Prediction Machines

My book with Ajay Agrawal and Avi Goldfarb is out now. It is called Prediction Machines: The Simple Economics of Artificial Intelligence.

We have written some pieces that provide little excursions into the book.

Also, we had a book launch and a video of it is available here.

Suffice it to say, if you like the material in this blog, it is a safe bet that you or your robot partner will like this book.

Boehmermann vs Erdogan – an update

You might remember the case of German comedian Boehmermann and the poem with which he demonstrated what you could, by then-German law (paragraph 103 StGB, the Criminal Code), *not* say about high-office holders abroad. Because the poem targeted him, and since wanna-be dictators like him tend to lack a sense of humor, Turkish PM Erdogan fell into the trap and took Boehmermann to court, demanding at the same time that all traces of it, and the earlier brilliant song that motivated it, be removed by the German government. Much hilarity ensued. Also much publicity for both, the song (now at more than 12 million views on youtube alone) and the poem.

That obscure paragraph 103 StGB with which Erdogan tried to silence Boehmermann goes back to 1871 and had been invoked only a few times previously. Equally obscure, and absurd, was paragraph 104a which stipulated that the government must decide whether it allows the complaint to go forward under 103. Merkel copped much opportunistic criticism, mostly from spectacularly ill-informed media writerlings and pollies, for her considered decision to let the suit go ahead, arguing correctly that it was not her job to decide whether Boehmermann had run afoul of paragraph 103 StGB.

I predicted then – confidently, because clearly the poem was meant to illustrate what you were *not* allowed to say — that Erdogan did not stand a chance. And sure enough he never did.

It must have been quite the lesson for Erdowahn. Be it only for the additional wave of ridicule it generated.

As of January 1, 2018, that silly paragraph in the criminal code (StGB), is gone for good. A  pity really because teachable moments for wanna-be dictators are few and far between.

Lessons to be learned: First, in a functioning democracy, satire can be used to speak truth to power even if that power feels perpetually offended. Second, a bit of knowledge of what laws say carries a long way. Spectacularly ill-informed opinions about what ought to be done, not so much. Third, a bit of sound game-theoretic reasoning carries a long way. Almost always.

Congrats Andreas and Kyoung-hee

Andreas and Kyoung-hee
Andreas & Kyoung-hee

This past weekend was a very busy one for our fellow blogger Andreas Ortmann. He embarked upon a new research program, his most ambitious ever. It is completely ‘blue sky’ exploratory work and is going to cost so much that the Australian Research Council will not fund the project. The review panel also noted that it lacks a control sample, that it did not include planned repetition, and that the use of models was apparently not entertained. So yes, Andreas has just tied the knot with my ex-schoolmate from MIT, Kyoung-hee Yu. In a room filled with economist and management scholars who managed to get along without incident (probably because of the lawyers and an anthropologist who kept the truce), Andreas and Kyoung-hee began their journey together with lots of wine, a delicious cake, and lots of rejoicing. Congratulations Andreas and Kyoung-hee. May your journey together be filled with happiness.

Migrants v babies

In today’s Age, Ross Gittins argues the economic case for less immigration based on the impact on the environment and aggregate demand. But he commits a surprising sin: he fails to consider what the best way to manage aggregate demand and the environment would be as opposed to pointing out what immigration does to them. Indeed, he claims that immigration accounts for half of our population growth. But it is population growth that is the bigger issue. And the obvious solution is to curtail the other half of population growth — the so-called natural half. Let me edit his article accordingly:

immigration birth adds more to the demand for labour than to its supply.

That’s because migrant families with children add to demand, but only the individuals who work add to supply.

Migrant families with children need food, clothing, shelter and other necessities. They also add to the need for social and economic infrastructure: roads, schools, health care and all the rest.

Another factor is that their addition to demand comes earlier than their addition to labour supply. The rate of unemployment among recent immigrants babies is significantly higher than for the labour force generally.

Of course, I have to spot there because the rest is about skilled migration and let’s face it, babies ain’t skilled. So the case here is for reduced natural population growth as opposed to immigration.

Indeed, that case becomes stronger. Here is Gittin’s argument on the environment:

It’s obvious that one of the quickest and easiest ways to reduce the growth in our emissions — and make our efforts to cut emissions more effective overall — would be to reduce immigration.

It is not obvious at all. It is simply wrong. If people cause emissions and emissions are a global problem, then at first glance moving people from one country to another neither helps nor harms the environment.

More on vertical mergers

I note by way of interest that Henry Ergas has responded to my comments taking issue with some of his initial thoughts on the ACCC’s draft merger guidelines. He notes that my (i) argument that studies of vertical mergers may be biased towards seeing them as favourable may not be so strong (especially as bad mergers get through: something I argued was the case with AGL-Loy Yang) and (ii) that AGL-Loy Yang wasn’t a bad merger. I can see his point on (i) but that reduces but doesn’ eliminate such biases. On (ii), I refer back to my recent work with Frank Wolak that actually looks back at what happened after AGL-Loy Yang and confirms that the ACCC’s predictions of 20 percent price rises were largely borne out. In any case, I reiterate my view that it is precisely because vertical mergers have ambiguous consequences that merits the ACCC spending extra time being extra clear in how they will look at them and such emphasis is not misplaced. (By the way, one of these days Henry Ergas and I might agree on something. Will that mean we are both right or both wrong?)

Game Theorist: Lots of Posts

Lots of recent posts over at Game Theorist: