Outliers and Spatial Dependence in Cross-Sectional Regressions

Outliers are a risk factor in any econometric analysis. They are often observations that exert an excessive influence on the results and lower our confidence in the estimations. As a consequence, the attention given to their identification and treatment in the context of time series is not surprising. Our intention in the present paper is to advance in this same direction but now focusing the discussion on the impact of outliers on cross-sectional specifications. In particular, we will analyse the behaviour of the most habitual misspecification tests in this field in the presence of outliers. With this objective, we present a series of analytical results that try to delimit the effects suffered by these statistics and complete the study with a Monte Carlo exercise designed to measure their effect more precisely. According to our results, the impact of outliers can be very important, especially when several coincide in the sample.

[1]  W. J. Dixon,et al.  Analysis of Extreme Values , 1950 .

[2]  Donald Geman,et al.  Boundary Detection by Constrained Optimization , 1990, IEEE Trans. Pattern Anal. Mach. Intell..

[3]  Robert Haining,et al.  Data Problems in Spatial Econometric Modeling , 1995 .

[4]  Shashi Shekhar,et al.  Detecting graph-based spatial outliers , 2002, Intell. Data Anal..

[5]  D. J. Unwin,et al.  Control Point Distribution in Trend Surface Modelling Revisited: An Application of the Concept of Leverage , 1987 .

[6]  Ursula Gather,et al.  Concepts of Outlyingness for Various Data Structures , 2002 .

[7]  Donald Geman,et al.  Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images , 1984, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[8]  Robert Haining,et al.  Error Propagation Through Map Operations , 1993 .

[9]  Anil K. Bera,et al.  Simple diagnostic tests for spatial dependence , 1996 .

[10]  H. Rieder.  Robust asymptotic statistics , 1994 .

[11]  Peter J. Rousseeuw,et al.  Robust regression and outlier detection , 1987 .

[12]  D. Griffith.  What is spatial autocorrelation? Reflections on the past 25 years of spatial statistics , 1992 .

[13]  P. Burridge,et al.  On the Cliff‐Ord Test for Spatial Correlation , 1980 .

[14]  Daniel A. Griffith,et al.  Error Propagation Modelling in Raster GIS: Overlay Operations , 1998, Int. J. Geogr. Inf. Sci..

[15]  J. Davidson.  Stochastic Limit Theory , 1994 .

[16]  R. J. Martin.  Leverage, influence and residuals in regression models when observations are correlated , 1992 .

[17]  F. J. Anscombe,et al.  Rejection of Outliers , 1960 .

[18]  Luc Anselin,et al.  Small Sample Properties of Tests for Spatial Dependence in Regression Models: Some Further Results , 1995 .

[19]  Douglas M. Hawkins,et al.  Robust kriging—A proposal , 1984 .

[20]  Peter J. Huber,et al.  Robust Statistics , 2005, Wiley Series in Probability and Statistics.

[21]  S. Kuhnt,et al.  Outlier Identification Rules for Generalized Linear Models , 2003 .

[22]  P. Moran.  The Interpretation of Statistical Maps , 1948 .

[23]  Peter J. Rousseeuw,et al.  Robust Regression and Outlier Detection , 2005, Wiley Series in Probability and Statistics.

[24]  Thomas S. Ferguson,et al.  On the Rejection of Outliers , 1961 .

[25]  Calyampudi R. Rao,et al.  Linear Statistical Inference and Its Applications. , 1975 .

[26]  Jim Freeman,et al.  Outliers in Statistical Data (3rd edition) , 1995 .

[27]  Daniel A. Griffith,et al.  Spatial error propagation when computing linear combinations of spectral bands: The case of vegetation indices , 2003, Environmental and Ecological Statistics.

[28]  R. Haining.  DIAGNOSTICS FOR REGRESSION MODELING IN SPATIAL ECONOMETRICS , 1994 .

[29]  A. Madansky.  Identification of Outliers , 1988 .

[30]  Daniel A. Griffith,et al.  Error Propagation Modeling in Raster GIS: Adding and Ratioing Operations , 1999 .

[31]  Raymond J.G.M. Florax,et al.  The Performance of Diagnostic Tests for Spatial Dependence in Linear Regression Models: A Meta-Analysis of Simulation Studies , 2004 .