What do we know about non-bank interconnectedness?

Zijun Liu and Jamie Coen.

Non-banks are clearly important in the financial system – according to the FSB, global non-bank financial intermediation grew to $75 trillion in 2013, roughly half of banking system assets. But how are they connected to banks, and what risks does this pose? Using a new granular dataset on the exposures of banks to non-banks, we gained some important insights into what these interconnections look like in the UK. Banks’ direct credit exposures to non-banks are currently small, but there is evidence that some non-bank financial institutions have entered the core of the repo network. We found little evidence in our dataset that hedge funds are conducting risky credit intermediation, but other non-bank financial institutions seem to be leveraging up via the repo market.

Interconnectedness between financial institutions can take many forms, as explained in the recently published Quarterly Bulletin article. Banks and non-banks (insurers, hedge funds etc.) may be directly interconnected via credit exposures, e.g. a repo or margin loan, or indirectly interconnected via fire sales and margin calls. Tarullo (2013) explains how both played a role in the financial crisis. We test the following hypotheses about how non-bank interconnectedness might differ from bank interconnectedness:

  1. Direct credit exposures of banks to non-banks should be low relative to banks’ capital, given that most of the exposures will be in the form of derivatives and repos, which are heavily collateralised.
  2. If the network of UK financial institutions has a ‘core-periphery’ structure (as described in this paper), then the non-banks should all be in the periphery and not in the core, given that they rely on the intermediation services provided by dealers.
  3. Hedge funds, if conducting risky credit intermediation via leverage, should be borrowing a lot of money from dealers via repos backed by corporate bonds (as explained by Pozsar et al (2010)).

This note is based on a new regulatory dataset where around 20 major UK banks and investment firms report their top 60 exposures to banks, other financial institutions and non-financial institutions via repo and derivatives, as at end-2014. As a robustness check, we have done the same analysis based on June 2014 data, and the results are qualitatively similar.

Our analysis confirmed that banks’ direct exposures to non-banks are low relative to banks’ capital. Chart 1 shows the net-of-collateral exposures of reporting banks to banks and non-bank institutions via derivatives and repos (based on the sum of exposures to top 60 counterparties). The exposures of reporting banks to non-bank financial institutions (excluding central counterparties) were £47bn. This appears low when compared to the amount of interbank exposures (£174bn) or reporting banks’ capital (£321bn). We only focus on securities financing and derivative exposures of banks to non-banks, because we believe that banks’ unsecured exposures to non-bank financial institutions should be very small.

Total exposures are calculated as the sum of exposures to top 60 counterparties of reporting banks. Exposures are measured by Exposures At Default (EAD).

The second hypothesis was about non-banks being in the periphery. Chart 2 on the non-bank repo and derivatives networks confirmed our understanding that the non-bank financial system tends to be much less concentrated than the banking system: most non-banks lie on the periphery and only interact with one or two dealers. However, the repo network has a number of non-banks in its core, who trade with multiple dealers on a regular basis.

Chart 2: Network properties of the repo and derivatives market

(a) The repo financial network


(b) The derivatives financial network


Red dots represent banks, blue dots represent non-bank financial counterparties. Lines represent repo or derivatives exposures. More connected counterparties are clustered in the centre of the network.

What is more surprising is that non-bank financial institutions (excluding central counterparties), rather than banks, are the largest counterparties in the repo network, measured by the gross notional of repos and reverse repos they have with reporting banks (Chart 3). In fact, four of the five largest repo counterparties in our dataset are hedge funds. This may be due to the fact that hedge funds that conduct relative-value trades tend to have large offsetting repo positions (e.g. hedge funds may use a combination of repo and reverse repos in government bonds to bet on the yield curve spread), but there is also a risk that hedge funds effectively become part of the interdealer repo market. If that happens, stresses at one hedge fund may spread to the entire repo market.

