Correlated Order Flow: Pervasiveness, Sources, and Pricing Effects

Jarrad Harford, Aditya Kaul
2003 Social Science Research Network  
We hypothesize that a combination of indexing, industry and broader market forces create common effects in order flow and returns. We test the relative contribution of each to common effects in large samples of both index and non-index stocks. Common effects are strong in index constituent stocks, but are economically inconsequential in non-index stocks. Once indexing effects have been removed, the common effects in index constituents disappear. Industry and broader market effects exist, but
more » ... fects exist, but contribute relatively little to common effects. Aside from identifying economic causes for the statistical common factors extracted in prior work, these results indicate that common effects are not pervasive. Finally, we show that common effects in order flow have real economic impact by linking them to common effects in returns. This effect of order flow on the correlation structure of returns has implications for diversification strategies. The results also show that specialists adjust prices differentially in the presence of aggregate order flow, supporting multi-asset models of price formation and suggesting that traders with discretion can minimize the price impact of their trades by timing these to coincide with index trading. Abstract We hypothesize that a combination of indexing, industry and broader market forces create common effects in order flow and returns. We test the relative contribution of each to common effects in large samples of both index and non-index stocks. Common effects are strong in index constituent stocks, but are economically inconsequential in non-index stocks. Once indexing effects have been removed, the common effects in index constituents disappear. Industry and broader market effects exist, but contribute relatively little to common effects. Aside from identifying economic causes for the statistical common factors extracted in prior work, these results indicate that common effects are not pervasive. Finally, we show that common effects in order flow have real economic impact by linking them to common effects in returns. This effect of order flow on the correlation structure of returns has implications for diversification strategies. The results also show that specialists adjust prices differentially in the presence of aggregate order flow, supporting multi-asset models of price formation and suggesting that traders with discretion can minimize the price impact of their trades by timing these to coincide with index trading. (2001) use factor analysis to reveal comovement in intraday order flow, returns and trading costs across the 30 Dow Jones constituents, and canonical correlations to establish that the common factors are related. 1 Chordia, Roll and Subrahmanyam (2000) document the existence of market and industry factors in average daily changes in five measures of liquidity across a broad sample of NYSE stocks (see, also, Huberman and Halka, 2001). 2 Common effects in longerhorizon trading volume are identified by Tkac (1999), via market model type regressions, and Lo and Wang (2000) , via factor analysis. Recent research has uncovered common effects in trading activity, returns and trading costs for individual stocks. Hasbrouck and Seppi Our study extends this literature by exploring the following unanswered questions about common effects or commonality (following Chordia, Roll and Subrahmanyam) in order flow and returns. First, how pervasive are common effects in order flow and returns, and what are their sources? Because the extant evidence on commonality is based on samples dominated by stocks in major indexes, we hypothesize that trading by index funds may be the major source of commonality. This implies that samples of nonindex stocks will exhibit little in the way of common effects. Second, are commonality and its effects getting stronger through time? If indexing is at the root of common effects, then these effects should be growing as indexing grows. Finally, is commonality in returns actually produced by commonality in order flow? Unless commonality in order flow affects returns, it is of little economic importance. Our approach is to hypothesize a hierarchy of non-mutually exclusive sources of common effects, comprising industry, index and broader-market factors. Using intraday data, we study common effects for stocks in the S&P 500 index and a large sample of non-S&P stocks for both 1986 and 1996. In each year, we examine both the end of the day (when index funds are most active) and the rest of the day. 1 Order flow is the difference between buy and sell volume or number of transactions. Because it indicates the net direction of trades, it is sometimes referred to as net order flow. 2 The existence of common factors in lower frequency (daily, weekly or monthly) returns and even volume has been known for as long as 30 years, as evident in the use of the market model or its variants.
doi:10.2139/ssrn.407708 fatcat:shu7hiim35ay3hkpbjjacgf2re