Being an Economist Means Not Having to Have Any Actual Knowledge of What You’re Talking About, Exhibit 37: Occupational Safety and Health

Robert Frank’s column in this morning’s New York Times is about the economics of occupational safety and health, or more precisely, about Robert Frank’s one obsession with economic theory, invidious comparisons of well-being.  Frank has gone totally hedgehog over the tendency to evaluate one’s condition in relation to others; it has been the subject of virtually all of his writing for the past decade or so.  To better highlight this one wrinkle, he has embraced the most orthodox of orthodoxies in every other respect, which means that his treatment of policy topics has become little more than an exercise in deduction from axiomatic principles.

To see what I mean, take a look at this latest piece on “the price of danger”.  Here is his summary of the issue as seen through the lens of economic theory:

Consider the familiar trade-off between wages and workplace safety. Because safety devices are expensive, additional safety means lower wages. Reducing risk to zero is impossible, so the practical question must always be this: How much safety is enough? Since Adam Smith’s day, classical economic theory has held that well-informed workers in competitive markets will navigate this trade-off sensibly. They will accept additional risk in return for higher pay only if the satisfaction resulting from their additional buying power is greater than the corresponding loss in satisfaction from reduced safety. Regulations that mandate higher safety levels make workers worse off by forcing them to buy safety they value at less than its cost.

Of course, as Frank notes, workers have clamored for safety regulation, and some degree of standard-setting and monitoring is universal in industrialized countries.  Hence the theory must be wrong.  Why?  The answer is—surprise!—the tendency to evaluate our well-being in relation to others.  Income is a rat race, so trading off risk against income makes workers worse off.  And that’s it—end of story.

Does it matter that every single point that Frank raises is contradicted by the evidence?  He apparently doesn’t care, since evidence doesn’t get any space.  Having a clever idea is enough.  If Frank were the only economist with this frame of mind, we could safely ignore him; alas, he has a lot of company.  Pick up any economics textbook and you will find sweeping judgments about all sorts of issues of law and policy based entirely on deductive theory, without any consideration of the history of the topic or factual details that people on the ground have regarded as important.  (Exhibit 1 would have to be rent control.)

So let’s make a list of some of the things that people who study occupational safety and health know and Frank omits from his analysis:

1. Safety is expensive, but so are accidents.  There is an enormous literature that documents that firms frequently fail to take measures on their own initiative to prevent accidents even though they would pass a cost-benefit test.  Not all such measures would pass, but many would.

2. Adam Smith’s theory of compensating wage differentials was highly disputed during the nineteenth and twentieth centuries.  John Stuart Mill believed that, in times of involuntary unemployment, higher risk would be accompanied by lower wages—the sweatshop effect.  The institutional economists identified with legal realism in the early twentieth century agreed.  True, there was a period of theoretical monoculture in economics during the 1970s to 1990s in which compensating differentials became the hegemonic view, but that perspective declined with changes in labor market analysis post-1990 or so.  Take a look, for instance at Labor Economics, the graduate textbook by Cahuc and Zylbergerg; their skepticism toward the model that Frank identifies with “economic theory” is palpable.

3. Frank simply assumes that wage bargains are not affected by the prior allocation of rights—that it makes no difference whether workers have no right to safe jobs and must give up wage demands to get them, or whether they do have this right and have to be bribed with hazard pay to accept greater risk.  Of course, in the real world willingness to pay and to accept are vastly different, and workers understandably want all sorts of mandated job rights, beginning with health and safety.  To be effective, such rights have to be codified and enforced.  I have not made up this point; you can find it in legal reasoning during the period at the turn of the twentieth century when courts began to find in favor of injured workers.  (My source for this was Atiyah’s Rise and Fall of Freedom of Contract.)

4. Wage compensation for risk will be incomplete if the threat of dismissal is used to control worker effort, the so-called efficiency wage approach to wage-setting.

5. The literature on the behavioral aspects of risk perception and response is a thick stew.  Workers tend to respond asymmetrically to safety norms, as suggested by prospect theory.  They often retreat into denial of unpleasant truths about the risks they face, as suggested by cognitive dissonance theory.  Workers are human beings, not cost-benefit machines.

6. The empirical record for compensating wage differentials, despite what its proponents (like Kip Viscusi) claim, is mixed at best.  Positive coefficients on occupational risk in wage-risk regressions depend on the choice of control variables; they show up in some subsamples, such as unionized workers, and not others.  They are different for white workers compared to black, men compared to women, and these differences fluctuate from one sample to the next.

But why get bogged down in all this detail, specific to the economics of occupational risks, when the point is to show how you can solve the world’s problems with a single, all-purpose axiomatic model?