David R. Henderson  

Response to Arnold on Zero Marginal Product

Youth Unemployment, Continued... Best Economics Books of the Ye...

In his latest response to co-blogger Bryan Caplan on ZMP, co-blogger Arnold Kling writes:

The point that third-world workers are employed refutes the Z in "zero marginal product." I would agree that American workers could compete for jobs at third-world wages. But it still could be that marginal product is well below first-world wages.

To be clear, Arnold's first statement above is his statement of Bryan's criticism and then the next two statements are Arnold's response.

I find Arnold's response unpersuasive. There's a lot of room between typical first-world wages and third-world wages. Let's put some numbers in here. A typical third-world-wage is below $2.00 an hour. A typical first-world wage is $15 an hour. So if a wage well from $15 to $11.25 an hour, most of us would think of that as still a first-world wage. Why do I choose $11.25 an hour? Because that's a 25% cut in wages. In an earlier post responding to Bryan, Arnold had written:

Even with discontinuous changes, the phrase "drastically enough" can be invoked to suggest that wage cuts could cure unemployment. But suppose that "drastically enough" means 25 percent or more. If you want to say that the PSST story of unemployment depends on wage stickiness because such a large wage reduction could take care of things, then fine. You have scored a debating point without practical significance.

So it's not just "a debating point without practical significance." Arnold has gone from saying that a 25% wage cut could solve the problem but that's of no practical significance to suggesting that only a cut to third-world wage levels would motivate employers to hire many unemployed Americans. But that's more like an 80 or 90% cut in wages.

Comments and Sharing

CATEGORIES: Labor Market

COMMENTS (12 to date)
aretae writes:


I haven't been following the ZMP debate very well, but as you phrase it, there's a huge problem with the anti-ZMP analysis. What's the added cost (due to regulation) per added worker. My understanding, having worked in industry, is that the cost to employing someone is something within spitting distance of DOUBLE their wages. I'm also certain that the regulatory burden is not directly proportional to the wage. If a $15/hr wage costs me the employer $30/hour, I'd expect that a $5/hr wage costs the employer upwards of $18/hr. In that case, ZMP is very easily explained by regulatory burden. Worker would have to pay me $10/hr to get my (employer) costs down to the $2/hr that an employee costs in the 3rd world.

Of course...I don't have formal econ training, and I could be mistaken. Please advise if I am.

Alex J. writes:

Another way to say what I think aretae is saying is that the de facto minimum wage is much higher than the official minimum wage (and also not a simple floor). To me, this means that "zero marginal product" is a confusing term. It would be better to say that there is a price floor that matters for more workers than is the case for just the minimum wage. Without the price floor, they'd be low productivity workers, in the face of it, they have zero productivity (since they are unemployed).

If you add in generous unemployment benefits, you're probably back to talking about "marginal" productivity, but I'm not sure term "ZMP" gets people thinking about leisure on the margin. It suggests (to me anyway) notably unskilled people rather than contented idleness, home production or black market work.

Nathan Smith writes:

@ aretae:

Thanks, I was about to make a similar point but you made it better than I could. At the low end of the wage spectrum, I doubt that the money wage is all that important of a constraint on hiring. Still, that's partly "endogenous": entrepreneurs are in the habit of thinking that business plans that depend on having masses of low-wage workers for an extended period of time in the US are a bad idea. Obamacare reinforces that; so does the likelihood that at some point, the US will enjoy a strong rebound. Also, think of Bryan's "Victorian ethnography": low-skilled US workers probably don't have the docility that Third World workers do.

David R. Henderson writes:

Yes, I think you are mistaken. SS and HI are proportional to wage rates for the relevant population. The FUTA tax that the federal government imposes to pay unemployment benefits is proportional to wages up to $7,000 annually. So all that goes against your claim.
If you throw in the risk of having to defend yourself from a lawsuit from an employee you fire, then it’s probably true that the cost is a higher % of a lower wage. Is that what you are thinking?
@Alex J.
Yes, I think ZMP is confusing or worse: it’s inaccurate.

Bob Montgomery writes:

With regulation (minimum wage, minimum required benefits, taxes), perhaps it isn't actually possible to employ someone full time for $11.25/hour.

For example, minimum wage in WA (where I live) is about $8.65 /hr. Add in the various taxes (L&I, payroll tax, etc.), required benefits, I'm betting you can easily get up to $15/hr.

See here:

Though that estimate of $7000 /yr for health insurance for a family is about half of what the actual cost is. My health benefits are ~$17,000 /yr, my employer pays about $13000 of that.

