Sunday, August 03, 2008

Network Effects: The Case of Civil War Blogs

How does information spread across a diverse population? When an objectively important economics working paper is released, how much time passes before the leading popular media outlets (i.e the New York Post) is aware of the study? How much has the Internet and networks such as SSRN and the NBER Working Paper series accelerating this diffusion?

Why is this interesting? If a tree falls in the forest and nobody hears it ... What is the difference between a tree and a research paper if nobody reads it? Nothing. The expectation of effective networks creates incentives for researchers to write papers. Now a cynic might ask whether the expectation that a "sexy" paper will get popular attention encourages creative researchers to engage in too much freakonomics? I have blogged on this subject before. If you recall my main point, there are huge financial incentives in academic economics if you remain "pure" and are branded a leading theorist or econometrician. Yes, being quoted in the New York Times is fun and offers a short run immediate payout but there are very high medium and long term returns for working on fundamental economic questions.

Returning to the theme of network effects, I'm happy to see that Civil War blogs are aware of my forthcoming book; A Sophisticated U.S Civil War Blog

Brett Schulte will need to take a close look at our book. Yes Civil War companies were formed locally. Men fought in groups of 100. In our work, we use variation in the characteristics of these communities to examine how "peer groups" affect such important outcomes as desertion, POW Camp survival, and ex-slave soldier literacy.

Many Civil War studies rely on soldier diaries as the basis of the analysis. We respect such diary based research but there are limitations to what can be learned from such a self-selected sample. Many black soldiers couldn't read or write. You can't expect to see many diairies written by such a sub-group. Our main empirical sources is a unique longitudinal data set that covers over 40,000 soldiers (white and black). I will have more to say soon; but go to to go to the Princeton Press weblink.

Before I forget, one other example of Network Effects is the network. Suppose that an economist issues two working papers a year and randomizes to release one as a NBER working paper and posts the other on his web page. What would be the difference in the diffusion of the two papers?

At the Martin Feldstein dinner last week, Jim Poterba asked the 350+ audience to consider what their life would be like without the NBER. Access to this network is valuable. I conjectured that the average NBER member would be willing to pay $15,000 for this access --- so Jim Poterba could collect $15 million a year to keep this operation going. Some people claimed that I was over-stating the value of the network but I disagree. Would cogent argument is that the SSRN network provides a substitute but it doesn't screen out anything.