The Impact of Interaction Effects among Neighbouring Countries on Financial Liberalization and Reform: A Dynamic Spatial Panel Data Approach
2013; Routledge; Volume: 8; Issue: 3 Linguagem: Inglês
10.1080/17421772.2012.760136
ISSN1742-1780
AutoresJ. Paul Elhorst, Eelco Zandberg, Jakob de Haan,
Tópico(s)Housing Market and Economics
ResumoAbstract Abstract Using data from Abiad et al. (2008), we estimate a dynamic spatial panel data model with country and time-period fixed effects to test for spatial cointegration in the financial liberalization index. Parameter estimates are obtained by reformulating the model in spatial first-differences. Next, by considering the error correction model representation of the dynamic spatial panel data model, we examine financial reform interaction effects and the extent to which a change in a single explanatory variable in a particular country affects financial reform in other countries. Finally, by comparing the performance of different specifications of the spatial weights matrix describing the spatial arrangement of the countries in the sample, we show that the popular regional leader matrix must be rejected in favour of an inverse distance matrix with a cut-off point of 3,000 km. RÉSUMÉ En utilisant des données de Abiad et al. (2008), nous procédons à l'estimation d'un modèle de données spatiales dynamiques, avec effets fixes sur le plan des pays et des périodes, pour tester la co-intégration spatiale dans l'indice de libéralisation financière. On obtient des estimations paramétriques en reformulant le modèle dans des premiers éléments de différence spatiale. Ensuite, en examinant la représentation du modèle de rectification des erreurs du modèle de données spatiales dynamiques, nous examinons les effets d'une interaction de réformes financières, et la mesure dans laquelle un changement d'une variable explicative unique dans un certain pays affecte les réformes financières dans d'autres pays. Enfin, en comparant les performances de différentes spécifications de la matrice des pondérations spatiales décrivant la disposition spatiale des pays de l'échantillon, nous montrons qu'il est nécessaire de rejeter la matrice leader régionale populaire au profit d'une matrice de distance inverse avec un point limite de 3 000 km. EXTRACTO Utilizando datos procedentes de Abiad et al. (2008), estimamos un modelo de datos de panel dinámico espaciales y los efectos fijos de período de tiempo y país para comprobar la cointegración espacial en el índice de liberalización financiera. Las estimaciones de parámetros se obtienen reformulando el modelo en las primeras diferencias espaciales. A continuación, considerando la representación del modelo de corrección de errores del modelo de datos de panel dinámico espaciales, examinamos los efectos de la interacción de la reforma financiera y el punto hasta el que un cambio en una sola variable explicativa en un país particular afecta a la reforma financiera en otros países. Finalmente, comparando el rendimiento de diferentes especificaciones de la matriz de ponderaciones espaciales que describen la disposición espacial de los países de la muestra, mostramos que la matriz de líderes regionales populares debe ser rechazada a favor de una matriz de distancia inversa con un punto de corte de 3000km. 摘要:我们使用阿比亚德等人(2008)提供的数据估算带有国家及时间段恒定效应的动态空间面板数据模型 ' 以测试金融自由化指数中的区域整合程度 。通过以一阶差分对模型进行重建 。接下来 ' 通过分析动态空间面板数据模型的纠错模型表示 ' 我们考查了金融改革互动效应 ' 以及某一特定国家中单个解释变量的变化对其他国家金融改革影响的程度 。最后 ' 通过比较用来描述样本中国家和地区分布的不同规格的地区权重矩阵性能, 我们得出 ' 为了获得分界点为3,000km的逆向距离矩阵, 必须丢弃常用的地区领导人矩阵。 Keywords: Financial reformfinancial liberalizationspatial cointegrationerror correctionJEL CLASSIFICATION: C21G3 Acknowledgments The views expressed in this paper do not necessarily reflect those of De Nederlandsche Bank. The authors thank two anonymous referees for their very helpful comments on a previous version of the paper. Notes 1. Measured as the coefficient of this variable times the probability that FL changes. 