Kevin Dowd 3 July 2019
Or maybe a couple of things. Still working on the baffling case of the logic of the Matching Adjustment, and I wonder if even Columbo could solve this one.
Let’s sweep aside the objections of the Doubting Thomases (here, here, here etc.) that MA is theoretically flawed. Perhaps IFoA president-elect John Taylor is right when he claimed that MA is not only a Fundamental Actuarial Principle but sound too.
But then I struggle with implementation. You take your credit spread, which is known. You decompose the spread into the Fundamental Spread, the risky bit, and the Matching Adjustment, which is the non-risky bit. I still don’t quite understand the bit about some of the spread above risk-free being risk-free, but that might be me so let’s move on.
Ah yes, you say. Implementation is easy. You just apply the regulatory formula.
Well of course! The regulatory formula says, in essence, that for a bond of a particular class (sector, credit rating, whatever) the regulator ‘knows’ the breakdown between the FS and the MA. I can accept that too, but the epistemologist in me still worries about how the regulator could know these things. How could any regulator know some quantitative feature of an ever changing empirical reality that eludes everyone else?
So presumably we must infer that regulators have access to some Hermetic Muse who speaks to them in secret to give them knowledge that mere humans cannot discover. I know that sounds a little far-fetched, but what else could it be? Eliminate all the other possibilities and whatever is left, however seemingly implausible, must be the answer. Consider the alternative: if regulators don’t have access to some hidden source of knowledge – of course, it doesn’t have to be Hermetic, it could equally well be the Egyptian God Thoth, the Fount of All Knowledge and curiously the inventor of both astronomy and astrology – then they couldn’t possibly be sure that they had the correct Matching Adjustment calibrations.
But then another thought pops into my head. The markets have lots of smart people who are always on the look-out for profitable empirical relationships. If you can identify the MA (and remember, the regulators tell them what the MA is, so the markets don’t have to work it out; they don’t even have to make sacrifices to some bird-brained idol; they can just use the divinely inspired regulatory formulas), you can establish, with certainty, two risk-free rates and then arbitrage them. You synthesise the risk-free bond with the higher risk-free rate, go long on that bond and short on the risk-free bond with the lower risk-free rate. You can then make a ton of money.
I can follow all this argument, but here is the bit that baffles me. If all this stuff happens, then wouldn’t these arbitrage activities reduce the gap between the two risk-free rates back to zero? And if that is the case, wouldn’t we be back again to just the one risk-free rate? And if there is only one risk-free rate, how can any of the spread above risk-free also be risk-free? And if that is correct, then how can the true MA be anything higher than zero?
So I end up going round in circles. Even if I assume that the MA is as we are told, it just keeps getting arbitraged away. But surely that can’t possibly be right, because otherwise the regulators would be wrong.
Only Thoth knows and he isn’t telling.