Abstract
The spatial lag of X model, a linear regression model extended
to include explanatory variables observed on neighboring cross-
sectional units, is the simplest spatial econometric model producing
flexible spatial spillover effects and the easiest model to parameterize
the spatial weights matrix, denoted by W , describing the spatial
arrangement of the units in the sample. Nevertheless, it has received
relatively little attention in the theoretical and applied spatial
econometrics literature. This study fills this gap by considering several
extensions of its basic form. It is found that the claim made in many
empirical studies that their results are robust to the specification of W
is not sufficiently substantiated. Especially the spatial spillover effects,
often the main interest of spatial economic and econometric studies,
turn out to be sensitive to the specification of W. In addition, it is
found that the common practice to adopt the same W for every spatial
lag should be rejected. These findings are illustrated using a cigarette
demand model based on panel data of 46 U.S. states over the period
1963 to 1992.
to include explanatory variables observed on neighboring cross-
sectional units, is the simplest spatial econometric model producing
flexible spatial spillover effects and the easiest model to parameterize
the spatial weights matrix, denoted by W , describing the spatial
arrangement of the units in the sample. Nevertheless, it has received
relatively little attention in the theoretical and applied spatial
econometrics literature. This study fills this gap by considering several
extensions of its basic form. It is found that the claim made in many
empirical studies that their results are robust to the specification of W
is not sufficiently substantiated. Especially the spatial spillover effects,
often the main interest of spatial economic and econometric studies,
turn out to be sensitive to the specification of W. In addition, it is
found that the common practice to adopt the same W for every spatial
lag should be rejected. These findings are illustrated using a cigarette
demand model based on panel data of 46 U.S. states over the period
1963 to 1992.
Translated title of the contribution | The SLX model: Extensions and the sensitivity of spatial spillovers to W |
---|---|
Original language | Spanish |
Pages (from-to) | 34-50 |
Number of pages | 16 |
Journal | Papeles de Economía Española |
Volume | 152 |
Publication status | Published - 2017 |