Spatial Models

Since the turn of the century, the interest for spatial models in marketing science has increased significantly. An econometric model becomes spatial if the behavior of one economic agent is codetermined by the dependent variable, the explanatory variables, and/or the error term observed on other economic agents, known as respectively the spatially lagged dependent variable, spatially lagged explanatory variables, and the spatially lagged error term, or shortly spatial lags. Except for these spatial lags, the degree of codetermination also depends on the set of agents affecting the focal agent; these mutual relationships among economic agents is generally modeled by the so-called spatial weights matrix W. It should be stressed that the term spatial needs to be read here in the broadest sense of the word, since spatial models are synonymous to several alternative terminologies used in marketing science, among which social interactions, word of mouth, peer effects, neighborhood effects, contagion, imitation, network diffusion and interdependent preferences