Minimum wage-unemployment studies

[tunk]

I’ve read Dube, Lester, and Reich’s study (“Minimum Wage Effects Across State Borders”). It’s interesting because they say that, historically, both studies that have shown disemployment effects from the minimum wage, and studies that have not, use methods with severe flaws. The former fails to take into account what they call “heterogeneity” in national unemployment statistics, and the latter, employing cross-state variations in minimum wage rates, fails to account for the possibility that minimum wage effects have a time-lag. They claim their method combines the different advantages of the two. I’m neither a professional economist nor a a statistician, so I couldn’t really evaluate it.

One could quarrel with the fact that, like Card and Krueger, they conveniently use data from a period of economic boom (around 1996-2005). But setting this aside, all they basically show is that when minimum wages are marginally increased (studied increases measure anywhere from 7 - 20%), almost nobody loses their job, a few people gain jobs, and reported incomes increase.

Needless to say, this doesn’t come within an inch of overturning basic economics. It’s perfectly understandable that an employer might not sack all of his existing employees when he is forced to give them a pay raise. Most minimum wage employers are small retail and restaurant outlets, for whom it such a tactic would be costly and damaging to employee morale. (Though the numbers do add up; employing 5 workers for $6/hour at 8 hours a day amounts to $240 a day.)

Yet it is undeniable that following a minimum wage hike, costs of production have been increased, particularly for the small businessman. Hiring low-skilled labour has been made more expensive relative to hiring highly-skilled members of wealthy, middle-class unions, automating, or shipping the job overseas (if possible). And a profit-maximizing employer will take this into account when he decides to expand, if he does at all. So you may not see a high rate of job losses, but neither will you see a high rate of job creation for for the unskilled, arguably those who need jobs the most.

Evidence for this abounds. The disparity between US black and white teenage unemployment rates was negligible before the first minimum wage increase in 1949. Sweden, though lacking an official minimum wage, has similar and quite strict labour requirements set by unions, and the youth unemployment stands at 29%. Australia is often cited as a case where the minimum wage is quite high - around was $15.15/hour as of July 1st, 2011 - and yet there is a low unemployment rate. However:

The first thing that must be said about the official unemployment statistics is that they cover up the real level of joblessness. Numbers of studies have established that when the “hidden unemployed” are included—for example, those who are deemed employed because they perform one hour’s paid work a week—the joblessness rate leaps to over 12 percent. While the official figures show there are around 600,000 jobless, the number stands at approximately 1.3 million when hidden unemployment is factored in.

[…] [I]n the one-time industrial city of Wollongong, an hour’s drive south of Sydney, unemployment stood at 9.6 percent, the highest in the country. Youth unemployment was a massive 30 percent.

(Google “hidden unemployment australia” for some of those studies.)

Some of the pro-minimum wage studies which merely survey employers (Card & Krueger come to mind) claim to show an increase in employment after a minimum wage hike. As Thomas Sowell points out in Basic Economics, any such study can only survey the businesses that manage to survive such a hike. There will be fewer businesses afterwards that have access to a larger pool of workers, and so it may appear as though employment has risen because empoyment per firm has risen. But with such a method you could arguably prove that nobody died during WWII.

Most importantly, as Dube, Lester, and Reich acknowledge:

[The] data do not permit us to test whether restaurants respond to minimum wage increases by hiring more skilled workers and fewer less skilled ones.

See Roger Garrison’s powerpoint presentation on this subject. Nobody has ever argued that if you pass a minimum wage increase, throngs of people will be thrown out of work onto the streets. Rather, what the minimum wage does is make low-skilled labour more expensive relative to other options. It decreases the demand for low-skilled labour and increases the demand for other kinds of labour or for labour-saving technology. The almost inevitable result will be a reduced rate of job creation for low-skilled or unskilled labour, and as a result a relatively high rate of joblessness among youth, minorities, the poor, etc. Simply demonstrating that more people may be hired or more income reported in the wake of a minimum wage increase does not even begin to address who is being hired and who isn’t, which constitutes the bulk of the case against the minimum wage to begin with. All that has been shown is what our side has argued all along: that most of its supposed benefits apply only to those who already have jobs or who already have vital work experience, those on the inside looking out. Things are very much different for those on the outside looking in.

And in any event, even if it could be shown that tiny increases in minimum wages have little to no unemployment effects as labour market oligopsony models predict (though such models are dubious, as Don Bellante points out), this wouldn’t much help the case of its advocates. They don’t want a 30 cent increase. They want a fullblown “living wage”, and have absolutely no evidence that such a thing wouldn’t do more harm than good. This is extraordinarily disingenous and hypocritical. If the minimum wage doesn’t affect the employment market, why don’t these people go ahead and call for a minimum wage that is ten times higher than what prevails? How about a milion dollars an hour? After all, it wouldn’t have any effect!

Here’s a cogent comment from Econlog on a related topic, “efficiency wages”:

I’ve never understood how people can seriously use efficiency wages to argue for a high minimum wage. Just because paying people more under some circumstances can increase their productivity enough to pay for the wage bump doesn’t mean that this is true under all circumstances. Taken on its own terms, an efficiency wage is not just some magic formula whereby paying people more increases their productivity. For instance, it’s not usually understood as an income effect (e.g., workers can afford good transportation and daycare which reduces absenteeism). Rather the theory of an efficiency wage is that it makes the particular job attractive relative to the other options and either the employee is grateful for the high wage (what Akerlof calls a gift exchange of excess effort for excess wage) or is simply terrified of getting fired from such a great paying job. This is important because if you think about efficiency wages as a “relative to other options” thing rather than an “income effect” thing, a high minimum wage will compress the wage distribution and therefore reduce any efficiency wage effects from the status quo ante.

For instance, if Costco and WalMart are in the same labor market but Costco pays $12 and WalMart pays $6, then Costco can reasonably expect more motivated employees as they’d realize how lucky they are to be at Costco. However if a living wage law raises the minimum wage all the way to $12, there’s no longer a relative advantage to being at Costco. If you think efficiency wage is about income effects then this change should increase productivity at WalMart, but if you think it’s about relative to the competition (as is the dominant model in the theoretical gift exchange literature) then this change should decrease productivity at Costco.