On Legislating ‘Living Wages’

by Don Boudreaux on January 15, 2006

in Prices, Reality Is Not Optional, Regulation

Jon Gertner’s cover article in today’s New York Times Magazine offers no surprises. It’s as predictable as what you’ll read here.

Gertner reports on the movement for a “living wage” – a movement working mostly at the local level, trying to persuade local governments to legislate minimum wages well above the national minimum wage of $5.15 per hour.

Of course, Gertner mentions the famous (or infamous) 1995 study by David Card and Alan Krueger in which the authors claim to find evidence that a hike in New Jersey’s minimum wage caused no fall in employment and might even have caused employment to increase. Card and Krueger are quoted, each predictably insisting that their data exposes the error in the standard economists’ analysis in which legislated minimum wages increase unemployment among low-skilled workers.

Gertner makes no mention of the many critical responses to Card’s and Krueger’s study. (He does quote critic David Neumark – once – but only on the success of the political movement for ‘living wages,’ not on the Card-Krueger study.) A good summary of the criticisms of Card-Krueger is this essay by Donald Deere, Kevin Murphy, and Finis Welch.  (I must also recommend this outstanding blog-post by my colleague Bryan Caplan, over at EconLog.)

Maybe Gertner can be forgiven, for Krueger told him that (in Gertner’s words) “Some recent surveys of top academics show a significant majority now agree that a modest raise in the minimum wage does little to harm employment.” If a “significant majority” of “top academics” accepts the Card-Krueger finding, why bother to talk with the lunatic fringe who cling to the archaic superstition that higher minimum wages cause greater unemployment?

But look more closely at what Krueger told Gertner – namely, “top academics” (we’ll assume those to be economists and not sociologists and experts on French literature) “agree that a modest raise in the minimum wage does little harm to employment.”

It seems as if Krueger is saying that economists now agree that “modest” increases in the minimum wage are justified, or at least not harmful. That is certainly Gertner’s reading of Krueger’s meaning.

But taken literally, the economist consensus that Krueger reports is neither new nor inconsistent with the traditional understanding that higher minimum wages reduce employment. This understanding has nothing to do with the extent of the effect that higher minimum wages have on employment; it has to do with the direction of the effect: higher minimum wages, lower rates of employment of un- and low-skilled workers.

Because more than 95 percent of American workers have skills sufficient to enable them to earn wages higher than the federal minimum, any “modest raise in the minimum wage” is unlikely to have an effect on employment that is more than modest.

But let’s assume that Krueger really means something more substantive – namely, that economists have been wrong to assert that higher legislated minimum wages increase unemployment – and that, therefore, the economists’ case against the minimum wage is without merit. (Again, this meaning is the one that Gertner takes from Krueger.)

Because there are margins other than pecuniary wage rates upon which employers and employees can adjust, it’s possible that higher legislated minimum wages don’t so much reduce employment as reduce the quality of available employment. Most obviously, employers of low-skilled workers can reduce (or not increase) fringe benefits, or they can work their employees harder (thereby extracting more output per hour from them).

If the brunt of employer and employee adjustments to higher minimum wages take place on these non-wage-rate margins, even the finest empirical studies will show little or no effect of higher minimum wages on the rate of employment.

But it would still be a grotesque error to conclude that low-skilled workers are generally better off as a consequence of the higher minimum wage.

Be Sociable, Share!



44 comments    Share Share    Print    Email


Russ Nelson January 16, 2006 at 3:04 am

…. and yet people make that error all the time.

Randy January 16, 2006 at 8:21 am

But the people calling for the "living wage" are the very people being hurt by it. The politicians who put forward such legislation are responding to the wishes of their constituents. In a democracy, people should get what they want.

If a living wage drives further automation, thus creating more jobs for higher skilled workers, and the lower skilled workers want the living wage anyway, why should I care? What is the net effect of the living wage on GDP?

mark January 16, 2006 at 8:39 am

"But the people calling for the "living wage" are the very people being hurt by it."

