Slower Chinese Growth Could Be Good for the Global Economy

One thing that we hear a lot is that global economic performance increasingly depends on the BRICSAM countries, particularly China which is now the world's second largest economy. If China's astonishing growth slows, this thinking goes, then they can drag down the rest of the world.

That could be true, but it doesn't have to be. There is a scenario in which slower measured Chinese growth is actually good for the global economy, and also good for the Chinese. I do not refer to beggar-thy-neighbor mercantilism, in which the rest of the world expropriates from China, but to an arrangement that is Pareto-improving in aggregate. To see why we just need to remember our Econ 101 national accounting device:

GDP = C + I + G + (X - M)

See that minus sign in there? If China increases its imports without anything else changing then its measured GDP growth would be negative. Yet this would in no way be a bad thing... everyone agrees that Chinese citizens should be consuming more, some of which should probably be imported goods, and many also argue that the macroeconomic imbalances contributed to by China's large trade surplus increases financial instability. Meanwhile, many countries outside of China would like to increase the exports in order to boost job growth. Narrowing the gap between 'X' and 'M' would be a positive for China and for the rest of the world as well. Some of this might be happening. Chinese consumption has been growing faster than GDP, and imports had been growing faster than exports until last month. A prolonged, multi-year trend of this sort would be good for everyone... and would also show up in the data as a slowdown in Chinese GDP growth.  

A broader point is that we often pretend that GDP measures one thing -- the well-being of a society -- when it's really measuring something different -- the composition of economic activity in a society. Well-being can increase under a variety of scenarios including the increase of imports, which drags down the GDP measure. Or GDP could increase in a way that doesn't benefit society, if e.g. the government spends $100mn building a skyscraper then another $100mn knocking it down. Broad measures like GDP are often useful as proxies for other quantities we're interested in, but not always.

(Terrible, No Good) GOTD

This, apparently, was Bloomberg's "Chart of the Day" (via, and HT to Leigh Caldwell on Twitter). In case you can't tell, the orange line is the U.S.'s legal debt ceiling, the white line is the spot price of gold.

It is a very bad, no good, terrible, very bad graph. Misleading at best. Can you tell why?

Here's a clue, but it won't work until tomorrow.

Insta-Classic

Andrew Gelman:

If some dude scratched my car, I wouldn’t be so quick to jump to the conclusion that he’s a rapist.

But would you think he was an atheist?

Context.

Apologies for the light posting. Hopefully we'll be back to normal soon.

Making Quant IR Credible

Phil Arena, IR assistant professor at SUNY-Buffalo, has a new blog. He focuses on the security side of IR, but some of it crosses over. For example, this post on the difficulty of isolating causal processes from associative stats analysis. Basically the idea is this: Suppose there are two possible states of the world, one in which A causes B and one in which C causes both A and B but is not directly observable. Standard stats methods would not be able to distinguish between the two.

I think he does a pretty good job of describing the problem, so I'm not going to rehash his post. Just go read it. And I think we're seeing an increased use of Bayesian stats, instrumental variables, experimental approaches, and network analysis to try to mitigate the problem. In other words, I think things might be improving as the discipline matures. I completely disagree with this, however:

If you ask Jas Sekhon, one of the most talented methodologists we have in political science, he'll tell you that the answer for IR scholars is to give up on quantitative work altogether, learn how to do credible qualitative, and start asking more policy relevant questions.

An argument for better stats (or better theory) is not an argument for qualitative methods, which must be made on its own merit. I like qualitative methods and value their inclusion in the discipline, but it's not as if qualitative analysis is definitionally error-free, and "policy relevant" is very much in the eye of the beholder.

I also think that his suggestion that we focus more on theory -- which is unsurprising, since he does formal theoretical work -- is pablum. True pablum, but pablum nonetheless. Of course creating credible, rigorous theory is important and perhaps under-valued in IR, but the whole point of using stats is to evaluate theory. We don't know if a theory is credible or rigorous until we put it to some evidence-based test. Internal logic is important, but is not the end of the story.

Nevertheless it's a good post, and worth thinking about.

Note: Updated slightly for clarity shortly after posting.

