What caused the LDI crisis? – Bank Underground

Gabor Pinter, Emil Siriwardane and Danny Walker

In September 2022 the interest rate on UK gilts rose by over 100 basis points in four days. These unprecedent market movements are generally attributed to two key factors: the 23 September announcement of expansionary fiscal policy – the so-called ‘mini-budget’ – which was then amplified by forced sales by liability-driven investment funds (LDI funds). We estimate that LDI selling accounted for half of the decline in gilt prices during this period, with fiscal policy likely accounting for the other half. Balance sheet segmentation and operational issues slowed capital injections into LDI funds by well-capitalised pension schemes, leading LDI funds to instead sell gilts. Our analysis shows that these frictions were most pronounced for pooled LDI funds.

What is LDI?

In the UK liability-driven investment (LDI) has historically been used by corporate defined-benefit pension schemes for asset-liability matching, whereby the schemes seek to match the duration of pension assets and liabilities. In a typical LDI fund, a single pension or group of pensions invests capital, after which the LDI fund often borrows funds to purchase gilts, either outright or synthetically using derivatives. The liability side of an LDI fund’s balance sheet consists of capital from the pensions, debt via repurchase agreements (repo) and interest rate swaps (paying floating). The asset side consists primarily of gilts, interest rate swaps (receiving fixed), and cash equivalents.

LDI deleveraging during the crisis

Chart 1, based on transaction-level data held by the Bank of England, describes the balance sheet of LDIs as of 1 September 2022, three weeks before the crisis commenced. It reveals that LDIs entered the crisis with £300 billion in assets, financed by equity from corporate pension schemes and debt through repo and derivatives.

Chart 1: LDI aggregate balance sheet on 1 September 2022

Sources: Bank of England, EMIR Trade Repository, MIFID II and SMMD data sets.

LDI leverage, measured by the ratio of assets to equity, began September at a level initially below 2. It then rose steadily throughout the month, spiking sharply in the week of the mini-budget to 2.7, before returning to its initial level by the end of October. Chart 2 analyses the factors driving LDI leverage during the crisis by plotting their cumulative gilt purchases, swap exposure, and repo borrowing from September to October 2022. Flows in the plot are based on par values and are indexed to zero as of 23 September, the day the mini-budget was announced, which is indicated by a vertical dashed line. The plot illustrates a rapid liquidation of gilts by LDIs following the mini-budget announcement. In the five weeks following the mini-budget, LDIs sold approximately £25 billion in gilts, with 30% of this activity occurring within the first five days after the announcement.

The plot also shows that the proceeds from these gilt sales were primarily used to reduce leverage by retiring repo debt. Our calculations suggest that 74% of the £33 billion of repo debt retired from 23 September to 31 October was covered by the proceeds from gilt sales, while the remainder was financed through cash reserves or equity injections from corporate pension schemes.

Chart 2: LDI funds deleveraged rapidly during the crisis

Sources: Bank of England, EMIR Trade Repository, MIFID II and SMMD data sets.

LDI selling accounted for at least half of the fall in gilt prices during the crisis, with fiscal policy likely accounting for the remainder

Given their substantial gilt sales, it is important to determine how much LDIs contributed to the decline in gilt prices following the mini-budget. This analysis is challenging because it requires separating the effects of LDI behaviour from the impact of the mini-budget itself. The ideal approach to address this identification issue would be to compare the price trajectories of two gilts that, while differentially held by LDIs, were equally exposed to the mini-budget’s fundamental shock. Our difference-in-differences research design approximates this ideal by comparing gilts of similar maturities but differing LDI holdings at the start of the month.

Chart 3 plots regression estimates of the impact of LDI selling on gilt prices, along with 95% confidence intervals. Each point on the plot represents the daily average percentage price gap between gilts heavily held by LDIs and those less heavily held, after adjusting for fundamental factors such as duration. Note that a 100 basis point increase in yields for a 20 year gilt maps to roughly a 20% price fall.

Chart 3: LDI forced selling led to gilt price falls of around 7%

Sources: Bank of England, EMIR Trade Repository, MIFID II and SMMD data sets.

