In 2013 investor confidence, still reeling from the effects of the ’08 crash and the Greek debt crisis was dealt another blow. Revelations emerged of egregious cheating and shady goings on in two of the world’s most widely used FX benchmarks: the ECB fixing rate and the London 4pm Fix . This prompted a lot of serious debate and culminated in a number of related initiatives including but not limited to a Financial Stability Board special working group on FX benchmarks, the UK government’s Fair & Effective Markets Review and the Global FX Code.
A truly delightful critique of the 4pm Fixing scandal penned by Dan Davies in 2014 explains the Benchmark rigging scandal with the help of a market for oranges. 
Davies argues the scandal was really the fault of the clients “demanding tighter and tighter pricing from their brokers without asking how those businesses would remain profitable and acting as if they benefitted from the protections of Best Execution without wanting to pay for them” which based on the story of oranges is plausible enough, but is probably the wrong way round. A more probable theory would suggest that Banks engaged in Bertrand competition, quoting ever tighter spreads to crush their competition in order to capture market share, which would be much more in character with bank behaviour. The value for dealers after all is in the information of the order flow.
The Fair and Effective Markets Review published their findings in 2015. The report defines an effective market as one where “market participants can form, discover and trade at competitive prices”. The opacity of FX markets means that participants must expend effort and resources to achieve this, an effort that most participants are unable and/or unwilling to make. There’s a justification for this. Most people are not buying dollars because they think the USD is undervalued, they are buying dollars because they need to purchase a commodity, an asset or service. This aspect of FX, participation on a non-profit motivated basis is often assumed to mean the clients don’t care about costs. When clients can work them out, they do care. Deeply.
This information asymmetry between clients and dealers was displayed in all its splendour by a 2017 ESRB working paper re-published by the ECB last year. The authors sampled half a million trades, transacted between 204 dealers and 10,062 clients. Clients above the 25th percentile were able to place their dealers in competition via multi-dealer platforms. They paid on average 2.5 bps for their EURUSD hedges. Meanwhile, the 75th percentile (the bottom quarter of the survey) paid on average 30 bps for their EURUSD FX forward hedges. Comparing the results between the best and the worst percentiles we can see the difference between competitive and monopolistic pricing. We have to wonder how many FX risks remain unhedged because the client does not have the resources to access a multi-dealer platform?
Three critical roles
Benchmarks fulfil three critical roles in promoting fair and effective markets. 
The primary function is to provide transparency on the likely terms participants will face when they come to market. Are price quotes high or low, relative to the benchmark? Judged on this criteria, taking a price once a day or only at certain times of the day does not really help mitigate the information asymmetry between dealers and clients. This function of a benchmark is the fair price function. It acts as a constraint on price gouging, and lowers the cost of entry.
A second function of a benchmark offers objectivity and independence. Dealing brings together two counter parties with mutually offsetting positions. For every buy there is a sell. Sellers seek to sell at the highest rate, buyers prefer the lowest price. The rate at which the positions are matched off will determine the value to the client of the trade. Use of an external, independent benchmark provides both parties with an assurance that the matching rate was not set to favour one counterparty at the expense of another. This feature of a benchmark plays a pivotal role in the emergence of peer to peer pre trade netting, with the first such buyside only network going live this year. An additional feature of pre-trade netting off market is that notionals can be reduced before going to market, reducing potential market impact.
The third function of a benchmark encourages participation. When a market is opaque, dealers will often sponsor a benchmark to advertise the likely cost of participating in the market. When participants have greater visibility on their expected costs, or when those costs are lower, because transparency has fostered competition, declining costs can unleash pent up demand. The ESRB study by Hao et al focused exclusively on EURUSD, the world’s most liquid currency pair. If there are large aggregate unhedged risks in EURUSD how much more unhedged risk could transparency release in less liquid pairs?
Transparency allows participants to understand the true cost of participating in the market. Without it, how are market participants to determine whether a price is fair. Many FX market participants were under the impression they were protected by duties of Best Execution, but the rules of engagement of the wholesale market have developed on the assumption that all participants are able to look out for their own interests. Discerning whether a price is competitive or not does requires effort. Benchmarks can fulfil this role, but only if they are specified with transparency in mind.
