Kevin Drum doesn’t like my take on the minimum wage
This kind of stuff bothers me on a bunch of different levels. Let’s count the ways:
- You either believe empirical studies or you don’t. If you have reason not to believe them, then let’s hear it.
- Intuition about supply and demand just flatly won’t work in this case. We’re talking about a market with (probably) low elasticities and a huge number of confounding factors that could push it in multiple directions. It’s easy to see that a small increase in the minimum wage could be overwhelmed by other factors and lead to either a very small or zero impact on employment levels.
- Are there jobs where it’s profitable to hire at $4.75 but not at $7.25? Well, there must be some, but we’re talking about such low skill levels here that there very well might not be many. That’s why empirical studies are so important. The effects are just too small to intuit.
- Is this really what we’ve come to? That we should provide a (probably very small) boost to the job market by allowing businesses to hire people for $9,500 per year instead of $14,500? Seriously? I mean, this is the ultimate safety net program, aimed squarely at working people at the very bottom of the income ladder. If we’re willing to throw them under the bus, who aren’t we willing to throw under the bus?
This is how I think the honest intellectual response goes: Basic economic reasoning tells me that the minimum wage cuts employment and that the more effective the minimum wage is in raising wages above market rates the more pronounced the effect will be. For years empirical studies seem to support this though I do not have deep knowledge of them.
Then David Card and Alan Krueger come out with a study showing the opposite. This had an impact on me primarily because the names Card and Krueger were attached to the study. They are smart, thorough guys and quite frankly that matters. Yet, I haven’t delved seriously into the empirical issues surrounding it.
I am aware of some of the common faults that people point out from the Card and Krueger work, but I haven’t really evaluated them carefully either. I could latch on to some of those criticisms as my “reasoning.” But, honestly they are not my reasoning.
My reasoning is that I had strong priors one way. I saw conflicting evidence that made me question those priors but that evidence wasn’t strong enough to completely persuade me.
Its also true that while my skepticism is a data point, people should keep in mind that it doesn’t carry a huge amount of weight because I don’t have a huge amount of evidence behind it.
There has also been some follow-up work that has confirmed Card and Krueger. My understanding is there has also been work to refute it. I read a well put together paper not long ago using border-states that seemed to show small effects but my impression at the time was that the variables of true interest were going in the expected direction, less than expected employment growth.
I stir all of that together with my causal empiricism about how businesses work and I come to the conclusion that “my best guess is that the minimum wage decreases employment opportunity for low skilled workers.”
My mind could clearly be changed on this but that’s where I am right now. Not to turn this into to much of a philosophy of science post, but I think there is too much hanging of hats on specific arguments, weaknesses in position, or points about analysis.
That’s not how we reason and that’s not how a good Bayesian should reason. All evidence “counts” just to varying degrees and priors matter. We shouldn’t pretend that unless we can find specific reasons to tear apart a study we have to accept it. Doing that just motivates us to be artificially critical.
We accumulate evidence over time and hopefully get closer and closer to the truth.
8 comments
Comments feed for this article
Friday ~ July 8th, 2011 at 3:35 pm
foosion
Remember the controversy about “reality based community” a few years ago? There’s a widespread perception that people of a certain political persuasion, such as those who oppose the minimum wage, act on instinct rather than evidence.
If your theory doesn’t fit the facts, do you change your theory or do you ignore the facts? Your answer is that you’ve thought about it and decided to go with theory, not because of any problems you can identify about studies which don’t support your theory, but because. well you really don’t come up with much.
Friday ~ July 8th, 2011 at 5:30 pm
Freddie DeBoer
Hmmm, I don’t think that’s quite fair. I think Karl is at once saying that he feels, in a provisional way, that the studies he is critical of have sufficient empirical refutation to support his preexisting, deductive skepticism, while also admitting that he hasn’t yet done the work necessary to evaluate the extant evidence and that there is certainly some confirmation bias leading him in the direction of skepticism. That’s my gloss, anyway. It’s a reasonable and honest position, I think, even though I disagree with him about the minimum wage.
Friday ~ July 8th, 2011 at 3:40 pm
ben
i don’t understand how conservatives and progressives can’t settle on a compromise of using government transfers to replace the minimum wage. no risk of creating unemployment and it could be done in such a way that no-one previously protected by minimum wage was worse off.
Friday ~ July 8th, 2011 at 6:47 pm
Josh W
Bryan Caplan has a couple good posts on Bayesian reasoning and the minimum wage that relates to your argument:
http://econlog.econlib.org/archives/2005/07/the_common_sens.html
http://econlog.econlib.org/archives/2005/05/infinite_contra.html
Friday ~ July 8th, 2011 at 7:41 pm
Yet even more on the minimum wage « Modeled Behavior
[...] Behavior | Tags: labor economics, minimum wage | by Adam Ozimek Pivoting off Karl’s recent posts, I want to throw my two cents in on the minimum wage. Actually, I’ll make that one cent, [...]
Saturday ~ July 9th, 2011 at 12:05 am
Tom Dougherty
As Karl Smith say, before the Card and Kruegar study all the empirical evidence was pointing the other way. So, it is not a matter of not believing empirical studies. And there were certainly a lot of questions regarding the strength of Card and Kruegar’s empirical evidence based on a survey of restaurants.
And it is true that a small increase in the minimum wage could be overwhelmed by other factors that could lead to zero impact on employment levels, but that is why it is said that ceteris paribus the minimum wage leads to lower employment. If you want to see the effect of the minimum wage you must hold other things equal. It just won’t do to say the minimum wage doesn’t cause less employment because its effect was swamped by other things going on in the economy.
There are always jobs that are profitable to hire at $4.25 and not at $7.25. Always. If employers could hire at $4.25 they could always certainly find things to be done for that wage. The question is could you find workers willing to work for that wage? With the white teenage unemployment rate around 21%, black teenage unemployment rate around 40% and the Hispanic teenage unemployment rate around 35%, I think employers could find low-skilled labor to work for $4.25 an hour. However, if employers could not find workers for $4.25, then they would have to raise their wage rate until they could find workers. But then if no one was willing to work for anything less than $7.25 an hour, then you wouldn’t need a minimum wage would you?
But why look at the lowering the minimum wage as allowing businesses to hire that would give a boost to the job market? Why not look at the other side of the equation as well and see the lowering of the minimum wage as giving prospective employees a chance at earning a wage and gaining job skills. If anyone is being thrown under the bus it is those people who are priced out of the market because their job skills are too low to earn the basic minimum required to be hired. We are saying to the most disadvantaged workers that you are better off being unemployed and remaining unemployable than to be gaining valuable job skills that can be developed which would lead to higher paying jobs in the future.
Saturday ~ July 9th, 2011 at 12:47 am
Khal Mojo
No doubt, ceteris paribus, an increase in minimum wage is likely to decrease employment, but ceteris paribus doesn’t exist in the real world, only in theory. We’ve had massive amounts of productivity increases in the last decade while wages had stayed stagnant. That means that either that flexibility has been siphoned elsewhere or wages should have risen but never did.
Saturday ~ July 9th, 2011 at 1:08 am
Tom Dougherty
But even if things are messy in the real world you still need to hold other factors constants to see the effect of x1 on y. Otherwise, you cannot say anything intelligent about x1’s effect on y if you don’t hold constant the effect of x2, x3, x4 and x5 on y as well. Econometrics comes in handy here.
Wages may have stagnated as productivity increased but total compensation did not stagnate. Total compensation and productivity rose together. By only looking at wages you ignore the rising share of benefits (particularly health benefits) as a greater and greater portion of a workers total compensation.