On April 1, Judge Richard Posner attempted to refute the findings of PRI's study Jackpot Justice: The True Cost of America's Tort System on the Becker-Posner Blog. [See: Is the Tort System Costing the United States $865 Billion a Year?--Posner, April 1, 2007.]

Judge Posner's comments, while a serious attempt to refute the study on the merits of the report, were riddled with errors ranging from misreadings, false accusations, inaccurate citations, to untrue claims.

What follows is our response to Judge Posner's comments in a point-by-point format. It reveals what we believe to be poor reading/comprehension of our report and a clear misrepresentation of our methodology. Despite the claim made by Judge Posner, our results are far from ficticious.

Response to Judge Richard Posner on "Jackpot Justice"
By Lawrence J. McQuillan and Hovannes Abramyan


Is the Tort System Costing the United States $865 Billion a Year?--Posner

A study published last month, and favorably summarized in an op-ed in the Wall Street Journal, estimates that the American tort law system costs the nation $865 billion a year. The study, entitled Jackpot Justice: The Cost of America's Tort System, was written by Lawrence J. McQuillan and other members of the staff of the Pacific Research Institute, which published the study. (The study can be downloaded at www.pacificresearch.org.) How did the authors arrive at that figure, and is it meaningful?

They begin by estimating that the nominal (that is, dollar expenditures as distinct from social costs) annual cost of the tort system, consisting mainly of attorneys' fees and other costs of administering the system plus the amount of money paid to tort claimants in judgments and settlements, is $279 billion, of which $128 billion is the amount paid out to claimants.

[PRI:  This is untrue.  We did not say that this is the “annual cost of the tort system.”  Instead, this is the total direct U.S. tort cost expressed in 2006 dollars.]

The estimate comes from a report (U.S. Tort Costs: 2003 Update) by Tillinghast-Towers Perrin, a consulting firm for the insurance industry, with the report's estimate updated to 2006. It is impossible to determine from Tillinghast-Towers Perrin’s report what the sources for most of its data are, and so the figures I have quoted must be taken with a grain of salt; indeed, so far as I can tell, they may be completely unreliable. They are almost certainly exaggerated, given the financial connection between the firm and the insurance industry.

[PRI:  The Tillinghast study is the industry standard for measuring direct U.S. tort costs and is the most frequently cited source for this information by the media, legislators, insurance experts, and public-policy scholars.  The data that Tillinghast uses to measure direct U.S. tort costs come primarily from A.M. Best, which compiles composite financial data for the U.S. insurance industry.  Rather than being “completely unreliable,” these data are considered the gold standard because they are subject to audit and are reviewed by state insurance regulatory agencies.  Tillinghast’s unique experience as a consulting company for insurers also has given them inside knowledge and historical experience with the industry, allowing it to better make estimates of self-insurance tort costs.  Rather than being a liability, the “connection between the firm and the insurance industry” provides the company with unique insight and information that it draws on to make its calculations.  The charge that their numbers are “almost certainly exaggerated” is unsupported and, in our opinion, uncalled for.]

The authors of Jackpot Justice know the difference between a cost, which in economic terms is a reduction in the amount of valuable resources, and a transfer of wealth from one person to another that doesn't reduce the total amount of resources but merely redistributes them. The $128 billion figure is a transfer, not a cost. But as the authors point out, the opportunity to obtain a wealth transfer can generate a cost--a cost incurred to obtain the transfer (incurring costs to obtain a wealth transfer, when socially unproductive, economists call "rent-seeking"). They assume, without analysis or evidence, that the entire $128 billion is translated into a cost.

[PRI:  It is perfectly acceptable to measure the costs associated with a transfer separate from the benefits.  This is our approach.  The costs associated with an excise tax, for example, can be measured separate from the benefits resulting from the expenditure of the resulting tax revenues.  Some expenditures will yield net social benefits, while others will yield net social losses depending on how the money is spent and the incentives it creates that shape future behavior.  The same applies to tort transfers.

