Scott Sumner  

What empirical evidence should we trust?

What about Capital?... Dr. Pritchett's Six Bitter Pil...

Noah Smith has a piece in Bloomberg discussing recent empirical research, some of which (not all) suggests that higher minimum wage rates do not have negative effects on employment. He then suggests that these studies also discredit the competitive model of labor markets, in favor of alternative (monopsonistic) models where employers have lots of market power.

Textbook writers and instructors should respond by changing the baseline model of labor markets that gets taught in class. Students ought to start with a model of market power, in which a few companies set wages below levels found in a competitive market unless prevented from doing so. That model is about as easy to work with as the traditional supply-and-demand setup, but matches the data much better.

I'm not totally convinced by this line of reasoning, for a number of reasons.

1. The replication crisis in the sciences, and the social sciences.

2. Conservative studies seem able to explain the very low levels of hours worked in Europe much better than progressive studies. And the studies that Smith cites are progressive studies. That doesn't mean progressives are wrong, but until progressives are able to explain Europe's labor market, I'll continue to have trouble taking them seriously.

3. Most importantly, Smith overlooks the fact that empirical research is just as unfriendly to the monopsony model of labor markets as it is to the competitive model of labor markets. AFAIK, almost all the empirical studies of the minimum wage suggest that higher minimum wages do not come out of profits, but rather are passed on in terms of higher prices. That result is 100% consistent with the competitive model of labor markets, and inconsistent with the monopsony model (which suggests that if employment doesn't fall then product prices do not rise.)

Thus empirical studies show that when a minimum wage increase forces grocery stores to pay higher wages, they pass on the increased costs in the form of higher prices.

Now I suppose that progressives could argue that demand curves don't slope downwards, and that the higher prices will not reduce sales. In that case, a higher minimum wage need not reduce employment. But as soon as you abandon downward sloping demand curves, you are faced with other dilemmas. For instance, why should progressives oppose "regressive" consumption taxes? After all, if demand curves don't slope downwards, then higher prices would not reduce consumption. And since living standards depend on consumption, regressive taxes would not reduce the living standards of the poor.

Of course we know that demand curves do slope downwards, and we know that regressive taxes tend to adversely impact the poor. What we need to figure out is whether higher minimum wages raise prices and reduce sales. So far, the empirical evidence suggests that they do.

To summarize, the empirical evidence on the effect on minimum wages on employment is mixed. The empirical evidence on the effect of minimum wages on prices is pretty clear---it raises prices. That means that, on balance, the empirical evidence is more supportive of the competitive labor market model than the monopsony model.

This doesn't mean that firms have no monopsony power---they almost certainly have some. The question is how much, and whether the short and long run labor demand elasticities differ.

I would add that the question of whether higher minimum wages are desirable is very different from the question of whether they affect employment levels. There are other important issues to consider, such as the impact of minimum wage laws on working conditions.

Comments and Sharing

COMMENTS (22 to date)
Iskander writes:

Scott, why does the monopsony model predict that higher costs are not passed on to customers?

I haven't seen anything that explains this and haven't been able to think my way through it. I don't doubt it, I would just like to see it explained.

Scott Sumner writes:

Iskander, Because if employment does not fall, then we can assume that the marginal cost of production does not rise. And if the marginal cost of production does not rise, then firms have no incentive to raise prices.

Don Boudreaux writes:


Nice post. I add four points that I've made in the past at my blog.

First, as Jim Buchanan, Donald Dewey, and other economists have pointed out, as long as demand curves for outputs are downward sloping, monopsony power is only a necessary and not a sufficient condition for minimum wages not to reduce the employment prospects of low-skilled workers. For minimum wages not to reduce these workers' employment prospects, employers with monopsony power must also have monopoly power (and not just the sort of such 'power' as is identified in models of monopolistic competition). That is, these employers must have the ability to keep the prices of the outputs they sell above average total costs. If they do not have this ability, then there are no excess profits out of which the higher labor costs can be paid.

Second, empirical studies typically fail to examine all the many ways that employers and employees can adjust to minimum wages. The list of such possible adjustments other than reduced hours of employment includes reductions in formal fringe benefits (such as paid leave), reductions in informal fringe benefits (such as workplace safety higher than what is minimally required by legislation), and changes in the nature of the jobs such that workers are worked harder in order to produce more output per hour. To the extent that adjustments such as these occur, minimum-wage-induced reductions in employment will be fewer or lower, but the standard textbook model really still holds.

