(This is an update on a post I had initially written at the start of the academic year. I figured it would come in handy, as many of us are busy writing our syllabi for the spring semester.)
If you teach in a policy school, a political science department, or in an economics department that grants Bachelor of Arts instead of Bachelor of Science degrees, chances are some of your students are not quite conversant in the quantitative methods used in the social sciences.
Many of the students who sign up for my fall seminar on the Microeconomics of International Development Policy or for my spring seminar on Law, Economics and Organization, for example, are incredibly bright, but they are not familiar with regression analysis, and so they don’t know how to read a regression table.
This makes it difficult to assign empirical papers in World Development for in-class discussion, let alone empirical papers in the Journal of Development Economics.
While I do not have the time to teach basic econometrics to students in those seminars, I have prepared two handouts for them to read in preparation for reading papers containing empirical results, which I thought I should make available to anyone who would rather not spend precious class time teaching the basics of quantitative methods. I have used both these handouts in my development seminar last fall, and my students said that they had learned quite a bit from reading them.
The first handout is a primer on linear regression, which shows analytically and graphically (and hopefully painlessly) what a regression does, and why it is such a useful tool in the social sciences. Perhaps more importantly, this handout also explains how to read a regression table.
The second handout primer on the identification of causal relationships in the social sciences, which discusses the distinction between correlation and causation and explains two ways in which social scientists go about making causal statements (i.e., randomized controlled trials and instrumental variables), with a few examples. I suggest supplementing this handout with a reading of Jim Manzi’s “What Social Science Does–and Doesn’t–Know” in City Journal as well as with Esther Duflo’s TED talk.
Of course, neither handout is a substitute for a course in econometrics or on research design, respectively, but these handouts are intended primarily for undergraduates or Masters students with little to no quantitative background.
(Update: This post by Tom Pepinsky also offers a very good introduction to the identification of causal relationships. HT to Chris Blattman for this great find.)
There’s No Free (Causality) Lunch
First of all, causality requires identification. Vector autoregressions (VARs) do not provide any automatic or free identification. To do policy analysis with a VAR (as opposed to agnostic forecasting) one has to make the same type of untestable identifying assumptions here as one does in the older, explicitly simultaneous equation, Cowles commission approach.
The most common way of identifying a VAR (ordering the variables and performing a Cholesky decomposition) is EXACTLY the same as using exclusion restrictions to identify a system of equations. Other structural VARS do NOT remove the need for identifying assumptions. VARS are not a free lunch.
That is Kevin Grier, in a post over at Kids Prefer Cheese.
I was looking forward to someone finally saying it, as many of the media discussions of the 2011 Nobel prize for economics somehow made it sound as though the work of Sargent and Sims allowed us to estimate causal relationships.
Not so. The estimation of causal relationships is difficult in the social sciences for the simple reason that we almost never observe the counterfactual. This is especially true in macroeconomics, where the opportunities to run an experiment are few and far between. And even in more micro contexts, we’re not completely sure about much.