The Silver Lining in the GDP Cloud

As an graduate of Econ 101 can tell you, gross domestic product is calculated according to a modest identity:

GDP = Consumption + Investment + Government + Exports - Imports

This measure of a nation's well-being is attractive for its simplicity, but it lacks nuance about the state of an economy. GDP does not measure the distribution of national income, the type of production or consumption, or the circumstances under which production and consumption take place. Today's GDP report illustrates some of this:

Growth in the last quarter was stifled by a 32.4 percent surge in imports, the largest since the first quarter of 1984, dwarfing a 9.1 percent rise in exports. That created a trade deficit, which sliced off 3.37 percentage points from GDP, the largest subtraction since the fourth quarter of 1947.

Via Felix Salmon, who notes:

Obviously, a trade surplus would be better than a trade deficit, especially in terms of generating employment growth domestically rather than abroad. But exports did rise, at quite a healthy clip. They were just eclipsed by this whopping rise in imports — which are a sign that there’s still a lot of demand in the economy.

This is a subject which came up at the Treasury blogger meeting last week: while no one at Treasury is exactly overjoyed at seeing imports rising so much faster than exports, any sign of increased economic activity is being taken as a good sign. Certainly this kind of thing is preferable to seeing the opposite happen, where exports fall and imports fall faster. Even if that would have a better effect on GDP.

So the second quarter GDP revision looks pretty bleak -- 1.6% annualized growth rate, rather than the 2.4% that was originally reported, which itself was down from 3.7% in the first quarter -- and it certainly isn't great. But the rise in exports shows the the U.S. retains some competitiveness, and the rise in imports demonstrates the presence of consumer demand in the face of a weakened dollar.

This is also one piece of evidence that we are not in a paradox of thrift world, nor that the aggregate problem is nominal, not real. That's not proof of anything, but it is one point. It also demonstrates how important trade is for the well-being of the country, even if it has a nil or negative effect on GDP.

For some of these reasons, the French government created the Commission on the Measurement of Economic Performance and Social Progress, which included such luminaries as Joseph Stiglitz and Amartya Sen, to create a different measure. Their report is interesting, although i doubt GDP will be displaced as the most common unit of measure any time soon. GDP's benefit is its simplicity and universality, even if those attributes sacrifice some nuance in the process.

Lies, Damn Lies, and Statistics

I enjoy the repartee I have with IPEZone's Emmanuel. We often disagree, but the back-and-forth is always fun. I must protest, however, to his recent challenge. In an update at the bottom of this post, Emmanuel criticizes me for not being able to use data:

Now Kindred's a good kid, but he is wet behind the ears as he tends to put words in my mouth and gets embarrassed for it. Anyway, I decided to look up the numbers for myself instead of relying on other's charts. Below are the figures for US income for those 25 and up for the years 2000 to 2008 from US Census Bureau tables P-16 and P-18. It doesn't matter whether it's the mean or the median or if you're male or female; real annual income has been falling there since 2000. It doesn't matter either if you're a college graduate or have a higher educational attainment.

I guess the furriner is more familiar with US stats than the American. Also at grad school, you are taught not to compare apples and oranges, but he does so by quoting another data series and naively suggesting differences have something to do with using "mean" and not "median" data.

There's a lot of wrong in just a few sentences there. First things first, I haven't put any words in Emmanuel's mouth; I've quoted him directly. The purpose of his post is to counter the argument that a college education is a wise investment. That's why it is titled "The Knowledge Worker Myth vs Blue Collar Reality", not "A Handful of Skilled Blue Collar Jobs Are Hot Right Now, But College Still a Good Investment Overall". That's why he writes sentences like "So much for the college myth." He is claiming at least one of two things: the college wage premium is collapsing, or the real return to a college education is falling. I am claiming the opposite.

