big data and economic research

Liran Einav and Jonathan D. Levin highlight how "big data" could change economic research. Being a regular data cruncher myself, this is a welcome contribution as (applied micro-)economists have been skilled in the use of large data for quite some time.

Some highlights:

• "Government administrative data are almost certainly under-utilized, both by government agencies and, because of limited and restricted access, by researchers" [my emphasis!]--hear, hear!
  
• "Many government agencies are increasingly smart about using data analytics to improve their operations and services. However, most agencies almost surely lag behind"...
  

For all my students, this is good news---quantitative skills and analytical minds will be in demand, probably increasingly so, whether you'll be working in a private firm, in the public administration or as an academic.

boys and girls

Maybe parents treat their female and male children differently. Previous evidence indicates that parents spend more time with pre-teen sons than with pre-teen daughters, a difference that is caused by fathers spending more time with their sons.

Michael Baker and Kevin Milligan have new evidence that this extra time from fathers only emerges with age, and is not present when children are very young:

It is [pre-school] girls, not boys, who systematically, in three leading developed countries [i.e., Canada, the UK, and the US], receive more of these time inputs from their parents

But does this matter for academic achievement? They show that these differences in time are indeed important:

in each country the boy-girl difference in inputs can account for a non-trivial proportion of the boy-girl difference in preschool reading and math scores

We need more experiments

The assistence job seekers receive from Unemployment Offices to find jobs is perhaps not as good as we might hope. A Study by Bruno Crépon, Esther Duflo, Marc Gurgand, Roland Rathelot, and Philippe Zamora (QJE; working paper version here) assesses youth unemployment policy in France using a controlled experiment:

"Job counselling" – a key French policy that prepares some job seekers for the recruitment process, and connects them with potential employers – seems to only marginally improve graduate’s chances of employment. Moreover, the evidence suggests that what’s good for one graduate may be bad for another: the beneficiaries of intensive job counselling are more likely to find employment simply at the expense of other job seekers. (Job placement and displacement.)

We need more experiments.

Weights in Regressions

Solon et al. have a new paper on the use of weights in empirical analysis. Not only does the paper have a great title, but it addresses one of the more hairy issues in empirical work, when and how to use weights:

We discuss three distinct potential motives for weighting when estimating causal effects: (1) to achieve precise estimates by correcting for heteroskedasticity, (2) to achieve consistent estimates by correcting for endogenous sampling, and (3) to identify average partial effects in the presence of unmodeled heterogeneity of effects. In each case, we find that the motive sometimes does not apply in situations where practitioners often assume it does.

There is, of course, no foolproof recipe:

In situations in which you might be inclined to weight, it often is useful to report both weighted and unweighted estimates and to discuss what the contrast implies for the interpretation of the results. And, in many of the situations we have discussed, it is advisable to use robust standard error estimates.

politikberatung durch ökonomInnen--Caveat emptor!

Haucap und Mödl, "Zum Verhältnis von Spitzenforschung und Politikberatung":

"Ökonomen agieren mit ihrer Forschung also im Wesentlichen auf zwei Märkten: Zum einen auf dem Markt für wissenschaftliche Publikationen im engeren Sinne, also den Fachzeitschriften, und zum anderen auf dem Markt für wirtschaftspolitische Beratung."

Sie zeigen:
* "nur jeder fünfte Artikel eines Ökonomen, der an einer Institution in Deutschland, Österreich oder der Schweiz forscht, [hat] einen besonderen Bezug zu spezifischen Problemen der Wirtschaftspolitik oder Ökonomie im deutschsprachigen Raum"

* "es [sind] primär nicht die forschungsstärksten Ökonomen ..., die sich in der Politikberatung engagieren"

Copyrighted Music

New research by Joel Waldfogel shows that the erosion of copyright does not necessarily lead to worse products. The study "Copyright Protection, Technological Change, and the Quality of New Products: Evidence from Recorded Music since Napster", published in the Journal of Law and Economics (working paper version), assesses the quality of recorded music since Napster.

Three different approaches are used (critics’ lists, music sales, and airplay data), assuming that if the music is better, it should be sold or played more often. Contrary to my own view, there is no evidence that the music released since Napster was of lesser quality. Copyright protection should therefore be weighted against producer surplus and consumer surplus.

boys and girls in the classroom

New evidence from Italy has strong implications for mixed high-school education. (See also this.) Anelli and Peri find that girls in single sex classes are more likely to study "hard" subjects (e.g., economics or engineering) at University than those in mixed-sex classes.

The policy conclusion are clear:

If an objective of schooling is to increase women’s career opportunities and thereby their salaries, our results would suggest that gender-separated classrooms would be an effective step in the right direction. Gender-separated classrooms would increase the probability of choosing high-earning majors for both women and men.

Similar evidence, for university students, have been obtained by Booth and Nolan: "after eight weeks in a single-sex environment, women were significantly more likely to choose the [risky] lottery".

While the Italian evidence is from a "natural experiment", the evidence by Booth and Nolan is from an actual field-experiment, following an earlier study in high-schools. Good schools should offer students the possibility of choosing single-sex classes or offer the possibility of a random lottery over single-sex or co-ed classes. (Which provides additional data for analyses of peer-effects and subsequent outcomes.)