Third, because in the U.S. the national minimum wage has been in place now for 80 years and is at no risk of being repealed, employers have long ago adjusted their business plans - their capital-labor ratios - to the existence of minimum wages. And employers expect occasional minimum wage increases. Therefore, even the finest and most carefully controlled empirical study of a minimum-wage hike today will not detect the employment-reducing effects of the long-standing expectation of minimum-wage hikes. Because employers have already adjusted to the reality of minimum wages - and to the reality of minimum wages being increased from time to time - any study that correctly finds little or no negative employment effect from this or that minimum-wage hike today nevertheless misses the negative employment effects of minimum wages overall.

Fourth, about monopsony power: it's more difficult to detect than, ironically, standard textbook models suggest. Suppose that Acme, Inc., competes for workers by offering unusually attractive fringe benefits and work conditions. And suppose that Acme, Inc., has a differential advantage over other employers at supplying to its workers such non-wage amenities, or that for Acme, Inc., the marginal cost of attracting X number of workers by supplying non-wage amenities is lower than is its cost of attracting X number of workers by increasing the wages it pays. Under such conditions, Acme, Inc., gains the power to lower its workers wages by some amount without losing all, or perhaps even any, of its workers.

An empirical study of this firm would conclude that Acme, Inc., has monopsony power. But this conclusion would be incorrect, for the 'power' that Acme, Inc., is detected to have over its workers is 'power' that Acme, Inc., purchased from its workers - workers who voluntarily agreed to Acme's employment terms.

Put differently, if (as is not unreasonable for many employers) Acme, Inc., values a steady workforce, it can purchase - with non-wage amenities - from its workers the ability to cut their wages without their quitting. The textbook-bound economist, seeing only the reduced wages and no mass exodus of workers from Acme, leaps confidently to the conclusion that Acme has monopsony power. Yet clearly, in this example, that conclusion would be mistaken.

Jason S. writes:

US economists have largely studied restaurant employment, where labor markets are apparently most monopsonistic: Other low-wage industries, like manufacturing of food products and most services, are close to perfect competition. So the literature may be underestimating the negative employment effects of minimum wages.

However, I also don't get the point about prices. Under monopsony, minimum wages raise marginal & average cost, right? Employment rises despite the rise in cost because the establishment-level labor supply curve is upward-sloping rather than perfectly elastic.

Michael Pomatto writes:

Studies that I've read suggest that any decrease in employment from min. wage hikes happens but is inconsequential generally. It suggests that there is relative inelasticity for the demand of labor. Some localized markets seem to experience greater effect. I agree then that we can assume that costs are passed on to consumers.

Miguel Madeira writes:

I think the best way to reconcile the two observations ("minimum wages don't seem to raise unemployment" and "minimum wages seem to be passed to prices") is to assume that workers who earn minimum wage are strong consumers of products made by workers earning minimum wage.

LK Beland writes:

What if firms were simply intermediaries? In some sense, the demand for labor is created by consumers, not firms.

In that case, one could argue that the minimum wage provides the suppliers--labor--market power over the consumers. Essentially, it lets the low-wage workers form a cartel.

Then, depending on the elasticity of demand on the consumer side, you could find a non-zero minimum wage that maximizes the cartel's income.

Does this make sense? It would explain a lot of the data, no?

Daniel Kuehn writes:

"The question is how much, and whether the short and long run labor demand elasticities differ."

Should that be labor supply or am I misunderstanding?

michael pettengill writes:

Doesn't lower incomes cut consumer spending, reducing GDP, killing jobs if prices are not cut to the bone, ie, the lower labor costs?

Cut wages in half, and prices must be cut in half to maintain the quantity of trade.

Of course, debt, which is labor cost time shifted, throws a monkey wrench into the mix. If wages are cut, prices must be cut, plus debt principle must be cut to cut the wage paid in the past to current wage levels.

GDP growth has been lower when wages are held down or cut, because wages put money in consumer pockets, not tax cuts, lower prices.

And by definition, cutting consumer spending cuts GDP, unless offset by saving future or past wages in debt.

Putting wages in equity is consumption, which for accounting purposes is spread over the life of the durable good consumed.

But debt is merely shifting consumption from the worker to someone who consumes more than they earn, or the reverse when repaying debt.

Higher minimum wages can't help but increase GDP, almost certainly increasing quantities consumed immediately. Why a business would layoff workers when the quantity of goods sold almost certainly increases is a puzzle. Perhaps they want to give Wal-Mart greater market share while reducing consumer choices.

I note the current attack on the CFPB is over the rule that loans be made only to workers who can repay the debt on time. Those advocating lower wages in general, also argue for more bad consumer debt to drive up or maintain quantities consumed even as reduced incomes and higher profits require reduced quantities consumed.

I grew up when debt to fund consumption was very very bad economic policy. Debt required offsetting durable goods, assets.