Pages 12-14 and endnotes 9-10 develop the theoretical framework underlying our empirical analysis and references scholarly literature.  We apply the standard, conventional assumption of “perfect rent dissipation” to arrive at our estimates.]

They further assume that 28 percent of the $128 billion transfer represents a deadweight cost, that is, a loss of value. They base this assumption on a study which found that increasing the corporate tax rate by $1 generates 28 cents in deadweight costs.

[PRI:  This is untrue.  The study, by Professor Dale Jorgenson of Harvard University, examines the increase in deadweight costs for each dollar of extra tax revenue, not tax rates.  The President’s Council of Economic Advisers also relied on this conservative 28-percent statistical estimate to measure the deadweight cost of the tort transfer system.  Our application of this number is not novel or radical.]

The basis of that finding was that a tax, like a monopoly markup, causes the taxpayer, like a consumer, to substitute for the taxed item or activity something that may cost society more to provide but looks cheaper because it's untaxed, or taxed at a lower rate. The authors of Jackpot Justice do not explain why a tort transfer would have the same effect. Of course the threat of tort liability might well alter the behavior of potential injurers--indeed, it is intended to do so--but it might alter that behavior in the direction of greater efficiency, by making potential injurers internalize accident costs. That is the objective of tort law, though imperfectly achieved. Without tort liability, firms would have weak incentives to invest in safety measures to benefit potential victims of the firms’ activities, unless the victims were either their employees or their customers.

[PRI:  The reason that tort transfers have the same effect as a tax is that they must be paid by someone, which increases the price of products, services, and inputs, resulting in fewer units sold, ceteris paribus, as substitutions take place, or they act as a tax on income, resulting in less output of other goods, ceteris paribus.  This is explained on pages 12-13 of the report.  The level of the tort tax varies, of course, by industry or goods sector.  We examined only the cost side of this transfer system.]

With the addition of 28 percent of $128 billion ($36 billion) to $128 billion in assumed rent-seeking expenditures, the authors jack up their estimate of the annual social cost of the tort system to $164 billion. To this they add another $36 billion, on the assumption (it seems---this part of the report is none too clear) that efforts to avoid the deadweight cost will cost that amount. This appears to be double counting, based on the assumption that $36 billion is spent every year in a futile effort to avoid a $36 billion cost.

[PRI:  This misrepresents and misunderstands our accounting methodology and the claim of “double counting” is untrue.  Each addition tallies a different cost component – one is the deadweight cost and the other is the rent-avoidance cost.  This is explained on page 14 of the report.]

It is possible, however, that some--maybe considerable--costs are being incurred to prevent that deadweight loss from rising from $36 billion to some higher level—for example, legal-counseling costs (not included in the attorneys' fees incurred in actual litigation) or costs in curtailed new-product development (see below). But there is no basis for supposing that the sum of such costs would equal $36 billion; nor do the authors make any effort to defend the figure or estimate the actual costs.

[PRI:  Again, this is untrue.  Pages 12-14 and endnotes 9-10 develop the theoretical framework underlying our empirical analysis and references scholarly literature.  In our analysis, we apply the standard, conventional assumption of “perfect rent dissipation” to arrive at our estimates.]

They add to their new total of $200 billion the $128 billion transfer, for a grand total of $328 billion. The addition is improper, since the transfer is not a cost. They are adding apples and oranges.

[PRI:  The assertion that we do not distinguish transfers from social costs is untrue.  We state throughout the study that we report two costs:  what we call “accounting costs” that include the transfers, and what we call “social costs” that do not include the transfers themselves.  We present the numbers both ways to allow for comparisons with the Tillinghast study (see endnote 11), which includes the transfers.  We took care to always delineate which cost we are referring to at any given point in the study.]

They borrow from another study an estimate that tort liability generates some 2,000 accidental deaths a year by discouraging the introduction of risk-reducing products and services. But they fail to offset against that figure (and its monetization) the number of accidental deaths that are prevented by the deterrent effect of tort liability. In the absence of liability, potential injurers would spend less on safety, at least with regard to potential victims with whom the injurers lack actual or potential contractual relations.

