Banking

‘No one length fits all’ – haircuts in the repo market


Miruna-Daniela Ivan, Joshua Lillis, Eduardo Maqui and Carlos Cañon Salazar

Funding markets are crucial for healthy and active financial institutions, and consequently for everyone in the economy. The repurchase agreement (repo) market plays a key role in bank and non-bank financial institutions’ (NBFIs’) daily activities by facilitating short-term financing and risk hedging. In this post, we use novel Securities Financing Transaction Regulation (SFTR) data to highlight new, and corroborate previous, stylised repo haircut facts.

What are repos? Why do they have haircuts? And why do we care?

A repo transaction is the simultaneous sale of, and forward agreement to repurchase, securities at a specific price, at a future date (Duffie (1996)). The seller receives cash in exchange for the securities, while the buyer earns a return on the cash lent. The agreed price of repo transactions is usually lower than the market price of the security used as collateral, and the difference is known as a ‘haircut’. Our analysis will show haircuts primarily respond to counterparty credit risk, but will also reflect a wide range of risk, eg market and liquidity risks.

Repo markets support price discovery and improve liquidity in a variety of markets, but they can also pose risks to financial stability. They often involve firms taking (or extending) leverage and can create liquidity and maturity mismatches. They can be exposed to run-like events – which the role of collateral can amplify. Given the role repo markets played during previous systemic episodes, eg the global financial crisis (GFC), the 2020 ‘dash for cash’, and the 2022 liability-driven investment (LDI) crisis, central banks and regulators monitor them closely.

From previous work we know some stylised facts about haircuts in the repo market, eg Kotidis and van Horen (2018), Julliard et al (2022), and Hüser et al (2024). Haircuts depend on counterparty characteristics, eg they tend to be larger for firms with higher (perceived) credit risk. Haircuts also depend on the characteristics of the underlying collateral. They are typically larger for longer maturity contracts, for higher Value at Risk (VaR) collateral, and where collateral is concentrated.

New data, new insights

Enhancing data collection has been a key part of post-GFC regulatory initiatives to better understand securities financing markets. Under the Securities Financing Transactions Regulation (SFTR) (introduced in 2019), the Bank of England receives transaction-level data on securities financing transactions from all UK-based participants. It aims to improve the transparency of markets and to enable better monitoring of risks.

SFTR data provide new insights into an area that supports core secondary market activity like the cash gilt market, including that by non-bank market participants. Haircut data in SFTR is more comprehensive and of higher quality than other existing sources. Improved market coverage and haircut reporting allows us to break down repo haircuts by collateral type (eg gilts), maturity, and sector. It also provides new information on non-gilt government-bond repo markets, enabling us to compare gilt haircuts to other core markets like US treasury securities.

Who’s in the market for (bilateral) gilt repo?

Activity in the bilateral gilt repo market is broadly concentrated in the dealer-to-client segment of the market. Dealers include prime brokers and other banks that typically extend cash lending to NBFI clients that rely on the bilateral gilt repo market for their liquidity and collateral operations. As in Hüser et al (2024), we find money market funds (MMFs) make up the largest share of average daily volume (around 33%). They are important (net) lenders of cash – primarily in the overnight segment of the repo market – to dealers. In turn, MMFs receive medium (7–20 year) and long-dated (>20 year) gilts.

In addition to previous findings, SFTR data show that hedge funds (HFs) make up the second largest share of average daily volume (around 30%), being active on both sides of the repo market; the average tenor of this lending is between one and two weeks. HFs have been increasingly active in government bond markets globally over the past decade, with an acceleration in activity over the last couple of years.

SFTR data largely corroborate this. Chart 1 illustrates the structure of the bilateral gilt repo market across financial sectors. Flows of the same (different) colour as the sector in the outer ring show cash lending (borrowing) by (from) that sector. The width of the flows – indicated by the numbers in the outer ring – reflect the size of outstanding cash lending and borrowing (in £ billions). We see a larger share of outstanding HF bilateral gilt repo cash lending compared to MMFs due to the longer average tenor. Other NBFI participants in the bilateral gilt repo market include asset managers (AMs), insurance companies and pension funds (PFs), as well as LDI funds and other financial institutions (OFIs).

New findings from SFTR data also indicate that non-dealer to non-dealer transactions represent a small fraction of total activity. Chart 1 shows that nearly all outstanding transactions involving non-dealer counterparties are intermediated by dealers. Inter-dealer activity is substantial, shown by the within-sector grey flow for dealers. This highlights their key role as cash lenders in the gilt repo market, which can be crucial during periods of stress.

