August 1, 2018

Abstract

August 1, 2018

Contents

August 1, 2018

Approved By

Ratna Sahay

Prepared by the Monetary Capital Markets Department (MCM). The team comprised Romain Veyrune, Asad Qureshi, Waesh Khodabocus, Miklos Vari, and Xingbang Weng.

Contents

  • INTRODUCTION

  • FOREIGN EXCHANGE MARKETS

  • A. Market and Pricing

  • B. Market Development Since 1979

  • C. Market Data Availability

  • NEW METHODOLOGY

  • A. Spot Transactions

  • B. Banknote Transactions

  • C. Non-Spot Transactions

  • PRELIMINARY ILLUSTRATIVE RESULTS OF TESTING THE NEW METHODOLOGY

  • A. Types of Official Actions Analyzed

  • B. Aggregate Results of Applying the Methodology

  • BOXES

  • 1. Methodology for the Computation of FX Benchmarks

  • 2. Market Data Availability Through DataStream and Bloomberg

  • FIGURES

  • 1. Deviation Between Banknote and Wholesale Market Exchange Rates

  • 2. Distribution of the Bid/Ask and High/Low Spreads

  • 3. Cumulative Distribution of Official Rate Deviation from the Mid of the H/L

  • 4. Forward H/L Cumulative Distribution

  • 5. Use of Official Exchange Rates

  • 6. Auction Rates

  • 7. Results of the Proposed Revisions of the MCP Policy

  • TABLE

  • 1. Developments in FX Markets

  • APPENDICES

  • I. Taxonomy of MCPs—Comparative Treatment Under the Current and Proposed New Policy

  • II. Applying the Proposed Methodology to Foreign Exchange Auctions

  • III. Computation of the Official Exchange Rate for Margin Requirements

Introduction

1. This background paper describes in more detail the proposed new methodology in the Main paper (Review of the Fund’s Policy on Multiple Currency Practices: Initial Considerations) to assess exchange rate deviations that may result in multiple currency practices (MCP). The current methodology relies, for spot transactions, on a uniform two-percent rule for permissible spreads which does not reflect current market realities and does not provide sufficient guidance on the definition of market exchange rates. The new methodology would retain the principle of commercial reasonableness, i.e., differentials in exchange rates arising from official action would only give rise to an MCP if they exceeded what could be attributed to the normal commercial costs of exchange transactions, but would rely on a market-based, country specific market norm instead of the current uniform two percent threshold. The paper also presents some illustrative preliminary results of the proposed methodology on existing country practices based on available information.

2. The new methodology is expected to help the MCP policy better reflect market realities. The introduction of a market-based norm is facilitated by developments in foreign exchange (FX) markets, which have contributed to greater uniformity of market practices across jurisdictions, and the availability of granular market data for most IMF members. Adding a tolerance margin for deviations from the market norm remains important to avoid minor deviations leading to MCP findings.

Foreign Exchange Markets

A. Market and Pricing

3. Official action can distort the system of exchange rates that is kept unitary by competition in the foreign exchange market. Competition between intermediaries both in the wholesale and retail markets reduces the opportunities for arbitrage and keeps rates in both markets together. However, it does not lead to a single rate in the foreign exchange (FX) market as each intermediary and each transaction has a slightly different cost structure; rather, it leads to a range of interrelated rates that move together. This range of rates defines the norm for the market. International standards for exchange rate benchmarks usually provide several rates, such as: (i) the average buying (bid) and selling (ask) rates (thereafter bid/ask); (ii) the intraday high (most depreciated) and low (most appreciated) rates (thereafter H/L); and (iii) the market opening and closing rates. Official action can result in exchange rates that do not reflect the cost structure in the FX market and, thus, deviate from the market norm.

4. Price formation in the FX markets differs depending on the settlement date of the transactions. Based on the global financial market standard for settlement dates,1 staff distinguish between two types of FX transactions: those that settle at or less than t+2 days (spot) and those that settle at more than t+2 days (non-spot transactions). The settlement date affects price formation in these two markets and, thus, the rates in the two markets cannot be directly compared.

Spot Markets: Transaction Settled at Maximum t+2 days

5. The interdealer (wholesale) market is an important component of FX market pricing. FX intermediaries, such as banks, are generally motivated by profits.2 They intermediate FX transactions to end users (i.e., the retail market), such as importers and exporters, under a no-loss constraint, and close positions arising from transactions with their clients (risk management) among themselves (i.e., the wholesale market). FX intermediaries also transact in the wholesale market to take advantage of arbitrage opportunities (proprietary trading), thereby keeping the system of exchange rates unitary.

6. In the end-user segment, the pricing of FX transactions depends on their size.

  • Large transactions, usually with corporations, are dealt with at the banks’ treasury level, as are banks’ transactions with other intermediaries, because they create positions for the intermediaries that should be managed without delay in the wholesale market to avoid exposure to exchange rate risk. Thus, these transactions are fully integrated in the pricing of the wholesale market rates due to their impact on banks’ FX positions.

  • Transactions in smaller amounts, typically with individuals, are managed by banks’ customer services at the branch level. Branches are assigned a budget and a schedule of rates for the day by the banks’ treasury, which depends on how the treasury plans to manage the exposures arising from small size transactions. The rate of retail transactions is the wholesale rate to which the intermediary’s overhead costs and profit margins are added, which depends on competition in the retail segments of the market. Based on the settlement date, under the new methodology, FX transactions involving banknotes would be considered a segment of the spot retail market, which have additional costs related to the handling of banknotes compared to electronic payments.3

7. Interdealer market rates define all other spot rates where an interdealer market exists. In the review of the MCP policy that started in 1979 (1979 review)4, staff considered “spot telegraphic transfers” as the core exchange transaction from which exchange rates for other transactions are derived in all markets. Staff proposes the interdealer market rates as the core exchange rates for spot transactions for three reasons: (i) this is the “real” market on which the spot rates are formed; (ii) this market can be monitored continuously based on reliable benchmarks; and (iii) official actions are typically reflected in wholesale-type transactions.

Non-spot Markets: Transaction Settled at More Than t+2 days

8. The FX non-spot market provides hedging opportunities vis-à-vis exchange rate risk. This market includes various products such as foreign exchange forwards, swaps and options. It can be considered as a market for future exchange rate transactions for which the exchange rate is locked in advance or guaranteed. Consistent with the treatment of spot transactions, the exchange rates prevailing in the wholesale forward market on the day of the transaction are considered as an adequate proxy for commercial reasonableness for forward transactions.

B. Market Development Since 1979

9. A major development since 1979 is the deepening and increasing sophistication of international FX markets. In 1979, the only data available to staff regarding FX market transaction amounts was the USD 5.3 billion daily turnover on the New York stock exchange. Spot, outright forwards, and swaps represented respectively 55.2 percent, 5.3 percent, and 29.5 percent of the total transaction amount. According to the latest BIS “triennial central bank survey” (Table 1), the average daily turnover in the foreign exchange markets was above USD 5 trillion, more than four times its level in 1995, underscoring the rapid development of FX markets in recent decades.

Table 1.

