The fraction of women in economics has grown significantly over the last forty years. In spite of this, the differences in research output between men and women are large and persistent. These output differences are related to differences in the co-authorship networks of men and women: women have fewer collaborators, collaborate more often with the same co-authors, and a higher fraction of their co-authors are co-authors of each other. Moreover, women collaborate more and do so with more senior co-authors. Standard models of homophily and discrimination cannot account for these differences. We discuss how differences in risk aversion and an adverse environment for women can explain them.