Several prior studies have estimated institutional commission rates for sporadic time spots, and these estimates are consistent with ours for matched time spots. For instance, Berkowitz et al. (1988) estimate the institutional commission costs of NYSE trading in 1985 to be 0.18 %, as compared to our average commission rate estimate of 0.31 %. Keim and Madhavan (1997) find an average institutional commission rate of 0.20 % of trades during the period of January 1991 through March 1993, as compared to our estimate of the average commission rate of 0.24 % for the same period. Jones and Lipson (2001) find the average institutional commission rate of NYSE trades to be 0.119 % in the second and third quarters of 1997, which coincides with our estimate of the average commission rate of 0.119 % for the same period.14 Note that the aforementioned researchers’ estimates of commission costs are based on their samples of NYSE institutional trades in different time spans, while our estimate of the average commission rate is for all NYSE trades. The above studies together with our research suggest that retail commission rates have declined more than institutional commission rates and that by 1997 both rates became similar in size.
Commissions vary over time and are not a constant proportion of stock price (see Jones 2002 for a complete survey of the history of commission regulation in the US). Per-share commissions are the dominant form of payment between brokers and their institutional clients. The negotiated rate per share charged to institutional investors depends on trade size and the research provided by the broker.15 Commission rates charged to retail traders differ across brokers, while discount brokerages allow lower commission with less research service. For instance, per-trade commission is charged for Internet trades by online discount brokers. For both institutional and retail traders, trade size is inversely related to the realized commission rate. The latter reflects the composition of small trades in total trading volume. A higher realized commission per share reflects the presence of more small trades.
In order to check this contention, we estimate the proportion of small trades in total trades of the Dow Jones Industrial Average (DIJA) stocks using trade data from NYSE TAQ (Trade and Quote) databank. We follow the approach of Barclay and Warner (1993) and Chan and Fong (2000) in defining orders with less than 500 shares as small trades. We then calculate the ratio of the number of small trades to total trades for each day. The daily result is aggregated into a quarterly measure for each stock, and then the cross-sectional average of all 30 Dow stocks is estimated.16 For the period 1993–2005,17 the quarterly average small trade ratio is 0.608 with a standard deviation of 0.034. Its correlation with ACR is 71.8 %, and 50.3 % with the logarithmic ACR. Both correlations are significant at the 1 % level. This strong correlation lends support to our notion of ACR as a proxy of trading noisiness.