Will Osborne’s UK National Living Wage Really Cost 60,000 Jobs?

It’s pretty dismal if you’re left-leaning in the UK right now.  Not only did Labour lose the election catastrophically and – adding to the shock – much more badly than implied by the polls, they’ve now gone nuts and elected a leader who can’t win, and even if he did advocates policies that belong in the 1970s.  Meanwhile Osborne is implementing a policy Labour should have been pushing during the election campaign, namely what is in effect a higher minimum wage, his so-called National Living Wage (NLW) for over 25s.  Of course, Osborne’s overall package is disastrous for many of the poorest households who will be worse off even with the NLW because of simultaneous cuts to tax credits.

If you’re following the debate around the NLW – for example as expertly hosted by the Resolution Foundation – it’s clear that the Big Question is how much effect the NLW (and increased minimum wages in general) is likely to have on (un)employment.  Now, based on logical argument (that being my favoured modus operandi), and, of course, because my philosophy is to question everything, I am highly sceptical of the mainstream line of reasoning that labour behaves like paper-clips.  Put up the price of paper-clips and you’ll sell fewer; put up the price of labour and unemployment will rise is the gist of it.  But this ignores the fact that increasing wages itself creates demand.  More on this later.

Much as I believe in the power of reasoned argument, I nevertheless recognise that it’s a good idea to first look at the strengths and weaknesses of the opposing position.  In this post I therefore want to focus on the meme that Osborne’s NLW will cost 60,000 jobs.  How well-founded is this estimate?  You’ll see it quoted frequently, for example, by the Resolution Foundation and on the Institute for Fiscal Studies’ (IFS) website and no doubt in mainstream media reports.  The original source is the Office for Budget Responsibility.  As far as I can tell the 60,000 figure first appeared in a report, Summer budget 2015 policy measures (pdf) which was issued around the time of Osborne’s “emergency” budget in July (the “emergency” being that the Tories had won a majority), when he bombshelled the NLW announcement.

So, I asked myself, being keen to get right to the bottom of things, where did the OBR boffs get their 60,000 estimate from?  Well, what they did was make a couple of assumptions (Annex B, para 17 on p.204), the key one being:

“…an elasticity of demand for labour of -0.4… This means total hours worked fall by 0.4 per cent for every 1.0 per cent increase in wages;”

They stuck this into their computer, together with the assumption that “half the effect on total hours will come through employment and half through average hours” and out popped the 60,000 figure.

But where does this figure of -0.4 come from?  They explain in Annex B.20:

“The elasticity of demand we have assumed lies within a relatively wide spectrum of empirical estimates, including the low-to-high range of -0.15 to -0.75 in Hamermesh (1991). This is a key assumption, with the overall effects moving linearly with it.”

The Hamermesh reference is given in footnote 3 on p.205, together with another paper:

“Hamermesh (1991), “Labour demand: What do we know? What don’t we know?”. Loeffler, Peichl, Siegloch (2014), “The own-wage elasticity of labor demand: A meta-regression analysis””, present a median estimate of -0.39, within a range of -0.072 to -0.446.” (my emphasis)

Evidently Hamermesh is the go to guy for the elasticity of demand for “labor”.  So I thought I’d have a look at how Hamermesh’s figure was arrived at.

I hope you’ve read this far, because this is where matters start to become a little curious.

Both papers referred to in footnote 3 are available online.  Here’s what Hamermesh actually wrote (it’s a screen print since the document was evidently scanned in to produce the pdf Google found for me):

150916 National Living WageSo what our guru is actually saying is that although the demand elasticity figure is between -0.15 and -0.75, as assumed by the OBR, his best guess – underlined, when that was not a trivial matter, necessitating sophisticated typewriter operation – was actually -0.3.

So why didn’t the OBR use the figure of -0.3?

Perhaps the answer is to do with the -0.39 they quote from the Loeffler, Peichl and Siegloch paper (pdf).  But this is what those guys actually wrote:

“Overall, our results suggest that there is not one unique value for the own-wage elasticity of labor demand; rather, heterogeneity matters with respect to several dimensions. Our preferred estimate in terms of specification – the long-run, constant-output elasticity obtained from a structural-form model using administrative panel data at the firm level for the latest mean year of observation, with mean characteristics on all other variables and corrected for publication selection bias – is -0.246, bracketed by the interval [-0.072;-0.446]. Compared to this interval, we note that (i) many estimates of the own-wage elasticity of labor demand given in the literature are upwardly inflated (with a mean value larger than -0.5 in absolute terms) and (ii) our preferred estimate is close to the best guess provided by Hamermesh (1993), albeit with our confidence interval for values of the elasticity being smaller.” (my emphasis)

Yep, the Germanically named guys from Germany came up with a figure of -0.246, not the -0.39 in the OBR’s footnote 3.  The OBR’s -0.39 is a rogue figure.  It must be some kind of typographical error, since they correctly quote the possible range ( -0.072 to -0.446) for the demand elasticity.  Bizarre, frankly.

It’s even more mysterious when you consider that the OBR would surely have used the elasticity of demand for labour previously.

Based on the sources they refer to it seems the OBR should have plugged -0.3 at most into their model, not -0.4.  This would have given a significantly lower estimate of the increase in unemployment attributable to the introduction of the NLW, that is, roughly 45,000 rather than 60,000.

Why does this matter?  It matters because the idea that a higher minimum wage will increase unemployment is one of the main arguments against it, frequently cited by those opposed to fair wages and giving pause to those in favour.  Here, for example, is what Allister Heath wrote recently in a piece entitled How the new national living wage will kill jobs in the Telegraph:

“…it is clear that George Osborne’s massive hike in the minimum wage will exclude a significant number of people from the world of work. There is a view that this might be a worthwhile trade-off: if millions are paid more, does it really matter if a few can’t find gainful employment any longer? Again, I disagree: there is nothing more cruel than freezing out the young, the unskilled, the inexperienced or the aspiring migrant from the chance of employment.

Being permanently jobless is a terrible, heart-wrenching state; the Government should never do anything that encourages such a catastrophe.”

Clearly, Heath’s argument (which I don’t in any case agree with) carries more weight the greater the effect of a higher minimum wage on unemployment.  But getting the numbers wrong isn’t the only problem with the OBR’s use of the demand elasticity of labour, as I’ll try to explain in my next post.