Ramblings on Random Research

Since I was travelling a lot here a summary of highlights of research I stumbled across over the last weeks.




Gonna start with a Zodiac paper! As Sinophiles and Game of Thrones fans know, being born in the year of the dragon brings luck. Chinese mothers actually, to some extent, try to time their births in order to get dragon kids. And *insert gong noise* dragon kids actually do end up with better high school testscores and higher rates of college entrace.


The authors test for whether this is driven by characteristics of the parents who self-select to give birth in the dragon year (more educated mothers and fathers controlling fertility better etc.). However this (contrary to what I expected) does not seem to be driving the kids outcomes here. Then the authors check the Chinese Social Survey and find that parents of dragon kids are more likely to believe in their kids’ chances to succeed and that this seems to be driving the better child outcomes. So the paper leaves us with a story about the power of myth and superstituation as well as with an uplifting ‘Believe in Your Kid’ story. Enter the Dragon!

One small technical remark on that paper: while I think the authors do a good job controlling for parental characteristics, work on Chinese immigrants in the US has shown that in that case it was actually fertility selection that led to better child outcomes. Also fertility control might be linked to positive non-observed parental characteristics. An additional test I would propose to the authors is to check the outcomes of non-dragon year siblings of dragon kids (older and younger seperate because of sibling peer effects). This could be done with the CFPS data I guess.

More Demography! “Diverging Destinies”, the idea that existing inequalities are exacerbated by trends in family structure has been a hot topic over the last years.  While original work mostly focussed on the US the idea is often easily applied to other national contexts. Maybe too easily as new work by Juho Harkönen shows. By looking at the link between single motherhood, education and child poverty in 33 countries, this is one of the first works to test the “Diverging Destinies” idea cross-nationally. While single motherhood has been going up universally it only augments inequality by education if more educated mothers systematically divorce less and single motherhood leads to big increases in child poverty. This is not a feature that is universally present in all countries and as the paper argues, the relationship of single motherhood to poverty is something policy can influence.

Even more (kinda) demography! Is Urbanization linked to bigger governments? This question hasn’t really been asked this way before and  new work argues … *drumroll* … well yea, it is. Of course the standard correlation causation yadda yadda disclaimer applies (using lagged government spending as an instrument is probably the best the authors could do to try to dispel doubts). They also show that the link doesn’t only hold cross-nationally but also between regions (in Germany and Columbia) and that urbanization is associated with a greater openness to redistribution in opinion surveys. Personally I see this as a factoid confirming my long-held pet theory that higher density means more externalities and hence more need for government, but make of that what you will.


And a hint for (demography) practicioners in the area of same-sex couple research: Don’t identify same sex couples by looking at sex-composition in survey data, coding errors might be driving your results (here)

Glaeser and Ponzetti have a timely theoretical contribution. They look at the “fundamental attribution problem” in leader selection, namely that we tend to overattribute political and economic results to leader characteristics, where in reality they are often due to luck or institutions. Introducing this attribution error into political choice models, shows that it leads to better leader behaviour, but also that it leads to a higher demand for dictatorship and to less demand for a free press and institutional investments than is optimal.

Another timely contribution: Novoknet, Piketty and Zuckman look at the development of inequality and property in Russia. They do a great job telling the story of the oligarchization of the country in graphs.

Here you see the increasing income share of the top 1%


The next graph shows the distribution of income gains since 1989 (notice all gains go to  the top 10%). Some people have raised questions about the measurement of income and the consumption basket before 89 regarding this graph and those issues are hard to deal with correctly, but I think the story of very unequal gains stands beyond a doubt.


They also do a great job capturing the comparatively extreme share of offshore wealth that post-communist Russia has accumulated.



Since we are looking at what happened after the Iron Curtain fell: The larger scale agriculture that was imposed in Eastern Germany led to its agriculture being more productive and its biodiversity being lower than in Western Germany (here). Switching to organic farming increased biodiversity in both regions, something policymakers may consider as they make agricultural policy.

“Causality is not Correlation” is the mantra everyone has learned to sing. But what to do about it? One way to try to get around the problem is to use instrumental variables. The main idea is simple: if you are interested in the relationship between two variables, let’s say income and consumption then find something that randomly makes income vary but that has no independent effect on consumption, like say a lottery win. Then you can use the variation in income induced by the lottery instrument to look at how it affects consumption. A new paper tries to look at the effects of child obesity on the child’s medical expenditures and the instrumental variable it uses is maternal obesity. The approach is laudable as we lack causal estimates of obesity (a household with obese children might have lots of characteristics that drive medical expenditures that we cannot measure and that are not related to obesity directly so we cannot measure the cost of obesity correctly). The findings are that childhood obesity really increases medical spending quite a lot (even more than we thought). I like this paper a lot yet I still have problems with the instrumental variable it uses. There is a good case to be made that the exclusion restriction is violated, in particular since more obese parents probably have higher medical treatment costs themselves and through the increased contact with the medical system, lower barriers of entry for getting their child treated (I think the authors did not control for parental medical expenses).

There was actually another paper out this week, which instrumented obesity. What they used was an individuals polygenic score (very simplified: genetic predispotion) for obesity to test for peer effects among siblings. So they only use the variation in obesity that is induced by genetic predisposition, while controlling for the siblings genetic predisposition (pfeeew still following?) More precisely they test whether having an obese sibling has potentially a direct causal effect (via role modelling, copying of behaviour etc.) on an individuals obesity and yes it does. Methodologically I find this paper superinteresting. The growing number of instrumental variables that come out of the new genetic data will be a big deal in the next decade of social science research. Let’s continue with genetics research then.

Todo Sobre Mi Madre: Since Mitochondrial DNA is usually maternally inherited, mutations that are deleterious to (ie suck for) males but are beneficial or neutral to females are not selected against. This is “The Mother’s Curse”, which may or may not, also be the title of the first single from my indie rock band “Oedipus”. New work shows this effect to be at work when looking at male and female fitness with respect to Leber’s hereditary optical neuropathy in 290 years of population data. The authors suggest the mother’s curse might also be something to look at when thinking about male-female life expectancy differences. Excellent paper for thanksgiving dinner discussions in any case.


Economists are doing what researchers like to do most, look at citations, this time to look at disciplines insularity.  So we can see that, yay, economics is getting less insular, but the shining stars of integrated social sciences are clearly sociology and political science as you can see in the graph below.


When looking at which disciplines are cited by others, you can see that the influence of economics has been increasing a lot.


I think John Lennon got it right when he sang “Imagine all the social science disciplines living in the world as one”. May he rest in peace.

In the economic history department Franck and Galor look at the long-run effects of the industrial revolution in France. And they are not all that great. Areas that industrialized faster are poorer nowadays and the channel seems to be that the prevalence of industrial labour led to lower investments in human capital. Oh and here is horsepower of steam engines by departments in 1860-65 map! The first french steam engine was installed in Fresnes sur Escaut, in case you wanna be the guy winning the next pub quiz or annoying everyone at a dinner party or in case you are already planning your historic steam engine trip of Europe!steam.PNG

Also: a higher earned income tax-credit leads to improvements in child-birthweight (here)

Probably 10% of adults you cross have taken anti-depressants at some point in their lives and that number is growing rapidly, yet they remain largely ignored by the social sciences. Katolik and Oswald present an overview of the literature that exists so far.

and: midlife-crisis is for real, which has me going for a drink, folks!



P.S.: I find how the authors of that last paper use the term “humans” in the title as if they were dealing with some alien species, terribly endearing.

About Sander Wagner
I am

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: