A road map to capture the spatial dependence underlying regions’ economic resilience

ABSTRACTRegions are embedded in complex webs of interactions that influence, among other things, their economic resilience. However, a general lack of attention is given to the spatial dependence underlying regions’ economic resilience. This paper advocates investigating such spatial interactions and, in doing so, provides a road map to guide researchers through the specification search. The road map is theoretically underpinned by the argument that regions’ resilience is influenced by local spillovers. The several articles in the resilience literature that incorporate spatial dependence are evaluated, and their shortcomings discussed. The paper provides an empirical illustration of the road map.KEYWORDS: resilience; regional economies; spatial dependence; specification searchJEL: R11R10R12View correction statement:Correction ACKNOWLEDGEMENTSWe gratefully acknowledge the valuable support of Godwin Arku in developing the ideas advanced in this paper. We thank two anonymous reviewers and the editor, J. Paul Elhorst, for their constructive comments and suggestions that significantly improved the final version of this paper.DISCLOSURE STATEMENTNo potential conflict of interest was reported by the authors.Correction StatementThis article was originally published with errors, which have now been corrected in the online version. Please see Correction (https://doi.org/10.1080/17421772.2023.2247933)Notes1 For an in-depth overview of the factors that influence region’s resilience, see Evenhuis (Citation2017).2 There are readily available spatial econometric approaches that can be incorporated (e.g., Elhorst, Citation2014; Fischer & Getis, Citation2010; Fischer & Nijkamp, Citation2014); however, the point is that these spatial methods have largely not been incorporated by scholars empirically examining regions’ resilience.3 The SARMA test is a Lagrange multiplier (LM) test developed by Anselin (Citation1988) to detect spatial autocorrelation (i.e., spatial dependence) in the response variable and error term based on OLS residuals.4 Similar spatial modelling approaches, including drawbacks, are present in Giannakis and Papadas (Citation2021).5 For the Social Explorer database, see https://www.socialexplorer.com/explore-maps6 The empirical illustration assumes all the classical assumptions of standard linear regressions are satisfied.7 Five-nearest neighbours indicate means that all regions (i.e., counties) are all influenced by their five nearest neighbours. No theoretical justification is used in this empirical illustration, as Texas counties are not the authors’ area of expertise, but in actually empirical spatial analysis spatial weights matrices should be theoretically grounded to varying degrees.8 Human capital is the share of the labour force aged 25 years old who have a post-secondary degree; diversification, using the diversity index (Brown & Greenbaum, Citation2017), is used to measure the industrial diversity of Texas counties.9 All spatial weights matrices are row standardised.10 The spatial models are estimated using maximum-likelihood estimation.