2. Securitized banking
2.1. The repurchase agreement
2.1.2. Market and volumes
2.2. The repo’s transformation
2.2.1. In the pre-crisis period
2.2.2. During times of instabilities and distress
2.3. The contribution of shadow banking and repos to the crisis
2.3.1. Short-term funding and shadow banking risks
2.3.2. Repo risks as a cause of instabilities
3. Studies on repo risks
3.1. Counterparty risk and the run on repos
3.1.1. ABX indices and the LIB-OIS spread
3.1.2. Data and empirical tests
3.2. Margin calls and bankruptcies in the repo market
3.2.1. The model
3.2.2. Model results
4.1. Basel III
4.1.1. Prevention of a run on repo
4.1.2. Reduction of liquidity risks
4.2. The Dodd-Frank Act
4.2.1. Prevention of a run on repo
4.2.2. Reduction of liquidity risks
4.3. Discussion and outlook
4.3.1. Summarizing evaluation
4.3.2. Political recommendation
4.3.3. The future of repos
In ancient China, collateralized loans have been used probably even before 1,000 BC, but with the Federal Reserve introducing repurchase agreements (or repos) in 1917 (Acharya, et al., 2010), secured loans became more complex. During the 20th century, repos gained in importance until right before the recent financial crisis, when they were seen as “perhaps the most important financial instrument in the world, after the basic cash equity and bond product” (Choudhry, 2006, p. 3). In addition, repos were considered of high relevance for short-term funding (Mancini, et al., 2014), as an essential requirement for liquid derivatives-markets (ICMA, 2013), and even as the predominant instrument in the shadow-banking system (Penney, 2011). The combination of this central economic role with the findings of Gorton and Metrick (2012), whereby the withdrawals from the repo market triggered the financial instabilities, lays out the reasons of the financial crisis. Two factors describe the connection between the crisis and repos in the most appropriate way. Firstly, the run on shadow banks and repos initiated the financial crisis. Secondly, systemic liquidity shocks had already destabilized the market prior to the crisis, and worsened the development later on.
The important role and the associated risks during the crisis prompted regulators to frame policies in order to keep the future untroubled by alike processes in the shadow banking sector. In fact, the Financial Stability Oversight Council, a Dodd-Frank Act entity, identified repos as one of the market components that needed strengthening the most (FSOC, 2011a). Nevertheless, while Valderrama (2010) covered the regulatory topic of repos’ systemic liquidity risks in her technical model, there is only very little research on the regulation covering other risks associated specifically with repos. The extensive regulatory measures proposed by the Dodd-Frank Act and the Basel III accord have not been assessed with respect to the repo market, whereas the shadow banking sector received more academic attention.
Richard Comotto, senior visiting fellow of the European Repo Council, specified the needed regulation in more detail, demanding a regulation that reacts on the consequences of financial development with a stimulation of positive externalities and simultaneously mitigating negative externalities (Comotto, 2012). This motivation is the driving topic of this paper, in which I am going to review the effects of Basel III and the Dodd-Frank Act regulation on the negative externalities connected to repo transactions more precisely. Thus, after the analysis of the connection between repos and the financial crisis, I will test whether, and under what conditions, regulatory policy achieves the reduction of risks, such as counterparty credit risk and liquidity risk. The knowledge about the regulation effects on repos will allow to make a statement about the danger of instabilities resulting in a new financial crisis.
This regulation assessment offers the major novelty to the research on repos. Former studies did not discuss - to a similar extent - the actual approaches concerning repos in combination with the causes and development of the recent financial crisis. This may be because shadow banking is rather hidden than transparent, complicating an empirical analysis as well as the data acquisition about volumes, shares and developments. Based on the existing review, I am going to highlight the necessity of regulating causal factors, such as counterparty risk, rather than a rash focus on symptomatic developments, like haircuts. The recommendation of an evaluation process for preventive or reactive regulations comes along with the important differentiation between the source and the symptom of financial instabilities. Hence, the elementary role of foregoing assessment for regulatory adjustments goes beyond the results of papers with similar topics.
Because the Dodd-Frank Act and Basel III are complex and sophisticated proposals, I will not evaluate the act as a whole, but the major included regulatory approaches will be assessed and - if effective - included in my recommendation. Whereas Comotto (2012) recommended a regulation that additionally strengthens positive effects of financial innovations, I will focus on his suggestion of mitigating negative externalities.
The contents of this paper are not only of interest to researchers, but also to regulators and financial institutions. Regulators that plan to intervene in repo trades may find chapter 2, 3 and 4 relevant, because chapter 2 and 3 cover on the one hand the relationship between the repo market and the financial crisis, and on the other hand the actual empirical and technical analysis which will qualify and lead to the results of chapter 4. This last section is relevant for regulatory authorities as well, since it reveals the actual consequences for repos due to the proposed regulation in the form of the Dodd-Frank Act and the Basel III proposals.
Furthermore, the conclusions from chapter 4 form a basis for regulation departments in financial institutions, like the regulatory risk management, to evaluate how regulations on repos may affect the own balance sheet and risk exposure.
2. Securitized banking
The term “securitized banking” was used by Gorton and Metrick (2012) to describe the combination of securitization plus repo finance, giving institutions the ability to fund much of the business of securitizing loans and holding/managing collateral. The authors attributed those activities to firms formerly known as “investment banks” (e.g. Bear Stearns, Lehman Brothers, Morgan Stanley and Merrill Lynch). Since securitized banking is a supplement to traditional banking activities, it is also crucial for commercial banks like Citigroup, J.P. Morgan and Bank of America. In fact, securitized banking was the key market that collapsed in 2008, causing the ensuing crisis and the taxpayer bailout of many important institutions.
In this section, I will address the transformation of repos and the question to which extent the repo market had influenced the burst and course of the financial crisis, even though the instrument was seen as panic-proof. The starting point is an assessment of the key characteristics of repos: their participants, elements, importance, and market volume.
2.1. The repurchase agreement
By using a repo, the repo borrower receives cash for selling a security or a basket of securities along with a commitment to rebuy it, or alternatively an asset with equal value, at a later stage. Involving an underlying security, the repo is a derivative by definition (Faure, 2011).
The quality and accessibility of financial data is important for researchers, regulators and companies. Unfortunately, the market for repurchase agreements is mostly hidden, which is why even “regulators lack sufficient data to thoroughly analyze all repo markets” (Financial Stability Oversight Council, 2014, p. 5). Although it is impossible to get information about the repo market in its entirety, I will present several estimations and facts from transparent parts of the repo funding market.
To be able to discuss the structure and complexity of modern financial markets, it is crucial to know which market participants act, and in which manner. The main market actors for repos split in borrowers, lenders and intermediaries – without strictly defined boundaries (Moorad & Choudhry, 2006).
Borrowers are financial institutions such as hedge funds, (investment) banks and securities houses. They use those activities to hedge interest rates, and especially to finance their assets and short positions.
Lenders, such as corporate treasuries, insurers, commercial banks, as well as mutual and pension funds, are the investors with large amounts of idle cash. They often buy tri-party repo from dealer banks, which again lend to other dealers and hedge funds by using the bilateral repo market. For those repo lenders, the money-market instrument is called a reverse repurchase agreement, because they buy the collateral from the repo borrower.
Custodian banks, or inter-dealer brokers and money brokers constitute the intermediaries in the repo market, acting as an agent for both the seller and the buyer.
Major elements: rate, haircut and margin call.
When the repo borrower, or seller, buys back the security, the lender, or buyer, also receives the interest, for its chosen credit. This interest, or repo rate, relates solely to the buying and selling price of that security for the cession period. Depending on the quality and liquidity of the collateral, the repo rate is often close to, but typically below the market rates. The repo rate is called special repo rate, if it is beneath the prevailing money market rates (Duffie, 1996). It will be explicitly discussed in chapter 3.1.
Repo defaults are more likely for securities with a longer maturity – such as many bonds, which are more price sensitive to interest rate changes than short maturity notes. To diminish this default risk, a haircut can be applied to the market value. This means a specified percentage is being subtracted from the market value of the used collateral (Hoshi, 2011). Acharya et al. (2010) were noting that there is also a possible haircut to the creditor. In this event, the haircut describes the margin demanded by the seller, protecting him against the potential value gain of the security in case the buyer fails to return the collateral on the purchase day. Chapter 3.1 will address the haircut approach in more detail.
The last important element of a repurchase agreement is the margin call. This is the amount that needs to be paid, if during the life of the repo the market value of the collateral decreases. The name illustrates that there is a lenders claim, telling the borrower his security has devaluated and therefore needs to be balanced. If these margin calls are made too frequently and aggressively, this can lead to financial imbalances, as seen in the case of Lehman Brothers. Lehman had to face daily margin calls, as the price of its mortgage-backed securities decreased, even worsening its financial situation. The consequences of excessive margin calls will be addressed in chapter 3.2.
The repos purpose
Maturity transformation is a key part of financial intermediation, and banks heavily rely on short-term funding, which became very important as an alternative to loans, credit from commercial banks and the issuance of commercial paper. Since “repos are the predominant form of short-term money market funding” (Mancini, et al., 2014, p. 1), the framework and features of repos are important for the whole economy. Although repos were not created as a hedging instrument, they can be utilized to hedge certain derivatives, such as forward rate agreements.
Acharya et al. (2010) named two reasons for investors to engage in the repo market: First to get higher interest rates compared with regular commercial bank deposits, and second those repo bank deposits are secured by debt used as collateral. They also say that while a central bank participates in the repo market mainly to implement its monetary policy, the primary security dealers are getting into the repo market mostly for market-making and risk management purposes. The connected collaterals high liquidity is an additional crucial benefit. Because the lender becomes the owner, the security can be held, sold or split without any further legal approval. Naturally, once to the “purchase day” comes, the lender has to rebuy the collateral (in case of selling it during the repo term), in order to return it to the repo borrower.
2.1.2. Market and volumes
Although Comotto (2012) classified repos as not being inherently a “shadow banking” tool, Jeff Penney, senior advisor at McKinsey & Company, said that “repurchase agreements are the largest part of the shadow banking system” (Penney, 2011, p. 1). This existing connection to the hidden shadow banking structure and mostly OTC activities averts official statistics on those derivatives and results in partially diverging estimations on the market structure and size.
Repo markets initiated in the G10 countries at different times. Namely, in the United States it developed during the 1920s, in Europe (besides the UK) in the 1970s and in the United Kingdom in the 1990s (Hördahl & King, 2008). The volumes for both of the US and European markets doubled between 2002 and 2007/08. Prior to the crisis, most monetary authorities used these transactions as a monetary policy instrument, which was also able to increase market depth, liquidity and price efficiency (Bank for International Settlements, 1999). Years later, investment banks were invested in repos as well. Since they need short term funding, they finance themselves to 50 percent by using repos (Hördahl & King, 2008). This already illustrates the repo diffusion for specific institutions and led some researchers to evaluate the market.
The overall market size calculation of $10 trillion by Gorton and Metrick (2012) is being fortified by the “Frequently Asked Questions on Repo” published by the International Capital Market Association (ICMA) (2013) suggesting a global volume of up to $10 trillion ($19.9 with double-counting for both lender and borrower) as well.
For the United States, estimations on primary dealers, which fund their repos via US Treasuries, reveal the repo development from 1996 through 2014, as illustrated in figure 1:
FIGURE 1: Repo financing over time
illustration not visible in this excerpt
Daily financing (Annual averages) by US Government securities primary dealers.
(Securities Industry and Fin. Markets Association, 2014)
The estimation for 2012 support those from Copeland et al. (2012), who sized the outstanding repo business of its primary dealers to under $5.5 trillion. In mid-2008, the report by Hördahl and King (2008) referred to an akin source with 19 primary dealers and 1,000 bank holding companies, which estimated the US repo market to have exceeded $10 trillion (including double-counting). This is inconsistent with the quotations of the Securities Industry and Financial Markets Association, provided in figure 1, and illustrates the difficulty to appropriately value repo markets.
In June 2008 as well, the volume of the EU repo market reached over $5 trillion (ICMA, 2014). According to further compiled values in the European Repo Market Survey the recent European repo business adds up to $3.9 trillion, suggesting a decline in market size during the crisis.
In the United States, half of all repo transactions are one-day transactions, or overnight repos (Acharya, et al., 2010). Also the US repo market is dominated by tri-party repo with estimated shares ranging from between 46 percent and 81 percent (Copeland, et al., 2012). Those tri-party repos, or repos which are cleared via a central counterparty, are collateralized with liquid assets like government and agency bonds to an extent of 80 percent (FSOC, 2011b). The European market, however, is with 90 percent predominantly bilateral (Mancini, et al., 2014) and it faces an increase of repos with a maturity of one month at least (ICMA, 2013).
2.2. The repo’s transformation
As banks heavily rely on short-term financing, the transformation of maturities is a major activity in the financial market. Still, financial imbalances can be fueled and even caused by short-term debt. In the following sections, I will give a detailed overview about the repo’s pre‑crisis situation, its effects on financial instabilities and the development during the crisis.
2.2.1. In the pre-crisis period
The transformation from traditional banking to shadow banking
In the 1960s nearly 60 percent of the relative share of total financial intermediary assets came from the traditional banking model (Edwards & Mishkin, 1995), which is based on granting and holding loans with insured demand deposits as the main source of return (Gorton & Metrick, 2012). These insured demand deposits imply that traditional banks are acting within a safety net, designed to prevent large runs on those institutions. Unlike that framework, shadow banks are detached from those safety features. In the following section, I will show how these non-banks (e.g. insurance companies, mutual and pension funds) replaced the traditional banking sector in 1994, reaching a share of 60 percent of total transactions with financial intermediary assets (Edwards & Mishkin, 1995).
