(FORTUNE Magazine) – The news from Washington is that the Bush Administration is making with the meat-ax in this year's budget planning, and yet OSHA will get the 6% increase it boldly requested. The news stories explain the request by stating that Secretary of Labor Liddy Dole (whose department runs OSHA) is determined to finally make the agency more effective. Terrific plan, eh? The Occupational Safety and Health Administration is obviously benefiting here from the Bushies' commitment to kindism-gentlism. In addition, it seems to be an agency whose steady record of failure weirdly works in its favor. If a huge crane topples in San Francisco, or steelworkers die in an explosion in Johnstown, Pennsylvania, or a construction worker is injured by a collapsing building while eating a turkey sandwich in New York City -- to cite three recent headline generators -- the reportage always seems to include a detail about OSHA sending around an inspector to check out safety violations. Obvious implication: If the agency were better funded, the carnage might have been avoided. The crack about OSHA's failures is empirically based. It refers, first of all, to the fact that scholars have traditionally had trouble finding aggregate data demonstrating that U.S. workplace safety has been positively and materially affected by OSHA. The data tell us that the normal workings of capitalism were enhancing workplace safety quite substantially in the years before OSHA got going. Between 1948 and 1971, annual deaths per 100,000 workers declined from 29 to 17. In the post-OSHA years (since 1971), there have been further steady declines, to nine per 100,000 in 1988. Whether that figure is lower than it would have been in OSHA's absence is an interesting question. W. Kip Viscusi of Duke University, a distinguished economist specializing in the economics of safety, has written that during most of OSHA's life, it generated no statistically significant safety gains. But in a contribution to a recent symposium on ''risk, safety, and $ capitalism'' (published in Society), he wrote that one could now discern ''evidence of a modest favorable impact.'' Which does not mean that OSHA is now a ''success.'' The modest impact is being created at an absurd cost, as evidenced by a stunning statistic in Viscusi's contribution to the symposium: ''Regulatory costs have run higher than $100 million per expected life saved.'' That figure is wildly out of line with other resource-allocation decisions involving Americans' safety. The figure makes no sense, if only because an equivalent amount of money invested in other ways (e.g., cancer research) could be expected to save a lot more than one life. The $100 million is out of line with the judgments made by workers themselves about the value of their lives -- judgments reflected in, for example, the pay premiums required to get people to take risky jobs. Econometric analyses of those premiums suggest that an average high-paid worker is implicitly valuing his own life at something like $5 million. A question insufficiently pondered by ''workplace advocates'' -- a phrase now applied quite regularly to fans of safety regulation -- is what was causing safety to increase before OSHA came along. The answer is developed brilliantly in Searching for Safety by Aaron Wildavsky, a political scientist based at the University of California at Berkeley. Wildavsky's answer is, au fond, capitalism and free markets. Severely capsulized, the mechanism described by Wildavsky works somewhat as follows. First, capitalism raises incomes. Second, because of the high income elasticity of demand for risk reduction, this rise works to reduce risk. In other words, safety is one of the things people tend to buy more of as their incomes rise; well-off workers will increasingly trade off pay for safety. So when markets are free and information is good, the system moves quite naturally toward a safer workplace. OSHA's regs are not in the same league.