Examining the WM/Reuters London Close through the Prism of Foreign Exchange Transaction Cost Analysis
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Examining the WM/Reuters London Close through the Prism of Foreign Exchange Transaction Cost Analysis ITG Analytics - April 2014 Responding to many client requests, the FX team at ITG Analytics reviewed trade data surrounding the WM/Reuters London Closing Spot Rate Service (“the fix”). By observing the factors that influence trading costs using ITG TCA® for FX’s rich quote data we found trade patterns that were unique. Consistent with academic literature,1 we show that volume and volatility around the fix spikes and the spread costs tighten temporarily. In addition, we see mean reversion of the FX rates on days when there is substantial price pressure shortly prior to the fix. Our analysis does not prove the allegations of manipulation brought about by some market participants. Nevertheless, our results suggest that the price movements create a real cost for asset managers. Utilizing order matching or algorithmic trading strategies around the fix can possibly remedy some of the cost burden. With or without collusion, the fix has become an irrational trading period that can cause prices to diverge from an equilibrium state without a concurrent change in fundamental risk. 1 Chaboud, A.P., Chernenko, S.V., Howorka, E., Krishnasami Iyer, R.S., Liu, D. and Wright J.,H. The High- Frequency Effects of U.S. Macroeconomic Data Releases on Prices and Trading Activity in the Global Interdealer Foreign Exchange Market, 2004, Federal Rserve System Board of Governors, International Finance Discussion Papers, Number 823. 1
Examining the WM/Reuters London Close through the Prism of Foreign Exchange Transaction Cost Analysis 1. Methodology and Intraday Patterns The WM/Reuters London Closing Spot Rates are published by the WM Company and Thomson Reuters using a sampling of data during the one minute period beginning 30 seconds prior to the top of the hour. Trillions of dollars in assets contractually tied to these rates trigger billions of dollars in turnover, especially at the end of each month. In our empirical study, we look at tradable quote data during the overlap of London and New York trading before and after the fix rate is calculated. Since the WM Company is responsible for the “fix” relies on Thomson Reuters traded rates, we focus on eight deliverable currency pairs where Thomson Reuters is dominant: GBP/USD, USD/CAD, USD/DKK, USD/MXN, USD/NOK, USD/SEK, USD/TRY and USD/ZAR. Figure 1 summarizes the intraday patterns of volatility and spreads between 11:00 and 17:00 GMT aggregated across all selected currency pairs in 2013. 2 As expected, we observe volatility spikes around the London Close fix at 16:00 GMT, just as we do at 13:30 GMT (major scheduled US economic reports) and 15:00 GMT (FX option expirations, economic reports).3 Figure 1: Cross-sectional average intraday volatility and spread patterns in 2013 (aggregated across all selected currency pairs) Average Intraday Volatility Pattern Average Intraday Spread Pattern 2.5 1.4 1.3 2 1.2 Normalized 1.5 Normalized 1.1 1 1 0.5 0.9 0 0.8 10:40 11:00 11:20 11:40 12:00 12:20 12:40 13:00 13:20 13:40 14:00 14:20 14:40 15:00 15:20 15:40 16:00 16:20 16:40 17:00 10:40 11:00 11:20 11:40 12:00 12:20 12:40 13:00 13:20 13:40 14:00 14:20 14:40 15:00 15:20 15:40 16:00 16:20 16:40 17:00 in GMT In GMT Sources: FXAll and ITG Sources: FXAll and ITG 2 The upward spike in volatility and the downward spike of spread around 16:00 GMT are observed systematically across all currency pairs. Normalization of volatility and spread by their daily values is done separately for each currency pair before aggregating across currency pairs. 3 In addition, volatility also tends to rise around arrival times of other news events (11:00, 12:00, 12:30, 13:00, 13:15, 14:00, 14:30 GMT). 2
Examining the WM/Reuters London Close through the Prism of Foreign Exchange Transaction Cost Analysis During the two spikes at 13:30 and 15:00 GMT, spreads widen in synch with volatility, which appears to be consistent with the risk-averse behavior of FX dealers as they charge more for the increased uncertainty in the markets. What is interesting to note with respect to the London Close fix at 16:00 GMT is that the trading risk premiums, proxied by the spread values, do not respond as expected in a typical high volatility environment. Goettler et al. (2009) 4, modeling such a behavior, noted that even at the times of high volatility, traders with “intrinsic motives” will increase liquidity and tighten spreads. There is certainly enough motivation to manage currency risk in the countdown atmosphere to keep spreads tight without considering front running or manipulation. 