Spatial Dependence in Asset Pricing

Abstract

We follow the seminal work of Paelinck (1978) who introduces spatial interdependence of, i.e. income, expenditure, investment, to classic Keynesian economic models, and estimate a spatial factor model. Asset prices may display characteristics of spatial dependence meaning spatial proximity can explain return comovements with neighboring firms. The spatial patterns are not associated with exposure to local factors but account for interdependence between returns based on individual unobserved characteristics transmitted spatially. Accounting for spatial return dependence would not change the classic factor model by introducing a new factor. Instead, it enables us to disentangle spatial spillover risk from market risk and idiosyncratic risk, which changes the risk allocation across those components. Our results show that returns display spatial patterns and the spatial volatility mostly matters arises during the Global Financial Crisis (GFC). The spatial spillover risk can contribute up to 28% to total asset risk. Our results imply that investors should consider spatially distributed local risks which otherwise will be hard to capture as firms do not have a single location and an exposure to a single region.

Keywords: Asset pricing, spatial dependence, spatial spillover risk.

JEL Classification: G12, C30