[PRI:  We calculate accidental deaths resulting from excessive liability that increases the cost of risk-reducing products and services.  The report that we borrow from determines the net lives saved as a result of tort reforms and does not assume an absence of liability.  Our annual estimate, therefore, accurately portrays a net loss of lives resulting from accidental deaths attributable to the current tort system.  Our report quantifies the costs of America’s tort system, and admittedly does not explore the benefits, of which there are many (see page 2); though a net positive number of lives saved at the margin does not appear to be one of them.  We encourage researchers to quantify the benefits.]

The authors of Jackpot Justice cite a respectable economic study by Daniel Kessler and Mark McClellan which finds that malpractice liability increases hospital costs by 5 to 9 percent; and they treat the entire amount as social waste. But as I said earlier, the aim of liability is to induce potential injurers to spend more on safety, and so the fact that they do spend more cannot be adjudged a failure to improve social welfare. Medical-malpractice law is in its administration rife with inefficiency, but it would be surprising if eliminating it entirely would be all social benefit and no social cost (nor in fact do the authors argue for eliminating it, as we’ll see).

[PRI:  It is untrue that Kessler and McClellan conclude malpractice liability increases costs by 5 to 9 percent.  Like the authors of the study on accidental deaths, Kessler and McClellan use liability-reducing tort reforms to determine the effect of excess liability on medical expenditures, and do not assume an absence of liability.  Kessler and McClellan conclude that these additional medical expenditures do not yield improved health outcomes thus they do not improve safety, but rather, are medically unnecessary wasteful expenditures driven by excess medical-liability concerns.]

The authors argue that products-liability law is responsible for the loss of $359 billion in new products.

[PRI:  This is untrue in two ways.  Our number reported in the study is actually $367 billion.  And we argue, based on the scholarly literature, that excessive liability is responsible for the loss of new products, not liability law in general.]

They base this on a study by the economists Kip Viscusi and Michael Moore which found that when the ratio of liability costs to sales revenues exceeds 5 percent, firms reduce their investment in R & D. The authors of Jackpot Justice identify product markets such as power tools, fireworks, and cigarette lighters where they believe (relying however on the questionable data in the Tillinghast-Tower Perrin reports) that this ratio is exceeded.

[PRI:  Several things about this statement are untrue.  We did not identify the product markets; rather, the markets were identified for us by Viscusi and Moore and Frederick T. Stocker (see pages 23-28 and endnotes 36-37).  Also, the data on liability burdens came from these researchers, not Tillinghast-Towers Perrin.]

Then, using Viscusi and Moore's estimate that 6 percent of the products of the industries that the two economists studied are new products, Jackpot Justice multiplies the output of these markets by 6 percent; with certain other adjustments, this calculation produces the estimate of $359 billion in lost sales.

[PRI:  This is a misrepresentation of the procedure we used to calculate our numbers.  We encourage those interested to read what we wrote in the study since the actual procedure is different and more nuanced than represented above.]

This is not, however, as the authors believe, a social cost. The social cost is the consumer surplus that the sales of the new products would have produced. Suppose that a product costs $10 and is sold in a competitive market for $10, but that consumers would pay $12 for it if it were priced at $12. Then if the product is not produced, there is a loss of consumer surplus of $2. That, not the $10 in lost sales revenue, is the social cost of not producing the product.

[PRI:  It is untrue that we do not understand what a social cost is.  But the example above was cleverly constructed to make us look bad.  If consumers would pay $10 for the product and it was sold for $5, then the value of lost sales revenue ($5) equals the loss of consumer surplus ($5).  We assumed that this is the case, on average, to use the value of output as a proxy for the social cost.

For the dynamic cost components, i.e., accidental deaths, health care expenditures, reduced access to health care, and lost sales of new products from less innovation, we used the value of output (either lost output or unnecessary output, depending on the cost component) as our common unit of measurement or yardstick.  We relied on U.S. Bureau of Economic Analysis data for the average value of output per employee.  We chose this yardstick because it is readily available and it allows for reliable calculations across different occupations and industries, which was necessary for a macro study of this kind.  We welcome suggestions from those who might have ideas on how to refine this approach/metric for these estimates.]

