Motus Mentis

There’s been a lot of argument lately about whether to raise the minimum wage, and I haven’t had much to say about it in these pages. But a comment in a recent thread got me digging into the topic more deeply than I had before, and so it’s time for a post about it, I think.

The subject came up almost by accident: our liberal commenter Peter K., a.k.a. The One Eyed Man, having made the audacious assertion that conservative thought was ‘mostly hokum’, offered to to take any conservative column or article of my choosing and expose its ‘fatuities’. I quickly chose one, more or less at random: a recent piece by Thomas Sowell on the minimum wage. I invite you to read it before proceeding.

Peter, to his credit, came back a day later with a substantial response. I reproduce it here below, and will add my own comments after that:

Ad hominem is not my thing, so I’ll leave it to others to decide if Sowell is a big dope or not. However, his piece is rife with errors in fact, logic, and history, and can aptly be described as hokum. Among the most egregious errors are these:

1) Sowell asserts that “there was a time when there was no federal minimum wage law in the United States. The last time was during the Coolidge administration, when the annual unemployment rate got as low as 1.8 percent.’ He is wrong on the facts ”“ the federal minimum wage was established by the Fair Labor Standards Act of 1938, five years after FDR took office ”“ and he repeats this error in his penultimate paragraph. More importantly, he is wrong about the conclusion he draws. The unemployment rate in 1938 was 19.0%, which is greater than every year which has followed it. By his logic, the federal minimum wage has been a smashing success, because it has led to lowered unemployment ever since it was first put in place.

2) Similarly, the peak in the minimum wage in real terms was in 1968, when it was roughly $11 in today’s dollars. The unemployment rate in 1968 was 3.6%, and it has been higher ever since then. Again, using Sowell’s logic, if a high minimum wage was a disincentive to employment, then 1968 would have been a year of record unemployment, and every year which followed would show lower unemployment correlated to the lower minimum wage.

3) The fatal flaw in Sowell’s argument is that there are many factors governing the employment rate which are far more determinative than the minimum wage. In the years following World War II, high school dropouts could get jobs at the factory and make a decent living. No more. The automation of manufacturing, global competition, the rise of China and other low wage countries, the decline of unions, the ascendance of skilled labor over unskilled labor, and the steady increase in productivity which has reduced the need for labor, are all far more consequential in determining the level of unemployment than where the minimum wage is set. Companies have been forced to learn how to make a lot more stuff with a lot fewer workers, which far overshadows any role the minimum wage may have had in the reduction of the labor force.

4) Sowell’s assertion “that people tend to buy more when the price is lower and less when the price is higher’ pertains to goods, and not to labor. If your burger shop has to increase wages by 10% to meet a higher minimum wage, and every other burger shop in town is faced with the same increase, then the price of burgers will increase, but not the number of workers required to make them (provided that the price does not increase to the level where people give up eating burgers). It is a transference of wealth from burger eaters to fast food workers, but is exogenous to the number of people working in burger shops.

5) Sowell writes that “a recent issue of ”˜The Economist’ showed Switzerland’s unemployment rate as 2.1 percent,’ and touts this as evidence that there is a correlation between no minimum wage and low unemployment. Again, he’s wrong on both the facts ”“ I’m looking at this week’s issue of the Economist, which shows the Swiss unemployment rate to be 3.2% ”“ and the conclusion he draws. There is no minimum wage in Sweden (8.6% unemployment rate), Finland (8.4%), or Denmark (5.6%). Sowell cherry picks data by picking the no-minimum-wage country with the lowest unemployment in Europe, ignores the other countries without a minimum wage, and then implies that the outlier case is the norm.

6) He makes much of the black unemployment rate in isolation from the overall unemployment rate. The more relevant criterion is the ratio of black unemployment to white unemployment, which has been falling steadily from Reagan and Bush I (2.3) to Clinton (2.2) to Bush II (2.1) to Obama (1.9). This argues for the success, not failure, of the minimum wage as applied to blacks.

* * * *

Raising the minimum wage to $10.10 will cause jobs whose worth to the employer is less than $10.10 to be eliminated. The CBO gives a two in three probability that the number will be between negligible and one million, so let’s pick the mid-point and say that 500,000 jobs will be lost. The CBO also reports that a minimum wage increase will raise wages directly for sixteen million people and indirectly for another nine million people. Moreover, the cost of providing food stamps and EITC payments will decline commensurately with an increase in the minimum wage, as more workers get priced out of these programs.

If all you care about is the total number of people working, at whatever meager wages they are able to get, and you don’t care much about how much money low income workers receive, then you might give preference to saving 500,000 (very) low wage jobs over the 25 million people who would benefit from a higher minimum wage, along with the reduction in social welfare spending which results from higher wages. This is the only valid argument against the minimum wage, and Sowell does not make it. However, it seems to me that the trade-off is clear: better to raise wages for 25 million people, as well as reduce government social welfare outlays, than keep 500K marginal jobs.

Conservatives against the minimum wage won’t accept the trade-off of higher wages for reduced employment, but there are plenty of instances when they are more than happy to accept reduced employment when the trade-off runs counter to policy goals they cherish. Maintaining unemployment insurance for the long term unemployed would save 200K jobs; Republicans voted it down. The sequestration cost 900K jobs, and Republicans insisted on it. Infrastructure spending would reduce unemployment greatly, and it gets voted down in the House every time. So when the trade-off is something conservatives don’t care much about (higher wages for low wage earners), they insist on the primacy of employing as many people as possible. When the trade-off involves policy preferences they care very much about (reducing deficits by curtailing unemployment insurance, budget sequestration, and deferring maintenance on the infrastructure), then the importance of employing as many people as possible dwindles to insignificance.

