A Model of Online Publishing: the Case of arXiv.org

Abstract

arXiv.org is a publishing platform, without formal refereeing, that is

mainly used in physics and math. Taking arXiv.org as an example, we show that authors can self-referee, i.e., among authors of the same quality, those with good papers are more likely to make papers more visible to readers to help readers evaluate papers. Therefore, authors play a role analogous to referees in the peer review process. Our work suggests that online publishing, like arXiv.org, provides an alternative to journals, where the traditional role of evaluating papers through peer review is replaced by authors' self-refereeing. We also address the broader question: do producers have incentives to reveal the quality of their bad products?

We propose that authors' self-refereeing is driven by their reputation

concerns. Given papers of the same quality, good authors make their

papers less visible than bad authors in order to benet from the reputation improvement in the future. Good authors are more likely to invest in their reputation because they are more likely, than bad authors, to have good papers in the future. In general, good papers are cited more than bad papers. This citation advantage is higher if a good paper's author invested in their reputation. The more limited readers' attentions are, the more likely authors are to self-referee, because authors with good papers incur a higher cost for their papers to be missed by readers. In Chapter 1, we identify properties of authors' strategic submissions, which motivates the model that we present in Chapter 2. In Chapter 3, we collect data and test the theoretical model of Chapter 2.