David R. Henderson writes:

@Bob Montgomery,
I checked the MIT web site you reference, and it is consistent with what I said in my comment above about the taxes being roughly proportional or even somewhat (with FUTA) regressive. Also they don’t come close to making $8.65 into $15. They do if you include health insurance but an easy way around that is to keep the worker’s hours under 30 a week. Of course, ObamaCare will change that.
But we’re getting away from the point. Even if you’re right, that means that many unemployed are not ZMP; they could have substantial MP that’s below $15 an hour (benefits and employer taxes included.)

aretae writes:


To your question, yes...that is what I am thinking. See, for instance, Bob's comment with actual numbers. I worked in the Health Insurance industry last time I wasn't self-employed...and so I was looking at that.

Mostly, my claim is that there are variable costs and fixed costs to hiring someone. Lawsuit defense, regulatory paperwork, health care (doesn't vary TOO much by salary, even over a 10x salary range, costs to employer probably vary by 50-100%), etc. Near minimum wage, the fixed costs probably dominate, while up in the 6-figure professional range...it's the other way around.

Undiscussed in my prior comment...many jobs have time-to-productivity. Hiring a programmer, for instance, you can't actually assume that (s)he'll be useful in the first 3 months (at a large firm). And so...there's investment there too.

If it takes (on average) 1 month to reach net productivity above no worker at all, and a full year to reach standard productivity (those numbers made up, but highly plausible to me, hiring and observing IT staff)...that further complicates life for the temporary worker, looking to move out after job prospects improve. It means that temp workers are highly unlikely to be net-positive value.

Daublin writes:

David, I'm struggling to see the inconsistency in your two Arnold quotes. Is both cases, he says yeah, if you cut wages drastically, you can find work for people.

Noah Yetter writes:

People! It's Zero NET Marginal Productivity!

It's true, Tyler and the people he's convinced use the term "ZMP" not "ZNMP" but if you actually READ Tyler's arguments it's clear. The point is not that these workers cannot produce anything at all under any circumstances, rather it is that they cannot be profitably employed.

I see a lot of confusion about Marginal vs. Average as well. Remember that workers' marginal productivity is diminishing! Just because I have N employees producing X on average does not mean that employee number N+1 will produce X as well. In fact that is extraordinarily unlikely! He will produce strictly less than X.

The costs of employing a worker are not just wages + taxes + regulatory burden. There's also management costs, and hiring costs. Hiring is extremely expensive!

I've noticed that only academics are skeptical of ZMP. Anyone who works in business understands it immediately.

David R. Henderson writes:

Daublin writes:
David, I'm struggling to see the inconsistency in your two Arnold quotes.
I tried to do it with a numerical example, but I guess that didn’t work for you. So here’s another way. First, Arnold says, essentially, “Well, sure they could be hired if their wages fell 25% or more,” and then suggests that this is irrelevant since this is such a big cut.
Then Arnold suggests that their wages would have to be cut by 80 or 90% to make them employable. See the difference? There’s a huge difference between 25% and 80%.
Noah Yetter:
Thanks. Could you point me to the places where it’s clear that Tyler’s just being careless and really means ZNMP?

Bryan Willman writes:

Counting tax rates misses some (important) part of the costs of employing someone. In particular it misses costs of regulation and fear of the behavoir of regulators.

Consider for example OSHA. By defintion, it only has authority over employers. So if I don't hire anybody, I have no OSHA compliance issues. I have no risk of prosecution or litigation by OSHA or the employees. I have no risk of an OSHA inspector showing up on a "tip" and finding "something" to give me a citation over. (A thing reported in various places.)

Consider employment law related to racial and gender harrassment, and the point that the employer is held liable for one employee's being a particular kind of jerk to other employees. This is a constant risk, since any hire may turn out to be a jerk. Likewise, most any non-hire may decide to sue for discrimination, but that can only happen if there was an opening in the first place.

All of that shows up as a drag on the bottom line, and all of it argues for outsourcing, offshoring, or automating basically wherever possible.

(Make no mistake, workplace safety and fair treatment are very important things. But they have a cost.)

Consider also local regulations pertaining to parking and the like. If I start a new venture that is to hire people, I have to have offices and most likely parking for them. If I can manage to do the same venture with no people of my own, and can just have computers in a closet, I'm way ahead. Office space, parking spaces, etc. are basically fixed regardless of whether the employee is paid $15/hr or $11/hr.

The real point of this is that "the real cost of employment" is not completely linear with the cost of wages, regardless of what tax rates say.

OK, so what does that have to with ZMP versus Sticky Wages?

Maybe they are the same thing, in that if the minimum fixed cost of an employee is, say, $10/hr, then even someone paid $0/hr has a $10/hr cost, and if the job only produced $9/hr in value, it makes no sense to hire them, even at wages of $2/hr or $0/hr. Sticky wages would have to unstick to negative to fix this.

The developing world of course has non-wage costs for employees too, and they may be higher than we imagine, but they are likely not as high as they are here.

David R. Henderson writes:

@Bryan Willman,
All good points.

Comments for this entry have been closed
Return to top