Developments in FX Markets

article image
1 Adjusted for local and cross-border inter-dealer dou We-counting {ie ‘net-net” basis). 2 The category ‘other FX products” covers highly leveraged transactions andfor trades whose notional amount isvariable and where a decomposition into individual pJain vanilla components was impractical or impossible. * 3Non-US dollar legs of foreign currency transactions were converted into original currency amounts at average exchange rates for April of each survey year and then reconvened into US dollar amounts at average April 2016 exchange rates. 4 Sources: Euromoney Tradedata: Futures Industry Association; The Options Clearing Corporation; Bl S derivatives statistics. Foreign exchange futures and options traded worichwde.Source: Bank for International Settlements.

10. Liberalization of FX markets since 1979 has reduced the number of countries where an interdealer market does not exist. An interdealer market is necessary to close the positions that arise from intermediaries’ transactions with their clients. However, monetary authorities occasionally replace the interdealer market by closing intermediaries’ position themselves. In the past, this was associated with regulation impeding interdealer market transactions, such as a full surrender requirement of FX export proceeds to the monetary authorities. Trends in FX market liberalization have supported the emergence of wholesale markets in most members. However, there are still cases where wholesale markets do not exist:

  • The country has no separate legal tender, i.e., there is no national currency and another country’s currency operates as legal tender (14 members).

  • Exchange rates are fixed at par or with a very narrow spread between the central banks’ FX buying and selling rates (10 members). A meaningful wholesale market is unlikely to emerge in these circumstances.

  • The member’s FX regulations explicitly or implicitly prohibit interbank transactions. The number of countries having a full surrender requirement to the central bank declined from 22 in 1999 (the first date for which information on surrender requirements is available in the AREAER online database) to 9 in 2016. Similarly, the number of countries with partial surrender requirements to the central bank declined from 44 to 27 during the same period.

11. One issue encountered at the time of the 1979 review was the difference in practices across domestic FX markets. FX market development since 1979 has been accompanied by the standardization of practices across markets that has greatly increased the comparability of FX benchmarks across members, together with technological advancement.

12. Electronic platforms contributed to standardization of transactions and collection of data on market rates. Electronic platforms are the preferred trading channel, with a share above 50 percent for all customer segments, and are available for all instruments and investors across the globe. Spot is the segment with by far the highest fraction of trades conducted electronically, at 64 percent. Due to the practical benefits of electronic execution, such as straight through processing5, most of the voice trades are, eventually, booked electronically. Some market reports suggest that as much as 95 percent of all spot transactions could, in fact, be electronic.

13. Ascertaining the significance of the banknote segment in the spot market remains a challenge. The 1979 review estimated that the banknote segment amounted to USD 100 million in June 1978 (compared to a 5.3 billion daily turnover for the rest of the market), based on information from “New York banknote dealers.” Intuitively, staff expected a decline in the relevance of this market segment due to the development of alternative, more convenient, payment methods.

14. Available evidence does not support a definitive conclusion regarding the significance of the banknote segment. There are estimates of the circulation of the euro and the USD outside the euro area and the United States based on banknotes shipments data, but not of market turnover.6 However, part of the shipment is due to the demand for banknotes as a store of value and, thus, does not necessarily reflect an increase in market turnover. Few countries, usually with a history of important use of FX banknotes in the spot market, publish volumes for this segment. For instance, the volume in one country declined from 2.7 percent of spot transactions in 2003 to 0.5 percent in 2017.

15. While derivatives were not a particular focus of the 1979 review, the development of FX hedging products since 1979 warrants a more detailed discussion in this review. Table 1 shows that all FX derivatives, including forward, swaps, and options experienced a rapid increase in turnover. Foreign exchange swaps account for 47 percent share of the global market, while outright forwards account for 14 percent. FX swaps are mostly traded among intermediaries and used for liquidity management, while forwards are mostly used by end-users.

C. Market Data Availability

16. Availability of wholesale market data based on a standardized methodology increased substantially since 1979. At that time, staff had to interview FX traders by phone to seek their assessment on what was the reasonable bid and ask spreads over the previous 12 months. With this method, only 36 currencies could be included in the 1979 review. Nowadays, standardized exchange rate benchmarks, including bid/ask, high/low, and open/close both on intraday and closing basis, are prepared and published daily by professional international benchmark administrators for all but five IMF members.7

17. Staff proposes using Thomson Reuters DataStream as the primary source of information for wholesale exchange rates. For the 14 members, for which data are not available through Thomson Reuters DataStream, data can be sourced from Bloomberg. Both companies prepare reputable, well-known, and widely used international benchmarks for exchange rates. Box 1 describes the methodology used to compute financial benchmarks by Thomson Reuters and Bloomberg.

18. Both data sources are equally reliable. Both Bloomberg and Thomson Reuters methodologies are subject to independent assessment of their compliance with the international standard for the computation of financial benchmarks of the International Organization of Securities Commissions (IOSCO), thereby ensuring the integrity of data. The ready availability of data from two independent benchmark providers increases the robustness of the proposed methodology.

19. The methodology relies on a predefined order of data sources to limit the risk arising from excessive reliance on a single data source. Data sourcing relies primarily on two independent data providers: Thomson Reuters DataStream is the primary source and Bloomberg is the secondary source. If data become unavailable for a member from the primary source, staff will source data from the secondary source. Both these sources have published independent business continuity plans to ensure uninterrupted services. However, if data for a member from either the primary or the secondary source cannot be used, representative exchange rates from other sources, including from the authorities may be used. Staff will assess the representativeness of exchange rates from such sources based on whether they have been computed in compliance with the IOSCO principles for the computation of financial benchmarks.8

20. Anecdotal evidence suggests that the typical difference between wholesale and retail rates is around 70 basis points. There is no standardized benchmark for retail rates, as the focus is on the wholesale segment where price formation takes place. However, central banks publish an average of the intermediaries’ FX rates for banknotes for customer information where banknotes play or used to play an important role (Figure 1). The median of the difference between wholesale rates and banknote rates was 70 basis points on March 26, 2018, reflecting overhead costs and the profit margin of intermediaries across countries. Banknote wholesalers contacted during the review indicated that they charge 15 to 20 basis points per 100 USD (30 basis points for higher risk location), leaving the typical profit margin at about 50 basis points.

Methodology for the Computation of FX Benchmarks

The World Markets/Reuters (WMR) computes and publishes FX reference benchmarks for 155 currencies. The WMR FX spot benchmarks are currently available for 155 currencies; coverage for forward rates is available for 76 currencies on Thomson Reuters. The WMR has become a de facto standard for foreign exchange (spot, forwards, and Non-Deliverable Forwards) rates and is most widely used in the market, index compilers, and by country authorities. The computation and governance of the benchmark is based on publicly available documentation1/ and is reviewed against appropriate international FX benchmark regulations and guidance, specifically the principles for the computation of the financial benchmark of the IOSCO.2/

The WMR computes and publishes FX reference benchmarks validated through pre-defined robustness checks aimed to limit manipulation. The benchmark provider ensures that the collected rates are validated based on pre-established robustness checks, outlier rules, and expert judgment. Suspicious contributions are discarded. The rates used are the median of the validated contributions from multi-contributor sources. Thomson Reuters undertakes a periodic review of the appropriateness of using specific data suppliers to provide the data for the calculation and validation of the benchmark rates. The review process is intended to, among other things, assist in ensuring that the benchmark rates are: (i) based on reliable and observable market data that reflect the economic realities of a given market, and (ii) themselves reliable representations of the FX market.