The Regulation Q from 1933, like the setup of very low interest limits, in combination with the prohibition of entering non-banking markets, diminished deposits at banks (Trotter, 2013). The additional competitive pressures from non-banks (i.e. money market funds – not being influenced by Regulation Q) and the huge reliance on traditional banking contributed to this issue and eroded the banks’ profitability. The decline in margins resulted in consolidations and US bank failures, which rose from below 20 in 1981, where they had been since 1960, to over 200 in 1988 and declined to the former level until 1994 (Edwards & Mishkin, 1995).
During the same timeframe, technological progress improved the availability and the quality of information, leading to a decrease of information asymmetry. This development made financing with commercial paper and money market funds more attractive. Similarly, returns from trading account income and commission/fee income became more relevant (Kodres & Narain, 2010). This shift in priorities, implied a major loss of customers for banks, which consequently searched for new opportunities, such as: financing not only with customer deposits, but also with intra banking funding, like repos. The evolution in the market allowed banks to diversify their activities: a broader range of services, international expansion, increasing their size and share (mergers & acquisitions), securitization, more debt and additional risk. Starting in the 1960s, when over half of the financing happened by using cash deposits, this share diminished to under 20 percent in 1995 (Edwards & Mishkin, 1995). With the relaxation of the Glass Steagall Act in the 1980s - until full repeal in 1999 - regulators were seeking changes in order to allow banks to engage in securities activities. This should reinforce the banks’ ability to pay more interest on deposits, in order to recollect the customers into their traditional core business.
Securitization and the demand for repo
Although activities with repurchase agreements had been known before the shadow banking development, figure 1 shows how repo markets became more important as the securitization unfolded. Namely, the rising mortgage volume, due to several political decisions, made repo funding even more relevant for financial institutions.
For instance, the New Deal’s National Housing Act as early as 1934 (Fishback, et al., 2001) and other recent governmental interventions, which planned to provide mortgage to the poor and minorities, forced the private sector to offer financial products for groups which had been classified by banks as too unsafe to qualify for the usual mortgage contracts - like a 30-year fixed rate mortgage.
More precisely, Gorton (2010) stated that institutions structured those mortgages with a maturity of two or three years and a fixed interest rate. After that period – at the “reset date” – the rate rose significantly so that the debtors had to reschedule their debt payments. With rising home prices, borrowers were able to refinance. As nobody was appropriately considering the risk of collapsing prices, the project was considered a success.
This optimistic view about home prices influenced its major funding source as well: the securitization instruments. In fact, the usage of repos was based on the securitization, which was financing about 80 percent of the subprime mortgages (Gorton & Metrick, 2012). The process of packaging loans into diversified bonds (origination) in order to be able to sell the risk of the borrower to investors, such as repo buyers, was an important step towards a more complex financial system. This origination resulted in major advantages for banks and investors (Reserve Bank of India, 1999):
i. key source of funding and revenue by means of fees,
ii. increased shareholders value plus help to reduce the exposure, and
iii. improved financial ratios increasing the flexibility under equity and capital requirements.
By buying bonds from the special purpose vehicle, which acquired the mortgage-backed securities, investors essentially benefit from:
i. the opportunity to achieve a higher return than vs. non-securitized products, and
ii. the ability to diversify their asset portfolio with regards to both risk and region.
Because they constitute major components of the research in chapter 3.1, it is important to illustrate the other new financial instruments, which are partially used as repo collateral, created by the securitization process – especially. Among these, Gorton and Metrick (2012) were specifying asset-backed securities (ABS), credit default swaps (CDS), collateralized debt obligations (CDOs), and collateralized loan obligations (CLOs).
Figure 2 shows that the market for CDOs, ABS and mortgage-backed securities (MBS) reached nearly $2 trillion in 2006 and dropped in the course of the financial crisis (Zandi, 2010).
FIGURE 2: Bond issuance over time
illustration not visible in this excerpt
Annualized bond issuance in billion dollars ordered by collateralized debt obligation, asset backed securities, commercial mortgage-backed securities and residential mortgage-backed securities.
The main reason for the rapid expansion in demand for repo (doubled in size from 2002 to 2007 – see chapter 2.1.) is the increase in funds managed by institutional investors, pension funds and mutual funds (Gorton & Metrick, 2012). The demand from states, municipalities, and nonfinancial firms also drove this trend.
2.2.2. During times of instabilities and distress
While the size of the European repo market has doubled in five years from 2002, it has decreased in the ensuing years, leading to a recent European volume of $3.9 trillion (ICMA, 2014). Figure 1 illustrates both this development and the drop in repo usage starting in 2009. The development behind that decline will be assessed in the following section.
Kodres and Narain (2010) wrote for the International Monetary Fund that due to low nominal interest rates, the financial crisis developed in an environment where investors and financial institutions were immoderately optimistic about asset prices and risks. After several indicators signaled that the financial markets were becoming more vulnerable, the growing concern about counterparty risk and a skyrocketing demand for liquidity in mid-2007 caused significant disruptions in the credit and money markets (Hördahl & King, 2008).
Hördahl and King connected the strong fluctuations in asset prices, leading to an increase in uncertainty regarding the value of collateral, especially to less liquid assets and collaterals with a value hard to determine. These repo markets came rapidly under pressure, even though the consequences for repo rates were less pronounced than for unsecured LIBOR. Moreover, Michaud and Upper (2008) highlighted a significant divergence of the LIBOR and OIS rates since late 2007 (figure 3). Due to the feature that OIS contracts are secured - compared to repos, the repo OIS rate spread is referred to as a more appropriate benchmark. For that period, Hördahl and King (2008) draw a line between the EU and the US repo markets: repos came under fire especially in the US, whereas the European and British repo pricing kept its low-key profile. Furthermore, they argue that in the US, the volatility of repo rates increased significantly, and that borrowing connected to maturities greater than one month turned out to be problematic.
FIGURE 3: LIBOR, OIS and GC repo rates
illustration not visible in this excerpt
Repo, LIBOR and OIS rates, three-month rates in percent.
(Hördahl & King, 2008)
Because of the increased demand for safe government securities, repo rates based on that collateral (GC repo rates) disproportionally declined compared to OIS rates with the same runtime, as illustrated in figure 3. In contrast, rates for repo transactions being connected to a collateral with a higher risk level rose.
Gorton and Metrick (2012) used the spread between the LIBOR and the OIS as a proxy to explain the run on repos in August 2007. They assume that the market slowly became aware of the subprime market risks, leading to doubts about the quality of repo securities and the bank reliability (chapter 3.1. addresses this paper in more detail). Other authors, like Hördahl and King (2008), as well as Lucas and Stokey (2011), affirmed that assumption by attributing the occurred run on the repo market to the lack of confidence in securities that did not meet highest quality standards, and to the mitigation of trust in the ability of investment banks to redeem their short-term loans. More recently Jeff Penney, senior advisor at McKinsey & Company, characterized the repo network to be vulnerable to investor panic (Penney, 2011).
In the course of the crisis
After explaining those disruptions in the repo market, it is yet to be clarified how repurchase agreements changed during the recent period of financial instabilities and in the course of the crisis. In the following part, I am going to examine that matter.
As the financial instabilities intensified, especially the US repo markets came under pressure. In September 2008, the entire US GC repo market was trading at rates that normally apply to Special Repos, while the GC repo rates in Europe had risen over the OIS rates. Hördahl and King (2008) explained this alteration with a decreased average spread between the GC repo rates and OIS rates by 25-30 basis points (bps). In fact, Krishnamurthy et al. (2014) found a drop in repo with ABS collateral from the pre-crisis level of $169 billion to about $14 billion in the first quarter of 2009. According to Hördahl and King (2008), this market shock happened especially for fixed-term repos, where trades with a maturity of at least one week vanished predominantly.
This standstill of repo activity in the fixed-term market was accompanied by the largely abolishment of repos based on corporate bonds and structured securities, suggesting that the market had concentrated on securities with the highest quality (Hördahl & King, 2008). For the two weeks following Lehman’s bankruptcy, Gorton and Metrick (2009) estimated that average haircuts, or initial margins, for non-US treasury collateral grew about 18 percentage points to the point of 43 percent. They argued that this change in pricing was the result of concerns about the securities’ illiquidity. Besides the changes in the haircut and the focus on securities like government bonds, Hördahl and King (2008) explained that the overall financing conditions had become stricter and that uncertainty about the future led to a decrease in the repo runtime. For instance, the repo utilized short-term funding, in the case for multinational German banks, decreased from 60 percent before the crisis to 48 percent before the Bear Stearns rescue in March 2008 (Deutsche Bundesbank, 2013).
Thanks to the enhanced liquidity provision by central banks in the US and in Europe, the constitution of the repo market improved. Longer maturities were available again, sales are on the rise as well and repo rates based on government securities normalized. This is what Hördahl and King already anticipated back in 2008 and what was supported two years later from Tett (2010). In fact, Gillian Tett argued that by mid-2010 the market had recovered from the crisis for the most part and, at least in Europe, had risen to exceed its pre-crisis size. In addition, both the way repos are traded and their complexity changed. The interbank activities started to shift away from interbank lending to non-bank financial institutions and security firms (ICMA, 2013), mitigating the interbank complexity.
2.3. The contribution of shadow banking and repos to the crisis
Besides that structural difference, Acharya et al. (2010) effectively observed a damaging run on shadow banks, but no wholesale run on traditional banks. But prior to this, the shadow banking system financed most of its leveraged positions with repos, leading to the runs’ emphasis on the repo market.
The crisis of 2007-2008 was, as Gorton (2010) argued, a return to the full-scale financial crises of the nineteenth century. Against the background of the brief overview on the development of repos before and during the recent financial distress, I highlight the repos’ crisis contribution specifically.
2.3.1. Short-term funding and shadow banking risks
Short-term financing risks
The innovation concerning securitization is that it empowers intermediaries to access massive pools of short-term cash seeking risk-free yield even without a deposit protection scheme by the government (Gennaioli, et al., 2013). When thinking about the financial crisis of 2007-2009, many believe that short-term debt increases the financial vulnerability or may even cause a crisis. This might happen because short-term funding exposes financial institutions to runs (Diamond & Dybvig, 1983), financial contagion (Allen & Gale, 2000), and rollover risk (Zhiguo & Xiong, 2010). In addition, Gennaioli et al. (2013) explained how bank runs quickly turn system-wide. That is because the existence of short-term finance depends on the perceived reliability of publicly traded securities held by several banks.
However, the empirical analysis by Benmelech and Dvir (2013) suggested that short-term debt was a consequence of weak financial institutions rather than the reason for their downfall or for bank failures. Nevertheless, having the case of Lehman Brothers in mind, where the investors lost faith in Lehman’s ability to repay its short-term debts, it seems that excessive reliance on short-term debt does potentially lead to the banks’ collapse (Ryan, 2013).
Shadow banking risks
A range of issues, highlighted by Comotto (2012), illustrate the greater systemic risk for shadow banks than for traditional banks: lack of regulation and arbitrage possibilities in between those gaps, agency problems, the complexity and the dimension of the shadow banking system, the lack of transparency, the (in parts) neglect of systemic and market risk in pricing, the emergence of excessive leverage, as well as the interdependence of shadow banks with each other and with traditional banking systems.
More or less all of these risks can be attributed to repos as well, since it is a major instrument in the shadow banking sector. The actual repo risks will be discussed in the following section more precisely.
2.3.2. Repo risks as a cause of instabilities
Risks associated with repos
Because wholesale liabilities - such as repos - are collateralized, they are akin to protected deposits issued by traditional banks. Nevertheless, Comotto (2012) named a number of reasons why repurchase agreements might be riskier: no official safety net, the greater funding dependency of shadow banks, lack of regulation, and the ever changes of many institutional cash balances. In addition, he argued that repos feature risks following from their pro-cyclicality, excessive leverage, and lack in transparency. Comotto explained furthermore, that the Lehman’s Repo 105 policy, an accounting trick to make a bank look like it is less reliant on debt than it actually is, strengthened the doubts about the transparency of repos. However, this lack in transparency might be overrated at least in Europe, since the International Capital Market Association argued in their FAQ on repo (2013) that after all, 40 percent of European repo market turnover might be cleared through risk reducing central counterparties (CCPs). By providing multilateral netting, as well as a credit-worthy counterparty, the integration of CCPs into the process reduces risk exposures and the accumulation of liquidity. Still, not only the risk decreases with the CCP-practice, but as Mancini et al. (2014) suggested: market actors assume CCP clearing as a safe part of the interbank market that can absorb the rising demand for hoarding of liquidity.
The repo as a crisis trigger
Repurchase Agreements might have lighted or even caused the financial crisis via different channels. Firstly, and of capital importance, the run on repo and the pro-cyclicality of haircuts (Gorton & Metrick, 2012), and secondly, the excessive repo use as an accounting maneuver (Repo 105), which some analysts assumed mistakenly to represent the proper method for repo accounting (Comotto, 2012).
Two academics, Gary Gorton and Andrew Metrick (2012) of Harvard University, coined the term that the ‘run on repo’ could have caused the financial crisis in 2007. They were arguing that the crisis was connatural to a traditional banking panic, but was instead triggered by a run on the repo market as a part of the securitized banking sector. The authors stated the massive deleveraging in the financial system, enforced by the deepened haircuts reducing the value of collateral to a great extent, as a major blow for the market. Companies from which repo funding was revoked by the enforcement of ever-expanding initial margins (pro-cyclical behavior, as the crisis has led to expected devaluation) had to deleverage by selling assets. The resulting fire sales intensified the crisis. Furthermore, featuring inter alia a considerable run on Bear Stearns’ repos in the beginning of March 2008, Acharya et al. (2010) were arguing, that panic selling was central to the financial crisis of 2007 to 2009. The money market mutual funds that financed Bear Stearns’ AAA-rated mortgage backed securities in the overnight repurchase agreement market refused to extent that funding. However, papers like that from Krishnamurthy et al. (2014) were characterizing the run on repo as a sideshow, because the vast majority of repos were collateralized by safe securities such as government bonds, instead of riskier securitized mortgage products. Picked by the ICMA (2013) as well, this argumentation indicates an overestimated impact of haircuts, especially in the European repo market. This might be because 80 percent of collateral in that market is government bonds, also it is not as dominated by one-day maturities as the US market is. In the end, the run on repo could have been driven by psychological factors, or as Krishnamurthy et al. (2014) noted: the troubles some systemically important dealer banks were facing, can be attributed to their difficulties in repo funding of securitized assets, to which they were predominantly exposed to.
As the first financial trembles started, Lehman knew it needed to reduce its reliance on debt. The Repo 105 policy was seen as an applicable method to reduce its leverage. Archarya et al. (2010) argue moreover that by the use of this loophole towards the end of every fiscal quarter since 2001, Lehman was able to keep its equity unchanged – reporting a lower leverage than it actually had. As a result, Lehman’s Repo 105 turnover were as much as $50 billion in some quarters (Acharya, et al., 2010). Activities such as the excessive Repo 105 usage by Lehman can increase the “reputational risk” leading to a loss in confidence of its funding sources. Besides this prestige issue, primarily the false statement about the leverage may cause crucial imbalances within the financial sector when it comes to shocks in reliance. This - besides other factors - boosted the incidents at the initial period of the crisis, at least in the United States. In Europe, these accounting options are not available and repurchase agreements must be accounted for in the standard treatment; which is discussed by Comotto (2012): keep the collateral on the seller’s balance sheet, as it retains the collateral risk and return when repurchasing at a fixed price as soon as the purchase day is reached.
Even though the run on repos and to some extent an excessive/accounting repo usage seem to have a considerable crisis participation, Mancini and Ranaldo (2014) suggested that the CCP-cleared repo market is better characterized as an interbank buffer rather than a financial shock trigger. A narrowed repo participation in the crisis seems to hold true for Europe, since the authors found that around 60 percent of euro interbank repos are cleared by a CCP, whereas in the US only a minority is CCP-conducted. Thus, the run on the shadow banking system, and therefore to some extent also the run on repos, can be described as “the core of what happened” (Krugman & Wells, 2012, p. 952) to cause the crisis.
3. Studies on repo risks
Certainly, the run on the securitized-shadow-banking system can be considered as one of the financial crisis triggers. This chapter will discuss this issue in more detail on the basis of two major studies – each with a technical analysis.
The first research explains the danger in the interbank market and how the run on repos developed. Beginning with the review of this paper from Gary Gorton and Andrew Metrick (2012), in which they were proposing a mechanism for the run on repo by analyzing two data sets. This study had an eminent impact on the repo related research because the authors were able to illustrate the connection between the trouble with structured products and repo transactions. They argued in particular that it is required to realize why spreads on non-subprime related assets rose dramatically in order to understand the crisis.
The second paper will derive different major parameters for systemic liquidity shocks and the bankruptcy of repo participants. In this technical study, Laura Valderrama (2010) clarified the relationship between liquidity risks, the collateral’s price, the repo haircut, the capital base and financial instabilities. As the research from Gorton and Metrick (2012), Valderrama (2010) illustrated how an increase in the repo’s haircut may cause a run on that financial product, resulting in bankruptcies of financial institutions.
Whereas Gorton and Metrick derived their conclusions from empirical regressions, Valderrama develops her equations from a rational investment-behavior-framework and economic restrictions.
3.1. Counterparty risk and the run on repos
While some literature, e.g. Collin-Dufresne et al. (2001), has discussed spreads on corporate bonds, Gary Gorton and Andrew Metrick (2012) released their study on securitized product spreads as the first to address that topic. Securitized products are widely used as a repo collateral; that is a reason why this research is very important when it comes to cover the repo market. Their paper had major influence on the regulatory debate regarding haircuts and central clearing of repos, topics I come back to in chapter 4.
In their paper Gorton and Metrick (2012) provided evidence for the effects of a risk increase at banks on those securitized assets (like securitized bonds, credit-default swaps and other assets) that are used as collateral in repo transactions. These effects are major factors for the run on repo - and hence for the crisis triggers as well. Therefore, this paper discussion is divided in four sections addressing: the two main state variables, the data set, the empirical tests plus its findings, as well as a critical acclaim.
3.1.1. ABX indices and the LIB-OIS spread
Gorton and Metrick (2012) used the ABX index as well as the LIB-OIS as the two main state variables. The former acts as a proxy for fundamentals (or pricing) in the subprime mortgage market and the latter proxies for primarily counterparty risk in the banking system.
ABX indices and housing prices
With the development of the subprime securitization, several subprime related indices started to emerge in the beginning of 2006. One of those indices is the ABX.HE (ABX) which forms the basis for Index CDS. This credit derivative was set up by dealer banks to allow investors to take positions in subprime mortgage backed securities. The sub-indices of the ABX reference 20 residential mortgage-backed security (RMBS) bonds within a specific rating-group.
Because many subprime-related instruments and securitized products do not trade in publicly observable markets, it is important to get information about the value of subprime mortgages by observing related indices, such as the ABX. These indices help to analyze the pricing of subprime risks, since it opens a liquid and publicly observable CDS-trading market.
By using ABX indices, the authors could provide a detailed look at the pricing in the subprime mortgage market, which was key for the collateral valuation in repo contracts.
LIB-OIS spread and the interbank market
The LIB-OIS spread is the rate for overnight unsecured interbank lending (OIS) subtracted from the rate paid on 3-month unsecured interbank loans (LIBOR). A decline in the central banks’ interest rates lowers the rates of both the LIBOR, estimated by leading banks in London, and the OIS, based only on the rates set by central banks.
Hence, a doubtful solvency of borrowing banks would lead to a LIBOR gain, indicating the increased perceived counterparty risk in the banking system (Acharya & Öncü, 2010). This circumstance makes the LIB-OIS an appropriate and important measure of counterparty risk and market-liquidity in the money market (Schwarz, 2014).
The spread was used by Gorton and Metrick (2012) primarily as a proxy for the repo market’s condition and for the health of the interbank market.
The crisis chronicle
Figure 4 examines a timeline with partly contrary ABX and LIB-OIS developments. Wherefore the authors analyzed the trend changes for the beginning of the crisis: from early 2007 to the end of 2008.
FIGURE 4: ABX and LIB-OIS developments
illustration not visible in this excerpt
ABX versus LIB-OIS. ABX is the 2006-1 BBB tranche.
(Gorton & Metrick, 2012)
While the LIB-OIS shows two small jumps in August 2007 and one significant in September 2008, the ABX continuously rose unaffected by the LIB-OIS peaks. Although Acharya et al. (2010) hinted at the freeze of the asset-backed commercial paper market as a driver for the first jump in the LIB-OIS spread, Gorton and Metrick (2012) did not find specific reasons for the massive LIB-OIS changes. Gorton and Metrick rather highlighted the independence of the ABX growth from the LIB-OIS development.
The first signals of particular danger in the interbank market during the observed period was during the summer of 2007. The LIB-OIS jumped from 13 bps in July 26 to 40 bps in August 9 and from this point to 96 bps on September 10. Starting from 669 bps at the end of June and reaching 1738 bps by the end of July, the ABX rise launched earlier than the first LIB-OIS peak.
Whereas both the ABX and LIB-OIS continued to behave differently in the first half of 2008, the real impact of the panic struck subsequently. The uncertainty in financial institutions and in the asset market led the ABX spread to rise further and pushed the meanwhile calmed down LIB-OIS to its historical record of 364 bps in October 10. This peak in the LIB-OIS spread was due to the possibility of an interbank market collapse, Gorton and Metrick (2012) concluded. The fall back to 128 bps by the end of 2008 came along with the relaxation in those doubts about the financial stability.
Gorton and Metrick reviewed the timeline for the crisis by showing that the subprime market recession began in early 2008 – according to the ABX data. The peak in the LIB-OIS in August of 2007 revealed the real bank’s balance sheet damage, caused by banks having many securitized mortgages and pre-securitized assets in their books.
3.1.2. Data and empirical tests
In order to give an accurate interpretation of the results, the quality of the data being used is a major criterion for researchers. In this section, I will present and discuss the used data and the main empirical findings of Gorton’s and Metrick’s (2012) paper.
Because this empirical analysis examines co-movement of spreads on various assets - due to the lack of a structural model of repo markets - these results are rather based on correlation than on causation.
To perform the analysis, Gorton and Metrick (2012) used two novel data sets: one with repo rates and haircuts; the other set includes information on 392 structured products and credit derivative indices, including classes of ABS, CDOs and CDS.
Dealer banks which convert the observed market prices into spreads, offered the broad data. In each case, those large financial institutions capture the spreads of interest by analyzing the bond or tranche. Many repo participants usually use this data for trading and portfolio valuation.
By using this information, Gorton and Metrick (2012) were able to provide a new perspective on the crisis contagion. In fact, they argued that in order to explain the crisis, it is crucial to clarify why spreads on non-subprime related asset classes rose significantly.
Methodology and basic tests
This section explains the methodology, and tests whether a few representative asset classes move with the subprime market or rather with the bank sector’s counter party risk. Hence, an answer will be given, how the spread on US non-subprime related asset classes changes with the development of the subprime related index (ABX) and the counter party risk (LIB-OIS).
The estimation is modeled for each asset with the following regression:
illustration not visible in this excerpt
Here Si,t is the spread on the asset i at time t, a 0 is a constant, a 1 is a time trend, and t is a weekly time index. The vectors ABX t and LIB − OIS t include the last four observations of the underlying index spread including the current period, and X t is a vector of control variables. Finally, e i, t is the disturbance term, capturing all factors which influence S i, t unlike than the regressors.
Subsequently, the authors took the first differences of this regression and normalize all changes by their previous period level. Since these observed levels vary heavily over time, the first difference (Δ S i, t = S i, t - S i, t-1) addresses the problem of omitted variables in econometrics with panel data. Because the individual heterogeneity, or time invariant omitted variables a 0, is expected to be time constant, the first differences remove a 0. Hence, this fixed effects model can be estimated by the following OLS-regression:
illustration not visible in this excerpt
Where the Δ constitutes the change of the related variable.
Based on the regression results for the asset classes of US non-subprime-related assets (AAA tranches), the ABX coefficients are probably not jointly significant (the F-Test, with the hypothesis that the coefficients on the ABX variables are jointly zero, suggests that). In particular, jointly significant effects of a one percentage point change in the LIB-OIS spread were measured. This let the spread of ABS backed by credit card receivables rise by 0.341 percent and the spread on ABS backed by student loans increases by 0.461. While a one percentage point change in the ABX index leads to a significant change in student loans (0.455 percent), the spreads’ reactions of ABS backed by: auto loans (-0.331 percent) and credit cards (-0.141 percent) were not significant.
Whereas the high grade and mezzanine CDOs are not significantly affected by either the LIB‑OIS or the ABX, this changes with other assets like ABS. On a more global perspective, the authors could find a significant effect of the LIB-OIS only. Asset classes such as the AAA-rated ABS backed by European car loans and the AAA-rated ABS backed by European consumer receivables are significantly and positively affected by the LIB-OIS spread at the 10-percent confidence level.
Thus, at least the counterparty risk, measured with the LIB-OIS spread, seems to have an impact on different assets like some loans and ABS. Nevertheless, there could still be a correlation between the subprime housing market (ABX) and other securitized assets that were used for repurchase agreements.
Credit spreads for non-subprime assets and rating classes
To test this possibility regarding non-subprime securitized assets, this section presents the joint-significance (via F-test) of the changes in the LIB-OIS.
For the period from January 2007 through January 2009, 66 percent of the US non-subprime asset classes, and 76 percent of the European non-subprime categories are significantly positively correlated with LIB-OIS changes at the 10-percent level. Most of this happened in 2007 with European structured products, whereas for these products in the US it is split across 2007 and 2008. Once more, ABX changes did not seem to be useful at predicting movements in spreads.
Besides the allocation into different asset classes, Gorton and Metrick (2012) analyzed assets on the basis of ratings as well. Resulting in a significantly positive correlation of 62 percent of the AAA products with LIB-OIS movements. This rating category was most widely used for repo collateral, but also more risky bonds can be part of a repo transaction. Those AA, A and BB rated products are also positively and significantly correlated with the LIB-OIS leveling at 28, 55 and 53 percent, respectively. For all ratings, the percentages are about equally divided between the sub-periods 2007 and 2008. Gorton and Metrick (2012) did not provide any results on the F-tests for the ABX index concerning rating categories; hence, I assume that there were no significant findings.
