The muddled thinking at the heart of the UK pensions debacle, by Publius
The Bank of England announced today that its £65 billion purchases of UK government bonds (known as gilts) in the previous week had stabilized the UK gilts market which, though no longer central to the global economy, can still cause substantial ripples through it. In today’s FT, legendary investor Sir Terry Smith tried to explain the essential fallacy underpinning what caused the positive feedback loop that threatened to destabilise that market last week and prompted the Bank to intervene. Writing for a general readership, he got his basic point across while skipping lightly over some of the reasoning behind it unless you understood it already. If you didn’t, here goes.
In many countries, investment vehicles, including pension funds, are required by law to value their assets at market prices, which are usually pretty easy to observe: Just look them up on the internet. These funds hold assets, of course, to pay liabilities down the road, sometimes in the far future, to customers who will one day draw their pensions from the savings they have built up at these funds. These liabilities are harder to value, but also have to be ‘marked to market’ to the extent possible. When interest rates are low, as they have been for well over a decade, asset prices tend to be high but liability valuations tend to be even higher. As a result, this marking to market exercise has usually implied that pension funds and similar investment vehicles are under water.
So far, so sensible. Marking to market is a rule designed to prevent rogues and rotters from claiming their investments are more successful than they actually were. The mischief comes when regulators and pension fund trustees and managers actually believe the results of such exercises, or perhaps over-interpret them, and take steps to correct a problem which doesn’t really exist, thereby making the problem much worse.
Here is a very simple example. Suppose I promise to pay you £1 million in twenty years’ time. The UK 20 year government bond yield is 4.288% today. But the UK 20-year bond pays intermediate cash payments, or coupons, every 6 months, so that number is a bit misleading. What I need is the yield on a bond that pays £1 in 20 years and nothing at all between now and then. This ‘zero-coupon’ yield is 4.525%. If you pay me £412,664 today, and I invest it in a 20-year UK government zero-coupon bond at 4.525% per annum, then by the magic of compounding, the sum owed to me at the end of the life of that bond will be £412,644 times 1.04525 to the power of 20, in other words exactly £1 million.
Let’s suppose I sell you this simple savings product. If I fail to pay you, your heirs or assigns the £1 million I owe you in 20 years, you can sue me. However, I’m not worried, because I know there are almost no conceivable circumstances in which I will be unable to pay you. The UK government issues its own currency and will never therefore be in a position where it cannot or will not pay me.
Suppose I wake up next Monday and check my phone for the latest news. I see that the interest rate on 20-year zero coupon bonds has risen to 5%, because these rates are set by supply and demand in what is usually a very liquid and orderly market. The market price of my 20-year zero-coupon gilt is now £1 million divided by 1.05 to the power of 20, or £376,889. I am sitting on a paper loss on the asset I bought a week before of about £36,000 or about 8.7% of my (i.e. your) original investment.
Do I panic? Not if I’m sensible. I know that my investment is locked in at 4.525%, whatever happens to the market, and therefore I absolutely will be able to pay you the £1 million in 20 years which I’ve promised. And of course in this simple example, the present value of my liability to you has fallen by exactly the same amount as my asset, so my deficit is still zero, because the correct rate at which to discount that promise I made to you is the zero-coupon risk-free rate, now 5%. So, I do nothing, for 20 years, and no-one need worry.
An observer looking at my balance sheet every day would conclude that I had bought a very volatile asset, one whose price fluctuates a lot from week to week. He would be right, but it doesn’t matter. For the purpose for which I bought it, my 20-year zero coupon bond is the one that completely eliminates risk.
So where does the mischief come in? Let’s suppose that instead of investing your premium in a zero-coupon bond, I invest in a broadly (i.e. globally) diversified portfolio of bonds, stocks and commodities. As I wrote two weeks ago there is a very high probability that such a portfolio will beat the UK bond handsomely over 20 years, delivering returns in excess of 2-6% over our benchmark bond yield of 4.525%. Let’s be highly conservative and pencil in 2%, so now you only have to pay me about £297,000 to have an excellent chance of your £1 million, with a really good chance of a much higher payout in the range of £1.5 to £2 million.
Again, suppose I wake up a week later to discover that it’s 2000, 2008, March 2020 or 2022 all over again and the value of your pension fund asset is valued by the market at way below the value of the zero-coupon bond that would have funded your £1 million. Because liabilities have to be discounted at the appropriate risk-free rate (the liability is certain to happen, after all), then my pension scheme is in deficit.
And here is the mischief: If I panic, because I’m muddled or the regulator is muddled, I might decide to sell some of your portfolio and put the proceeds into the zero-coupon bond. I am selling at a bad time, because the assets I invested in are all lower in value. Now you are much less likely to receive your £1 million in 20 years. Curiously, if I do nothing, you will almost certainly get the higher payouts originally promised.
Why is that? Because stock and bond prices more generally, not just UK zero-coupon bonds, obey the same laws of value: When discount rates go up, prices tend to fall. But expected returns on those same assets are usually higher than they were last week, so you’ll be fine. Sometimes, it’s true, bad news can be of the permanent variety, but at the broad level of stock market indices those times are incredibly rare: in fact in the very long-run, like 20 years to 100 years, they appear to be non-existent. This extraordinary fact was first properly documented by Professor Robert Shiller of Yale University in 1982, for which he richly deserved his 2013 Nobel Memorial Prize in Economics: Stock market index returns, year on year, can be incredibly volatile, as can annual returns on long-term bonds. Nevertheless, long-term returns over decades on these assets don’t seem to be volatile at all: Companies still make profits, commodity prices revert to the mean, and broadly diversified portfolios of so-called risky assets just don’t seem to be very risky.
The explanation for this excess volatility puzzle is not yet fully established, perhaps may never be. That’s not the point. The point is that pension funds that take action to eliminate deficits that arise solely because of marking to market are doing the wrong thing, as are funds that pay to ‘hedge’ mark-to-market swings in their deficits. Last week, it emerged that several large pension funds were doing exactly that, and because of the hedging contracts they had entered into were obliged to sell valuable assets into a bear market in order to fund their margin payments on the insurance products they shouldn’t have bought in the first place. The managers of those funds, their advisors, and the regulators who made them do it should all be fired: At best, they don’t understand what they are doing.