The sources of data used for the benchmark, by order of priority and depending on availability, are: (i) actual traded rates; (ii) rates of orders placed in matching order systems; and (iii) banks’ quotes. In less liquid markets, the benchmark is predominately based on bid and offer rates, or on prior transactions data.

The primary infrastructure used to source spot exchange rates consists of various FX trading platforms used globally. The information on actual transactions is collected from the platforms of Thomson Reuters Order Matching, EBS, and Currenex. Quotes are collected from Reuters dealing system. Other sources, including central bank spot rates and official fixings, are also used as and when necessary to fill gaps or reach a minimum number of contributions. Data sources are regularly checked for reliability and representativeness. Wherever possible, a multi-contributor source of rates is used in preference to a single contributor unless an “official” fixing is being used.

The methodology is based on the principle that a market in each currency pair represented by the benchmark must genuinely exist, and that the market must be active. The standardized benchmarks, including bid/ask, high/low, and closing both on intraday and closing basis are available. The computation process and validation methodology, however, depend on the liquidity of each market and are grouped in three categories.

  • Trade Currencies: The top 22 highly liquid currencies referred as “Trade Currencies.3/ The transactional trade and order rates are sourced from liquid platforms, such as Thomson Reuters Order Matching, EBS, and Currenex for validation. The process performs continuous validation, and identifies outliers.

  • Nontrade Currencies: All other currencies not identified above are assumed as “Non-trade Currencies.” The primary source of rates will be actual traded rates taken from Thomson Reuters Matching, and EBS, as detailed above, and the secondary source will be the rates on the order matching systems. Commercial bank quotes from Reuters Dealing System will be captured where neither Trades nor Orders are available. Hourly snapshots of quoted rates are taken over fixed periods and medians are calculated.

  • Locally-traded Currencies: The benchmark administrator sources quotes and transacted rates from locally-traded markets, where offshore trading is not permitted. The published spot rates reflect activity in that market’s time zone and is for 9 currencies. 4/

  • CB or Official-fixing Currencies: These currently include 15 currencies whose rates are sourced from central banks, in the absence of any continuous and reliable availability of wholesale market quotes.5/ The benchmark administrator reaches out to market contributors, global clients and other counterparts to assess their reliability and representativeness.

The Bloomberg BFIX FX reference rates represent prevailing FX market prices and are published based on Bloomberg Generic Composite (BNG) every 30 minutes. The Bloomberg “BFIX” are FX rates that are half-hourly snapshots of the BGN, using a time-weighted average price (“TWAP”), around the fixing time. The Bloomberg Generic Composite rate (BGN), is a sophisticated pricing algorithm that produces accurate indications of bid and ask quotes that are derived from hundreds of quality sources, including indicative and executable price quotes from money-center and regional banks, broker-dealers, inter-dealer brokers, and trading platforms. However, data are not based on any actual market trades but on indicative rates contributed by market participants.

1/ WM FX Reuters Benchmark: Spot and Forward Rates Methodology Guide. Thomson Reuters is the benchmark administrator for the WM/Reuters Rates and thus has primary responsibility for all aspects of the benchmark determination process.2/ Thomson Reuters IOSCO Statement of Compliance.3/ The currencies referred as “Trade Currencies”, include AUD, CAD, CHF, CNH, CZK, DKK, EUR, GBP, HKD, HUF, ILS, JPY, MXN, NOK, NZD, PLN, RON, RUB, SEK, SGD, TRY and ZAR. For purposes of the methodology, the data sourced from the benchmark providers for the calculation of the H/L for a member would be based on the relevant time zone of the member.4/ The currencies are: CNY, IDR, INR, KRW, MYR, PHP, THB, TWD and BRL.5/ The currencies are: AOA, BOB, BZD, CRC, ETB, HNL, HTG, JMD, KGS, MRO, MVR, NIO, SRD, TTD and VEF.
Figure 1.
Figure 1.

Deviation Between Banknote and Wholesale Market Exchange Rates

(March 26, 2018)

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Source: IMF staff calculations.Note: Deviation is calculated as follows: (mid of buy and sell banknote rate / mid of H/L)-1, except for one country, where it is (mid of buy and sell banknote rate / mid of electronic transfers)-1.

New Methodology

21. Staff proposes replacing the current two percent rule with an indicator that reflects the range of exchange rates traded on a given day. Staff considered, but did not retain, the option to improve the application of the current two percent rule. The challenges with the application could arguably be partly addressed, for example, by choosing the mid of the bid/ask of the day, as a proxy for the market exchange rate to which the two-percent would be applied.9 These changes would address the current challenges with contemporaneity and the identification of the market rate discussed in the Main paper. However, this approach could still result in MCPs arising from more than two-percent differences under stressed market conditions (Figure 2). The introduction of an indicator that reflects market conditions better is, thus, necessary to increase the relevance of the norm in different market circumstances.

Figure 2.
Figure 2.

Distribution of the Bid/Ask and High/Low Spreads

April—October 2017

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Source: Bloomberg, Reuters and IMF staff calculation.

A. Spot Transactions

22. The market indicator that defines the market boundary or maximum deviation between the rates of two transactions in the market is determined by the highest and lowest rates in the market on a given day (H/L). The bid/ask spread provides a country-specific indication of transaction costs for the FX market of each member. However, rates arising from official action may well be outside the bid/ask spread while being in line with the rate of transactions that actually happen in the market, which could lead to “false” findings. As official or reference rates are usually computed at a precise time of the day, they can exceed the bid/ask spread if market rates change after the computation of the reference rate. On the other hand, the H/L adjusts to exchange rate developments during the day, reducing the likelihood of such deviations. Based on the data reviewed by staff, official rates are 3.5 times more likely to deviate from the bid/ask spread than from the H/L range. FX market participants, contacted by the staff, also considered the H/L norm as being most representative of the market rates. Although the H/L range would be different for each country, this approach would be consistent with the principle of uniformity of treatment, as the results would be determined based on the same methodology, applied consistently across the membership.

23. The main limitation of the H/L range is that it can be susceptible to manipulation, but benchmark providers have some built-in safeguards. As the H/L range determines the “market boundaries”, manipulation can occur through wash or round-tripping trades 10 that could be used to widen the H/L range for the trading day deliberately, regardless of whether based on traded or on quoted rates. However, benchmark operators have built-in robustness checks as described in Box 1, which provide safeguards against potential manipulation.

24. Staff considers the H/L range of a given market during the trading day as an adequate norm to identify deviations from the market exchange rate. As shown in (1), the requirement is to keep rates arising from official actions (R) in the range of rates defined by the H/L of a given trading day. The High and the Low are both a bid for a counterparty and an ask for another counterparty.