Usually the people calling for it are not the people who earn below it. Some of them may want it but they're an electorally weak group. The people who have the power to get these measures past the lobbyists are quite often they are unions whose members earn above the minimum wage, and desire it for the reasons you state Randy.

I think about it this way. If we shipped all the people earning below the minimum wage to another country, then would it make sense to trade with that country? It would so we should all be better off if those people are allowed to work.

N.B. The same unions who oppose the minimum wage also oppose free trade – I think the American expression is "go figure".

spencer January 16, 2006 at 9:58 am

Here is the data on the minimum wage
and the number of people employeed at
or below the minimum wage for the US
since 1979. Sorry, the data does not go
back further and is only available on an
annual basis.
minimum min wag
wage employment
1979 2.90 6912
1980 3.10 7773
1981 3.35 7824
1982 3.35 6496
1983 3.35 6338
1984 3.35 5963
1985 3.35 5538
1986 3.35 5060
1987 3.35 4698
1988 3.35 3927
1989 3.35 3162
1990 3.69 3228
1991 4.14 5283
1992 4.25 4921
1993 4.25 4332
1994 4.25 4128
1995 4.25 3656
1996 4.38 3724
1997 4.88 4754
1998 5.15 4427
1999 5.15 3340
2000 5.15 2650
2001 5.15 2174
2002 5.15 2146
2003 5.15 2100
2004 5.15 #N/A

As the data shows the correlation between increases in the minimum wage and the number of people employeed at or below the
minimum wage is positive. Actually, if you look at the data the only time minimum wage employment increases is when the minimum wage increases. When the minimum wage is left unchanged the number of minimum wage employees falls. Sorry the data is so tight and hard to read — will be glad to send you the data and/or a chart to clearly demonstrate the points.

Interestingly, if you look at the data on teenage participation rate you get exactly the same relationship. Over the last quarter century the teenage participation rate has fallen sharply except in the years when the minimum wage was raised.

If you theory and comments on the impact of the minimum wage are so good, why does the data say they are wrong?

John Pertz January 16, 2006 at 10:10 am

O.K Spencer then if the min wage is not harmful to the working class then alow me to ask a simple question. Why not raise the fed min wage to twenty dollars an hour, hell if there are no effects why not 30, 40 or even 100 dollars an hour? I am dieing for a response.

Stefan Karlsson January 16, 2006 at 10:56 am

I don't think a modest increase in the minimum wage, like the proposed increase from $5.15 to $6.25 that was recently defeated in the Senate would have much effect on unemployment, simply because continuing inflation have eroded the ral value of minimum wages so much that even after such a increase the minimum wage would be fairly low (In France the minimum wage is nearly $10 per hour). But this on the other hand also means that very few would actually get any wage increase from such a minimum wage increase.

Spencer's statistics on how the number of people who live on minimum wages increases after minimum wage increases do not show what he thinks they show. The reason the number of people on minimum wage go up is that before the increases they affect much more workers. The relevant comparison between 1995 and 1997 is not the number of people who lives on $4.25 in 1995 and the number of people who lives on $5.15 in 1997, but the number of people who lives at or below $5.15 in 1995 and the number of people who lives on $5.15 in 1997.

If you lowered the minimum wage to 25 cents, the number of workers on minimum wage would likely fall down to zero or nearly zero, but this sure doesn't mean that fewer are employed

spencer January 16, 2006 at 10:58 am

I am not making an argument that it is either good or bad. I am just asking why the data never supports your case.

The counter argument — supported by the evidence — is that moderate increases in the minimum wage does more good then bad.
The argument I hear here it that it does no good, only bad.

spencer January 16, 2006 at 11:07 am

Stefen — the data is not of the number of people that live on the minimum wage — it is the number of people that are paid the minimum wage. Since an extremely large share of the people that are paid the minimum wage are teenagers and/or people that have multiple jobs and/or other family members employeed the two data sets would be very different

Stefan Karlsson January 16, 2006 at 11:28 am

Spencer, by "living on" I meant "paid the minimum wage", not that it is necessarily their only source of income. And it have no relevance for my basic point about how employment of low wage workers likely falled as the increases that your statistics showed dealt with two partially different groups of workers.

