"Are Drivers of Air-Bag Equipped Cars More Aggressive? A Test of the Offsetting Behavior Hypothesis" by Steven Peterson, George Hoffer, and Edward Millner, in The Journal of Law 6' Economics (Oct. 1995), Univ. of Chicago Press, P.O. Box 37005, Chicago, 111. 60637.

Anybody who doubts that people are strange ought to consider the research on car safety measures. There are conflicting find- ings, of course, but one line of research strongly suggests that people react to new safety devices, from seat belts to studded snow tires, by driving with greater abandon. In the curiously apt jargon of the economist, they are said to increase their "driving intensity." So much so, apparently, that they may totally negate the worth of the safety measure.

That is precisely what Peterson, Hoffer, and Millner, all of Virginia Commonwealth University, find in their study of the effects of air bags (which will be required for both front-seat occupants in 1998 autos). They made a list of the car models (excluding station wagons) to which air bags were added between 1989 and '93, then checked insurance industry records of personal injury claims. Sure enough, claims for the air bag models rose "significantly." (Interestingly, however, claims dropped for models that were "upsized.")

The authors then looked at data on the 207 fatal car crashes in Virginia during 1993 that involved late-model cars. "Of the 62 crashes involving cars equipped with air bags, 53 were . . . initiated by the driver of the air-bag equipped car." That's strong evi- dence, the authors say, that the reassuring presence of air bags promotes hot-dogging on the road.

It gets worse. There were 33 crashes in which the driver was the sole fatality, and 16 of these drivers were protected by an air bag. This is further evidence, the authors say, that protection offered by the air bags is offset by a new recklessness.

Even worse, drivers protected by air bags seem to be a danger to passengers. There were 13 single-car crashes in which a passenger was killed but the driver was not. In nine of these, the driver had an air bag.

The authors leave readers to draw their own conclusions. (It's important to note that traffic fatalities have been declining for decades.) One obvious possibility is that cars ought to be equipped with air bags for everybody but the driver.