Chart 3: Weighted degree distributions

(a) The repo network


(b) The derivatives network


The weighted degree for a firm is the sum of all exposures reported to that firm. Exposures are measured as gross notionals for repos and Exposure At Default for derivatives. Central counterparties are excluded.

The third hypothesis was about hedge funds borrowing via repos against low-quality corporate bonds. For hedge funds to be considered as shadow banking entities performing credit intermediation via leverage, they must be borrowing to invest in credit products such as corporate bonds. This is clearly not the case as shown in Chart 4. Hedge funds, while having the largest notional repo portfolios, are the only non-bank financial institutions that provide net funding to reporting banks (although the amount of net lending is very small compared to gross notionals).  Asset managers, insurance companies and pension funds, who are traditionally not leveraged, all borrow from banks via repos on a net basis.

Positive bars represent net lending of cash via repos to reporting firms, and negative bars represent net borrowing of cash via repos from reporting firms.
Positive bars represent net lending of cash via repos to reporting firms, and negative bars represent net borrowing of cash via repos from reporting firms.
Positive bars represent net lending of cash via repos to reporting firms, and negative bars represent net borrowing of cash via repos from reporting firms.
Positive bars represent net lending of cash via repos to reporting firms, and negative bars represent net borrowing of cash via repos from reporting firms.

Chart 5 breaks down the net repo positions of non-bank financial institutions by collateral type. Hedge funds have large repo borrowings against equities (repos include margin loans), but only have limited borrowings against highly-rated corporate bonds, and even have a small positive net lending against non-highly-rated corporate bonds. This may be because hedge funds prefer to invest in relatively liquid instruments (which include most equities but exclude most corporate bonds) so they can adjust their positions quickly without significant costs. Hedge funds are significant net lenders of repos against government bonds, which may suggest that they are short-selling government bonds via repos as part of their overall investment strategy.

The majority of repo borrowing by asset managers, insurance companies and pension funds is against government bonds. This is consistent with anecdotal evidence that these financial institutions are borrowing against their holdings of government bonds to hedge their liabilities or to post initial margin on their derivative positions. The ‘Other’ category also has significant repo borrowings against equities, mostly driven by one institution. In sum, hedge funds are not doing leveraged credit intermediation on any significant scale, while asset managers, insurance companies and pension funds seem to be repo’ing out their government securities to obtain leverage.

So far we have assumed that non-banks do not transact directly with each other. This assumption is based on market intelligence, but we do not have any hard evidence to back it up. If transactions are taking place between non-banks directly without going through a dealer, this will currently be off authorities’ radar. This should change, however, as mandatory reporting requirements for over-the-counter derivative transactions are implemented in Europe. The European Commission has proposed similar requirements on securities financing transactions. Globally, the Financial Stability Board will be responsible for ensuring that data are collected and aggregated in a consistent way. These policy initiatives should enable authorities to better assess the risks from non-bank interconnectedness going forward.

There is another important limitation to our findings, which is that our results are based on top 60 counterparties of UK-based deposit-takers and investment firms only. It is possible that hedge funds outside the top 60 counterparties of reporting firms, or hedge funds that borrow from non-UK dealers, have significant long positions in corporate bonds financed via repos, which would be beyond the scope of our dataset.

These findings, despite being based on a snapshot of data, have important financial stability implications.  Banks’ direct exposures to non-banks do not seem to pose a big risk currently, but there is evidence that some non-bank financial institutions have entered the core of the repo network. This may warrant measures to ensure these firms are resilient, or to ensure that markets are resilient to any distress at these firms. We found little evidence that hedge funds are conducting risky credit intermediation, but there is evidence that other non-bank financials are obtaining leverage via the repo market. Regulators should monitor this through time to ensure financial stability risks in the non-bank financial sector are adequately managed.

Zijun Liu works in the Bank’s Financial Stability, Strategy and Risk – Capital Markets Division and Jamie Coen works in the Bank’s Financial Stability, Strategy and Risk – Banking & Insurance Analysis Division.

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.

If you want to get in touch, please email us at bankunderground@bankofengland.co.uk