How Is This Research Useful to Practitioners?
When risk-off sentiment is pervasive in the stock market, investors attempt to reduce their small-cap stock holdings and increase (or reduce to a lesser extent) their large-cap stock positions. To measure this “flight-to-quality” pressure in the stock market, the authors compute the order flow differential (OFD), which is the difference between the value of buyer- and seller-initiated trades over a quarter, standardized by the aggregate dollar volume for large-cap stocks minus the similarly computed buy-versus-sell trade value difference for small-cap stocks.
The OFD is found to increase around business-cycle peaks, when macroeconomic conditions are likely to deteriorate, and to decrease around business-cycle troughs, when macroeconomic fundamentals are likely to improve. Moreover, OFD seems to exert significant predictive power over the most covered macroeconomic indicators—for example, the real GDP growth rate, the industrial production growth rate, the change in real corporate earnings and dividends, the change in short-term interest rates, and the change in the term structure spread (used as a proxy for the rate curve’s steepening or widening).
The authors’ results indicate that an increase in OFD forecasts a slowdown in output growth, a decline in the short-term interest rate, and a steepening of the slope of the yield curve. These findings are consistent with the authors’ hypothesis that because of countercyclical investment decision making throughout the business cycle, investors tend to rebalance their portfolios in favor of large stocks in advance of economic slowdowns and in favor of small stocks in advance of economic booms.
These results have implications for asset pricing in terms of explaining the cross section of stock returns. Stocks, which are consistently selected for investment when uncertainty and risk aversion are prevalent, are expected to carry a negative risk premium over the volatility of OFD. The reason is that for such “safe haven” stocks, investors would agree to pay higher prices, thus requiring a lower return, especially around business-cycle peaks.
Investment practitioners (especially chief economists/strategists) who forecast macrofundamentals to determine asset allocations can enhance their prediction models by incorporating the OFD variable into the set of parameters used to pinpoint the exact location of the economy across the business cycle.
How Did the Authors Conduct This Research?
The authors perform several tests of the OFD’s significant predictive power in the stock market with respect to future macrofundamentals and the performance of production factors, driven by the hypothesized countercyclical behavior of investment decision makers. In the first test, the authors run multiple regressions of macroeconomic indicators on the OFD of intraday trades executed with NYSE stocks over 1993–2006, after controlling for liquidity effects, current business conditions, and such systematic risk factors as size, value, momentum, and short-term reversal.
To account for possible feedback effects, the authors perform bivariate Granger causality tests via vector autoregressions with four lags on quarterly data for variables in the initial multiple regressions. They find that OFD significantly passes (1) the one-way causality test for the economic growth variables and the short-term interest rate and (2) the two-way causality test for the term structure spread. To check for robustness, the authors make out-of-sample predictions of macrofundamentals by estimating rolling regressions of quarterly macroindicators on lags of quarterly OFD levels, resulting in further evidence in support of OFD’s out-of-sample forecasting ability.
Finally, to perform asset-pricing tests of the explanatory power of the OFD risk loading at the cross section of returns, the authors apply the Fama and MacBeth (Journal of Political Economy 1973) methodology whereby estimations include OFD among other regressors — namely, six other market risk premium factors and four business-cycle variables.
The flow of investors’ money from one investment segment to another, courtesy of changes in marketwide risk aversion, constitutes a market dynamic that usually imposes itself, either positively or negatively, on the market valuation of different securities. Both academic research and practice have historically focused on identifying the driving factors behind the movement of liquid funds, with a view toward predicting price changes in risk-on and risk-off stocks.
But the authors’ analysis views the order flows between risky small stocks and safer large stocks from a different angle, attempting to verify the usefulness of examining order flows in accurately predicting an economy-wide business cycle. One assumption implicit in their analysis is that the majority of investors can predict with relative precision the future state of the economy and preventively rebalance their portfolios accordingly, protecting their portfolios’ performance against economic slowdowns while benefiting from anticipated expansions.
Although theory relates smart money to market efficiency and the correction of market anomalies, empirical evidence of financial bubbles and market crashes indicates that those who represent smart money, more often than not, do not consider the upswings or downturns of the business cycle and simply fail to recognize whether the stock market is overbought or oversold.
Another limitation of this research is that the authors postulate that risky stocks are sufficiently distinguished from safer stocks on the basis of their market capitalization only. This approach is a simplification that should be enhanced with such other dimensions of risk assessment as sector of activity, corporate earnings variability, and stock return volatility.