Meeting the economists

I’ve been at the Conference of Economists in Melbourne for the last few days. I econobloggers, Stephen Kirchner and Andrew Leigh[1], as well as Andrew’s co-author Justin Wolfers and lots of old friends and sparring partners. I gave a paper on Learning and Discovery, an attempt with Simon Grant to unravel the knotty problem of unknown unknowns. You can read a more accessible summary Unknownunknowns0509here

Olivier Blanchard gave quite an interesting paper on European unemployment, on which I’ll try to comment further. I also found out what’s wrong (and right) with the bootstrap (an econometric technique) and saw an interesting ranking of economics departments, which showed that one of my former homes was ranked #2 in Australia on per capita research output just at the time when it was closed down as part of the reform process.

6 thoughts on “Meeting the economists

  1. I read your summary with great interest PrQ. I have some questions: When does an unknown become a known? Does the act of considering an unknown then move it to the realm of the known? Or are the conditions specified for application to the particular domain

  2. To finish my interuppted post (damn VB). Are the conditions stringently set, themselves being unable to be flexible enough to recognise the unexpected? I guess what i’m trying to get at is where does it end for a judgement to be made? I know you referred to this point in your paper, but isnt this crux of the matter?

  3. It seems to me that you can never really model the unknown unknowns. Because even if you do, then there’s always the unknown unknown unknowns to worry about, and they are not part of your model of the unknown unknowns by definition (the are unknown “unknown unknowns”). And even if you do get a handle on those, some unknown unknown unknown unknown will come by and bite your bum. It is an infinite regress.

    It seems the best chance is to bound the probability of all unknownN scenarios based on previous experience. For example, with war, you could investigate empirically how often surprising bad stuff happens. Although you can’t prepare for it, you’ll at least know how likely it is. Maybe that’s the point of your paper (I only read the summary).

    I believe it was Hitler aho said something to the effect that going to war is like opening a door to a dark room: you never know what you’ll find inside.

  4. Most surprise events are predictable, and, perhaps more suprising, most are in fact predicted. Even the idea of terrorists hijacking aircraft to fly into skyscrapers had been predicted in prior US terrorism reports. It’s just that we don’t hear (or choose not to hear) the predictions.

    After the first Gulf War Colin Powell said, “No [military] plans survive first contact with the enemy.” The US defense services (Army, Airforce, Marines, etc) each has an agency devoted to learning from experience, called the Centers for Lessons Learned. If anyone ought to have foreseen the aftermath of the Iraq war it should have been the US military. The aftermath was entirely predictable, and indeed was predicted. In Britain, Ken Clarke, former Conservative Minister (and currently again a contender for the party leadership), in his speach to the House of Commons before the vote on the war, predicted just the calamity we now see.

  5. I understand from a newspaper article that Peter Abelson presented a paper on how govts. in Australia subsidise housing to the tune of some $7b. annually. Are the papers available from a website?

Comments are closed.