2. Trading Activity and Market Participants Trading at the fix during the final minutes before the rates are determined can be intense, especially at month’s end (see Figure 2). Asset managers are accountable for matching WM/Reuters London Closing Spot Rates when they use an international index as a benchmark. This creates acute pressure to match the fixing rate or outperform it without adding undue risk. Many clients shed responsibility for this risk by accepting a foreign exchange bank’s guarantee to match that rate for a set fee, usually two basis points or less. On the other end of the spectrum, some institutional investors enter the closing period with large positions and actively participate in trading around the fix. In the first case, the bank’s foreign exchange trader’s role shifts from that of a risk-neutral market-maker to a cross between an informed trader and a market-maker. In the second, the asset manager is the informed trader with inside information (the large position that is about to be traded). In this two-player race, each has incentive to increase the frequency of their trades as time runs out on the 4pm London close. Our review of the data confirms this pattern. Figure 2: Cross-sectional average trading activity around the fix (across all trading days in 2013) 4 Goettler, R., Parlour, C. and Rajan, U., Informed Traders and Limit Order Markets, 2009, Journal of Financial Economics, Vol. 93(1), pp. 67-87. 3
Examining the WM/Reuters London Close through the Prism of Foreign Exchange Transaction Cost Analysis 1.8 1.6 Number of Trades (Normlaized) 1.4 1.2 1 0.8 0.6 0.4 0.2 0 15:40 15:41 15:43 15:45 15:47 15:48 15:50 15:52 15:54 15:55 15:57 15:59 16:01 16:02 16:04 16:06 16:08 16:09 16:11 16:13 16:15 16:16 16:18 UK Time Sources: FXAll and ITG 3. Return Patterns Although the spreads compress around the fix, indicating that it has become cheaper to trade, there are still costs created by increased volatility. As a result, institutional investors can see significant adverse price action in the form of implementation shortfall even if they are trading on their own behalf. Figure 3 below illustrates the average FX rate moves for the 10% largest positive and negative returns between 15:45 and 16:00 UK time across calendar year 2013 for the selected currency pairs. Regardless of whether or not the base currency was being bought or sold, we see significant average rate drifts from 15:45 until 16:15. The absolute average currency moves are in the ballpark of 10 to 25 basis points across all considered currency pairs. Most of these moves end quite abruptly after 16:00 and partially mean revert in the first few minutes after the fix. Figure 3: Average cumulative returns (starting from 15:30 UK time) on days with the 10% largest positive (on the left) and negative (on the right) FX rate drifts between 15:45 and 16:00 4
Examining the WM/Reuters London Close through the Prism of Foreign Exchange Transaction Cost Analysis 10% largest positive returns 10% largest negative returns between 15:45 and 16:00 UK Time between 15:45 and 16:00 UK Time 30 5 25 0 20 -5 Cum Return (in bps) Cum Returns (in bps) 15 -10 10 -15 5 -20 0 -25 -5 -30 15:45:00 15:46:30 15:48:00 15:49:30 15:51:00 15:52:30 15:54:00 15:55:30 15:57:00 15:58:30 16:00:00 16:01:30 16:03:00 16:04:30 16:06:00 16:07:30 16:09:00 16:10:30 16:12:00 16:13:30 16:15:00 15:45:00 15:46:30 15:48:00 15:49:30 15:51:00 15:52:30 15:54:00 15:55:30 15:57:00 15:58:30 16:00:00 16:01:30 16:03:00 16:04:30 16:06:00 16:07:30 16:09:00 16:10:30 16:12:00 16:13:30 16:15:00 GMT Time GMT Time GBP/USD CAD/USD USD/DKK GBP/USD CAD/USD USD/DKK USD/NOK USD/SEK USD/MXN USD/NOK USD/SEK USD/MXN USD/TRY USD/ZAR AVG USD/TRY USD/ZAR AVG Sources: FXAll and ITG Sources: FXAll and ITG The mean reversion patterns after the fix appear to be partial but very consistent in the above charts. We see a price reversion of 2 to 3 basis points on average which corresponds to size-adjusted spread costs for deal sizes that are distinctly larger than $15mln for all selected currency pairs. Focusing on the USD/CAD pair, we see from Table 1 that the mean reversion does not occur 100% of time; however, the probability of a large price reversal (>1bp or >2bps) is high. For instance, after having observed a very large negative price move in USD/CAD within the last 15 minutes before the fix (the average return is 13bps in Figure 3) one can expect to see a rate increase of more than one basis point in 80 of 100 cases. Table 1 also confirms that the price reversal is quite rapid and typically does not last beyond two to three minutes after the hour. Table 1: Percentage of days when the USD/CAD return between 16:00 and 16:00+t UK time has opposite sign from the large return between 15:45 and 16:00 5
Examining the WM/Reuters London Close through the Prism of Foreign Exchange Transaction Cost Analysis All after-fix return After-fix return in excess of 1bp After-fix return in excess of 2bps t 10% largest positive 10% largest negative 10% largest positive 10% largest negative 10% largest positive 10% largest negative 30sec 77.4 73.3 54.8 50.0 38.7 30.0 1min 71.0 83.3 64.5 80.0 41.9 46.7 2min 80.6 93.3 64.5 76.7 41.9 46.7 3min 77.4 83.3 64.5 66.7 48.4 53.3 5min 71.0 73.3 61.3 56.7 58.1 36.7 15min 51.6 76.7 48.4 63.3 38.7 60.0 Sources: FXAll and ITG 4. Final Thoughts Does the price reversal around the London fix prove manipulation or “banging the close”? It is impossible to tell at this point. If institutional investors are actively trading in a herd- like manner in the FX market, how could one distinguish their trading from the banks compensated for assuming the FX risk of their institutional