This demonstrates that the spread gains on non-subprime assets and related derivatives was correlated with the interbank counterparty risk (LIB-OIS) rather than with the subprime housing market (ABX). These higher non-subprime asset spreads are equivalent to decreased prices, which display a reduction in the worth of securities used in repo contracts. In practice, as the lenders doubt about bank stability rose, this translated into an increase in repo haircuts, illustrated in figure 5.
FIGURE 5: Repo haircuts
illustration not visible in this excerpt
Repo haircuts on different categories of structured products.
(Gorton & Metrick, 2009)
Overall, the counterparty risk, rather than the subprime mortgage market, seems to drive the movement of repo rates and haircuts. However, one question remains: Are there differences on what influences repo rates and haircuts?
Because of the existence of transaction costs for selling collateral in case of default and the uncertainty about the value and volatility of the collateral, measures of interbank counterparty risks are relevant for all market participants. Higher repo rates would occur because the lending and its underlying security are no longer risk free. To test how the average repo rate changes, if the ABX, LIB-OIS or expected volatilities change, Gorton and Metrick (2012) estimated a modification of equation (2):
illustration not visible in this excerpt
where R j,t constitutes the average spread of repo rates to the OIS for the collateral class j. The vector VOL j,t represents the last four expected volatilities for this class j, and the other variables are defined as in equation (2).
Changes in the ABX , the VOL , LIB-OIS or X do not significantly affect repo rates. Similar to the findings for credit spreads, the variable for counterparty risk in the banking sector is the only significant factor correlating positively with repo rates. In addition, the movements in repo spreads are significantly linked to the change in LIB-OIS only. This is indicated by the F-tests for the joint significance of LIB-OIS, the ABX and VOL changes.
Nevertheless, a crisis is not caused by higher rates themselves, instead the withdrawal from banks by lenders constitutes the systemic event for a run on repo. In the repo market, this withdrawing is represented by higher haircuts that are applied to the market value; figure 5 shows that this happened.
The following econometric model will test for several possible reasons for a repo haircut increase: to adjust for counterparty risk (LIB − OIS t), to balance subprime risks (ABX t) or to compensate for the uncertain value of the collateral by the time it can be sold (VOL j, t).
illustration not visible in this excerpt
This is the same regression framework that is used for repo spreads, but the dependent variable H j,t constitutes the average haircut for all assets in class j.
This time, the only significant variable is the proxy for implied volatility, defined as the difference between expected volatility today and realized volatility over the previous four weeks, which is significant for most of the haircuts of securitized bonds classes. In fact, an implied volatility change for AA-AAA rated residential mortgage-backed security or commercial mortgage-backed security and for AA-AAA rated collateralized loan obligation has a significant effect on the repo initial margin of collateral based on those asset classes.
Because a withdrawal from the bank is the equivalent to a haircut increase, which constitutes the run on repo, it was important to determine the volatility effect in order to understand the actual run on repo. Gorton and Metrick argued furthermore, that the reason for not specifying the exact causes of the run is the fact that the gains in the VOL and the LIB-OIS increase happened simultaneously.
LIB-OIS: Counterparty risk or market-liquidity risk?
Until this point, Gorton and Metrick (2012) referred to the LIB-OIS spread as a proxy for counterparty risk. In contrast, Schwarz (2014) found that market-liquidity can explain much more of the LIB-OIS than other risks may. This section discusses both positions and tries to determine the primary LIB-OIS driver.
When using different data, it is not surprising that the outcomes might differ. But for this part, Gorton and Metrick used Schwarz’ data and methods from 2009 (revisited in 2014 by Krista Schwarz) to decompose the LIB-OIS. For the part of Gorton and Metrick’s sample period, they used the credit-tiering/distinction measure construction of Schwarz:
Where the time-varying credit measure is the daily average of intra-daily credit-tiering. This measure is calculated by the quintile-average borrowing rate on each day t, at each time-of-day h for the maturity m. The difference in brackets is the average spread in overnight borrowing rates between “high-risk” and “low-risk” banks on day t at time h.
This credit measure can be interpreted as a proxy for credit risk, as shown in Schwarz (2014). Some (but not the majority) of the variation in LIB-OIS from January 1, 2007 until April 30, 2008 can be explained with this credit-tiering model.
Gorton and Metrick (2012) modified their regressions (2) and (3) as well as their volatility measure by replacing the ΔLIB−OIS with analogous changes in credit t. This led to almost the same results as for the LIB-OIS variable. The credit-tiering measure was significant in 60 cases at the 1-percent level and additionally for 28 cases at the 5-percent level (out of 176 non-subprime related US bond spreads). Since the LIB-OIS spread is also influenced by other factors, these new results suggest that Δ credit t might be superior to LIB−OIS changes as a proxy for counterparty risk. However, Gorton and Metrick justified their measure choice with the fact that the credit-tiering data is not available for the entire sample period they wanted to cover and because for the case of the repo haircut measure in regression (4), the credit-tiering measure was insignificant in all cases.
Gorton and Metrick also decomposed the LIB-OIS spread into a credit-predicted variable and its weekly changes as well as into a residual (which includes inter alia market liquidity) and its weekly changes. They found that the credit-predicted component was frequently offering significant results, while the residual of predicting repo rates and credit spreads was less frequently significant. This suggests that the important component for the LIB-OIS spread is the counterparty-risk rather than the market-liquidity.
Thus, the LIB-OIS is indeed an appropriate proxy for counterparty risk, at least in the observed period. Even though Gorton and Metrick (2012) used data from Schwarz (2014), they could fortify their position in this section by showing that the counterparty risk was the main driver for repo rates and credit spreads.
Contrary to the interpretation by Schwarz, Gorton and Metrick showed that the counterparty risk is forming the major factor for their findings, whereas the liquidity component has no significant effect.
When the danger in the interbank market increased, the higher spread between the 3-month LIBOR and the overnight index swap (OIS) (see figure 4) signaled the lender’s doubt about the solvency of borrowing banks. The subsequent run on repo was the result of the jump in repo haircuts (due to uncertainty about collateral values), equivalent to massive withdrawals from the banking system. Also the increase of repo rates (due to concerns about counterparty risk), to still be able to attract funds during the period of high counterparty risk, is another reason. These results were, and still are, crucial for the research on the financial crisis and the repo market. Especially because of their influence on the regulatory debate on the pro-cyclicality of haircuts, which develops the idea of a minimal but mandatory haircut. Nevertheless, there are some questions and critical arguments I will discuss.
Whereas the counterparty credit risk constitutes a major driver resulting in financial instabilities, the causal development might have received its main impulse not in form of deregulation of the commercial banks, but due to the government’s unsustainable housing policy (Trotter, 2013). However, since regulation, that reduces counterparty risk and weakens a new financial crisis, is dependent on financial activities rather than on specific government policies, the actual cause is secondary.
Maybe the weakness with Gorton and Metrick’s data that comes to mind first, is its apparent sole focus on structured securities. Even though US treasuries constitute the largest pool of repo securities in the US market (ICMA, 2013), Gorton and Metrick did not offer data on US treasuries. The effect on repo haircuts might be overestimated, since government securities are not as much affected by uncertainty about their value as structured bonds might be.
The fact that 50-60 percent of the outstanding US repos are structured as tri-party transactions (ICMA, 2013) implies the possibility that a considerable share of tri-party repo investors decide to stop financing a dealer rather than demanding higher haircuts (Task Force on Tri-Party Repo Infrastructure, 2009). This argumentation is supported by the findings of Copeland et al. (2012), whereby during the crisis, haircuts in the tri-party repo market did not increase (a lot).
While Krishnamurthy, et al. (2014) also computed a sharply contracted repo funding for transactions using non-treasury securitized assets as collateral, they do not support the “broadbrush picture” (Krishnamurthy, et al., 2014, p. 55) of Gorton and Metrick. This is, because for the shadow banking system repos are only a small share of the short-term funding of securitized assets, they argue. Others argue that repos play a major role in shadow banking (Penney, 2011) and are crucial for short-term funding (Mancini, et al., 2014).
An additional lack of clarity is the difficult transfer of the results from Gorton and Metrick (2012) to the European repo market, which is structured differently from the US market. One reason is that 80 percent of the repos in Europe are collateralized with government bonds. The other major reason is, that the European market has much longer-lasting maturities, than the largely overnight US market, making haircuts less important in Europe (ICMA, 2013).
Overall, Gorton and Metrick (2012) might have overestimated the dynamics of the repo market in the form of deepening haircuts leading to a deleveraged US financial system and a decline in the US money market’s liquidity. However, probably both excessively increasing haircuts and a reduction or withdrawal of credit facilities by market users had pushed the repo rates and had played a large role in the US crisis development - even though this role might be less relevant than Gorton and Metrick concluded. In the end, counterparty risk and uncertainty regarding the collateral-valuation seems to have driven a big share of the growth in repo rates and haircuts.
3.2. Margin calls and bankruptcies in the repo market
Rising repo haircuts played an - at least moderate – role when it comes to the crisis development. Nevertheless, the way repo haircuts, collateral prices, and stability buffers relate to liquidity shocks and bankruptcies is yet to be seen. In fact, financial instabilities due to an excessive use of margin calls already happened in the past. Namely, Lehman Brothers’ liquidity situation worsened as it faced daily margin calls, because the price of its mortgage-backed securities decreased.
Even though Gorton and Metrick (2012) did not find market liquidity to be able to explain a big share of the interest rate spreads, Schwarz (2014) concluded that in this respect market liquidity effects are even more important than credit risk is. Because Schwarz (2014) used a spread that includes two bonds with identical credit risk but different liquidity, her measure might be more accurate when it comes to this risk weighting, instead of the focus on the LIB‑OIS. Using this as an occasion, the underlying paper from Laura Valderrama (2010) checks on a technical basis whether and how the asset’s liquidity increases solvency risk within a repo transaction and how systemic risk can cause the failure of a repo borrower. The model illustrates the risk that the cash lender may face a liquidity shock, forcing it to liquidate parts of the underlying security, corresponding to a margin call for the repo borrower,
Because repo lenders may use the received collateral to pledge it in an additional contract (ICMA, 2013), lenders can become borrowers themselves. This connection makes margin calls dangerous for an even bigger share of the market.
The following part explains the risk of repo participants that face a margin call demanded from its creditor. It is divided into three different sections. Beginning with the introduction of the model of financial contagion, in order to discuss the main results in 3.2.2. Closed by the discussion in chapter 3.2.3, which will also address the question on how that study can be reviewed today - four years after its publication.
3.2.1. The model
Valderrama (2010) designed her analysis with three periods and four types of investors. The investors divide into a repo-market lender (repo buyer), N borrowers (repo sellers), a noise-trader and a deep-pocketed outside investor. The market participants will be introduced in the following section.
The regulated repo lenders use a 2-period debt to invest that cash at in one unit of a 2-period asset via a 1-period reverse repo from N repo borrowers for an individual amount of . The cash balances are funded in the 1-period unsecured market (wholesale or deposits).
Gorton and Metrick (2009) explained that, while the repo rate reflects the borrower’s risk of default, a haircut needs to be implemented to cover the risk of the asset changing its market price ( . Hence, the long-term asset is more risky, making the repo lender to pay a haircut and to hold enough capital to absorb changes in price.
The repo borrowers are N identical unregulated investment banks, each holding one unit of a 2-period asset . At , the security is valued at and it is financed by pledging it in a short-term repo transaction. Hence, the borrower faces a maturity mismatch in its balance sheet and the haircut = for the long-term asset used as collateral to fund the short term repo. Thus, the larger the haircut for the borrower, the more difficult it gets to finance the 2-period asset . Additionally, the repo borrower raises external funds for .
The noise traders are subject to a liquidity shock , forcing them to sell s units of the long-term asset to other noise traders. Where reflects the units of the assets being liquidated, due to a maximal liquidity shock. These trade flows initiate the asset’s price fluctuations. The buyers of those securities have the demand function . Where denotes the ability to trade quickly without affecting the price. Since under the market clearing condition equals , the lowest possible long-term asset price is -. Hence, the lender has to hold capital to absorb losses from the liquidation of the collateral when its price equals . This condition and the debtholders’ haircut constraint gives the minimum level of capital to be held at the initial time: .
3.2.2. Model results
A repo market becomes illiquid due to systemic risk, when a large part of the financial sector faces stress (Acharya, et al., 2010). Hence, systemic risk also covers the risk of a fast and deep infection from struggling financial institutions to other market players. In this section I present (the derivation of) Valderrama’s (2010) formal condition for financial contagion as a channel through which these instabilities in terms of a margin call can also trigger a run on the repo market, because the lender reduces its lending to repo borrowers.
When we assume that the haircut constraint is binding, i.e. , we can insert this equation for the minimum level of capital into , resulting in: . As the asset was initially valued at , needs to equal . Additionally, the maximum capital loss needs to be equal to the initial capital buffer . Hence, the lenders debt that will be invested in the asset needs to be strictly bigger than the possible loss . Aggregating the equations , , and leads us to
which denotes the condition for financial contagion to arise. This illustrates that financial contagion is more likely with a high liquidity of the asset, denoted by . For a high haircut for the 2-period security, the initial capital buffer is larger, but also a small liquidity shock may already trigger a margin call.
