Bobby L. Dexter (Chapman) has published Rethinking "Insurance," Especially After AIG, 87 Denv. U. L. Rev. 59 (2009).  Here is the Conclusion:

In light of general judicial rejection and scholarly criticism, the Service officially abandoned its economic family theory in the captive insurance company arena in 2001. As promised, however, the Service continues to analyze certain captive insurance transactions to ascertain whether they incorporate risk shifting and risk distributing, despite the fact that courts have questioned the legitimacy of a risk shifting inquiry and thereby weakened its determinative force. Commentators do not criticize the risk distribution requirement as inherently useless, but they do accurately note that the standard can be used to demarcate true “insurance” only in those contexts in which reliable probability data exists with respect to the designated risk pool.

To the extent the Service or any other entity focuses only on the mere existence of a large number of apparently independent, homogenous risk units in the designated pool (with very little or no regard for predictive data with respect to that pool), latent investment activity may successfully cloak itself in the garb of “insurance.” AIG’s so-called “debt insurance” on credit default swaps has brought home that truth and, given the concentration of risk within behemoth AIG, that truth arrived with resounding and devastating force.

Contextual self-insurance represents a healthy and viable means of de-concentrating insurable risk generally, and for companies deemed “too big to fail”—or those operating in certain industries—limited contingency reserving is far more palatable than multi-billion-dollar bailouts and the attendant enhancement of corporate agency costs. Indeed, both contextual self-insuring and limited contingency reserving promise to maximize the common good, whether measured on the national or individual community scale.