Lawrence J. McQuillan and Hovannes Abramyan respond in detail point-by-point to Richard Posner's critique of the PRI "Jackpot Justice" study.
Their main objection is that Posner repeatedly characterized their report as a measure of the costs of the liability system, rather than that of the costs of excessive liability, but the original study can be at least partially blamed for that confusion given that some elements of the $865 billion figure in press release headlines were obtained by measuring total costs. But McQuillan and Abramyan do take down a number of Posner statements that misrepresent study findings.
I still take issue with PRI's use of Rubin & Shepherd. Some tort reforms, such as noneconomic damages caps, improve social welfare by reducing randomness and increasing accuracy in the litigation system. Other reforms, such as collateral source offsets or economic damages caps, operate as wealth-transfers to tortfeasors, and can be expected to decrease deterrence arbitrarily. And indeed, that is what Rubin & Shepherd (and several other studies, like Klick & Stratmann's work on infant mortality) find. All in all, the early version of the Rubin/Shepherd study found a loss of 2700 lives in 2004 vis-a-vis a baseline of a tort system without legislative constraints. But the total loss of lives attributable to "excess liability" is higher than that number, because the Rubin/Shepherd number is dampened by the number of lives cost by inefficient liability reductions. If every state adopted collateral-source reform, that wouldn't be an argument against the benefits of non-economic damages caps just because the Rubin-Shepherd number dipped below 2700 or even if it dipped below zero. (For example, imagine a highly inefficient reform that made it a felony to bring an unsuccessful malpractice suit. There's little question that such a law would cost lives; the numbers might even swamp the 2700-life figure in Rubin/Shepherd. But that the Rubin/Shepherd methodology would then conclude that the combination of efficient and inefficient reforms cost lives on balance would not be an argument against the efficient reforms.)
On the other hand, however, the PRI study measures the cumulative effect on the workforce of deaths over the last 24 years. This is another place where their study's measurements are inconsistent with measurements elsewhere in the study. Why measure cumulative loss of workforce when one is not measuring cumulative loss of innovation? In some places, the study is measuring the marginal cost of a broken liability system; in other places, the study is measuring the cumulative cost of decades of damage caused by the liability system. One measure or the other is alright, as long as it is made clear that this is what is being estimated, but it is problematic to add the two calculations in the same number.