Leveraged investment funds: A framework for assessing risks and designing policies
Published as part of the Macroprudential Bulletin 15, January 2025.
This article develops a framework for assessing risks and formulating policies for leveraged alternative investment funds by integrating entity-level information for investment funds with transaction-level data on derivatives and repurchase agreements. Combining both types of data allows us to better understand the use of leverage in alternative investment funds and assess its implications for financial stability. Using a comprehensive set of risk metrics, our analysis identifies hedge funds and liability-driven investment (LDI) funds as the most vulnerable to leverage-related risks. We demonstrate the usefulness of our framework for risk assessment by analysing the sensitivity of leveraged funds to interest rate shocks. We find that LDI funds may face significant liquidity needs and mark-to-market losses. Hedge funds appear to be more resilient to this type of shock, but depending on their investment strategy, they could be sensitive to other risk factors. Our framework allows us to flexibly analyse other risk scenarios and to evaluate regulatory measures in terms of both their effectiveness and their precision in addressing potential vulnerabilities arising from leverage.
1 Introduction
The use of leverage by investment funds can present risks to financial stability. Leverage amplifies exposures of funds and magnifies profits and losses. Exposures can be increased through financial leverage (for example through collateralised borrowing in the repo market) or through synthetic leverage (by using derivatives). In the event of adverse market shocks, mark-to-market losses reduce the net asset value of investment funds. This results in losses for investors and increased credit risk for counterparties. In addition, leverage entails liquidity risks, as leveraged funds may face liquidity needs to meet margin calls on their derivatives exposures to compensate for mark-to-market losses (variation margin) or increased counterparty risk (initial margin). Declines in the value of collateral used for repo borrowing increase the need to post additional collateral or make up for the valuation losses with cash.
Recent market stress episodes have highlighted how liquidity needs and mark-to-market losses can lead to forced asset sales by leveraged funds. Leveraged entities faced severe liquidity squeezes during recent market episodes such as the COVID-19 market turmoil in March 2020, the collapse of Archegos Capital Management in March 2021 (European Systemic Risk Board, 2021; European Securities and Markets Authority, 2022) and the UK mini-budget crisis of September 2022. Increased liquidity needs and mark-to-market losses on leveraged positions forced those funds to sell assets, consuming market liquidity and exerting downward pressure on prices. Such deleveraging created downward spirals, where declining asset prices triggered additional liquidity demand, leading to additional forced sales, driving prices further down. This dynamic propagated shocks to other entities with similar exposures, as observed for example in September 2022 for GBP funds pursuing LDI strategies (European Systemic Risk Board, 2023).
In the EU investment fund sector, high leverage is typically associated with alternative investment funds (AIFs).[2] The Alternative Investment Fund Managers Directive (AIFMD)[3] constitutes the EU regulatory framework for AIF managers, established as part of the regulatory response to the global financial crisis (for an overview of the reporting framework, see European Securities and Markets Authority, 2019). This Directive includes harmonised reporting requirements at fund level and direct supervision of managers by national competent authorities (NCAs). The AIFMD covers a highly heterogenous group of funds, which employ different strategies and can invest in multiple asset classes, including hedge funds, real estate funds and private equity funds. AIFs are often marketed to institutional or other professional investors and generally face no regulatory restrictions on the level and sources of leverage.