RightThinker January 16, 2006 at 12:35 pm

I have this question for liberals: if raising the minimum wage is the solution to helping the working poor, why should any increase be modest? Why should we economic doctors every stop doling out the cure? Right-thinking people like us oppose ALL tarriffs; we don't ask the Congress to lower all tarriffs 25%, we demand their elimination. If liberals believed in their own remedy, they'd ask Congress to raise the minimum wage to astronomical levels.

spencer January 16, 2006 at 12:45 pm

Sorry rightthinker I asked my question first. When you answer mine I'll answer yours.

SK Peterson January 16, 2006 at 1:25 pm

Part of the problem with the data you have (and I think what Stefan is trying to get across) is that the number of people employed at the minimum wage rate is not as important as the number of people who could be employed if the wage rate was lower. While the data suggest that increases in the minimum wage are accompanied by fewer people actually making the minimum wage it is simple correlation; other variables such as the total number of persons looking for work (the total labor supply), the growth in this labor supply pool over time, and the number of persons who are unemployed over long periods of time, i.e low-skilled workers whose marginal productivity is lower than the wage rate, need to be factored into the relationship between employment and the minimum wage. It's not so much that your data are incorrect as they are incomplete.

As to Card and Krueger, as I recall they did their study in the greater Philadelphia region where the prevailing introductory wage rate (like working at McDonalds, Wal-Mart, etc.) was already above the national minimum wage, so an increase in the minimum wage did have little impact on employment in the area they studied. However, that same wage rate does have an impact on employment in other parts of the country where the prevailing introductory wage rate is lower. While this would suggest that minimum wage rates should be "regionalized" or "localized", it suggests, rather, that minimum wages should be eliminated in order for locales to more quickly adjust to changing economic conditions by accomodating changes in the wage rate and labor supply.

Sorry for the long post.

Kevin Feasel January 16, 2006 at 1:31 pm

spencer: There are two problems with the data set. First, these are not real hourly wage rates, so $5.15 in 1998 is not $5.15 in 2005. Second, the same group of people is not being tracked.

With the wage rate, even if you assume that real wages don't go higher, we do have non-zero inflation in the United States, so it makes sense that the absolute number of people who make the legislatively-mandated minimum wage drops as time goes on. According to the Bureau of Labor Statistics CPI calculator, $5.15 in 1998 dollars is worth $5.97 in 2004 dollars. So if you have somebody doing a job valued at $5.15 an hour in 1998 dollars, this person will get approximately $6 in 2004. There are no major population spikes, large changes in the birth rate, or anything like that during this period, and business cycles get masked by the number of changes in the nominal minimum wage, but I think that inflation would be at least a valid theory that explains the trend.

Now, to explain the spikes when the minimum wage goes up, let's have a thought experiment. There are 6 people in my economy. Currently, 2 make $1 an hour, 2 $3 an hour, and 2 $5 an hour. I pass a minimum wage law that sets the minimum wage to $1 an hour. I have 2 people working at minimum wage. But now I set the minimum wage to $3 an hour. Of the two people who make $1 an hour, let us say that one is replaced and the other kept (say because one person is a hard worker who seems like he'll be a good employee later even though he's costing the company a little now, and the firm cuts costs to make up some of that defecit). Now we have 3 people making the minimum wage, even though I have lowered total employment. Then, because I run a high-inflation economy, at time two, the two guys who made $3 now get $4. So I have a chart like this:
t=0, $1 2
t=1, $3 3
t=2, $3 1

My little made-up economy now has a minimum wage chart that looks very much like the data you present. But in my thought experiment, the minimum wage certainly hurt one person, as he is now unable to find a job. By adding in people who were previously making more than the minimum wage, we could mask this.

