Global Financial Chains
Abstract:
The use of global financial hubs and tax havens to move capital internationally obfuscates true international investment positions. As a result, available data on these financial relationships is deeply flawed. Using information on bond ownership structures, Coppola et al, 2021 restate data on gross positions to account for tax haven usage. We provide and estimate a general model of global capital flows where capital can be sent indirectly from origin to destination through one or multiple third countries in a manner which reduces transaction costs. We use the model to obtain conditions under which the empirical method of Coppola et al, 2021 correctly recover the true underlying bilateral positions. We characterize the bias arising when these conditions are not met. We then relax these conditions and use the restated matrices to recover estimates of the true origin-destination bilateral positions, as well as a cost matrix that underpins these flows. Our findings confirm the role of tax havens in international financial transactions and augment the recharacterization of China's net positive position. Our method also uncovers a substantial intermediation role for non tax haven hub countries (Switzerland, Japan, Great Britain). The implicit cost matrix we recover exhibits a strong gravity relationship. Our estimation produces five data products: a restated matrix of global capital positions, a matrix of bilateral transfer costs, a matrix of bilateral network transfer costs, an enlarged version of the IMF Coordinated Portfolio Investment Survey (CPIS), and a version of the CPIS which excludes government bonds.