Liquidity Risk (Mis)Management: The Failure of Silicon Valley Bank and the Liability-Driven Investment Episode in UK Gilt Markets

Silicon Valley Bank (SVB) failed in March 2023.  Three prominent economists affiliated with the Golub Center discussed their ideas of what happened and what’s next immediately following SVB’s demise. Now former Golub Senior Fellow, Laura Kodres, discusses the similarities between what happened with British pension funds in autumn 2022 and SVB in a new article, Liquidity Risk (Mis)Management:The Failure of Silicon Valley Bank and the Liability-Driven Investment Episode in UK Gilt Markets. To access the full study, view/download the PDF here.

October 12, 2023 | by Laura E. Kodres ^[1]

In September and October 2022, the Bank of England (BoE) intervened in the gilt market, buying up gilts and providing a new liquidity facility for a subset of gilt market participants—Liability Driven Investment (LDI) fund managers—to halt a potential fire sale of long-dated gilts. Such a fire sale would have impeded the normally functioning of the UK’s government market—raising yields and widening bid-ask spreads—with likely knock-on effects to the broader economy through a rapid and large tightening of financial conditions. Although the BoE’s gilt purchases and liquidity support appeared to be necessary, it has raised a number of questions about how a relatively small set of actors could cause such a large problem and what can be done to avoid a repeat of the event. Importantly, that small set of actors were non-bank financial institutions (NBFIs) that took on leverage without properly assessing and preparing for its liquidity implications. Sound familiar?

The backdrop to the LDI episode was that on September 22, 2022 the then-new UK government released its “Growth Plan.” The Plan was interpreted by market participants as unsustainable, increasing UK debt and the cost of issuing it. Gilt yields rose precipitously, with larger jumps—over double—than those in the early days of the COVID pandemic, particularly for long-dated securities. A subset of gilt market participants, defined-benefit pension schemes using LDI strategies, were ill-positioned for the abrupt move and the liquidity strains that resulted from their leveraged gilt repo positions and other derivatives. The collateral and margin calls were too much for some of the LDI fund managers to handle. The liquidity buffers held by the defined-benefit schemes’ LDI fund managers were depleted, requiring outright sales of gilts and other assets to raise the necessary funds. The BoE, sensing a fire sale with even greater yield repercussions stepped in to buy gilts and provide extra means of generating the needed liquidity through a new repo facility.

So, why were UK pension schemes using leveraged LDI strategies and how did they become so liquidity impaired?

UK defined-benefit pension schemes had been underwater for quite some time.^[2] The measure used to assess funding gaps varies over time, in part because the schemes’ asset values are mark-to-market, but the liabilities are discounted at a long-term interest rate and so move more slowly. Moreover with the extended low interest rate environment ending only recently, the present value of liabilities had been elevated, contributing to the underfunded position. Both the volatility and underfunded status were quite distasteful for corporate sponsors, who need to report on their pension scheme’s status in their annual reports. 

Asset managers and pension fund advisors offered UK schemes a way out—LDI strategies. At their most basic level, these strategies attempt to match the risk sensitivity of the value of liabilities owed to future pension retirees with the those of the assets on which they depend. An LDI strategy can lower the volatility of the reported (accounting-based) funding ratio by investing in assets that have cash flows that mirror the liability payments—fixed-income assets, like gilts—as then the liability’s discounted value and asset’s market values move more closely together. However, while solving the volatility problem, the returns to these fixed-income assets are relatively low—and have been since at least 2008—and so this “matching” principle does not address the underfunding status of UK defined-benefit pension schemes. 

To help rectify the funding shortfall, LDI advisors suggested that the fixed-income assets be used to generate cash (through a repurchase agreement) and then the proceeds used to buy other assets, including natural hedges such as long-dated conventional gilts and indexed-linked gilts, or other more risky (and potentially less liquid) higher-earning assets, including collateralized loan obligations, private equity, infrastructure projects, climate bonds, and so on.^[3] The cash from gilt repos can be also used to satisfy initial margin requirements for derivatives—either to hedge risks associated the future liabilities or to take on additional outright risk.^[4]That is, leverage can be employed to raise returns.^[5] With conventional gilts and index-linked gilts in plentiful supply on pension scheme’s balance sheets, these high-quality assets were cheaply and frequently used as repo collateral.

