JEL classification: C21 C31 C33 C36Keywords: Spatial autoregression Increasingly many parameters Central limit theorem Rate of convergence Spatial panel data a b s t r a c t This paper develops consistency and asymptotic normality of parameter estimates for a higher-order spatial autoregressive model whose order, and number of regressors, are allowed to approach infinity slowly with sample size. Both least squares and instrumental variables estimates are examined, and the permissible rate of growth of the dimension of the parameter space relative to sample size is studied. Besides allowing the number of parameters to increase with the data, this has the advantage of accommodating some asymptotic regimes that are suggested by certain spatial settings, several of which are discussed. A small empirical example is also included, and a Monte Carlo study analyses various implications of the theory in finite samples.

Pseudo maximum likelihood estimates are developed for higher-order spatial autoregressive models with increasingly many parameters, including models with spatial lags in the dependent variables and regression models with spatial autoregressive disturbances. We consider models with and without a linear or nonlinear regression component. Sufficient conditions for consistency and asymptotic normality are provided, the results varying according to whether the number of neighbours of a particular unit diverges or is bounded.Monte Carlo experiments examine finite-sample behaviour.JEL classifications: C21, C31, C36

Tests are developed for inference on a parameter vector whose dimension grows slowly with sample size. The statistics are based on the Lagrange Multiplier, Wald and (pseudo) Likelihood Ratio principles, admit standard normal asymptotic distributions under the null and are straightforward to compute. They are shown to be consistent and possessing non-trivial power against local alternatives. The settings considered include multiple linear regression, panel data models with fixed effects and spatial autoregressions. When a nonparametric regression function is estimated by series, we use our statistics to propose specification tests, and in semiparametric adaptive estimation we provide a test for correct error distribution specification.These tests are nonparametric but handled in practice with parametric techniques.A Monte Carlo study suggests that our tests perform well in finite samples. Two empirical examples use them to test for correct shape of an electricity distribution cost function and linearity and equality of Engel curves.JEL classifications : C12, C14, C31, C33

Autoregressive spectral density estimation for stationary random fields on a regular spatial lattice has many advantages relative to kernel based methods. It provides a guaranteed positive-definite estimate even when suitable edge-effect correction is employed, is simple to compute using least squares and necessitates no choice of kernel. We truncate a true half-plane infinite autoregressive representation to estimate the spectral density. The truncation length is allowed to diverge in all dimensions in order to avoid the potential bias which would accrue due to truncation at a fixed lag-length. Consistency and strong consistency of the proposed estimator, both uniform in frequencies, are established. Under suitable conditions the asymptotic distribution of the estimate is shown to be zero-mean normal and independent at fixed distinct frequencies, mirroring the behaviour for time series. A small Monte Carlo experiment examines finite sample performance. We illustrate the technique by applying it to Los Angeles house price data and a novel analysis of voter turnout data in a US presidential election. Technically the key to the results is the covariance structure of stationary random fields defined on regularly spaced lattices. We study this in detail and show the covariance matrix to satisfy a generalization of the Toeplitz property familiar from time series analysis. JEL classifications : C14, C18, C21

We examine a higher-order spatial autoregressive model with stochastic, but exogenous, spatial weight matrices. Allowing a general spatial linear process form for the disturbances that permits many common types of error specifications as well as potential ‘long memory’, we provide sufficient conditions for consistency and asymptotic normality of instrumental variables, ordinary least squares, and pseudo maximum likelihood estimates. The implications of popular weight matrix normalizations and structures for our theoretical conditions are discussed. A set of Monte Carlo simulations examines the behaviour of the estimates in a variety of situations. Our results are especially pertinent in situations where spatial weights are functions of stochastic economic variables, and this type of setting is also studied in our simulations.