Redrawing the Map of Global Capital Flows: The Role of Cross-Border Financing and Tax Havens, by Antonio Coppola, Matteo Maggiori, Jesse Schreger, and Brent Neiman:

We start with the dataset of global mutual fund and exchange traded fund (ETF) holdings provided by Morningstar and assembled in Matteo Maggiori, Brent Neiman and Jesse Schreger (2019a, henceforth MNS). For each position in the data, we link the security’s immediate issuer to its ultimate parent. The resulting data can then be used to create a mapping that transforms cross-border positions from a residency to nationality basis and that sheds light on how global firms finance themselves. …

First, in the case of bonds, positions are almost always reallocated away from Bermuda, the Cayman Islands, and other tax havens. Under nationality, these positions are often associated with developing countries like Brazil, China, India, and Russia, which may reflect the fact that developing countries find it easier to issue offshore than onshore, where the legal system and institutional quality may be of concern to foreign investors. Reallocating positions from tax havens to developed countries is also common, though, perhaps because tax havens allow them to access international investors with less onerous rules governing the withholding of taxes on interest payments. These patterns may also reflect tax-driven profit-shifting, whereby one unit of a company raises money at a low interest rate in a low-tax regime and loans it at a higher interest rate to an affiliated unit in a high-tax regime.

Second, in the case of equities, we find that many developed-country investments in tax havens are actually associated under nationality with China. Many of these positions are in securities issued through Variable Interest Entities (VIE), a structure designed to avoid China’s capital controls and the legality of which may rest on tenuous ground. Relatedly, we see a large share of equities reallocated by our algorithm away from Ireland and to developed countries, an adjustment reflecting the popularity of “tax inversions” there.

Third, in the case of asset-backed securities, for several investor countries, we find large reallocations toward the domicile of the investor, often because the underlying assets are found there. For example, our reallocation matrix records that 73.4 percent of U.S. investment in Cayman Islands’ asset-backed securities should instead be thought of as U.S. domestic investment, largely because those securities are backed by U.S. mortgages.