How high is too high: minimum wages edition

Radio NZ listeners tuning in to hear me, Tim Hazeldine and Laila Harre talking minimum wages and living wages this morning might find some of these references useful.

[Update: the audio is here; RNZ's write-up is here.]

Let's start with living wage proposals.

New Zealand's Living Wage proposal is poorly targeted and badly constructed.

For example, look at page 24-25, where they describe how they came up with the weekly expenditure on food. Prices come from the supermarkets with the highest sales volumes, rather than from any examination of where poorer people actually shop. Shelf prices are used, ignoring specials or coupon prices. I'm not in a low income decile, but when things we buy are on special, we stock up. It would be surprising if poor people didn't carefully check coupons and specials when shopping. Food cost estimates that ignore that people adjust their weekly bundle according to what's on special that week likely are missing something important.

As for targeting, look at Treasury's analysis (and here). 63% of households earning below the targeted $18.40 are single adults with no dependants; almost all teens and a majority of those in their 20s earn below $18.40/hour. Those in most need would see most of any living wage abated through clawbacks in other benefits.

Treasury notes that disemployment among teens and young adults is to be expected. And if the point is to focus on those in long term low income, only a quarter of those in the bottom decile in 2002 were there in 2009.

Treasury concludes that the Living Wage project's method was geared to produce a number around two-thirds of the national average expenditure of a two-adult, two-child household. It's blunter than most things I've seen from Treasury.

How does New Zealand's minimum wage stack up?

First, here's growth over time, or at least through 2013, from Treasury's analysis of the Living Wage. The recent increase continues to well outstrip inflation.

If we look internationally, New Zealand's current minimum wage, at 66% of the median hourly wage, is very high. Here's how Treasury illustrated things, using 2009 OECD data:

So, New Zealand's minimum wage is one of the world's highest, in relation to the median wage. This matters when we think about American evidence that hikes in American minimum wages have had little effect on overall employment. A minimum wage that's 40% of the median wage will have rather less effect than one that's 66% of the median wage.

Think about it this way. Imagine the government set a minimum price on new cars. The median car is, say, a Civic that sells for $40,000, and a cheap Kea goes for $20,000. If the minimum price were set at 40% of the median, you wouldn't notice much. At 50%, it would have some effect. And at 66%, even more.

What about minimum wages as antipoverty tool?

Work by Tim Maloney and Gail Pacheco (published version here) showed that a much larger fraction of minimum wage workers in New Zealand come from high decile families than is the case in other countries. This will in part be due to New Zealand's relatively high minimum wage. Here's what the distribution of minimum wage workers, by decile, looks like:

Increases in the minimum wage are poorly targeted if the point is to help those in most need. Only about 40% of minimum wage workers are in households in the bottom 3 deciles.

A ten percent increase in the minimum wage, even if we assume zero effect on employment or hours worked, only reduces poverty rates by 0.08 percentage points in their simulations. If it results in a 3% reduction in hours worked, poverty rates only drop by 0.05 percentage points.

Work in Canada by Aninda Sen and coauthors showed that Canadian minimum wage hikes resulted in increased poverty: a 10% hike there in the minimum wage yields about a 5% increase in the percentage of families living below Canada's Low-Income Cut-off line. The intuition runs as follows. In households above the LICO where the second-earner's earnings are critical in keeping them above LICO, disemployment due to minimum wage hikes easily knocks folks below the line. But there aren't that many families that are close enough below the line that a 10% hike in the minimum wage would bring the family over the line.

And it's also worth remembering that unhappiness from job losses can trump unhappiness from not being paid as much as you'd like.

Back to New Zealand now. The Cabinet paper on this year's $0.50 hike includes Treasury's advice that the hike is poorly targeted and that disemployment effects on youths could have longer term consequences for labour market connectedness. They don't point to it, but here's some evidence.

So, if minimum wage hikes are a bad way of helping the poor, what would I suggest instead?

• Flatten some of the EMTR ranges where clawbacks compound to provide very little return to the next dollar earned.
• Fix housing. Fix housing fix housing fix housing. The gap between before-housing-cost and after-housing-cost measures of poverty is striking. Look at Table 2 in The Initiative's latest report on poverty. In 1982, 18% of children lived in households that earned less than 60% of 1998's median household income, before housing costs. That number dropped to 10% by 2014. Great news, right? Except, if we look at after-housing-cost measures instead, we get a rise from 12% to 17%. 
• Recognise the difference Working For Families, a wage subsidy programme targeted towards those in work with kids, has already made. Earnings subsidies like this have the advantage of not encouraging employers to fire people. 

Previously:

• Minimum wages and living wages
• Prior posts on minimum wages are indexed here.

Addendum: The background notes provided to the panellists pointed to Marc Amlinger's summary of the effects of Germany's recently adopted minimum wage. It notes drops in the number of 'mini jobs' (which they estimate at 133,000 losses) but no particular decline in the number of other jobs. The Ifo Institute for Economic Research's Hans-Werner Sinn, by contrast, puts the number of minijob losses at 170,000-180,000, with expectations of longer term losses of 900,000.

And think too about the refugee crisis. Germany's hosting now huge numbers of refugees, for whom German would be a second language at best, and who need jobs. How many minijobs might have been created for these workers? Is the number of jobs lost the right counterfactual when there should have instead been lots of entry-level low paid positions created with the big influx of refugees?

More on living wages and minimum wages

Fairfax's Richard Meadows asked me a few questions about living wages and minimum wages. Some of them made it into his excellent story, here. I agree with everything Matt Nolan said there too.

[Update: Nolan's extended comments are here. And I'm really pleased to see that Richard's linked the more extensive commentary from both of us.]

Here are the bits of mine for which he didn't have room, for those who are interested.

To what extent do childless people subsidise families under the current model?

The tax and benefit system strongly benefits lower-to-middle income individuals with children, with the cost most heavily borne by higher income individuals both with and without children, but also by poorer people without children. In addition, a substantial part of the government's non-transfer expenditures go towards the health and education of children and young adults, with the burden falling the same way.

Is the subsidisation of families considered a “good” thing for the economy as a whole?

Some subsidisation of lower income families has public good aspects. Ensuring that families have the resources to make at least some investment in their children's health and education helps reduce the chances that those children will themselves be in poverty later in life. But much of the current Working-for-Families scheme is a straight transfer to middle-income families with children. Chicago economist Aaron Director lent his name to one of the iron laws of transfer policy: Director's Law holds that transfer programmes mostly work to the benefit of the middle class. Education, 20-hours-free-ECE, WFF, zero-interest student loans: Director would say these are the sorts of policies we should expect where the middle-income median voter has children.

I would question the premise of your question though. If we think that it's good to give money to low-to-middle-income working people with kids, then WFF is good. If we think that others would benefit more, then it isn't. I can't tell you whether it's better to take money away from childless people, both rich and poor, to give to low-to-middle income people with kids. Economists can tell you a bit about the tradeoffs involved and the likely effects of the policy on different groups of people - both intended and unintended. But once we've taken proper account of these tradeoffs, then we start moving into value judgments to decide whether the policy's worthwhile.

Are there any unintended consequences or distortions caused by the status quo?

Poverty campaigners often criticise WFF for not doing enough for non-working families. I think they miss the point of WFF: it's intended as a family wage supplement for lower income working families. The distortion then isn't in that WFF helps encourage the lowest-income families, typically single mothers, to re-enter the workforce: that's the intended consequence.

The high effective marginal tax rates on middle income families under Working-for-Families strongly penalises second earners, often women, deciding to re-enter the workforce when the children get a bit older. For some, this then can become a trap: over reasonable income ranges, the combination of income taxes and Working-for-Families clawback rates mean that they don't get to keep much from an extra hour's work. Higher income families don't get the WFF transfers, but also don't face the distortions caused by these high effective marginal tax rates. This could be an unintended consequence of WFF. We should note, though, that while there appeared to be a reduction in female labour force participation rates shortly after the introduction of WFF, as best I know, there has not been a study looking at micro data that can tease out whether this was a causal relationship.

Let's imagine replacing WFF with an $18.40 minimum wage. What impact might raising the minimum wage to $18.40 have on unemployment?

An $18.40 minimum wage would be extraordinarily destructive. While the family envisioned in your scenario is the kind of family that Living Wage campaigners have used in setting their models, a substantial proportion of those earning below $18.40 are second-earners in higher-earning families or young adults starting out in the labour market. If the goal of the policy is to improve the lot of lower-income working families, this is better done through schemes like WFF than through across-the-board measures like minimum wages. While WFF may have some effect on employment rates by discouraging second-earners from working, and through the higher tax rates required to fund the transfer system, it can encourage some lower income workers into the labour market by making work more attractive than benefits. An $18.40 minimum wage would cause substantial unemployment among some of our most vulnerable cohorts. And while those who did not lose their jobs would be better off, the poor are often the consumers of the goods and services produced by the working poor, and some of the cost of the higher wage rates would be passed along to those customers.

While recent American data suggests little effect of minimum wages on overall employment, American minimum wages tend to be about 40% of the median wage. The current New Zealand minimum wage is more than 60% of the median wage, and an $18.40 minimum wage would be almost 90% of the median wage. The unemployment effects of an immediate shift to an $18.40 minimum wage would be large, rapid, difficult to reverse, and especially destructive to the lifetime work prospects of some of our most vulnerable cohorts.

On balance, is it likely to do more harm than good?

On balance, it is likely to be ridiculously and uselessly destructive. The modelling here really needs to account for the sharp rise in unemployment we'd see with a rise in the minimum wage of such a magnitude.

Would the cost of goods and services simply rise accordingly, and be passed on to consumers?

Whether workers would bear the burden through increased unemployment or reduced on-the-job benefits, customers would bear the burden through increased prices, or shareholders bear the burden through reduced returns, would vary greatly industry-to-industry. In some cases, firms and customers would absorb most of the cost increase, at least in the short term. In the long term, the biggest burden will be borne by those workers who want to work but who cannot provide services worth $18.40 an hour to any employer. They'll be out of work. And if we take the economics literature on happiness at all seriously, we will do far more harm to this cohort than we would benefit those who stay in work. Being unemployed is far worse than being paid less than one might hope for.

Would there be any positive effect on employment, consumer spending and business growth if taxes were cut universally? [imagining a scenario in which WFF is abolished and a tax-free income threshold is established in its place.]

Scrapping WFF and replacing it with a universal tax-free threshold costing the government as much would have a few effects. Most directly, we would stop providing the substantial transfer from childless families to those with children. We didn't have this transfer before WFF, so ending it isn't unimaginable or the stuff of the dark ages. We'd likely see a positive labour supply response from second-earners currently subject to very high effective marginal tax rates. I would expect stronger beneficial effects from ending the very high effective marginal tax rates under WFF affecting middle-income families than from the slight reduction in taxes paid by the poorest cohorts.

While a $10,000 tax free threshold sounds nice, it might not be as beneficial as an equivalent reduction in the tax rates affecting lower and middle-income earners. The tax-free threshold would benefit everybody, but would only reduce the tax burden of working an extra hour for those people earning under $10,000 per year. For an equivalent cost to the government, we could reduce tax rates in the lower-income bands to let workers keep more of the next dollar earned.

Could it be dangerous to remove the targeted support for families and replace it with universal cuts?

A lot of families have come to rely on WFF support. They've taken out mortgages, made employment decisions, and maybe even made decisions about family size, with WFF calculations at the back of their minds. There's a reason that Key's incoming National government didn't scrap WFF in 2008 despite having campaigned against it in 2005: it would be politically dangerous. At the same time, the longer WFF stays in place, the harder it would be ever to get rid of it: more families will become locked into choices that depend on the existence of WFF.

Further, one reasonable benefit of Working for Families is that it makes employment pay for some of our lowest income cohorts: single mothers with low skills and low wages. One alternative could be to turn Working for Families into a wage subsidy scheme targeted more directly toward those on lower incomes, whether or not they have children, while enhancing access to childcare facilities.

Could other parental subsidies (paid parental leave, childcare assistance, greater chance of accommodation supplement) provide enough support in the absence of WFF?

I think I've covered that off above, but I'll add one bit on accommodation supplements:

In the current housing market, we shouldn't really think of the accommodation supplement as something that helps poor families. Where cities make it very hard to get consent to build new housing, generous accommodation supplements mostly serve to help landlords earn more. When cities again make it easier to build new houses, then the accommodation supplement could do the job it's supposed to be doing. But for now, it's not helping the group it's intended to help.

The Living Wage, in one chart.

I love this chart from Treasury's advice on living wage proposals. 

There's a morbid part of me that wishes the thing would be implemented as a minimum wage - the resulting substantial increase in unemployment would do a good job of settling certain empirical debates about the effects of minimum wages. I don't really want it implemented: some data points are just too expensive to acquire.
But wait, there's more!

Other key points in the Treasury information release:

• The group that produced the $18.40 figure based it on what would be needed to sustain a family of two adults and two children. Treasury notes that families with two adults and two children make up only 6% of families currently earning below the $18.40 living wage. Three-quarters of families earning below the living wage have no kids; sixty-three percent are single adults with no dependents. Twenty-nine percent of low-earners are in families with family income greater than $60,000.
• After taking into account abatement of income-tested benefits for those with kids, the living wage would do far more to subsidise those without children. The biggest benefit would go to families with two low-earners with no children, conditional on both of them keeping their jobs.
• The proposed living wage is just shy of the median wage. Employment effects are then likely to be large: MBIE reckoned 25,000 job losses. We would also expect reductions in staff benefits and reduced hours.
• If the goal is to improve outcomes for low-earning families with young children, it is better to consider some mix of:
• Shifting WFF towards parents with younger children
• Targeting ECE subsidies more strongly (I agree entirely; see here)
• Fix benefit abatement rates to encourage 3-5 days of work;
• "Making our system of service interventions for children aged 0-5 years more focused and integrated."
• The policy would further hinder manufacturing.
• Minimum wage increases have substantially outpaced CPI but have not helped increase average wages; we shouldn't expect this hike to do better.
• We'd reduce the incentive to acquire skills because the policy would attenuate the returns to upskilling.

I love it when Treasury makes it very clear that some proposed policy is a very bad idea.

Previously:

• Guaranteed income and living wages
• Minimum wages and living wages
• Living wage mandates revisited
• Living wages and raising rivals' costs
• Living wages - Canadian economist(s) edition
• Minimum wages - addressing the more sensible critiques

Living Wages - Canadian economist(s) edition.

Simon Collins's Herald piece on living wages makes for interesting reading.

He opens with a story of a couple who both work shifts cleaning at a mix of buildings, some housing government-owned entities, some private. They both earn $14.10 per hour, less than the union's $18.40 living wage recommendation. He then points out the current version of Labor candidate pledges:

Grant Robertson pledged to set a timetable to pay the living wage to all government workers and contractors.

David Cunliffe promised to "roll out a living wage as a minimum for public servants and, as we can afford it, through the contractor process".

But the third contender, Shane Jones, refused to commit to the policy, and Prime Minister John Key said it would cost $2.5 billion and destroy 26,000 jobs.

If Labour puts in a $18.40 minimum wage for government workers, the featured family would likely only be getting this for their shift cleaning at a school unless the school contracts out facilities maintenance. And the school might shift to contracting out to keep costs down unless the government topped up its budget to make up the difference. If contractors also have to pay it, then the featured family does better in the short to medium term. But recall that if the potential benefits are large, so too are the incentives to shift to renting serviced facilities and so to have cleaners and other maintenance staff out from under the living wage mandate. So either it doesn't do much, or it gets circumvented. Collins also notes in passing that Ofa, one member of his featured family, is a delegate for the Service and Food Workers Union. I expect that the unions as a whole do well out of the measure, if it's extended to contractors, because it blunts the force of contracted outsourcing in keeping costs (and wages) down.

Collins then works through some of the costing estimates on living wage mandates, correctly noting that increasing the minimum wage to $18.40 would be very expensive. He then quotes me on the likely disemployment effects of an $18.40 minimum applied only to the government sector:

Those taxpayers would have less to spend, but low-paid state workers would have more. Even right-wing blogger Eric Crampton, a Canterbury University economist, wrote this week that the net effect would be minor: "Lots of people queue for jobs in the high-paying sector, but they'll take lower-paying jobs in the private sector."

I do expect that there wouldn't be much change in aggregate employment with a living wage mandate applied only to government workers and contractors because I expect that the government has close to a vertical labour demand curve for such workers and tasks.* There would be job rationing - in other words, more people wanting to work as cleaners in the public sector than there are available jobs - and the unions could extract higher dues as consequence. But Councils and Government would be likely, in the short term at least, just to pay more and make it up with increased taxes. In the medium to longer term, I still would expect a shift to government departments taking up tenancies in serviced buildings, but where the cleaning budget is a small part of the overall calculus, the effects mightn't be large.

Collins nicely does cite the literature on that living wage mandates are very poorly targeted and that we could do rather better by increasing targeted benefits. Then he cites U Vic's Morris Altman:

Morris Altman, a renowned Canadian economist who moved to Wellington's Victoria University in 2009, argues that a living wage is "a moral imperative situated in the natural rights of individuals".

His research suggests that a wage rise can actually pay for itself by raising productivity through motivating workers to work harder and stay in their jobs, and by inducing employers to introduce new technology and train workers to work smarter.

But that is only true, he warns, if wages are raised at a rate that productivity can keep up with. "So one has to be ultra-careful about by how much one increases. If it's a radical increase, that might be too much to deal with in the short-term," he says. "You might need a bit of an adjustment period to get productivity up."

I choose to take it as a compliment that Collins seems to have assumed that I'm Kiwi.

I haven't read Morris's work on living wages. I'd disagree pretty strongly with him on moral imperatives and natural rights, but I'm pretty sure neither of us gets to trump the other on that kind of question. And I can believe that, in some cases, salary increases can be self-financing - that's the general basis underlying efficiency wage theories (which also typically generate equilibrium unemployment). But we expect that firms choosing to increase wages on this kind of basis do so because they expect the salary increase to be worth the cost. I'm a bit curious why we'd expect those results to hold where employers are forced to pay more, but I'll perhaps have to look up his book this summer.


* I do hesitate a bit here though. I remember when the University set up a sustainability framing for a change in how they handled departmental waste collection. Instead of cleaning staff going into each office every night and emptying the bin, academic staff were asked to bring their trash and recycling bins to a central waste bin on each floor and those central bins would be collected every night. Maybe you could make some kind of sustainability case for it where staff who hate the cost of shuffling off doing a trash run every night might instead produce less waste. I'm not sure I believe it. But I am pretty sure they were able to cut the costs of building cleaning because of the policy change. At a minute per room for unlocking, collection, and relocking... well, it adds up. So there's often a margin, even where we don't expect there to be one.