The AIFMD mandates detailed regulatory reporting for the universe of AIFs, which can help to identify risks and assess policies. The data collected under the AIFMD include information on fund exposures, leverage levels, liquidity profiles and risk concentrations, enabling authorities to identify and assess risks that AIFs may pose to financial stability. Using the data, NCAs are required to perform an annual assessment of leverage-related risks for AIFs. When risks related to leveraged AIFs are identified as systemic, NCAs have the power to impose limits on the amount of leverage used by an asset manager, or to place other restrictions on the management of the fund under Article 25 of the AIFMD (European Securities and Markets Authority, 2021). Such powers have only been used on two occasions, namely to address leverage risk in Irish and Luxembourg GBP LDI funds and Irish property funds.[4]
Combining entity-level data gathered under the AIFMD with transaction-level information offers a unique opportunity to deepen the understanding of leverage-related risks in the AIF sector and to assess policies for addressing such risk. Transaction-level datasets offer detailed and timely information on exposures, while entity-level data include measures of an investment fund’s characteristics and resilience. For instance, when estimating the impact of an interest rate shock, transaction-level data on repos and derivatives would allow potential losses to be calculated. However, it would not allow these losses to be compared with the investment funds’ net asset value or available liquid assets. This could only be achieved by merging the data with entity-level data, such as AIFMD data.[5] Merging both types of data is therefore crucial for a comprehensive assessment of financial stability risks. A combined dataset of this kind can also be used for assessing policies that target systemic risk, for instance by constraining leverage at entity level or by imposing margin or haircut requirements on certain transactions.
This article offers initial insights from an ongoing European Systemic Risk Board (ESRB) analysis, where AIFMD and transaction-level data are used in combination to identify leverage-related risks and to evaluate policies aimed at tackling those risks. Our initial risk assessment focuses on interest rate risk scenarios and highlights the impact on the liquidity risk faced by leveraged funds, as well as the effect on their net asset value (NAV). We find that LDI funds are particularly vulnerable in the event of an interest rate shock. While LDI funds would face large liquidity demands relating to derivatives (EUR LDI funds) and repo borrowing (GBP LDI funds), hedge funds are more likely to remain resilient under the same type of shock, as their repo and derivatives positions offset each other. Looking ahead, our approach of using a combined dataset allows us to consider various risk scenarios tailored to the specific risk profiles of (cohorts of) funds. It also allows us to evaluate policy measures in terms of their effectiveness in addressing potential vulnerabilities arising from leverage in AIFs and their possible unintended effects.[6]
2 Building a comprehensive dataset of leveraged alternative investment funds in the EU
We enhance the detail of AIFs’ exposures by merging entity-level data with transaction-level data on derivatives and repos. Under the AIFMD, entity-level data on AIFs’ exposures to broad asset classes are collected, along with information on regulatory assets under management (AuM) and NAV. However, these data are only available infrequently[7] and do not give information on individual positions. To overcome these shortcomings, we use granular transaction-level data on derivatives (gathered under the European Market Infrastructure Regulation, EMIR[8]) and securities financing transactions (gathered under the Securities Financing Transactions Regulation, SFTR[9]) reported at entity level and link them with AIFMD data using legal entity identifiers (LEIs). Chart 1 shows a stylised balance sheet of a leveraged alternative investment fund. It illustrates that in order to reconstruct the granular exposures of a leveraged AIF, it is necessary to rely on both EMIR and SFTR data.
Chart 1
Stylised balance sheet of a leveraged alternative investment fund and data sources
Sources: AIFMD, SFTR and EMIR data.
Notes: The different colours indicate the different sources of the data. Percentages refer to the value of the positions (cash, initial margin, etc.) of all leveraged AIFs compared with the sum of all assets held by these AIFs. For derivatives, the sum of exposures over total assets is displayed. SFTR and EMIR data provide transaction-level information, whereas AIFMD data only contain information on exposures at the level of asset classes. Owing to discrepancies across the three datasets, the sum of assets is not equal to the sum of liabilities. Since we average across all funds in our sample, this chart does not reflect the high degree of heterogeneity within and across fund types.
We use end-of-2023[10] data for the analysis and focus on highly leveraged EU-domiciled AIFs. We identify leveraged funds as AIFs with a gross leverage ratio (AuM/NAV) above 3.[11] The two most important groups of AIFs in the resulting sample are LDI funds[12], with an aggregate NAV of €85 billion, and hedge funds, with an NAV of €13 billion (see first section of Table 1). We distinguish between hedge funds with a relative value strategy and hedge funds with other strategies.[13] In addition, we categorise LDI funds according to their base currency, which is indicative of their geographical focus and the origin of their investors. The residual group of other leveraged AIFs is composed of fixed income funds and mixed funds.
