A comparison : US and Europe repo market

(1)

A comparison: US and Europe repo market

Master Thesis

Nadia Seevathian

10874526

Thesis supervisor: Dr. R. (Rafael) Almeida da Matta

Master in International Finance, Amsterdam Business School

29 August, 2016

(2)

1 Abstract

This thesis provides a descriptive and quantitative comparison of the two largest Repurchase agreement market in the world, US and Europe. A major obstacle to the existing literature is the nonexistence of a detailed and comprehensive dataset at trade level. Not surprisingly, many of the recent contributions are based on a novel, but often market-specific (mostly the US) dataset. Therefore, only a limited number of studies focus on the European market and little is known about its difference and similarities with respect to the US market. This study attempts to partially fill this gap by taking an alternative approach; it uses several publicly available data sources in order to make an in-depth comparison between the US and the EU repo markets. It does so by providing an extensive description of their mechanics, market participants and legal and accounting features. The comparison shows that although repo is a global product, the country wide differences result in different usages and market patterns which in turn result in different source of systemic risk. The banking model can be seen as the back bone underlying the differences between these two countries. Europe is governed by universal banking model and as a result banks play a crucial role in the repo market, while the US which is still suffering from the post- Glass-Steagall Act, is still driven by specialized banking. A clear split between investment banks versus commercial banks exist, leading to the centricity of broker-dealers.

Local regulations such as the Volcker rule, under Dodd-Frank Wall Street Reform and Consumer Protection Act will only widen this gap further. Consequently the US is characterized by tri-party repo, while in Europe central cleared counterparty’s repo is on the rise. Furthermore legal differences and the fragmented securities settlement infrastructures within EU, make the EU cross-border repo a heterogeneous product compared to the homogeneous domestic US repo.

(3)

2

1. Introduction

Repurchase agreement (Repo) is viewed as the foundation stone of the financial market. First it is a key source of short-term funding and the market is growing compared to the more risky alternative, the unsecured deposit market, acts as the primary channel through which central bank target bank reserves and transmit monetary policy and last due to the collateral component it is critical for secondary market liquidity in treasuries and other securities, and hence plays an important role in the pricing and price discovery of cash and derivatives instruments. The lack of a full dataset, netting and accounting treatment makes it difficult to get an accurate measure of the total size but several estimates exist. According to ICMA survey1_{December 2015 the}

estimated volume of repos and reverse repos in Europe (EU) amounts to about Eur2.75 trillion2_.

While PWC and BNY Mellon 2015 estimated the United States (US) market size at $3.7 trillion ($1.6 trillion in tri-party as of June 2015 and $1.9 trillion as per October 2014 based on primary dealer volumes) an estimate based on Finadium research. Compared to the GDP of EU Eur10.4 and US $17.9 trillion the market is significant.

Repo has been in the limelight since the financial crisis. The product and its usage by market participants is not well understood and hence viewed as an obscure financial instrument. The failure of the 164-year old Lehman Brothers Holdings on September 15, 2008 because of heavy reliance on short-term funding and the subsequent misuse of a specific US GAAP provisions by Lehman via the usage of ‘Repo 105’ to move 50 billion of assets off its balance sheets, was key in drawing attention on the potential issues with repo as a financial product. Additionally the famous paper by Gorton and Metrick (2009) which shows that the financial crisis was the result of a repo ‘run’. Although this has been refuted by many other researches such as Krishnamurthy, Nagel, Orlov (2012) who showed that ABCP played a greater role than repo. All these papers contributed to the increased focus on repo.

Interestingly in terms of financial crisis impact the US repo market got a serious a blow. Not only has the US repo market shrunk quite a bit from around $4.5 trillion in the first quarter of 2008 to $3.7 but the tri-party repo segment also get to get a serious make-over, where the clearing banks were obliged to reduce intraday credit by 90% as per the Triparty reform

recommendations paper published by the Payment Risk Committee. While EU only experienced

1

http://www.icmagroup.org/Regulatory-Policy-and-Market-Practice/short-term-markets/Repo-Markets/repo/latest/

2_{ICMA Dec 2015 reports a market size of Eur 5.5 trillion referring to both lending plus borrowing} positions.

(5)

4

the spill over effect of the US repo market. Whereby the loss of confidence caused the financing conditions to tighten and demand for high rated and liquid government bonds to increase. This difference in impact prompted the central question of this work; what is the difference between these two repo markets.

Furthermore the comparison of these two markets has never been more relevant than now. As a result of the financial crisis financial regulators realised how little is known about the repo market and the possible implications it might have for financial stability. Since then there has been a raft of regulations, to name a few, Basel III, Dodd-Frank Wall Street Reform and Consumer Protection Act, European Market Infrastructure Regulation (EMIR) and Markets in Financial Instruments Regulations (MiFIR). These are transforming the financial market as we know it now and there are a lot of speculations at this stage what the end result will be for US and EU. Will the US repo market move to a more European model? Can the US market participants learn from EU or the other way around? This comparison establish a good basis to further analyse the future consequences.

As it is a very broad topic we first identify the key dimensions we want to compare. Within each dimension we select the core aspects we want to focus on. The method used is to research a wide array of academic journals, surveys, regulatory publications and documents, accounting and legal journals and articles and other researches to find the differences and the reasons. This analysis makes use of solely publicly available data, which has been researched and selected per dimension to either support or to identify the differences.

The comparison begins with the legal dimension. Documentations are scarce on this topic. The Repo Master agreement, Global master agreement, Supplemental notes and legal articles for interpretations of the MRAs have been studied. While under the English law the legal definition for repo, is a sale and repurchase contract, the New York law does not have a legal definition but created provisions to exempt it from the automatic stay under the US Bankruptcy Code. This has implication in case of bankruptcy which is in turn governed by the law of the insolvent party’s jurisdiction rather than the governing law of the agreement. Consequently this implies that EU cross-border repo is a complex, expensive and heterogeneous, even when with similar term, product. While in the US domestic repo, with same term, would be homogeneous.

To compare the accounting dimension, IAS 39 for EU and SFAS 140 for US have been studied. EU accounting treatment reflects the economic, rather than the legal nature. While for the US since there is not really a legal nature defined, it just reflect the economic nature. The definition of when a ‘transfers’ has actually taken place is different. For IAS 39 it is based on ‘risks and rewards’ and control’ while for SFAS is based on ‘control’. Despite definition

(6)

5

differences repo is accounted in almost the same way under IAS and SFAS. As IAS is principle based while SFAS is rule based, the language used and application guidance tend to be very specific under SFAS. This might be a reason why SFAS tend to be subject to loopholes as compared to IAS.

Who are the main market participants and why, has been researched through a wide range of papers, including the history of banking in the US and EU. As for the data on EU side we researched a wide range of regulatory reports submitted by banks to the ECB to find out that the Dutch ‘Socio-Economische’ report data could be used to get an insight on banks repo

counterparty sector. The US repo market is dominated by broker-dealers, while in EU banks have this central role. The study of the repo market participants highlight the stride between bank and non-bank in the US. Where we see that EU is characterised by universal banking model, while the US still suffering from the post- Glass-Steagall Act, characterised by specialised banking. Additionally broker-dealers are insured by SIPA, indemnification per investor is up to $500,000 in case of broker-dealers’ failure compared to banks which are insured by FIDC up to $100,000 (as from 2008 $250,000).