(1)LOWRHIGH

R is the exchange rate arising for official action;

HIGH is the highest wholesale exchange rate of the day, i.e., the most depreciated rate of the day;

LOW is the lowest wholesale exchange rate of the day, i.e., the most appreciated rate of the day.

25. To avoid capturing insignificant deviations, an additional tolerance margin could be envisaged. While each deviation from the market norm challenges the unity of the system of market exchange rates, staff is of the view that the revised MCP policy should not capture small deviations from the norm as they do not have significant economic consequences. In practical terms, the tolerance margin (m in (2) below) on a given day in a given market could be set as a percentage of the mid of the high and low rates (MH/L). An MCP would be found if the exchange rate arising from official action exceeds the H/L and the tolerance margin.

(2)|RMH/LMH/L|m%

26. Staff considers a tolerance margin of +/- 2 percent around the mid of the H/L as adequate based on the empirical distribution of official rate deviation from the mid of the H/L ((3) below and Figure 3). Typical distribution threshold for tail events would be reached for +/-2 percent; 95 percent of the deviations are less than 2 percent from the mid of the H/L. In addition, +/- 2 percent minimizes the change in the current policy with respect to members, who have already included it in their policy (e.g. for exchange taxes).

(3)Min{LOW;MH/L2%}RMax{HIGH;MH/L+2%}
Figure 3.
Figure 3.

Cumulative Distribution of Official Rate Deviation from the Mid of the H/L

April—October 2017

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Sources: Bloomberg, Reuters, and IMF staff calculations.

27. The addition of the tolerance margin would avoid findings due to differences in the cost structure of the spot rates which are compared. As shown in Figure 1, the +/- 2 percent tolerance margin would accommodate the typical spread between the wholesale spot rate and the banknote rate (about 70 basis points) in most cases studied during this review.

28. The tolerance margin would also address the cases in which no independent wholesale market exists. Generally, there would be a negotiated market segment unless the authorities fully intermediate the wholesale markets. In such a case, the monetary authorities capture all FX inflows in the market via surrender requirements or other means, and distribute the proceeds to end-users for the payment of current account and other transactions. In these cases, the wholesale reference rate is the mid-rate between the rate at which the authorities capture FX revenues (buying rate) and the rate at which they redistribute it (selling rate). A typical illustration would be a fixed exchange rate regime with a buying rate and a selling rate and no market outside the central bank. In such circumstances, the new methodology would require that any rates arising from official action remain in the +/- 2 percent tolerance margin from the mid of the official buying and selling rates.

29. A single deviation from the norm because of official action would be considered an MCP. Staff considered the possibility of finding an MCP only if the rate arising from official action deviates from the norm for a minimum number of consecutive occurrences. However, it was considered inappropriate to retain this option because of practices that lead to infrequent breaches of the norm over a longer period (e.g. bilateral payment agreements that have quarterly settlement).

30. Once an official action is identified, deviation from the norm will be monitored on a continuous basis. Once the rate arising from official action is identified, the ready availability of data and the relative ease of applying the methodology will make it possible for staff to monitor the exchange rates continuously to detect deviations promptly. (See Box 2 on the availability of market data through the Fund’s subscription to benchmark providers.)

Market Data Availability Through DataStream and Bloomberg

Market data is available through the Fund’s subscription to Thomson Reuters DataStream and Bloomberg. The current subscription of the Fund through IMF Library (CFS) covers sufficient licenses to Thomson Reuters DataStream and Bloomberg. These licenses are accessible to various users throughout the IMF in area departments and other functional areas. There is a budget proposal approved to add licenses during the coming fiscal year, whereby almost all research officers in country teams would have access to the required market data. The current service agreement provides both close of day and intraday data for exchange rates covering bid/ask, mid-rates, and intraday H/L. The table below details the availability of instruments on FX market data.

article image

B. Banknote Transactions

31. Official actions that affect banknote exchange rates are rare. They include the buying and selling of banknotes by the central bank with the public in general or the setting of exchange rates for specific transactions in banknotes (e.g., concessional cash FX rates for specific purposes). Such transactions, cannot be considered FX intervention as the central bank directly trades with the public and not with intermediaries. Central banks that buy and sell banknotes usually operate fixed exchange rate regimes11 and apply spreads that result in deviations between the banknote and wholesale rates that are lower than the +/-2 percent permissible margins.

32. Banknote transactions would be treated as spot transactions. As for all spot transactions, the methodology would require that banknote rates resulting from official action remain within the H/L or the permissible margin of the wholesale spot market. The main difference between the exchange rates of banknotes and other spot transactions is due to the additional costs related to the use of banknotes, which result in larger spreads for these transactions. These additional costs usually cannot be quantified and continuously monitored and, thus, applied in the methodology. However, based on information available to staff, the permissible margin of +/-2 percent is expected to provide ample room to absorb the additional costs of banknote transactions. That said, if the authorities can provide representative data on banknote rates in their FX market, such data could be used to apply the methodology. Staff would compare the results of the methodology using the H/L data provided by the benchmark providers and the data on banknote rates provided by the authorities, and the finding would be based on the data that results in the most favorable treatment for the member.

C. Non-Spot Transactions12

33. Official actions can affect exchange rates of transactions with maturities longer than spot. Consistent with the proposed methodology on spot transactions, such rates would be considered under the policy if the authorities use them in transactions or require that market participants use them, and if the official action creates market segmentation. Official actions can impact FX derivatives.13 In addition, official actions can also take the form of exchange guarantee schemes akin to forwards or options, while not explicitly structured as such.14

Forwards

34. The market norm for forward exchange rates arising from official action would be the daily H/L of the forward wholesale market. The market norm and the tolerance margin would be applied to the assessment of the exchange rate as in the case of spot transactions. An MCP would arise if the forward rate arising from official action deviated from the H/L of the forward market and the tolerance margin.

35. The primary source of information for these rates is Thomson Reuters DataStream. It provides outright H/L forward rates quoted by interbank market participants for 76 currencies, where forward markets exist. Three can be sourced from Bloomberg as secondary source. Those forward exchange rates can be monitored continuously. As data come from the most liquid market, 98 percent of the H/L were less than 2 percent in 2017 (Figure 4). In cases where wholesale forward market exchange rates from the benchmark providers cannot be used, forward exchange rate data that are representative of the market from other data sources, including from the authorities, can be used (see also footnote 8). If representative data from alternative sources are not available, staff propose using theoretical prices. The theoretical forward rate is computed by adding the difference between local and foreign interest rate for the maturity of the forward to the spot wholesale market rate.15 To avoid an MCP, the rates arising from official action should remain in the +/-2 tolerance margin of the theoretical price.

Figure 4.
Figure 4.

Forward H/L Cumulative Distribution

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Source: Thomson Reuters and IMF staff calculations.

36. Market forwards and theoretical prices can be determined for most members with a national currency. Calculated forwards prepared by the benchmark administrators are based on a standardized methodology and the interest rate data available (bank offered rates and government securities rates). Market and calculated forward rates can be monitored on a continuous basis for 178 members. The rest of the members do not have national legal tender.

