Space-Time Spatial Panel Model
Also known as: ST-SPM, spatiotemporal panel model, space-time panel econometrics, dynamic spatial panel model
The Space-Time Spatial Panel Model extends standard spatial panel econometrics to jointly account for cross-sectional spatial dependence, temporal autocorrelation, and unit-level heterogeneity. It allows outcomes in one location and time period to be influenced by outcomes in neighboring locations and by the location's own past, making it the canonical framework for dynamic spatiotemporal panel data analysis.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use the Space-Time Spatial Panel Model when your data cover multiple geographic units observed across multiple time periods and you have strong theoretical or empirical reason to expect both spatial spillovers and temporal persistence. It is the appropriate choice for regional economic convergence, disease spread, housing markets, and environmental outcomes where neighboring areas and past values matter simultaneously. Do not use it when you have purely cross-sectional data (no time dimension), purely time-series data (single unit), or when the spatial dependence is negligible (confirmed by Moran's I). Avoid it in small panels (few units or few periods) where the spatial lag parameter becomes poorly identified.
Strengths & limitations
- Simultaneously controls for spatial dependence, temporal autocorrelation, and unobserved unit heterogeneity.
- Produces direct, indirect, and total effects that distinguish own-unit impacts from spillover impacts.
- Can accommodate a wide range of spatial structures through the weights matrix specification.
- Generalizes most other spatial models (spatial lag, spatial error, spatial Durbin) as special cases.
- Widely supported in R (splm, SDPDmod) and Stata (xsmle), enabling replication and extension.
- Requires a correctly specified spatial weights matrix; misspecification biases all estimates.
- MLE estimation is computationally intensive for large N (many units) because it involves repeated log-determinants of the N×N matrix (I - rho*W).
- Dynamic spatial panel GMM estimators require careful choice of instruments and lag depth to avoid weak-instrument problems.
- Interpretation of spillover effects is non-trivial and requires computing partial derivatives rather than reading off regression coefficients directly.
Frequently asked
What is the difference between the spatial lag, spatial error, and spatial Durbin variants in a panel context?
In the spatial lag (SAR) panel, the dependent variable in neighboring units enters the right-hand side; in the spatial error (SEM) panel, spatial dependence affects only the error term and coefficient estimates remain unbiased if ignored but standard errors do not; in the spatial Durbin panel (SDM), both the dependent variable and the covariates of neighbors appear. SDM is the most general and nests the other two as special cases.
How do I choose the spatial weights matrix W?
Choose W based on the theoretical mechanism generating dependence. Contiguity (queen or rook) suits administrative regions sharing borders. Inverse-distance weights suit processes where influence decays with physical distance. K-nearest-neighbor weights are appropriate when units vary greatly in area. Always row-standardize W so each row sums to one. Sensitivity analysis across two or three plausible W matrices is strongly recommended.
Can I include a lagged dependent variable and a spatial lag at the same time?
Yes. The dynamic spatial panel model (tau ≠ 0, rho ≠ 0) is the full space-time specification. However, including both simultaneously requires GMM-style instruments (Arellano-Bond adapted for spatial settings) because the spatially lagged dependent variable and the temporally lagged dependent variable are both endogenous. MLE is only valid for the static (tau = 0) version.
How do I interpret a significant spatial lag coefficient rho?
A significant positive rho indicates positive spatial spillovers: a unit with high values is surrounded by units with high values, and vice versa. But rho alone is not enough for policy inference. Compute the matrix of partial derivatives to separate the direct effect (own-unit change) from the indirect effect (change propagated to neighbors and back), because the two can differ substantially in magnitude.
What software implements space-time spatial panel models?
In R, the splm package handles static spatial panels (ML and GM estimators) and the SDPDmod package adds dynamic spatial panel estimation. In Stata, the xsmle command covers both static and dynamic spatial panels. The spdep and spatialreg packages in R provide complementary diagnostics. Python's spreg module in PySAL is growing in capability for cross-sectional spatial models but panel support is more limited.
Sources
- Elhorst, J. P. (2014). Spatial Econometrics: From Cross-Sectional Data to Spatial Panels. Springer. ISBN: 978-3642403408
- LeSage, J., & Pace, R. K. (2009). Introduction to Spatial Econometrics. CRC Press. ISBN: 978-1420064247
How to cite this page
ScholarGate. (2026, June 3). Space-Time Spatial Panel Model. ScholarGate. https://scholargate.app/en/spatial-analysis/space-time-spatial-panel-model
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Geographically Weighted RegressionSpatial analysis↔ compare
- Space-Time Spatial Lag ModelSpatial analysis↔ compare
- Spatial Durbin ModelSpatial analysis↔ compare
- Spatial Error ModelSpatial analysis↔ compare
- Spatial Lag ModelSpatial analysis↔ compare