Because, it doesn't look at the cost of the system itself, the "deadweight loss." Instead, it includes transfer payments to compensate people for legitimate losses (often completely outside the court system itself).
For instance, they include payments that don't involve the legal system at all. Say somebody smashes his car into the back of your new SUV and his insurance company sends you a $5,000 check to fix the damage. That gets counted as a tort cost in Tillinghast's number. Critics say it's just a transfer payment from somebody who wasn't driving carefully to somebody who has been legitimately wronged. How is that evidence of a system run amok? . . . . Just how open to interpretation are tort costs? Consider a study performed by the Rand Corp.'s Institute for Civil Justice back in the 1980s. The study covered 1985, and estimated that tort costs that year were between $29 billion and $36 billion. Tillinghast's figure for that same year is $83.7 billion.
To come up with its estimate, Tillinghast wades through data that insurance companies file with regulators, adding up the multitude of payments to people or companies claiming losses on anything from a fender bender to a hospital mishap. It adds in the costs of mounting defenses and investigations in tort cases, and administrative costs in the insurance industry.
As the article appropriately notes, the whole point of the tort system is to provide incentives not to engage in undesirable behavior:
How do you calculate the value to society of a tort system that deters negligent behavior? Isn't it a good thing, for instance, if fears of liability prevent an executive from fudging his quarterly numbers, a surgeon from walking into an operating room after having a few drinks, a driver from tailgating?
Nobody would deny, at least I wouldn't, that the tort system does not work perfectly. Just like other parts of the court system, such as the criminal justice system, judges and juries sometimes make the wrong decisions on the merits. Just like other parts of the system, the costs of litigation and procedural rules can skew the incentives of all parties to results different than what they would have been if decided on the merits in the settlement process that resolves most cases.
But, if you focus solely on administrative costs and inappropriate transfer payments, rather than all payments made by insurance companies, you get a very different picture of what is wrong with the system. The focus turns to limiting litigation costs (as No Fault Auto Insurance and the Worker's Compensation system do), and on maintaining competition in the insurance market, rather than on trying to beat down awards to legitimate victims (as damage caps do). You turn to preventing people from committing torts in the first place with underwriting rules that require best practices and penalized the insured for not following best practices and causing harm as a result, and on mitigating damages to the tort victim once a tort is committed, rather than putting procedural barriers in the way of recovery. You consider laws that penalize people with weak claims, rather than people with strong claims that juries agree with at trial.
The absence of tort law, by allowing people to harm other people's persons and property with impunity, is really both anti-business and anti-consumer.
Put another way, what Tillinghast measures is not the cost of the tort system, but the cost of torts themselves (which is underinclusive since it doesn't include torts for which there are no claims made for any reason). What does economics tell us is the best way to cut those costs? To create strong disincentives to commit torts. Do the incentives of the tort system work? A long term improvement in safety, at home, in the workplace and on the road, strongly suggest that they do.