A model of the editorial process in academic journals

Editors of academic journals make their acceptance or rejection decisions about submitted papers based on their own prior assessment of the intrinsic quality of these papers augmented by the information in reviewer recommendations. In this paper, we theoretically analyze the editorial process of academic journals, and in particular, the editors’ extraction of information about intrinsic paper quality from reviewers. We assume that, if a reviewer’s own research is close to the research area of a paper, he is likely to have greater expertise in evaluating that paper (i.e., be an “expert” reviewer) but is also more likely to be positively or negatively biased with respect to it. On the other hand, “generalist” reviewers, whose own research is further away from the research area of the paper, are likely to be unbiased about it; however, their expertise in evaluating the paper is likely to be lower as well. We further argue that the editorial decisions of journals will deviate considerably from the socially optimal rule of accepting good papers and rejecting bad papers if the above potential reviewer biases are not taken into account by editors when choosing reviewers. We show that editors can make better editorial decisions if they choose the appropriate type of reviewer to evaluate a paper (in the one reviewer case) or the appropriate combination of reviewer types (in the two reviewer case), based on their own prior assessment of submitted papers. We also show that, if the editor can aggregate the information contained in multiple reviews efficiently, two reviewers are better than one as long as the cost of using an additional reviewer is moderate; however, two reviewers may be worse than one if the editor adopts ad hoc decision making rules such as requiring both reviewers to recommend acceptance of a paper before the journal can accept it.