Let me begin by saying that Peter’s first point in item 1) is quite right: Mr. Sowell is simply wrong about the dates. He claimed that the last time there was no Federal minimum wage “was during the Coolidge administration”, which ended in 1929, but the first Federal minimum wage appeared in 1938. Even the Davis-Bacon Act, which established minimum wages for Federal contractors, was signed into law by Herbert Hoover in 1931. How Mr. Sowell and his editors could have allowed such a plain factual error to appear in print, I have no idea.

Peter also said this in 1), however:

The unemployment rate in 1938 was 19.0%, which is greater than every year which has followed it. By [Sowell’s] logic, the federal minimum wage has been a smashing success, because it has led to lowered unemployment ever since it was first put in place.

This is itself a rotund ‘fatuity’. This spike in unemployment was due to a great recession that began in 1937. What caused unemployment to fall soon after, and to remain low for many decades, was of course the outbreak of World War Two, and the enormous wave of growth that followed it. Nobody in his right mind, and certainly not Thomas Sowell, would imagine that the effects of a low minimum wage on overall unemployment — which are, in the best of times, faint enough that they are the subject of continuous debate — would have been salient during that great upheaval.

Peter’s points 2) and 3) push back, generally, against the idea that minimum-wage laws have any strong effect on overall employment rates. In this I think he is right. Minimum-wage earners are only 1.2% of the workforce (there are in fact more people making less than minimum wage). This sets a very low ceiling on the effect that even the worst case imaginable — the immediate firing of every minimum-wage worker as soon as the minimum wage rose by a single penny — could possibly have on overall unemployment.

In 4) Peter makes the argument that if burger-flippers cost more to hire, the burgermeisters will simply pass along the increase to the customer. This may also be true, but only up to a point, and to the extent that it is true, there are other effects to consider. We’ll come back to that.

In 5), Peter takes Sowell to task for ‘cherry-picking’ Switzerland as an example of a nation with no minimum wage and low unemployment. (It’s true that Switzerland is just unnaturally great in almost every way.) But there are many ways of looking at this. Here’s one, from blogger Scott Summer, as cited in Forbes:

Regarding the minimum wage, here is some data for Western Europe:

There are nine countries with a minimum wage (Belgium, Netherlands, Britain, Ireland, France, Spain, Portugal, Greece, Luxembourg). Their unemployment rates range from 5.9% in Luxembourg to 27.6% in Greece. The median country is France with 11.1% unemployment.

There are nine countries with no minimum wage (Iceland, Norway, Sweden, Finland, Denmark, Austria, Germany, Italy, Switzerland.) Five of the nine have a lower unemployment rate than Luxembourg, the best of the other group. The median country is Iceland, with a 5.5% unemployment rate. The biggest country in Europe is Germany. No minimum wage and 5.2% unemployment.

Still want to raise our minimum wage to $10? Germany used to have really high unemployment. Then they did labor reforms to allow more low wage jobs, combined with subsidies for low wage workers. Now they don’t have high unemployment.

It’s complicated: the EU countries that don’t have minimum-wage laws nevertheless have de facto minimum wages, based on labor-management negotiations, that cover most workers. (Also, the more generous the welfare state, the more people will take advantage of it, which obviously puts a downward pressure on employment rates.) But as I said above, I think the main point to take away from Peter’s response as regards overall unemployment rates is that they are simply not very sensitive to adjustments in the minimum wage. I believe he’s right about this, and I believe he’s right to criticize Sowell for making so much of it.

In 6), Peter takes up the question of the black/white unemployment ratio. I went to the Bureau of Labor Statistics, and using their data I created a spreadsheet reflecting unemployment rates going back to 1972, the first year for which they offered unemployment data broken out by race. I added rows for overall unemployment, unemployment by race, unemployment among 16-24-year-olds, unemployment among 16-24-year-olds by race, and the minimum wage in constant dollars. This enabled me to put together some graphs (with thanks to Google Charts).

This one, which plots the minimum wage, overall unemployment, and the white/black unemployment ratio, is telling, I think:

 
We see not only that the minimum wage is not visibly correlated with total unemployment, but also that the black/white unemployment ratio remains remarkably constant at roughly 2:1, regardless of total unemployment or any changes in the minimum wage.

The claim is often made is that minimum-wage increases disproportionately affect young workers. In the next graph we see the minimum-wage rate again, and another curve showing the ratio of total unemployment to unemployment among 16-24-year-olds:

 
As you can see, the unemployment rate of 16-24-year-olds hovers quite steadily at around twice the overall rate, despite changes in the minimum wage.

Finally, here’s one more — the minimum wage again, this time with a curve showing the ratio of unemployment rates among blacks vs. whites in the 16-24-year-old cohort:

 
Again, the ratio between black and white unemployment rates in this age group seems fairly constant (at around the same 2:1 ratio that exists in overall unemployment rates), and it’s hard to see any correlation with the minimum wage.

Mr. Sowell’s argument rests heavily on the effects of minimum-wage increases on both overall unemployment and unemployment among young people, particularly blacks. Before I was provoked by Peter’s response to delve into the statistics more deeply than I had before, I had found this line of argument much more persuasive than I do now. So I believe I have to say, Peter, that I stand corrected on this point, and I think Mr. Sowell does too.

There is much more to say about all of this, however. The final paragraphs of Peter’s response make a positive argument for raising the minimum wage, and the second part of Peter’s challenge to me was to present a counterargument to the case for a minimum-wage increase made in this New York Times editorial. I’ll get to that in a subsequent post.