Score-driven dynamic patent count panel data models

Abstract

In this paper, we propose the use of Dynamic Conditional Score (DCS) count panel data models. We compare the statistical performance of the static model with different dynamic models: finite distributed lag, exponential feedback and different DCS models. For DCS, we consider random walk or quasi-autoregressive dynamics. We use panel data for a large cross section of United States firms for period 1979–2000, and the Poisson quasi-maximum likelihood estimator with fixed effects. The empirical results suggest that DCS has the best statistical performance.