First, do no harm. (Mercatus Reports: Commentary).

Klick, Jonathan ; Stratmann, Thomas

WHEN CONGRESS PASSED THE OCCUpational Safety and Health Act in 1970, then secretary of labor James Hodgson heralded it as the "most significant legislative achievement" for workers in a decade. The agency created to implement the provisions of the Act, the Occupational Safety and Health Administration, took as its mission "to save lives, prevent injuries, and protect the health of America's workers."

In carrying out that mission, OSHA's staff of more than 1,000 inspectors visited work sites 35,897 times in 2001, discovering nearly 80,000 safety violations amounting to $82 million in fines. It remains an open question, however, whether OSHA really does "protect the health of America's workers."

OSHA claims that the issue is clear-cut, declaring that its activities make America's workplaces safer, and there are a few academic studies supporting that view. Unfortunately, there are just as many studies suggesting that OSHA enforcement does little to change the level of safety enjoyed by workers.

Statistical analysis of the issue is difficult because OSHA targets much of its enforcement at industries and firms that, by the nature of the work they do, have the most dangerous conditions for workers. Because of that, it is difficult to isolate the causal effect of OSHA enforcement. On the surface, it appears that the firms OSHA inspects and fines tend to have the highest rates of worker injuries and deaths, but it is not possible to conclude from this relationship that OSHA enforcement leads to more accidents and fatalities. Instead, it is more likely that accidents and fatalities lead to more OSHA enforcement.

Enforcement's effects

In a new Mercatus Center research paper, "Offsetting Behaviors in the Workplace," we attempt to get around this "selection" effect through the use of an instrumental variables technique to determine exactly what greater OSHA enforcement accomplishes. The key to our technique is to find variables that affect how frequently OSHA performs inspections and what level of fines it levies, but do not otherwise affect the safety of workers. With such variables, we can effectively model the level of OSHA enforcement independent of the underlying safety of the targeted industries.

It turns out that OSHA tends to target industries that are less economically important, and performs more inspections and assesses more fines in those industries with relatively low gross state industry products. Thus, gross state industry product is our first instrument. Another instrument that proves to be useful is whether or not OSHA has granted "final approval" to a state agency to control worker safety regulation within the state. Under the Occupational Safety and Health Act, states may apply to take over worker safety regulation. After a period of state control, if OSHA determines that state enforcement is "at least as effective" as federal enforcement, final approval is granted, and the state agency can receive funding for up to 50 percent of its costs from the federal government. However, OSHA can revoke final approval if it determines that the state agency is no longer effectively protecting workers. After final approval is given, the level of inspections increases and the level of fines tends to d ecrease.

Using those instruments, we examined the effect of worker safety inspections and fines on the fatality rates at the industry level within the states comprising OSHA's Region IV (Alabama, Florida, Georgia, Kentucky, Mississippi, North Carolina, South Carolina, and Tennessee) and Region IX (Arizona, California, Hawaii, and Nevada) for the period 1992-2001. Surprisingly, we found that fines have no statistically significant effect on death rates and increasing inspections actually leads to significantly higher fatality rates. On average, we found that an additional 100 inspections in a given state-industry in a particular year leads to between 1 and 2.5 additional deaths in that industry. That result is robust to a wide range of statistical specifications.

OSHA efforts and worker efforts

What accounts for such a surprising result? The intuition is quite simple. If worker safety depends on firm inputs and worker inputs, when the firm increases its efforts because of OSHA enforcement, the worker rationally substitutes away from his own efforts. That is, if the firm is doing more to protect the worker, the worker has less incentive to protect himself.

On net, that substitution implies that increasing firm safety inputs could lead to more, less, or the same amount of worker safety depending on worker preferences. In this case, it appears as though workers more than offset the additional safety measures induced by increased OSHA enforcement.

Our finding resembles of the so-called "Peltzman Effect." In a 1975 issue of The Journal of Political Economy, University of Chicago professor Sam Peltzman speculated that automobile safety devices might induce individuals to drive more recklessly because safety equipment such as seatbelts protected the drivers from some of the costs of their risk-taking. He found evidence of that effect, though he did not find that drivers more than offset the benefits to themselves of the improved safety.

The policy implication of this offsetting behavior in the workplace is striking. If workers effectively undo safety regulations, it is doubtful that OSHA can do much to "save lives, prevent injuries, and protect the health of America's workers."

Jonathan Klick is the Dorothy Donnelley Moller research fellow at The Mercatus Center and a Robert A. Levy Fellow in Law and Liberty at the George Mason University School of Law. He can be contacted by e-mail at iklick@gmu.edu.

Thomas Stratmann is a faculty scholar at The Mercatus Center and a professor of economics at George Mason University. He can be contacted by e-mail at tstratma@gmu.edu.