Identification in Some Random Coefficients Panel Data Models (With Application to Quantile Regression)

This paper considers a random coefficients panel data model with individualspecific intercepts (or fixed effects). The identification of the distribution of random slope coefficients is established in two settings: when random slope coefficients are conditionally independent from individual-specific intercepts; and when individual-specific intercepts are allowed to depend on random slope coefficients, too. The identification result requires only two observations per each unit. The paper considers a quantile regression panel data model with fixed effects as a special case of random coefficients panel data model and provides sufficient conditions under which conditional quantiles functions are identified. All identification results are constructive, and estimation procedure based on these results is proposed.