37. Staff would need to assess whether the domestic interest rate used by the benchmark administrator is the most representative one in the available data. The valuation tools of the benchmark administrators allow replacement of some inputs with other data that may not be publicly available but accessible to the authorities and the country team. Staff may use alternative data sourced from local markets if considered more representative of the actual funding rate in that market for a given maturity.

FX Swaps

38. Exchange rates of FX swaps arising from official action would be assessed with the methodology. Under the proposed methodology, the spot rate would be assessed as any other spot rates, while the forward rate arising from official action—or the implicit forward rate in foreign exchange swaps involving interest exchanges—would be compared with the market forward rate or the theoretical price at same maturity, calculated based on the interest rate differential in the market. The official forward rate arising from the FX swap provided by the authorities should remain in the permissible margin of the calculated forward rate to avoid an MCP.

Options

39. With respect to conditional transactions, the authorities may provide the market with an option to purchase or sell FX at a given rate at or by a future date.16 The FX option would be exercised if the exchange rate appreciates or depreciates more than a given level, i.e., the strike rate. FX options that are provided to some market participants could give rise to MCPs because they may give a subsidized hedge to some counterparties only, thereby segmenting the market. In contrast, if the option is provided to market participants without segmenting the market (i.e., via a competitive auction meeting the criteria for auctions), the option would be considered as an intervention and, thus, would not be assessed under the MCP policy.

40. The methodology would be implemented only if the option is exercised.17 The “effective exchange” rate, in the case of a currency option, is the strike price plus the option premium. If the option is exercised at an effective exchange rate that deviates from the equivalent market transaction with similar terms, an MCP would arise. The methodology would examine whether, at the inception of the option, the effective exchange rate of the contract was within the H/L of the effective exchange rates of comparable market options or the tolerance margin of +/- 2 percent from the mid of the H/L. The market effective exchange rate will be obtained from wholesale market transactions with similar terms, when available from benchmark providers. When data from benchmark providers are unavailable, representative exchange rates from other sources, including from the authorities could be used; in absence of such data, a theoretical price of the option with similar terms will be used.

41. In practice, the assessment of options would have to rely mainly on theoretical prices. Option markets are often not deep and standardized enough to provide relevant price benchmarks. However, both Reuters and Bloomberg18 provide the possibility to calculate theoretical option prices for a wide range of contracts, including “exotics.” The default setting is based on the data available to the benchmark administrator and their assumption regarding expected volatility. The default setting uses the “Black-Scholes formula” for simple contracts—a standard model, widely used in financial markets. Staff propose using Reuters as the default calculator for calculation of theoretical option prices under the methodology, and, if such calculator is not available, Bloomberg’s calculator would be used.

42. Staff would assess whether the default parameters used by the theoretical price calculator of benchmark administrators are the most representative for the country. Staff may use alternative data sourced from local markets to change the parameters of the calculation, in particular implied volatility and the local currency interest rate, to obtain option prices. Alternative implied volatility should be based on prices formed in the domestic market that reflect expected exchange rate volatility. Alternative local interest rates should be based on actual cost of funding for the relevant maturities. In the absence of such information, the assessment would be based on the theoretical option price derived from the default setting of the calculator, depending on the type of option.

Preliminary Illustrative Results of Testing the New Methodology

This section presents the preliminary results of the application of the new methodology to spot transactions. The new methodology takes into account the refined application of the concept of official action, as laid out in the Main paper. These tests are based on information currently available to staff and market conditions prevailing at the time when the tests were conducted. These preliminary results are meant to show the potential impact of the proposed changes on existing and possible new MCPs. These preliminary results are merely illustrative. A full jurisdictional analysis, including discussions with the authorities, would be made if and when the new policy enters into effect, before an MCP finding can be made.

A. Types of Official Actions Analyzed

43. Staff tested the combined effect of the proposed methodology and the refined concept of official action as described in the Main paper on different types of official actions that can result in MCPs.19 This section presents the results of the tests on the following types of MCPs: (i) official exchange rates; (ii) dual and multiple FX markets; (iii) auctions; and (iv) official actions increasing the cost of foreign exchange transactions (see Appendix I for the main types of MCPs in these categories).

Use of Official Exchange Rate in FX Transactions

44. Official exchange rates are often used in transactions of the authorities.20 Staff analyzed whether these official exchange rates (often called reference rates) could give rise to MCPs based on the refined concept of official actions and permissible spreads. The analysis considered only those official exchange rates which are used in FX transactions (Figure 5).21 As a result, staff established that 108 members used an official exchange rate for transactions with FX end-users or foreign exchange intermediaries.

Figure 5.
Figure 5.

Use of Official Exchange Rates

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Sources: 2016 AREAER and survey of country teams. 189 members and 3 territories.

45. Staff identified 38 members using the exchange rate of the previous days in official transactions, including 5 members calculating official exchange rates with a more than one-day lag. Out of these, 27 members compute their official rates based exclusively on market exchange rates—i.e., they use either a combination of interbank and dealer-client rates or any one of these rates. Five members determine their respective rates based inter alia on the results of FX auctions and the remaining five retain some discretion in the determination of the official rate. Information is unavailable on the lag used by 1 member.

46. The methodology would reduce the number of MCP findings due to official exchange rates. Currently, 12 countries have MCPs arising from the use of official exchange rates that deviate or may deviate from other exchange rates. Based on available information, staff identified thirteen instances in which official rates deviate from the H/L. Among those, four also deviate from the mid of the H/L by more than +/-2 percent and would result in MCPs under the new methodology. This would represent a decrease by two thirds of the number of existing MCPs arising from the use of official exchange rates. None of the current MCPs assessed based on potentiality are captured by the new methodology. The four cases where the official exchange rate is outside the permissible margins emerge from the systematic review of all members’ practices.

47. A few official exchange rates calculated with a lag have been excluded from the findings. Of the members using lagged official rates for transactions, twelve members have their official rates outside the H/L range. Out of these twelve cases, six also deviated from the mid of the H/L by more than +/-2 percent. These cases were nonetheless excluded from the results, in line with the proposal not to consider them under the MCP policy if they meet certain criteria (see Box 4 in the Main paper).22

Dual and Multiple FX Markets

48. Official action can lead to market segmentation, and give rise to MCPs. While not necessarily targeting specific transactions or economic agents, an official action could impose specific markets by regulation for certain transactions or hinder arbitrage between market segments, which challenges the system of market exchange rates. Staff tested the proposed revisions to the policy on known segmented markets.

49. Parallel markets are not the only case of more than one market in a single jurisdiction. There are few cases where two legal markets coexist due to different trading infrastructures (some transactions are required to take place exclusively on a specific platform). A further case is where capital controls effectively segment legal FX markets. Exchange rates may differ across segments if arbitrage between the segments is hindered.

50. Staff has identified 32 members having dual or multiple FX markets. The number of such markets was determined based on a survey of the country teams and includes dual/multiple markets reported in the 2016 Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER), except when the dual rates arise from an auction (results for auctions are presented separately below) or an official rate.

