Wages and health insurance

Over at Marginal Revolution, Alex Tabarrok catches David Leonhardt in a fundamental error concerning wages and benefits:

David Leonhardt makes an interesting argument about why employers don't choose employee health insurance carefully. The argument is interesting because it is wrong but in a subtle way.

The bottom line: The cost of insurance comes mostly out of employees' paychecks. If insurance costs more, employees are generally paid less. If insurance costs less, employees are paid more. The cost of insurance does not have a big effect on employers’ overall compensation costs.

That’s why no one should be surprised that employers don’t make for good consumers of insurance. And it’s why insurers are not operating in a very competitive marketplace.

The premise is correct, employee compensation comes out of wages. The subtle mistake is to forget that this is only true in equilibrium. Imagine that a single employer was able to buy for his employees equal quality health insurance at a lower price. Would wages at that firm rise? No, an employer only has to pay workers what they could earn in another job. If other firms aren't paying more then this firm need not raise wages even though its costs have fallen. Thus an employer that reduced health insurance costs while keeping real compensation the same could pocket the savings as profit. It's only when other firms follow suit--also in an attempt to cut costs and earn excess profits--that wages at all firms rise, eliminating the excess profit everywhere.

Tabarrok's argument really boils to down to this: we have an example of a cost saving opportunity. We know that profit-maximizing firms do not pass them up. So Leonhardt must be wrong.

Leonhardt distracts us with a red herring. It is true that, in or out of equilibrium, to attract a worker you need to pay him in wages and benefits (as valued by him) at least as much as he could get elsewhere. An employer can reduce the wage and increase the value of the benefit by the same amount without changing the package's attractiveness to the worker. That is, the "cost value of insurance comes {snip} out of employees' paychecks." I would characterize Leonhardt's error this way: he is suggesting a principal-agent problem exists; the one who chooses the insurance is not the one who pays. But he is forgetting why "cost value of insurance comes {snip} out of employees' paychecks." And he is not allowing for a difference in cost and value. If an employer can find an insurance plan at the same value for less he will buy it.

But that doesn't mean that profit-maximizing employers "make for good consumers of insurance." In a world where the employees you have today are not the employees you have tomorrow, you have no incentive to choose plans that create incentives for employees to engage in healthy behaviors. That begs the question of why there is employer-based health insurance provision, but that is answered by the favoritism towards them in the U.S. personal income tax code.

It's significant that healthy behaviors are where there actually is evidence for large health care savings.