Why the University of Washington’s Study on Seattle’s Minimum Wage Increase is Not Flawed

Late last month, a team of economists at the University of Washington released their findings on the past couple minimum wage increases in Seattle. Unsurprisingly, this was covered by just about every news outlet. To my surprise, many are divided over what the Seattle economists found. Economists at UC Berkeley did their own study and had different findings. This led to a lot of people, including economists, calling the University of Washington’s study flawed. The only problem is that their study isn’t as flawed as critics claim.

Why Some People Harshly Question the University of Washington’s Study

In 1994, David Card and Alan Krueger published a study that showed that a minimum wage increase from $4.25 to $5.05 in New Jersey, which contradicted the notion that higher minimum wages causes unemployment in lower wage jobs (a common theory in labor economics). This was a study we used in my econometrics class at the University of Oregon and it was interesting to see the flaws in this study when used for a $15 minimum wage comparison.

Another reason why some people are questioning the University of Washington study is because of a study done by UC Berkeley (Berkeley), which claims that the $15 minimum wage did what it was supposed to do, which was raise wages on low-wage jobs.

Differences and Findings from Berkeley and Washington

Berkeley’s focus was very different from the University of Washington (Washington) study. Berkeley focused on the restaurant industry, much like Card and Krueger’s study of New Jersey did. The difference with Berkeley’s study is that they “scale outcome measures so that they apply to all sectors, not just food services.”

The Washington study focused on all sectors of employment. The study by Berkeley’s economists showed an increase in food service wages, which they claimed was the policy’s goal. The study by Washington’s economists showed an overall decrease in the number of hours worked in low-wage jobs, which led to an average decrease in wages of $125 per month in 2016.

It’s notable that the Washington study agreed that there was zero effect on the restaurant industry across all wage levels. Another difference between Berkeley and Washington’s studies was that Berkeley used the Bureau of Labor Statistics’ Quarterly Census on Employment and Wages (QCEW) and Washington used data from Washington’s Employment Security Department.

Issues with the Berkeley Study

One issue with the Berkeley study is that it did what Card and Krueger did for their study: They cherry picked an industry known to have low wages and put their focus on it. Of course, when we think of low-wage jobs, we tend to think about someone working at McDonalds, a fast food restaurant.

Earlier, I mentioned that the Card-Krueger study had flaws when comparing it to a $15 minimum wage. An increase from $4.25 to $5.05 isn’t a sharp increase by any stretch; it’s not even a dollar of an increase. This makes it a bad comparison not only because it’s not a sharp increase in the minimum wage, but also because it’s only applicable to New Jersey and Pennsylvania.

Another issue with the Berkeley study was that it created a “synthetic control” by using data from other counties, many of which are in other states, that did not raise their minimum wages. The Berkeley economists claimed that results of the restaurant industry are comparable to the result of all other minimum wage jobs. The issue with this is that it creates similar issues to the Card-Krueger study.

That is, I doubt sampling from counties that are “similar” to Seattle is going to produce something accurate regarding what would have transpired had Seattle not raised its minimum wage.

Lastly, the Berkeley study had this to say about studies that included hours worked in their research:

“Employers who react to the minimum wage increase by reducing employee hours will thus impart a negative effect on our wage measure. In the presence of negative effects on hours, our estimated effects on wages represent a lower bound on the true wage effect. However, studies that have hours data (including Seattle Minimum Wage Team 2016a, b), find a very small hours effect.” (8)

In addition to this, the Berkeley study included an appendix (found on page 20; Appendix A) that was entitled, “Why minimum wage increases produce little to no employment effects.” Let’s talk about the “very small hours effect” before getting into why you shouldn’t have a two-page argument in your “research” to convince people why you’re right. One of the findings in the Washington study was that hours worked by low-wage workers decreased by an average of 9 percent under the $13 minimum wage.

For perspective, the average low-wage worker working 25 hours per week would now be working 2.25 hours less. If they were working 40 hours per week, they would now be working 3.6 hours less. This doesn’t sound like much, but that’s a good 9 to 14.4 hours per month. Had a low-wage worker lost this many hours when the minimum wage was $11 per hour, they would lose anywhere from $104.17 to $136.37 per month. While the economists at Berkeley are likely making much more than the average American worker, anyone who is working a low-wage job will likely tell you that losing $100 can be pretty nasty if they’re living paycheck-to-paycheck.

