Spatial GARCH Models for Unknown Spatial Locations - An Application to Financial Stock Returns

Abstract

Finding a suitable weight matrix in spatial GARCH models is a challenge when the actual locations are not known. Thus, we introduce an estimation procedure for spatial GARCH models when the locations are unknown. We suggest to use balance sheet data of companies as proxy for the spatial distance between companies. We provide a simulation study underpinning the finite sample estimation performance of the method. As an empirical illustration, we use one year stock returns between 2019 and 2020 of the companies listed in the Dow Jones Global Titans 50 index. We find, that research and development expenses are most relevant for determining similar firms with regard to stock return volatility.

Keywords: Spatial GARCH, spatial weight matrix, stock returns, balance sheet, funancial risk spillover, unknown locations