Leverage via both repos and derivatives can accentuate changes the value of the pension schemes’ LDI portfolio—and, importantly, elicit heightened liquidity risks. If gilt yields move against the LDI scheme by a given amount, the repo counterparty will request more gilts or cash as collateral to cover their loss in value; if derivative prices move beyond some pre-set levels, the over-the-counter (OTC) derivatives counterparty (most likely a bank) or the central counterparty (CCP) will require increased initial margin to hold the position. In both cases, the repo holder or derivatives user needs to come up with (lower-valued) gilts, so more of them, or larger amounts of cash (so needs to sell assets to generate it) in the very short term.

Hence when gilt yields moved up dramatically and prices fell, LDI fund managers were asked to post more collateral and cash at their bank counterparties for the levered positions they held on behalf of pension schemes. Although pension schemes had approved liquidity buffers for use by their managers, some did not have enough. The schemes’ trustees needed to give managers more gilts or cash from their non-LDI portfolios and were not operationally set up to do this on a moment’s notice. Documentation, decision-making structures, and the operational means of transferring the collateral were not in place. When the Bank of England’s market intelligence discovered that an estimated £50 billion gilts were to be sold, on September 28, 2022, the BoE agreed to buy gilts outright for a fixed period of time so the bank counterparties could get the collateral in place. The Bank of England would have preferred to allow the private sector to solve this issue, but the LDI advisors were constrained by their net asset value (NAV) requirements and could not take on the additional leverage that would have been implied by the use of repos with their counterparties. Moreover, their bank counterparties and other dealers did not have the capacity (or desire) to take on the mostly indexed-linked gilts and corporate bonds as collateral for repos, while using other more conventional gilts from their portfolios that they, in turn, could use as collateral at the Bank of England’s normal repo facility. During the next few days, the Bank of England realized that some of the assets—namely indexed-linked gilts and corporate bonds—were what was available but not eligible for the Bank of England’s facilities. On October 10, they then extended its Asset Purchase Facility (APF) to include purchases of indexed-linked gilts and created a new repo facility to accept corporate bonds.^[6]

So the main factors that contributed to the surprise were: the Bank of England’s (and perhaps the bank counterparties’) incomplete information of the degree of leverage taken on by LDI managers on behalf of their pension scheme clients and the insufficiency of their liquidity buffers; the concentration in illiquid indexed-linked gilts and corporate bonds that were excluded from the APF; and the lack of operational capacity and the ability of trustees to move assets quickly to replenish their LDI liquidity buffers. These factors are common to many stress periods, though there are some nuances specific to UK defined-benefit pension schemes and their advisors. 

What can be done to prevent such episodes and accompanying gilt fire sales in the future? 

The fixes require both the private sector and the relevant UK regulators to take ownership of their respective roles in contributing to financial stability and a healthier pension industry. 

1. Margin models, liquidity buffers and collateral management should be improved so that they permit economically effective risk management and the appropriate use of derivatives by UK pension schemes. 

These improvements apply not only to LDI fund managers and their clients, but more generally to any NBFIs using similar leveraged positions. With NBFIs increasing their use repos and derivatives, there is a general lack of understanding about the need for (and expense of providing) collateral. Bank counterparties should be more forthcoming about how they calculate needed margin, the haircuts applied, and when they are willing (or not) to accept various forms of collateral. For leveraged LDI strategies, very few of the contracts were cleared on a centralized counterparty (CCP), so a move to CCP clearing for both repos and derivatives could improve margin calculation transparency, lower default risks, and allow for the posting of gilts (albeit at a cost that would need to be set against the benefits).

• The degree of leverage that is undertaken by pension schemes should be better monitored and, if need be, the UK Pension Regulator should be able to limit leverage on a case-by-case basis if convinced that pension scheme trustees are unable to make informed decisions.