Economies are zero sum. Cut wage incomes, cut consumption. Increase wage incomes, and consumption MUST increase. If the two are not linked, the economy ceases to be sustainable. E.g., Saudi Arabia and Venezuela, where consumption is much higher than wage income.

Rajat writes:

Defenders of the monopsony model might argue that Europe's minimum wages are set too high, whereas the US's are not.

I agree especially with Don's third point - firms get used to periodic increases in minimum wages and don't necessarily react when they actually occur. Kind of like the interest rate does not define the stance of monetary policy.

A few other points:

1. The monopsony model predicts small increases in minimum wages raise employment, not just keep it stable.

2. Nick Rowe has made the point that if wages are set below the competitive level, we would not observe involuntary unemployment - yet we do.

3. Relatedly - and maybe it's wrong to mix micro and macro - I find it hard to reconcile the monopsony model with your musical chairs model (which I support), in which employment rises when w/NGDP falls and the reverse. If (real) wages were below competitive levels due to monopsony power, why would effectively reducing the real wage (or at least, real purchasing power over society's resources) raise employment?

Rajat writes:

Actually, scrub my third point. Even in monopsony, a down shift in labour supply due to NGDP rising faster than sticky nominal wages would increase employment.

Scott Sumner writes:

Don, Very good points.

Jason, Average costs rise, but not marginal costs.

Daniel, I meant labor demand, which may be more elastic in the long run.

Rajat, Just to be clear, I was not referring only to Europe's minimum wage laws---some European countries don't even have any.

Everyone, You can't have it both ways. If minimum wages are passed on to consumers, then employment falls.

Alternative writes:

My guess is that a lot (not all) of firms would be better off raising wages to attract more and better workers a lot of the time. They don't do it because managers are afraid of this sequence of events:

1. There will be a recession in the future and their workforce will become overpaid.

2. It will be very hard to reduce nominal wages .

3. There will be layoffs, including managers. Or the whole restaurant closes or whatever.

So employment does not go down after the min. wage hike because management was underpaying workers to hold on to the option of paying low wages during the next downturn. The higher min. wage takes away that option, but does not make them want to employ fewer people right now.

Of course min. wage laws are less effective than preventing equilibrium nominal wages from going down at all. Like if they could keep some nominal quantity that tracks national income on a fairly smooth upward trajectory. Someone should advocate for a policy like that.

Mark Z writes:


"You can't have it both ways. If minimum wages are passed on to consumers, then employment falls."

Only if the goods being produced by minimum wage employees have elastic demand. If demand is inelastic (and markets are competitive) the employers can indeed pass the cost of the minimum wage onto customers without employment falling.

So if, say, fast food, being an inferior good, is much cheaper than the alternatives, and this is not changed by the price increase caused by the minimum wage increase (and assuming predominantly low income or cost-conscious customers), demand for fast food may remain more or less unchanged even as prices increase.

I don't know what empirical evidence there is on this, but it is often said that taxes on fast food are ineffective at reducing consumption for this reason.

Paul writes:

It's possible that there's a difference between short run employment effects and long run.

It is suggested that gas taxes don't immediately impact car efficiency - because you don't rush out the day the tax comes in and buy a new car. But when you next buy a car, you might buy a more efficient one. So near term gas demand is relatively inelastic, but over time it is more elastic.

Perhaps, similarly for minimum wage increases. If I put up the minimum wage today, it's unlikely a restaurant owner will rush out and replace her workers with kiosks to order food. But maybe she'll be more disposed to that the next time she replaces her point-of-sale systems. So near term inelasticity (you've still gotta run the business) but medium term you make capital investments that are less reliant on labour.

By the way, this isn't necessarily bad if a new equilibrium is reached that uses greater capital, and therefore has increased productivity. It's only a problem if there aren't other alternative businesses opening up, and therefore not enough alternative employment. In fact, in a perfect world employers WOULD invest more capital and increase the productivity of their workers, thereby making their workers worth the new minimum wage.

Scott Sumner writes:

Mark, You said:

"Only if the goods being produced by minimum wage employees have elastic demand. If demand is inelastic (and markets are competitive) the employers can indeed pass the cost of the minimum wage onto customers without employment falling."

That's not accurate. As long as demand curves slope downward, employment will fall.

Radford Neal writes:

Maybe employment could go up after a minimum wage increase because the increase constitutes free advertising that employers are paying at least that much? Then people who hadn't realized how much McDonalds had been paying before (suppose it was at least the new minimum) now do know that, and may apply for a job.

DeservingPorcupine writes:


Basic question here. You say

This doesn't mean that firms have no monopsony power---they almost certainly have some.
In a competitive market (like, say, fast food) is the monopsony power you speak of just that of normal employment frictions/transaction costs, or do you have something else in mind?