Bankruptcy of repo borrowers
Shocks may not only introduce instabilities into the market, they can lead to bankruptcies of some financial institutions. Defaults worsen the overall doubtful environment even further, such that the development can become systemic. But how is this bankruptcy probability changed by the level of the collateral’s price, the repo haircut or the institution’s interconnectedness?
A liquidity shock in the market for asset results in the partial deleveraging of asset , This means that repo borrowers will live on, if they manage to rollover their portfolio funding. To be able to do so, the borrower’s capital base has to be as big as the modified haircut constraint by the repo lender:
The capital base can at least cover the partial deleveraging of the asset , preventing the repo borrower from the case of bankruptcy.
The borrowers deleveraging is denoted by , where the cash transfer represents the exceeding short-term debt over the short-term asset . When we solve equation (7) for and implement the cash transfer, we get:
The price can be interpreted as the minimum price that prevents the repo borrower from its liquidation. If the price in is too low, the repo lender will take possession of the repo collateral prompting the borrower’s bankruptcy.
Still, these equations do not give any implications on the needed haircut range for neither a high enough haircut to protect the lender from a drop in the collateral value nor on the haircut a borrower is able to pay. In order to implement these relations into a single equation, we solve (from above) for and use as the collateral price for the high level equilibrium; where = decreases with the margin call demanded by the lender and the risk-aversion of the deep-pocketed investor that buys the liquidated asset. This determines the right-hand-side, while the left-hand-side results from = , for in case of the low level equilibrium:
Under a liquidity shock in asset the haircut level has to be sufficiently high to balance a possible collateral devaluation. While the repo borrower faces an additional bankruptcy risk for high intermediate values of the haircut, it can leverage furthermore increasing the vulnerability to a margin call if the haircut is set too low.
When using the left hand side to take a look at liquidity shocks again, the relationship between haircuts and the propagation of liquidity shocks can be illustrated in more detail. Hence, replacing in equation (9) results in the estimated haircut equal to:
Akin to the liquidity shock being too large, a very low price of asset gets the repo borrower into trouble due to the increasing haircut. Moreover, the lender’s risk-aversion pushes the haircut just as a low asset liquidity does. These connections illustrate that the haircut depends on the repo lender’s risk attitude and on how sellable the security is. Thus, the haircut is mainly influenced by factors, which directly affect the repo lender, rather than the repo borrower. In addition, an increase in the haircut demanded by the lender for the long-term collateral, leads to a lower haircut for the short-term asset.
In addition, the equation (8) for the minimum price, that prevents the repo borrower from bankruptcy, can be extended with the cash transfer and the lender’s risk-attitude such that it provides conditions for a margin call resulting in a failure of the repo seller:
It is more likely that the repo seller may go bankrupt, if the interconnectedness of the repo buyer is large. Moreover, an increase in the long-term asset’s haircut has two effects. On the one hand it lowers the expected margin call, reducing the effect of this liquidity shock. On the other hand will decrease and consequently diminishes , resulting in a smaller buffer for the case of a solvency shock.
Summarizing the results, Valderrama (2010) found implications about the collateral’s price, the repo haircut, the capital base and financial contagion. Furthermore, she anticipated the statement from Gorton and Metrick (2012) suggesting that an increase in haircuts is equivalent to withdrawals from the banking system, because both events are strongly connected to liquidity risks. An empirical analysis by Valderrama (2010) fortified this proposed relationship even on a global basis. It confirmed that the grown liquidity pressure in the US interbank market, due to the declining real estate prices, diminished bond prices for emerging markets.
Another aspect is the effect of fire sales of assets, which start to become system-wide. Because the maximum capital loss of the borrower, due to margin calls, depends only on the number of traders liquidating the asset and the asset’s trading speed, the systemic liquidity issue spreads as the borrower’s capital buffer needs to increase with the number of fire sales.
In addition, the author illustrated the rising probability of bankruptcy of the repo borrower with an increasing interconnectedness in the system. This is a suggestion some researchers could not affirm, such as Adrian and Brunnermeier (2009) who did not find a change in the systemic risk for the case of a large institution being split into n identical segments.
As the research has been published in 2010, the progress dynamics of the financial crisis might make the paper appear insignificant for recent interpretation. When we have a look at the development of the repo market for the last four years, the share of tri-party repos with US Treasuries increased from 30.2 percent in June 2011 to 39.9 percent in June 2014 (Federel Reserve Bank of New York, 2014). In this period, the usage of agency MBS decreased from 32.1 to 28.4 percent. Thus, the gain in US Treasury collateral-usage outweighs the drop in the mortgage-backed securities issued by government agencies. Both government bonds and agency MBS are highly liquid (Campbell, et al., 2014) and less liquid assets like Agency Debentures and CMOs decreased its share. On the basis of this development towards a bigger share of collateral with high liquidity, the model by Valderrama (2010) would suggest that the today’s tri-party repo market does not need as high bankruptcy-avoiding repo haircuts for the short-term contract as it was the case with less treasury-usage. However, the Federel Reserve Bank (2014) provides data showing that margins levels did not change at all – staying at 2 percent (median value) for US Treasuries and Agency MBS in the respective period.
Nevertheless, there are two reasons why it is not justified to conclude that in case of tri-party repos the new repo market structure mitigates the significance of the relationships provided by Valderrama. Firstly, changes in parameters like the lenders risk attitude and the haircut for the long-term contract may balance the effect of an increasing liquidity. Secondly, the anticipated haircut decrease depends on the collateralized asset used for the short-term contract, which is hard to observe due to the fact that a big portion of repos are arranged on a bilateral basis, especially in Europe, and is not transparent in this respect even for tri-party repos.
In the end, after four years, the model remains meaningful. In fact, the paper does still provide important results for economic policies and further research. Mainly because it is based on rational investment-behavior-framework and economic restrictions which do not change dramatically. Hence, Valderrama’s (2010) equations continue to be applicable to provide relations between different repo parameters and will therefore be revisited in chapter 4 to highlight implications of regulatory policies on the repo market.
When it comes to findings with those regulatory implications, two main conclusions can be drawn. On the one hand, the suggestion of mandatory repo haircuts can mitigate the broad effects from fire sale liquidations. On the other hand, the illustration that systemic risk changes if a large firm is split into n identical institutions, questions the view that those clones are less systemic than the former structure was.
In the previous chapters, I explained the repo’s development and role during the financial crisis and events of systemic shocks. This, obviously, opens up the question on how to regulate the repo market, so that instabilities and shocks can be mitigated effectively to prevent crises.
In their research, Gorton and Metrick (2012) described two systemic events causing the instabilities in the market. Firstly, the shock to the repo market, illustrated by the LIB-OIS as a proxy for counterparty credit risk. Secondly, a rise in liquidity risks due to an overuse of margin calls, resulting in the failure of Lehman and increasing haircuts even for the safest securities. In this section, I will use their framework to evaluate the Basel III regulation (Basel Committee on Banking Supervision, 2011a) and the Dodd-Frank Act (U.S. Securities and Exchange Commission, 2010). In addition, I will answer the question whether counterparty and liquidity risks are mitigated successfully.
Moreover, both regulations will be reviewed by means of a technical analysis. In this approach, I will mainly use the model by Valderrama (2010) to give precise ratios and requirements that should be fulfilled to reduce the risk of a bank’s failure and to minimize systemic liquidity risks for the financial sector.
After the discussion on whether and how resolved policies regulate the repo market, the ensuing section will use those results to present a regulatory revaluation, regulation recommendation and outlook on the repo’s possible future.
4.1. Basel III
The Basel III accord is the first regulatory policy I will consider. The Basel Committee agreed upon Basel III at the end of 2010. It was scheduled to be implemented from 2013 until 2015 by the participating countries. While it is not the former, it is the one which offers more precise and technical approaches compared with the Dodd-Frank Act, therefore it is more significant for this analysis. To avoid pro-cyclicality, the closing process of implementation is delayed until its finalization in 2019 (Basel Committee on Banking Supervision, 2011a).
Aside from regulating haircuts and capital requirements, Basel III offers different rules directly impacting the repo market. Breaking down the main objectives to two pillars, they can be named as: firstly, the loss and shock protection via capital requirements and strict definitions of the calculation methods. Secondly, the proposal of two new quantitative minimum standards with different risk horizons to improve the liquidity management.
The capital requirements of 8 percent changed from the former uniform calculation to a formula with a focus on the tier 1 capital share. The ratio can now be increased up to 15 percent via additional buffers for all banks plus extra buffers for systemically important financial institutions (SIFI). Additionally, Basel III defines the capital tiers on whether the bank shall continue its activities or if it only needs to be able to repay lenders and senior creditors, which has not been defined by neither Basel I nor Basel II.
The second pillar, the liquidity requirements, can be divided into two parts. On the one hand the Liquidity Coverage Ratio (LCR) (coming in 2015) wants banks to survive at least 30 days without new liquidity. This avoids a case similar to Bear Stearns, which did not get any more money from banks and through repos, resulting in a bankruptcy situation. And on the other hand the Net Stable Funding Ratio (NSFR), or Northern Rock Rule, by which the stable funding must be at least equal to illiquid assets to achieve a surviving despite an assets-liabilities maturity mismatch.
Both Basel III approaches, the capital ratio and the liquidity requirements will be assessed by means of the liquidity shock model introduced by Valderrama (2010).
Additional Basel III propositions cover supervisory haircuts and a central clearing of repo. Paragraph 61 from the Basel III accord allows banks to choose between those defined initial margins and a rate calculation based on the effects of credit risk mitigation under the financial collateral comprehensive method. But supervisory haircuts are mandatory for bilateral repos with non-cash collateral, which is transformed into cash-equivalent.
Basel III is also trying to transform a majority of the bilateral repos into cleared tri-party repos. The new clearing requirements will therefore give an incentive to trade repos via central counterparties.
4.1.1. Prevention of a run on repo
In the light of the recent financial turbulences the call for regulatory interventions increased significantly, not least in consideration of systemic shocks. Besides the Dodd-Frank Act, the Basel III accord is trying to mitigate counterparty risks as well. Namely, the management of mandatory haircuts and a reliance on tri-party repos may prevent a party’s default and minimize the effects for its counterparty. Whether the attempts made by the Basel III regulation avoid systemic withdrawals from the banking sector, or diminish consequences and scope of bankruptcies, will be discussed in the following section.
Supported by Gorton and Metrick’s (2012) findings, the Basel III regulation has the key for the shock in the repo market in mind: the haircuts. Because tri-party repo offers additional safety - as the operational risk is transferred to a clearing house - this regulatory approach applies to bilateral repos only.
If collateral risk increases, tri-party repo lenders are more likely to give up on their lending activity, while bilateral repo lenders may continue lending - but with higher haircuts. Hence, an effective regulation of the bilateral market would need to set those haircuts to a specific level, or decrease its volatility through other channels. As a big share of the repo market has OTC character, the administration situation for supervisors is difficult. Nevertheless, the Basel Committee decided to define supervisory haircuts for those bilateral repos depending on the rating of the asset class.
Dependent on the collateral’s maturity and rating, paragraph 151 the Basel Committee defines the specific haircuts:
TABLE 1: Supervisory haircuts for asset classes
illustration not visible in this excerpt
Standardized supervisory haircuts in percentage (%).
Assumptions: daily mark-to-market, daily remargining and a holding period of 10-business day.
(Basel Committee on Banking Supervision, 2011a)
While a worse collateral rating or expanded maturity increases the supervisory haircut, both lead to growing securitization exposures, corresponding to the exposure to a special purpose entity. Nevertheless, when fixing rates, although affected by many different parameters, factors like risk or market-changes try to find other channels to price the collateral accordingly. While Gorton and Metrick found that counterparty risk does not affect repo rates, margin calls might replace haircuts reflecting this risk. However, the margin call is an instrument in case of a needed revaluation of existing repo contracts after the sale day, rather than for the change of proxies for risk, such as the LIB-OIS or its volatility.
Even though they did not recommend specific rates or formulas, Gorton and Metrick’s (2012) empirical model suggests that a change in the counterparty risk (the implied volatility as a proxy) has a significant effect on the repo haircut. Namely, a rise in the implied LIB-OIS spread’s volatility of 1 percentage point of an A-AAA rated asset backed security, made up of auto and student loans, results in a haircut increase of 0.036 bps (Gorton & Metrick, 2012). Since this development cannot be balanced in a fixed haircut framework, the incentive for revaluation is reflected by 0.00036 percent of the collaterals worth being notionally demanded from the repo borrower.
As Gorton and Metrick argued, the counterparty credit risk was one of the main drivers of changes to repo haircuts. Thus, it is questionable whether supervisory haircuts, dependent on only maturity, rating and issuer, are set appropriate while ignoring other repo parameters. In addition, fixed haircuts might lead to re-pricing methods after the repo sale date, like a contract clause allowing specific margin calls in case of parameter changes by means of a common model. Regulatory haircuts are also covered by the Dodd-Frank Act, which will be evaluated more technically in chapter 4.2.
Central clearing of repo
The setup of supervisory haircuts rather treats the symptoms than fights the causes. As contagion risk triggers a haircut increase, the fighting of causes would mean to reduce this risk instead of prohibiting particular market haircuts. Paragraph 97 of the Basel III accord introduces a series of new adjustments and measures related to counterparty credit risk.
Copeland et al. (2010) showed this, during the financial crisis, a tri-party framework prevented repo haircuts from increasing. Thus, an important proposal for repurchase agreements is the movement to tri-party contracts, such as with central counterparty (CCP) clearing houses. Reviewing the European Parliament resolution of April 16th 2013, Villafranca et al. (2013) expressed their belief that Basel III wants to increase the incentive to trade repos via a CCP. By making capital requirements for bilateral repos more challenging, institutions may clear more derivatives. This transfers the counterparty risk to the clearing house, because a CCP acts as an intermediary and manages securities independent from the financial situation of the repo parties (Villafranca, et al., 2013).
When a dealer bank lends money to a hedge fund via repo and then uses this security to buy a repo, the first dealer bank would have a net of zero (ignoring haircuts). However, when it comes to a bank run, the lenders want to withdraw their cash. But the dealer bank cannot instantly get these resources from the hedge fund. To make necessary resources available, Gorton and Metrick argue that a repo trade through a clearinghouse would be a possible solution.
This is applicable as well for the run on repo, where the withdrawal is reflected by the rising haircuts due to uncertainty about the solvency of the repo buyer. In this case, an AAA-rated CCP would guarantee the return of the lender’s money because it holds the security itself (Penney, 2011). Therefore, the haircut increase, driven by counterparty risk, would not be needed, because the collateral does not include the repo borrower’s solvency situation anymore.
With a repo market based on cleared trades, the run on securitized banking, as Gorton and Metrick’s defined it, would not have had the same big impact. In addition, more features of the CCP structure are discussed in chapter 4.2.
4.1.2. Reduction of liquidity risks
The liquidity situation of a financial institution is crucial when it comes to spill-over effects and systemic risks for the whole market. In reality margin calls may occur more often than expected, thus intensifying the liquidity situation of the repo borrower.
Basel III is a package with several different regulatory approaches, which try to cover the liquidity risk due to shocks. Some of the major features to increase the protection against shocks are capital and liquidity requirements. These propositions will be reviewed with reference to liquidity and solvency constraints emerging from a rational investment and treasury framework introduced by Valderrama (2010).
Besides regulatory haircuts, which constituted one of the main recommendations by Valderrama’s model, another important regulatory approach is a capital buffer to balance margin calls. When introducing a mandatory regulatory capital, equity must be held as a percentage of the risk-weighted assets. Valderrama (2010) used her equations derived in chapter 3.2. to calculate this additional capital base.
The capital buffer needs to cover the liquidity shock to the extent that the situation were the same as if no shock had occurred. Therefore, the regulatory capital needs to equal the maximum loss given the default. The loss that endangers the bank’s solvency is given by:
which needs to be differentiated with respect to the initial shock , leading to:
and when solving for s:
Where derives the units of the long-term asset being sold for which the capital loss is at its maximum with the probability of occurrence. The additional regulatory capital , which ensures the bank’s solvency, is determined by inserting into . The results are:
The relationship suggests that the capital buffer needs to increase in case of a serious liquidity shock or when the asset’s liquidity is low.
To provide an assessment on the capital requirements demanded by Basel III, I will use this equation to compare the regulatory capital , proposed by Valderrama (2010), to the one introduced by the Basel accord. The numerical example uses the following parameters:
In the model, the capital buffer would need to reach 1,45 percent to cover the loss after a possible margin call. Because the chosen haircut at 2 percent is the median margins level for US Treasuries and Agency MBS (Federel Reserve Bank of New York, 2014), the Basel III framework would demand to hold 0 percent for that debt. Thus, in case of a liquidity shock, the bank would have no regulatory capital under Basel III, even though it would need over one percent based on this technical model. Although government bonds are seen as a save asset compared to others, the risk weighted capital requirements under Basel III might be set to low for assets with a low risk profile.
This interpretation changes for repos backed by ABS without an investment grade, where the median haircut is at 8 percent (Federel Reserve Bank of New York, 2014). In this case 6,89 percent (for ), which is not entirely covered by the minimum tier 1 capital ratio of 6.0 percent, but by the total capital buffer of at least 8.0 percent (Basel Committee on Banking Supervision, 2011a).
Of course, the overall regulatory capital is dependent on inter alia the diversification, maturity and the market risk of the portfolio, which is not covered with that simplified model. Generally, an approach with regulatory capital can constitute a proper measure to reduce liquidity risk in repo markets - but the estimation’s value relies on the model and definitions.
As the Basel Committee on Banking Supervision (2011a) wrote, a regulatory capital is a required, but not a sufficient condition for banking sector stability. Another important factor of the Basel III accord is the setup of liquidity ratios. To date, there have not been harmonized liquidity standards on a global basis. With the ongoing adoption of the Basel III rules, the liquidity requirements can help to prevent a competitive race to the bottom by proposing international standards for financial institutions across different regions.
Since funding via repos is the major factor in the short-term money market (Mancini, et al., 2014), the LCR features a good approach for the model to offer a more significant informative value for repos, than the NSFR - which rather describes maturity mismatch.
Because liquidity risk is defined as the probability of a funding gap between the value of the security and the exposure, liquidity risk does not apply to highly-liquid securities like government or agency bonds. This definition also implies that a Liquidity ratio would need to balance a possible liquidity shock due to illiquidity. In the case of a triggered margin call, the stable funding has to equal the value of this liquidity shock (Valderrama, 2010):
The liquidity buffer is positively correlated with illiquidity (equivalent to a shrinking ) and with the liquidity shock’s volatility (determined by ). A rising haircut would lead to a lower level of the liquidity ratio needed.
In case of a repo backed by assets with AAA-rating and its haircut ranging at 2 percent, the liquidity ratio model (with the same parameters as before and ) would require to hold 14 percent of the collateral’s value as high-quality liquid assets like government bonds.
Against the background of the LCR formula and parameters looked at as an example of the Basel Committee (2011b, pp. 2-8), I applied specific bank’s values for reserves and AAA-rated repo collateral of $5b each, a security’s haircut of 2 percent, $0b ($5b) of AAA-rated exchange traded equities and $15b ($20b) of total net cash outflows. The resulting LCR of 99.34 percent obviously takes an entirely different approach compared to Valderrama’s model. The much higher ratio can be attributed to the calculation method whereby the overall stock and the cash outflow play an important role. Furthermore, the Basel III calculation is highly responsive to those parameters. For instance, the LCR changes to 74.63 percent if the bank holds additional stock (values in the brackets) – such as exchange traded equities in an AAA-rated company - with the same value of the repo collateral.
Those differences and the changes due to the stock structure suggest that the liquidity requirement proposed by Valderrama might be oversimplified. In particular, the model does not imply the situation of the bank’s portfolio or the prospective outflows. Valderrama argued furthermore, considering the bank’s cash inflows, that the bank has repos with illiquid collateral uncovered by the liquidity ratio, even though the model shows that illiquidity increases the need for a liquidity ratio. However, the Basel III accord does not demand a liquidity buffer for not rolling-over illiquid assets to liquid collateral, because it assumes that repos covered by illiquid securities do not change the liquidity situation tremendously.
Thus, it is unclear how the Basel III liquidity ratio decreases systemic liquidity risk in the repo market. Moreover, it is difficult to evaluate regulations on the liquidity ratio, because analytical repo models are very responsive to parameters which lack in transparency. In consideration of the current policy design, it seems that the Basel Committee shares the opinion about the importance of a liquidity ratio, but it does not have the repo market in its major field of vision.
Additionally, the Basel III framework includes several approaches to prevent pro-cyclical effects due to the regulation policy. Firstly, national authorities will monitor risk indicators for system-wide trends and will be able to react accordingly with countercyclical policies. Secondly, private banks will calculate their specific buffer requirement by means of their credit exposures. Thirdly, restrictions will be implemented if those banks do not meet the requirement set by the demanded countercyclical buffer. Finally, the conservation buffer, set at 2.5 percent of risk-weighted assets (FSOC, 2011b), promotes the buildup of capital in good times that can be called for in stressful periods.
The Basel Committee on Banking Supervision (2011a) named the pro-cyclical dynamics the failure of risk management and capital frameworks prior to the crisis. The paper from Valderrama does also imply some dynamics related to cyclical movement. In fact, as the equations on the liquidity and capital buffer move negatively with the pro-cyclical haircut, the buffers behave anti-cyclical.
This means that the different buffer requirements of the Basel III accord lead to decreasing haircuts. Because, as illustrated by Gorton and Metrick (2012), periods of financial distress with uncertainty about the market liquidity result in increased repo haircuts, the regulation has an anti-cyclical effect.
4.2. The Dodd-Frank Act
The Dodd-Frank Wall Street Reform was signed into federal US law mid-2010. Ryan (2013) even described the Act as the most extensive legislation for investment banks since the New Deal. Even though Acharya et al. (2010) criticized the Dodd-Frank Act for not offering a significant mention of the repo market (repos are even exempted from the relevant Volcker Rule in Section 619), the reform still provides several channels through which it controls repos as well. The fact that the Financial Stability Oversight (FSOC), a Dodd-Frank Act entity, is repeatedly covering tri-party repos, shows that the execution of the Dodd-Frank Wall Street Reform does have repos in mind. To react to the financial crisis, the US Treasury’s proposal started the process to reform the financial regulatory system.
An important feature of the Dodd-Frank Wall Street Reform, according to Hoshi (2011), is the establishment of the Financial Stability Oversight (FSOC) as a systemic risk regulator. The FSOC monitors and regulates even nonfinancial companies (if they are seen as systemically important), and as the case may be, it responds to systemic risks. In fact, Section 113 of the Dodd-Frank Act attributes financial instability to maturity mismatch and reliance on short-term funding (Public Law, 2010). Although this new responsibility is an important feature, there are also other regulations, which might affect repos more profound.
The Federal Deposit Insurance Corporation (FDIC) received new jobs as well. Critical for repo contracts is the task to set up regulatory capital ratios, which may affect counterparty credit risk. Furthermore, the FDIC may unwind SIFIs when they are in financial difficulties. When it comes to repos, in a case of a default the FDIC would sell the collateral to an acquiring firm, or give it to a temporary bridge bank (FDIC, 2011). A process like that would mean that lenders are not permitted to sell the collateral from the repo contract made with the bankrupt repo borrower. This makes repos more similar to collateralized loans and might have significant effects on the counterparty risk of repo contracts.
Another implication coming from the FDIC is the possible charge of higher haircuts. This may happen since banks will have to pay deposit premiums on all types of debt, including repos. Whether those increased haircuts and the transition towards more tri-party repos, which are traded for instance via a CCP, mitigates the risk of liquidity shocks, will also be addressed subsequently.
4.2.1. Prevention of a run on repo
Gorton and Metrick (2012) concluded, that it was the interbank counterparty risk rather than the subprime housing market, which had caused the decrease in the price of securities. Especially large non-bank financial institutions experienced these effects. With this rise in the doubt about the bank’s stability, the haircuts increased significantly, resulting in a run on the securitized banking sector. In particular, the volatility concerning the LIB-OIS and the counterparty risk are starting points for regulators in order to prevent repo market shocks.
With capital requirements and the creation of a resolution process for financial institutions, I name just a few of the Dodd-Frank Wall Street Reform propositions that control factors, which can trigger or enforce counterparty credit risk. Subsequently, I will discuss whether those capacities provide a significant influence to prevent systemic shocks by means of the framework highlighted by Gorton and Metrick (2012).
While a ranking as undercapitalized would mean to cast doubt about the solvency, a financial institution has an incentive to reduce its capital levels, because of the high cost of capital (Ryan, 2013). Therefore, a regulatory capital requirement is needed in order to make banks satisfy the requirements for a stabilized market.
Because the findings by Gorton and Metrick were greatest at non-banks, they state that significant parts of the formal banking sector moved to non-bank financial companies. This clarifies the importance of regulation not only focusing on the traditional banking sector. In fact, the Dodd-Frank Act wants the FDIC to establish minimum risk-based capital requirements also for non-banks. In section 171 the capital requirement’s objectives are defined as to – at least - address risks explicitly arising from repos and concentrations in a too big to fail classification and assets valuated based on models. Moreover, the requirement should not be less than the “generally applicable” (Public Law, 2010, p. 1436) capital ratio based on risk. Because Basel III became the generally applicable minimum capital requirement, the FDIC needs to set up rules, which at least build on the Basel III accord (FDIC, 2014). The agency defines each of the capital ratio standards that an institution needs to satisfy to be well/adequately capitalized, elsewise it is ranked as undercapitalized:
TABLE 2: Proposed capital requirements
illustration not visible in this excerpt
Capital ratios in percentage (%). Ratios are effective from January 1, 2015.
A higher capital to work with in periods of distress lowers the default risk for the counterparty. Even though, Gorton and Metrick did not include capital requirements in their analyses, the LIB-OIS measure is a proper proxy for counterparty risk. Hence, the LIB-OIS spread is expected to decrease with an implementation of higher capital requirements. However, since the new ratios will take effect in the beginning of 2015, it is not yet possible to assess this policy accordingly. However, as Schwarz (2014) showed with her time-series regression, changes in credit risk can explain parts of the LIB-OIS movement:
Where determines the LIB-OIS spread and is the daily average of intra-daily credit-tiering to proxy credit risk, both measured on day . represents a spread on day between highly liquid German federal government bonds and less liquid KfW agency bonds, denoting the market liquidity. A one percentage point increase in results in a one-month LIB-OIS spread increase of 2.6 percentage points (3.28 percentage points in the three-month LIB-OIS). The effect, while also significant at the one-percent-level, is not important for this assessment, because it reflects bonds which would have a risk-weighting of zero and it is rather about liquidity than counterparty risk.
Whereas Gorton and Metrick (2012) argued that the LIB-OIS is affected by the credit risk alone, the results by Schwarz suggest that the market liquidity does also drive questioning the spread’s information value regarding solely credit risk. Because Schwarz was rather evaluating the early stages of the crisis, a conclusion might be that the market liquidity was a major factor for the LIB-OIS spread early on (Gorton & Metrick, 2012), with the counterparty credit risk steadily gaining significance during the financial crisis.
The proposed capital requirements will decrease counterparty credit risk due to a higher capital base, which will lower the LIB-OIS spread at least after the initial crisis period. According to the findings by Gorton and Metrick (2012), this decrease in the spread means that the probability of a return of the run on repos is successfully lowered by the capital buffer introduced in the context of the Dodd-Frank Act Wall Street Reform.
The previous attempts that should lower counterparty credit risk by Basel III and the Dodd-Frank Act were focused mainly on prevention, than on ex post reaction. Although a preventive regulation is key for market stability, bridge banks offer a different credit risk approach by improving the collateral administration of (pending) bankrupt financial institutions. For the case of a bank’s financial failure or default, the FDIC can establish a bridge bank to maintain systemically important operations (FDIC, 2011). The bridge company may buy securities to avoid their liquidation, protecting the collateral’s value and overall stability. The FDIC needs a clear authority to implement a liquidation facility, because private institutions have no real incentive to create a bridge bank on their own (Adrian & Ashcraft, 2012).
According to the FDIC this process may provide several benefits: certainty to the market, financial stability and maximization of the asset’s sale value. But does it also lead to a decreasing counterparty credit risk?
Since the bridge bank takes over the failed institution and continues certain essential operations during the resolution process, it eases the consequences and may ultimately avert bankruptcy. These effects would decrease the counterparty risk that would have pushed the haircut elsewise, because the probability of the default would be higher in case of no support during the resolution process. This certainty about the improved process should decrease the haircut. However, Gorton and Metrick (2012) did not address the idea behind bridge banks at all, therefore the conclusions based on this evaluation remain vague.
4.2.2. Reduction of liquidity risks
The research from Valderrama (2010) was focused on a macroeconomic framework proposing facilities to minimize liquidity risks. As shown by the author, an increase in margin calls due to liquidity shocks may bring the repo seller into financial distress. This situation can make the whole repo market unstable, when combined with high counterparty credit risk. Two ideas out of the Dodd-Frank Wall Street Reform that may reduce the liquidity risk, are a secured creditor haircut and a tri-party repo framework based on clearing houses. Both propositions are known from the alike approaches of the Basel III regulatory framework, but this time they are evaluated by means of Valderrama’s model.
Secured creditor haircuts
Under Section 215 of the Dodd-Frank Act, the FSOC has been instructed to evaluate the potential impact on secured creditors if a haircut change was imposed. The published FSOC study suggested to give the power of decision about the haircut amount to the entity responsible for the resolution of the (if so) failed firm (FSOC, 2011b). This would facilitate the FDIC to decide over the haircut to a specified limit - on a case-by-case basis. Additional approaches from the FDIC are affecting repos by increasing deposit premiums on all types of debt.
Thus, it is likely that repo haircuts change in the US due to regulation rather than as a market solution. Having this in mind, the following technical evaluation of the secured creditor haircut intervention will show whether this is beneficial for the repo market under the framework presented by Valderrama (2010).
As determined in chapter 4.1. the additional regulatory capital would need to cover the margin call at the highest probability of default. This relationship resulted in:
When setting the capital buffer equal to zero, solving for the haircut leads to:
The regulatory haircut gives the condition under which the bank’s capital buffer and credit rating is unchanged. Because the haircut was determined under the capital buffer assumption, the secured creditor haircut policy is akin to the requirement of additional regulatory capital.
If the haircut increases above the determined level, the needed buffer is negative. A negative capital buffer means that, for the case of a retained liquidity, additional uncovered debt can be absorbed at no extra cost. When the additional regulatory capital is positive (smaller haircut than posed), the credit rating of the bank will downgrade, making the additional debt more expensive and forcing the bank to add on capital buffer.
As Valderrama (2010) argued, an inadequate haircut increase may cause a rise in systemic risk. Against the background of the FDIC mandate to set those haircuts, it is essential that the FDIC has every information needed to determine the optimal haircut. Reflecting this issue to the proposed haircut model, the FDIC would need to know the size and probability of liquidity shocks, the collateral’s liquidity and its valuation. While it is likely that the FDIC obtains valid data about the latter, the entity will struggle to determine the security liquidity and details on the potential shock in particular. In their annual report, the FSOC (2014) supported this assumption, assessing the regulators to be unable to effectively analyze all repo markets.
Whereas Valderrama proposed mandatory haircuts for repo securities to mitigate the effects from the fire sale liquidation of collateral via haircuts, I think the implementation is not very simple. It is doubtful that an external agency is able to set haircuts more appropriate than the market mechanism would – especially with a haircut calculation based on a model less simplified and, thus, with a lot of data required. Hence, liquidity risks should be addressed by other regulatory measures as well.
Central clearing of repos
Besides regulatory haircuts, and alike the Basel III accord, the Dodd-Frank Act also wants to transfer the derivative market into a cleared framework with three parties involved. As demanded in section 813 of the reform, the Board of Governors of the FED, the Commodity Futures Trading Commission and the Securities and Exchange Commission published a report about the Risk Management Supervision of Designated Clearing Entities. The report proposes core principles for the derivatives clearing organization to address, including inter alia risk management, default rules and procedures, system safeguards, reporting and settlement procedures (Board of Governors of the Federal Reserve System, et al., 2011). This framework applied to repos, reduces risk exposures and liquidity hoarding, as argued by ICMA (2013).
Even though the research from Valderrama does not mention a central clearing of repos, according to the ICMA (2013) a CCP can change the liquidity risk situation by concentrating risks in the CCP institution. The effects of a CCP framework on margin calls are key to assess the Dodd-Frank Act regarding market liquidity risk, because margin calls significantly define the liquidity situation of the repo borrower. But since Valderrama’s model does not cover the clearing measure, though it is a major component of the Dodd-Frank policy and many studies ascribed various benefits to CCPs, the effects of clearing on liquidity risk are presented based on the analyses by Pirrong (2012).
Pirrong referred to the Black Monday from 1987, where the Federal Reserve intervened in order to keep the central counterparty Chicago Mercantile Exchange alive. Prior to this, the CCP faced $2.0 billion intraday margin calls, leading to a delay in the actual payments, and thus releasing rumors about the central counterparty to default. This case illustrates the risk when concentrating liquidity issues on one single clearing house, which has also been noted by the ICMA (2013).
Furthermore, Pirrong questions the Fed’s ability to repeat such support, because the central bank became more careful with bailouts and moral hazard. Moreover, many CCPs clear in different currencies, requiring coordinated liquidity support in case of market stress.
The risk is not only due to the CCPs liquidity, the bank offering the needed credit lines to the CCP might also be in distress. For this case, or if the bank has concerns about the CCP’s solvency, the central counterparty will struggle to receive sufficient funds.
These concentrated liquidity risks are akin to those in a bilateral repo market allocated between the repo parties, but they also reflect the unsafe effects of CCPs. Because a liquidity risk transfer from multiple repo dealers/investors to a single central clearing party does not mean the risks have been reduced in any way, the CCP fails to address the liquidity risk issue.
4.3. Discussion and outlook
In the case of financial innovations, “there are positive and negative externalities. Regulation must try to constrain the latter but, at the same time, nurture the former” (Comotto, 2012, p. 25). I will address the question, regarding effects on negative externalities, whether regulatory policy achieves this aim and what precise measures should be applied. An outlook will then forecast possible consequences for the development of repos due to regulatory changes.
The Basel Committee leaves the supervisory framework at the nation-wide regulation, but it recommends much more precise policies than the Dodd-Frank Wall Street Reform does. Even though, a detailed regulatory measure might help hypothesizing that it is highly reasonable, I will show that this conclusion does not apply to every policy. In fact, only a few approaches result in verifiable positive effects on repo market risks and on sources of instability. Others have to be evaluated furthermore due to their recent design and implementation methods.
4.3.1. Summarizing evaluation
Both the Dodd-Frank Act and the Basel III proposals will likely make repurchase agreements more expensive and increase haircuts. This does not mean that those effects are necessarily harmful for financial markets. However, because there is still the need of repo finance running smoothly to ensure daily functioning of financial markets, regulators should act with caution. Based on the analysis from the chapters 4.1 and 4.2., I will therefore shortly assess capital requirements, regulatory haircuts, central counterparties, liquidity ratios, countercyclical buffer, and bridge banks.
Capital requirements are the central component of reasoned regulation (Edwards & Mishkin, 1995). The grave capital ratio measures in the Dodd-Frank Act and even more in the Basel III framework support this assessment. By means of the models explained earlier, capital ratios seem to keep their significance today.
In general, regulatory capital constitutes a proper approach for liquidity risk reduction associated with repos. Nevertheless, for securities with a very low underlying risk (e.g. government bonds) and governmental ambitions aside, the capital requirement of zero is set to low. Additionally, the counterparty risk in repo markets is lowered with the introduction of adequate capital requirements. Since regulatory capital directly covers repo risks, as the source of financial distress, it constitutes a preventive measure.
Whereas some papers, like from Copeland et al. (2012), concluded that tri-party repo haircuts did not rise during the financial crisis, the Dodd-Frank Act and the Basel III proposals do include regulatory haircuts. Nevertheless, the importance of this regulation is backed by the findings of Gorton and Metrick (2012), and Valderrama (2010), but it actually lacks in the embedding of the repos complexity.
As already assessed by the FSOC (2014), regulators have difficulties to find out about repo market details, which are needed to set up proper regulatory haircuts. Thus, the haircut measure by regulators is likely to be off target when it comes to the reduction of counterparty and systemic liquidity risk. In addition, since haircuts increased due to uncertainty in the financial system, regulatory margins treat consequences of instabilities, rather than the actual cause.
Although, it is difficult to transfer every bilateral repo to a repo with central clearing, a structure where cleared repos are more profitable than regular bilateral agreements will be beneficial for the transformation of the future repo market.
In fact, a CCP structure effectively reduces the counterparty credit risk, rather than systemic liquidity risks, because it transfers the credit risk from the repo borrower to the AAA-rated central counterparty and eases the management of resold assets for the case of a withdrawal of the lender. As Pirrong (2012) described the concentration of risks as a dangerous exposure to excessive margin calls, liquidity risks are not lowered necessarily when clearing the repo trade. However, a central clearing for repos can help to prevent the kind of runs that started the financial crisis in the first place (Penney, 2011).
The German central bank expects repo markets for liquid assets to constitute an important factor to satisfy the required liquidity ratios and that repos will be increasingly long-term contracts due to liquidity regulations (Deutsche Bundesbank, 2013). Contrary to this positive evaluation, Adrian and Ashcraft (2012) emphasized a rise in the credit cost to finance companies, leading to a re-structuring in order to exploit accounting loopholes.
Similar to these opposed expectations, the assessment of liquidity ratios based on Vaderrama’s model did not give clear results. But both, Valderrama and the Basel Committee try to address liquidity risks inter alia via liquidity requirements, which offers a starting point for further evaluating research on that regulatory topic.
Adrian and Ashcraft (2012) claimed that regulatory reforms may change the haircuts to be less cyclical, which has already been confirmed by Vaderrama’s (2010) technical analysis.
As shown in chapter 4.1., the different proposals by the Basel III regulatory framework decrease haircuts. This represents a countercyclical effect in periods of financial instabilities.
One element of the regulation is to prevent risks to spread furthermore, another element is the treatment of financial institutions with instable status. Addressing the latter, a bridge company may therefore manage the resolution to ensure market stability. However, the exact effect of bridge banks on counterparty risk, and especially on systemic liquidity risk, remains unclear.
4.3.2. Political recommendation
As noted by Lucas and Stokey (2011), it is unlikely that bank regulation is able to eliminate liquidity crises entirely. But proper policies can make these events infrequent and if instabilities occur, the effects can be managed accordingly. Regulation can be everything between effective and counterproductive. Because of this uncertainty about the effects and the complexity of developments during crises, an evaluation of the regulation consequences is essential. The findings of Gorton and Metrick (2012) and Schwarz (2014), whereby market liquidity has a high influence on the LIB-OIS, but during financial distress it is steadily replaced by counterparty risks, can be concluded in the way that regulation should adapt consistently during crises. Namely, the crisis development can be slowed down by decreasing liquidity risk and then the crisis diffusion can be mitigated with a reduction in counterparty credit risk.
Against the background of complex and changing financial relationships, this political recommendation will start with a focus on the importance of the academic evaluation and analysis process of proposed regulations. Regulation-assessments should be mandatory and independent before a public institution is permitted to intervene into market mechanics.
The more precise measures should lead to a reduction in counterparty credit risk and liquidity risk. Additionally, a central clearing framework makes periods of financial distress more unlikely and will also help regulators to react to crises, due to a better access to repo market information and activity.
The findings suggest that regulations should only be implemented if the possible consequences were accordingly assessed after their proposal. Some policies can be as effective as the regulator plans them to be, but others might just fail at lowering risks when it comes to distress in the repo market.