Chart 1: Financial sector cash lending network in the bilateral gilt repo market (stock data as of December 2023)

What do haircuts look like?

Looking through different collateral types in the bilateral repo market, haircuts typically increase with the market risk (interest rate and liquidity) of the underlying collateral (Chart 2). Government debt securities – widely recognised as the ‘safest’ non-cash assets – have an average haircut of close to zero in bilateral transactions between dealer banks and NBFIs. Consistent with Julliard et al (2022), we find that within the government bond repo market, haircuts reflect, in part, the risk of fluctuations in the collateral price (interest-rate risk). Longer-dated collateralised bonds generally involve higher haircuts largely due to being more sensitive to interest rate changes. Meanwhile, repo backed by riskier debt securities – such as debt issued by banks and NBFIs – attracts average haircuts of over 5%, but this remains far smaller than the 35% average haircuts on main index equity repos.

Chart 2: Volume-weighted average bilateral repo haircuts by collateral type (data from 15 June to 1 July 2023)

Examining the government bond repo market in more detail, SFTR data show that gilt repo haircuts are generally near-zero (between 0%–2%).  Meanwhile, haircuts on repo transactions backed by US Treasury securities are at or below 0.5% for most of our sample, with a notable increase (to nearly 1.5%) around the time of the Spring 2023 banking sector turmoil (Chart 3). This slight variation in haircut levels across government bond repo provides some further evidence that the level of haircuts also reflects the degree of market liquidity in the underlying collateral securities.

Chart 3: Volume-weighted average bilateral haircuts on government bond repos

Near-zero bilateral haircuts in gilt repo are partly driven by market structure and competitive dynamics, beyond prudent risk management considerations, as highlighted by the PRA’s 2023 thematic review of major banks’ fixed income financing. Zero haircuts can also correspond to a set of transactions (generally referred to as ‘nettable packages’) whereby a dealer has a nearly identical amount of cash received/payable with the same client over the same tenor, which can be netted out (Hempel et al (2023)).

But not all bilateral gilt repo is transacted with zero haircut. In fact, less than half of bilateral repo volume (around 40%) is so. There is high heterogeneity beyond trades with near-zero haircut pricing, which is largely driven by counterparty characteristics, such as credit risk considerations (Gorton and Metrick (2012); Copeland et al (2014); Krishnamurthy et al (2014); Mancini et al (2016); and Boissel et al (2017)).

SFTR data corroborates previous evidence showing that haircuts tend to increase with counterparty credit risk. We see this by comparing haircut levels across different NBFI participants in the gilt repo market both in normal times and in stressed financial market conditions. Focusing on the LDI crisis as a case study, Chart 4 shows that haircuts charged on LDI funds and PFs increased procyclically during this stress episode and remained elevated for some time, largely reflecting higher credit risk. This procyclicality amplified market dysfunction by raising the cost of repo borrowing for NBFIs (see Pinter (2023)).  For identified HFs in SFTR data, however, haircuts on repo gilt transactions have remained unchanged over time, and close to near-zero levels.  

Chart 4: 25th–75th percentile distribution of haircuts on bilateral gilt repo and reverse repo transactions across counterparty sectors

What implications can we draw from the analysis?

In this post we use novel SFTR data to show that variation in haircuts is largely driven by counterparty credit risk considerations, although other factors play a part, including the characteristics of the underlying collateral securities. So, no one haircut fits all.

We find evidence of procyclical haircut fluctuations in the LDI and PF sectors during the 2022 LDI crisis. These were slow to return to pre-crisis levels on the back of heightened credit risk. Further work using SFTR data is required to analyse the resilience of sterling markets, given the potential of haircut procyclicality to exacerbate liquidity demands during stress and amplify market dysfunction.

On the other hand, the low level of haircuts faced by HFs may suggest that substantial leveraging could be taking place in a large bilateral gilt repo market segment without appropriate risk mitigation. This merits continuous monitoring given its potential financial stability implications.


Carlos Cañon Salazar and Joshua Lillis work in the Bank’s Market Intelligence and Analysis Division and Miruna-Daniela Ivan and Eduardo Maqui work in the Bank’s Market-Based Finance Division.

If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.



READ SOURCE

Business Asia
the authorBusiness Asia

Leave a Reply