Wednesday, RiskProf posted an interesting discussion on the zone of fuzziness around what it means when a government-run program is required to have actuarially sound rates.
His concern:
The national flood insurance plan is supposed to have prices that are actuarially sound-yet it has a large deficit. If there was truly actuarial pricing there would be little or no deficit. In addition, a number of state plans are operating at a significant deficit. In fact, according to some figures I just saw but can’t quote, a major insurer is claiming that there is about $5.5 billion of losses in excess of premiums in state wind pools. This is a massive subsidy to homeowners on the coasts.
A somewhat flip response would be to point out the traditional standard of “adequate, not excessive, and not unfairly discriminatory”…and to speculate that in certain programs there is an inherent tendency to focus on the “not excessive” portion of the mantra.
But that’s not why I’m writing this post. 
What actually caught my eye in RiskProf’s post was the following comment:
What is missing from Section 2(5) is the loading for the risk the firm takes by committing capital to the market. However, shouldn’t the government account for the risk in some way to recognize the fact it is creating a potential tax payer liability? That would be actuarial pricing.
I want to say that the answer to that question is “yes, but…”
Oversimplifying a bit, you could say that there are three or four major components of a price:
- The loss cost — the best estimate of what the expected future loss is for the policy
- An expense component — to reflect a need to pay the actuaries, provide commissions to the producers, and to keep the lights on in the home office
- A profit load — to reflect fair compensation to the capital providers/investors/owners for the use of their money to capitalize the risk; and
- A margin element — to reflect assorted market inefficiencies or other constraints.
In this scheme, “risk” would be a component of the profit load. The return demanded by investors is (again, oversimplifying) a function of three main components:
- The expected “risk free” rate of return;
- Compensation for the parameter risk in the pricing (the risk that the models are wrong — i.e., how much variability can there be in the return on investment due to making a mistake in calculating the loss cost or expenses); and
- Compensation for the process risk associated with the risk — how much can the return on investment vary just because of pure randomness.
The process risk is the source of my discomfort with RiskProf’s assertion, largely because it’s something I have had to think about quite a bit in the past couple of years.
The process risk component will vary from insurer to insurer depending in part on the portfolio of risks they write. When you look at the act of deciding whether to write a new block of business, you should be considering the incremental change to your overall risk brought about by writing that additional business.
Let’s say you’re a property insurer who writes business only in North Dakota. For some reason you’re considering writing an “ordinary” risk in South Florida. I submit that it is theoretically cheaper for you to write that risk than it would be for a competing insurer who already has a book of business dominated by a large presence in that part of the world.
If you have a loss on that one policy, yes, it will hurt, but it will likely barely register in your overall results, and it is unlikely to be correlated with the results on your overall portfolio of business, since events in Florida tend to bear very little relationship to those in Florida. That one risk won’t cause a material blip in your results, and therefore investors (or your cat treaty reinsurer) shouldn’t demand much in the way of additional return due to this lack of incremental volatility
However, for that competing Florida-heavy business, adding that risk will likely have a direct impact on how big a big loss will be. That risk will be highly correlated with the other risks insured by the company. Plus, there could be a potential that you need to acquire additional capital to support the incremental risk (lest the ratings agencies be made unhappy)….
So, at least from a pure cost-of-process-risk basis, that incremental Florida risk is cheaper for you, a North Dakota-heavy insurer, than it is for your Florida-dominated competitor.
(Yes, I know there are some infrastructure costs that, in reality, would complicate the comparison. However, that’s not actually relevant to this discussion, and therefore I’m going to ignore that very valid point.)
So…if that one Florida risk is cheaper for an ND-focused insurer as compared to a FL-heavy insurer, what happens when you throw (say) the federal government into the mix?
Considering just how many fiscally-uncertain matters the feds play in, I’d bet that the variability associated with that one risk…even several hundred identical such risks, is still incredibly uncorrelated with the other risks borne by the government. For example, how is the cost of supporting the war in Iraq (or the cost of extricating ourselves from it) correlated to the risk of the next Katrina striking this year?
If the federal government should seek compensation for the process risk assumed by writing insurance business, I would argue that the appropriate compensation would be lower than it would with a private insurer.
Even the idea that a federally-underwritten insurance program needs to charge for process risk is one that I’m not sure I entirely agree with. Perhaps it’s something I haven’t thought through enough, but this sort of risk-based compensation for the assumption of risk seems in my mind to imply a profit motive…and I’m not sure that the government should be considered a for-profit operation, even in an ideal world where recurring budget deficits and the national debt were not significant concerns.
Aside from building and maintaining appropriate reserves for future downturns, presumably any profit made by the federal government would be returned to the citizens, in the form of reduced taxes or increased other services (the latter being more likely than the former).
However with all that said, I would generally agree that
- A federal government run insurance program needs to have some acknowledgement of a pool of resource that can be drawn upon in the event of catastrophe;
- A state-run program would be exposed to significantly more process risk than a federally-run program; and
- When politicians are involved, inefficiencies tend to rise, and the “not excessive” rule-of-thumb is likely to dominate the “adequate” principle.
One last tangentally-related thought:
I realize that I have reveled a little bit in the hype of media coverage about various insurers pulling back from the coasts. However, I’m not sure that such changes are necessarily bad signs. Ideally, the market should view these as opportunities.
If it has become too expensive for a couple of large insurers to feel comfortable writing in certain areas, doesn’t that open the door a bit wider for other insurers for whom risk is comparatively cheaper?
All it takes is time for those insurers to see the opportunity…or for the big dogs to potentially realize that such risk isn’t as expensive as they once thought.
Risk pools and emergency regulations arguably have a place in offering consumers/taxpayers some stability while the market seeks its new equilibrium. However, left to its own devices, given time and as little government interference as possible, the free market will eventually find that equilibrium, and the market will work on its own.