Not only the evaluation of the regulation details is crucial, the general necessity of the policy needs to be assessed by politically independent institutions. For instance, if we include every single risk of high-impact, but very unlikely, events into normal market prices, the credit availability and financing situation can be weakened significantly (Comotto, 2012). Supporting this, Mancini et al. (2014) showed that hasty liquidity provisions can decrease the secured interbank lending volume, while the approaches were intended to reduce repo rates. In addition, government interventions, such as the US housing bubble and its structured investment vehicles, can even buildup an unsustainable inflated market, which was a reason for the financial crisis, as argued by Trotter (2013).
As a consequence, either financial products should initially pass a benefit cost analysis, or under a libertarian approach, regulations should pass this evaluation (Posner, 2013). As regulation can have broad negative effects and financial markets need to keep their flexible product portfolio, the libertarian approach is superior.
It is not constructive if the most respected regulation assessments are executed by regulators themselves. Therefore, my political recommendation starts with an independent, academic and more profound evaluation of the proposal’s effects, rather than of the repo contract itself. With researches like recently from Gennaioli et al. (2013), who suggested that risk-weighted capital ratios are vulnerable to ratings misinterpreting or ignore the underlying risks, capital buffers and the work of rating agencies need an evaluation, additionally, on an ongoing bases.
After the assessment of the regulatory details and effects, capital requirements seem to offer an effective measure to mitigate counterparty and liquidity risks. But capital requirements should be adapted to underlying risks and maturities, rather than to political objectives. A distortive capital requirement of zero for certain assets such as government bonds is, therefore, not advisable.
However, contrary to liquidity risk, counterparty credit risk steadily increases its effect during financial distress. Hence, the liquidity risk should be covered early on, for instance with a liquidity ratio. But for now, the proposed liquidity requirements offer a starting point for further research, rather than a proper and realizable measure.
Central counterparties constitute a facility featuring the needed conditions to implement and manage regulatory measures. Since regulators do not have the amount of information needed, an entity which has access to more market information should be involved. The CCP, as an institution between the repo seller and buyer, does obviously have access to information about these contracts due to the concentration during the clearing process, even though they are traded OTC. This availability of information makes haircut regulation on the CCP level as promising as on an institutional level, for which case Valderrama (2010) showed that it reduces systemic risk. Although the trades are conducted anonymously, Mancini et al. (2014) illustrated that CCP-cleared repos are seen as safe agreements that can handle the rising demand for liquidity hoarding and high-quality assets in periods of financial distress. Thus, the CCP can help to determine the optimal regulatory haircut and liquidity ratio to cover counterparty and liquidity risks.
Moreover, a new and deep CCP structure will be easier to supervise and provides additionally several benefits for cash lenders and borrowers (Penney, 2011). For instance the reduction of counterparty risks to participants is stated recently by the Financial Stability Oversight Council (2014) and already back in 2008 by Hördahl and King.
Even though, details should still be determined, the proposed regulations would reduce both liquidity and counterparty credit risks. Namely, liquidity risks decrease due to capital and liquidity regulation, and counterparty credit risks can be covered by capital requirements and regulatory haircuts. However, liquidity ratios are still a topic for further evaluation.
More incentives, or even commitments, to clear bilateral repos via a central counterparty bring their own positive effects and add the benefit of the CCP as an effective channel for further regulation like on haircuts and liquidity ratios. Therefore, capital requirements and the CCP framework are highly desirable and constitute major factors for both a preventive and a reacting repo regulation.
4.3.3. The future of repos
After the repo market contracted in 2008 due to the financial crisis, by mid-2010 the market had recovered for the most part and, at least in Europe, had risen to exceed its pre-crisis size (Tett, 2010). The different regulatory measures will have a significant impact on risks, features and costs attached to repos, but it is also likely that the overall repo framework and market structure changes. There will also be new challenges for regulators, like a close coordination between FSOC departments as well as on an international level. Especially with respect to the Basel III regulations and the banking union as a central contact in Europe.
Based on the recent developments and regulations, shifts in market size, the repo usage and the securities used as collateral will change the prospective repo.
Basel III increases the incentives to trade repos via a central clearing house rather than on a bilateral basis, which will likely decrease the share of bilateral repos and transfer them to CCP-traded contracts.
Kodres and Narain (2010) expect the banks to return to more traditional activities, resulting in less leverage and a reduction in the overall financial system. In fact, the level of the outstanding repo business for 65 European institutions, which participated in the latest ICMA survey, decreased from $8.5 trillion in June 2010 to $7.6 trillion in June 2014 (ICMA, 2014) (both double-counted), giving evidence for the declining market to hold true for repos.
Because a stricter regulation strengthens supervisors and regulators, central banks, which are instructed to implement some regulations and act as intermediaries, will become more important (Hördahl & King, 2008). These broader activities might boost the repo usage, since central banks can use repo trades as an instrument to withdraw cash from the banking system.
This aspect was also covered three years later by Gu and Haslag (2011), noting unconventional usage of repos, where the federal reserve can use this instrument to respond to the liquidity problem and to redistribute liquidity across different markets in order to implement the efficient allocation. This would widen the applicability of repurchase agreements in the US. In Europe, the European Central Bank has already used this policy frequently.
 Two of the papers about the shadow-banking-regulation topic are from Adrian and Ashcraft (2012) and from Ricks (2010).
 Loans are not fungible, so they are illiquid. By packaging them into diversified bonds (origination), which can be sold, they get more liquid. This process is called securitization.
 In a tri-party repo a custodian/clearing bank assures the collateral on behalf of the lender (Krishnamurthy, et al., 2014). Bilateral repos are typically between broker/dealers and hedge funds without an intermediating third party custodian (Mancini, et al., 2014).
 Many other contracts feature the possibility to sell a collateral in case of a defaulted debtor only.
 The purchase day is the day the repo is terminated. The day of the repos initiation is called the sale date. (Acharya, et al., 2010)
 ‘Shadow banking’ is defined, for regulatory purposes, as non-banks performing traditional banking activities. Or how Ricks (2010, p. 3) noted: “shadow banking refers simply to maturity transformation that takes place outside the terms of the banking social contract”.
 Primary dealers are banks and brokers which are authorized by the central bank to directly trade government bonds. They may account for as much as 90 percent of the US repo market (Copeland, et al., 2010).
 The repo’s constant growth until its drop due to the financial crisis, will be discussed in chapter 2.2.
 These are features like a deposit insurance and the central bank acting as the lender of last resort.
 The Regulation Q was part of the Glass-Steagall Act of 1933 during the Great Depression. It limited, inter alia, the interest rates that banks pay. This led to the emergence of the money market mutual funds, as they can bypass such regulation while keeping major characteristics of bank deposits.
 The Glass-Steagall Act separated permissible activities for commercial banks, investment banks and insurers.
 The Act (Title I) makes unsecured loans, which are designed to facilitate home rebuilding, insured by the federal government. (Fishback, et al., 2001).
 For example, the continual changes to the Community Reinvestment Acts (Liebowtiz, 2008) or as George W. Bush emphasizes in 2004: “Not enough minorities own their own homes. […] One thing I’ve done is I’ve called on private sector mortgage banks and banks to be more aggressive about lending to first-time home buyers.“ (Gorton & Metrick, 2012, p. 7)
 The London Interbank Offered Rate (LIBOR) is the referential interest rate for interbank lending.
 An Overnight Index Swap (OIS) leads to an exchange between a fixed short-term interest rate and the average overnight interest rate during the maturity. In contrast to the unsecured LIBOR, OIS contracts are considered to possess very low credit risk and have been untroubled by the increase in counterparty credit risk and liquidity requests seen since the start of the crisis. (Hördahl & King, 2008)
 General collateral (GC) repos are repurchase agreements with collateral accepted by the majority of intermediaries in the repo market and at a very similar repo rate (GC repo rate). General collateral is liquid and has a high quality - like many government bonds.
 A special repo rate is lower than the GC repo rate, because the special repo is subject to exceptional specific demand. (ICMA, 2013)
 A basis point is a unit equal to one hundredth of one percent. It can denote changes in a financial instrument.
 In this context (although it is not a typical shadow bank), they stated the run on the Washington Mutual, which was the sixth largest US-firm in the asset management business when the run occurred in September 2008.
 To face a rollover risk means that the institution has to roll over the matured debt into new debt. Which can be connected to possible higher interest and new terms.
 The evidence, suggesting this interpretation, is based on the East Asian financial crisis in 1997.
 The author describes “agency problems” as conflicts of interests, which may emerge in securitization-based credit intermediation. Since traditional banks lend from their own balance sheet, these misalignments do not exist for those institutions. But the conflicts of interests could lead to supply of poorly underwritten credits and structured securities, which may provoke market collapse.
 Despite this phrase, Gennaioli et al. (2013) spot the problem not in the realization of neglected risks by itself, but rather in the raise in the total amount of risk-taking.
 The repo 105 procedure is the usage of the cash from a short-term repo sale to deleverage (ICMA, 2013).
 A CCP registers transactions between two parties and may also net them.
 Primary dealer repo financing fell from $6.5 trillion in 2008 to $4.4 trillion in 2009 (Acharya, et al., 2010).
 In the examiner’s report Valukas (2010) quotes Martin Kelly, who calls Lehman the last of the Consolidated Supervised Entity firms which continue using Repo 105, leading to a potential for „reputational risk“.
 Until September 2008 the LIB-OIS remained historically high but unexcited. The ensuing events at Fannie Mae, Freddie Mac, Lehman, and AIG resulted in a rapid interbank market decline and significantly increased the LIB-OIS spread. This change lasted until the end of 2008 (Gorton & Metrick, 2012).
 Gorton and Metrick (2012) used the term “contagion” specifically to mean the crisis spread from subprime-housing assets to non-subprime assets that are independent from the housing market.
 The vector X t includes the same control variables as chosen by Collin-Dufresne et al. (2001); besides the company-specific control variables, because most of the series by Gorton and Metrick (2012) are not related to specific firms. See their research for detailed data on these control variables.
 Alternatively, the within estimator would be the more efficient procedure, if the authors expected the disturbance term to be homoscedastic and time-invariant.
 The F-test checks whether the coefficient of determination of the regression measures zero. In the case of its rejection, we can assume that the chosen regression model can explain the dependent variable, or its variance.
 For details on the specific values (in absolute terms, since the percentage differences are difficult to interpret across multiple weeks) and on the volatility calculation, see Gorton and Metrick (2012).
 They might fear the fail of the borrower, forcing them to sell the collateral for the up-to-date price. In this case, the withdrawal rather constitutes an increase in the initial margin.
 That is the last date Schwarz’ (2014) series is available.
 For details on their point estimate, see Gorton and Metrick (2012).
 Since the collateral’s worth shrinks with an increased haircut, the debt decreases respectively.
 The market liquidity term can be interpreted as the early liquidation cost due to imperfect market depth (Valderrama, 2010).
 Here, the haircut constraint measures the haircut weighted share of the risky assets without its deleveraged part, denoted by .
 The haircut needs to be between the equations with low and high level equilibrium prices.
 is the outcome when: asset supply , where , equals the asset demand . Then replacing parts with to simplify the result from the quadratic formula.
 For more details on the supporting empirical evidence see Valderrama (2010, p. 27).
 In this part I present relationships relevant for my interpretation rather than technical derivation.
For a detailed calculation on how to get the equation for the maximum loss see Valderrama (2010).
 The values are based on those used by Valderrama (2010) illustrating effects due to interconnectedness. Because I compute with US Treasuries, which are highly liquid (Campbell, et al., 2014), I raised the parameter for the collateral’s liquidity from 0.5 to 0.7.
 Government bonds issued by members of the European Union are excluded from risk-weighting, see Capital Requirement Regulation (CRR) Art. 114 (4) in Corrigendum to Regulation No. 575/2013 (European Parliament/Council, 2013). The same applies for US-banks trading US Treasury securities.
 Adding the conservation buffer to the minimum common equity tier 1, the ratio of 7.0 percent would already be sufficient (Basel Committee on Banking Supervision, 2011a).
 The model offers a condition for the demanded liquidity ratio under the LCR standard, forcing the bank to hold high-quality liquid assets at least to the extent that they cover the net cash outflows under a short-term shock (Basel Committee on Banking Supervision, 2011a).
 Repos are a substantial topic in the recent FSOC report (2014): such as the declining intraday credit exposures in tri-party repos, the market development and the repo as a source of financial instability. In the annual report to Congress from 2011, the FSOC even identified repos as the third major component of the market infrastructure that require strengthening.
 A broad literature attributes positive effects to central clearing, including decreased counterparty risk (Deutsche Bundesbank, 2013), risk management standards (Penney, 2011) and resilience in times of crisis (Mancini, et al., 2014).
 Hasty in the sense that the consequences have not been evaluated sufficiently.
 The banking union is a broad authority for the ECB to monitor and regulate EU-banks from a central entity. In order not to go beyond the work’s scope, I focused on the Dodd-Frank Act and Basel III.
 The authors name the procedure a repo, because of its alikeness: the central bank buys an asset, then sells that asset to an agent and retires the money (Gu & Haslag, 2011).
- Quote paper
- Sebastian Bieder (Author), 2014, The Repo as a Part of Shadow Banking. An Empirical and Regulatory Discussion, Munich, GRIN Verlag, https://www.grin.com/document/301171