In recent months there has been a renewed interest by asset owners and trustees in the costs of foreign exchange transactions undertaken on their behalf by external counterparties. The initial salvo was fired in the US by several high profile pension funds that undertook litigation against their foreign exchange counterparties, to recover costs associated with the facilitation of foreign currency trades.

Historically most asset owners have ignored the costs of foreign exchange transactions. They recognized that transaction cost analysis (TCA) was difficult to undertake given the lack of underlying information driving currency transaction costs. Unlike equity markets, currencies are not exchange-traded assets - foreign currency transactions take place within a fragmented, highly decentralized over-the-counter market. As foreign exchange trade information is not readily observable within such a market, the underlying liquidity information necessary to build an accurate transaction cost analysis process has remained elusive. In practice, many asset owners assumed what could not be measured could not be managed.

Despite the challenges posed by foreign exchange transaction cost analysis, asset owners and trustees maintain a fiduciary responsibility to seek and deliver best execution to their investors across all asset classes. As such, every manager must have an appropriate systematic process in place to audit, analyse, and evaluate the quality of execution being received, and attribute the costs being charged to investors across each asset class and security managed.

Morgan Stanley’s FX TCA framework aims to provide clients with the ability to:

•    estimate transaction costs and market impact ex-ante (pre trade)

•    identify the fair-value benchmark price against which the effective price received should be measured,

•    attribute transaction costs to key underlying market factors such as order size, depth of market liquidity, order flows and volatility and

•    generate comprehensive TCA reports on every transaction done with Morgan Stanley and other counterparties

The TCA framework serves to challenge several traditional execution practices that have influenced foreign exchange execution across many investment portfolios. One legacy practice identified is the utilization of a fixed point in the trading day, the most common being the 16:00 London Fix, as a universal benchmark for currency execution. The problem here is that a clear distinction must be made between performance benchmarking and execution benchmarking. In performance benchmarking the goal is to mark-to-market the value of a portfolio or index on a daily basis in order to consistently report and measure its value. By contrast, execution benchmarking serves to measure the quality of execution obtained on one or more transactions in order to verify and validate that best execution is being achieved on behalf of investors.

A subtle dilemma arises when a fixed point in time price is employed as “the benchmark”. The problem is that one should not expect the ‘best’ prices to arrive systematically at any particular point in time each and every day. Indeed, it is more likely that the “best” prices will tend to arrive somewhat randomly at different times of day. Further more, the notion of “best” for a buyer will tend to be the opposite for a seller. It is quite likely that the best prices to buy currency arrive at different times than the best prices to sell. The legacy of benchmarking to point-in-time fixing rates is unfortunate because although fixing rates serve a valid purpose for portfolio valuation, they should by no means be construed as representing best execution in FX.

[Graph 1 description:: Sample paths for EURUSD from March 2011, with prevailing exchange normalized to zero at 07:00 London each day, show how random variations in market prices prevent the “best” price from systematically appearing at the same time each day. Instead, price levels tend to be distributed around points in time. Horizontal bars marking the 16:00 London fix illustrate the breadth of price dispersion in EURUSD around this time each day. Green and red dots show how the best time to buy or sell EURUSD, respectively, varies day-to-day and is never the same for both buyers and sellers.]

To address the need for a viable benchmark for currency execution, Morgan Stanley’s TCA framework identifies a fair-value benchmark as being the flow-weighted average price (F-WAP) obtainable over a specified time-window. Like volume-weighted average pricing (V-WAP) in exchange traded securities transactions, F-WAP provides an average price directly linked to the liquidity profile typically seen in FX markets.

There are two major benefits to using F-WAP as an execution benchmark. The first has to do with the fact that F-WAP is measured over a window of time, which can be calibrated to coincide with the time a particular investor actually needs to trade currency (such as when they are actually trading international equities). The fact that this window of time can be customised means that the F-WAP price can always be made to reflect the actual trading opportunity set available to each individual investor. Rather than a “once size fits all” benchmark price, that may or may not correspond to the timing of an investor’s currency trades, the F-WAP provides a summary measure of the prices the investor could have traded on when they actually were trading.

The second benefit has to do with the fact that as an average, the F-WAP possesses the statistical property of representing the central-tendency price over an execution window. By definition, this means that the F-WAP not only represents a fair price that an investor could have obtained during the time they were trading but it also represents a price that avoids extreme prices during the window. It stands to reason that in any trading opportunity set measured over a window of time, some proportion of prices observed will constitute worse execution prices while the other proportion represents better execution prices. Because the F-WAP is a central tendency price, it splits the difference, making it an appropriate benchmark for both buyers and sellers of currency.

[Graph 2 description: A viable execution benchmark should reflect the trading opportunity set during the time window when FX trading actually needs to take place. The top diagram shows the path of EURUSD on a randomly selected day along with the 24-hour F-WAP and the F-WAP measured during the DAX exchange hours. Note how each F-WAP benchmark price is approximately centered in the range of prices observed during its respective time-window. Each benchmark was derived by flow-weighting the observed EURUSD prices by liquidity. The bottom diagram illustrates the typical profile of daily liquidity for EURUSD with DAX exchange hours highlighted in green.]

The F-WAP benchmark also helps to reveal the pitfalls associated with another legacy practice known as standing instructions. Under standing instructions, currency execution that needs to be conducted to settle international security transactions is conducted by a counterparty once confirmation of the security trades is known. This practice can lead to a delay in execution that can extend 36 hours or more. The pitfall arises because the currency execution can take place at a time far removed from the original window of time when the underlying securities were traded, exposing the portfolio to the risks of currency volatility and price drift.

Unless the asset owners’ currency exposure is being actively monitored and managed at all times by a trader under standing instructions, delayed execution essentially introduces an unmanaged and uncompensated FX settlement risk into the portfolio through price slippage.

The risks of delayed execution in EUUSD can be seen clearly in the first graph above; An investor executing underlying security trades early in the London trading day, but who waits to execute currency for settlement until 4PM that day, or even 4PM the next day, exposes the portfolio to considerable EURUSD price drift and slippage. QSI analyzed the effects of delayed execution in an international equity portfolio, taking into account all currencies in the MSCI World index, and found that the potential cost of delayed execution could be as high as 500 basis points on an annualized basis.

To fulfill fiduciary responsibilities, and evade sub-optimal execution, it is preferable that asset owners avoid standing instructions that do not define explicit shortfall constraints on a transaction (e.g. shall not be filled at a worse price than prevailing at the time an underlying security order was submitted). Pairing FX execution with every equity transaction helps to eliminate delayed execution slippage and assist with proper auditing of FX transactions for best execution. In addition, utilizing trading algorithms such as TWAP and Morgan Stanley Fix, which delivers the F-WAP, further reduces the chance that investors trade on the worst prices during an execution window.

Paul Aston, is an Executive Director and co-head of the Quantitative Solutions & Innovations (QSI) group, which is part of Morgan Stanley’s fixed income FXEM sales and trading desk.