It's Always the Other People Who Are Biased

Inspired by Daniel Kahnemann, Bryan Caplan presents an answer to the question, "Why are so many people so economically illiterate?" Caplan speculates that people confronted with an "economic literacy"-type question substitute a simpler question for it. For example,
Does the minimum wage help low-skill workers? (1)
get translated into
Would I be happy if employers gave low-skilled workers a raise? (2)
which often elictits the answer "yes". If it is then interpreted as an answer to question (1), we have an economically illiterate view, according to Caplan.

I guess there's a lot of truth to his model of constructing answers to econ questions, but perhaps it's worth pointing out that, according to what I've read and heard from him, Caplan also uses a simplified question translation, namely
Does a rise in wages lead to less demand for work, according to a supply-and-demand graph? (3)
to which the answer is yes. As for the truth, I haven't delved into the literature and won't, but according to this piece by Eric Crampton, not known as a raving socialist and opposed to minimum wages on grounds of liberty, minimum wages may not negatively affect employment if they are set at less than 45% of the average wage.

Lesson: When scoffing at other people's heuristics, be aware that your mind, too, may be working with a simplifying model. Disclosure: My heuristic in this post is, "trust what Crampton writes is roughly true".


Eric Crampton said...

Net effects on minimum wages are always a balancing of the increased income for low income folks who keep their jobs versus the disemployment losses for those who lose theirs. The disemployment effects are likely small where minimum wages are less than 40-45% of the median. That still doesn't make the welfare effects clear-cut as the incidence of minimum wage changes is often on the consumers of products produced by minimum wage workers, and those consumers are disproportionately poorer themselves.

I just can't get worked up about minimum wages as causing problems where they're sufficiently low (40-45% of median or so) EXCEPT where they're also applied to the disabled, like the significantly mentally or physically disabled where disemployment effects are likely to be large and where it's not implausible that the job has intrinsic value.

LemmusLemmus said...

Is it the average (mean) or the median? Given the way wages are distributed, that should make quite a difference, shouldn't it?

Eric Crampton said...

Double-checking now in case my memory was wrong... I'd posted average, and the place from which I was mainly drawing, here, said average. So revise the above. Thanks!

Eric Crampton said...

Ideally, we'd be tracking minimum wage relative to the median, not the average. There's no way that changes in exec compensation at the top, which boost the average, do much to affect productivity at the bottom; the median wage ought better track how binding the minimum wage is as a floor. The pieces I was looking at were before some of the big ramp-up in difference between median and mean earnings. So where 40-45% of the average might have been the limit in Canadian data or earlier US data, the cut-point might be lower than that now in US data with median wage stagnation relative to the average. So the rough (and likely wrong, but I don't know in which direction) rule of thumb I keep in my head is lower end of that range for the average, higher for the median. Hence the sloppy slipping between mean and median in the comment.

Anonymous said...

The underlying assumption of linear demand and supply schedules that make for the easiest presentation of the model is, at least, incoherent. Even in keeping with neoclassical logic, as soon as we recognize that leisure has value of its own, the supply schedule should be S-shaped; and there are good reasons to suspect the demand curve is better captured by a sine-like function (without crossing to a negative quadrant) - in any case, you get multiple possible resting points for the system, including high-wage/high-employment-rate combinations.

'Fallacy of composition'-type arguments, with their multiplier relations, offer reasons for why that can be expected.

Combining that with a high marginal propensity of consumption for lower income-classes (and, obviously, the nature of a minimum wage as below a certain threshold when compared to either mean or average income) makes it quite difficult to argue a compelling case against some minimum wage (and the concurrent easing of distributional problems) on the basis of economic theory as soon as the most simplified attempts at macro-modelling are left aside.

@Eric: Where does the claim originate that minimum wage workers are largely employed in the production of goods and services for people with a rather low income? Many of them work in the production of services that are more likely to be consumed by people in middle and even high income brackets.

Anonymous said...

P.S.: In fact, acknowledging Sonnenschein-Mantel-Debreu, the demand schedule may have any shape at all; argueably, the linear declining function is the least likely of all shapes.

In any case, multiple resting points for the system result.