Testing Panel Data Regression Models with Spatial Error Correlation

Department of Economics, Texas A&M University, College Station, TX 77843-4228, USA
Journal of Econometrics (Impact Factor: 1.53). 02/2003; 117(1):123-150. DOI: 10.1016/S0304-4076(03)00120-9
Source: RePEc

ABSTRACT This paper derives several lagrange multiplier (LM) tests for the panel data regression model with spatial error correlation. These tests draw upon two strands of earlier work. The first is the LM tests for the spatial error correlation model discussed in Anselin (Spatial Econometrics: Methods and Models, Kluwer Academic Publishers, Dordrecht; Rao's score test in spatial econometrics, J. Statist. Plann. Inference 97 (2001) 113) and Anselin et al. (Regional Sci. Urban Econom. 26 (1996) 77), and the second is the LM tests for the error component panel data model discussed in Breusch and Pagan (Rev. Econom. Stud. 47(1980) 239) and Baltagi et al. (J. Econometrics 54 (1992) 95). The idea is to allow for both spatial error correlation as well as random region effects in the panel data regression model and to test for their joint significance. Additionally, this paper derives conditional LM tests, which test for random regional effects given the presence of spatial error correlation. Also, spatial error correlation given the presence of random regional effects. These conditional LM tests are an alternative to the one-directional LM tests that test for random regional effects ignoring the presence of spatial error correlation or the one-directional LM tests for spatial error correlation ignoring the presence of random regional effects. We argue that these joint and conditional LM tests guard against possible misspecification. Extensive Monte Carlo experiments are conducted to study the performance of these LM tests as well as the corresponding likelihood ratio tests.

Download full-text


Available from: Badi H. Baltagi, Aug 30, 2015
  • Source
    • "See, e.g. Lee and Yu (2010a). Baltagi et al. (2003) considers a static spatial panel model where the error term is a SAR model. Xu and Lee (2010) shows that the maximum likelihood estimator is inconsistent when heteroskedastisity exists in the error term for static panel data models and proposes an alternative GMM estimation method. "
    [Show abstract] [Hide abstract]
    ABSTRACT: We consider a class of spatio-temporal models which extend popular econometric spatial autoregressive panel data models by allowing the scalar coefficients for each location (or panel) different from each other. To overcome the innate endogeneity, we propose a generalized Yule-Walker estimation method which applies the least squares estimation to a Yule-Walker equation. The asymptotic theory is developed under the setting that both the sample size and the number of locations (or panels) tend to infinity under a general setting for stationary and {\alpha}-mixing processes, which includes spatial autoregressive panel data models driven by i.i.d. innovations as special cases. The proposed methods are illustrated using both simulated and real data.
  • Source
    • "International Business Review (2014), (Baltagi et al., 2003; Blonigen et al., 2007; Hall & Petroulas, 2008; Ledyaeva, 2009) or a dummy variable to capture a common border (Coughlin & Segev, 2000). But institutional theory suggests that foreign firms may well consider near and distant in more nuanced terms, and that a variety of economic and political factors may also impact upon firms' perceptions of the distances between alternative locations – particularly in an emerging economy context. "
    [Show abstract] [Hide abstract]
    ABSTRACT: We investigate how different conceptions of distance impact upon one of the fundamental decisions made by foreign investors, the choice of foreign direct investment (FDI) location within the selected host country. We argue that the attractiveness of host country locations to foreign investors depends not only upon location-specific attributes such as labor costs, but also upon the location's proximity to alternative locations. We provide theoretical rationales for how and why alternative concepts of distance might impact upon firms’ FDI location decisions, and explicitly model different measures of geographic, economic and administrative distance. Empirically we illustrate the use of a number of spatial regression models with a new dataset on FDI in Chinese prefecture-cities, and have shown, in this context, that geographic distance is not the ‘best’ measure of distance to use. We find clear evidence of spatial dependence between the cities based upon economic distance, with weaker evidence related to administrative distance. The distinctive contribution of this paper is to emphasize that city-level policy to attract FDI is more likely to succeed if the prefecture-city is economically (and administratively) close to alternative city locations, while any policy expenditure may fail to attract FDI inflows if the prefecture-city is distant from other city locations.
    International Business Review 08/2014; DOI:10.1016/j.ibusrev.2013.12.002 · 1.51 Impact Factor
  • Source
    • "Those serial correlation tests are composed of the residuals or R-square. In advance, Lagrange multiplier test (LM test) with combination of R-square is applied in hypothesis and testing for dependence both within and between the cross-sectional units (McCoskey and Kao, 1998, Westerlund and Edgerton, 2007), heteroskedasticity test in panel data (Baltagi et al., 2003, 2007, Debarsy and Ertur, 2010, Furno, 2000, McAleer and Medeiros, 2008, Tse, 2002), bootstrap tests for serial correlation (Godfrey and Tremayne, 2005, and Godfrey, 2007), for example, Godfrey (2007) implemented the F-statistic from LM test with three bootstrapping tests in dynamic regression models of the higher-order autocorrelation models with nonnormal disturbances and proposed that the F-statistic can improve control of finite sample significance levels have been examined. However, LM test is suitable for the large number cases, but no researchers can define how many samples can be viewed as large number. "
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper provides a basic investigation of an R-square and sum-square error (SSE) in a linear regression model with the errors following a first-order autoregressive process in which the autocorrelation coefficients are non-zero. The consideration and measurement of the model are difficult to control, thus a computer stimulation is necessary to corroborate how the R-square and SSE are affected by the autocorrelation coefficients. The evidence reveals that the R-square and SSE differ in the ranges of positive and negative autocorrelation coefficients. The results show that it would require one to verify the estimators including the R-square or SSE for testing non-zero autocorrelation coefficients.
Show more