Two Reasons The American Action Forum’s New Minimum Wage Study Shouldn’t Convince Anyone

minimum_wage_theory

Fox News reports:

A new report from the conservative-leaning American Action Forum, shared exclusively with FOXBusiness.com, shows hiking the minimum wage hurts hiring.

The study looks at the 19 states that have minimum wages above the national rate of $7.25 an hour, as well as the 31 states in which the minimum wage is equal to the national average. The report finds that in 2013, a $1 increase in the minimum wage was associated with a 1.48 percentage point increase in the unemployment rate.

What’s more, this $1 hike also led to a 0.18 percentage point decrease in the net job growth rate, a 4.67 percentage point increase in the teen unemployment rate and a 4.01 percentage point decrease in the teenage net job growth rate. Overall, the AAF reports that high state minimum wages increased unemployment by 747,700 workers and reduced job growth by 83,300 jobs.

The study, “How Minimum Wage Increased Unemployment and Reduced Job Creation in 2013,” is published on AAF’s website.

I’ve read it, and I don’t think it’s persuasive, for two reasons: First, it’s an outlier, and second, its methodology compares apples and oranges, because it compares entire states rather than economically similar regions.

1) In 2009, Hristos Doucouliagos and T. D. Stanley – both PhD economists who are published experts in meta-analysis (they literally wrote the book on the subject, har har har) – “conducted a meta-study of 64 minimum-wage studies published between 1972 and 2007 measuring the impact of minimum wages on teenage employment in the United States. When they graphed every employment estimate contained in these studies (over 1,000 in total), weighting each estimate by its statistical precision, they found that the most precise estimates were heavily clustered at or near zero employment effects.”1

Doucouliagos and Stanley concluded, “Two scenarios are consistent with this empirical research record. First, minimum wages may simply have no effect on employment… Second, minimum-wage effects might exist, but they may be too difficult to detect and/or are very small.” Here’s their graph:

minimum-wage-effects-on-tee

So this AAF study is, at best, an outlier. The evidence from 64 minimum-wage studies shows that the minimum wage either has no effect on teen unemployment, or that whatever effect it does have is extremely small.23

2) In 2010, in a study published in The Review of Economics and Statistics (pdf link), Arindrajit Dube, T. William Lester, and Michael Reich showed that minimum wage studies “that do not account for local economic conditions tend to produce spurious negative effects due to” regional effects in employment “that are unrelated to minimum wage policies.”4

In other words, if you don’t control for regional differences in employment, it will look as if the minimum wage is correlated with higher unemployment; but the moment you account for regional differences, that finding disappears.

This is why the best studies of the minimum wage compare contiguous counties in neighboring states. The counties used to test minimum wage effects are, as much as possible, within a single economic region, except that one is in a state that has just raised its minimum wage. In other words, good studies compare apples with apples. The AAF’s hamhanded study, in contrast, compares entire states, as if the only economic difference between (say) Oregon and Georgia is minimum wage levels. They’re comparing apples and oranges.

(To be fair, the AAF study also controls for high school graduation rates. But that is literally the only confounding factor they consider. Nothing else – not region, not college graduation rates, not industry – is controlled for.)

How did they make such an amateur mistake? Possibly because this study was conducted by an amateur. Ben Gitis, who conducted the AAF study, graduated from college (undergraduate) less than a year ago, according to his AAF bio. Gitis majored in econ, but he’s not an economist, and he doesn’t know how to objectively measure effects of the minimum wage.

  1. I’m quoting John Schmitt’s excellent overview of the debate; pdf link. []
  2. Another recent meta-analysis, by Dale Belman and Paul Wolfson (pdf link), had similar results, concluding “that there is a negative and generally statistically significant employment effect which is between small and vanishingly small.” []
  3. To be fair, economists themselves seem to be split on the question of what effect the minimum wage has. However, a majority agreed that the minimum wage’s positive effects outweighed any negative effects. []
  4. In 2011, Dube and Reich replicated their finding using a different dataset, this time focusing on teen unemployment. Allegretto, Sylvia A., Arindrajit Dube, and Michael Reich. 2011. “Do Minimum Wages Really Reduce Teen Employment? Accounting for Heterogeneity and Selectivity in State Panel Data.” Industrial Relations, vol. 50, no. 2, pp. 205-240. []
This entry posted in Economics and the like, Minimum Wage. Bookmark the permalink. 