51. Consistent with the proposal in the Main paper, the methodology was applied to 10 of the 32 dual markets that were either reported as legal or have an undetermined legal status.23 The reference used for the H/L and the mid of the H/L was the primary (largest by volume) wholesale spot market. The mid of the H/L in the secondary market was considered as the rate arising from official action (R in Equation 3). In the absence of H/L for the secondary market, the test used an alternative representative rate (usually a single rate).24

52. The application of the methodology and the proposed revisions to the MCP policy capture fewer cases of MCPs due to market segmentation. There are currently eight MCPs arising from dual or multiple markets (first column of Figure 7). Five out of the 8 MCPs arise from illegal markets and, as such, are excluded in line with the proposed revision to the policy. The systematic review of all countries that report dual or multiple FX markets and for which data is available, including countries currently having no MCP findings, revealed that exchange rates of the secondary markets in five countries are outside the H/L and permissible margins.

Figure 6.
Figure 6.

Auction Rates

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Source: IMF staff calculations.
Figure 7.
Figure 7.

Results of the Proposed Revisions of the MCP Policy

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Source: IMF staff estimates.

Auctions

53. As proposed in the Main paper, FX auctions (including multi-price auctions) would not be considered under the policy as long as they do not segment the FX market. Auctions that meet certain criteria are not considered as segmenting the market,25 while auctions that do not meet these criteria, would be assessed as other legal dual markets.

54. Staff reviewed the auction regulation of 33 members to identify those which are inconsistent with the criteria. 26 The determination was based on a review of the publicly available information and a survey sent to the country teams (Appendix II). The survey indicates that 19 auctions appear to result in market segmentation by providing FX for specific purposes and only partially based on the price proposed.

55. The methodology has been applied to auctions that appear to be segmenting the FX market. The reference used for the H/L and the mid of the H/L, to which the permissible margins are applied, is the spot wholesale market. Considering the bid rates at the auction as reflective of the secondary market (auction), the weighted average rate (WAR) of successful bids represents the rate arising from official action (R). Figure 6 illustrates the integration of the auction rate with the rest of the market rates. In all cases, the auctioneers publish the marginal rate of the auction and the WAR of successful bids.

56. The new methodology, together with the proposed exclusion of auctions that do not segment the market from the definition of official action, reduces the number of cases where the auction exchange rate is beyond the permissible margins. Of the 33 members operating an auction, 19 either do not provide access to all FX intermediaries in the wholesale market or include a criterion for the allotment which is different from the bid rate. Applying the test to those 19 auctions, the number of findings declines from the current 8, all of them based on potentiality, to 2. One finding overlaps with a current MCP.

Official Action Increasing the Cost of the Foreign Exchange Transactions

57. Official actions that increase the cost of FX transactions would continue to be captured under the MCP policy. These include exchange taxes (4 MCPs in 2017), margin requirements, and other types of unremunerated compulsory local currency deposit requirements prior to FX purchases (3 MCPs in 2017). The exchange rate arising from official action is calculated as the deposited local currency augmented by the interest forgone due to the deposit requirement. The application of the methodology consisted of checking whether the effective exchange rate resulting from official action remains in the H/L and the +/-2 tolerance percent margin around the mid of H/L (see Appendix III)

58. The number of cases where official action results in an increased cost of FX transactions exceeding the permissible margins declines to 3 from 7 with the application of the methodology. The two members currently maintaining taxes on FX transactions exceeding 2 percent are captured by the new methodology. Authorities of three other members maintain a compulsory deposit requirement. Official action effectively increased the cost of buying FX beyond the H/L range for two of these members and outside the +/- 2 percent permissible margin for one of them (see Appendix III).

B. Aggregate Results of Applying the Methodology

59. Figure 7 depicts the MCPs as of end-2017 and the preliminary results discussed above.27 The first column of the chart shows 35 of the 45 MCP findings in 2017, classified into the four MCPs categories described above—i.e., different exchange rates for distinct transactions, dual and multiple FX markets, auctions, and official action increasing the cost of foreign exchange transactions. The 10 MCPs not included on the chart arose mostly from practices that could not be tested during this review due to lack of data. Among the 35 current MCPs, 18 are based on potential rather than actual deviations from the current two-percent rule. The second column illustrates the findings based on the application of the H/L norm only. The third column presents the findings resulting from official actions that are out of the H/L and deviate by more than +/-2 percent from the mid of the H/L.

60. The proposed changes to the policy lead to fewer but better targeted findings. The H/L norm, in combination with the proposed changes in the definition of official action under the new policy, would reduce the number of findings to 33 (second column of Figure 7). The number of MCPs would decrease to 14 with the addition of the +/-2 percent tolerance margin (last column of Figure 7). Exchange rates of members with MCPs based on potentiality remained within the permissible margin, while the systematic review of all members flagged a few cases that were not previously identified as MCPs.

Appendix I. Taxonomy of MCPs—Comparative Treatment Under the Current and Proposed New Policy

article image
article image
article image
article image
article image

Appendix II. Applying the Proposed Methodology to Foreign Exchange Auctions

1. Central banks often organize FX auctions to buy or sell FX with market participants. The use of auctions is prevalent within the IMF membership: 38 members are reported to operate auctions in 2016, according to the AREAER.

2. The revised policy proposes to exclude from the MCP policy auctions that do not segment the FX market. The assessment of auctions under the revised policy would consist of two steps: First, it will be identified if the auction is organized according to the criteria that ensure that the auction does not segment the FX market. This step is based on a review of the auction regulation. Second, for auctions that fail to meet the criteria the quantitative methodology proposed to identify MCPs would be applied.

3. The rest of the appendix is organized as follows: (i) it presents the conceptual framework that establishes the relationship between such FX auction and the broader FX market; (ii) proposes criteria to identify auctions organized according to best practices (auction regulation screening); and (iii) discusses the preliminary results of the application of the MCP test for auctions that do not meet the proposed criteria. The presumption is that auctions that fail the screening are auctions that segment the FX market as some FX end-users have a preferential access to FX. These are in fact, allocation systems, for which the term “auctions” is a misnomer.

Conceptual Framework

4. In the absence of official actions (other than organizing the auction), auction rates are expected to be integrated with the interbank (wholesale) market rates. At the time of the auction, FX intermediaries are confronted with the choice of participating in the auction or refraining from participation because they expect to find better opportunities in the interbank market. Assuming that the auction is opened only to authorized dealers (i.e., corporations, whose business is to buy and sell FX for a profit as opposed to end users of FX), the rate offered by each participant in the auction should be in line with the rates at which it can re-sell FX to other intermediaries (i.e. the wholesale market) or to its clients (i.e. the retail market). At times, intermediaries may have to close a position at a loss (i.e. bid below the reservation rate), but, on the longer term, intermediaries should not make a loss on FX dealing.

5. Figure II.1 below shows how participants in the auction are expected to bid compared with the market rate.

  • The most competitive bid represents the participants’ reservation rate. They correspond to the amount of FX that intermediaries absolutely need to buy or sell even at a loss to close a position. This rate should be close to the worst alternative in the market, i.e., the Low if the central bank buys FX and the High if the central bank sells FX.

