The Portfolio Flows of International Investors

The level of net flows into a country or region is influenced by past equity returns in that country or region.

Where are international investors putting their money on a daily basis? And, what's the relationship between those decisions and the returns in a country? In The Portfolio Flows of International Investors: Part I (NBER Working Paper 6687), Kenneth Froot, Paul O'Connell, and Mark Seasholes take advantage of a unique set of data from State Street Bank and Trust Company to answer those and related questions about international (cross-border) stock market investments. The daily records that they use start in mid-1994, continue through 1998, and contain more than 3 million trades from 46 countries.

This, or any, study of portfolio flows is complicated by the fact that a randomly chosen sample of flows should not be correlated with anything, because half of all flows are "buys", half are "sells", and a randomly selected trade may be either. Thus, meaningful results only can be obtained if the focus is narrowed to one particular group of investors. In this case, the authors look specifically at cross-border flows -- those originating outside a given country (for example, the investments of U.S. fund managers overseas.)

By first studying just the flows themselves, the authors find that daily net flows (that is, buys minus sells) are slightly correlated across countries but are more strongly correlated within regions. In fact, regional correlation in Latin America, East Asia, and other emerging markets increased during the recent Asian crisis. Both inflows and outflows show high levels of persistence. In other words, a large inflow today likely will be followed by more large inflows over the coming week. Net flows also are highly persistent.

The authors then ask whether investors follow stock market returns. They examine the interaction between cross-border flows and equity returns and confirm that the level of net flows into a country or region is influenced by past equity returns in that country or region. This is evidence of "positive feedback trading behavior" -- high returns in a country today predict high, positive flows in the future.

Finally the authors ask whether there is information in the flows that can help in predicting future stock market returns; that is, do cross-border flows actually predict future equity returns? In emerging markets, they find strong evidence that a positive shock (or sudden increase) in cross-border inflows predicts an increase in equity returns over the next 60 days. This may be surprising, because recent theoretical work suggests that foreign investors are at an informational disadvantage vis-a-vis local investors. Still, the authors find no such relationship between inflows and returns in developed markets.