Could Canadian Bond Funds Add Stress to the Financial System?

Introduction

At the end of a hockey game, spectators face a dilemma. Should they follow the crowd rushing out of the arena and risk getting stuck in traffic? Or should they wait until the traffic clears and have a quick and smooth drive home?

In a sense, bond fund investors form a large crowd, like one at a sporting event, in financial markets. Because they are a crowd, fund investors cannot exit and sell their bonds at the same time without consequences. This raises concerns about what would happen to market liquidity if the investors all rushed to the exit at once.

A bond fund is a mutual fund that pools the money of many investors. Fund managers use the money to buy mainly fixed-income securities. The funds invest in relatively less-liquid assets but let investors redeem their shares for cash on any given day. In normal times, investments in bond funds are more liquid than investments in bonds partly because fund managers can match redemptions with cash from new investors.

But this is not necessarily true in times of stress. Experience shows that investors tend to redeem their shares at the same time when they sense higher risk. The liquidity mismatch between the fund’s assets and liabilities can amplify stress in financial markets if investors all try to “rush the exit,” causing fund managers to collectively sell assets that are less liquid.

We quantify the effects of a rush of investor redemptions in a hypothetical scenario where funds experience losses because interest rates are rising rapidly. To study this issue, we use Ceto, a new stress-testing analytical framework for Canadian investment funds (see Figure 1 and Arora et al. [2019] for more information).

In our risk scenario, interest rates in Canada rise rapidly by 100 basis points. For comparison, we use Ceto to analyze the scenario first with data from 2018 and then with data from 2007. Table 1 shows the results of this analysis.

Table 1: Bond funds play a larger role in intensifying shocks to the financial system

When investors require higher returns (price discounts) to buy corporate bonds, the cost of providing liquidity in this market rises. We find that the required returns are larger when we use 2018 data than they are with 2007 data. The greater results for 2018 reflect the rapid growth in the size and risk of the portfolios managed by bond funds since 2007 (Arora, Merali and Ouellet Leblanc 2018; Bédard-Pagé 2019).

Our results are sensitive to key assumptions about the mix of assets that fund managers use to meet redemptions and whether broker-dealers or long-term investors accommodate bond sales. Depending on our assumptions, the increase of required returns estimated for 2018 ranges from 63 to 137 basis points.

Our analysis suggests that the collective actions of bond funds have a greater impact on fixed-income market liquidity than they did in the past. As bond funds continue to grow, Ceto provides a necessary stress-testing framework to assess vulnerabilities and help us make informed policy decisions related to financial stability. The Bank of Canada will continue to monitor these funds and how they affect the financial system.

Stress testing bond funds

We want to assess how a rapid increase in interest rates caused by economic or financial conditions outside of Canada could affect bond funds. This is one of the key risks to the Canadian financial system identified in the Financial System Review (BoC 2018).

For bond fund managers, rising interest rates mean that their fixed-income portfolios lose value. To quantify these potential losses, we create a risk scenario that is severe but has happened in the past and conduct stress tests on the bond funds using Ceto. We calculate how an increase in interest rates of 100 basis points (over one quarter) would affect the value of bond fund holdings.

The Ceto stress test covers only bond funds, defined as open-ended mutual funds with large holdings of Canadian corporate bonds. Arora et al. (2019) contains information about the underlying data and explains how we identify bond funds.

Bond funds now hold 23 per cent of all Canadian corporate bonds, up from 12 per cent in 2007. On average across funds, holdings of Canadian corporate bonds represent 52 per cent of the funds’ portfolio. Our sample consists of 243 bond funds at the end of 2018 with total holdings of Can$354 billion in assets under management. Table 2 shows the growth of bond funds since 2007.

Table 2: Bond funds now hold more corporate bonds

*Excludes money-market funds and fund of funds
**According to book value

Investors redeem more after poor performance

Bond funds subject to our risk scenario experience losses in their portfolios. If they were to face an interest rate increase of 100 basis points, we estimate that they would incur average portfolio losses of 6.2 per cent in 2018 compared with 5.7 per cent in 2007. It is well documented that investors in mutual funds tend to redeem their shares in response to portfolio losses. Based on the historical relationship between fund flows and fund performance, Ceto predicts Can$68.5 billion in funds would be redeemed after losses of 6.2 per cent in 2018. This is a large increase relative to Ceto predictions for 2007 (Can$14.2 billion in funds redeemed) in part because bond funds had grown in size and had increased exposures to corporate bonds since 2007.

Funds rebalance portfolios to meet redemptions

When investors wish to redeem shares, fund managers must choose a mix of cash and bond sales to honour their commitments and provide cash to the investors. A common liquidity-management strategy fund managers use in normal times is called horizontal slicing. In this strategy, fund managers use liquid holdings (cash and government bonds) to meet redemptions.

Previous staff analytical notes (Arora and Ouellet Leblanc 2018; Arora, Fan and Ouellet Leblanc 2019) find that bond fund managers typically use liquid holdings but will also sell less-liquid assets (corporate bonds) to maintain the liquidity of their portfolio when they fear that they will continue to receive redemption requests. This liquidity-management strategy is called vertical slicing.

When we assume the fund manager will use vertical slicing, we find that an increase in the interest rate of 100 basis points in 2018 would cause greater sales of corporate bonds than in the past (Chart 1). Bond funds would sell Can$31.7 billion of Canadian corporate bonds over one quarter if this occurred in 2018 compared with Can$5.0 billion in 2007. This is equivalent to 5.5 per cent of corporate bonds for 2018 and 1.8 per cent for 2007. These comparisons are useful because they show us that bond funds could play a larger role than they have in the past in intensifying shocks to the financial system.

Chart 1: Impact of interest rate increases on corporate bond sales

Selling bonds depresses market prices

In Ceto, asset sales from bond funds can only be purchased by broker-dealers and long-term investors, such as insurance companies and pension funds. These market participants balance a trade-off between the required returns (price discounts) and the cost of financing when they decide to purchase corporate bonds. Financing in Ceto is provided by broker-dealers. For example, a large pension fund holding Can$1 billion in corporate bonds may be willing to buy more bonds if the discounted bond prices offer a good potential for returns relative to the cost of financing. When investors require higher returns (price discounts) to buy corporate bonds, the cost of providing liquidity in this market rises.

In our risk scenario using data from 2018, we estimate that the required return (price discount) increases by 93 basis points when bond funds sell Can$31.7 billion of Canadian corporate bonds to meet investors’ requests to redeem. In contrast, the same scenario in 2007 generates an increase in required returns of 44 basis points. The difference between 2007 and 2018 suggests that bond funds could have a greater impact on fixed-income market liquidity, as measured by the increase in required returns, than in the past.

Chart 2 shows that the estimated increase in required returns in 2018 would be greater than what was experienced during the oil price shock and the taper tantrum but less than during the global financial crisis.

Chart 2: Impact of bond fund asset sales on market prices

What if bond funds sell fewer corporate bonds?

Our results are sensitive to assumptions about the mix of cash and asset sales used to honour requests for redemptions. We analyze an alternative response (mixed liquidity-management strategy) where some bond funds sell more of their liquid assets to minimize the effect that the sale of corporate bonds has on prices (horizontal slicing), while some other bond funds choose vertical slicing to prevent their portfolio from drifting away from their target allocation. The mixed liquidity-management strategy is plausible if some fund managers prefer using liquid assets to meet redemptions when they anticipate a short-lived stress episode.

In our risk scenario using data from 2018, we estimate that a mixed liquidity-management strategy produces corporate bond sales of Can$17.8 billion, a decrease of 44 per cent relative to our baseline assumption (vertical slicing) (Chart 3). Chart 2 also shows that the required return declines by 30 basis points because some funds use liquid assets to meet investor redemptions.

Chart 3: Impact of mixed liquidity-management strategy

What if long-term investors do not buy corporate bonds?

For our analysis, we assume that long-term investors take advantage of discounted prices and buy corporate bonds being sold by bond funds. However, long-term investors may not be able or willing to increase their holding of corporate bonds in times of stress. We analyze an alternative assumption where broker-dealers are the only buyers of corporate bonds. In this case, broker-dealers provide liquidity to bond funds at a larger discount because they are subject to higher regulatory costs than long-term investors are.

Using data from 2018, we estimate that the required return would increase by an additional 44 basis points compared with our baseline assumption, to 137 basis points, if broker-dealers were the only buyers of bonds sold by the funds (Chart 2). This result indicates that asset sales by bond funds could exert greater pressure on market prices than our baseline assumption suggests if long-term investors do not buy corporate bonds.

Conclusion

Using Ceto, a new stress-test framework for investment funds, we quantify the effects of assets sales by bond funds in a scenario where interest rates rise rapidly. Because bond funds have grown in size and have increased their exposures to corporate bonds, we estimate that asset sales would have a much larger impact on fixed-income market liquidity than they did in the past.

How much market liquidity deteriorates is sensitive to our assumptions about the liquidity-management decisions of bond fund managers and the role of long-term investors in times of stress. Because bond funds continue to grow, Ceto provides a necessary stress-testing framework to quantify the effects of collective selling by bond funds on the financial system.

References

1. Arora, R., G. Bédard-Pagé, G. Ouellet Leblanc and R. Shotlander. 2019. “Bond Funds and Fixed-Income Market Liquidity: A Stress-Testing Approach” Bank of Canada Technical Report No. 115.
2. Arora, R. and G. Ouellet Leblanc. 2018. “How Do Canadian Corporate Bond Mutual Funds Meet Investor Redemptions?” Bank of Canada Staff Analytical Note No. 2018-14.
3. Arora, R., C. Fan and G. Ouellet Leblanc. 2019. “Liquidity Management of Canadian Corporate Bond Mutual Funds: A Machine Learning Approach.” Bank of Canada Staff Analytical Note No. 2019-7.
4. Arora, R., N. Merali and G. Ouellet Leblanc. 2018. “Did Canadian Corporate Bond Funds Increase their Exposures to Risks?” Bank of Canada Staff Analytical Note No. 2018-7.
5. Bédard-Pagé, G. 2019. “Non-Bank Financial Intermediation in Canada: An Update.” Bank of Canada Staff Discussion Paper No. 2019-2.
6. Bank of Canada. 2018. Financial System Review (June).

Acknowledgments

We thank Don Coletti, Jean-Sébastien Fontaine, Toni Gravelle and Virginie Traclet for helpful comments and suggestions. We are also thankful to Adriano Palumbo for excellent research assistance.