How much does it cost institutions to trade the common stocks in their actively managed portfolios? The hard element to estimate in transaction costs (which also include commissions, as well as taxes and SEC fees on sales) is, of course, market impact. When it undertook to measure the cost of trading, Bankers Trust chose to define market impact as a stock’s price movement from receipt of an order by the trading desk until execution, with the effect of changes in market level in the interim adjusted out.
This definition of market impact reflects the cost of not being able to execute orders when entered at the then-prevailing price. While many factors unrelated to the bank’s presence in the market can affect any one security’s price over the course of a trade, across a large sample much of the impact of these other factors will average out.
The bank’s study examined all purchases and sales generated by actively managed employee benefit portfolios over the six-month period December 1977 to May 1978. Purchase transactions (numbering 2,330) had a median time lapse between entry and execution of 5.7 days. Sale transactions (numbering 1,172) had a median time lapse between entry and execution of 3.9 days.
The dollar-weighted transaction cost of purchases significantly exceeded the cost of sales, despite the fact that taxes are paid on sales but not on purchases. The reason was the market impact component, which reached 0.92 per cent for purchases versus 0.18 per cent for sales, overwhelming the tax differential. Although the transaction costs of broker-initiated and bank-initiated sales were not significantly different, broker-initiated purchases cost significantly less than bank-initiated purchases.
Attempts to build a model that could predict transaction costs on the basis of such factors as market capitalization, liquidity ratio, number of shares in the trade, etc., were disappointing.