One weakness of the study is that settlements and fees are more likely to be reported in federal class actions. If, as tort reformers argue, state courts are more prone to excessive settlements, an upward trend might not be apparent in the available data.

Even if inflation-adjusted settlements and fees have not risen on average during the last decade, the total cost of class actions is going up because more are being filed. The number of federal class actions doubled between 1997 and 2002, and Federalist Society surveys of corporations indicate that the number of state class actions also has risen dramatically in recent years.

Another limitation of Eisenberg and Miller's study is that averaging results

across states tends to conceal problems with particular jurisdictions. One of the most notorious is Madison County, Ill., which sees more class actions per capita than any other county in the United States and had a record number in 2003. Madison County made news last year with a $10.1 billion award in a tobacco class action, accompanied by lawyers' fees of $1.8 billion.

Eisenberg and Miller found that "the amount of client recovery is overwhelmingly the most important determinant of the attorneys' fee award." That may seem fair, but it's not when damages are absurdly high, or when a settlement is a payoff to make the lawsuit go away rather than appropriate compensation for real injuries. Basing fees on the total "client recovery" is also unreasonable when the settlement is overvalued (as it probably will be when it consists of coupons that might not be used) or when the amount received by each plaintiff is negligible.

I'm not sure whether fee regulation is the right solution, but I thought I might as well put in my 73 cents' worth.