On the geography of international banking : a case for spatial econometrics?

The aim of this study is to examine the determinants of international banking. In doing so, it utilizes data from the Bank of International Settlements (BIS). In particular, the primal variable of interest is drawn from BIS Consolidated Statistics and refers to banks' foreign claims. is defined as the sum of "international claims" (cross border claims & local claims of foreign affiliates in foreign currencies) plus "local claims in local currency" of bank foreign affiliates (branches and subsidiaries). This sum is called "foreign claims" and inter-office positions are netted out. σ review of the literature on international banking and finance would suggest that international banking and finance on the one hand and trade and FDI on the other are all intertwined. The second most striking conclusion is, however, that all these activities are geographically confined as far as the significant role of distance makes it clear. Banks in particular do not seem to extend too far from their home markets. Geography seems to be important and might be even more significant than only the effect of distance would suggest. If spatial dependency is present in banks' foreign claims, and there are strong indications in the literature that this might be the case, accounting for it would reveal possible indirect channels in which borrowing from and lending to foreign countries through international banking might be affecting the risks that the banking system of a country is exposed to. The keyword here is "indirect effects" or financial spillovers that might be operating in a fashion that subsequently connects countries beyond those immediately involved in borrowing and lending relationships with each other. The geographical aspects of international banking will be explored by the means of global and local statistics of spatial dependency in an initial exploratory data analysis level. While pursuing the analysis at an econometric modeling level, priority will be given to models containing spatial lags (although other alternatives may be considered) as these naturally lend themselves to the exploration of possible spatial financial spillovers and regional feedback mechanisms.