3 Exploring the risks posed by leveraged alternative investment funds
3.1 Leverage levels, sources of leverage and risk metrics
We explore the risks posed by leveraged AIFs using metrics suggested by the Financial Stability Board (FSB) in a recent consultation report on addressing leverage in non-bank financial intermediation (NBFI).[14] Many of the metrics considered by the FSB and in this article are constructed as the ratio between an exposure metric (e.g. gross borrowing) and a resilience metric (NAV or cash). First, we consider metrics for gross and net leverage that take into account all sources of leverage (see second section of Table 1). We then move on to metrics that are informative about risks related to repo borrowing (third section) or derivatives (fourth section).
Table 1
Descriptive statistics and risk metrics
(see table rows for units)
Sources: AIFMD, SFTR and EMIR data.
Notes: The leverage metrics are averages across AIFs weighted by NAV. In contrast to gross leverage as measured by AuM/NAV, net leverage computed using the commitment method takes into account that netting and hedging arrangements may reduce overall exposure. Gross synthetic exposure is measured as the sum of notional values. Gross repo borrowing divided by NAV or cash is computed for AIFs with positive repo borrowing. Similarly, initial margin posted divided by NAV or cash is computed for AIFs with positive synthetic exposure. (This is different from Chart 2, where the NAV of AIFs that are not active on these markets is included in the respective leverage ratios.) Since only a small number of other hedge funds and EUR LDI funds are active on the repo market, the value of some metrics cannot be shown for confidentiality reasons.
Hedge funds pursuing relative value strategies tend to have the highest levels of leverage, followed by hedge funds with other strategies. Relative value hedge funds exploit arbitrage opportunities arising from minor discrepancies in security pricing, which typically involve gaps of only a few basis points. Therefore, they often employ significant leverage to enhance their returns. Gross leverage is close to 30 for hedge funds with relative value strategies and remains very high even after netting and hedging arrangements are taken into account, as indicated by a leverage measure above 20 using the commitment method (see second section of Table 1). Leverage is also very high among other hedge funds with a net leverage ratio of 8, followed by LDI funds and other leveraged funds.
Chart 2
Sources of leverage
a) Synthetic and financial leverage | b) Synthetic leverage by asset class |
|---|---|
(x-axis: synthetic leverage; y-axis: leverage through repos) | (y-axis: synthetic leverage) |
 |  |
Sources: AIFMD, SFTR and EMIR data.
Notes: Synthetic leverage is measured as gross synthetic exposure (EMIR) divided by NAV (AIFMD). Leverage through repos is measured as gross repo borrowing (SFTR) divided by NAV (AIFMD). The size of the circles is proportional to the NAV of the AIF groups. In panel a), leverage through repos is set to zero for other hedge funds to comply with data confidentiality guidelines. In panel b), we count currency and equity derivatives as “other derivatives” in the case of relative value hedge funds because of data confidentiality. HFs: hedge funds.
All types of leveraged AIFs have sizeable synthetic exposures, especially to interest rate derivatives, but only GBP-denominated LDI funds and hedge funds with relative value strategies use repo borrowing extensively to increase exposures. The relative importance of synthetic leverage and financial leverage through repo borrowing is illustrated in the scatter plot in panel a) of Chart 2. Panel b) disaggregates synthetic leverage by the asset class of the underlying. While LDI funds are almost exclusively exposed to interest rate derivatives, hedge funds and other AIFs are exposed to a wide range of derivatives, including currency derivatives, equity derivatives and credit derivatives.
Risks related to repo borrowing appear to be highest among hedge funds with relative value strategies, followed by GBP LDI funds. For relative value hedge funds, gross repo borrowing is 20 times larger than the NAV and more than 37 times larger than cash holdings (see Table 1), which indicates significant potential losses and liquidity risks related to leveraged bond positions. These ratios are much smaller, but still sizeable, among GBP LDI funds.