  • Then, participants are expected to post bids in line with the average of the market rates (trade bid), which corresponds to what they could trade for a normal profit in the broader market.

  • Finally, there may be some opportunistic bids, corresponding to what they are ready to commit while not having the resources available before settlement because their proposed rate is well above the market.

Figure II.1
Figure II.1

A Stylized Auction for a Central Bank Purchasing Foreign Currency

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Source: IMF staff calculations.

6. This conceptual framework indicates that the marginal rate of the auction should remain in the H/L range. In a well-calibrated auction, the central bank selects bids that correspond to rates ranging from those of the reservation bids (most competitive rates) up to the trade bid rates and reject the opportunistic bids (non-competitive rates), keeping the marginal rate and the auction weighted average rates in between the High and Low rate of the auction day. Deviations could happen in the two following circumstances: (i) the intended allotment is notably less than the position that participants have to close, allowing the central bank to pick only reservation rates and, thus, to overperform the market; and (ii) the central bank has an excessively large allotment objective compared to the position of the market and is forced to pick opportunistic bids, thereby underperforming the market.

7. The relationship between market rate and reservation rate also depends on the depth of the FX market. In the case of a liquid FX market, the reservation rate will be equal to the market rate as banks can sell large volumes of FX in the market at competitive prices. In case of less liquid FX markets, the reservation rate might be lower than the market rate (as shown above). The reason is that a bank might be willing to sell FX at a rate lower than the market rate to sell large quantities, knowing that the sale of large quantities in the market could result in a sharp appreciation of the rate.

Auctions Regulation Screening

8. To distinguish auctions that do not segment the FX market from other auctions, the proposed policy would consider whether the auction regulations meet two criteria: (i) access; and (ii) allotment, as explained below:

  • Access: All intermediaries in the wholesale FX market can participate in the auction which is not open to end-users of FX (i.e., the clients of the intermediaries).

  • Allotment: The rate is the only criterion to allot the bid at the auction. In practical term, bids are ranked according to the exchange rate offered and FX is awarded to the bids according to their ranking in the limit of the intended allotment with the bids offering the best exchange rates satisfied first. In addition, the authorities do not set constraints on the range or level of exchange rate that can be offered in bids because such restrictions could prevent arbitrage between the interbank wholesale market and the auction, thereby segmenting the market.1

9. The 2 percent limit introduced by some members to avoid that the auction gives rise to an MCP would need to be removed to satisfy the allotment criterion. This will support the functioning of the FX interbank market and enhance the auction performances.

10. Figure II.2 below shows the results of the preliminary assessment for the 33 countries where auction regulations were available. Among them, 14 had auctions that seem to satisfy the criteria. Information obtained from auction regulations was complemented by staff experience in those countries. Among the 19 auction that did not meet the criteria, 17 failed the allotment criterion, while only two failed the access criterion. None failed both at the same time.

Figure II.2
Figure II.2

Preliminary Results of the Auction Regulation Assessment

(Number of countries)

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Sources: IMF staff assessment.

Application of the Methodology

11. Once auctions that are organized according to best practices have been identified based on the analysis of the FX auction regulations, the methodology is applied for those auctions that segment the FX market. This section describes these tests. Figure II.3 presents the preliminary results of the tests applied to the auctions that do not meet either the access or allotment criterion.

Figure II.3
Figure II.3

Results of the Quantitative Tests

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Sources: IMF staff calculations.

12. First, the implementation considers that there is a deviation from the market-based norm (H/L) whenever the weighted average of the successful bid at the auction is not within the high-low range of the interbank market.2 The weighted average of the successful bid rates at the auction is considered as reflecting the secondary market that is the auction. A weighted average of the successful bid rates within the high-low range indicates that there is some room for arbitrage between the auctions and the wholesale market.

13. Second, to retain only the most significant breaches of the norm, in cases where the weighted average rate is not within the high/low range, it will be assessed whether the weighted average rate deviates by more than+/- 2 percent from the mid of the FX wholesale market H/L. The country would be considered to have an MCP when this second threshold is exceeded. While six countries have current MCPs arising from FX auctions, the new methodology identifies only two countries where the weighted average exchange rate exceeds both the H/L and the permissible margins.

Appendix III. Computation of the Official Exchange Rate for Margin Requirements1

1. Margin deposit requirements, if remunerated below market rates due to official action, would be assessed under the MCP policy following the methodology at maturity. Thus, the effective exchange rate of the transaction (R), that includes the forgone interest would be compared with the wholesale spot market H/L and the permissible margin. When the authorities require that margin requirements are remunerated below market rates they raise the all-in cost of buying FX, and change the effective exchange rate that market participants face.

R should be computed in the following way:

R=DC*(ikig)*G*T

where,

  • DC is the amount of domestic currency required by official action to be deposited to obtain one unit of foreign currency.

  • ikis the annualized domestic currency market interest rate.

  • igis the annualized rate at which the cash margin is remunerated.

  • (ik – ig) represents the difference between market rates and the remuneration of the deposit margin.

  • G is the margin requirement. G=0.5 means that a market participant wishing to exchange 100 unit of the domestic currency must deposit 50 unit of the domestic currency as cash margin.

  • T is the period during which the deposit must be kept. Since interest rates are annualized, it is necessary to adjust the formula for cases where participants will be required to keep the deposit for less than a year.2 For instance, the participants are required to keep the cash margin for 3 months, T=0.25.

2. For example, the authorities may require banks to provide deposit margins equal to 100 percent of the amount of domestic currency to be exchanged (G = 1), up to a year (T=1), for no remuneration (ik = 0), while the market interest rates is 14 percent (ig = .14). The amount of domestic currency (DC) demanded is the market exchange rate.

R=MH/L*(1.14)*1*1

3. This would result in the effective exchange rate to be outside of the High-Low range and to deviate by more than 2 percent from the mid of the H/L (see Figure III.1). Under the proposed policy, an MCP would be found.

Figure III.1
Figure III.1

Margin Deposit Requirement 1

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Source: IMF staff calculations.

In another example, banks are required to provide cash margins equal to 50 percent of the amount to exchange (G=0.5), for three months (T=0.25), for no remuneration (ig = 0), while market interest rate is around 13 percent (ik = 0.13).

4. The effective exchange rate at which market participants can transact is outside of the High-Low range but remains within the 2 percent margin. Under the proposed policy no MCP would be found.

Figure III.2
Figure III.2

Margin Deposit Requirement II

Citation: Policy Papers 2019, 015; 10.5089/9781498318860.007.A005

Source: IMF staff calculations.
1

The definition of spot transactions can be found in the Model Code of the Association Cambiste Internationale, which is the standard benchmark for global FX and OTC markets. (https://acifma.com/Why-The-Model-Code)

2

State owned banks may have different motivations.

3

Under the current policy, banknote transactions are considered “other” (i.e., non-spot) transactions.

4

See Background Paper I.

5

Straight through processing streamlines the transaction-related processing time. It allows information that has been electronically entered at the trade level to be transferred during multiple points in the settlement process without manually re-entering the same information repeatedly, thus optimizing the settlement process.