The sum of $328 billion and $359 billion is $687 billion, which is almost $200 billion short of the authors' grand total of $865 billion. The excess malpractice costs and accidental-death costs they estimate at less than $50 billion, so there is still a big gap. I can't figure out how they fill it.

[PRI:  All the numbers are reported in Table 4, page 28, and add up properly.  Again, the $359 billion stated above is actually $367 billion.  The “excess malpractice costs and accidental-death costs” are actually more than $170 billion in total, as reported in Table 4, not “less than $50 billion” – so, again, another untruth.]

So far in the report, there is nothing about the benefits of the tort system.

[PRI:  Again, our report quantifies the costs of America’s tort system, and admittedly does not explore the benefits, of which there are many (see page 2).  We make this point very clear and encourage researchers to quantify the benefits.]

To estimate those benefits, the authors compare the percentage of U.S. GDP that is accounted for by our tort system with the percentage of GDP accounted for by the tort systems of other developed countries. They base all these percentages on the dubious Tillinghast-Towers Perrin report. The U.S. percentage is estimated at 2.2 percent, twice Germany's and roughly three times Japan's and the United Kingdom's. The average for the foreign countries in the comparison is 0.9 percent, so the authors of Jackpot Justice conclude that the benefits of the U.S. tort system are equal to only 0.9 percent of our GDP. The possibility that our more costly system might generate greater benefits (though not necessarily equal to the greater costs) is ignored. But a more serious weakness is the implicit assumption that a tort system generates benefits exactly equal to its costs.

It might generate much greater benefits. Politics, which the authors assume lead to greater than optimal liability, might instead lead, in the countries they compare to the United States, to less than optimal liability. And even if investment in the U.S. tort system has been carried to the point at which the last dollar spent on the system generates exactly one dollar (no more) in benefits, the total costs may be far below the total benefits because average cost may be much lower than marginal cost. This is apart from the possibility that politics may have prevented our investing enough in the tort system to equate benefits and costs at the margin.

[PRI:  We present two ways to determine excess costs (page 31).  As we discussed in the report, we were admittedly forced to use a second-best approach due to a lack of scholarly studies on this subject.  Our approach uses a cross-country comparison.

Our cross-country approach assumes that the average direct tort cost ratio for our economic rivals with a comparable standard of living is a good approximation for what optimal direct tort costs as a percentage of GDP ought to be in the United States.  It is untrue that we assume that benefits equal these costs.  Nowhere do we say this.  We do not use the cross-country approach to estimate the benefits of the current American tort system.  As we make clear, quantifying the benefits is not our objective.  We do not make the assumption that “a tort system generates benefits exactly equal to its costs.”  Nowhere do we state this.  As GDP and trade grow, we expect tort costs to increase even if they remain the same percentage of GDP.]

The authors' estimate of the benefits (= costs) of the average foreign tort system, when subtracted from the $865 billion "cost" of our system, results (with some further adjustments) in an estimate of an annual excess of costs over benefits of almost $600 billion. The figure, however--the authors' estimate of the net social loss created by our tort system--is, as I have tried to show, fictitious.

[PRI:  Once again, we do not equate benefits with costs.  Nowhere do we say this.  That claim is untrue.  Furthermore, in our calculations, we do not subtract from $865 billion; rather, we subtract from $737 billion, the total social cost (see Table 4), since we understand that the transfers themselves are not technically an economic cost.  Once again, our methodology is misrepresented.  Finally, it is untrue that we claim to estimate “the net social loss” since we clearly state that we do not examine the benefit side.  We state on page 1 that this is a first approximation - an estimate - of the true costs of the current U.S. tort system based on the most reliable scholarly research available today.  The estimate will be refined as advances in procedures and studies emerge over time.

Sadly, what is fictitious are the many untrue claims made here regarding our study.  The review is freighted with factual untruths, misrepresentations, and inaccuracies.  We counted a total of 24.  It is a classic example of how knee-jerk criticism on the blogoshere has turned much of it into the “blog-o-smear.”]