A spatial panel data model for estimating the impact of social and economic determinants on opioid mortality rates in the US

This paper employs a spatial Durbin panel data model, an extension of the crosssectional
spatial Durbin model to a panel data framework, to estimate the impact of a set of social
and economic determinants on opioid-induced mortality in the US. The empirical model uses a
pool of US states over six years from 2014 to 2019 and a k=8 neighbor matrix that represents the
topological structure between the states. Calculation of direct (own state) and indirect (cross-state
spillovers) effects estimates – based on Bayesian estimation and inference – reflects a proper
interpretation of the marginal effects for our nonlinear model that involves lags of the dependent
variable vector. The study provides evidence for the existence of spatial effects working through
the dependent variable vector and points to the importance of larger indirect effects of Asian
and Hispanic/Latino minorities on the one side and the population age groups 35-44 years and
65 years and older on the other. This finding echoes the first law of geography that ”everything
is related to everything else, but near things are more related than distant things” (Tobler
1970). Space – largely neglected in previous research – matters for gaining a valid and better understanding
of why and how neighboring states contribute to opioid-induced mortality in the states.