The Hydra That Never Dies

The other week, the State of California reached an agreement to raise its minimum wage to $15/h from the current $10/h. The U.S. nation-wide (if not world-wide) movement of Fight For 15 celebrates its success and marches on to achieve nation-wide minimum wage hikes. This issue really is like the Greek mythological Hydra creature that grows another pair of heads whenever one is cut off - no matter how much empirics, theory or anecdotal evidence is thrown against it, the conviction that Minimum Wage hikes is a panacea for all kinds of problems persists. I've called mercantilist emphasis on trade-deficit hydras in the past, but this one must be worse.

You might think of this post (as I do) as a message to my economically illiterate friends on the left. Here we go.

Background

Economists have for decades claimed that the MinWage is a harmful policy that causes unemployment - especially for low-skilled immigrants and the poor. The story is most-commonly explained by a straight-forward perfectly-competitive S&D analysis where a price floor such as MinWage is set in a space of downward-sloping demand curves and upward-sloping supply curves. This means that some employment that previously would have occurred, is now legally banned from occurring - and so the policy creates unemployment.

Since the early 1990s, most famously by Card & Krueger in a study of fast-food restaurants in New Jersey & Pennsylvania, some economists have empirically found that hikes in the minimum wage were not associated with a reduction in employment, and sometimes even an increase in employment.

Political movements during the early 2000s to create a "living wage" and more recently the Fight For 15 campaign and Bernie Sanders' presidential campaign (not the mention the Friedman calculation of Sanders' economic impacts) have brought these discussions back into the media spotlight from the quiet and calm disputes of academic economic journals.

How Big of A Deal is The Minimum Wage?

First of all, we have to put the discussion into perspective. In most OECD countries, the proportion of full-time workers earning the MinWage is something like 2-3% (or 5-7% if we include part-time workers) of the workforce. That is, hardly anyone makes these wages, and so the political discussion amounts to making a mountain out of a molehill.

This has two implications that become important further down: 1) it explains away the most superficial defence of the MinWage ("see, it doesn't cause any unemployment!" - or the more honest point that the unemployment caused is negligible). Of course studies concerning the entire population, or quick-and-dirty comparisons of unemployment rates and minimum wage laws across states or countries, will show virtually no effect, since the changes of MinWage is a fraction of a tiny fraction. 2) the direct corollary of the above is, of course, that its value and its ability to achieve the ends of ending poverty and creating wage justice or lifting the hard-working poor out of poverty is very narrow indeed.

To The Juicy Stuff

The initial Card & Krueger study has all the elements of a bad economist, despite their respective merits elsewhere. The phone survey of employers is dubious, the sample is small, the selection of fast-food restaurants is a very narrow share of the market (see Sowell's hypothetical example) and it compares employment only, not hours worked.  Indeed, when Neumark & Wascher in 1995 re-creates the study with payroll data, they find the negative unemployment effects loud and clear.

But, my left-leaning New-Keynesian friends object, there are theoretical reasons for why a higher MinWage increases employment. Many versions of the Efficiency Wage Theorem pop up here; it states that a higher-than-average wage reduces vacancy & turnover costs, which implies lower training costs for employers and it increases the motivation of workers. The reason for the latter is that the difference to their next-best option is now higher. However reasonable this is in theory, it only works for a specific employer at a certain time; the distance to the ocean does indeed increase if you lift one boat upwards, but it is meaningless when a minimum wage tide lifts all boats.

But, my Post-Keynesian friends object, a higher wage for the lowest-paid has to come from somewhere (profits, increased prices etc) and so is essentially a redistribution of income from high-income earners (that are prone to save a lot) to low-income earners (that are prone to spend a lot - in economist speak: a higher marginal propensity to consume, MPC). In effect, a MinWage hike will then stimulate the economy via more spending, create more jobs further down the track that in turn requires more workers - so any initial unemployment is offset! This is a crucial misconception in most Keynesian and Post-Keynesian models (and a conflict that goes way beyond this issue), that spending steers and governs the economy. For a whole bunch of reasons bigger than the MinWage debate, those are highly doubtful at best, and outright incorrect at worst (see for instance Mulligan on consumption functions and variable MPCs or this recent paper on its stability).

