## February 20, 2013

### Minimum wage and unemployment

With the recent proposal to raise the minimum wage, I noticed that California currently has one of the highest minimum wages (\$8.00/hour) and one of the highest unemployment rates (9.8%) in the country.

##### Minimum wage (in dollars per hour) by state as of Jan 1, 2013. Data source: National Conference of State Legislatures.

However, this does not appear to be a general relation across states. Several of the southern states with no state minimum wage also have unemployment rates that are above the national average.

##### Unemployment rate (in percent) by state as of Dec 2012. Data source: BLS.

A regression of the unemployment rate on a constant and the minimum wage has an R2 of only 0.002.

##### Scatter plot of raw data and regression line. Horizontal axis: minimum wage as of January. Vertical axis: unemployment rate as of December.

Of course, it is difficult to use data like these to assess the effect of changing the minimum wage on the unemployment rate. The reason is that there are important differences in a variety of characteristics across states that led some states to choose a higher minimum wage than others. Those differences could themselves have important effects on the unemployment rate that do not come from the minimum wage itself. But the one thing that we can safely conclude from the diagram above is that any relation between the minimum wage and the unemployment rate certainly isn't obvious in these data.

By the way, here is a more detailed plot of last year's unemployment rates at the individual county level.

##### Source: BLS.

Posted by James Hamilton at February 20, 2013 07:01 AM

digg this | reddit

I think the real issue is related to the impact on the folks who are employed at or near the minimum wage, those who are unemployed but would accept a job at the minimum wage, and employers who hire people at or near the minimum wage.

The BLS notes that ~ 5% of hourly wage earners, earn at or below the minimum wage. Thus, it's unlikely that a change in the minimum wage will have a large impact on aggregate unemployment.

In 2011, 73.9 million American workers age 16 and over were paid hourly rates, representing 59.1 percent of all wage and salary workers. Of that number, 1.7 million earned exactly the prevailing Federal minimum wage of \$7.25 per hour and 2.2 million had wages below the minimum. Together, these workers made up 5.2 percent of all hourly paid workers.

Although workers under age 25 represented only about one-fifth of hourly paid workers, they made up about half of those paid the Federal minimum wage or less. Among employed teenagers paid by the hour, about 23 percent earned the minimum wage or less, compared with about 3 percent of workers aged 25 and over.

http://www.bls.gov/opub/ted/2012/ted_20120306.htm

Posted by: tj at February 20, 2013 07:34 AM

before the rethuglicans show up, I did the same thing as Jim above (same minwage data), and used the unemployment rate for 16-24 year olds. the r^2 is .05 but the relationship is negative. see image at link below.

http://i.imgur.com/uLtNba3.png?1

Posted by: Nick at February 20, 2013 07:42 AM

US states are not iid.

Posted by: Anonymoous at February 20, 2013 07:58 AM

The direct effect of a minimum wage is clear: folks whose productivity falls below this will lose their jobs. Currently inner city unemployment, with low high school graduation rates, etc., is tragic. Raising the minimum wage will do nothing good for the 25% urban teen unemployment.

There have been repeated calls to increase the use of Friedmans's negative income tax (currently the earned income credit) which eliminates the going-back-to-work penalties. Everyone knows folks who simply cannot afford to return to work, since benefits simply disappear, some drastically. Replacing the minimum wage increase and further unemployment insurance increases with a sliding scale benefits package/increased earned income credit would be a much better idea.

Posted by: pete at February 20, 2013 08:03 AM

An idea... You might plot the ratio of the minimum wage to the local living wages to get a better context for the minimum wage.
\$8 in California might have a lower ratio than \$7 in New Mexico for example.
I think there are living wage calculators online for every county or town in the US.

Posted by: Edward Lambert at February 20, 2013 08:05 AM

Question. Unemployment and minimum wage is a basic relationship – quantity and price. It’s obvious from the graph (and from the regression result) there is no correlation. With a little imagination anyone could go to the literature and round up a handful of causal variables that would explain some of the variation in unemployment. Suppose you used 3 variables, each statistically significant, and the R2 was between .3 and .5. Suppose you then entered the minimum wage variable and found it to be of correct sign and now statistically significant with t-ratio of 2 or better. (Theoretically I believe this result possible.) Here’s the question: Would a good econometrician consider this valid? The deeper question is this: If a variable (minimum wage in this case) fails so badly in a simple regression (the t-ratio in your simple regression can hardly be different from zero), is there ever any justification to adding it as the last variable of a multiple regression, whether after the first explanatory variable, after the first two, etc., and then claiming you’ve found something for it after it preforms so dismally by itself?