Keeping in mind that leverage, used judiciously, can help to improve returns in an under-funded plan, it is important for it to be properly calculated. Leverage taken through repo borrowing to invest in riskier assets is different than the “synthetic” leverage in derivative contracts, which can be risk reducing if used to hedge. The Pension Regulator, the primary regulator of pension schemes in the UK, needs to better understand how leverage is employed and whether it is, in fact, excessively risky in the context of a particular pension scheme. Repos and derivatives should remain available.

• For the mitigation of the systemic component of excessive (collective) pension scheme leverage, broader policy tools are needed by the Financial Policy Committee (FPC). The BoE should remain ready to provide emergency liquidity assistance in extreme cases. 

Currently, the FPC, the overseer of financial stability in the UK, has no macroprudential tool to apply to leverage build-ups in NBFIs. However, it has the ability to suggest changes to relevant regulators—and has recently provided such advice to The Pension Regulator regarding LDI funds’ liquidity buffers. Consideration of a more formal binding leverage ratio or other constraints (e.g., margin haircut floors) for pension schemes employing leverage may be required after a thorough cost/benefit analysis. The expense of eliminating all tail risks would hinder efficient capital mobility and so the BoE should remain ready to provide liquidity during extreme stress—an accepted role for a central bank. It is important to note that the UK pension schemes had enough high-quality assets to post or sell during this episode, and indeed had improved their funding status in the run-up to the episode. The issue was the timeliness with which the assets could be transferred to LDI funds to be tapped by the counterparties, not their existence.  

• Pension fund accounting may well be a driver for leveraged LDI strategies, but accounting changes are unlikely to be an effective tool to guide good risk management. 

Instead of aiming to alter the accounting measures of funding as some have suggested, it would be more productive to improve the risk management capacity of pension schemes’ trustees and provide more informative reporting by corporate sponsors.^[7] The ability for a specific number at a specific date to represent the uncertainty surrounding long-term asset returns and whether they will generate enough cash flows for promised pension liabilities is a bridge too far. That said, there are other ways of presenting pension scheme funding information to stakeholders of corporations that will incentivize better risk management and convey useful information. Measures that correct for cyclical variations and illiquidity, and use of derivatives for hedging, can all lower volatility of reported figures and provide more transparency about a pension scheme’s long-term health. 

• Coordination among regulators needs to improve in order to provide a holistic assessment of both risks to financial stability and to corporations and their pensioners. 

Given that the relevant UK regulators have different mandates, tools, and information, it is incumbent upon them to share relevant information and use their different areas expertise to address financial stability (FPC), bank (counterparty) health (Prudential Regulation Authority), market integrity (Financial Conduct Authority), and individual pension fund scheme health (The Pension Regulator). Since the main regulatory bodies have overlapping members, aside from The Pension Regulator, it should be possible to coordinate better (see figure). Prior to the LDI episode, there were data gaps (some in the process of correction),^[8] but going forward it is important to collect data to better assess the risks to financial stability, which would include liquidity buffers, leverage, concentration of asset holdings, and counterparty interconnectedness at a minimum. Who should do so and how frequently needs to be sorted out. However, extending the mandate of the Pension Regulator to include financial stability (a suggestion from the House of Lords)^[9] would risk diluting and complicating the ability of the FPC to carry out its mandate and lessen its effective operation.^[10]

• Education and training for pension scheme trustees requires an holistic approach, as the LDI advisors do not have the incentive to provide unbiased advice.

Because the LDI consultant business model is to profit from the advice given to pension scheme trustees, consultants are not in an unbiased position to provide basic training that could lead the trustees to use other (perhaps their own in-house) risk management solutions. Moreover, LDI advisors are not under the Financial Conduct Authority’s regulatory purview. A first step is to extend the perimeter to ensure advisors are themselves competent and are avoiding conflicts of interest. A second step is a more effective training program provided by an arms-length third party for trustees. There are many options for this and serious consideration should be given to requiring trustees to have a certificate or license. It is important that professional trustees of all pension schemes, no matter how small, are in a position to ask the right questions of their advisors and fully understand the answers. Pension schemes trustees need to be able to independently assess the appropriateness of their strategy, which may or may not include the use of hedging, leverage, or higher risk/return assets. 

