Trends in NBER Working Papers
I received an email from Jim Poterba this morning on the occasion of the 30,000th NBER working paper. Excerpting, it said:
The NBER reached a milestone this morning with the distribution of our 30,000th working paper. The series was launched in 1973 by labor economist Robert Michael to disseminate research prior to journal publication and to facilitate distribution of data appendices and related supplemental material. Working paper number 1 was Education, Information, and Efficiency by Finis Welch.
The series began on a modest scale, reflecting the small number of NBER affiliates at the time. There were 41 working papers in the first year, and it took 12 years to reach the 1000 paper mark. Originally, working papers were printed and had bright yellow covers. Packets of papers were mailed occasionally to libraries, leading economics departments, and research institutes. As the number of NBER researchers expanded, the volume of working papers rose. Eventually, a shift to digital distribution became essential for accommodating the expanding number of studies.
In 2020, when many economists ratcheted up their research output to address the many new questions posed by the COVID-19 pandemic, a record 1,713 working papers were distributed. The annual average for the last five years was 1,322. More than 25,000 subscribers receive the “New This Week” email each Monday, and there were more than 2.9 million paper downloads in 2021. Twitter has become an increasingly important channel for calling attention to working paper content.
The NBER working papers provide some insights on the changing structure of economic research. For example, 60 percent of the papers distributed during the series’ first decade had a single author, while 35 percent were coauthored and 5 percent had more than two authors. In the last decade, only 11 percent had a single author; 56 percent had three or more. The number of working papers per NBER affiliate per year, which was more than 1.7 in 1980, has trended down. It averaged about 0.95 at the turn of the century, and was 0.78 in most recent five years excluding 2020.
In the image at the top of this post, you can see that there exist some older retrospectives on NBER working papers.
The most interesting trend is that people seem to be involved in about as many papers per year as they were before, but with more coauthors, so that papers on CV has not had a big trend, but if only fractional credit were given for papers with coauthors, it would look like productivity had gone down. But I think that papers are more ambitious than they used to be, in rough proportion to the increased number of coauthors. (In the cross-section, I think the number of citations goes up roughly with the square root of the number of coauthors, but the time-series trends would be different from that.)