11 Responses to Two Reasons The American Action Forum’s New Minimum Wage Study Shouldn’t Convince Anyone

  1. 1
    RonF says:

    The CBO issued a report that stated that raising the minimum wage to $10.10 would certainly raise the wages of many people, but would cause an overall job loss of 500,000 people. That study was certainly not performed by an amateur.

    Perhaps the discrepancy can be explained by the fact that we seem to be comparing apples and oranges here. A careful reading of your post shows that the main study you cite references teen unemployment, whereas the CBO study and (I presume by the way it was cited, I haven’t read it) the AAF study reference overall unemployment. A study that concentrate on teen unemployment do not either prove or disprove that raising the minimum wage will have no effect on overall unemployment.

  2. 2
    Elusis says:

    Ron, I believe the CBO’s study said it would cause job loss of between ZERO and 500,000 jobs. A… difficult spread to work with in any meaningful way, to say the least.

  3. 3
    Hugh says:

    I see you’ve taken to heart the critique that in your comics, when a man argues with a woman, the woman’s always right…

  4. 4
    Eytan Zweig says:

    Elusis – actually, it’s even vaguer than that – they say that there’s a 66.6% chance of a loss of 0-1,000,000 jobs. They don’t say what will happen in the rest of the 33.3% cases, but since a loss of 500,000 jobs is the average (presumably, mean) estimate, they must be allowing for possibilities of both a greater job loss and a job gain.

    In other words, they seem to be predicting that if the minimum wage is increased to $10.10 then there will be some effect on employment which is more likely a loss than a gain but they’re really not committing themselves to a number.

  5. 5
    Ampersand says:

    I see you’ve taken to heart the critique that in your comics, when a man argues with a woman, the woman’s always right…

    Heh. Funny you mention that, since this cartoon was done in 2008.

  6. 6
    Ampersand says:

    A few points on the CBO analysis:

    1) My first point completely applies to the CBO analysis. Their estimates of job loss are an outlier, and like any other outlier should not be seen as more important than what the larger body of work finds.

    2) My second point also applies to the CBO analysis, albeit to a lesser degree. Because they seemingly put a lot of weight on research from years ago, before economists fully understood how important accounting for regional effects are, their resulting numbers probably do not sufficiently control for regional effects. (I say “seemingly” because the CBO chose not to publish a full account of their methodology in this case).

    3) The CBO in this case didn’t do a “study,” in the sense of doing original research. They didn’t conduct any original research, nor did they do a meta-analysis. They read already existing research and issued an opinion which made it clear that they weighed the theoretical research (which says that the MW will raise unemployment) much more heavily than the empirical research (which basically finds no effect). If they had done a review concentrating on studies that relied on actual observations, they would have found little or no effect on unemployment.

    4) 500,000 jobs sounds like a big number, and I sort of hate saying this because I don’t want to seem like a cold-hearted jerk. But 500,000 jobs is not that big compared to the scope of the US economy. That’s about three months of job growth if we assume that job growth will continue to be lousy. Put in terms of unemployment, if 500,000 additional people entered the unemployment rolls tomorrow, that would cause unemployment to rise from 6.7% to a bit under 7%.

    According to the same CBO report, raising the minimum wage would mean 16.5 million low-income Americans would get a raise, and perhaps another 8 million in a “ripple effect.” It would bring about 900,000 families out of poverty, and improve life for many millions more. All in exchange for an increase in unemployment that will be made up for in three months. 500,000 is a lot of people, and being unemployed is terribly painful; but over 20 million is also a lot of people, as are the 900,000 who the CBO says would be lifted out of poverty. In a cost/benefit analysis based on the CBO’s estimates, it seems to me that the benefits still outweigh the costs.

