When Does a Labor Economist Ask for a Raise?

Little persuasion happens in 280-character snippets, but people willing to explain their thinking and answer each other’s questions can still accomplish a lot by clarifying their views and identifying the underlying sources of disagreement. So I was delighted yesterday when the Cato Institute’s Alex Nowrasteh took the time to walk me through his understanding of how wages are set in labor markets.

I’ve transcribed the exchange below for ease of reading, and would offer a couple of observations:

First, it’s important to recognize that Alex’s argument is the one underlying much of libertarian economic thinking, and thus the economic policy agenda that has dominated on the right-of-center in recent decades. Does labor economics say that firms know the marginal value of each worker and compete to pay him exactly that? Will capital rush in to any void and create new employment opportunities that put each worker to his most productive use? If you answer yes to those questions, you’ll have a very particular understanding of how the economy is operating and what if any public policy might be appropriate.

Second, there’s an apparent inflection point when we move from Alex defining his view to me asking how this would work in practice. This is when he stops answering my questions, starts questioning whether I understand the difference between marginal and average, and then logs off. That was genuinely disappointing. If you’re going to have confidence in a very formal and abstract theory of how the labor market functions, you should be able to apply it to the actual world in which economic actors operate.

Below, I describe the questions I asked and had planned to ask Alex, and explain why I see them as important. My hope is that he will continue the dialogue—I’d welcome any corrections on my own economic analysis or my understanding of his, and we’d be delighted to post a further elaboration from him here on The Commons or to link to something he might write elsewhere.

The Twitter Discussion

AN: There is no substantive economic debate here. On one side, you and Lind reject economic basics. On the other side are people who are 1) calling you out on it and 2) trying to explain econ basics to you.

OC: Are you saying “marginal value product” is by definition equal to whatever a person is paid, or that it’s independently knowable?

Say I sell widgets for $20, and hiring another person lets me produce an extra widget each hour. I pay him $10 per hour. What’s his MVP?

AN: “Are you saying “marginal value product” is by definition equal to whatever a person is paid, or that it’s independently knowable?”

No, nobody is saying that. As I wrote in my blog post: [link] [continuing] So, in your example, the MVP is $20.

OC: But I’m only paying him $10 per hour; no one else in town is offering more. Is your assumption that someone else will come along and build a new factory and bid his wage up to $20?

AN: You’re assuming the conclusion and working backward to show that the theory, which is grounded in reality, is wrong. Your hypothetical is classic example of “begging the question,” a species of fallacious circular reasoning. Also guilty of the appeal to fiction fallacy. [continuing] You can’t make up a fictional story, beg the question, and say “see, I told you I was right. Labor economic is wrong.”

OC: Sorry, I guess I thought “person is currently earning less than his MVP” was a basic/realistic scenario that labor economics could handle. Are you saying that in labor economics everyone is always earning their MVP already?

AN: You want labor economics to handle fiction with a predetermined ending that you wrote to suit your opinion?

Ok, this is what would happen: The worker wouldn’t take that job in the first place and that position would go unfilled at that wage.

OC: Why wouldn’t he take the job? Because he knows his own MVP (perhaps he does some widget consulting on the side…) and so is insulted by the offer?

AN: Because he can get paid more elsewhere. A firm could profitably hire him away for a wage >$10 and <$20. The worker is rational and self-interested so he’s going to take those offers.

[continuing] Oren, in your example, let’s say the other workers employed by the firm in question also have an MVP of $20 each and are paid $20. In your understanding of the world, why wouldn’t the employer just cut their wages down to $10 an hour?

OC: Seems unfair I can’t assume a worker earning less than MVP while you get to assume some other firm stands ready to hire him away. But so stipulated. My imaginary firm was paying everyone $10/hr. Sounds like we should assume the firm knows and pays each worker his individual MVP.

AN: You assumed the conclusion and presented it as a thought experiment. I’m not assuming a conclusion, just assuming that there is more than one potential employer and there isn’t a perfect monopsony in the labor market.

OC: I guess my question then is: aren’t I losing a lot of money? If I pay workers who can make one widget per hour $20, and I sell the widgets for $20, how do I cover my material costs? My overhead? My very-important-to-me shareholder return? This seems like a bad business model.

AN: Wait a second, do you not know the difference between average and marginal?

OC: No, I’m good there. Are we paying each worker his individual MVP or are we paying them all the $20/hr of my new hire?

AN: Since you know the difference between marginal and average, you tell me. What would theory predict?

OC: Well I’m torn. I think I’m going to have to pay everyone their individual MVP, otherwise I’ll have a big adverse selection problem. But that doesn’t seem especially workable. Either way I’m concerned about how I’ll afford my materials. There’s about $10 worth in each widget.

AN: Wait, I thought you said you knew the difference between average and marginal?

