An increasing number of economists publish in teams. This paper attempts to examine what makes two economists collaborate and how much productivity gain comes with collaborations. I find that as the difference in the average publication quality, as measured by the journal quality index, gets smaller, two economists are more likely to collaborate. Moreover, similarity in log citation counts and being at the same institution both increase the chance of collaboration. There is also evidence of horizontal sorting (two authors of similar productivity) as well as vertical sorting (two authors of fairly different productivity). On average, a two-person team benefits from a 9% collaboration gain whereas a three-person team benefits from a 20% gain.