|Dr Paul Lynch|
Why do extended warranties on household goods exist? Why does income protection on unsecuredf loans exist? As these forms of insurance are never free, then why do some people prefer a certain loss as opposed to a possible loss? The answer is of course that some people are risk averse and uncertainty can make them feel uncomfortable. The insurance premium in these products therefore represents a price for certainty.
When trading equities, buyside institutions are typically faced with a simple choice; trade the equities as agency or principal. With the huge advances made in recent years by algorithmic trading, agency orders are fast becoming a quantitative technology service, but from the end user's perspective the principal alternative appears relatively unchanged. However, what goes on once the principal transaction has occurred can typically be the same as the agency algorithmic product, except that the volatility of the algorithmic model is being underwritten by the bank.
A principal transaction represents the exchange of risk from the buyside to the sellside. Typically, the prices at which these exchanges/trades are determined are the prevailing bid price for a sell order and the prevailing ask price for a buy order. Hence the trade involves the payment of half the bid/ask spread, as the arrival price of the transaction is actually the mid price. This spread payment is complemented with a commission, often in excess of 15 basis points, which may or may not include a bundled product such as research. f
Thousands of these principal transactions are executed every day, with the customers enjoying an execution with zero volatility, and with all the execution risk now transferred to the investment banks that were counterparty to the transactions. To facilitate this business investment banks have to put up capital (and they rightly deserve to be rewarded for providing such a service) but the question always remains; what is the fair price of the risk premium in a principal trade?
Of the many principal trades that occur, some users will have good knowledge or indicators that the stock will soon move in a favourable direction, whilst many others will simply want to get rid of positions that may take too long to unwind as agency. Obviously, a lot of these principal trades will be done by people who simply do not want any execution volatility at all.
"If the bank cannot retain enough commission to warrant risking its capital then eventually the client will be given unattractive principal quotes..."
The first group, the clients with alpha rich order flow, are continuously monitored by the investment banks who study the retention rate of their clients who trade on a principal basis. If the bank cannot retain enough commission to warrant risking its capital then eventually the client will be given unattractive principal quotes (when requested) to drive away unprofitable business. However, the clients who do not want to wait for the end of an agency execution to receive the execution price represent a good profit opportunity. With many such trades being undertaken, along with any necessary futures hedging, a lot of the risk from market volatility is reduced and the bank just needs to unwind the resulting positions to realise the profit. The unwind may involve crossing the position with natural order flow that comes onto the trading floor. More often than not, however, the position will be unwound by a trader, possibly using an algorithmic trading tool.
Algorithmic trading is becoming an increasingly popular execution method offered by banks to their external clients. One of the strongest selling points of these algorithms is that the client is being given the same tools that are used by the banks' inhouse traders. Hence it is not an unrealistic assumption that the resulting positions that the banks accrue from principal trading are unwound using the same algorithms that are already being offered to the same client at a fraction of the cost. The most likely algorithmic strategies that would be used by the bank in this unwind would be ARRIVAL (trade relative to the mid price at the start of the trade) or PARTICIPATION (trade relative to liquidity on the exchange). A question therefore remains; why doesn't the buyside just use the same algorithm at a fraction of the price and take the execution risk?
Portfolio of risk
One of the main reasons why the buyside may be unwilling to take this execution risk is that the traders are regarding each principal trade as being independent from all others. Whilst the investment bank may rely on doing a hundred of these principal trades on any given day to diversify the market volatility, the buyside trader may only see it as one stand alone trade. However, were the buyside traders to consider the individual trade as one of a hundred that may occur over a six month period, then they would also benefit from the same diversification, albeit over time as opposed to the same day.
With increasingly sophisticated post trade analysis tools being offered by investment banks to the buyside institutions, an assessment of the performance and volatility of participation and arrival price strategies can be conducted. This evaluation should, over a sufficiently large sample, be similar to the banks' own execution performance on unwinding principal positions using the same algorithms. Therefore, these tools now enable the clients to evaluate statistically the risk premium that investment banks charge when dealing as principal and, ultimately, the algorithms and analysis tools will be used as leverage to drive down the price of principal trades.