A second theory, which most parties agree holds some validity, but which is nevertheless inadequate to explain the magnitude of the present crisis, is based on the underwriting cycle.

The theory of the underwriting cycle, simply put, is that insurers engage in price competition which causes falling premiums. Because insurers often don’t know whether a particular price was correct until years later, when all claims resulting from that policy have finally been settled, it is easy to "guess" incorrectly, and continue price cutting until premiums become inadequate. Later, when the losses roll in, prices are raised to reflect the true, nature of the risk. Income rises, and new capital enters the market to buy up insurance stock, undervalued from prior losses. The companies return to financial health, and a new round of price cutting begins.

The particular severity of the cycle in this decade is explained by the extremely high interest rates of the 1979 to 1981 period. During this time, insurance companies cut rates in search of market share, and used the extraordinarily high interest income to cover losses from claims. When interest rates plummeted later in the decade, the insurers were left with thousands of new policies written at inadequate premiums. Insurers accused of irresponsible "price gouging" under the collusion theory are accused here of irresponsible competition.

This theory has some validity, and to the extent that interest rates have stabilized and policies been rewritten at new, higher premiums, it may help to explain what seems to be a gradual easing of the insurance crisis since its 1986 peak.

Yet if the problem was nothing but the underwriting cycle, exacerbated by insurance price cutting aimed at capturing premium dollars to invest at high interest rates, why have its effects been concentrated in a handful of liability lines? Why has there been no corresponding crisis in automobile insurance, [12] fire insurance, or at least in other "long-tail" lines such as workers’ compensation insurance? There must be something more.