To investigate the risks from synthetic leverage, we compute risk metrics based on the initial margin posted instead of risk metrics based on gross synthetic exposures. Risks associated with derivatives depend not only on the notional exposure, but also on the type of derivative and the underlying. In addition, gross synthetic exposure does not take the netting of positions into account. For these reasons, gross notional exposures might overstate leverage-related risks for funds, especially for interest rate derivatives. Metrics based on initial margins seek to address these issues. The initial margin amount is calculated based on the potential future exposure of the derivatives portfolio and is therefore a measure of typical variations in its value. For this reason, comparing the initial margin posted to the loss-absorption capacity of AIFs (the NAV) or their available liquid assets to cover margin calls indicates the solvency and liquidity risks associated with a fund’s derivatives portfolio.
According to initial margins posted, relative value hedge funds, other hedge funds and EUR-denominated LDI funds face the highest levels of solvency and liquidity risks associated with the use of derivatives. While hedge funds with relative value strategies have very large gross synthetic exposures compared with their NAV, the ratio of initial margin to gross exposures is generally very low. This reflects the hedging and netting of market risk taken into account when calculating initial margins. For such strategies, gross synthetic exposures may thus overestimate the associated risks (see Table 1). If we instead consider metrics based on initial margins, the solvency and liquidity risks of relative value hedge funds do not stand out compared with other hedge funds or EUR LDI funds. For GBP LDI funds and other leveraged AIFs, the ratios of initial margin posted over NAV and over available cash are lower, pointing to smaller risks from synthetic leverage for these groups of funds.
3.2 Assessing the resilience of leveraged funds to simple interest rate shocks
To assess the resilience of leveraged AIFs to interest rate shocks, we estimate the impact of such shocks on the mark-to-market value of their leveraged bond and interest rate swap positions.[15] Given the importance of interest rate derivatives and repo borrowing for most leveraged AIFs, we consider a parallel upward shift of all interest rate curves, regardless of currency and collateralisation, similar to that in Jukonis et al. (2022). We then compute the impact on the NAV and the liquidity shortfall. Results aggregated to AIF categories are shown in Chart 3.
A large interest rate shock could lead to substantial losses for LDI funds, raising solvency risk. In the case of a 300 basis point upward shift in yields[16], the average loss to investors for GBP-denominated LDI funds is 31%. For EUR-denominated LDI funds, we find a somewhat smaller effect of 26%. Interestingly, the impact on GBP LDI funds works almost only through revaluation of leveraged bond positions from repo borrowing rather than through interest rate swaps, whereas the opposite is true for EUR LDI funds (see Chart 3, panel a). Despite the large size of the shock, we find that only very few LDI funds would have a NAV below zero in this scenario. The effect of a parallel shift of the yield curve on hedge funds and other leveraged funds is negligible, both at the fund level and aggregated to AIF categories. For relative value hedge funds, the low impact is likely related to offsetting positions in repo and derivatives markets. However, our analysis focuses on one risk factor (interest rate risk), while hedge funds might be exposed to other risk factors (credit spreads, cross-asset correlation) not factored into our stress scenario.
Chart 3
Effect of interest rate shocks
a) Impact on NAV (300 bp shock) | b) Impact on liquidity (100 bp shock) |
|---|---|
(y-axis: impact on NAV in %) | (y-axis: liquidity needs and shortfall in EUR billions) |
 |  |
Sources: AIFMD, SFTR and EMIR data, and own calculations.
Notes: We compute the effect of interest rate shocks on the valuations of bonds pledged as repo collateral and of interest rate swaps. The impact on the NAV is defined as the total change in valuations divided by NAV. To calculate liquidity needs per AIF category, we consider changes in liquidity due to margin calls and demand for additional repo collateral and add up liquidity needs among AIFs with negative changes in liquidity. To calculate the liquidity shortfall per AIF category, we compare the changes in liquidity at the AIF level with available liquidity (either cash or cash and money market fund shares) and add up the liquidity shortfall among AIFs for which the liquidity need exceeds available liquidity. For data confidentiality reasons, the decomposition of the impact on NAV cannot be shown for other hedge funds in panel a), the impact on NAV through repo collateral is set to zero for EUR LDI funds in panel a), and information on hedge funds cannot be shown in panel b). bp: basis point.
Milder shocks could still trigger liquidity issues for funds. For more than half of LDI funds, cash is not sufficient to cover margin and collateral calls in response to a 100 basis point upward shift.[17] The total liquidity shortfall is €6.6 billion and would be covered by the redemption of money market fund shares and sales of unpledged bonds (see Chart 3, panel b). Such a reaction could transmit stress to money market funds and/or to sovereign bond markets through procyclical asset sales. We find either no shortfalls or only negligible shortfalls among hedge funds and other leveraged funds.
4 Conclusion
This article explores the risks from leverage in the alternative investment fund sector by considering risk metrics and a stress scenario. Hedge funds and LDI funds have the highest leverage levels among AIFs. Whereas LDI funds are vulnerable to parallel shifts of the yield curve, this is not true for hedge funds. Our results indicate that most LDI funds are resilient in terms of NAV losses, but sizeable liquidity shortfalls still seem likely. For a comprehensive analysis of hedge funds’ vulnerability to interest rate risk and other types of risk, future work could consider additional scenarios, such as scenarios tailored to evaluating the specific risks of investment strategies commonly pursued by hedge funds.
Our analysis highlights how entity-level data, combined with transaction-level data on derivatives and repo agreements, can be used for risk monitoring. Granular data on derivatives and securities financing transactions provide macroprudential authorities and market supervisors with valuable information on the use of leverage by non-banks. This information can be combined with entity-level data to improve risk monitoring and policy assessment. Recent recommendations proposed by the FSB to mitigate leverage-related risks for NBFI entities have emphasised the usefulness of combining various granular data sets and using a range of metrics to assess risks from NBFI leverage (FSB, 2024). As well as helping authorities in the supervision of individual firms, such data are especially useful for assessing risks to financial stability, including by identifying groups of leveraged entities with similar risk exposures or concentrated positions.
The exploration of risks from leverage lays the necessary foundations for an assessment of policy options to contain these risks. Examples of such policies are direct leverage limits, minimum haircuts, higher initial margins and minimum shock absorption capacities such as the yield buffer requirement for LDI funds. Both the risk metrics and the stress scenario considered in this article are possible ways to quantify the effects of policies on the resilience of funds.
References
Central Bank of Ireland (2022), The Central Bank’s macroprudential policy framework for Irish property funds, November.
Commission de Surveillance du Secteur Financier (2024), Macroprudential measures for GBP Liability Driven Investment Funds.
European Securities and Markets Authority (2019), “EU Alternative Investment Funds”, Annual Statistical Report.
European Securities and Markets Authority (2021), Guidelines on Article 21 of Directive 2021/61/EU.
European Securities and Markets Authority (2022), “Leverage and derivatives – The case of Archegos”, ESMA Report on Trends, Risks and Vulnerabilities Risk.
European Securities and Markets Authority (2024), “Assessing risks posed by leveraged AIFs in the EU”, ESMA Report on Trends, Risks and Vulnerabilities Risk.
European Systemic Risk Board (2021), EU Non-bank Financial Intermediation Risk Monitor 2021.
European Systemic Risk Board (2023), EU Non-bank Financial Intermediation Risk Monitor 2023.
European Systemic Risk Board (2024), EU Non-bank Financial Intermediation Risk Monitor 2024.
Financial Stability Board (2024), Leverage in Non-Bank Financial Intermediation: Consultation report, December.
Ianiro, A., Leonello, A. and Ruzzi, R. (2025), “Measuring synthetic leverage in interest rate swaps”, Macroprudential Bulletin, Issue 26, ECB.
Jukonis, A., Letizia, E. and Rousová, L. (2022), “The impact of derivatives collateralisation on liquidity risk: evidence from the investment fund sector”, Working Paper Series, No 2756, ECB, December.
Molestina Vivar, L., Weistroffer, C. and Wedow, M. (2023), “Burned by leverage? Flows and fragility in bond mutual funds”, Journal of Empirical Finance, Vol. 72, pp. 354-380.
van der Veer, K., Levels, A., Lambert, C., Molestina Vivar, L., Weistroffer, C., Chaudron, R. and van Stralen, R. (2017), “Developing macroprudential policy for alternative investment funds”, Occasional Paper Series, No. 202, ECB, November.
The findings presented in this article are based on a broader ESRB project, which will be published later in 2025. We would like to thank Filippo Bucchi, Pierre-Emmanuel Darpeix, Dorota Okseniuk, Franck Raillon, Martin Saillard and Annegret Schäfer, who contributed to the ESRB project.
Note that some UCITS funds also take on high leverage (see Molestina Vivar et al., 2023; European Systemic Risk Board, 2024).
Directive 2011/61/EU of the European Parliament and of the Council of 8 June 2011 on Alternative Investment Fund Managers and amending Directives 2003/41/EC and 2009/65/EC and Regulations (EC) No 1060/2009 and (EU) No 1095/2010 (OJ L 174, 1.7.2011, p. 1).
All new Irish property funds will need to comply with a 60% leverage limit, while existing funds are granted up to five years to comply. See Central Bank of Ireland (2022).
Similarly, AIFMD data provide information on the NAV and available liquid assets of an investment fund but do not contain information on the duration of bond and interest rate swap positions. Therefore, AIFMD data alone are not sufficient to estimate the impact of interest rate shocks on NAV and liquidity..
Our article builds on previous work by van der Veer et al. (2017), European Systemic Risk Board (2023) and European Securities and Markets Authority (2024). It is related to the work in this edition of the Macroprudential Bulletin by Ianiro, Leonello and Ruzzi (2025), who also emphasise the role of interest rate risk. However, their focus is on developing a new methodology to measure synthetic leverage in interest rate swaps, whereas we explore risks from leverage in AIFs more broadly, including risks from leverage through repos.
AIFs report on a quarterly, semi-annual or annual basis, depending on the fund’s characteristics.
Regulation (EU) No 648/2012 of the European Parliament and of the Council of 4 July 2012 on OTC derivatives, central counterparties and trade repositories (OJ L 201, 27.7.2012, p. 1).
Regulation (EU) 2015/2365 of the European Parliament and of the Council of 25 November 2015 on transparency of securities financing transactions and of reuse and amending Regulation (EU) No 648/2012 (OJ L 337, 23.12.2015, p. 1).
AIFMD data as at the end of 2023 are the most recent data available.
This is based on Article 111 of the AIFMD, which defines “leverage on a substantial basis” as commitment leverage above 3. We also use three as the threshold to identify highly leveraged AIFs but because of the low data quality of the commitment leverage variable, we prefer to use AuM/NAV as a leverage metric. AuM is a measure of total exposures, including synthetic exposures.
LDI funds align their investment portfolios with an institution’s future liabilities, such as pension payouts, by hedging risks such as interest rate fluctuations and inflation.
The AIFMD provides a more granular classification of hedge fund strategies, but we restrict ourselves to the distinction between relative value strategies and other strategies to keep this article concise. Relative value hedge funds seek to capitalise on pricing inefficiencies between related financial instruments. They stand out because they obtain the highest levels of leverage and use repo borrowing extensively.
See FSB (2024).
We approximate changes in the valuations of interest rate swaps and bonds used as repo collateral based on three assumptions. 1) For bonds with a fixed coupon rate, we consider a shift of the yield to maturity and not of the yield curve. 2) For floating rate notes, we approximate the effect as zero. 3) Interest rate swaps can be viewed as a combination of a long position in a bond with a fixed coupon rate and a short position in a floating rate note. Since we do not have position-level information on unpledged bonds, we may underestimate the impact of an interest rate shock on NAV.
The size of the shock is due to the 300 basis point yield buffer requirement for GBP-denominated LDI funds, which was introduced in Luxembourg and Ireland under Article 25 of the AIFMD in response to the 2022 gilt market crisis (CSSF, 2024).
We follow European Systemic Risk Board (2023) and combine margin calls on derivatives and repos.