First, let me offer my apologies for such a long time between posts. No Excuses!
If you have never stood in freezing cold water up to your waist in the wee hours of the morning waiting for smelt to run you are much more sane than I! I went to college in the Upper Peninsula of Michigan and smelt dipping is a tradition for the locals there. Every night, just after the ice melts on Lake Superior, small sardine-like fish called rainbow smelt run up into the streams connected to the lake in the millions (yes millions). When you stand in the middle of these streams in your waders you can literally feel them banging against your legs in waves.
You catch these fish by using a 5 gallon plastic barrel with the bottom cut out and replaced with screen mesh. The size of the gaps in the mesh is important because if its too small you catch too many fish that you can’t use. If its too big you only get a small number of the biggest fish. There is a dynamic that needs to balanced to catch enough fish to make it worthwhile to stand all night long in freezing water but not too many really small worthless fish. Now, here is the insurance part of all this….Insurance companies have to balance this “Smelt Dip Dynamic” as they make decisions about risk selection.
One of the most critical issues associated with the profitability of any insurance program, captive or traditional, is the selection of the participating risks in the group of insureds. How you set the criteria for selection determines the quality of risk and the number of risks that might qualify. This decision process has a lot of moving parts. Insurance theory has several, often competing, criteria that all come together to form the operational characteristics of an insurance company. On one hand the theory of large numbers says that you have to have a lot if risk units mixed together in order to have predictable loss results. On the other hand we know that some risk units just don’t have the loss control characteristics that make them good risks.
One approach to risk selection would be to accept only a very exclusive group made up of accounts representing the very best loss ratios. This might work if you have a large enough group of acceptable risks in the group. Beside the statistical shortcomings of a small group of insureds there are the fixed costs of an insurance program that must be covered as one of the expenses of the program. In order to cover costs and alleviate some of the statistical problems of a small group we would have to raise prices but then these excellent risks might be able to go somewhere else and get a better price.
Another approach would be to take all comers and build a very large risk group that would have very predictable statistical performance but because there were no criteria to entry would end up with some very bad players in the pool. The experience of the good loss performers would offset the cost of the poorer performers but the overall premiums would have to be higher to compensate for the bad risks. Again, good risk could find a cheaper deal and the bad risk wouldn’t be able to find a better price anywhere. The result, adverse selection.
Enter “The Smelt Dip Dynamic”. Imagine that the risk selection criteria is correlated to the size of the mesh in the bottom of our 5 gallon plastic barrel. It needs to be wide enough to let the very bad accounts flow through without being captured, but small enough to catch enough good fish to make the process worthwhile, both from a cost perspective and from a statistical perspective.
Captives differ from traditional insurance programs in how they determine how big to make the mesh. Because captive participants are typically larger accounts that take a significant share of frequency layer losses they already have a larger statistical base to predict losses on. Captives also have a lower expense load than a traditional insurer so there is less cost that has to be distributed in the program. The end result is that a captive can pick and choose the very best risk to be a participant in the program and still maintain a reasonable degree of statistical credibility.
In our analogy, captives are only after the biggest and best fish they can find and they set their underwriting criteria to ensure that happens.