At any rate, the alleged "boost" is minor, considering that it amounts to maximum some 10%-points(empirical differences in MPCs) of the maximum wage change (around 50% over several years) for a very small proportion of the workforce (say 5%); 0,1 x 0,5 x 0,05 = 0,0025, that is 0,25% of a "boost", spread over 5 years, a rounding error in the national accounts (and this is overstating the effect, as Worstall has showed, and left-leaning economist Reich agrees). As the more genuine economists of the MinWage proponents like Krueger, Reich, Burtless and even Dube admit, there will be some employment loss. In Los Angeles, after their having recently hiked the minimum wage to $15/h, according to the the report for the LA City Council, it will even contract the economy. Reich concludes that even after having accounted for a maximum amount of spending (assuming that all current-poor workers save nothing at all), the GDP of Los Angeles will fall, and a couple of thousands will be unemployed. That is, after making too strong assumptions about saving and consumption patterns, even the most left-leaning of the well-known economists who support increasing the minimum wage, still come up with a MinWage scenario worse than the status quo. Some employment boost.

Moreover, Mark Perry at the American Enterprise Institute, whose beautiful Venn Diagrams I have written about before, produced this chart of the unemployment in Seattle after its recent MinWage hike in April last year:

Temporal correlation is not causation, but if I was advocating MinWage hikes and saw this happen after my policy was finally introduced, I would perhaps be a bit more modest about the urgent need and alleged benefit of my policy.

But it helps the most vulnerable people in society, my SWJ friends claim, economists or not. That's the alleged and desired benefit, isn't it. But Murphy et al's recent study from the Canadian Fraser Institute emphatically concludes that the "benefits" of a minimum wage are certainly not reaped by the most vulnerable or worst-off people in society:

the vast majority (84.9%) of young minimum wage earners in 2014 were a son, daughter, or other relative living with family. More generally, 56.9% of all minimum-wage earners in 2014 were living with family, while only 2.2% of minimum wage earners (24,000 workers total) were unmarried heads of household with at least one minor child.  [...] 19.9% of all minimum wage earners were married to a spouse who was also employed. Of this group, 91.3% had spouses who either earned wages higher than the minimum wage or were self-employed. Taken together, these data undercut the popular image of minimum wage earners consisting of breadwinners supporting a family. (pp. 4-5)

Murphy et al also found the so called displacement effect where a higher minimum wage attracts students or other higher productive labour for jobs that they previously would have discarded. This exacerbates the problem for MinWage supporters, since the so-called vulnerable workers they are trying to help, now face better and more productive competition from other workers.

There's heaps more to dig through, but as Neumark & Wascher concludes in their massive study from 2006:

The oft-stated assertion that recent research fails to support the traditional view that the minimum wage reduces the employment of low-wage workers is clearly incorrect.

What About Australia?

Australia, then, is allegedly the upside-down world where economic theory once and for all was disproven. Here, comparatively high minimum wage laws are associated with significantly lower unemployment levels than in most other places of the world. This chart from Ben O'Neill at UNSW serves to illustrate the superficial point:

The Australian case at first glance does look like it overturns the conventional wisdom (or should at least make us evil economists slightly concerned). The "puzzle" is however quickly resolved when the kind of minimum wage structures between the two countries is accounted for. The U.S. minimum wage is a static one that applies for every worker regardless of age and industry (states do however differ in their minimum wage laws). O'Neill shows further that Australia has a sliding MinWage that is "using age as a proxy for productivity". That is, younger workers (below the age of 16) face much-lower minimum wage restriction than that for every age bracket up to 21 years of age (here for details). This means that the low-skilled and unexperienced employees where a high minimum wage would be most binding, is not preventing employment in Australia, since those particular groups of employees are exempt. Of course, the complicated lists of MinWages according to class of employment and age probably doesn't get it right all the time, but it is bound to be a closer proxy for productivity than the U.S. flat rate. O'Neill says:

The graduated minimum wage system prevailing in Australia is a tacit admission by the government that minimum wages are known to cause unemployment.

Andrew Leigh at Harvard also found a negative elasticity of demand for labour in comparing Western Australia with the rest of Australia between 1994-2001. Australia does not seem to be any different.

Conclusion

No, my dear friends on the left. Minimum Wage hikes are not strictly beneficial or a no-brainer. They are quite likely to harm the very people you are trying to help. Increasing the wage of Paul by making Peter unemployed or attracting brilliant physicist-student Pamela to compete for that job, doesn't help Peter. On the contrary, it makes him pretty unemployable.

The "New Minimum Wage Research" is not obvious or uncontested or straight-forward. Apart from its many methodological and empirical flaws, its proponents don't even support minimum wages as high as California or Seattle has recently enacted ($15/h). The harm, as more properly understood in Australia, is still there, and unemployment for the people you are trying to help is a very likely outcome.

As Worstall concludes:

Making us all poorer in aggregate is really not the point nor purpose of public policy.