This supplementary appendix is divided into seven sections. In Section S1 we describe the computational algorithms. In Section S2 we deal with inference, both from a large-N, T perspective and from a large-N, fixed-T perspective. In Section S3 we treat the issues of misspecification of the number of groups G and its choice. In Section S4 we study two extensions of the baseline model, which allow for unit-specific heterogeneity and for group-specific coefficients, respectively. In Section S5 we deal with several other issues, including the connection with mixture models, and how to incorporate prior information in estimation. In Section S6 we report the results of a simulation study. Lastly, in Section S7 we show a number of additional results related to the empirical application.