Posted by: JBH at February 20, 2013 08:59 AM

You'll want to re-do the regression with the 0's moved to the federal minimum wage of \$7.25 (and move those below \$7.25 up to that level).

Posted by: winstongator at February 20, 2013 10:06 AM

I've done what you are proposing by making teen employment a function of adult --over 24 years old -- employment and the minimum wage.

Guess what, adult employment dominates the equation and the minimum wage barely has any impact.

The data is overwhelming, minimum wage changes do not produce significant changes in minimum wage employment.

Posted by: Spencer at February 20, 2013 10:30 AM

My understanding is that the federal minimum wage dominates state minimum wage laws, so any state with a minimum wage below \$7.25 should be reported as \$7.25 in your dataset.

Posted by: Ryan at February 20, 2013 11:10 AM

winstongator:

I did what you requested with my data (16-24 y/o unemployment). see the image at the link below:

http://i.imgur.com/QEw4F6B.png?1

r^2 = .03. the relationship now makes empirical sense though (higher minwage = higher UR)

Posted by: Nick at February 20, 2013 11:16 AM

Q1: Suppose the value of the marginal product of a (low skilled) worker is distributed as {chi-square(1) + \$4} so that the mean is \$5 but a given individual can have a higher/smaller value.

Now change the minimum wage from \$7 to \$9. What happens to the aggregate unemployment rate?

Q2: Now change the minimum wage from \$7 to \$5. What happens to the aggregate unemployment rate?

A2: A decent amount (potentially).

Q3: Is there any empirical evidence on the effects of minimum wages on unemployment rates when the minimum wage is lowered (not raised)?

A3: No clue.

Posted by: mike at February 20, 2013 11:50 AM

To those of you trying to do some simple regressions to determine cause and effect. You may want to check the literature. They've already done this and included a host of controls you are missing. Check Mezie's post on this for some details.

The only thing I would like to add is that many of you ignore the employer as if employers were some bottomless pit of profit. Seriously, place yourself in the shoes of a business owner. Presdient Obama has promised to:
*force you to pay as much or more for your workers' healthcare,
*remove tax deductions,
*increase your contribution to payroll taxes,

and now,

wants you to pay more in wages to low skilled workers.

That same president claims to be a job creator.

How would you feel? Would you be concerned that the government seems to be taking more and more of your profit? Would you feel like going out and hiring a bunch of new workers?

As an employer, you have no idea how much each worker is ultimately going to cost.

Posted by: tj at February 20, 2013 11:54 AM

At a smaller level, there have been some studies of employment in counties adjacent to places where the minimum wage was raised. I know one looked at what happened along the Delaware River but I forget the details. There was no effect on employment, meaning no drop in employment when the wage rate went up though the economies are highly connected.

Posted by: Jonathan at February 20, 2013 12:27 PM

What would happen to unit sales of bananas if the government mandated they must be sold no lower than \$9.00/lb? What would happen to the unit sales of gasoline if the government mandated it could not be sold for less than \$9.00/gal? What would happen to the unit sales of chewing gum if the government mandated that it could not be sold for less than \$9.00/pack?

What would happen to the units purchased of the lowest level of labor if the government mandated labor could not be purchased for less than \$9.00/hr?

Posted by: Ricardo at February 20, 2013 01:48 PM

One would probably want to divide the minimum wage by median wage for the locality but it would still be unlikely to show anything. The plains and the south have very different unemployment rates but not that different median wages.

