MSCA TRAINING - Data Science @BBU | Digital Finance MSCA - Anil BERA Seminar

The 4EC doctoral training course "Foundations of Data Science" is tailored specifically for doctoral students in finance, providing a comprehensive
and practical foundation in data science as applied to financial analysis and decision-making.
This course equips participants with the essential skills and knowledge required to effectively extract, transform, analyze, and model financial data using Python.
The starting point of the training school will be on the 20th of March, at 14:00 Romania time, 13:00 CET, with a scientific seminar of WP1 joint with Cluj Economics and Business Seminar Series (CEBSS) (https://econ.ubbcluj.ro/cebss.php) and WP5. We have the pleasure to have as a guest speaker Prof. Anil BERA (Jarque-Bera normality test) (Professor of Economics, College of LAS; Affiliated Professor of Statistics, College of LAS; Adjunct Professor of Finance, College of Business Adjunct Professor of Agricultural and Consumer Economics, College of ACES, University of Illinois).
This scientific seminar is compulsory for the PhD students, as part of the Foundations in Data Science training.
Please notice that is an hybrid event. The seminar will be held in Room 118 of the Faculty of Economics at Babes-Bolyai Univerisity.

The details of the presentation are to be found below:

Title:
The Last Word on Testing General Nesting Spatial (GNS) Model

Abstract:
Three decades ago, Anselin et al. (1996) was unceremoniously published in the Regional Science and Urban Economics (RSUE) that literally changed the landscape of testing spatial regression models. By our last count, it has been cited by more than 2,600 papers in applied spatial econometrics out of which more than 1,200 papers have explicitly used its testing tools as the principal model selection methodology. The crux of the paper was to identify the source of spatial dependence: lag and/or error. Since then, much water has flown through the space of spatial econometrics. More recently another source of dependence, namely the Durbin specification that takes account of the exogenous interaction effects among the cross-sectional units (locations) has become very popular in the econometrics literature. Elhorst (2017) pointed out that conducting statistical tests while ignoring the Durbin term may spuriously lead to strong evidence in favor of interaction effects among the dependent variable or among the error terms. He further cautioned that the tests in Anselin et al. (1996) or the analogous ones developed for the spatial panel set up in Elhorst (2010) may not be very helpful in finding the right model as they do not account for the exogenous interaction effects. In this paper, we modify Anselin et al. (1996) by including the Durbin specification and attempt to identify the three main sources of spatial dependence, namely, lag, error and Durbin. This work can also be viewed as an extension of Koley and Bera (2024) that took account of lag and Durbin terms but did not consider the spatial error dependence. We conduct extensive simulation experiments to investigate the finite sample behavior of our suggested tests. We also provide an empirical analysis using the City of Chicago housing prices data, to illustrate the usefulness of our procedures in practical applications.

This event is organized as part of the MSCA Industrial Doctoral Network Digital Finance. The course is open to participants beyond the network, as long as capacity allows. In case of insufficient capacity, doctoral candidates from the network will have priority. Please note that our online stream is available to all.