At the peak of the crisis, our preferred estimates indicate that LDI selling resulted in gilt price discounts of approximately 7%. This estimate could be considered a lower bound, as it excludes any spillover effects that LDI selling may have had on the overall level of gilt yields. A back of the envelope calculation based on this estimate suggests that LDI selling accounted for half of the total decline in gilt prices following the mini-budget, with the fiscal policy announcement likely accounting for the remainder. Of course, it is worth noting that in the absence of the Bank’s gilt market intervention, the gilt price fall could have been more severe so LDI selling might have accounted for a greater share.

Why did LDIs sell in the first place?

Our preceding analysis shows that LDIs sold gilts to reduce leverage after the announcement of the mini-budget, resulting in large fire sale discounts in the gilt market. But why were LDIs and their investors, namely corporate defined-benefit pension scheme, unable to avoid gilt liquidations? This question is fundamental for understanding the root causes of the crisis.

One potential answer is that pension schemes lacked sufficient assets to recapitalise the LDIs. However, a closer examination of the combined balance sheet of the LDI-pension sector suggests that this explanation falls short. At the height of the fire sale, our analysis shows that debt held on LDI balance sheets amounted to no more than 15%–20% of the combined LDI-pension sector’s balance sheet. Furthermore, this low level of financial debt was supported primarily by sovereign debt, investment-grade corporate credit, and developed-market equities.

Given that corporate pension schemes seemingly had ample assets to fully secure the debt on LDI balance sheets, we hypothesise that the internal contracting structure between pensions and LDIs effectively created a form of slow-moving capital. To understand the nature of the contracting friction, consider a pension with £100 of capital that wishes to purchase £150 of gilts using debt. The simplest arrangement for the pension would be to hold the debt on its own balance sheet, as depicted in Portfolio 1 of Chart 4. A second, more complex arrangement would be for the pension to invest £30 of capital into an LDI fund that purchases the £150 using £120 of margin debt. This situation is depicted as Portfolio 2 in Chart 4.

Chart 4: Example of balance sheet segmentation

Source: Bank of England.

On a consolidated basis, Portfolio 1 and Portfolio 2 might look the same, but their reaction to market downturns could vary significantly. Should there be a 20% decline in gilt prices, Portfolio 1 would remain stable; its £50 of margin debt would still be far exceeded by the £120 value of its gilts. Conversely, the same drop would push the LDI in Portfolio 2 into technical default, since its margin debt is collateralised only by its own assets, not those of its pension owner. Theoretically, the pension could transfer £70 of its cash reserves to bolster the LDI’s balance sheet. Yet, institutional barriers (eg, trustee approval) might impede timely collateral replenishment, forcing the LDI to instead sell its gilt holdings.

Pooled LDI funds had the largest structural issues

The previous example underscores how the siloed or segmented nature of the LDI’s balance sheet, coupled with procedural delays in transferring additional collateral, can precipitate forced sales. This mechanism also potentially explains why UK life insurers, despite their economic similarity to UK pensions, did not experience the same level of stress.

While the structural issues associated with balance sheet segmentation affected all types of LDIs, they are easier to detect within pooled LDI funds, which invest on behalf of multiple pensions. This is because recapitalising a pooled LDI fund requires co-ordination among several pensions, a task that becomes increasingly difficult during a crisis. In line with this idea, our analysis indicates that, compared to single LDIs, pooled LDIs sold roughly 11 percentage points more of their gilt holdings by the end of October, even after accounting for differences in balance sheet composition and manager effects. We further show that selling by pooled LDIs caused gilt price discounts of roughly 10%.

Policy implications

In the aftermath of the crisis, there has been a large debate about how LDIs should be regulated going forward, and the Financial Policy Committee has recommended that the Pensions Regulator takes action in the UK. One set of proposals involves liquidity and leverage restrictions for LDIs. However, our results imply that excessive LDI leverage was likely not the core issue driving the fire sale, as LDI debt was relatively small compared to pension assets. Instead, our analysis points to segmentation between LDI and pension balance sheets as a key driver of the crisis. Regulation designed to improve operational arrangements between pensions and LDIs, such that their balance sheets are better integrated, may therefore be most effective at avoiding crises of this kind. Of course wide-scale operational changes will take some time to implement, which means other measures are helpful in the shorter term.

Danny Walker works in the Bank’s Governors’ offices, Gabor Pinter is an economist at the Bank for International Settlements and Emil Siriwardane is a professor at Harvard Business School.

If you want to get in touch, please email us at [email protected] 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.