A Regression Mystery, Solved

All the economists who read this blog will no doubt be familiar with the popular instrumental variables (IV) regression technique, which is used to estimate the coefficient of endogenous variables. But who established the technique as a solution to the identification problem? IV regression was first mentioned as a solution in the appendix of a 1928 book by Philip Wright; however, there’s considerable debate surrounding the identity of the appendix’s author. Authorship has been attributed, at various times, to both Philip Wright and his son, the statistician Sewall Wright. A few years ago, James Stock and Francesco Trebbi set out to answer that question by analyzing the writing style in the appendix. The authors concluded that the evidence “points toward Philip as being both the author of Appendix B and the man who first solved the identification problem.” [%comments]


Chris

Well I'm glad that's finally solved!

RandFan

I guess today's topic makes clear that this blog is written by and for economists. Most of the topics are so interesting and clever that I tend to forget that.

scott Cunningham

I'd be interested in hearing why John Snow shouldn't be given the credit of first offering IV as a solution to identification problems for his work in the mid-1800s in epidemiology. Paul Grootendorst argues this, for instance.

http://individual.utoronto.ca/grootendorst/pdf/IV_Paper_Sept6_2007.pdf

Starting on page 3:

Cholera transmission
Although IV theory has been developed primarily by economists, the method originated in epidemiology. IV was used to investigate the route of cholera transmission during the London cholera epidemic of 1853-54. A scientist from that era, John Snow, hypothesized that cholera was waterborne. To test this, he could have tested whether those who drank purer water had lower risk of contracting cholera. In other words, he could have assessed the correlation between water purity (

Me

Or they could have worked together. Father writes it, son figures it out.

scott Cunningham

To Me: this was actually what the two authors were trying to test. They had writing samples of father and son through various econometric methods looking at phrases of words concluded the Appendix was actually written by the father. It's a similar method as that used to identify the identity of the anonymously written book PRIMARY COLORS, fyi.

Eva

Neat little piece of info, thanks!

ginny

...estimate the coefficient for endogeneous variables. WTH

Wayne

An instrument variable is a pretty ingenious way to try to isolate the effect of variable X on variable Y in the situation where not only does X have an effect on Y, but Y also has an effect on X.

It's discussed in Freakonomics; the number of cops on the street is very closely connected to the crime rate, but since crime levels also affect police hirings, Levitt uses an instrument - mayoral elections - to tease out the effect of cops on crime.

It's not ideal - if randomized experiments are the gold standard, IV is the silver - but it's really clever, all the same!