ANALYSIS: What Paul Krugman Gets Wrong When He Claims Minimum Wage Hikes Don’t Cost Jobs
Paul Krugman opined today in the New York Times that new research shows higher minimum wages don’t cost jobs. In his view workers will work harder, so businesses break even at higher wages. He concludes the government should force companies to pay significantly higher wages. If it doesn’t hurt jobs, why not?
Unfortunately Krugman is wrong about the evidence. Good intentions or not, such wage mandates would hurt the workers he wants to help. I explained why on Twitter:
Some thoughts on @NYTimeskrugman's piece today saying minwage research shows gov't pay setting won't hurt workers 1/? http://t.co/75iiQjhl91
— James Sherk (@JamesBSherk) July 17, 2015
Krugman draws pretty strong inferences from #minimumwage literature and says we no know higher minimum wages don't cost jobs 2/?
— James Sherk (@JamesBSherk) July 17, 2015
That conclusion is highly debated. @arindube & co-authors have compared counties that cross state borders and find little effects from MW /3
— James Sherk (@JamesBSherk) July 17, 2015
They argue pre-existing state-specific trends mean you can't use states as control groups. They say county-pairs control for these trends /4
— James Sherk (@JamesBSherk) July 17, 2015
It is an intuitively appealing argument. You would think that neighboring counties are more comparable than neighboring states /5
— James Sherk (@JamesBSherk) July 17, 2015
However, Neumark et al actually examined how closely cross-border counties resemble each other. They found they weren't that similar /6
— James Sherk (@JamesBSherk) July 17, 2015
E.g. This compares urban Leon County, FL (i.e. Tallahassee, population 275k) w/rural Grady County, GA – pop. 25k. Not that similar /7
— James Sherk (@JamesBSherk) July 17, 2015
.@arindube and his co-authors responded arguing that cross-border counties are comparable, and Neumark et al. responded in turn /8
— James Sherk (@JamesBSherk) July 17, 2015
You can find those papers from Reich et al here http://t.co/cs1PMveRmN and from Neumark et al here http://t.co/jqsZvbgBJM I've read them. /9
— James Sherk (@JamesBSherk) July 17, 2015
Neumark et al.'s latest critique quite persuasively (& counter-intuitively) shows cross-border counties are not great controls /10
— James Sherk (@JamesBSherk) July 17, 2015
I suspect @arindube & his co-authors are working on a response. But current evidence not great for their approach /11
— James Sherk (@JamesBSherk) July 17, 2015
Also, Neumark says that (a) they could not replicate one of @arindube & co. results, and (b) they ignored requests to share data & code /12
— James Sherk (@JamesBSherk) July 17, 2015
See footnote 10. I consider @arindube a serious economist w/strong priors (which I can't blame others for), but that undercuts his case. /13
— James Sherk (@JamesBSherk) July 17, 2015
Moreover the results of other cross-border county studies are … weird. If you think the method is good then you have to accept … /14
— James Sherk (@JamesBSherk) July 17, 2015
… the Hagedorn et al result that extended UI benefits accounts for ~80% of the jump in unemp. in the recession http://t.co/fQgiaW0cX0 /15
— James Sherk (@JamesBSherk) July 17, 2015
I was a skeptic of 99 weeks, but not even I find that plausible. Yet they use same data source as @arindube – QWI /16
— James Sherk (@JamesBSherk) July 17, 2015
Moreover, the papers find directly opposing results on employer wage sensitivity. The MW paper finds wages have little effect on jobs /17
— James Sherk (@JamesBSherk) July 17, 2015
The UI paper finds small increases in wages driven by increased reservation wages massively reduce hiring. Both can't be true /18
— James Sherk (@JamesBSherk) July 17, 2015
The Neumark conclusion that cross-border counties arent great controls makes more sense than @arindube & Hagedorn's approach being right /19
— James Sherk (@JamesBSherk) July 17, 2015
That is also why I have avoided using the Homes result that #RightToWork laws increase manufacturing jobs by 1/3 http://t.co/RxqXl8iT9r /20
— James Sherk (@JamesBSherk) July 17, 2015
While it is a strong argument for a policy I prefer, if the cross-border counties aren't great controls it isn't a reliable estimate /21
— James Sherk (@JamesBSherk) July 17, 2015
In short, the economic evidence does not conclusively show that employment does not respond to wages, as @NYTimeskrugman claims /22
— James Sherk (@JamesBSherk) July 17, 2015
But I am partly Bayesian. I do see the @arindube et al results as evidence historical MW hikes have had smaller net effects than thought /23
— James Sherk (@JamesBSherk) July 17, 2015
Even if that is true @NYTimeskrugman's claims don't hold, for two reasons. (1) Historical MW didn't affect many workers /24