Broker-dealers kept the price of repo very attractive by outsourcing the collateral management to a clearing bank and as a result created a safe and lucrative new deposit-like call – tri-party repo for the risk-averse return seeker shadow banking investors. This is why tri-party repo is the main mode of trading repo in the US. While the CCP is a pure interdealer platform that is just

complementing the tri-party market. Compared to EU where banks rule the financial market, balance sheet optimization is the ultimate goal, repo traded via CCP outweigh tri-party.

While repo publicly available data is in general a constraint, in Europe it is dire. ICMA semi-annual surveys are only published in pdf format. To analyse the time series the 30 pdf surveys had to been studied.

When compared with its alternative unsecured borrowing the European Money Market survey data show signs of complementary usage of repo versus unsecured funding despite the rise repo and falling unsecured funding trend. While US commercial bank data shows an opposite trend with a rising unsecured borrowing trend. Last Basel III will change the banking model, reliance on short-term funding will fall and repo duration will rise.

The remainder of this paper is structured as follows: section (2) is the literature review section (3) provide a high level definition of the repo market to get started, section (4) we explain the data being used in this document, section (5) is the comparison which is further split in the following subsections; subsection (5.1) looks at the legal differences between US and EU repo, subsection (5.2) we analyse the accounting differences between US and EU, subsection (5.3) map

(7)

6

the market participants in US and EU, subsection (5.4) unfold the tri-party repo market for both US and EU, subsection (5.5) unfold the centrally cleared counterparty repo market for both US and EU, subsection (5.6) we take a brief look at the bilateral market, subsection (5.7) is a quantitative analysis of the repo rates in both US and EU, subsection (5.8) we summarises the academic papers on repo market vulnerabilities and subsection (5.9) we look at the impact of regulations, we end with section (6) assessment and recommendations.

2. Literature Review

The majority of researches on repo tend to be on an analysis of the financial crisis, vulnerabilities of the repo market and the market structure such as Gorton and Metrick 2012, Copeland Adam et al 2012, Krishnamurthy, Nagel, and Orlov 2012, Baklanova Viktoria et al 2015, Loriano Mancini et al 2014, Charles. The closest in terms of comparison is the work of Hördahl and King 2008 which analyse the financial turmoil of 2007-2008 by looking at repo rates for US, EU and UK. No academic journals could be found that focus purely on the differences between the US and Europe repo market. High level information is scattered in a handful of journals such as Charles Boissel et al 2014, Loriano Mancini et al 2014, Brunnermeier Markus K 2009 and several ICMA researches. Information is usually in the form of a brief few liner to put into perspective some other aspect in the journals. On the accounting side the few journals that can be found are destined to US GAAP such as Ong Kingsley T.W. and Eugene Y.C. Yeung 2010 and Smalt W Steven and J. Marshall McComb II 2015, but include one paragraph on IFRS to distinguish with US GAAP criterion; ‘who retains the risks and rewards’ versus ‘control’. Chircop Justin et al 2012 question the current way of accounting repo under US GAAP as secured borrowing and propose an alternative way; to account repo as sales, which is basically what Lehman exactly did with Repo 105/108. On the legal area hardly any work could be found not even single sided ones. We reviewed a wide spectrum of researches, surveys and regulatory documents on both the US and Europe and to arrive at this detail comparison.

3. Repo definition

A repo transaction is structured legally as a simultaneous agreement between counterparties to engage in a sale of securities on an initial date, with a repurchase of the securities by the initial seller at a later date. In terms of definition, the transaction on the party selling the security(in the initial transaction) is termed repo agreement, while the transaction on the party buying the

(8)

7

security and is committed to deliver it at agreed future date enters a reverse repo agreement. The outer shell bears close resemblance to a secured cash loan with fundamental legal differences. A repo transaction provides significant protections to creditors, the counterparty holding the securities or also known as the cash lender, may liquidate the securities held, and accelerate or terminate the agreement. There is a full transfer of ownership (known as ‘transfer of title’) throughout the life of the transaction. However interest or dividend paid on securities sold under a repurchase agreement is rebated to the seller of the securities.

On the collateral side we have two types of agreements, “special collateral” repo is when the borrower and lender have agreed that only a single designated security is acceptable on the repo and that the borrower has no right to provide substitute securities. While in a “general

collateral,” the borrower of funds has an option to choose among a basket of possible assets, subject to some limitations in the agreement, the securities that the creditor is to receive and may also have the right to substitute securities during the term of the repo.

The market value of collateral can exceed or fall behind the value of the loan. Typically a collateral value greater than the loan value would be preferred. The amount by which the loan is over-collateralized or under-collateralized is expressed in the initial margin or haircut. A haircut is a percentage discount deducted from the market value of a security that is being offered as collateral in order to calculate the Purchase Price. Initial margin is a percentage premium added to the market value of a security that is being offered as collateral in a repo in order to calculate the Purchase Price (ICMA: Haircuts and initial margins in the repo market 2012). Very often we see that both margins and haircuts being used interchangeably in documentations, while they quite distinct. Haircuts indicate how much money was lent relative to the asset value of the collateral at trade start, while margin refers to maintaining the value of collateral should market prices adversely change after the contract is signed, this occur during the life of the trade. Example of a bilateral repo - Assuming a 5 days duration repo.

(9)

8

Attribute Formula/Definition Amount(Eur)/%

Initial sale proceeds Analogous to principal disbursed at start 10.000.000 Market value of

collateral Which is equivalent to Securities in our diagram 11.000.000 Repurchase price Analogous to principal paid back at maturity 10.002.778 Repo rate

(Principal at maturity – principal at start)/principal at

start*100 *360/5 2%

Haircut (1 - principal at start )/ collateral market value 9% Initial Margin Market value of collateral/purchase price *100 110% Additionally two broad classes of repurchase agreements, based on differences in settlement are recognised – bilateral and tri-party agreements. A triparty repo involves a third party, which is a clearing bank. The clearing bank provides back-office support to both parties in the trade, by settling the repo on its books and ensuring that the details of the repo agreement are met. In contrast, in a bilateral repo, each counterparty’s custodian bank is responsible for the clearing and settlement of the trade. Bilateral repos can be centrally cleared counterparty or non-CCP.

4. Data

This work makes use of solely publicly available data. The advantages is that this type of data is readily available and normally comes from reliable sources. Unfortunately the flip side of the coin is the data lack granularity and hence restrain detail analysis. The sources of data are described below.

US

Financial Accounts of the United States – Is a quarterly publication by the Federal Reserve, providing statistics on the US flow of funds, balance sheets and integrated macroeconomic accounts. Data is available in pdf and csv layout. We make use of this data to study the volume of repo per sector and repo funding in comparison with interbank funding for the US.

Fed tri-party and GCF data - All the primary dealers are obligated to report their repo position via Form FR2004. This data is then summarised and published on a monthly basis on

(10)

9

the Fed site3_{. We make use of this data to study the volume, collateral breakdown of tri-party and}

GCF repo.

SIFMA data - The Securities Industry and Financial Markets Association (SIFMA) compile data from various sources and publish those on their websites for informational purposes only. They also publish tri-party and GCF repo data. Their data source for tri-party data is the Federal Reserve Bank of New York and GCF repo is from the Depository Trust & Clearing Corporation (DTCC). All the historical data are published in a convenient layout in excel.We make use of this data to study the volume, collateral breakdown, maturity breakdown of tri-party and GCF repo.

Depository Trust & Clearing Corporation (DTCC) – We make use of the Treasury GCF repo rates under section 5.7 to compare US and EU monetary policy impact on repo rates.

FED - Effective Federal Fund rate (EFFR) is calculated as a volume-weighted median of overnight federal funds transactions. This rate is used under section 5.7 Repo rates to compare US and EU monetary policy impact on repo rates.

Ycharts website – We make use of overnight USD Libor in section 5.7 Repo rates as a reference rate for unsecured interbank deposits.

Europe

ICMA survey - International Capital Market Association (ICMA) conduct a semi-annual survey, which is send to several financial institutions (72 in Dec 2015), mainly banks( 68 out of 72 in Dec 2015). The list of questions ask the financial institution to disclose their repo position and its breakdown on a particular date. Financial institutions disclose this information out of goodwill. The survey begun in 2001, but there are only a few big organisations that consistently responded for all years. Since it is not a mandatory reporting requirement one can question the quality of data gathered. Like in December 2015 there was a mention on improvement in tri-party repo volume, as banks have been including security borrowing and lending transactions under the tri-party repo volume. Despite the data quality issue the high level trend should be reliable and it is a data source that is being widely used. The data is published only in PDF version, we sometime inputted the data manually in excel or use the graphs from ICMA. We make use of this data for several variables, like for the repo volume, maturity break down, collateral breakdown and so on.

(11)

10

EMM survey - Euro money market survey is produced by the European System of Central Banks (ESCB) on an annual basis. In Dec 2015 149 banks all throughout Europe participated in the survey. Reported data was based on the quarterly turnover during the second quarter of 2015 in each of the money market segments. Compared to ICMA survey which is a stock based data, the ECB money market survey might provide better estimates as it is a flow based data. Data is available in both pdf and csv. We make use of this data to compare the size of the unsecured and secured short-term funding for Europe and the breakdown of bilateral, tri-party and CCP shares of the total repo market.

EMMI website - European Money Markets institute publishes current and historical rates for Eurepo. Eurepo is define as ‘the rate at which, at 11.00 am Brussels time, one bank offers, in the euro-zone and worldwide, funds in euro to another bank if in exchange the former receives from the latter the best collateral within the most actively traded European repo market. Range of maturities include tomorrow-next, up to one month, one month and beyond. The Eurepo rate is used in section 5.7 Repo rates to analyse the movements in Europe repo rates as compared to the US. Additionally Euro Overnight Interbank Average (EONIA) rates is used in section 5.7 Repo rates as a reference rate for unsecured interbank deposits. EONIA is calculated each business day in the Eurozone by averaging the overnight interbank deposit rates at which the banks in the EURIBOR panel have traded between midnight and 18:00 that day, with each rate weighted by the total volume of business done at that rate by the panel banks. Because it is the average price of overnight interbank deposits, EONIA is virtually a risk-free rate of return.

European Central Bank – we make use of the ECB deposit rates, the Main Refinancing Operations (MROs) rates, Overnight deposit rate to corporations and loan over 1M to Corporation 3months Floating rate under section 5.7 Repo rates to compare US and EU monetary policy impact on repo rates.

Monetary statistics Dutch Central bank – Within the Nederland all Monetary financial institutions(MFI) are obliged to submit several different reports on a monthly, quarterly, half-yearly and half-yearly basis to the Dutch Central Bank. This information is gathered on behalf of the European Central Bank. Additionally the Dutch Central Bank uses this information for country risk, overview of the OTC derivative market and for the quarterly sector accounts. A

consolidated data sets is also published on the Dutch Central Bank website.

We make use of this data to get an insight on the sector of the repo counterparty interacting with MFIs in the Nederland as a representation of Europe since this data could not be found at Europe level. World Bank and Focus – We get the macro economic variables data for both US and EU for section 5.7 Repo rates.

(12)

11

5. Comparison

5.1 US and EU repo legal differences

In most jurisdictions, some form of legal documentation is used to establish the terms and conditions of repurchase agreements. This legal documentation often comes in the form of a Master Repurchase Agreement or Global Master Repurchase Agreement which defines the terms or is a model legal agreement designed for parties transacting repos. Most US repo dealers use the Master Repurchase Agreement (MRA), governed by New York state law, for domestic US counterparties, and the Global Master Repurchase Agreement (GMRA)4_{, governed by English}

law, with international counterparties. Each document was developed in consideration principally of the market practices and legal developments in New York and London,

respectively. The GMRA is widely used however it is also worth noting that in Europe there are other standard master agreements as well which are used in domestic repo markets such as Italy, France and Germany. The European Master Agreement (EMA) attempts to replace the domestic master agreements in the euro area, but its use remains limited. To further complicate matters tri-party and central counterparties will have their own master agreement as well. For this section we will look at MRA and GMRA.

MRA is stated to be governed by New York law and GMRA is stated to be governed by English law. The law or laws of the jurisdiction in which a specific proceeding is maintained will generally govern, including its choice of law rules. Applicable laws relating to the transfer of property interests in the repo assets and applicable bankruptcy laws. The fact that in insolvency, the laws of the insolvent party’s jurisdiction prevail over the governing law of the agreement is an aspect that create a legal difference of primordial importance. We take a look at the impact:

Before 1982, under the New York law, if a repo participant would become insolvent the bankruptcy courts were forced to classify the transactions as either sales or loans. However the bankruptcy of Wall was a turning point. In September 1982 in the case of Lombard-Wall the Federal Bankruptcy Judge Edward Ryan ruled that the open repo transactions held by that dealer were loans. This created great uncertainty and upheaval in the securities market,

4_{For more details on the layout of MRA and GMRA and Annexes please refer to}

(13)

12

because this meant that the repo buyer losses all rights on the collateral held. As a result market participants lobbied for a change in the bankruptcy law which was adopted by Congress in 1984. As a result of this change it would mean that repo is exempted from certain provisions of the US Bankruptcy Code that normally apply to pledges, in particular, the automatic stay on

enforcement of collateral in the event of insolvency. This guarantees that a repo buyer would be able to liquidate securities involved in repo transactions and prohibit bankruptcy trustees from unilaterally forcing repo buyers to turn over underlying securities. Unfortunately the question of whether a repo transaction is a loan or a contract for a sale and a subsequent repurchase remain unanswered under the US Bankruptcy law.

