Labor and Labor Costs: Using Theory to Understand the Data
In an interview on 1 October 2018 EconTalk host Russ Roberts and economist Noah Smith were discussing wage stagnation and workers’ share of income in the US. It’s an interesting conversation but there was one part of the segment that is worth pointing out. Between 47:00 and 49:00 Smith claims that the more concentrated industries with larger firms tends to drive down labor costs. He states:
So, there is actually quite a lot of evidence in terms of correlation; relatively thin evidence in terms of causation. That doesn’t mean that the evidence is wrong or that the phenomenon isn’t happening, or that we won’t be able to prove the phenomenon is happening. But it just means that it’s hard to know in this area.
And what you see is that in those industries where the product side is more concentrated, you’ve seen labor’s share of income also fall more than in industries where it’s less concentrated. So, in less concentrated–in industries that have become less concentrated, that have not become more concentrated, you see that labor’s share of income tends to be higher than in industries that have become more concentrated. So, there is a correlation for you.
Later Smith and Roberts get into a disagreement about lowering costs for large companies. When mentioning large firms like Tyson, Target, and Walmart, Smith points out that labor is roughly 60 percent of costs, claiming that the overwhelming majority of the time, driving down costs to increase profit margins means lowering labor costs, which equates to falling pay for workers. Roberts replies that you can lower costs without lowering labor costs, which Smith dismisses, “…on what evidence? Do you have any evidence, any evidence at all? Because I have evidence showing that labor costs go down. That point is that I’ve already presented evidence that they are holding down wages.”
Roberts says he has intuition, to which Smith snickers. But his intuition is sound (starting around 58:00):
I’m an employee of Target, K-Mart, local firm; and my firm goes out of business. Now, I’m stuck working at–I have a smaller set of retailers who can employ me.
My argument, by the way, would be that it’s easy to apply technology to lower all the costs. And let’s think about that slowly. I can employ technology to reduce my inventory costs. It’s obvious that all these firms have done that. I can also employ technology to lower my labor cost. And the way I do that, is I pick a different mix of workers.
I find a technology — to take an obvious example — if I’m McDonald’s, and I can have my cashier push a button rather than actually work a cash register, I don’t have to have as skilled an employee. I’m going to have lower wages as a result. But I’m not squeezing anybody. I’m taking a different kind of worker than I had before. But the point I was trying to make before you jumped down my throat with that empirical evidence … is that — and this is my Tyson’s point, also: Just because I can’t work at K-Mart any more — K-Mart actually is out of business — doesn’t mean that I’m stuck working at Walmart.
Just because my Tyson opportunities have gotten smaller, my chicken-plucking opportunities are getting smaller, is that most of the people who are working in these industries we are talking about don’t have very specialized skills in these areas. They have lots of employers they can work for. Their wages are low because their skills are low. Not because they are being squeezed. That’s my point.
Smith concedes that point before moving on to other complications of the argument. Now, while Smith correctly points out that wage growth has stagnated (slowed) in the last 20 years, Roberts’ point is important. As long as labor is mobile enough to switch between firms and industries, then we must look at the heterogeneous nature of labor in the same way we should with capital. Different kinds of labor and different sets of skills must “fit” into the proper arrangement of production. Workers are not getting “squeezed” if their skills allow them to shift into other areas of employment. Of course, there is always the case of frictional unemployment for various reasons. But with a change in technology to reduce costs, a different “mix” of less-skilled labor can be inserted into a given company so that a reduction in the pay of that labor is commensurate with the reduction of the skill level relative to the previous employees.
Thus, we can see that while the empirical evidence seems to favor Noah Smith’s position that labor is receiving a smaller share of income in industries with less but larger firms, the intuition by Russ Roberts can better point to how this perspective can be misinterpreted from the correlations observed. Put simply, theory can help explain causality when interpreting the data.