Posted by: Lord at February 20, 2013 02:22 PM

tj and JBH In a very simple freshman micro course a hike in the minimum wage should lead to a drop in employment. But the real world doesn't always cooperate, which is why the empirical data concerning the real world effects of the minimum wage are simply inconclusive. For example, there are income effects that shift the labor demand curve (per Menzie's partial differentiation example). There is also the curious but well documented phemonenon of workers autonomously increasing their marginal productivity to match increases in wages. So raise the wage 10% and productivity increases 10%. Lower the wage 10% and productivity falls 10%. And the freshman micro course assumes that buyers of labor's services are competitive. That's not always the case. Where employers have monopsony power it's a simple exercise to show that increasing the wage will increase the wage bill and employment. There's also the possibility that a higher minimum wage will induce highly skilled people to enter the labor force. If this happens then total employment need not go down. Of course, it's an open question whether this is desirable social policy, but the pure economics are ambiguous. Finally, a minimum wage reduces employment search and transaction costs. A minimum wage that is so low as to be a non-binding constraint means workers have to internalize a lot of the search and transaction costs associated with finding a job. This can actually increase the unemployment rate.

The bottom line is that as long as we're just talking about minimum wages that are within reasonable ranges the net effect of changing the minimum wage cannot be known a priori.

And tj you cannot simultaneously claim that the minimum wage reduces employment and say that increases in the payroll tax increase the wage bill for employers. Those two claims are inconsistent. The claim that a binding minimum wage can only reduce the wage bill implicitly assumes that the elasticity for labor demand is very low. Again, refer back to Menzie's math. But if the elasticity for labor demand is very low, then all of the burden (technically the "incidence") of the payroll tax falls on labor, not capital. So you need to get your stories straight.

Posted by: 2slugbaits at February 20, 2013 02:24 PM

2slug, this actually came up in the context of the study I mentioned above: it was motivated by the realization that much was being discussed about the effects of minimum wage increases but not much direct research. The work I referred to compared NJ and PA when state minimum wage rates differed. I think the work was by Card & Krueger. Arindrajit Dube has published papers on this as well. One looked at San Francisco, with its own wage law. And there is a more general comparison paper by Dube, Lester & Reich.

So there is real data at the local level and varieties of larger scale data, all of which show no loss. The last paper mentioned says "no detectable loss".

Posted by: jonathan at February 20, 2013 05:10 PM

Oops...

The claim that a binding minimum wage can only reduce the wage bill implicitly assumes that the elasticity for labor demand is very low. Again, refer back to Menzie's math. But if the elasticity for labor demand is very low,

Should be when the elasticity is high.

Posted by: 2slugbaits at February 20, 2013 05:47 PM

I look at your county by county map of unemployment and wonder if the unemployment rate is correlated with population density.

It sure looks like denser areas have higher unemployment rates.

Posted by: DS at February 20, 2013 07:35 PM

It might be more helpful to compare minimum wage relative to the standard of living in that state to unemployment. Making \$7.50 in Connecticut is pretty different from making \$7.50 in Alabama.

Posted by: plc123 at February 20, 2013 08:49 PM

There doesn't seem to be any significant correlation here. However, if a similar scatter plot is used but instead has weighting for the population of each state, does this change r^2 in any meaning for way. California, Texas, New York, and Florida alone would make up nearly 1/3 of the weighting by population in such a case.

Posted by: Tristan Bruno at February 20, 2013 09:05 PM

Crappy low-wage employers like Walmart and McDonalds love the EITC. They love food stamps. They love Medicaid. Almost 80% of Walmart employees in many locations receive food stamps. In many states, Walmart employees constitute the largest number of Medicaid recipients. They love any program by which taxpayers subsidize their labor costs. They hate the minimum wage which comes out of their own pocket. They prefer corporate welfare on the backs of taxpayers.

Posted by: Joseph at February 20, 2013 10:22 PM

It is funny that the relation between (especially youth) unemployment and minimum wage is so difficult to illustrate empirically, but so obvious from an analytical perspective. Nobody really questions the fact that raising minimum wage leads to an increase in unemployment, otherwise we will set a \$1000 minimum right now. But when the wage is set near the market level then it is empirically quite impossible to assess it's impact.

Posted by: Florarie online at February 20, 2013 11:23 PM

Am I missing something here? If you have omitted variables (of which there many if you are only regressing unemployment against minimum wage, as you claim) disrupt the properties of OLS? Of course you are going to get a terrible fit if you ignore all the other variables that influence unemployment state by state and just leave in one: minimum wage. The whole point of regression analysis is analyzing the influence of an IV on DV **while holding the other variables constant.** Since you have no other variables in your regression you *aren't* holding them constant when you test the coefficient of the lone variable in your regression.