2. See Anselin (2006 Anselin, L. 2006. "Spatial econometrics". In Palgrave Handbook of Econometrics, Edited by: Mills, T. and Patterson, K. Vol. 1, 901–969. Basingstoke: Palgrave. [Google Scholar]), LeSage & Pace (2009 LeSage, J. P. and Pace, R. K. 2009. Introduction to Spatial Econometrics, Boca Raton, FL: CRC Press/Taylor & Francis Group. [Crossref] , [Google Scholar]) and Kelejian & Prucha (2010 Kelejian, H. H. and Prucha, I. R. 2010. Specification and estimation of spatial autoregressive models with autoregressive and heteroskedastic disturbances. Journal of Econometrics, 157: 53–67. doi:10.1016/j.jeconom.2009.10.025[Crossref], [PubMed], [Web of Science ®] , [Google Scholar]) for recent overviews of the spatial econometrics literature for cross-section data, and Anselin et al. (2008 Abiad , A. , Detragiache , E. & Tressel , T. 2008 A New Database of Financial Reforms , IMF Working Paper No. 08/266 . [Google Scholar]), Lee & Yu (2010 Lee, L.-F. and Yu, J. 2010. Some recent developments in spatial panel data models. Regional Science and Urban Economics, 40: 255–271. doi:10.1016/j.regsciurbeco.2009.09.002[Crossref], [Web of Science ®] , [Google Scholar]) and Elhorst (2010a Elhorst, J. P. 2010a. "Spatial panel data models". In Handbook of Applied Spatial Analysis, Edited by: Fischer, M. and Getis, A. 377–407. Heidelberg: Springer. [Crossref] , [Google Scholar]) for spatial panel data. 3. This means that we abstract from the problem that the liberalization index is bounded from below and above. 4. For this purpose we use a Matlab routine that has kindly been made available by Jihai Yu. 5. If W has more than just one eigenvalue that is equal to 1, say p, the number N−1 must be adjusted to N−p. 6. Since the numerical magnitudes of these two calculations of the indirect effect are the same, it does not matter which one is used (LeSage & Pace, 2009 LeSage, J. P. and Pace, R. K. 2009. Introduction to Spatial Econometrics, Boca Raton, FL: CRC Press/Taylor & Francis Group. [Crossref] , [Google Scholar]). 7. The dataset is an updated version of the data used by Abiad & Mody (2005 Abiad, A. and Mody, A. 2005. Financial reform: what shakes it? What shapes it?. American Economic Review, 95: 66–88. doi:10.1257/0002828053828699[Crossref], [Web of Science ®] , [Google Scholar]). There are alternative measures of financial liberalization available, like those of Williamson & Mahar (1998 Williamson , J. & Mahar , M. 1998 A Survey of Financial Liberalization , Princeton Essays in International Finance No. 211 . [Google Scholar]), Bandiera et al. (2000 Bandiera, O., Caprio, G., Honohan, P. and Schiantarelli, F. 2000. Does financial reform raise or reduce saving?. Review of Economics and Statistics, 82: 239–263. doi:10.1162/003465300558768[Crossref], [Web of Science ®] , [Google Scholar]), Laeven & Claessens (2003 Laeven, L. and Claessens, S. 2003. Financial development, property rights and growth. The Journal of Finance, 58: 2401–2436. doi:10.1046/j.1540-6261.2003.00610.x[Crossref], [Web of Science ®] , [Google Scholar]) and Kaminsky & Schmukler (2008 Kaminsky, L. G. and Schmukler, S. 2008. Short-run pain, long-run gain: financial liberalization and stock market cycles. Review of Finance, 12: 253–292. doi:10.1093/rof/rfn002[Crossref], [Web of Science ®] , [Google Scholar]). However, the coverage of these alternatives is more restrictive and therefore we consider the dataset of Abiad et al. (2008 Abiad , A. , Detragiache , E. & Tressel , T. 2008 A New Database of Financial Reforms , IMF Working Paper No. 08/266 . [Google Scholar]) as the most comprehensive source of information on financial liberalization at this moment. 8. Data before 1976 are incomplete, while data for 1975 are used to cover the (spatially lagged) dependent variable lagged in time. 9. Except for USINT, the US Treasury Bill rate taken from IFS of the IMF, because this variable would be perfectly collinear with the time-period fixed effects. 10. Zandberg et al. (2012 Zandberg , E. , De Haan , J. & Elhorst , J. P. 2012 The political economy of financial reform: how robust are Huang's findings? , Journal of Applied Econometrics , 27 , 695 – 699 .