Tuesday, 14 June 2011

Unions like minimum wages

Dr. Bill Rosenberg, economist* at the NZ Council of Trade Unions, didn't like my piece in the Dominion Post. Recall that in my Dom piece, I argued:
  • Unemployment rates for youths are higher than they have been since 1986; these figures are worse if you account for the drop in youth labour force participation. And youth unemployment is particularly bad when compared to adult unemployment, which is nowhere near the levels seen when youth unemployment reached its prior peak.
  • Labour unions like to cite recent US evidence that US minimum wages have little to no effect on employment, but we can't extrapolate from studies where minimum wages are a third or less of the average wage to a labour market where the minimum wage is half the average wage. This applies to both Doucouliagos's meta-study and Dube's big panel study on county border pairs; Dube's suffers the additional problem that a study of an industry with relatively inelastic labour demand doesn't extend well to a broader economy (note that I said "isn't likely to be representative of a country as a whole.")
  • The unions also cite Hyslop and Stillman's excellent piece, but forget to note that the labour market was seriously overheating during the period of their study; ability to fog a mirror was sufficient to get a job in 2005.
  • The problem wasn't with minimum wages per se, but with minimum wage levels that get into the really rather dangerous range above 45% of average wages.
So what does Bill Rosenberg say in defence of high minimum wages?
  • The ILO concluded in '98 that "Whether a minimum wage has a negative or a positive effect depends on many factors such as its relative level, the structure of the labour market, and the country concerned."
    • ...which was kinda my argument: a minimum wage of a third of average wages mightn't have much effect while one that's a half of average wages will
  • Hyslop and Stillman found no effects of youth minimum wage increases in the early 2000s
    • ...utterly ignoring the overheated labour market that followed, or the negative effects on employment Hyslop and Stillman found in the short period prior to the labour market's overheating
  • Doucouliagos's meta-study found no effects of minimum wages in the US, and neither did Dube's.
    • ...again, utterly ignoring that minimum wages there are a third of average wages while they're a half here. It's also completely hilarious that while the unions fall over themselves to cite Chris's work on minimum wages, they ignore his work showing that unions kill company profit rates and depress physical capital formation. I like Chris's work overall. His results from the US are depressing and speak poorly of journal refereeing processes, but they just don't extend to places where the minimum wage is a lot higher. The dose makes the poison - it's true in toxicology, and it's also true of minimum wages.
  • Crampton says Dube's study doesn't apply to NZ because it looked only at restaurant workers, but more than a third of 15-24 year old workers are employed in retail, accommodation and food services and "about three-quarters are in industries that similarly can't be easily outsourced."
    • I didn't say that Dube didn't apply to NZ; I did say that we can't extrapolate from Dube to an overall economy. We can't extrapolate from Dube to NZ not because of industry choice but rather because minimum wages are so much higher in NZ than in the US.
Rosenberg makes a couple of more substantial points.

First, that the 2008 minimum wage changes applied only to 16-17 year olds; 18-19 year olds have been under the adult rate since the changes analyzed by Hyslop and Stillman. This is true. And I would be surprised if 16-17 year olds were driving everything here (though I'm likely soon to be getting sufficiently disaggregated data to be able to do something on this). What seems more likely is that the prior changes became binding when the recession hit. The 18-19 year old cohort analyzed by Hyslop and Stillman didn't go through any major recession since being subject to treatment. When late 2008 hit, the previously non-binding price floor became very binding.

Second, Bill argues that youth unemployment rates are artificially high because the denominator has been dropping: absolute numbers of unemployed youths have been growing more slowly than absolute numbers of unemployed adults, it's just that youths have been switching over more quickly to study (or other options) and dropping out of the labour force. But this doesn't really help Bill's case. Why? Recall that the unemployment rate is the number of people unemployed over the number of people in the labour force. Rosenberg seems to be arguing that all the dropping out of the labour force is among the folks who'd otherwise have found jobs - they're dropping only out of the denominator. But that seems highly unlikely. Kids who are discouraged from the labour market fail to show up in the denominator but had expected to have shown up in the numerator as well. If they'd expected only to be in the denominator, they might well have entered the workforce.

* ...but the PhD is in psychology, isn't it? I'd normally just smirk to myself on this one, as he's been doing it forever. But as he called me Mr. Crampton throughout while using that somewhat misleading tag-line, I'll make an exception. I don't much care about titles; the students usually call me Eric. But I didn't go to Evil Economics School for five years (after four years at Not Nearly Evil Enough Undergrad) to be called Mr by someone with a PhD in psychology and a misleading tag-line. Gotta draw the line somewhere.


  1. Jordan is using selective graphics. Do you want to slap him down

  2. Oh that one's fun: early data series truncation, including 20-24 year olds.