So, a more valid minimum wage set would involve two things: real wages, and times series tracking of individuals. These would include invidiuals who currently make less than the new minimum wage and people who are unable to get a job at the present minimum wage because of lack of skill (we'll ignore other forms of unemployment, as they are not due to the minimum wage). By taking into account these groups of people, any effects of a minimum wage change would be visible. But what you have right now is adding new people to the pool just because their incomes happen to be the new minimum wage. Economic theory has nothing bad to say about these people–nobody thinks that these people would get hurt, so including them does not test the theory.

But this is a perfect place to end with one of the many wonderful FA Hayek quotations: "Mr. Keynes's aggregates conceal the most fundamental mechanisms of change." Here, too, do aggregates conceal reality–or at the very least, these aggregated data as presented should not be taken at face value.

the Radical January 16, 2006 at 2:03 pm

<<<"a modest increase in the minimum wage, like the proposed increase from $5.15 to $6.25">>>

So a 20%+ raise is a modest raise, I can't wait till my boss finds out.

<<<"(France the minimum wage is nearly $10 per hour)">>>

We all know what an economic powerhouse the "workers pardise" of France is.

spencer January 16, 2006 at 2:10 pm

Sorry Kevin, if you do the real minimum wage you get exactly the same results.
Moreover, if you do the real minimum wage and the teenage participation rate issue I also discussed the connection is even stronger. From the late 1950s until around 1980 the real minimum wage was generally rising and teenage employment and the teen participation rate rose. After 1980 when the real minimum wage generally fell the teen participation rate generally fell.
This is generally what you would expect, that the supply of teen labor would partially be a function of its real wage.

Yet, your theorical model is close. Your approach probably is the right direction to go in explaining why the number of people employeed at the minimum wage rises after a minimum wage hike. Probably what happens is that people previously paid the minimum wage or something between the old and the new minimum wage are now paid the new minimum wage. If the old minimum was $5 and the new is $6 all the people that use to earn between $5 and $6 are now paid $6.

This is what I think actually explains the jump in the number of people paid the minimum wage after it goes up.

But this still does not conflict with the arguments made in the article that moderate rises in the minimum wage do not negatively impact employment and that is the argument our good host is making.

the Radical January 16, 2006 at 2:29 pm

The question here is not whether artificial wage rates help or hurt those who earn them nor is what wage rate constitues a "living wage." The question is why, in a free economy, does the gov't interfere in any agreement between an employer and employee in regard to compensation. Helping or harming anyone at any skill level is not the business of the government, protecting their liberty is. So long as neithr party is subject to coercion or threat to life, liberty or property there is neither reason nor constitutional justification for any gov't at any level to legislate wages. If myself anf my employer come to a free agreement that I be paid X compensation for Y amount of work and I am compensated X times Y then that compensation is fair and just. Whether or not I can "live" on that amount is my problem, not my Congressman's

Material well-being aside, all workers at all levels are harmed by price floors on any goods, labor included. When one of us looses the ability to go into the market and deal freely with others in the market by government fiat we all become loosers.

Daniel January 16, 2006 at 2:46 pm

spencer, do you understand what the other part of his argument is? He is saying your dataset is measuring the wrong people in the right column. For example, take these two lines:

1990 3.69 3228
1991 4.14 5283

Now, it is possible that in 1990, there were more than 5283 people employed at $4.14 or less – but that is not what the 3228 figure measure, it is measuring the number of people employed at $3.69 or less. So the two numbers are comparing two different groups of people. In order to be meaningful, we would need to know the number of people in 1990 being employed at $4.14 or less (NOT just at $3.69 or less) so that we could compare it to that number in 1991. Otherwise, the comparison is meaningless – you've got a moving target group in the right column every time the minimum wage changes, since the measurement of the number of people is based on the left column.

