I just spotted a comment from ‘JKingdom3’ on our letter to the FT last month (Capital created by matching adjustment is entirely artificial, see Eumaeus “Our Reply to Rothesay“, 5 November 2019).
Kingdom wonders whether we have considered every angle, arguing that “In a world where firms seek to maximise their profit subject to the constraints they face, the “correct” assets required to meet this will be the assets corresponding to the least cost to the shareholder”. He contrasts investing
1. … $74.41 in the risk-free asset, hold no capital over ten years, and pay the policyholder $100 in ten years’ time with certainty; and
2. … $67.56 in the risky asset, and hold the minimum capital needed to meet the 99.5% requirement each year over the ten years.
He suggests that option 2 is a good deal for policyholders, and a better deal for investors. Correct or not?
There are more than a few things wrong here. Taking the policyholder angle first, the probability of exhausting artificial capital is either what he says, i.e. around 1 in 200 years, or much higher than that. Suppose the former. Then there is something wrong with pensions pricing. As I put it to Justice Zacaroli here (p.10), how is it possible that the spread of an illiquid pension portfolio can be at best 40bp over gilts, when the spread of the illiquid assets duplicating the portfolio is 130bp over gilts? If the default risk of the assets really is as low as 1 in 200, why isn’t the pension portfolio (which is an asset for the pensioners) trading at a similar spread? How is that a good deal for pensioners?
Or suppose the former, i.e. that the default probability of the asset portfolio is higher, perhaps much higher, than 1 in 200. Which seems more plausible. After all, the 1 in 200 number has been made up by some actuary at some life insurance firm, after strong pressure from the board to minimise capital, and agreed to by some actuaries at the PRA, under similar pressures from the firm and from their senior management. Is it really plausible that we can earn a 90bp+ return over risk free over 20-30 years without any risk whatsoever? All the evidence suggests that consistent excess returns, while not impossible, are very difficult to earn systematically. Neil Woodford was the last one to persuade investors that it was possible, and look what happened there.
Then take the shareholder angle. Clearly it is attractive for shareholders to borrow (from pensioners) at close to risk free, lend at well over risk free, and receive a guaranteed profit up front, courtesy of the PRA. For current shareholders, that is. But current shareholders are not prospective shareholders. The matching adjustment business model gives you a profit for each portfolio that you get MA permission for. Once the MA game runs out, and that will eventually happen as all pension schemes end up in the hands of MA firms. But that is the end of it. Any future shareholder who buys the firm at MA value will have overpaid by as much as they have paid.
So have we considered every angle? Either the default probability set by some actuary is correct, or not. If correct, pensioners are being ripped off. If not correct, (prospective) shareholders are being ripped off. A nice game for existing shareholders, mainly private equity players. Not so nice for the rest of us, no?