-
This study investigates the efficiency of the dissemination of working papers in economics.
-
Results show strong evidence that the NBER WP series suffers from overcrowding and congestion. [analogous to traffic/network congestion; congestion pricing for papers…?]
-
Increases in the number of weekly released working papers reduces downloads, abstract views, and media attention for each paper.
-
Working papers from more prominent authors suffer just as much as those from less prominent authors.
[supplement] Using data on the NBER working paper series, we show that the dissemination of economics research suffers from a congestion problem:
An increase in the number of weekly released working papers on average reduces downloads, abstract views, and media attention for each paper. Subsequent publishing and citation outcomes are harmed as well. Papers written by prominent authors are not immune to this congestion effect.
Finally, suggestive evidence on viewership and downloads implies that working papers substitute for the dissemination function of publication.
Our results highlight how readers face time and cognitive constraints, with increased congestion in working papers leading to real impacts on how research is consumed.
[Keywords: economics working papers, economics research]
…A final advantage to studying the NBER WP series comes from how the NBER disseminates papers. Throughout the course of each week, NBER affiliates submit their working papers to the NBER without knowing how many other papers have been submitted that week. Then, on Monday of the following week, all submissions are released together and distributed to subscribers. Hence, these weekly releases generate plausibly exogenous variation in the “crowdedness” of the working paper space. As such, any systematic variation within the calendar year in both research productivity and quality can be accounted for in models with higher dimensional time fixed effects (eg. week-of-year fixed effects), with only idiosyncratic variation remaining.
…This effect is particularly sharp in the first several months of the paper’s release. Using data from Altmetric, a company that tracks academic papers across news outlets, blogs, and social media (eg. Twitter), we find that NBER WPs also receive less media attention when the number of weekly NBER WPs increases. Doubling the number of weekly releases reduces a paper’s probability of being covered in the media by over 30%. Further highlighting the importance of these dissemination findings, we find that the NBER WP version of eventually-published papers receives more downloads and abstract views than their published counterpart. Thus, working versions of papers substitute for the dissemination function of the publication process, yet suffer from idiosyncratic variation in the crowdedness of the working paper space…Perhaps most importantly, we also find that publication prospects and citations are harmed from this working paper congestion. Doubling the number of weekly NBER WPs reduces a paper’s probability of publishing by over 4%.8 Interestingly, we find no effects on the “quality” of the publishing journal, suggesting a net loss in publishing outcomes for papers when released with a greater number of peer papers. Subsequent citations drop as well: doubling the number of NBER releases reduces citations by ~7.5%…Despite the highly selected sample of NBER papers, we find that 26% of NBER WPs never publish.
…Furthermore, we estimate quantile regressions for viewership and citation outcomes to find that the higher quantiles of the distribution experience sharper losses in viewership, suggesting that if anything, ex post more “important” papers suffer more from working paper congestion.