As a result to further protect the Buyer in the scenario that a repo transactions get re-characterized as a secured loan contrary to the intention of the parties, the MRA contains paragraph 6 “Security Interest” and several other provisions. Paragraph 6 of MRA – “ Although the parties intend that all Transactions hereunder be sales and purchases and not loans, in the event any such Transactions are deemed to be loans, Seller shall be deemed to have pledged to Buyer as security for the performance by Seller of its obligations under each such Transaction, and shall be deemed to have granted to the Buyer a security interest in, all of the Purchased Securities with respect to all Transactions hereunder and all Income thereon and other proceeds thereof”

Under the English law the wording used in the documentation and the intention of the parties as to the structure selected is important in labelling a repo as a sale/repurchase versus secured loan transactions. In order for repo transactions to take effect as sale/repurchase transactions under English Law there must be an absolute transfer of title of the securities from the Seller to the Buyer. An absolute transfer is recognised by the following wording and intention:

a. Buyer’s obligation to resell securities to the seller should be an obligation to resell securities equivalent to the securities purchased rather than the actual securities which were originally purchased by the Buyer. So the word ‘equivalent’ is very important and is opposite to a secured loan arrangement where the borrower has the right to take back the assets charged once the loan has been repaid. MRA/GRMA comparison:

Paragraph 1(a) of MRA – “From time to time the parties hereto may enter into transactions in which one party (“Seller”) agrees to transfer to the other (“Buyer”) securities or other assets (“Securities”) against the transfer of funds by Buyer, with a simultaneous agreement by Buyer to transfer to Seller such Securities at a date certain or on demand, against the transfer of funds by Seller. Each such transaction shall be referred to herein as a “Transaction”...”

(14)

13

Paragraph 1 of GMRA – “From time to time the parties hereto may enter into transactions in which one party, acting through a Designated Office, (“Seller”) agrees to sell to the other, acting through a Designated Office, (“Buyer”) securities or other financial instruments (“Securities”) (subject to paragraph 1(c), other than equities and Net Paying Securities) against the payment of the purchase price by Buyer to Seller, with a simultaneous agreement by Buyer to sell to Seller Securities equivalent to such Securities at a date certain or on demand against the payment of the repurchase price by Seller to Buyer”

b. Income paid during the term of the repo should belong to the buyer. However if there is an agreement to transfer any income received from Buyer to Seller then the transfer should be based on an amount equal to income paid on equivalent securities. MRA/GRMA comparison: Paragraph 5 of MRA – “Seller shall be entitled to receive an amount equal to all Income paid or distributed on or in respect of the Securities that is not otherwise received by Seller, to the full extent it would be entitled if the Securities had not been sold to Buyer. Buyer shall, as the parties may agree with respect to any Transaction (or, in the absence of any such agreement, as Buyer shall reasonably determine in its discretion), on the date such Income is paid or distributed either (i) transfer to or credit to the account of Seller such Income…..”

Paragraph 5 (a) Income Payments of GMRA – “where: (i) the Term of a particular Transaction extends over an Income Payment Date in respect of any Securities subject to that Transaction; or (ii) an Income Payment Date in respect of any such Securities occurs after the Repurchase date but before Equivalent Securities have been delivered to Seller….the Buyer shall on the date such Income is paid by the issuer transfer to or credit to the account of Seller an amount equal to (and in the same currency as) the amount paid by the issuer”

c. In the event that following a Seller’s default or insolvency, the Buyer sells the securities on the open market for less than the contractual repurchase price, any residual claim of the Buyer against the Seller should be based on damage suffered and not by reference to the amount of unpaid indebtedness. Since the concept of recoveries being limited to the amount “lent” is one which is consistent with a loan transaction and would not normally be applicable to a sale/repurchase transaction.

(15)

14

Wording all alone is not sufficient under English law, the intention is equally important. If for instance transactions are documented in one way but is implemented in a way inconsistent with the documentation then the courts may disregard the label and look at the features of the implementation to determine the true intention of the parties.

The above extracts shows a clear difference in the wording of the MRA and GMRA. In MRA the word ‘Equivalent’ is completely absent, same for the word ‘Loan’ in the GMRA. For this reason MRA deriving from New York law is generally not suited for the London market as it is worded in ways which involves a danger for repo transactions to take effect as secured loan transactions rather than sale/repurchase transactions. Hence under the English law the wording of the GMRA consistent with the intention fully guarantees that a repo be classified as

sale/repurchase while under the New York Law the bankruptcy exemption clause safeguard the interest of the repo buyer. Basically different route to the same end result.

In EU the fact that the laws of the jurisdiction or choice of law prevails in bankruptcy or transfer of property interest makes cross border repo a legally cumbersome product compared to the US. EU is not one state with one law as the US. Each of the 19 EU members using the euro currency have its own legal systems, hence transfer of property interests, bankruptcy laws and procedures are not fully harmonized. That is also the reason for the coexistence of the many different master agreements and why the European Master Agreement is poorly used. Conflicts of law in cross-border collateral arrangements is an additional aspect to consider. For example the collateral taker might be dealing with a counterparty that is incorporated in another

jurisdiction additionally the collateral securities might be legally recorded in another jurisdiction, or even in a range of different jurisdiction. Hence ambiguity might arise which part of the collateral arrangement is covered by which legal system. The legal differences result in repo heterogeneity and as a result the law of one price does not hold for cross-border EU repo market with the same term, compared to US where it does.

5.2 US and EU accounting treatment

Listed companies in Europe are required to adopt the International Financial Reporting Standards (IFRS), while in the US adoption of the US Financial Accounting Standards Board’s Statement of Financial Accounting Standards (SFAS) is mandatory. The relevant accounting standard for Repurchase Agreement is Statement of Financial Accounting Standards (SFAS) 140 Accounting for Transfers and Servicing of Financial Assets and Extinguishments of Liabilities

(16)

15

and SFAS 166(an amendment on SFAS 140), under US GAAP and IAS 395_Financial

Instruments: Recognition and Measurement, under IFRS. Both accounting standards provide a guideline whether a repo should be accounted for as sales or secured borrowings. Under both IAS 39 and SFAS 140 repo transactions are accounted for as secured borrowing while we have seen under the legal form repo is being categorised as a sale/repurchase. This is where repo earns its reputation as a controversial product as we will see there is a clear inconsistency between the legal and the accounting form.

Under IAS 39 when an entity transfers an asset (transferor) an evaluation should take place whether the transferor should continue to recognise this transferred asset in his book or not. This evaluation is based on two concepts: how much risks and rewards of the ownership of the financial asset is retained/transferred or how much control is retained/transferred. The concept of control is only secondary, only in case there is no substantial risks and rewards

transferred/retained. If the transferor transfers substantially all risks and rewards of the ownership or has not retained control then the transferor can derecognise the financial asset in his books. In the repo context derecognition of the financial asset would mean accounting repo as a ‘sales’ while continuing to recognise the asset would mean accounting repo as a ‘secured borrowing’. To be able to understand what this means for repo we need to look into the definition of transfer

risks and rewards and retained control.