6

Estimation of euro currency in circulation outside the euro area, External Statistic Division, European Central Bank, Eurosystem, April 2017. Location of U.S. currency: How much is abroad, Federal Reserve of New York, bulletin, 1997.

7

Staff conducted a survey of data available at the IMF on (i) interbank market rates; (ii) official rates; and (iii) parallel market rates. It relied on information available to the country teams, including from member countries’ authorities. Through the survey, data were made available for four out of the five countries, reducing the list of members with no exchange rate information to one.

8

While there would be a strong presumption for using data from Thomson Reuters DataStream and Bloomberg in the methodology, if data from these sources are not available, or the data that are available appear to be manifestly irregular or inaccurate, staff can use other exchange rate data which are representative of the market. Alternative data sources that could be used by staff would include the following: (i) publicly available independent data source, (ii) publicly available data from the authorities, or (iii) data from the authorities. In all these cases staff would determine whether the data are representative of the market. For non-spot transactions the methodology may also rely on theoretical prices as reference rates, as further described below.

9

Fundamentally, however, the bid/ask spread is an imperfect benchmark for a market-based norm under the MCP policy. By analogy with the principle that applied to the spread around parities, the 1979 review focused on spread between buying and selling exchange rates, which was understood at the time as the bid/ask spread in the market. However, the buying and selling exchange rates of the par value were single official rates that defined the boundaries of the market while the bid/ask spread represents the average of buying and selling rates in the market and as such do not define market boundaries. In most cases (86 percent), the bid/ask spread is lower than 2 percent with most of the distribution concentrated between 0 and 75 basis points. However, the number of instances in which market spreads exceed 2 percent is far from insignificant. Two circumstances explain higher spreads: (i) less liquid markets in which the cost of doing FX business is higher; and (ii) stressed market conditions (such as large depreciation of the exchange rate), which could happen in liquid as well as less liquid markets.

10

Wash or round-tripping—an attempt to inflate transaction volumes or prices through recurring trades of the same FX between a handful of larger players.

11

This is the case for two central banks and for two regional central banks.

12

Due to evolving trends in FX market sophistication, assessment of non-spot transactions may become more frequent under the new policy. There might be circumstances in which market related information on prices may be difficult to obtain or assessing non-spot exchange rates may be complex. Thus, further refinement of the methodology and guidance may be needed after more experience with assessing non-spot transactions is gained.

13

FX forwards are mostly used by end-users to hedge exchange risk and fund FX transactions. FX Swaps are frequently used for liquidity management and funding both in FX and local currency.

14

FX options allow the holder the right (but not the obligation) to buy or sell FX from the option seller at or by a certain time in the future, for a set price. Two types of exercisable variations are generally offered; ones exercisable at any point before expiration, or ones that can be exercised only at time of expiration date.

15

The intermediaries selling FX to a client at t+>2 immediately borrow local currency at interest rate (i) with a t+>2 maturity and use the proceeds to buy the FX at the spot rate. The FX purchased at (t) is invested in the money market at a remuneration of (i*) until t+>2. The cost for the intermediary, is, thus, the interest rate differential (i-i*), which is passed on to the client with a margin representing the intermediary profit. In practice, forward market exchange rates are often different from what would result from the interest rate differentials.

16

Such options can also take the form of exchange rate guarantee schemes.

17

Staff recognize that the availability of an option to hedge exchange rate risk, in itself provides a value to the beneficiaries, even if the option is not exercised.

18

Thomson Reuters provides FX Option Calculator as FXOC and Bloomberg provides similar calculator as OVML; both include multiple pricing models depending on the type of option being evaluated.

19

The exercise captured some measures that do not necessarily correspond to the existing MCP findings.

20

Usually, an average of the market rates—mostly interbank—is used to determine the official exchange rates. The market exchange rates can be actual traded rates or quoted rates. Not every member publishes the methodology for calculating the official rates.

21

Exchange rates used for accounting and valuation purposes are not covered by the MCP policy in the absence of an exchange transaction in which the official rate would be used.

22

Under the proposed revisions to the MCP policy (see Main paper), lagged exchange rates should not lead to MCP findings in the following circumstances: (i) the lagged official exchange rate is computed as a weighted average of transactions between intermediaries and/or between intermediaries and their clients; and (ii) the transactions included in the calculation should have occurred no more than one day before.

23

In the Main paper, staff propose to exclude illegal parallel markets entirely from the policy and address these cases through the policy on exchange restrictions.

24

For countries with dual or more markets, except for the primary FX market, data on other market segments are not always readily and/or publicly available. In most cases, either the central bank or the private sector monitor the different segments of the market and prepare representative rates for each of them. The reference rate is usually based on a survey of the operators in the parallel market, which are often FX bureaus. There is typically no standardized methodology to compute these rates, which are often opening or closing rates, or a simple average of the rates reported by the operators. Transactions in the parallel market are sometimes described as “cash” transactions (25 out of 31 reported by country teams) although it could not be confirmed that these markets are purely banknote markets, as “cash” is a term often used to describe spot transactions in the informal market. For the current assessment, in one case, data on the H/L for the dual market was available from the benchmark providers. In the remaining cases data was obtained from country teams, and included opening and/or closing rates, or the midpoint of bid-ask rates, whichever was available and considered representative for the methodology.

25

As described in the Main paper, the auction should (i) grant access to all intermediaries in the country’s wholesale FX market so that they can all make purchases for themselves and on behalf of their clients; (ii) ensure that the bid exchange rate submitted by participants is the only criterion used to determine whether the bid or the offer to buy or sell FX is selected; and (iii) not set constraints on the range or level of the exchange rates that can be submitted, as this may segment the auction from the rest of the FX market.

26

Data for 20 of the 33 members operating an auction were obtained from public sources or from country teams. For some auctions, data were not available as the authorities do not publish their auction results and did not share data with staff. For the rest, no auction was organized during the review period.

27

The results are based on data for the period April to October 2017.

1

This however, does not mean that the FX purchased at the auction can be used inconsistently with the general FX regulations.

2

Staff considered using the mid of the successful bids at the auction as the exchange rate arising from official action that would be compared with the H/L of the wholesale market and the permissible margin. However, staff is of the view the weighted average exchange rate of the successful bids is a superior representation of the auction results and its relationship with the wholesale FX market. Furthermore, the weighted average of the successful bids is published by the authorities, and thus it is easily accessible for staff, while the mid of the successful bids is generally not published and would require additional information provision by the authorities.

1

Here the focus is on the case where the cost of acquiring FX is increased as a result of official action. Similar considerations would apply to a case where acquiring FX is subsidized.

2

The maturity of the margin might not be stipulated by the authorities. In this case, staff could use the average duration of margin deposit based on information by market participants or those subject to the margin requirement.

Review Of The Fund’s Policy On Multiple Currency Practices: Initial Considerations
Author: International Monetary Fund. Legal Dept., International Monetary Fund. Monetary and Capital Markets Department, International Monetary Fund. Strategy, Policy, &, Review Department, and International Monetary Fund. Research Dept.