VoxEU.org | June 18, 2018
Currency is a far more important factor for cross-border capital flows than is typically assumed. This column demonstrates that investors exhibit a strong bias toward securities denominated in their home currency even when investing in bonds issued by developed countries, implying that the majority of foreign firms that do not issue foreign currency bonds typically do not borrow from abroad directly. A key exception to this pattern is the US, as the global taste for dollars enables smaller firms that issue only dollar-denominated bonds to access foreign capital. The benefit that the dollar’s status offers the US appears to have increased since the Global Crisis at the expense of the euro.
What determines which countries and firms receive capital from foreigners and benefit from capital account liberalisation? Are cross-border capital flows allocated similarly to domestic investment? A large literature has studied related questions in the context of equity portfolios. For example, home-country bias – the finding that investors disproportionately hold securities issued by domestic firms – constitutes one of the most robust patterns in global portfolios (French and Poterba 1991). But data limitations have left international bond portfolios largely unexplored, even as international bond investment has grown rapidly. We obtained from Morningstar the security-level details on nearly $27 trillion in positions held by mutual funds around the world, and bring these micro data to bear on these and related classic questions in international finance and macro. We find that the currency in which bonds are denominated plays a critical role in determining which firms get foreign capital and in shaping the global distribution of cross-border flows.
In recent work, we demonstrate that investors exhibit a bias toward corporate bonds denominated in their own currencies, even when borrowing from developed countries with a strong rule of law, low inflation, and fully liberalised capital accounts (Maggiori et al. 2018). This home-currency bias is so strong that the currency of denomination of the assets has more predictive power in our data for who holds which bonds than does the nationality of the issuer, the focus of a voluminous literature on home-country bias.1 In much the same way that the finding of home-country bias led to its use as a critical calibration target for models in international macro and finance, we believe that models aiming to capture the external wealth transfers associated with currency movements and other key dynamics in open-economy frameworks, must also aim to generate the critical sorting by currency that we document.