Finite reinsurance: a strange and sometimes manipulative thing

Thanks to AIG, the weird and wonderful world of finite reinsurance has come under broader scrutiny recently. (You may recall that a finite reinsurance policy between AIG and Gen Re was the method used to inflate AIG's earnings in the case that came to the courts in 2008.)

Now, thanks to the Big Picture, further amusing documents have achieved more general publication. I don't agree with much of the thrust of the post - which frankly contains altogether too much credit derivatives related hysteria. But the extra light on finite reinsurance is welcome.

When is finite reinsurance a valid business tool, and when does it verge on fraud? This is difficult to answer because finite reinsurance is a very sophisticated tool that can be used in myriad ways. But let me illustrate a good and a bad situation.

Good finite reinsurance. Suppose a company has a liability with a known size but uncertain timing. Asbestos-related claims are a commonly cited example: the firms knows it will have to pay workers for past exposure to asbestos, and it can estimate the size of those claims reasonably well, but it does not know when the claims will be presented as the sickness has a long and uncertain gestation period. The uncertainy thus created weighs on the share price, even though the company has every intention of paying and the resources to do so. Therefore it purchases a finite reinsurance policy whereby it pays a premium equal to (roughly) the present value of the expected claims to a large, well capitalised reinsurer. The reinsurer takes two risks: one small (that the claims will be larger than expected: this is unlikely as typically the risks insured under finite schemes have rather little uncertainty in claims); and one larger (that the claims will be presented earlier than expected, and hence the invested premium will not have grown sufficiently for them to make a profit). From this we see that finite schemes are often about transferring timing or investment risk rather than the risk of uncertainty in claims.

Bad finite reinsurance. Consider the effect of the scheme above though. Before the reinsurance, the firm had a known hit to earnings in the future but with uncertain timing. Afterwards, it has a stream of expenses - the premium payment or payments on the policy - but no uncertainty. Earnings have been smoothed. Clearly we can extend that effect more broadly via policies which pay out money in the future for an appearance of risk reduction today (buying surplus for an insurer, i.e. flattering their capital position) but where all of the risk comes back in later years, or via policies which move current profits into later years, smoothing earnings. Accounting rules do not permit you to arbitrarily reserve whatever amount of current earnings you like against some future risk, especially a very unlikely and hard to quantify one, but finite reinsurance policies achieve the same effect.

Finite reinsurance can therefore be used, quasi-legally, to manipulate earnings for many companies. It can also be used to manipulate insurance companies' capital position. If ever there was an area of finance crying out for better regulation, I'd say it was insurance.