Let's get more active

By far the best single way to get unemployment down is to keep up a good run of economic growth at a robust pace. Not only does the unemployment rate go down, but the longer and stronger the expansion, the more it succeeds in bringing more marginalised groups into employment.

I've shown previous versions of the graph below: here's the up to date version, which shows (yet again) that less favoured groups suffer distressingly high rises in unemployment when the business cycle goes pear-shaped, but conversely a long expansion works its magic on everyone, even on those who tend to be on the outer. It's not perfect - rates of unemployment for some groups, notably Maori and Pasifika, are still too high - but the long post-GFC expansion has seen big falls in the unemployment rate across all groups..

Because growth alone will not deal to everything there's always a role for 'active' labour market policies that try to make the labour market work better. Residual unemployment for example could be down to things like skills mismatches: employers are looking for people with skills they don't have. Or it could be down to regional immobility: who's going to move from Northland to a job vacancy in Auckland given the cost of putting a roof over your head in Auckland? And there can be other mismatches preventing people willing to hire from signing the deal with people wanting to work.

There's a brilliant new bit of research out looking at active labour market policies, and which ones work. They're not just any old active policies: they're ones where there was a control group experiment, where you can see how those who went through the programme fared compared to those who didn't, which is how the researchers can tell if it made any difference. Here's the more readable version for the intelligent public - 'Understanding what works for active labour market policies', on the excellent Vox site - and full-blown policy tragics can get the more academic version here.

Here's the key result in terms of the impacts on earnings and employment:

If we focus on the median impact on earnings, wage subsidies and independent worker assistance ['Support to micro-entrepreneurs and independent workers'] show the greatest impact relative to the control group, with improvements of 16.7% and 16.5%, respectively. Vocational training programs have a median impact of 7.7%, while employment services show an almost negligible impact. The median impact on employment outcomes exhibits a similar pattern

Here's the graph they drew to show the results, though to be honest it's hard to see the scale of the impact with the vertical scale they've chosen to use.

They also discovered a variety of other things that make complete sense: "Individualised coaching or follow-up of the participants, training exclusively focused on a specific industry, and the provision of monetary incentives to trainees all correlate with better outcomes in vocational training programmes (the most frequent ALMPs in our dataset)", and, unsurprisingly, they also found that you get better outcomes when you run these programmes in good times rather than in the pits of a recession.

They also found (noting that cost data isn't uniformly available) that you get what you pay for. As the graph below shows, employment services programmes are cheap, but useless, while the most effective option, helping people to do their own thing, costs the most.

There's a lot we could learn from this. Currently we're down the wrong end of the OECD league ladder when it comes to what we spend on active labour market policies (as I discussed here). This research gives us a pretty clear steer on what we should do to up our game.

It's also left me wondering about our policy institutions. This research works because all over the world people have invested in randomised control trials - experiments where you try something out and compare it with what happens where it wasn't rolled out. But I can't think of many home-grown examples where we've had a go at economic policy field experiments: charter schools, maybe, but they got sat on.

My feeling is that we're too fond of the over-dirigiste 'one size fits all' approach. And our chronic political oppositionism makes both the public service and the decision-making politicians far too wary of experiments that might go wrong - even when they might hold valuable information about what works and what doesn't. There's too high a political price to pay for what will be pilloried as 'failure', and it's hindering finding out what might or might not make a real difference to those hardest to get into work. We should move on: it's time to join the adult policy world.