I wrote these two problem sets as a teaching assistant for the graduate Industrial Organization course (Economics 543) at Penn State. Both cover demand estimation, from logit and nested-logit models to random coefficients (BLP), with full solutions in R and Julia.
Estimating multinomial and nested-logit demand from market shares, following Berry (1994): inverting shares to recover mean utilities, computing own- and cross-price elasticities, recovering marginal costs from the Bertrand first-order conditions, and simulating the price effects of a two-firm merger and of full industry collusion.
Full random-coefficients (mixed-logit) demand estimation in the tradition of BLP: simulating consumer heterogeneity, computing predicted shares by Monte Carlo integration, inverting shares for mean utilities with the BLP contraction mapping, and estimating the nonlinear parameters by GMM using Gandhi–Houde (2023) differentiation instruments and an analytic gradient.