Kent writes:

Here's the thing I never understood about these arguments:

If a business can so easily raise prices on consumers when minimum wages increase, why didn't they price their products higher before the increase?

The assumption must be that the higher price will be sufficiently offset by a reduction in demand to actually reduce profits.

But then, logically it would follow that by reducing prices, they would increase demand sufficiently to increase profits.

Therefore, increasing minimum wages should naturally lead to a decrease in prices in order to maximize profits. QED

Mark Z writes:

Scott, for a perfectly inelastic good the demand curve doesn’t slope downward, it’s just a vertical line.

S D writes:

Many years ago I was in a restaurant in Belgium.

Belgium has a minimum wage but also very high employer social security contributions so the cost of employment is more or less the highest in the OECD.

Restaurants have as few staff as possible, and you wait a long time for an order.

They also substitute capital for labour. Many restaurants have moved to electronic ordering systems which communicate by bluetooth with the kitchen.

Anyway - this particular day in the restaurant I was in the electronic system broke down for whatever reason and couldn't be fixed. The manager came out and basically said that anyone who didn't have food on their table wouldn't be getting it. The kitchen would close as they simply couldn't revert to a manual system. Please go away and come back another day!

So we went somewhere else to eat.

VP writes:

This was a super interesting post, because it's something I deal with every day. I operate several small businesses in California with about 80 employees, most of whom earn the minimum wage.

I'm not an economist, and I don't understand all the points being discussed, but I can tell you what I do when the wage goes up every year.

1. Raise prices. For every dollar increase in the minimum wage, I raise my prices 8 to 10%. In some of my businesses, I am not able to pass along the increase because of the competitive nature of the industry, so I resort to some of the other measures below.

2. Reduce non-wage benefits. I used to pay three weeks of vacation a year, have a fully-funded profit-sharing plan, provide a dental and eye-care reimbursement, provide all my employess with lunch every day, pay 100% of their health insurance premiums, and many more. Over the past 15 years, I have reduced or eliminated all of these perks.

3. Forgo wage increases for higher earning employees. I've been unable to raise wages for my higher earning employees for some time. When they come to me asking for a raise, I generally say no. I am able to keep good employees around because we have a great company culture and it is an enjoyable place to work.

4. Forgo maintenance on equipment or put off investing in new equipment. I have lots of equipment that needs replacing. We just keep repairing old equipment and dealing with the headaches. This often results in poorer product quality, but often to a degree that is not noticeable to consumers so we tolerate it. If I didn't have to invest in higher wages, I would buy the equipment in a heartbeat.

5. Automate. I may not be able to afford to replace current equipment, but it usually pencils out if I can purchase equipment that automates our systems. I was able to acquire equipment that could reduce our workforce significantly. It didn't make sense 5 years ago, but now it's a worthwhile investment.

6. Reduce hours of current staff. I may not actually let people go, but I can do some creative shifting of schedules, and often just ask salaried employees to take care of certain tasks.

7. Cost-cut mercilessly. I look for any and all ways to save on the products and services that I use to run my business. So you ask, don't businesses do this all the time anyway? The answer is no. There is limited time in the day, and the opportunity cost of trying to save a few bucks here and there is often not worth it. If a service is working for you, then why mess with it? But when forced by increasing labor costs to save elsewhere, you start cost-cutting with a vengeance.

8. Let people go. I never do this immediately. It's too risky to terminate employees, especially in California, unless they are performing so poorly that they are detrimental to the business. And really, the employees I hire are determined by the business coming in the door. Sometimes my price increase will allow me to maintain my sales at the same level, but with decreased volume. Therefore, I don't need as many people and can cut staff. It takes time to determine this, and at some level, you need a minimum amount of personnel to run the business, unless you automate.

9. Cut executive salaries, accept lower profitability, reduce return to investors. These are things that I am sure many people are delighted to see. Most often it never gets to this point because I will exhaust every other option before doing this, as I am sure is the case with most business owners.

And these are just some of the things that can be done. Sometimes we see the need for new staff, and we just forgo the hire indefinitely since we can't afford it. Also, I can often replace a salaried employee with another hire that is more productive at the same wage.

The other point I'd like to mention is that I think many underestimate the power of the black market for labor. When your company is of a certain size, you have to play by the rules. The legal penalties for doing so are too risky. But I know of many small businesses that just hire people and pay them cash under the table. Some of the difficulty in measuring the true effects of the minimum wage should take account of the fact that many people work below the minimum wage--it just isnt measured.

I realize my experience is just one data point, but thought it might be helpful to share.


Return to top