— James Sherk (@JamesBSherk) July 17, 2015
All-time high of the fed. min. wage came in 1968 at ~$8.50 in 2015 dollars (using the PCE — see the CBO report if you don't trust me) /25
— James Sherk (@JamesBSherk) July 17, 2015
The post-war average was around $6.80 or so ($2015). Historically most workers weren't affected. State hikes also didn't cover many /26
— James Sherk (@JamesBSherk) July 17, 2015
Econ 101 also says that a non-binding minimum wage has little employment effects. The hikes in Krugman's studies bound few workers /27
— James Sherk (@JamesBSherk) July 17, 2015
He and others are now proposing very binding minimum wage hikes. $15/hour is the 40th percentile of the US wage distribution! /28
— James Sherk (@JamesBSherk) July 17, 2015
We have very little evidence on that. What we have suggests trouble ahead: American Samoa, Puerto Rico, and now Los Angeles /29
— James Sherk (@JamesBSherk) July 17, 2015
Congress imposed 2007 minimum wage hike on A. Samoa. The equivalent of a $20/hr in the territory. Unemployment jumped from 5 to 36%! /30
— James Sherk (@JamesBSherk) July 17, 2015
See here for details on what happened to American Samoa: http://t.co/gt0rxWfbjY /31
— James Sherk (@JamesBSherk) July 17, 2015
~Universally acknowledged that PR's troubles are partly (not entirely!) caused by the Federal minimum wage: 77% of PR's median earnings /32
— James Sherk (@JamesBSherk) July 17, 2015
Freeman found in 1990s that the MW hike reduced Puerto Rico's employment by 9%(!). Not innocuous. /33
— James Sherk (@JamesBSherk) July 17, 2015
Now Los Angeles. WSJ reports garment manuf. industry plans to leave the city before $15/hour takes effect: http://t.co/cWRRxeGK08 /34
— James Sherk (@JamesBSherk) July 17, 2015
"Five years from now, there won’t be manufacturing in the city anymore" said one supplier. Say goodbye to 40k+ jobs /35
— James Sherk (@JamesBSherk) July 17, 2015
Contra @NYTimeskrugman we have little empirical evidence that large binding min. wage hikes don't hurt employment/jobs/workers /36
— James Sherk (@JamesBSherk) July 17, 2015
(2) Krugman argues that better morale, less turnover, etc. produces offsetting cost savings from higher pay. The problem w/this is … /37
— James Sherk (@JamesBSherk) July 17, 2015
… efficiency wages only work for above-average pay, not if everyone pays them. 90s Krugman put it well: http://t.co/V9zDZeEn2z /38
— James Sherk (@JamesBSherk) July 17, 2015
When everyone pays a higher MW labor substitution happens. Higher pay draws more into labor force. Firms then hire the most skilled /39
— James Sherk (@JamesBSherk) July 17, 2015
See lower turnover because firms employing different – and higher quality — workers. Great for those guys, not so much previous workers /40
— James Sherk (@JamesBSherk) July 17, 2015
Larger dis-employment effects among low-wage workers than the economy overall. E.g. see this JOLE paper: http://t.co/S4U2FJgKtM /41
— James Sherk (@JamesBSherk) July 17, 2015
Giuliano (2003) found MW hikes caused relative employment to shift towards teenagers from affluent zip codes and away from adults. /42
— James Sherk (@JamesBSherk) July 17, 2015
I think this also explains Clemens & Wither finding that minimum wage hikes reduce low-wage workers avg. monthly earnings by $150/mo. /43
— James Sherk (@JamesBSherk) July 17, 2015
Unlike most MW studies they used SIPP data to track workers over time. See here for their paper http://t.co/BZfPrHtP1p 44
— James Sherk (@JamesBSherk) July 17, 2015
They compared low-wage workers who made less than the new MW and workers who made a bit more than the new MW, in the same state. /45
— James Sherk (@JamesBSherk) July 17, 2015
Controls for the state specific trends that @arindube et al argue biases previous work. They find huge job losses among affected workers /46
— James Sherk (@JamesBSherk) July 17, 2015
I think they find this precisely because they use panel data. Those most hurt by minimum wage hikes are those currently making it. /47
— James Sherk (@JamesBSherk) July 17, 2015
If firms have to pay more they want to hire higher quality labor. Understandable. But bad for existing low-skilled workers /48
— James Sherk (@JamesBSherk) July 17, 2015
.@NYTimeskrugman conflates better turnover among new workers w/improvements in existing workforce. But they aren't the same people. /49
— James Sherk (@JamesBSherk) July 17, 2015
Policymakers assuming binding min. wages hikes have no effects can hurt a lot of people. Like the 40k soon to be ex-LA garment workers /50
— James Sherk (@JamesBSherk) July 17, 2015
It would be wonderful if raising wages and living standards was a simple as forcing businesses to pay more. It isn’t. Taking Paul Krugman’s advice will hurt a lot of vulnerable workers he and I want to help.
Copyright 2015 The Daily Signal