Here’s a follow-up to yesterday’s post (written immediately after I finished writing yesterday’s post, but not posted until today – so what is written below takes no account of whatever comments might have been offered in response to the earlier post). I break up these posts in order to avoid making already overly long posts even longer.
There are at least two additional problems with calls, such as those that are often made by EconLog commenter ThomasH, to use empirical data as a guide for determining whether or not to use minimum-wage legislation as a policy to raise the incomes of low-skilled, or poor, workers. The first problem is a “practical” one centered on the limitations of quantitative data; the second problem is a theoretical one centered on a failure to trace out the full implications of basing policy on such analyses of data.
(1) The “practical” problem centered on limitations of the data.
I surround “practical” with quotation marks to signify my sense that this practical problem runs so deeply that it might well be properly classified as a theoretical problem. But let’s stick with calling it merely a “practical” one….
Yesterday’s post mentioned the great difficulty of deciding which workers will, and which will not, be reckoned to be members of the sub-group of individuals whose collective fortune is either enhanced or diminished by the minimum wage. Again, if I understand ThomasH correctly, if the income of these workers as a group rises as a result of the minimum wage, the minimum wage should be considered to pass a relevant cost-benefit test: its benefits will, in such situations, be held to outweigh its costs, thus allowing the minimum wage to be classified as a socially worthwhile policy. (To his credit, unlike the typical minimum-wage advocate, ThomasH recognizes that the minimum wage does indeed price some workers out of jobs and, hence, also out of incomes.)
Let’s here, though, assume for the sake of argument that this problem has somehow been solved: we all agree that the welfare only of the relevant sub-group of workers matters for determining the goodness or badness of minimum-wage legislation, and we agree also on just what sorts of workers are, and which sorts aren’t, in the relevant sub-group of society whose group income will be used to determine if the minimum wage is a good or a bad policy. Even so, the data that we gather on this group income will hardly be sufficient to tell us if the minimum wage is a good or bad policy.
First, the minimum wage causes over time a significant change in the structure of production. It artificially encourages, as time passes, methods of production that are either more capital intensive or more high-skilled-labor intensive than otherwise. So given that minimum-wage legislation has been in place in the United States for almost 80 years, as a practical matter at least some of the people who we would ideally want to include in our relevant sub-group of workers will not today be workers at all. How many are the people who, were it not for the minimum wage, would today have jobs but, because of the minimum wage, do not have jobs? How many of these people are not even in the labor force? How many of these people choose instead (because their prospects of getting a job in the formal economy are so small) to spend more years in school, to remain unemployed (for example, by living with parents or with their children), or to work in the underground economy (on which reliable data are very difficult to acquire)? Because we can’t practically know that community-college-student Jane Jones and living-with-his-parents Sam Smith are out of the labor force because of the minimum wage, we will not know to include their (lack of) incomes in our empirical calculations of the consequences of the minimum wage.
Second, suppose that in our calculations of the consequences of a hike in the minimum wage we count on the negative side of the ledger the income that Joe Martin loses when he gets laid-off (or not hired) as a result of the higher minimum wage. Such a negative counting is appropriate. But it is also likely to be an underestimate of the loss that Joe Martin suffers because of the minimum wage, for it is practically impossible to measure accurately the resulting fall in the net present value of Joe Martin’s lifetime stream of income. When the minimum wage causes Joe Martin today to suffer unemployment, not only does Joe Martin lose current income, he loses also the prospect of gaining on-the-job skills that he would have gotten had he not been priced out of a job. These skills would have increased the net present value of Joe Martin’s lifetime stream of income. But by how much? It’s virtually impossible to know today – and, hence, impossible to include this fall in the NPV of his lifetime stream of income in our cost-benefit assessment of the income-distribuiton effects of minimum-wage legislation. Yet this effect on Joe Martin’s lifetime stream of income (and on the lifetime streams of incomes of all the other workers who lose jobs today because of the minimum wage) clearly is a negative effect of the minimum wage that should be – but cannot accurately be – included on the negative side of the ledger.
(On the other side, perhaps we should include on the positive side of the ledger the increases in the net present values of the lifetime streams of income of those workers who, unable to find employment because of the minimum wage, go instead to trade school or college. Without question, the number of such workers in reality who, as a result of the minimum wage, earn higher incomes over their lifetimes is greater than zero. Yet it seems sensible to assume that the number of these ‘benefited’ workers is smaller than is the number of workers, who like Joe Martin, suffer a decline in the NPVs of their lifetime streams of earnings. The reason is that such ‘benefitted’ workers have, even without the minimum wage, the option to attend trade school or college in order to increase the NPVs of their lifetime streams of earnings. Presumably, the great majority of people who find it beneficial to pursue these routes as means of increasing the NPVs of their lifetime streams of earnings will do so without having been prompted by being priced out of jobs today by the minimum wage. In contrast, the minimum wage denies to the Joe Martins of the world opportunities that they would otherwise have chosen.)
Third, many workers who keep their jobs after a minimum-wage hike will suffer worsened job conditions. In some cases job conditions will become more dangerous. (For example, a lawn-care company might equip its workers with somewhat less safe lawn-mowers than it would have used had this company not been forced to pay its workers the higher minimum wage.) Even if we decide not to count on the negative side of the ledger the dollar cost to workers of, say, bosses who become, as a result of the higher minimum wage, less tolerant of workers who show up late for work or of workers who send personal text messages while on the job, surely we should count on the negative side of the ledger the cost (in dollar terms) to workers of any increases in the probability of a higher minimum wage causing them to be be killed or maimed while on the job. Yet, practically, how do we do so? To include such a figure accurately on the negative side of the ledger is much easier said than done.
(2) A theoretical problem centered on a failure to trace out the full implications of basing policy on an analyses of data on the minimum wage.
If the goal is to increase the collective well-being of a sub-group of people – tolerating, in the process, imposed harms on some members of this sub-group – then why focus only the effects on this sub-group of the minimum wage? Why not also consider price ceilings as part of the policy package (or as an alternative to raising the minimum wage)? That is, if the goal is to increase the aggregate real income of some sub-group of favored workers, why not consider imposing ceilings on the prices of the goods and services that such workers buy? Sure, such ceilings will cause shortages, but it’s an empirical question, is it not, if the welfare effects of these shortages will be positive or negative for the sub-group of workers who are the object of our sympathies? How do we know if the benefits to our favored sub-group of workers is higher if we impose a minimum wage compared to if we instead impose price ceilings on the goods and services that they purchase? Surely a genuinely “scientific,” cost–benefit-based, empirical analysis of policy would not be one jumped to before considering all available options. So why leap so quickly to raising the minimum wage? Why not, instead, compare the welfare effects (on our preferred sub-group of workers) of a higher minimum wage with those of price ceilings on the goods and services that such workers routinely purchase?
This question isn’t frivolous. If price controls are to be kept alive as a policy option subject to their passing some empirical cost-benefit test, then there is no reason whatsoever to keep alive only one particular price control (namely, the minimum wage) while rejecting others. A true scientist – someone devoted without bias to implementing policies based only on “the data” – would not leap so dogmatically to the assumption that the minimum wage is the only, or even the best, means of using price controls to raise the real incomes of some select group of people.