What would be more reasonable would be if you took the number of people from 1979 to 2004 and looked at the number in each year employed at or below $5.15 an hour in 2004 dollars. Then we could have a meaningful conversation. So, instead of measuring (for 1979) the number of people employed at $2.90 or below, you would count the number employed at $1.87 or below ($5.15 in 1979 dollars) for that year. Then for the next year you would count for $2.08, then $2.36, etc. It might be reasonable to adjust for population, as well.

spencer January 16, 2006 at 2:58 pm

the Radical — You are making a valid argument. If I agree or disagree does not matter.

There is a wealth of good evidence that moderate increases in the minimum wage does not harm employment and does more good then bad. But our host is saying, do not confuse me with the facts — i am right.
If that is the position he want to take, fine. But if so, do not expect me to treat him as a serious economist. Moreover, why should the taxpapers of virginia continue to support him.

All i am asking him to do is provide some evidence or facts to show that his point of view is valid.

mike January 16, 2006 at 5:12 pm


I do not deny that you are making a fine point. However, isn't Don's post indicating something different than you are measuring with the BLS data you reported? He believes that even if minimum wage increases have minimal (or zero) disemployment effects, that total compensation might be unchanged or fall, even as wages rise, or some other margin might be affected. I suspect this is harder to measure, as I know the BLS does not capture information on working conditions, and do not know if they capture information on total compensation for low-skilled workers.

My only other comment at this time is that even if minumum wage hikes do not raise unemployment and do make those low wage workers that receive it better off, I cannot see how aggregate welfare is increased when wages are legislated rather than market determined. Even if the cost to consumers is microscopic, in aggregate the losses would have to be at least as large as the wage gains. Of course, I have not thought hard enough about labor markets with extreme long-term frictions to make this case convincingly for all situations, but I imagine you get my point.

I enjoy the data work you bring into the discussions here, it makes this one of the most enjoyable blogs I read. Thanks!


Randy January 16, 2006 at 6:47 pm

I think there is plenty of proof that employers do adapt to mandated increases to the cost of employment (Social Security, Medicare, regulations, minimum/living wages, etc.). The evidence; outsourcing, inflation, illegal immigration, unemployed teens. Who gets hurt by all this? The people the progressives said they were trying to help. Who gets helped? Progressive politicians.

John Pertz January 16, 2006 at 6:58 pm

"There is a wealth of good evidence that moderate increases in the minimum wage does not harm employment and does more good then bad."-SPENCER

This is simply not true. One study does not mean that there is a "wealth of good evidence." I would not be so quick to even label that one study as "good". If anything it is controversial but CERTAINLY NOT CONCLUSIVE. You are saying that Boudreaux is not a good ecnonimist because he refuses to abandon his beliefs because ONE study makes an EXTREMELY controversial argument that min wage laws are welfare enhancing. The big problem with that study is that its thesis has already been disproven. The economists from Princeton found that there was not a decreese in employment after the passage of min wage increases. However, this does not mean that people were harmed. The businesses simply passed the min wage increases on to their customers. Therefore this becomes alot like the Walmart debate. The min wage when it is not decreasing unemployment is therefore enhancing the welfare of unskilled laborers at the expense of impoverished consumers. What class of people on average do you think were consuming at the fast food restaurants that were survyed by the Princeton economists? The affluent were certainly not harmed by the increase in prices at Pizza Hut or Taco Bell because they do not eat there for the most part.

Poster February 17, 2006 at 11:26 pm

I live in Ontario, Canada, and the increase in the minimum wage (7.45/hour to 7.75 for adults, 6.95/hour to 7.25 for students under 18) has led to many minimum wage employers downsizing their staff.

McGuinty said that this was done to help the poor, but why would someone want to give a job to someone who actually needs it when they can give it to a teenager who can be paid less?

Instead of raising minimum wages, why not raise the legal working age to 18? This would restrict labour supply, and would let deserving individuals earn a decent wage. Teenagers can always volunteer for charities, many employers view this as a positive thing.

Previous post:

Next post: