Excellent presentations at the GEN conference

Yesterday we had a cracker of an annual conference from the Government Economics Network.
There have been individual star turns at previous conferences - notably Harvard Professor Raj Chetty's 2011 presentation, "The long term impact of teachers", which is one of the best, if not the outright best, economics speeches I've ever listened to (links here if you haven't come across it before) - but this year's conference had the highest overall standard of the three conferences run so far.

Professor John FitzGerald led off with "Contributions of economic analysis to policy making success and failures". John had the perfect professional background to talk to this, having spent a dozen years in Ireland's Department of Finance (he and I were colleagues there in 1976,  and I knew him socially around the Dublin traps in any event) before moving to Dublin's Economic and Social Research Institute (the ESRI). There's nothing he doesn't know about the use of economic analysis for policy in Ireland (where there were especially serious and instructive mistakes made), the UK and across the EU, and it showed.

He'd evidently also put in some time thinking about how his ideas would be relevant to New Zealand's circumstances. Keep an eye out for his speaking notes when they go up on the GEN website - they're well worth reading. One of the guys sitting near me had brought along a copy of John's ESRI working paper, 'Restoring Credibility in Policy Making in Ireland', which gives you a flavour of some of his ideas (as well as taking you through the near-death experience of the Irish economy).

John is one of the most articulate speakers you could hope to listen to - as Kim Hill, who was compering events, said, we'd have listened to him reading the phone directory - and it showed the value of choosing speakers with high communications ability.

Professor Anthony Scott of Melbourne  University's Institute of Applied Economic and Social Research followed him, on "The role of financial incentives in improving performance in health care". This was good stuff, too, though the state of play (as he summarised it) is that there are few rigorous studies of the links between incentives and changes in medics' behaviours (though what there is, seems to say that well-designed incentives can be material) and less again on the links between medics' behaviours and ultimate health outcomes.

One point I took away is that it can often be a good policy idea to incentivise improvements in quality, rather than achievement of absolute standards of quality. Absolute standards can throw superfluous rewards at the top performers (who are already over the absolute threshold) but be beyond the practical reach of the worst or the most dysfunctional performers. Improvements, though, can be made by anyone.

If there was a drop-off during the day, it was the panel discussion on "Future strategies - how can enhanced economic analyses help shape more effective policies?". Panel discussions are always a bit of a risk, and this one just didn't fizz. Maybe there wasn't much more to say on this after John's presentation, maybe it was just one of those things where the conversation didn't get its own momentum going, but it any event, despite everyone's efforts, I don't think it gelled.

But then things picked up again with two terrific presentations on methodology.

Dr David McKenzie, lead economist at the Development Research Group within the World Bank's Finance and Private Sector Development Unit, gave a presentation on "The value and use of randomized experiments to learn about employment and productivity policies" which first went through the rationale for randomised experiments and then showed us how three had worked out in practice.

One was the randomised use of wage subsidy vouchers for women graduates of community colleges in Jordan (ineffective - employment temporarily picked up while the subsides were available, and then dropped back to square one). One was randomised access to vocational training courses in Turkey (marginally effective on employment - it worked when the courses were provided by private sector trainers who faced stiff competition in the training market. Funny, that). And the third was randomised introduction of modern management practices into Indian textile manufacturing (very effective - a 20% increase in total factor productivity relative to a control group).

And we finished up with Professor Adam Jaffe, Director of Motu Economic and Public Policy Research, talking about "Regression discontinuity made easy" (according to the conference flyer - I think he headed his actual speech with something else which I didn't keep a note of, but the topic was the same in any event). You'd have to think this would be one of the more forbidding presentations, right up there with "Oxidative Phosphorylation For Beginners", and would be a tough challenge either to deliver or to listen to. It turned out to be a marvel of exposition - the most technical of topics laid out with an lucid explanation of how it works; why, when and how you could use it; and a comparison between it and randomised experiments, so it happily dovetailed with David McKenzie's presentation.

This was a really good day all round. Full marks to the organising sub-committee in particular, and to GEN's executive in general.