    5) A lot of the CBO’s numbers are based on the idea that the higher you raise the minimum wage, the bigger the unemployment effects, to such a large extent that (the CBO says) raising the MW to $9 will cause an increase in unemployment of only zero to 200,000 (with 100,000 being most likely), while raising it to $10.10 would cause an increase in unemployment from zero to a million (500,000 being most likely). That’s a huge difference that I don’t think can be justified based on empirical evidence. They also determine that the effects increase over time. These are their major methods of finding high unemployment losses, but their method for determining this is mysterious, and afaik these are purely theoretical claims that actually contradict some of the empirical findings.

    * * *

    More reading on the CBO report. The second link is to a response by an economist who has published a lot of research on this exact issue in peer-reviewed journals, and whose research was cited in my original post. The third link is to a post co-written by Danial Furman, the current Chairman of the President’s Council of Economic Advisers. I realize that could make him a not-credible source, but he has a doctorate in economics from Harvard and has been willing to go against the left in the past (see his opinions on Wal-Mart and labor). Nonetheless, the opinions of anyone working for a Presidential administration must be taken with a grain of salt.

    Why CBO differs from other economists on the minimum wage
    The Troubling Fine Print In The Claim That Raising The Minimum Wage Will Cost Jobs | ThinkProgress
    Congressional Budget Office Report Finds Minimum Wage Lifts Wages for 16.5 Million Workers

  7. 7
    Ampersand says:

    Perhaps the discrepancy can be explained by the fact that we seem to be comparing apples and oranges here. A careful reading of your post shows that the main study you cite references teen unemployment, whereas the CBO study and (I presume by the way it was cited, I haven’t read it) the AAF study reference overall unemployment. A study that concentrate on teen unemployment do not either prove or disprove that raising the minimum wage will have no effect on overall unemployment.

    The AAF study concentrates on teen unemployment (which is where it finds its most spectacular numbers), although it also compares overall unemployment between states. Sorry my original post did not make that clear.

    I think economists on all sides of the debate agree that any unemployment effects of raising the MW should be larger and more noticeable among teens than among the general population. (This is why economists who want to argue that the MW raises unemployment often concentrate on teens, or on other groups believed to be especially sensitive to the MW, such as fast food workers). So in fact, a lack of a measurable effect among teens is significant evidence that there’s no measurable effect at all.

  8. 8
    Chris says:

    RonF:

    “The CBO issued a report that stated that raising the minimum wage to $10.10 would certainly raise the wages of many people, but would cause an overall job loss of 500,000 people.”

    What a way to phrase their conclusion. I didn’t know that “many” was a number that can be easily compared to the number 500,000.

    As Amp pointed out, “many” in this case could be anywhere from 16 to 20 million people. A simple cost-benefit analysis tells us that raising the minimum wage will have more benefits than drawbacks. But today’s GOP has no interest in performing simple cost-benefit analyses. (See also: the push for voter ID laws.)

  9. 9
    Sebastian says:

    I see you’ve taken to heart the critique that in your comics, when a man argues with a woman, the woman’s always right…

    The observation was not that the men are always wrong, it was that the more privileged character is wrong. Given that the man’s skin is much darker, the rule definitely holds.

    And no, don’t say that this is due to light coming from the left. If you look at the globe, there is no shading. If you examine the floor near the characters’ feet, the light is overhead.

    Furthermore, we have straight light hair, vs undefined, and normal body type vs overweight. Clearly the man is less privileged, thus right.

  10. 10
    Ampersand says:

    I agree with you the man is dark-skinned (I thought of him as African-American when I drew this), but I’d say that both of them are on the chubby side, not just the man. He’s a little chubbier, but given how much more leeway men have to be fat than women, once could reasonably claim that she’s probably “more oppressed” for being fat than he is.

    I think that the way you’re analyzing this lacks any objective foundation; after all, if I had made the character on the right a dark-skinned man and the character on the left female, you could have said “Given that the person on the left is a woman, the rule definitely holds.”

    Anyway, I was a little disappointed that you didn’t respond to my lengthy response to you in the previous thread where this came up. Please feel free to go back to that thread and respond, if you want to.

  11. 11
    Ampersand says:

    And no, don’t say that this is due to light coming from the left. If you look at the globe, there is no shading. If you examine the floor near the characters’ feet, the light is overhead.

    I don’t say this to disagree about the dude’s skin color, but come on – the light is clearly coming from the left (and from above). I mean, look at his pants. Look at her shirt. :-p