Do hedge funds support liquidity in the Government of Canada bond market?

Introduction

In March 2020, shutdowns associated with the spread of COVID‑19 led to turmoil in financial markets globally. Asset managers sold large volumes of fixed-income securities within a short time period, while dealers did not have sufficient capacity to purchase securities using their own balance sheets (Fontaine et al. 2021). This episode highlights how the growth in asset managers' assets under management relative to the aggregate intermediation capacity of dealers has increased the likelihood that a sudden spike in asset managers’ needs for liquidity could exceed the supply. Such instances can lead to stressed liquidity conditions in fixed-income markets (Bank of Canada 2022). To learn more about this vulnerability, staff at the Bank of Canada have conducted a range of analytical projects to better understand the behaviour of mutual funds, pension funds and life insurers in periods of market turmoil (see, for instance, Ouellet Leblanc and Shotlander 2020 and Bédard-Pagé et al. 2021).

In this analytical note, we focus on hedge funds and study their trading activity in the Government of Canada (GoC) bond market. We focus on the GoC bond market because of its importance for the Canadian financial system, given the use of GoC bonds to price other securities, to facilitate secured funding and to finance the federal government (Fontaine, Selody and Wilkins 2012). For our analysis, we use transaction-level data from October 2019 to December 2022. While the data do not include transactions in the primary market or with the Bank of Canada, our analysis offers insights into hedge fund activity in the secondary market.

We find that outside the peak period of market turmoil from March 9 to 20, 2020, GoC bond transactions of hedge funds are typically in the opposite direction to those of other market participants—that is, when other market participants sell a bond, hedge funds buy that bond, and vice versa. This finding shows how hedge funds’ trading activity could support market liquidity. Their transactions typically promote a two-sided GoC bond market, which can help dealers more easily find the bonds or cash needed to fulfill the transactions of their different clients. On the days that hedge funds trade in the opposite direction, we find that their transactions tend to earn higher returns than the average returns of GoC bonds. This outperformance could be compensation for supporting market liquidity or could be related to other investment strategies. Nevertheless, we find that hedge funds can at times amplify declines in market liquidity, because in the peak period of market turmoil in March 2020, hedge funds sold GoC bonds, just as other market participants did.

Hedge funds have greater investment flexibility than other asset managers

The classification of an asset manager as a hedge fund can be difficult because hedge funds’ strategies and the terms and conditions they offer can differ significantly from one fund to another. Like other asset managers, hedge funds pool together and invest capital on behalf of their clients. A key distinction, however, is that hedge funds face fewer regulations that address the risks they can take in their investments. This provides them with greater flexibility than other asset managers have to take on leverage, short sell securities and use derivatives. They have this flexibility because their prospective clients are typically limited to investors that meet certain net income, financial asset or minimum investment requirements, and because they typically offer redemptions at long frequencies (see Ontario Securities Commission 2007 for details).

We identify hedge funds in our data by merging our data with the Investment Industry Regulatory Organization of Canada’s list of Bare Trustee Agreements. This list classifies counterparties into different groups, including hedge funds. We then manually reclassify some counterparties as hedge funds based on our own research into their business models (see the Appendix for details).

Hedge funds are active in the Government of Canada bond market

We start by examining hedge funds’ transaction volume in the GoC bond market to get a sense of their market presence. Chart 1 compares the daily share of client-to-dealer transactions that come from hedge funds and other types of clients. Throughout our analysis, we define clients as market participants other than dealers or banks. The chart shows that hedge funds are second to wealth managers for the largest share of transaction volume in the GoC bond market, representing around 28%.

Chart 1: Hedge funds have the second largest share of transaction volume

Hedge funds typically transact in the opposite direction of other market participants

Two-sided markets can help dealers more easily fulfill the transactions of their different clients, potentially supporting market liquidity. To assess whether the transactions of hedge funds promote two-sided markets, we estimate the extent to which hedge funds trade GoC bonds in the opposite direction to other clients. Our measure of opposite direction transactions is the ratio of hedge funds’ net daily transaction volume for each GoC bond relative to that of other clients. We calculate the average of this ratio across bonds on each day. We exclude the period between March 9 and 20, 2020, to get a better idea of the typical behaviour of hedge funds. Our measure does not consider whether hedge funds initiate a transaction. It is plausible that hedge funds demand liquidity while they transact in the opposite direction of other clients. Nevertheless, our measure is useful for assessing hedge funds’ contributions to two-sided markets.

Chart 2 shows the median of the opposite direction ratio over our sample period for hedge funds and for other types of clients. Hedge funds have a median ratio of around -14%, which means that hedge funds typically trade 14% of the volume of GoC bonds transacted by other clients, but in the opposite direction to other clients. Another interpretation is that, all else being equal, without hedge funds, dealers would have to intermediate an additional 14% of transaction volume from other clients, using their own balance sheets. Most other types of clients’ transactions are typically either not in the opposite direction or have smaller opposite direction ratios.

Chart 2: Hedge funds’ transactions are in the opposite direction to those of other clients

Our results show that, on average, over the sample period, the trading activity of hedge funds typically promotes a two-sided GoC bond market. Nonetheless, they could be transacting in the opposite direction for many reasons. Market intelligence discussions find that hedge funds engage in a range of relative value and other investment strategies in the GoC bond market. One common theme of these strategies is that they involve taking opportunistic positions in GoC bonds that may be temporarily mispriced. Temporary mispricing can occur if the demand for a bond from other clients exceeds the supply offered by dealers, or vice versa. This presents an opportunity for hedge funds or other market participants to capitalize on the mispricing by transacting in the opposite direction. On certain days, however, hedge funds’ transactions may cause a temporary mispricing, rather than correct it, if their demand for a bond exceeds the supply available.

Hedge funds generate excess returns when transacting in the opposite direction

In this section, we investigate our hypothesis that hedge funds transact in the opposite direction to capitalize on temporary imbalances between the supply and demand of GoC bonds. We evaluate whether hedge funds’ transactions generate higher returns on days when they transact GoC bonds in the opposite direction of other clients.

We follow the methodology of Czech et al. (2021) to construct a GoC bond portfolio based on the bonds that hedge funds bought and sold the most on each day they transact in the opposite direction of other clients. We then calculate the excess returns of each day’s portfolio over different horizons (see the Appendix for details). This approach is only a proxy to assess hedge funds’ excess returns because their strategies may involve assets other than GoC bonds. Nevertheless, the approach is useful to assess whether hedge funds are capitalizing on imbalances in the GoC bond market.

Chart 3 shows the excess returns from hedge funds’ GoC bond transactions over a 1-, 5- and 10-day horizon. These excess returns are statistically significant and increase up to the 5-day horizon but lose significance and return close to zero at the 10-day horizon. These results suggest that on days when hedge funds transact in the opposite direction, they could be capitalizing on temporary supply and demand imbalances because their transactions generate excess returns over a short horizon and then decline toward zero.

Chart 3: Hedge funds earn excess returns when they transact in the opposite direction

Although the magnitude of hedge funds’ excess returns is small, leverage can enhance these returns. Hedge funds can use more leverage because they are subject to fewer regulations than some other types of asset managers. The repo market is commonly used to take on leverage when trading GoC bonds. Indeed, Chart 4 shows that hedge funds represent around 36% of the daily volume of repo transactions from clients to dealers where GoC bonds are the collateral. This is the largest share when compared with those of other types of clients, providing some indication of hedge funds’ potential use of leverage to magnify their returns from trading GoC bonds.

Chart 4: Hedge funds have the largest share of repo transaction volume

While leverage can enhance returns, it can also increase vulnerabilities for financial markets given the potential for enhanced losses. Faced with large losses, market participants may reduce their leverage and liquidate some of their positions. Such liquidations can further amplify price movements and strain market liquidity (see footnote 9 in Bank of Canada 2023). Nevertheless, our results are consistent with the interpretation that hedge funds transact in the opposite direction of other clients to capitalize on imbalances in the GoC bond market. This behaviour supports a two-sided market and consequently can support market liquidity.

In the peak of turmoil, hedge funds contributed to net selling

In periods of market turmoil, markets often become one-sided because a range of market participants can experience greater needs for liquidity and therefore sell assets at the same time. In this section, we focus on the episode of market turmoil during the spread of COVID‑19 in March 2020 to examine whether hedge funds contributed to the one-sided market for GoC bonds.

Chart 5 shows the net transaction volume of GoC bonds from hedge funds and other clients. March 9 to 20 is the period of peak illiquidity in the GoC bond market: the bid-ask spread proxy of GoC bonds reached a high of more than 35 basis points. During this period, hedge funds were net sellers of GoC bonds just as other clients were.

Chart 5: Hedge funds sold Government of Canada bonds, on net, during the peak of market turmoil

The limitations of our data create challenges for determining the exact reasons for hedge fund behaviour during this period. However, lessons from other asset managers in Canada and abroad could offer insights into why hedge funds were net sellers of GoC bonds. In the United States, hedge funds active in the Treasury market reduced their leverage to unwind relative value strategies (Barth and Kahn 2021). Past analysis by Bank staff finds that Canadian pension funds had to meet margin calls (Bédard-Pagé et al. 2021), while mutual funds in Canada experienced large redemption requests (Ouellet Leblanc and Shotlander 2020). Concerns around redemptions, increased margin calls or a desire to reduce leverage are all potential reasons for hedge funds’ sales of GoC bonds. Regardless of the exact reasons, periods like March 2020, when different market participants transact in the same direction, can add to strains on market liquidity.

Conclusion

While GoC bond transactions of hedge funds are typically in the opposite direction to those of other market participants, we find that during the peak period of market turmoil in March 2020, hedge funds sold GoC bonds, just as other market participants did. This shows that hedge funds can at times contribute to one-sided markets and amplify declines in market liquidity. These results help to advance Bank staff’s understanding of the asset management sector and of asset managers’ behaviour in periods of market turmoil.

Appendix

Data

We use data from the Market Trade Reporting System (MTRS). MTRS contains a record of transactions at Canadian broker-dealers registered with the Investment Industry Regulatory Organization of Canada (IIROC), the self-regulatory organization for the Canadian securities market. The data contain secondary market fixed-income cash and repo transactions of broker-dealers. Transactions in the primary market or with the Bank of Canada are excluded from the dataset. Counterparties of broker-dealers are assigned legal entity identifiers (LEIs).

We use transactions from October 2019 to December 2022 because the LEI information is reliable from October 2019 onward. We restrict our sample to only transactions between broker-dealers and their clients and drop transactions with banks and other broker-dealers. To group LEIs into different types of asset managers, we first use IIROC’s list of Bare Trustee Agreements, which provides information on entities’ investment offerings. We look up the entities from this list in the Global LEI Index to find their corresponding LEI, and then match this information by LEI to our transaction data. We then manually reclassify entities if applicable, using our judgment based on research of the entities’ public websites. We also combine mutual funds, exchange-traded funds and some other asset managers that offer pooled investment products into a group called wealth managers.

We gather daily GoC bond prices from the FTSE Russell 200 index. Specifically, we calculate the mid-price of each bond as the average of the bond’s bid and ask prices. We also exclude from our sample bonds that have a remaining term to maturity of less than one year.

Methodology of excess return calculations

We follow the methodology from Czech et al. (2021). For each day and each GoC bond, we calculate the hedge fund sector’s order flow, which is the total net transaction volume divided by the total gross volume. We then rank the order flow of the bonds on each day from highest to lowest and construct a long-short portfolio that purchases GoC bonds in the top third and sells those in the bottom third. On each day, we calculate holding period returns of this portfolio from 1-day to 10-day horizons. That is, we have 10 return series, 1 for each holding period. Finally, we filter our sample to days when hedge funds transact GoC bonds in the opposite direction of other market participants.

To measure excess returns, we regress each return series against a constant and three yield curve factors that explain movements in bond returns:

• level factor—the average return across all GoC bonds on a given day weighted by each bond’s amount outstanding
• slope factor—the 10-year minus the 2-year GoC bond returns
• curvature factor—the average of the 10-year and 2-year GoC bond returns minus the 5-year GoC bond return

For the slope and curvature factors, on each day we select GoC bonds with durations closest to 2, 5 and 10 years. The constant in our regression represents the average return that is not explained by the yield curve factors. We repeat this methodology for other types of clients.

Robustness of the opposite direction transaction measure

To ensure the approach used in our analysis to measure opposite direction transactions is robust, we consider three alternative methods and find that our overall conclusions hold. The three alternative methods are the following:

• We calculate the same measure, but after excluding bond transactions one week before and one week after the bonds were auctioned to ensure our results are not based on trading patterns unique to primary auctions.
• We calculate the share of weeks in our sample where hedge funds’ transactions were in the opposite direction of those of other clients.
• We estimate the extent to which a type of client contributes to offsetting transactions. For each day and each bond, we calculate the share of transactions that could be offset by clients alone. We then remove a client type and recalculate the measure. We find that the share of offsetting transactions declines the most after removing hedge funds compared with when removing other types of clients.

References

1. Bank of Canada. 2022. Financial System Review.
2. Bank of Canada. 2023. Financial System Review.
3. Barth, D. and R. J. Kahn. 2021. “Hedge Funds and the Treasury Cash-Futures Disconnect.” Office of Financial Research Working Paper No. 21-01.
4. Bédard-Pagé, G., D. Bolduc-Zuluaga, A. Demers, J.-P. Dion, M. Pandey, L. Berger-Soucy and A. Walton. 2021. “COVID‑19 Crisis: Liquidity Management at Canada’s Largest Public Pension Funds.” Bank of Canada Staff Analytical Note No 2021-11.
5. Czech, R., S. Huang, D. Lou and T. Wang. 2021. “Informed Trading in Government Bond Markets.” Journal of Financial Economics 142 (3): 1253–1274.
6. Fontaine, J.-S., C. Garriott, J. Johal, J. Lee and A. Uthemann. 2021. “COVID‑19 Crisis: Lessons Learned for Future Policy Research.” Bank of Canada Staff Discussion Paper No. 2021-2.
7. Fontaine, J.-S., J. Selody and C. Wilkins. 2009. “Improving the Resilience of Core Funding Markets.” Financial System Review (December): 41–46.
8. Ontario Securities Commission. 2007. “Hedge Funds.” CSA Staff Notice 81-316. January 12.
9. Ouellet Leblanc, G. and R. Shotlander. 2020. “What COVID‑19 Revealed About the Resilience of Bond Funds.” Bank of Canada Staff Analytical Note No. 2020-18.

Acknowledgements

We thank Jean-Philippe Dion, Jean-Sébastien Fontaine, Toni Gravelle, Alexandra Lai, Stephen Murchison, Andreas Uthemann, Adrian Walton and Jun Yang and for helpful comments and suggestions. We are also grateful to Anne Génard and Malcolm Fisher for research assistance, to Alison Arnot and Jordan Press for editorial assistance and to Philippe Audet-Cayer and Maxime Beaudet for translation assistance.