Can you give us the actual specification of your regression, as well as relevant tests (Ramsey RESET test?). This seems like lazy regression analysis to me...

Posted by: Matt at February 21, 2013 12:37 AM

One may test the correlation between unemployment and corporate indebt ness or unemployment and total private indebt ness the correlations are working better.
One may as well look at the correlation between the minimum wage 8\$ actual, 5\$ in 1929 and the Dow during the same period.
« The participation of the workforce to the stocks options is not exceeding 3000 000 people out of the 120 millions counted in the work force » J. Mac Coy actuary US treasury 1928.

Posted by: ppcm at February 21, 2013 01:05 AM

One may test the correlation between unemployment and corporate indebt ness or unemployment and total private indebt ness the correlations are working better.
One may as well wonder wether a correlation may be established between the minimum wage 8\$ actual, 5\$ in 1929 and the Dow percentage growth during the same period.
« The participation of the workforce to the stocks options is not exceeding 3000 000 people out of the 120 millions counted in the work force » J. Mac Coy actuary US treasury 1928.

Posted by: ppcm at February 21, 2013 01:08 AM

Apart from that, even if a higher state minimum wage would lead to higher unemployment in that state, this would say nothing about the effect of a higher minimal wage on a national level. The reason is that the effect would almost certainly be because (some) employers would move to another state with a lower minimum wage. But with a national minimum wage rise this would not be possible. So if anything, a higher national minimum wage would be good when an effect would exist, because it would remove the distortion in the labor market arising from differing state minimum wages. One could say that this would shift the same thing to the international level, but that effect would certainly be much less, because inter-state labor and corporate mobility is much is much stronger than international mobility.

Posted by: Christiaan Hofman at February 21, 2013 04:31 AM

JBH at February 20, 2013 08:59 AM: and Matt at February 21, 2013 12:37 AM: I report a simple regression of the unemployment rate on a constant and the minimum wage. Yes, it is always possible that the outcome would be different with more variables added. But the inference also becomes less reliable if you pick and choose among various specifications, particularly with only 48 observations. In any case, adding more variables would not solve the underlying problem of simultaneous equations bias which I was describing.

Anonymoous at February 20, 2013 07:58 AM: It is not necessary to assume i.i.d. explanatory variables in order to calculate a correlation between x and y. The regression R^2 which I report is simply the square of that correlation.

Posted by: JDH at February 21, 2013 04:45 AM

2slugs

The payroll tax effects all workers. The minimum wage only effects low skilled workers. Apples and oranges. The level of an employer's attachment to a low skilled worker is low. Low skilled workers are easy to replace. The level of attachment to a high skilled worker is high. High skilled workers are hard to replace. As a result, an increase in the cost (wage or tax increase) of a low skilled worker increases the probability of an employer initiated separation, relative to high skilled worker.

jonathan
see the Neumark and Wascher NBER working paper I cited in menzie's post. They show that the methodology of Duhe and others using adjacent county data is flawed. Here's the intuition - If you restrict your study to adjacent counties, then you should be able to show that those counties explain more variation in your dependent variable then simply selecting a random draw of counties. Neumark and Wascher show the adjacent counties (and other methods) of restricting the data set, are no better than a random draw. Hence the title "Throwing out the baby with the bathwater". Neumark and Wascher also show that in the few cases where the data indicates that these restrictive methods are appropriate, the results flip back to the standard result found in 85% of the literature - an increase in the minimum wage has disemployment effects for the groups most effected by the minimum wage.

Let me re-iterate, the minimum wage is a good thing and needs to be adjusted from time to time. However, given Obama's pledge to increase the cost of electricity, his health care plan that will shift costs to employers, his never-ending efforts to increase the tax bill of employers, then you have to wonder if now is the best time to tack yet another cost on employers.

I think it is appropriate to refer to Obama's policies toward business as Obama's "War on employers". Recall how left coined the phrases "war on women", "war on hispanics", "war on the elderly", etc.

