Abstract
We develop a structural multivariate spatial regression model allowing us to incorporate both inter- and intralocation effects among different variables. The existing multivariate spatial regression approaches are not able to simultaneously account for these effects. The currently available models either ignore the intralocation effect between the variables, which may result to bias in estimation of the other effect, or estimate a reduced form without separating the two effects.We employ this model to investigate the comovements of international stock and bond markets using geographic neighborhood and bilateral trade between countries to define countries’ proximity to each other. Our results show that eliminating the within-country effect between stock and bond returns may lead to estimation bias and misrepresentation of cross-country feedback effects of these variables. We find a strong spatial dependence between stock returns, particularly for countries that have large trades with each other. The spatial dependence between countries’ bond returns is also highly significant but the magnitude of the effect is smaller than that of stock returns. We show that this correlation is mainly due to the global comovement of the bond markets rather than the interdependence of the countries through the proximities employed in this paper. Moreover, we find a positive within-country and a negative cross-country dependence between stock and bond returns. (Less)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have