IAS 39 – 21 - “An entity has retained substantially all the risks and rewards of ownership of a financial asset if its exposure to the variability in the present value of the future net cash flows from the financial asset does not change significantly as a result of the transfer. An entity has transferred substantially all the risks and rewards of ownership of a financial asset if its exposure to such variability is no longer significant in relation to the total variability in the present value of the future net cash flows associated with the financial asset”

A common problem with IFRS is the usage of words such as ‘substantially’, ‘reliably’, ‘unlikely’ and so on which are nowhere defined. However under the Application guidance a clear example is given. An example of a transfer with substantially all risk and rewards of ownership transferred would be a sale of a financial asset with an option to repurchase the asset at fair value on

repurchase date. While the conventional repo which is a sale and repurchase transaction with fixed repurchase price plus lender’s return is categorise as one where the transferor retained

5_{IAS 39 should be replaced by IFRS9 before Jan 2018, but for this thesis we make reference to IAS39} as it is the current active rule.

(17)

16

substantially all the risks and rewards of ownership. Hence explaining why the transferor (Seller) should keep/continue to recognise the security on his balance sheet and repo transaction booked as a secured borrowing. The legal absolute transfer as per previous section 5.1 is not the same as the accounting true sale that all risk and rewards and control associated with ownership, are transferred.

Under SFAS 140 the main evaluation criteria for assessing whether a transferred asset should be recognised or not in the book of the transferor is based solely on control. If control has been surrendered by the transferor (Seller) then the asset can be derecognised or vice versa. The definition of control under SFAS 140 is very different than the definition under IAS 39. Under IAS 39 it is about the practical ability of the transferee to sell the transferred asset. That is how freely or independently without attaching restrictive conditions on the transfer can the transferee dispose of the asset in an active market. This would indicate the extent the transferor has

surrendered control. While under SFAS control is as follows:

SFAS 140 - 9 “A transfer of financial assets in which the transferor surrenders control over those assets is accounted for as a sale to the extent that consideration other than beneficial interests in the transferred assets is received in exchange. The transferor has surrendered control over transferred assets if and only if all of the following conditions are met:

a. The transferred assets have been isolated from the transferor—put presumptively beyond the reach of the transferor and its creditors, even in bankruptcy or other receivership. b. Each transferee (or, if the transferee is a qualifying special-purpose entity (SPE), each holder of its beneficial interests) has the right to pledge or exchange the assets (or beneficial interests) it received, and no condition both constrains the transferee (or holder) from taking advantage of its right to pledge or exchange and provides more than a trivial benefit to the transferor.

c. The transferor does not maintain effective control over the transferred assets through either (1) an agreement that both entitles and obligates the transferor to repurchase or redeem them before their maturity or (2) the ability to unilaterally cause the holder to return specific assets, other than through a clean-up call.”

One can debate that criteria (a) and (b) is applicable to repo however (c) as explained under the Implementation guide really place it in the secured borrowing category. Under SFAS 140 – 47 the fact that the transferor has an agreement that entitles and obligates him to repurchase the transferred asset is seen as effective control being maintained by the transferor. Additionally the

(18)

17

following characteristics of the repurchase are considered, the repurchased assets being the same or substantially the same, based on an agreed fixed terms, repurchase before the maturity date and the repurchased being agreed at the same time as the transfer. SFAS 140-47 even explains what “substantially the same” means that is same issuer, same maturity, identical contractual interest rates, same risks and rights and same aggregated principal amount. All these

characteristics show that the transferor has not surrendered effective control and because repo as a transaction doesn’t meet all the 3 key criteria (a), (b) and (c) above it should be accounted for as secured borrowing under SFAS 140. As we saw in section 5.1 repo is not categorised as sales or secured borrowing under New York Law, hence in a way there is not conflict between the legal form, which is non-existent, and accounting definition when it comes to SFAS as compared to IAS.

Additionally if we compare the wording of SFAS to IAS, it is noticeable that SFAS is more specific or precise in its wording and guidance and this precision might be the reason why SFAS tend to be subject to loopholes. Only to illustrate the phrase ‘an agreement that both entitles and obligates the transferor to repurchase or redeem them before their maturity’. As this sentence says to ‘repurchase before maturity’, it is obvious why market participants would misinterpret and account repo-to-maturity as a sales. Repo-to-maturity is defined as a repurchase agreement where the settlement date of the repurchase leg is equal to the maturity date of the financial asset, hence the transferor would not need to reacquire the financial asset as the asset would just mature on the settlement date. The Financial Accounting Standard Board (FASB) issued on June 12, 2014 an Accounting Standards update, Transfers and Servicing (Topic 860) to provide clarity on the accounting treatment for repo-to-maturity, should be accounted as secured borrowing.

The difference between accounting as ‘sales’ versus ‘secured borrowing’ is depicted via the below example: For simplicity it is assumed that the securities are in the available for sales book. The Seller starts with an initial security position of $9.985.000.

If repo is accounting for as sales then the security is credited or removed completely from the book of the Seller and is recognised in the book of the Buyer instead. The only position the Seller is left with is a cash position of $10.000.000, which is the sales proceeds and a realised

(19)

18

profit of $15.000 since the Seller originally bought the security for $9.985.000 but sold it for $10.000.000.

Despite the difference in definitions under IAS 39 and SFAS 140 the end accounting entries for secured borrowing looks very much the same under both regime.

Secured borrowing accounting is very different from sales. Under secured borrowing the Seller record a liability of $10.000.000, which get add up in the leverage ratio, and the original security remain in the position of the Seller but disclosed as pledged assets. Additionally the interest payable is accrued and recognised in the profit and loss. While the Buyer record an asset for the money lent out but the securities is not recognised by the buyer.

(20)

19

When looking at position level we see why Lehman would have preferred to report repo 105/108 as sales because then they could get rid of the liability and security position which would otherwise be on the balance sheet under secured borrowing.

5.3 US and EU Market participants

The microstructure of both markets is quite different. What come through all documentations is that the majority of Euro repo transactions are conducted in the interbank market, reflecting the dominating role of banks in the European financial sector. While in the United States Securities dealers play a central role and most repo transactions are part of the shadow banking system6_.

Figure 1 – US market participants

Source: Combined diagram of Baklanova Viktoria et al 2016 and Loriano Mancini, Angelo Ranaldo, Jan Wrampelmeyer 2014

Figure 1 shows the market participants landscape of both US and EU side by side. As depicted on the US market, security dealers are key intermediaries between cash lenders and borrowers. A securities dealers encompasses businesses engaged in dealing with securities or other financial

6_{ICMA defines Shadow banking as traditional banking activity conducted by non-banks. This bank-like} activity falls partially or entirely outside the scope of prudential capital and liquidity regulation and beyond the safety nets provided by deposit insurance or lenders of last resort.

(21)

20

instruments, Investment banks and broker-dealers. Security dealers can be divided into primary dealers and non-primary dealers. Primary dealers7_{are banks or securities broker-dealers that}

serve as direct trading counterparties of the New York Fed in its implementation of monetary policy. They are required to participate in all auctions of US government debt and to make reasonable markets for the New York Fed when it transacts on behalf of its foreign official account-holders. Data on non-primary dealers is unknown, who they are exactly. It is uncertain how big that market segment is. Several estimates have been made in different papers.

According to the Federal Reserve Bank of New York (2010) the top 10 dealers account for 86% of the collateral posted, while a recent survey by Baklanova Viktoria et al 2015 assumed that primary dealers account for 80% of the total tri-party repo activity and up to 95% of the total repo market.

On the Euro landscape the interbank repo segment play a central role. Interbank include banks and financial institutions. The main forms of trading in the interbank repo market being bilateral non-CCP-based, bilateral CCP-based and triparty. We will look at each form in turn in later sections for this document.

The financing segment is composed of cash borrowers such as banks, broker-dealers and hedge funds (left hand side of the diagram), borrowing from “cash-rich” entities/lenders like central banks, retail banks, money market funds (MMFs) and securities lenders. It is a market mainly driven by short-term financing, however according to FSB (2012a) there is also a growing long-term element in the European market.

The question is why security dealers is central in the US as compared to Europe. One possible answer lies in Universal banking model; which permits commercial banks to conduct investment banking activities. Europe has a long history in universal banking, according to Richard Tilly (1998) Belgium might have been the pioneer back in 1822. This might explain the strong

presence and the central role of banks in Europe. While in the US the Glass-Steagall Act in 1933 dismantle universal banking and forbade commercial banks from dealing in securities and investment banks from taking deposits in the US on the ground that trading activities are inherently more risky than traditional banking. The act was repeal in 1999 but by then well-established security dealers had already found a niche and gain a significant market share. This might explain why security dealers played such a central role originally. However the fact that they still kept their key role may be attributed to the competitive prices they have been offering

7_{For a list of the primary dealers please visit:}

(22)

21

for repos and hence have managed to lure a great many investors, consequently made repo their new ‘deposit’. The stride between bank and non-bank has been a constant factor in the US banking history. In a similar way MMMF rose from obscurity to prominence in the 1970’s, given the interest-rate regulations banks were unable to offer the same return as the MMMF’s

instrument as a result MMMF market share rose from almost zero to 36% from 1970’s to 1980’s. We take a look at the statistics of the main players in the US market on the asset and liability side. The data is from the flow of funds8_{, we plot the average of 2010 – 2015 for both assets and}

liability. We see indeed that security brokers and dealers dominate the market on both asset and liability side. On the asset side Money market mutual funds account for 16% of the cash supply on average.

Figure 2 – US repo asset and liability investment breakdown per market participants

Data source: Financial Accounts of the United States (also known as flow of funds)

Another reason why broker-dealers might be convenient from the client perspective has to do with the risk-averseness nature of repo as a product. According to Financial Stability Oversight Council (FSOC) Annual report 2011, institutional investors, which in 2011 account for about

(23)

22

two-thirds of assets under management in MMFs, exhibit extreme aversion to absorbing even small losses. As a result I believe Securities Investor Protection Act (SIPA) is a major reason why brokers-dealers are preferred by cash-rich entities. SIPA was enacted in 1970. Within SIPA Congress established the securities Investor Protection Corporation (SIPC), a private corporation similar in purpose to the Federal Deposit Insurance Corporation (FDIC). SIPC compensates investors who suffer losses due to the bankruptcy of securities dealers. After the SIPA trustee uses a number of methods to maximize the number of securities and the cash in the customer property fund, he then distributes the contents of that fund to customers. If the securities in the fund are insufficient to meet customer claims, the trustee allocates the securities available on a pro-rata basis and indemnifies each customer for up to an additional $500,000 worth of securities and cash. SIPA does not indemnify customers for any loss in the market value of securities that occurs after the filing date. While FDIC guarantees bank deposits up to a maximum of $250,000(before 2008 limit was $100,000 only) per depositor over the sum of all savings/deposits account per insured bank, for each account ownership. Hence another good reason to prefer repo with broker-dealers over bank deposits.

Publicly available data on the sector of the market participants in Europe repo market is quite hard to find. Both ICMA and EMM do not gather this data. The majority of repos in Europe takes place in the interbank market Bakk-Simon et al (2012), however we also want to know if there are also repos conducted with counterparties other than banks and who are they. On the ECB website there are aggregated data on sector accounts9_{however information on the}

counterparty sector of repos trades is not available. As a result we take an alternative route. All banks within the Netherlands submit monthly and quarterly reports to the Dutch Central bank (DNB) for Social and Economic reporting. The DNB publish part of this data on their website10_{. According to the Social Economic reporting manual}11_{buy/sell back is also included}

under the category Repo. Since buy/sell back is also a type of repo, only difference is that the sale and repurchase leg are two separate contracts instead of one, this should not matter.

Furthermore unlike the US Money market funds is also included under MFI, making comparison a bit complicated. However according to Bakk-Simon et al (2012) MMFs balance sheets

represented only 4% of the balance sheets of the MFIs in the euro area, hence should have a 9_{http://www.ecb.europa.eu/stats/prices/acc/html/index.en.html} 10 http://www.dnb.nl/statistiek/statistieken-dnb/financiele-instellingen/banken/binnenlands-bankbedrijf-monetair/index.jsp 11 http://www.dnb.nl/binaries/Handboek_Sociaal-Economische_Rapportages_2015_(09-07-2015)_tcm46-324219.pdf

(24)

23

minimal impact. We make use of this data to get a feel of the counterparty sector. We plot the data from table T5.2.9 for MFIs repo counterparty break down in Europe and outside of Europe.

Figure 3 – MFIs in the Netherlands repo transactions counterparty sector

Data source: Dutch Central Bank

In the Netherlands we see completely zero repo activity between MFIs, SPVs and Non-financial public companies, some repo activity between MFIs, Insurance and Pension funds and Non-financial private companies but this is very minimal compared with Other Financial Institutions (OFI)12_{. However the category OFIs include CCPs and hence would include the interbank repos}

transacted with the CCPs. As pointed out by Bakk-Simon et al (2012) most of the shadow banking activities takes place under the grouping other financial activities, excluding the CCPs activity. The interconnection between shadow banking and regulated banking system takes place via the funding that MFIs take from OFIs. As these funding is mainly short-term it might be susceptible to runs, after excluding the CCPs part out of the OFI this amount would be much lower. Naturally we cannot extrapolate and generalise the Dutch data over the whole of Europe. According to Bakk-Simon et al (2012) Luxembourg has the biggest OFIs sector, around 50% of the non-MFIs deposits are from OFIs, Spain 40%, Netherland, Belgium, Portugal, France, Ireland, Italy all hover around 20-25% and Germany the least with 10%. We can conclude that for Europe besides the interbank repo segment, OFIs are the second supplier of funds via repo transactions.

12_{OFI sector comprises all financial institutions other than those included in the sectors monetary} financial institutions (MFIs) and the insurance corporations and pension funds.

(25)

24 5.4 Triparty

The key difference between European and US tri-party repo lies in the role of the tri-party agent. Comparison is not straight forward because of the progressive reform taking place in the US tri-party system since the 2008-09 financial crisis.

Market structure

In both US and Europe the main difference between tri-party and bilateral repo is that there is a third party, an agent, in between the buyer and the seller. The agent perform the collateral management on behalf of the counterparties. It does not change the relationship between the two parties and doesn’t participate in the risk of transactions.

In Europe the agent will automatically select, from the securities account of the seller sufficient collateral that satisfies the credit and liquidity criteria and concentration limits pre-set by the buyer. Subsequently, the tri-party agent manages the regular revaluation of the collateral, margining, income payments on the collateral, as well as (in the case of most European tri-party agents) substitution of any collateral which ceases to conform to the quality criteria of the buyer, substitution to prevent an income payment triggering a tax event and substitution at the request of the seller. The buyer is allowed to re-use the collateral, for which he is the legal owner, in other transactions within the tri-party system. If the buyer were to re-use the collateral outside the tri-party system, the transaction would cease to be a tri-party repo and would no longer be maintained by the agent.

To describe the US side we would like to first look at the transition this market has been since the financial crisis. Before the financial crisis the US tri-party repos were characterised by daily unwind. Unwinding consists of sending the cash balances back to the investor’s balances account and the securities back to the collateral provider’s securities account on the balance sheet of the clearing bank. At the same time, the clearing banks extend intraday credit to the dealer since the securities are no longer financed by the tri-party investors. While in Europe, the need to unwind tri-party repos daily has been avoided by the use of direct substitution and margining. The huge intraday credit exposures of the US tri-party agents raised systemic risk concerns. As a result FRBNY published a white paper on May 17, 2010 over Tri-party Repo Infrastructure Reform with 16 recommendations to strengthen the tri-party repo market and lower systemic risk. The key recommendations would be to reduce reliance on intraday credit and the risk of asset fire sales.

(26)

25

The timing of events prior to the proposed reforms: Morning: Trade agreement

Before 10 am - New trades are conducted – Trade agreed upon in the morning does not settled until the afternoon, around 5 PM.

Afternoon: Collateral allocation

In the afternoon, after the close of Fedwire Securities Service(at 3.30PM) and Depository Trust & Clearing Corporation(DTCC)(at 4.30PM), the dealers can allocate collateral to each trade using clearing-bank provided optimization tools and collateral eligibility criteria of the buyer.

Overnight lockup:

Overnight from 6.30PM – 8.30AM, the clearing banks locked collateral in cash investor accounts, prohibiting dealer access.

Next morning: The “unwind”

Between 8 and 8.30 AM the next morning, the clearing banks “unwind” the tri-party repo trades. All repos are unwound, including terms repos (non-maturing) and open repos that are rolled over, because the unwind is mainly done for operational ease. Term trades are “rewound” every evening, at the same time as the initial leg of new repos are settled.

Tri-party repo is considered “general collateral” financing, meaning that an investor may care about the class of collateral it receives, but not about the specific issue. According to PWC the Bank New York Mellon would be exposed to intraday credit for up to 9-10 hours of credit risk daily – amounting to $1.44 trillion in secured credit daily.

According to the latest update, June 24, 2015 on the FRBNY website and the PWC & BNY Mellon December 2015, the following progress has been made since the reform proposal: Three-way deal matching and auto-substitution has been implemented. Three-way deal matching helped to establish that clearing banks received identical instructions from both dealers and investors prior to settling a tri-party repo trade. With auto-substitution the US market moved closer to the EU market, this help replace the need for a full-scale unwind. The unwind for maturing tri-party repo trades has been moved to 3.30PM as a result the secured intraday exposed reduced from 10hrs to around 2.5 hrs and hence reducing the intraday credit extended to 3% of all tri-party repo collateral, which much lower than the Task Force’s original target of no more than 10%. One remaining element of the Task Force’s reform roadmap, is the full alignment of General Collateral Finance (GCF) repo settlement with the new triparty settlement process, which is not yet complete.

(27)

26 Volume

By now it is common knowledge that tri-party repo represents the most prevalent form of repo contract in the US, however by how much exactly leads to puzzlingly diverse estimates. In Figure 4 we give an overview of a few estimates. While in Europe tri-party repo account around 10% of the European repo market as per ICMA and EMM, depicted in Figure 5.

Figure 4 – Estimates of US tri-party repo market size

The total repo market is composed of a) tri-party and GCF repo b) bilateral repo. Insufficient data on the bilateral segment of the repo market makes it difficult to get a precise value for the total size of the US repo market. Consequently it is hard to know the percentage of the tri-party repo to the total market size. SIFMA survey derived it estimates based on feedback from 15 Primary dealers over a questionnaire distributed by the Bond Market Association and another survey conducted by the Risk Management Association’s member banks, where 19 banks reported their volume on 30 June. While the other three journals all made use of an estimation technique which is based on the Federal Reserve FR2004 data collected from primary dealers. Those data include both repo and securities lending activity against cash, Tri-party repo data include GCF repo as well. Many of these estimates make assumption that primary dealers repo account for up to 90% of the whole market and then estimate the size of the bilateral section. Despite the fact that SIFMA data is netted and is from 2004, there is a big difference between these two types of estimates. The SIFMA survey might point to a weakness in the way the US tri-party market size is estimated based on strong assumption that may result in an

overestimation of the market size. Unfortunately only detail trade level data can solve this mystery, so for now base on the above researches we will also assume that the tri-party repo is 50% or more of the US repo market.

(28)

27

Tri-party repo arrangement was created in the late 1970s by Salomon Brothers, a securities dealer, in cooperation with its Treasury securities clearing bank, Manufacturers Hanover in trying to solve the “double” financing costs issue (Antoine Martin and Susan McLaughlin 2015). Its usage began to grow in the mid-1980s and it became the panacea for the ‘double-dipping’ issue that arose with “hold-in-custody” (HIC)13_{repo arrangements, which was quite a scandal and}

resulted in the default of numerous dealers. Triparty repo was a great financial innovation as it would offer both the safety of delivery and the cost effectiveness of HIC repo. The broker-dealers who seems to earn a very narrow margin on repos of US government and federal agency securities would outsource the management of collateral to a triparty agents in order to save on settlement costs and to benefit from the economies of scale offered by the triparty agents. According to ICMA the size of the fee depends on the volume of business transacted, hence no wonder why bigger is better in the tri-party world. Additionally in the US both the seller and buyer share the fee paid, which make the costs even more attractively low. As compared to EU where only the Seller bears the cost of the fee. Given the low costs and investment required to start operating in the tri-party market and the attractive repo rates compared to deposit rates in the 80s, the tri-party market saw a surge of new participants such as MMFs and corporate treasurers which further increased the significance of this market segment. By that time US broker-dealers would rely on the US tri-party repo market to finance the majority of their securities inventories.

13_{HIC repos arose in the 1970s as a way for repo borrowers to avoid the inconvenience of delivering} repo collateral; instead of sending the collateral to the repo lender over Fedwire Securities, the dealer would offer to hold the collateral in custody on its books on behalf of the lender. However, the defaults of dealers like Lion Capital Group in 1984 and ESM Government Securities and Bevill, Bresler, and Schulman in 1985 caused large losses for some HIC repo lenders and made it clear that HIC repo arrangements did not adequately protect the collateral pledged to investors. The industry realized that it would be more secure and efficient for an independent third party to segregate and price collateral and to perfect a robust security interest in collateral by the lender through a segregated account in the lender’s name (Antoine Martin and Susan McLaughlin 2015).

(29)

28

Figure 5 – Europe tri-party market share of the total repo market

Diagram source: Euro money market survey September 2015

In 2003 and 2004 the European tri-party market was pretty small, however the size has remain more or less the same from 2005 to 2015, which is quite a long period. This might indicate that there is structural barriers limiting it’s grow or that it is a very specific market segment. The first reason is the strong presence of banks which also provide custodial service to asset management firms and then dealing in repos for extra return, hence tri-party agents will be bypassed. Second looking at ICMA Dec 2015 domestic tri-party account for 36.9% while cross-border is 63.0%, out of which 46.4% with non-eurozone counterparty (before the crisis this percentage was much higher 61.6% in Dec 2005). Based on these number we look in the EU tri-party repo history. Tri-party repo was introduced in 1992 in Europe, when bond houses like Morgan Stanley, Merrill Lynch, Salomon Brothers, Lehman, and Goldman Sachs took the first step toward exporting tri-party to Europe by approaching Euroclear. The top five tri-tri-party agents in Europe are Bank of New York Mellon (BONY), Citibank, Clearstream and Citybank. In Europe the Seller bears the cost of the fee, which indicate that it is seller or liquidity seeker driven. Hence tri-party in Europe is largely between investment banks (as sellers) and commercial banks, central banks conducting investment operations, multilateral development banks and money market mutual funds (as buyers or cash investors). Some central banks allow commercial banks to use tri-party systems to manage the selection, delivery and maintenance of collateral for monetary policy transactions (eg Germany, France and Italy), this might explain the small domestic share. Third, explanation for the stagnation in market share has to do with the costs of cross-border trades. Since the repo domestic market was already concentrated with banks, the clearing banks saw opportunities to

(30)

29

enter the cross-border European market via tri-party repos. Given Europe history when each country had its own currency, central bank, stock market and hence its own central securities depository, as a result linkages are not yet well established between these national settlement systems, or between them and the ICSD. All the 19 countries within EU has its own central securities depository while in the US there would be just a few. The fragmented clearing and settlement landscape result in higher trading costs for cross-border repos. According to a research on economies of scale conducted by Schmiedel, Malkamaki, Tarkka (2006) the average unit costs of cross-border securities transfer is $40.54 compared to EU domestic $3.11 and US domestic $2.90. The fundamental objective of Target2-Securities (T2S) project, a single pan-European platform for securities settlement in central bank money, is to integrate and harmonise the highly fragmented securities settlement infrastructure in Europe. The project plan, being supervised by the ECB, can be found on the ECB website14. The project will go live in 5 waves, starting 22 Jun 2015 until 18 Sep 2017. T2S should improve the securities settlement landscape. Collateral

When it comes to collateral Figure 6 shows that government securities are equally important for both US and EU tri-party market. However this was not always the case.

Figure 6 – Collateral breakdown US and EU tri-party repo

Data source: EU ICMA survey December 2015 and US FRBNY tri-party repo data December 2015

The US tri-party has always been dominated by Treasury and Agency debt (about two-thirds of that market). However when compared with Copeland Adam, Antoine Martin, and Michael Walker (2010) and Copeland Adam, Antoine Martin, and Micheal Walker (2014) for the

14

(31)

30

crisis level we see that Private ABS has disappeared, Agency MBS and Corporates have

decreased, while US Treasuries excluding strips has increased slightly which confirms the move towards safer collateral. While European tri-party repo before the crisis was normally used to manage non-government bonds and equity. It was attractive for dealers to use tri-party for collateral with heavy overheads, difficult to value and expensive to transfer. Unfortunately ICMA survey between 2001 – 2004 lack proper collateral break down for the tri-party market. However as from June 2005 this was fixed and in ICMA Dec 2005 we can see that the share of equity was rising sharply to 20.4% from 11.9% in June 2005 and 15.5% in Dec 2004 while the share of EU government bonds was falling to 23% from 25.7% in June 2005 and 44.2% in Dec 2004. From June 2007 onwards ICMA report a sharp increase in government bonds from 22.6% to 43.6%. This is when the collateral distribution started to change and become what it is today, 62% of safe collateral made of Government, public & sub-national and supranational.

Maturity

We do not have data on the maturity breakdown for US tri-party repos, this lack of knowledge on the share of overnight, open, and term repos is related to the fact that clearing banks themselves did not need to know the term of the tri-party repos they were settling, before the 2010 tri-party repo reform, because of the daily unwinding. However we do have this statistics for the primary dealers for 2015 quarter 4. For Europe we take the data from ICMA survey December 2015.

Figure 7 – Tri-party repo maturity break down for both US and EU

Data source: SIFMA primary dealer triparty repo data and ICMA survey

In Figure 4 for the US we saw that tri-party repo is estimated to be more than 50% of the repo segment, while on the reverse repo segment tri-party account less than 20%. Since every repo

(32)

31

should have a reverse repo, as it is a closed system, this implies that non-interdealer reverse repo (for which we don’t have any data) is substantially bigger than the inter-dealer reverse repo. Additionally according to previous assumption made in the researches that tri-party account for 80% or more of the primary dealers total, we can conclude based on Figure 7 that more than 60% of the tri-party repo is overnight while tri-party Europe repo are mainly open ended. Surprisingly enough for the US repo the proportion of term to overnight have not changed much since 2006, as the data is coming from a regulatory form FR2004 quality should be good. While for Europe ICMA survey asked participants for the first time in 2005 to break out contracts with more than 12 months remaining maturity. As a result we observe a sudden jump in 2007 and 2008 had no information at all and then from 2010 onwards there has been an upward trend in the open tri-party repos. So it is difficult to interpret, if this upward trend has to do with change in market behaviour or just improvements in data quality. In any case the share of open repo is significant in 2015. The explanation might be related to trading cost, given the expensive trading costs of EU tri-party repos overnight trades might be too costly, while open contracts provide the flexibility of being terminated on demand and will effectively continue to roll overnight indefinitely.

5.5 Central Cleared Counterparty (CCP)

Unlike tri-party repo in a centrally cleared market, participants have exposures to the CCP (also known as clearing houses) instead of bilateral exposures to each other. By definition a central counterparty interposes itself between counterparties to contracts traded in one or more financial markets, becoming the buyer to every seller and the seller to every buyer and thereby ensuring the performance of open contracts. A CCP becomes counterparty to trades with market participants through novation, an open-offer system, or through an analogous legally binding arrangement CPSS_IOSCO (2012). Participants also need to be direct members of the CCP. Both liquidity taker and provider need to post initial margins with the CCP on the net amount of the collateral due. The initial margins act as a buffer to mitigate potential risks. The initial

margins or haircuts imposed by CCPs are very high compared to current market practice and the remuneration of cash margin paid to members is low. Furthermore following mark-to-market valuation of individual position participants may be asked to post variation margins. All in all CCPs are expensive to use compared to tri-party.

A comparison : US and Europe repo market