Summary

Despite post-Global Financial Crisis progress, this piece shows that excessive leverage and its associated liquidity risks continue to require the attention of policymakers to avoid financial instability and harm to average individuals. NBFIs appear to routinely be at the heart of recent distress episodes. The use of LDI strategies by defined-benefit pension schemes is not, in and of itself, problematic but the number of actors that failed to use the strategies appropriately and the lack of holistic oversight requires attention. This piece suggests several policy paths (and a few paths that should be avoided), in order to ensure financial stability and to better serve the pensioners that ultimately depend on their schemes for their future incomes.

References: 

Bank of England, 2018, Financial Stability Report, November 2018. https://www.bankofengland.co.uk/financial-stability-report/2018/november-2018

Bank of England, 2022, Letter from the Governor Bailey to the Chancellor Hunt, November 4, 2022.  https://www.bankofengland.co.uk/letter/2022/november/asset-purchase-facility

Buiter, W., S. Cecchetti, K. Dominguez and A.S. Serrano, 2023, “Stabilising financial markets: Lending and market making as a last resort,” VOXEU, February 6, 2023. https://cepr.org/voxeu/columns/stabilising-financial-markets-lending-and-market-making-last-resort

House of Lords, Letter to the MPs Andrew Griffith and Laura Trott, February 7, 2023. https://committees.parliament.uk/publications/33855/documents/185115/default/

Kodres, L., 2023, “Too Many Cooks, Not Enough Risk Management: Gilt Market Dysfunction and Liability-Driven Investment (LDI),” May 13, 2023. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4445632

Lim, C.H., I. Krznar, F. Lipinksy, A. Otani, and X. Wu, 2013, “The Macroprudential Framework: Policy Responsiveness and Institutional Arrangements,” International Monetary Fund: Washington DC, Working Paper 2013/16. https://www.imf.org/en/Publications/WP/Issues/2016/12/31/The-Macroprudential-Framework-Policy-Responsiveness-and-Institutional-Arrangements-40789

Pension Protection Fund, PPF 7800 Index, January 31, 2023. https://www.ppf.co.uk/sites/default/files/2023-02/PPF_7800_Update_February_2023.pdf

---

^[1] Dr. Laura E. Kodres is an Associate at the Golub Center for Finance and Policy at the Sloan School of Management, Massachusetts Institute for Technology. She can be reached at lkodres@mit.edu. The views expressed in this piece are her own and are not to be associated with the Golub Center. She is grateful for comments received from Ed Golding, Debbie Lucas, and Maximillian Fandl. 

^[2] Pension Protection Fund (PPF) reports show chronic underfunding going back to 2010, with nearly 90 percent of all schemes in deficit in early 2015. https://www.ppf.co.uk/sites/default/files/2023-02/PPF_7800_Update_February_2023.pdf

^[3] Indexed-linked gilts refer to gilts indexed to inflation, providing an inflation hedge for those who hold them.

^[4] Use of derivatives is sometimes referred to as “synthetic” leverage since very little initial margin is needed to command the notional principal of the contact.

^[5] Kodres (2023) describes in more detail the use of leverage in LDI strategies, its benefits and its pitfalls, and some policy solutions to the LDI episode. 

^[6] See Buiter et al (2023) on the best practices for asset purchase programs and a discussion of their various uses under stress. The letter from Governor Bailey to Chancellor Hunt dated November 4, 2022 explains the BoE actions and its new facility, Temporary Expanded Collateral Repo Facility (TECRF). 

^[7] House of Lords, Letter to the MPs Andrew Griffith and Laura Trott, February 7, 2023. Some have suggested that the interest rates used to discount the liabilities are inappropriate; others suggest that the mark-to-market of the assets elicits excessive volatility. 

^[8] BoE Financial Stability Report, November 2018. 

^[9] House of Lords, Letter to the MPs Andrew Griffith and Laura Trott, February 7, 2023. 

^[10] Lim et al (2013) shows division of macroprudential mandates across regulators leads to less timely and effective implementation.