Stan Liebowitz (UT-Dallas) has posted his latest on the file sharing debate, “The Oberholzer-Gee/Strumpf File-Sharing Instrument Fails the Laugh Test.” (HT: Craig Newmark). Having covered the file-sharing debate on this blog, including Professor Strumpf’s remarkably unprofessional and uncharitable treatment of Professor Liebowitz, I greatly admire Stan’s ability to stay above the fray and stick to the econometrics. In my view, his most recent critique of the O/S instrument (Number of German Students Vacationing) and the results it generates is systematic, careful, and devastating. Among other things, Liebowitz demonstrates that the “O/S regression result implies that, from average levels, it would take a mere 2 million Germans in this cohort turning off their computers to completely eliminate US file-sharing even though approximately 100 million other file-sharers are still online with their files available to American file-sharers.” This is a paper that should be read broadly. Here is the abstract:
I examine the key instrument (German kids on vacation) used by Professors Oberholzer-Gee and Strumpf in their analysis of the impact of file-sharing on record sales, published as the lead article in the Feb 2007 JPE. Their measured relationship between the instrument (German students on vacation) and the variable that it is instrumenting for, American downloading, is seen to have outlandish implications in the often overlooked first stage regressions. The coefficient implies that if German secondary students all go to school, American file-sharing would drop to zero. A nonsensical result of this sort indicates an important error somewhere in their data or analysis. The instrument is also shown to be related to American record sales, contrary to the claims of Professors Oberholzer-Gee and Strumpf, and contrary to the requirements of their analysis. Their data on file-sharing, which Professors Oberholzer-Gee and Strumpf state is representative of worldwide file-sharing, is actually biased according to some of their own statistics which they failed to examine, considerably overstating the share of German files. Finally, I demonstrate that German students on vacation cannot have a measurable impact on American downloading (and thus American record sales) negating its potential usefulness and implying that the approach taken by Professors Oberholzer-Gee and Strumpf could never have provided useful information about the impact of file-sharing on record sales.