Now let’s talk about why you shouldn’t have a page and a half of convincing in your report. First, this isn’t the only place this happens in the paper. There are casual defensive remarks that quite a few economic papers have in order to address potential issues. But it’s the only place where the reader is expected to take Berkeley’s study for what the team believes it’s worth. Secondly, there are some things in there, such as productivity increasing from a higher wage (efficiency wage) and costs being passed to consumers (like with taxes), that are taught in the very first economics class anyone takes.

I understand that the audience isn’t necessarily as versed in economics or econometric modeling as I am (and I’m not that well-versed), but I was expecting to see an estimated model. At least Washington did that for us and explained every component of it so that even those that don’t understand economics at the PhD level can somewhat understand what’s going on.

Why Washington’s Study Isn’t Flawed

First and foremost, I cannot stress enough how important it is to realize that Washington’s study claimed to have similar findings on restaurant employment at all wage levels. If Josh Hoxie from Fortune was right when he claimed that this study was utter BS, then he would have to acknowledge that UW’s economists were absolutely wrong to say that there was an estimated effect of zero when analyzing employment in the restaurant industry at all wage levels, which is similar to what Berkeley’s economists found.

Another reason why Washington’s study isn’t as flawed as Hoxie claimed it to be was due to the fact that Washington’s economists had unprecedentedly detailed panel data. For those who don’t understand econometrics jargon, panel data is data gathered about individuals that changes over time. For example, each individual included in the study had their wages change in 2015 to $11 per hour and then to $13 per hour in 2016. The kind of data Berkeley’s economists had access to wasn’t bad data, but it was federal data that wasn’t nearly as detailed as the state data Washington’s economists had access to.

Berkeley said that hours worked would show very little effect. This makes me more confident in the Washington study even more. Washington’s data is probably a lot more reliable in terms of hours worked than the nonexistent data provided by the Bureau of Labor Standards, which doesn’t report this.

The reason why Washington’s economists didn’t rely on the restaurant industry is pretty logical:

“Using restaurant or retail employees or teenagers as proxies for the entire low-wage labor market might lead to biased minimum wage effects. Intuitively, a sample mixing jobs directly affected by the minimum wage with others for which the price floor is irrelevant would generally skew estimated impacts towards zero.” (8)

In layman’s terms, to solely rely on one or two industries, like Berkeley’s economists did due to a lack of detailed data, would tell us that there was little to no impact on employment. In fact, that is not only what happened with the Berkeley study, its economists also stated that the policy had done what it was put in place to do: Raise wages. The reason why Washington claimed wages had dropped was largely due to the decrease in hours worked.

If you read what Hoxie said about the fact that nearly 40 percent of the workers were not included in the study, you may have noticed that he didn’t explain that they were likely excluded because of missing data. This would occur if someone started working a low-wage job in Seattle in 2016 and didn’t work in that industry or city during 2015. Any economist will attest that this can happen when you have missing data.

Lastly, it’s important to recognize the fact that the control group used wasn’t from another state; Washington’s economists used other neighboring areas that didn’t raise its minimum wages in order to evaluate the effects. This is smart because the best cities that I would assume are great comparisons to Seattle’s economy would be Bellevue and Tacoma, both of which have corporations located within their respective areas. I would know. I lived in the area for 10 years.


I can’t determine what shocks me more: The fact that so many economists are refuting this obviously sound working paper by Washington’s economists or the fact that so many economists and opinion writers are politicizing Washington’s working paper. So the paper’s outcomes didn’t show that the road to $15 isn’t as great as people had hoped it would be. So what? This doesn’t mean that it wouldn’t have the ability to work somewhere else. What this implies is that Seattle is showing signs of overall wage decreases due to employers cutting the hours their employees work, which is a normal response when an economy isn’t doing as well as expected.

The Washington Post had this to say about an economist at MIT:

“”This strikes me as a study that is likely to influence people,” said David Autor, an economist at the Massachusetts Institute of Technology who was not involved in the research. He called the work “very credible” and “sufficiently compelling in its design and statistical power that it can change minds.””

I don’t think this is being politicized because all of these economists and policy experts think it’s as flawed as they claim. I think it has more to do with the fact that leftist economists and policy experts who have fought and supported this kind of policy so passionately may have been wrong about it and they’re scared.

It would not only make sense, but it would be understandable to see Berkeley’s economists worry about Washington’s findings since California is set to have a $15  state minimum wage by 2023; California isn’t set to have a $13 state minimum wage until 2020, which is less than three years away. Hopefully, more economists come forward either in defense or with constructive criticism of Washington’s study that doesn’t automatically classify it as BS or an outlier. Either way, now is the time to start watching Seattle closer than we’ve ever watched a minimum wage policy before.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s