THE PORTFOLIO FLOWS: The Interaction Between Flows and Returns 3

Pursuing this line of thinking one step further, we investigate the covariances of idiosyncratic (country specific) portion of flows from equation (1). This is the portion of flows that the Brennan and Cao local-information story emphasizes. Table 6 shows that country specific flows differ from regional factors in several important ways. First, the country-specific flows affect prices less than the regional factor does.

Estimated CVRs are smaller and less statistically significant than those reported in Tables 5. Second, with a few exceptions, country-specific flows seem unrelated to past returns. Third, idiosyncratic flows have only modest predictive power—positive or negative—for future returns. Overall, the results suggest that idiosyncratic flows behave according to our simplest null hypothesis: that flows are relatively uncorrelated with each other and with returns. This is directly at odds with the Brennan and Cao story, which implies a strong correlation between idiosyncratic flows and local returns.
Tables 7 and 8 are analogous to Tables 5 and 6, except that they focuses on excess currency (not equity) returns. Table 7 results are similar to Table 5 in that flows predict currency returns in emerging markets, but not in developed markets.

Vector autoregressions

While the covariance results tell us broadly about predictability, we can learn more about the structure of flows and returns from a vector autoregression. Specifically, we ask two questions: i) do returns predict flows over and above the predictions of lagged flows?; and ii) do flows predict returns over and above the predictions of lagged returns?

One way to address these questions is to consider a simple structural model of flows and returns. Our structural model assumes the following. First, the decision to buy more of a country’s equity depends on past inflows and past returns. Past inflows matter because they are correlated with the disparity between price and value, as perceived by investors. This assumes that there is information about future value in informed investors trades. Past returns enter because some investors are not informed and cannot observe inflows. These investors therefore rely on past returns as a proxy for information.

Second, the price set by market makers is a function of current and past inflows. Current inflows positively affect prices because current inflows may contain information about value. However, lagged inflows may also matter. With current inflows given, the larger are past inflows, the more prices have already risen. If the past price increase already impounded all of the information into prices, then past order flow should have no further impact on returns and the coefficient on lagged flows should be zero. payday loans reviews

However, order flow may increase prices temporarily due to transient price pressure. In such a case, prices rise with current inflows, but then decline when the inflow stops. Past inflows will therefore have a negative impact on current returns. Alternatively, if the inflow contains enough information, future prices may continue to rise (as others learn that information) even after the inflow has stopped. In this case, past inflows will have a positive impact on current returns.