes in the asset mix as well as investment return assumptions however any changes due to the latter will, unlike before, be transparent and should be justified.

11. Are we there yet? Yes, it seems like starting in 2015, with the FSB proposals we will see the uniform adoption of the above principles to valuing pension fund liabilities by SA pension funds...

South African pension funds have been far slower to fully adopt all the principles espoused by Exley, Mehta and Smith. 

Almost 18 years post the publication of Exley, Mehta and Smith's paper it seems like 2015 may be the year finally achieve uniformity and consistency and that with the FSB's latest proposal the stage is now set for South African pension funds to embark on the road to convergence where all pension fund liabilities will be valued in a way that is consistent with pricing implied by the bond market for those cashflows and for any margins due to higher expected investment returns to be separately quantified and justified.

12. ...creating more volatility in a typical pension fund's financial position...

A pension fund's financial soundness assessment must be explicitly related to bond market returns. These returns change daily. As we have shown above, if these bond yields fall then the liability value will increase. An increase in the liability value (without a corresponding increase in the value of assets) will result in the financial position of the pension fund deteriorating. Again, this makes sense to us in that if future returns that can be earned on monies set aside to pay future pension are lower, then more money must be set aside today to be sure of being able to meet all future obligations.

In particular, the FSB's proposals call for the investment return assumption to be set in a way similar to that proposed by Exley et al. The proposal goes further and expects any assumption used to be realistic and guided by past experience or reasonable expectation of changes in the future.

Now, as bond market yields change so too will the liability value of  the pension fund.

13. ...and more volatility for the companies sponsoring those pension funds...

This greater volatility should be seen together with the further requirement that where the funding position declines to below 95% then the pension fund trustees must submit a plan to the FSB for restoring the funding level to at least 100% within 3 years.

There are only two ways to restore the funding level:

-  to take more investment risk in search of higher returns on the money that has already been set aside or

- for the sponsoring company to commit to setting aside more money. 

The second option is likely to be unpalatable to many companies since it introduces uncertainty into their financial planning - an uncertainty which is linked to the vagaries of investment markets that are mostly beyond their control.

14. ...leading to a greater need to consider measures to mitigate the impact of the FSB's proposal.

One consequence will be that sponsors are increasingly likely to find that their existing investment strategies will not help protect them against volatile liability values which in turn could require them to cough up more money for their pension funds, potentially to the detriment of other business opportunities. 

Investment strategies that offer protection against volatile liability values will be those that invest in assets whose value rises in response to declines in bond returns. These strategies are know as Liability Driven Investment (LDI) strategies. 

But there is no free lunch; unfortunately LDI strategies can be lower yielding but the good news is that they need not be.

15. Pension funds should act but also be careful not to throw out the baby with the bath water...

South African pension funds find themselves in an enviable position of having healthy funding levels, Trustees would be wise to move now to protect those positions but should do so judiciously and without panic. Moving to a low yielding LDI strategy without considering the alternatives could jeopardise all the good work of the past few years and condemn pensioners to years of poor pensions -with no prospect of inflation-beating increases.

But this need not be the case. UK pension funds have led the way by implementing LDI strategies without compromising on their ability to pursue higher returns. South African pension funds would be wise to do the same. Before committing to a low-yielding LDI strategy, trustees and their advisers simply must explore opportunities for yield enhancement.  

But in doing so they should consider whether the LDI manager they are talking to is best placed to deal with another bit of regulatory action on the horizon - the FSB's draft proposal on regulating many of the investment instruments used by investment managers offering LDI strategies. That however is a subject for an entirely different blog.

Note: some pension funds have already embraced the principles of Exley et al and these should see little change as a result of the FSB's proposals.

Tweet