clients? How could one separate those trades from noise traders hoping to catch one of these directional moves? A telltale characteristic of conventional trading patterns around the fix is that the institutional investors normally start legging into the trade earlier than the banks that trade on behalf of their clients. Bank traders with access to nearly unlimited liquidity would trade much closer to the actual fix in order to reduce their price risk. The idea is highlighted in the work by Kumar and Seppi5 (1992), which models “punching the settlement price.” The institutional investor and banking communities seem to be equally important contributors to the move observed in the charts. The costs to pension funds and mutual funds cannot be ignored. On aggregate, 17 basis points of implementation shortfall for up to 20% of all days can cost the asset management community millions of dollars of un-invested funds. Can we propose any solutions to this dilemma? Can the market reduce volatility and the adverse price movement? Two tentative solutions that have been proposed by the banking industry deserve consideration: order matching and algorithmic execution. Using a third party dark pool, an intermediary can match off trades which would alleviate volatility and prevent information leakage. The other idea, employing algorithms, could help clients match or outperform the fixing rate while protecting their anonymity. A combination of the two, matching all available orders and pushing the excess 5 Kumar P. and Seppi D.J., Futures Manipulation with “Cash Settlement”, 1992, Journal of Finance, Vol. 47, pp. 1485-1502. 6
Examining the WM/Reuters London Close through the Prism of Foreign Exchange Transaction Cost Analysis positions out to an algorithm that creates a fair fixing rate would be another solution that would require less regulatory intervention. Alternate fix times had been suggested in the past, but WM/Reuters London Closing Spot Rate Service has asset values contractually tied to it. Other fixes are available, but they would have to be written into the contracts as benchmarks. That solution does not solve the problem of directional moves. Central bank oversight would also help, if only to remind noise traders or traders seeking excess profits that their trading behavior might be called into question. In the end, banks are given private information in return for providing the service (matching the fixing rate). If the customer’s order is large, and no collusion is involved, it is likely that the market will move in the direction they are trading. It is also likely that a majority of index funds will be on the same side of the trade, exacerbating the situation. Without a matching exchange, this cannot be classified as manipulation. It is a momentary use of market power while providing a risk-bearing service. Additionally, it would be impossible in an unregulated, OTC market to differentiate one bank’s trading pattern from another bank that is actively trading the fix for a client. Both scenarios would involve frantic trading patterns on behalf of each trader, informed and uninformed, attempting to beat the clock and, with some luck, the WM/Reuters London Close fixing rate. From a transaction cost analysis perspective, the costs from the order arrival time until trade execution are on average 17 basis points 20 percent of the time. We would also warn our clients that volatility is at one of the three peaks of the day and trading during periods of high volatility is not recommended. If an institutional investor is using the fix to offset currency risk without being contractually obligated to that rate then they are paying a lot for the service ($1,700 per million), especially if they lack contractual obligations. Clients with a mandate to target the fix rate should either use a matching service to avoid information leakage and thus reduce volatility, or work towards a reform of the system that has created an expensive, irrational trading environment. 7
Examining the WM/Reuters London Close through the Prism of Foreign Exchange Transaction Cost Analysis ©2014 Investment Technology Group, Inc. All rights reserved. Not to be reproduced or retransmitted without permission. #41714-15219 The opinions, positions, and/or predictions taken or made in this document reflect the judgment of the individual author(s) and are not necessarily those of ITG. These materials are for informational purposes only, and are not intended to be used for trading or investment purposes or as an offer to sell or the solicitation of an offer to buy any security or financial product. Nothing contained herein should be relied upon as a representation, guarantee, or warranty as to the reasonableness of the assumptions or the accuracy of the sources used by the author(s). These materials do not provide any form of advice (investment, tax or legal). ITG Inc. is not a registered investment adviser and does not provide investment advice or recommendations to buy or sell securities, to hire any investment adviser or to pursue any investment or trading strategy. All trademarks, service marks, and trade names not owned by ITG are the property of their respective owners. 8
You can also read