One of the problems that users of the 4 pm Fix have is that orders must be passed to dealers well in advance of the actual fix. Holding out to customers to deliver a known quantity of an asset, at an unknown price at a fixed point of time entails risk. In order to hedge this risk dealers have to build up inventory of the asset before the Fix. Banks were criticised for the manner in which these risks were managed leading to claims of front running, collusion etc. but the problem is structural. If there are more buys than sells in aggregate, the Fix will be biased higher, and the imbalance must be purchased in the market beforehand. This means that market impact occurs before the Fix, and dissipates afterwards. These effects are predictable. 
One market solution has focused on how inventory risk can be managed through the construction of the Benchmark. Created by Dr Jamie Walton, Siren Efficient Execution benchmarks are derived from the work by Almgren and Chriss on optimal execution. The algorithm optimises the trade-off between market impact costs and price uncertainty adapted from a framework that is widely used in equities to trade Market On Close orders. The SIREN benchmarks produce much smoother trading trajectories with much lower market impact.
Market impact has been called “Finance’s analog of the Heisenberg Uncertainty Principle”. As the authors note, you cannot identify what the price impact will be after you trade until you trade, and then, you cannot determine what the price would have been, had you not traded. This circularity frustrates a lot of effort to identify market impact. However when it comes to FX, HUP does not necessarily have to hold because participants can compare their quotes to an independent stream.
When we created New Change FX the scandal around the Fix was just breaking out. We were invited to provide FX transaction cost analysis for a large hedge fund. The only data available to measure their performance came from the platform on which the hedge fund had done their trading. This was circular. The prices the client observed had already been changed by their participating on the platform. In FX, prices move in anticipation of order flow, ex ante. Asking for a price, changes the price before you trade. The solution was to build a benchmark from the ground up, aggregating a sufficiently wide set of data sources, collecting data anonymously and unobtrusively; and streaming the data at sufficiently granular increments, in real time.
Agency business models
One of the growth areas in FX in recent years has been the development of agency business models. Banks and non-bank providers hire out their technical expertise in the form of execution algorithms, and their networks, in the form of direct market access. Each dealer and vendor (via credit intermediation) provides a unique pool of liquidity. Comparing the performance of different algos or execution services against an independent mid-rate allows these vendors to demonstrate the quality of their service, against a neutral and objective benchmark. Moreover, if several algorithms are being compared, they can both be measured by the same yardstick.
Every platform or venue has their own skew that results from the imbalance between buyers and sellers on the platform. Because there is sampling error across platforms, a benchmark can quantify the size of the skew, and its persistence. Determining whether that skew is temporary or permanent is key. Temporary effects arise from an order that pushes the market away from equilibrium. A buy order will hit an Offer, and Sell order will hit a bid. Provided the size of the order is less than the available liquidity to meet the order, there should be no change in the market equilibrium rate. The mid-rate should remain the same. If the platform mid-rate changes while the benchmark mid-rate stays the same, this would indicate a more serious problem. An order or series of orders which consumes more than the liquidity available at that rate will cause the vendor mid-price to move. This is a permanent market impact. This effect can persist over time.
In a principle transaction, market impact prior to execution (the change in mid-rate between the order submission time and execution fill time ) is borne by the client. Market impact post trade is borne by the dealer, unless the client needs to come back to the market again. In an agency transaction, all market impact effects, both pre and post, are costs that are borne by the client. Execution providers use independent Benchmarks to demonstrate how well market impact costs are managed.
One of the reasons why the industry suffered such reputational damage in the last few years was a lack of transparency. New initiatives have come to market that embed independent benchmarks in their processes. The market is beginning to embrace transparency. Lower spreads for more participants has the potential to unleash a new wave of pent up demand for currency products. This should have positive effects for both customers and dealers alike.
 ESRB working paper Discriminatory Pricing in OTC FX derivatives. Hau,Hoffmann, Langfield, Timmer 2017
 Benchmarks in Search Markets, Duffie, Dworczak and Zhu 2017
 Peer to Peer FX platforms challenge Wall Street https://www.ft.com/content/087a03dc-ec43-11e9-a240-3b065ef5fc55
 Did Reform Fix the Fix NBER working paper ITO and Yamada 2017
 Grinold &Kahn,Active Portfolio Management 1997