Dube, Lester, Reich focused on the short run. The paper I linked (Sorkin) agreed with their findings in the short run

> This short-run employment response is in line with both our estimates in this paper, as well as recent work studying the restaurant industry, such as Dube, Lester, and Reich (2010), Neumark, Salas, and Wascher (2014), and Allegretto et al. (2016)

The disagreement is over the long run effects:

> We find that the short-run employment effect captures only a small share of the long-run employment effect generated in the model. ...

> much of the recent research suggesting that minimum wage hikes barely reduce the number of jobs in the short run should be taken with caution. The longer-run disemployment effects are potentially large.

So when you say "no effect was found" it's because they were only looking in the short run.

>Dube, Lester, Reich focused on the short run.

Umm... no. From the abstract:

>Our findings are robust to allowing for long-term effects of minimum wage changes

Read the paper. Seriously.

That paper does not give a lot of motivation for why it uses the control that most strongly changes the results: heterogeneous (space) time variance. Why would it look like minimum wage increases decrease employment when considering US macroeconomic trends, but not when considering finer grained trends? If you increase your minimum wage you're losing out on employment compared to the US as a whole but locally there's no change in employment levels.

It does give the motivation:

"We show that this heterogeneity is spatial in nature. We also show that in the presence of such spatial heterogeneity, the precision of the individual case study estimates is overstated. By essentially pooling all such local comparisons and allowing for spatial autocorrelation, we address the dual problems of omitted variables bias and bias in the estimated standard errors"

That states an intuitive statistical fact about spacial unevenness of economics but it does not necessarily correspond to a reasoning why you should add it as a variable.

Based on the abstract of this paper, we don't know whether adding a variable will decrease or increase the bias: http://www.sas.rochester.edu/psc/clarke/405/Omit.pdf

Also worth mentioning is the inverted funnel of Doucouliagos and Stanley, which suggests that there's considerable publication bias in favor of the traditional economic narrative.

[1] Page 31 of https://pdfs.semanticscholar.org/49f8/49647aec895a598869113b...

[2] Annotated version of the key figure here: http://ethicussions.blogspot.com/2014/05/inequality-and-its-...

> It controls for most things.

I am not suggesting the results are one way or the other, but let's all keep in mind that given any statistical phenomenon, we can control for the change.

i.e. if there is an increase in automation due to the minimum wage, which in turn displaces workers, a biased or naive researcher can "control for automation and technology trends"-- effectively eliminating the perceived effect of the price floor.

I'm not saying that this is what is happening here, as I haven't read the studies, but I wouldn't be surprised if this were the case given that they are claiming the law of demand doesn't apply.