Posted by: tj at February 21, 2013 06:23 AM

Jacob Mincer, long ago, showed that the impact of minimum wages on unemployment is meaningless. The basic reason is the "unemployment" measures people search for a job, not the number of people who lost jobs. What is worse, if unemployment went up a lot (not just a small amount), it would indicate that the minimum wage is making workers better off (it would show an increase in the value of looking for and getting a job, pulling workers into the workforce). If you saw lots of shoppers waiting for a store to open, it might be because there a a limited number of really great deals inside. Or, it could indicate the store "fired" lots of shoppers and pushed them out of the door. If the minimum wage reduced unemployment, it could well mean it makes jobs not worth looking for, which means job searchers are worse off (as are those losing jobs). The bottom line is that if you believe the minimum wage causes people to lose jobs, look at how it effects employment and not the number of people who want jobs. I noticed some citing Dube et al, who looked at the effects of minimum wages on contiguous counties across state lines. To do this, they used many, many fixed effects and most likely overdifferenced the data. Redoing their regressions, I found their fixed effects took over 96% of the variation in the minimum wage (in their main regressions in Table 6). With only 4% of its variation left, it is not surprising the minimum wage had any effect.

Posted by: Walter Wessels at February 21, 2013 09:02 AM

To follow my previous post, what would happen to unit sales of automobiles (and dollar sales) if the government mandated they must be sold no lower than \$9.00/lb? Obviously this would be foolish for the government to even consider because there would be no impact. So our conclusion must be that the government will only set a minimum wage when it is higher than the market wage and that means higher than current unit sales of labor demand. Is this rocket science? No amount of Sudoku will change basic truth. Isn't this Econ 101?

Posted by: Ricardo at February 21, 2013 09:06 AM

The huge swatch of low unemployment across the Great Plains suggests that the answer is to have a steadily shrinking population.

Posted by: Michael Cain at February 21, 2013 09:20 AM

A good post from Cafe Hayek including a link to a review of Card and Kruger's book and a discussion of their study and its flaws. As someone commented (don't remember where) if an experiment "proved" water runs uphill, it would be natural to ask why that would be so. The Card and Kruger study seems to "prove" a similar phenomenon. Why would higher prices not suppress demand (or cause some other comparable effect such as reduction of benefits). Seems like magical thinking.

Posted by: Rich Berger at February 21, 2013 10:01 AM

I read through the comments and there are a few variables we could add to the regression...
ratio to living wage in area
% of families on food stamps or medicaid
median incomes in area
population density

Posted by: Edward Lambert at February 21, 2013 10:02 AM

"To follow my previous post, what would happen to unit sales of automobiles (and dollar sales) if the government mandated they must be sold no lower than \$9.00/lb? Obviously this would be foolish for the government to even consider because there would be no impact."

I looked at one car, the Honda Civic, which weighs 2,716 lbs. Your rule would set a minimum prices of \$24,444. The price range for the 2013 model: \$17,965 - \$26,465. No cars would be sold for less than \$24,444, which would affect sales.

Next example of your penetrating and creative analysis, please. This is amusing.

Posted by: Rich Berger at February 21, 2013 11:38 AM

Forget theoretical impacts of raising the minimum wage, let me tell you about some practical impacts it's going to have on our company. We employ several college students at minimum wage and if the minimum wage is raise to \$9 per hour (like Pres. Obama has suggested), we won't be able to work hiring these students and instead will have to pay our existing employees some overtime in order to make up the slack. Although this change will be cost neutral to us, it will make it that much harder for college students (interested in working in environmental science-related fields) to gain valuable experience in this field because it will be that much harder for them to find an entry-level positions. The bottom line is that raising the minimum wage will hurt younger and less experienced workers more than any other age group.

Posted by: Kenneth Kowalsky at February 21, 2013 12:12 PM

These two articles from Ezra Klein might warrant a comment:

http://www.washingtonpost.com/blogs/wonkblog/wp/2013/02/19/forecasters-keep-thinking-theres-a-recovery-just-around-the-corner-theyre-always-wrong/

http://www.washingtonpost.com/blogs/wonkblog/wp/2013/02/19/janet-yellen-explains-our-crummy-recovery-in-three-charts/

Posted by: Steven Kopits at February 21, 2013 01:12 PM

tj The payroll tax effects all workers. The minimum wage only effects low skilled workers. Apples and oranges.