[Crossref], [Web of Science ®] , [Google Scholar]) show that some of Huang's conclusions, notably those referring to the role of democracy, are not robust if the expanded dataset of Abiad et al. (2008 Abiad , A. , Detragiache , E. & Tressel , T. 2008 A New Database of Financial Reforms , IMF Working Paper No. 08/266 . [Google Scholar]) is used instead of the dataset of Abiad & Mody (2005 Abiad, A. and Mody, A. 2005. Financial reform: what shakes it? What shapes it?. American Economic Review, 95: 66–88. doi:10.1257/0002828053828699[Crossref], [Web of Science ®] , [Google Scholar]). See also Burgoon et al. (2012 Burgoon, B. M., Demetriades, P. and Underhill, G. R. D. 2012. Legitimacy and the political sources of financial liberalization. European Journal of Political Economy, 28: 147–161. doi:10.1016/j.ejpoleco.2011.10.003[Crossref], [Web of Science ®] , [Google Scholar]) for a further analysis of the role of political factors in financial liberalization. 11. The estimators can be modified for a spatial weights matrix that changes over time; their asymptotic properties have been investigated by Lee & Yu (2012 Lee, L.-F. and Yu, J. 2012. QML estimation of spatial dynamic panel data models with time varying spatial weights matrices. Spatial Economic Analysis, 7: 31–74. doi:10.1080/17421772.2011.647057[Taylor & Francis Online], [Web of Science ®] , [Google Scholar]). However, the problem that this matrix is not truly exogenous remains (see below). 12. As we pointed out before, ignoring time-period fixed effects, when present, may lead to an upward bias up to 0.45 (see Lee & Yu, 2010 Lee, L.-F. and Yu, J. 2010. Some recent developments in spatial panel data models. Regional Science and Urban Economics, 40: 255–271. doi:10.1016/j.regsciurbeco.2009.09.002[Crossref], [Web of Science ®] , [Google Scholar]). The difference between Abiad & Mody's coefficient estimate of 0.44–0.55 and our estimate of 0.151 falls within this range. In his replication study, Huang (2009 Huang, Y. 2009. The political economy of financial reform: are Abiad and Mody right?. Journal of Applied Econometrics, 24: 1207–1213. doi:10.1002/jae.1093[Crossref], [Web of Science ®] , [Google Scholar]) also adds time-period fixed effects. In addition, he employs Pesaran's (2006 Pesaran, M. H. 2006. Estimation and inference in large heterogeneous panels with a multifactor error structure. Econometrica, 74: 967–1012. doi:10.1111/j.1468-0262.2006.00692.x[Crossref], [Web of Science ®] , [Google Scholar]) common correlated effect pooled (CCEP) approach to allow for the possibility of error dependence across countries. Remarkably, he finds that the coefficient estimate of the diffusion effect falls to a negative and significant number of –0.14. This would mean that countries turn away rather than follow the regional leader, a conclusion that must be rejected in view of our estimation results. 13. Feedback effect=direct effect−coefficient estimate=−1.212−(−1.207)=−0.005. 14. In order to draw inferences regarding the statistical significance of these effects, we used the variation of 1,000 simulated parameter combinations drawn from the multivariate normal distribution implied by the maximum likelihood estimates (see LeSage & Pace (2009 LeSage, J. P. and Pace, R. K. 2009. Introduction to Spatial Econometrics, Boca Raton, FL: CRC Press/Taylor & Francis Group. [Crossref] , [Google Scholar]) for mathematical details). 15. See http://www.cepii.fr/anglaisgraph/cepii/cepii.htm. 16. For values greater than or equal to 3,000, every country in the sample appeared to have at least one neighbour. To avoid 'islands' for values that are smaller than 3,000, we also assumed that every country at least interacts with its nearest neighbour. 17. Estimation results for the dynamic spatial panel data model formulated in levels (with or without time dummies) are left aside here, because this model again turned out to be non-stable.
Referência(s)