So which of us is right? I am, of course. The data I previously reported shows this quite clearly in a few easy-to-read graphs from BLS data. Emmanuel appears to prefer Census data, but as we will see that doesn't matter. Other than the mean/median distinction, the two sources appear to disagree because the time period Emmanuel selects is misleading. His starting date is the peak of the previous business cycle (2000), while his end date is the trough of the most recent one (2008). Tables P-16 and P-18 go back to 1991, so this must have been a deliberate choice on Emmanuel's part. The series I posted go back at least to 1991 as well. The longer series show an inflation-adjusted increase over the past two decades in the return to college education for both men and women, mean and median. Over the same period, inflation-adjusted median income of high school graduates declined by nearly 10%. (The same figure increased for women... from $16k to $18k.) Male and female college graduates made nearly double what those with only a high school education made across the time series. If you compare peak-to-peak, incomes for college graduates went up from 1999-2007 as well. The spread between the college-educated and those who are not also increased across the series, and those without degrees have seen especially nasty upticks in unemployment in the last two years or so.

I'd produce a nifty graph here but the Census data is a mess, it's late, and I'm tired. Please read the tables for yourself. A quick glance is good enough. Or look at the BLS data, which tells the same story. I'm not the one talking about apples and oranges. Whether you're looking at absolute or relative gains, means or medians, male or female, the pattern is clear: the more educated earn much more than the less educated, that disparity has increased over time, and job security (as well as non-wage benefits and compensation) is also much higher for the well educated. The "Knowledge Worker" is anything but a myth, which is probably why Emmanuel's pet countries like Singapore have emphasized higher educated so much.

The only way Emmanuel can reach any other conclusion is by tweaking starting dates in a pretty egregious way, and changing his argument midstream. As I say: lies, damn lies, and statistics. Whether it's intentional or not is not clear to me, but just as in a previous tangle (see comments), Emmanuel loves to make broad claims and then say "That's not what I meant" when questioned. As before, he meant what he insinuated, and what he insinuated was wrong.

The Importance of Starting Points

This is a very good object lesson on why starting points matter when making time series comparisons, especially cross-sectionally:

But is Iceland’s post-crisis “miracle” real? No. It is an illusion created by the starting date Krugman chose for his figure. If we shift it back just one quarter – the quarter before Latvia and Estonia’s GDP peak – Iceland’s performance no longer stands out...

Iceland’s massive devaluation improved the country’s trade competitiveness, while imposing huge losses on its krona-based savers. Ireland’s inability to devalue protected its citizens’ euro-based savings, but has forced it to improve competitiveness in other ways, such as through wage cuts. Of the lessons that can reasonably be drawn from Iceland’s experience over the past decade, the benefits to tiny statelets of having a currency to debase is hardly one of them.

Graphs and more analysis at the link. One lesson is that a whole lot of time series stats are questionable, especially those coming from partisan think tanks or public ideologues. Another is that sometimes it's really hard to make appropriate comparisons, because choosing a starting date is arbitrary by nature. I don't mean to pick on Krugman here (this time) because I don't think he's doing anything intentionally wrong. The point is to be careful when producing, or consuming, statistical information.

RIP, David Blackwell

David Blackwell, pioneer in probability theory, game theory, and other mathematics, has died. A truly remarkable man, from a poor Midwestern town (close to where I grew up, incidentally). First black tenured faculty member at UC-Berkeley (after being turned down at Princeton), first black member of the National Academy of Sciences, and an early applier of mathematical game theory to conflict situations at the RAND Corporation. RIP.

It Depends on Your Posterior Distribution

Nerdy joke-question of the day comes from Double-D:

If you're doing a PhD in Applied Statistics, specialising in sampling theory, how many times do they make you write your dissertation?

(This was the basis of an actual conversation between me and a colleague, where we were arguing over what I believed to be an entirely sensible generalisation from a single case. He pointed out that he had a degree in statistics and thus could be assumed to have expertise in the area; I countered that he only had a single degree in statistics and would have to take his finals at least 30 times before I could be confident he hadn't passed them by luck).

Probably depends on whether you're a Bayesian or not.

Hans Rosling Ted Talk

These are two of the most interesting, engaging and funny Ted Talks I 've ever seen. They were given by Hans Rosling, a Professor of International Health from Sweden in 2006 and 2007. From Ted:

You've never seen data presented like this. With the drama and urgency of a sportscaster, statistics guru Hans Rosling debunks myths about the so-called "developing world."

Dr. Rosling tackles international development and the use of statistics and data in the first lecture and poverty and development in the second lecture. They truly are a must see.

(ht: Andy Pennock)