Separating Hyperplanes
Much about this blog has bothered me since the beginning. I've looked at all the various blogging engines and I hate them all. For a while I considered building my own blogging engine, but ultimately decided not to because it turns out as an economist I care more about comparative advantage than clean minimalist web design.

The new look sports a material design lite template—you can check out the MDL project here.

One of the things I often lament is that too many of my posts are walls of text with no visuals, so I decided on a photo-heavy theme. Of course, producing good graphics for every post is more work than I can afford to put into blogging these days, so my compromise was to build random photo picker. Ultimately all of the photos came from an old project from my publishing days that I never was able to publish, so here they are.

Hopefully I've not left any important components out of the redesign. I guess if the RSS isn't working none of you will see this, but in the off chance you do, let me know if anything broke, and I will fix it.
5/23/2016 07:40:00 PM
John Cochrane wrote an op-ed in the Wall Street Journal about the benefits of deregulation that was quickly obliterated by Brad DeLong.

As Dr. Kelso would say, John Cochrane spent 4 years in undergrad and 5 years in grad school, so we can assume he is at least 9. It's extremely hard for me to understand how an 9 year old could make such an absurd claim: that a little bit of deregulation could single-handedly push the US from GDP of $55,000 per capita to $400,000 per capita.

Forget the econometrics for a moment and just look at the graph:
Cochrane is making a claim that is massively out-of-sample. Which is to say, Cochrane's claim is totally unsupported by any of the data he has provided.

Statistics cannot be used to make claims about out-of-sample events. Economists sometimes weasel around this limitation by adding some theoretical assumptions: for example, if we assume that the relationship is continuously differentiable, then it follows that for some small range just outside of the sample, a naive extrapolation of the in-sample trend is a good approximation. But Cochrane is claiming that his relationship holds light years outside of his sample, with zero theoretical or empirical defense of this claim.

Brad DeLong offers some econometric arguments against Cochrane's claims. I say just look at Cochrane's data:
Graph taken from Brad DeLong's post
I've added the green and red boxes to highlight the countries that have higher doing-business scores than the US—those in the green box have higher GDP per capita while those in the red box have lower GDP. The vast majority of countries with better doing business scores have lower not higher per-capita GDP. I count 13 countries in the chart with higher doing business scores than the US, of those just 2 have higher GDP per capita. Even without any statistics, the overwhelming conclusion we can draw from this data is that increasing our doing business score is no guarantee of higher GDP at all.

What's important to remember here is that the Cochrane's critics don't even disagree with his premise that the US would benefit from many kinds of deregulation. In a follow-up post, Noah Smith points us to his own arguments for that point. What we're criticizing is the claim that deregulation alone could push the US to GDP of $400,000 per capita, a claim that makes economists as a whole look stupid, if not dishonest.

I've always regarded Cochrane as a competent, prominent economist, and no doubt the readers of the Wall Street Journal do to. I don't follow finance econ so much, but I've usually found Cochrane's blog enlightening. I guess sometimes even the most highly competent economists say stupid things. One hopes that they wouldn't do so on a medium as widely read as the WSJ, and that they would admit their errors once it is pointed out to them and they have had time to reflect.

Anyway, in hopes of ending this post constructively, here's some competent, honest analyses on how much we can expect to benefit from various kinds of deregulation. There's lots to like about this analysis. Reasonable people can disagree over whether some of those estimates are a bit too high or a bit too low, but they are all defensible. All of the usual caveats about model specification and endogeneity and omitted variables and all the rest apply. But what's really great about it is that it actually identifies real-world regulations we could repeal or modify, not hand-waiving about hypothetically making it easier to do business in the US without naming a single regulation to change. That's much more helpful! That's how you start a constructive discussion about deregulation.
5/08/2016 04:17:00 PM
That was the subject of a poll I conducted recently:

I promised to graph the result in whatever language won. So here it is in R:

6 in 10 Pirates recommend R over other leading brands.
I will confess that this is base R, not the renown ggplot library.
5/08/2016 02:30:00 PM
For many Americans, filing taxes on April 15th a couple weeks ago was the first they've heard of any requirements to buy health insurance. But here's the real kicker: if they didn't know until they got slammed with the tax penalty on April 15th, that means they will also get slammed with an even bigger tax penalty a year from now, because it is too late now to sign up for coverage for all of 2016. They learned this month that they will need to buy insurance in December in order to avoid a tax penalty in April of 2018, fully two years from now.

I get worried when our behavioral nudges require this much foresight and planning. This isn't a one-time start up issue. People are constantly dropping in and out of other kinds of coverage so every year you have a new batch of people who previously had insurance through their employer, for example, discovering for the first time that they need to buy individual coverage to comply with the law.

It doesn't have to be this way. Why not just make the coverage requirement period start in May and end the following April? That way people could sign up for coverage immediately upon discovering they owe a penalty for not doing so, thus sparing them from having a second year of penalties. I strongly suspect this would dramatically increase uptake on the exchanges, resulting in a healthier risk pool and decreasing insurance premiums for everyone.
4/28/2016 10:00:00 AM