This is ersatz economics. The relevant economic variable is the difference between the elasticity of labor demand and the elasticity of the labor supply curve. Those are the two factors that determine the incidence of a tax and the effect of a minimum wage on the total wage bill. Low skill, high skill, medium skill...those are just subjective categories. A low skill worker with an elastic labor supply curve facing an inelastic labor demand curve can enjoy a higher wage than a high skill worker with an inelastic labor supply curve and an elastic labor demand curve.

Posted by: 2slugbaits at February 21, 2013 03:55 PM

2slugs

Low skill, high skill, medium skill...those are just subjective categories.

Nonsense. Consult the literature on signaling screening. Firms don't post a vacancy/wage/credentials hoping to draw forth low skilled applicants. They create a job description and required credentials with the goal of screening out the low skilled from the pool of applicants. Thus, labor markets vary by skill level.

A low skill worker with an elastic labor supply curve facing an inelastic labor demand curve can enjoy a higher wage than a high skill worker with an inelastic labor supply curve and an elastic labor demand curve.

Again you fail to be pragmatic and make no sense.

A low skill worker with an elastic labor supply curve implies -
Labor supply is the number of workers willing and able to offer their labor. The empirical counterpart is the number of workers searching for a job. Unemployed workers, by definition, are those individuals who report they are looking for a job. The studies you cite claim there is no relationship between changes in the minimum wage and unemployment, (i.e. the number of workers looking for a job) so how can the elasticity of labor supply for low skilled workers be elastic when, according to the studies you cite, an increase in the minimum wage has no relationship to the number of workers looking for a job? Major inconsitency in your position! ;)

The minimum wage is a good thing. It needs to increase from time to time. Why do you ignore that fact that now is not the time to raise it? Simply put, Obama's policies are already raising(or will raise) the non-wage costs of employers. Let's give job creators a chance to digest all these new costs and regulations before burdening them with another cost increase.

How about an automatic increase every X (5?) years based on an index of (output prices - input prices) associated with the employer's industry? Workers in industries that are growing will get a greater inrease in wages than those that are in decline.

Posted by: tj at February 21, 2013 05:29 PM

Ricardo: "Isn't this Econ 101?"

I'm beginning to think that Econ 101 does more harm than good. At least we don't allow Physics 101 students to work on nuclear bombs. Econ 101 students we allow to rule the world.

Posted by: Anonymous at February 21, 2013 05:31 PM

" Econ 101 students we allow to rule the world."

In physics, you require students to test and prove their work. In economics, if the data is conflicting, people just double down on their fixed beliefs. Theory is important, not empirical results.

Steve

Posted by: steve at February 21, 2013 08:40 PM

Labor is not like commodities. Let's go back to econ 101 and look at the flow chart of the economy. You will see that households provide labor to firms, but they also consume the production of firms. Commodities feed the production but they don't consume the production.
Commodities compete for the consumption dollars of the labor that produces them.

Posted by: Edward Lambert at February 22, 2013 07:51 AM

tj The studies you cite claim there is no relationship between changes in the minimum wage and unemployment, (i.e. the number of workers looking for a job) so how can the elasticity of labor supply for low skilled workers be elastic when, according to the studies you cite, an increase in the minimum wage has no relationship to the number of workers looking for a job?

No, the studies all referred to the unemployment rate not the number of unemployed. In fact, if you'll refer back a few days I specifically pointed out that one of the effects of a higher minimum wage is that it induces more people to enter the labor force, which can increase the measured rate but does not mean that there is an actual reduction in the number of people employed. In fact, the study you cited even pointed out that one effect of raising the minimum wage appeared to be a substitution of higher socioeconomic white middle-class teenagers displacing lower socioeconomic non-teenage workers at the minimum wage. As I pointed out, this is something that has always bothered me about minimum wage hikes.

But I think you missed my point. I was not arguing that minimum wage workers have highly elastic labor supply curves; I was trying to point out that the wage and payroll tax burden are not determined by subjective categories but by the relative elasticities of labor demand and supply curves. If we can believe the paper you cited, then the labor supply curve at the minimum wage is kinked, with it being very inelastic for non-teenagers and very elastic for teenagers. And this is probably what causes a lot of the ambiguities in many of the empirical studies.

You should also keep in mind that there are three different variables. One is the unemployment rate, and that is a function of both the number of people working and the number of people looking for work. The minimum wage can impact both sides of the problem and lead to ambiguous results. Something akin to JDH's comment about simultaneous equations bias. The second variable is the number of people actually employed. Usually the only way this increases with a minimum wage hike is if there is some kind of monopsony power in the labor market, or if there is a strong income effect. The third variable is the wage bill. This can increase even if the number of employed workers decreases.

I don't think there's any disagreement about the uneven jumps in the minimum wage. The only explanation is political. Republicans want to hold off on minimum wage increases because the longer they can go between increases the lower the real wage. And Democrats kind of like the idea of infrequent hikes because it gives them an issue every 3 or 4 election cycles.

Posted by: 2slugbaits at February 22, 2013 07:57 AM

Anyone care to take a stab about how and why the real minimum wage was near \$9.25 an hour in the late 60s and this didn't cause mass unemployment and kill all the job creators?

Posted by: Nicholas at February 22, 2013 10:39 AM

sure ill take a swing at it if you are willing to expand the question to include the 50s

back then we taxed marginal income above a certain level at 70, sometimes 80% or more...we kinda had a quasi "maximum wage" ;)

Posted by: mock turtle at February 22, 2013 10:10 PM

When I check other countries, eg, Switzerland, with a monthly minium of 3700 USD and Brasil with a Minium of 300 USD and then check unemployment rate I think with a lot of interpretation you can say that unemployment might have something to do with minum salary but you also see that a huge number of factors are more important and have more impact

Posted by: sven at February 23, 2013 10:09 AM

There is always a lot of partisan screaming about how raising the minimum wage reduces employment but as your data suggests there is no obvious connection between the two. Further - is it more relevant to look at discouraged workers as well? Individual state or county unemployment rates likely have more to do with the relative competitive advantage of workers, resources, and capital employed relative to other similarly equipped markets able to serve the same constituent base of customers.
Then there is always the notion that some workers may prefer to live unemployed in Southern California than scratch out a minimum wage lifestyle in North Dakota or Minnesota. I mean just because a minimum wage job is available doesn't mean a labor unit (person) will opt to choose the lifestyle where that job is located.

Posted by: Steve at February 25, 2013 05:58 PM

1- Workers who can't find minimum wage jobs can find employment in the uncovered sector.  Thus without controlling for the size of the uncovered sector, raw unemployment numbers don't accurately measure the impact of a higher minimum wage.

2- In the absence of a minimum wage, low-skilled workers will trade off higher wages for employment that better prepares them for later (career) employment.  A minimum wage prevents such deals from being made.  Thus the impact of the minimum wage is not only measured by its employment (unemployment) effects, but also on the quality of matches between jobs and skills.

3- If a higher minimum wage makes it more difficult for low-skilled workers to obtain employment, parents may subsidize other activities (entertainment, travel, school) for their children which prevent them from appearing on unemployment rolls. Moreover, within families a considerable amount of negotiating, manipulation and game-playing take place.  Those whose bills are paid (or subsidized) by a third party are almost certain to respond differently to changing market conditions than one would predict with a simple supply and demand model.

4- Lower-skilled adult workers are somewhat mobile and can move from areas where the minimum wage is 'high' relative to other wages to areas where the minimum wage is 'low' relative to other wages (and perhaps ineffective).  Such mobility would cause minimum wages to explain city/county/state of residence but not unemployment.

5- Low-skilled workers are more likely to work in non-manufacturing jobs and produce services that do not compete with imports, either from abroad or from other cities/counties/states.  Examples are bagging groceries, serving soft drinks at MacDonald's or answering the telephone in an office.  The lack of import competition lowers the demand elasticity for low-skilled workers. However, as the minimum wages rises higher and higher, a growing number of affected workers are employed by import-competing firms ... so the greater will be the demand elasticity for labor and employment (unemployment) effect.

6- The cities/counties/states that increase minimum wages may be the same ones where unemployment among low-skilled workers is already lower than average.  Thus, instead of capturing inelastic labor demands by employers, the statistical analysis measures the supply elasticity of legislation by elected officials.

Posted by: Hosni at March 16, 2013 05:13 PM