OC: Sure, marginal refers to the change in the economics when another unit is produced, average is the sum across all units divided by the total quantity produced. I’m honestly unsure of how you’re going to apply your MVP theory to my shop full of workers, though, which is why I ask.

AN: I’ll give you an example. I don’t regret this conversation at all, it was polite and fun. But at this point, the marginal cost has exceeded the marginal benefit so I must “quit” and bid you good night.

OC: OK, well I appreciate your clarifying some of your thinking on MVP. It would be very helpful to know whether you’re suggesting I pay all my workers the same wage (if so, what?) or if each gets paid his own MVP. I’d thought that would be a straightforward question. Maybe tomorrow.

AN: It’s the marginal value. But it isn’t my thinking at all, I can’t take credit for it. It’s just your common dime store every day labor econ. Not my creation, I can’t take credit for it.

OC: I’m sure you’re not being intentionally opaque, but “the marginal value” is unclear. Each worker is paid his own “marginal value product,” or all workers are paid the $20/hr you told me to pay this new hire?

The Key Questions

Am I supposed to pay my workers the same or as individuals? A major challenge for MVP theory, it seems to me, is that it fails to describe how we observe labor markets operating in the real world. Firms tend to pay all workers in a given role the same amount, or at least keep them in fairly narrow “bands,” and those bands also appear remarkably consistent across firms. This should not be possible in the world Alex describes, where firms know the value of each worker and compete to pay him accordingly. One possible outcome would be extraordinary variability in worker pay. Another would be that each firm adopts a particular salary band and hires only the workers within that band, leading to extraordinary sorting across firms, with each selecting the workers whose actual productivity fits right within its particular band. Neither seems prevalent in practice. And so the question remains: How should I pay my workers? Do I determine the individual productivity of each and pay him accordingly, or do I pay them all the same? And how will they and other firms react to my choice?

How do I make money paying a worker the market price of the good? It seems unlikely that a worker’s wage would equal the market price of what he produces, ignoring the many other variable costs like materials that might also go into the production process. At a minimum, I think we would need to know the marginal profit that the firm expects from the worker. So in my imaginary firm, where I sell widgets for $20 and each one has $10 of materials in it, I can’t pay someone more than $10 to make one.

What to do with all those workers whose productivity cannot be measured on a per-unit basis? My firm has widget designers, widget machine-tool technicians, widget-factory custodial staff, widget marketing executives, and of course yours truly, widget CEO. How much do we get paid? If I understand correctly, adherence to MVP theory would require that firms determining the value of all such contributions to each widget bid each wage up to its proper level. This seems difficult. How much should go to the engineers versus the line workers? Of course, “true” productivity will come into play over time as firms adjust and offer more or less relative compensation for various roles to optimize their talent mix. But as a manager, I’m probably going to ask myself, “what am I going to have to pay to hire the people I need”—that is, how many prospective workers are out there, how urgently I need them, and what other firms are paying is going to have a lot of influence on what wages I offer. Put colloquially: the right time to ask for a raise is not when (and only when) your marginal value product has gone up.

Does MVP increase with price? One might think that I have no choice but to hold wages at the level that allows me to earn a profit on my $20 widgets, but this takes the $20 price as a given. If I have difficulty attracting workers at the wage implied by my price, I could raise my price. This of course assumes I have some pricing power in the market, but as anyone who has ever shopped for almost anything knows, most firms do in fact set their own prices and they vary considerably. If I raise my price and pay workers more, other firms might react by doing the same—unless of course there are plenty of extra workers available after I’ve hired mine. How this plays out will depend not only on the labor market but the end-product market as well. That it does play out seems clear—just look at the wage increases in response to the “tight labor market” of 2018-19, far above reported productivity gains.

How much profit will I earn? Finally, we get to the question of return on capital. How much I and other firms can pay our workers must depend in part on how much we intend to return to our shareholders at year’s end. In the situation above where I need to increase wages to attract workers, I might do that without raising price by simply accepting a lower profit margin instead. My willingness to do this will depend on my capital structure, need for access to capital markets, shareholder expectations, long-term strategy, and more. An related scenario is a firm accepting losses in pursuit of growth and market share. Even when taking a loss on the marginal unit, or while in a start-up phase producing no units, it still manages to set positive wages.

In summary, the wages offered by my firm and others to a given worker will depend on not only his measurable productivity but also the other workers available to me, the other options available to him, my other input costs, my approach to setting wages for a given role, the relative wages I pay workers in different roles, elasticities in the end-product market, my need to attract capital and its cost and availability, and other factors besides.

And so my questions to Alex boil down to this: When assigning attribution to the forces that set wages in the market, what weight goes to productivity, and which others factors count as well? I think we agree that productivity is the single most important factor. But I think many others matter too. Does he? It’s crucial that policymakers avoid reliance on simple models disconnected from the real world.