≡ Menu

Science Involves More than Measuring What Is Available to Measure

Below is a comment that I left at EconLog in response to Thaomas’s comment on this post on minimum-wage research by David Henderson.

First, here’s Thaomas’s comment:

While it is obvious that increasing the EITC is a much better way to increase incomes of low wage workers than increasing the minimum wage, it’s too bad the study did not (at least the abstract did not) estimate the income loss to the income gains of those who did not lose hours of employment. Was the total wage bill lower?

And here’s my response:


You here again assume that the question of the validity of minimum-wage legislation is one that can, or should, be settled by naive cost-benefit analysis: count the dollars gained by workers whose monetary incomes rise because of the legislation against the dollars lost to workers whose monetary incomes fall because of the legislation.

With respect, while this approach seems on its face to be very scientific and objective, in reality it is neither. Below are four of the many problems that I believe are evident with your approach.

First, your approach ignores non-monetary losses and gains unleashed by minimum wages. Such losses are likely far greater than any such gains. The biggest such loss that is missed by your method is the value of the work experience that workers who lose jobs are denied. The loss of current incomes for many of these workers is likely not the most serious cost; the most serious cost is the inability to get job experience and, thus, being cursed to remain longer in the pool of least-skilled workers.

In principle, an omniscient being could attach an accurate monetary value to these losses, but in reality no human being can reliably do so.

Second, your approach ignores minimum-wage-induced changes in jobs’ non-wage characteristics. Many workers whose monetary incomes rise as a result of minimum wages will find that their jobs have become more demanding – their supervisors less tolerant of activities such as personal texting – their employers more reluctant to offer improved workplace amenities such as cleaner break rooms and less harsh workplace lighting. It is practically impossible for third-party observers even to document all such changes in the nature of jobs, and even more difficult to attach to each of these changes a monetary value that must then be subtracted from the higher monetary incomes in order to arrive at a figure useful for accurate cost-benefit analysis.

Third – and least importantly (but it nevertheless warrants mention) – your approach ignores subjective-value differences. Would a suburban teenager’s extra $35 in weekly take-home pay be worth achieving if it comes at the expense of a fall of $30 in the weekly take-home pay of a single inner-city mom?

Fourth, if your cost-benefit method is to be used to determine if minimum-wage statutes (or instances of minimum-wage hikes) are or are not justified, why stop there? Why not agree to let the government, say, attempt to raise the pay of cardiologists by imposing a minimum wage for cardiologists? Some cardiologists, as  a result of such government intervention, will see their incomes rise while others will see their incomes fall as they are obliged to pursue some other occupation (which might or might not be in the field of medicine). If the government imposes such a minimum wage for cardiologists and the results are that the total pay of cardiologists rises, would you conclude that a policy of setting a minimum wage for cardiologists is justified?

Or suppose that the government sets a minimum price for the sale of garden hoses, with the empirical result being that the total revenue earned by sellers of garden hoses increases. Would you conclude that this garden-hose policy passes a cost-benefit test?

Saying that ‘we must be scientific and objective and compare the costs of a policy to its benefits’ is much easier than actually carrying out such a comparison. Sometimes such a comparison in real-world instances of policy changes is possible to do in a useful way, but not always. It is not remotely possible for analyzing minimum wages.


Next post:

Previous post: