Fixed-income dealing and central bank interventions

Central bank asset purchases at the onset of the COVID‑19 pandemic

In March 2020, the spread of COVID‑19 led to turmoil in global financial markets (Fontaine et al. 2021). The reason for this was that large asset managers needed to convert assets to cash to meet redemptions and margin calls. As a result, asset managers sold large volumes of fixed-income securities to dealers who had both:

• limited ability to take on inventory
• limited risk tolerance

Trading became increasingly difficult and costly, with some markets completely seizing. For example, Chart 1 shows investors sold a high volume of bankers’ acceptances in Canadian markets at the height of the crisis.1

Chart 1: Investors were selling bankers’ acceptances to dealers in March 2020

Sources: Market Trade Reporting System 2.0 and Bank of Canada calculationsLast observation: June 30, 2020

To calm markets, many central banks intervened using several measures, including asset purchase programs. These programs keep markets functioning well by allowing dealers to expand their inventories. Chart 2 shows how the Bank of Canada’s asset purchase programs expanded its balance sheet beginning in March 2020.2 Although asset purchase programs are common, existing literature has not fully considered the optimal policy for central banks purchasing assets when dealers face binding inventory constraints.

Chart 2: The Bank of Canada’s balance sheet grew rapidly during the COVID‑19 crisis

Source: Bank of CanadaLast observation: December 30, 2020

A model of fixed-income dealing and central banking

We develop a simple model of a fixed-income market that includes certain key features.3 Many fixed-income markets, including Canada’s, rely on a network of dealers to match buyers with sellers. This is similar to many other real-life markets, such as for used cars, where private parties can trade with each other, but many choose to use a dealer instead.

By acting as a central source of trading in the bond market, dealers serve an important function—reducing difficulties buyers and sellers may have in finding each other. But relying on dealers for trading can create economic issues. Dealers incur costs, for example, from holding inventories, while also charging a markup to their clients to earn a profit. In our model, we assume that buyers and sellers cannot interact directly; instead, dealers act as intermediaries.

Our model generates predictions that match stylized facts about the March 2020 market turmoil. Several analyses show that it became extremely difficult to sell fixed-income securities during March 2020, and agency trading—where dealers only match without taking any net position—increased sharply (Fontaine et al. 2021; Kargar et al. 2020). In some cases, markets broke down completely and securities could not be traded.

We represent a financial crisis in our model by using a much greater number of sellers than buyers, resulting in dealers taking on a large inventory. Much like in practice, dealers in our model face inventory costs from purchasing too many bonds and become unwilling to buy without a large drop in price. Lower prices bring more buyers into the market, but prices may not move enough for dealers to want to take on enough inventory for sellers to meet their cash needs. In the more extreme scenarios in our model, dealers are unwilling to buy additional securities, representing a market breakdown.

To correct market disruption, central banks intervene with dealers. In our model, the central bank can do this either through asset purchases or collateralized loans, which are similar to repurchase (repo) agreements. These actions have different effects on dealers’ leverage, which can affect how the central bank intervenes most effectively.

To intervene, the central bank announces a price and then accepts quantities of sales or loans that dealers agree to at that price.

We show that central bank interventions can improve welfare in a crisis. They do so by allowing dealers to:

• buy more bonds from sellers
• sell these bonds to the central bank immediately, rather than holding them as costly inventory

In turn, prices do not drop—and fixed-income yields do not increase—as much as they would without central bank intervention, and sellers suffer fewer losses. In Chart 3, we show that the impact on dealers’ buying and selling is largest when:

• competition among dealers is limited
• inventory costs are high

Chart 3: Impact of optimal central bank interventions on quantities traded by dealers

Chart 3: Impact of optimal central bank interventions on quantities traded by dealers

Note: This chart illustrates the quantities bought and sold by dealers, before and after optimal central bank intervention. The panel on the left shows the impact of the costs of inventory. The panel on the right shows the impact of the number of dealers in the market.
Source: Bank of Canada calculations

In many cases, limited competition reduces the effect that changes to privately generated costs have on dealer inventories, such as repo funding costs. However, in our model the optimal amount the central bank intervenes per dealer is larger when competition is limited, significantly lowering the costs for individual dealers and making it easier for them to take on more inventory. Chart 4 shows that, for the same reasons above, the price increase is largest (and yields fall the most) when:

• inventory costs are high
• competition among dealers is limited

Chart 4: Impact of optimal central bank intervention on dealers’ trading prices

Chart 4: Impact of optimal central bank intervention on dealers’ trading prices

Note: This chart illustrates the bid and ask prices for the asset, before and after optimal central bank intervention. The panel on the left shows the impact of the costs of inventory. The panel on the right shows the impact of the number of dealers in the market.
Source: Bank of Canada calculations

Policy implications

Traditionally, central bankers have been guided by Bagehot’s principles (Bagehot 1873), which are summarized as follows: “[T]o avert panic, central banks should lend early and freely (ie without limit), to solvent firms, against good collateral, and at ‘high rates’.” (Madigan 2009)

Our model suggests that central banks may wish to purchase assets or lend at more favourable rates if no other concerns exist. For example, a central bank can intervene effectively by pricing purchases at an asset’s fair value, encouraging dealers to purchase securities from sellers. In addition, we show that a risk-neutral central bank can improve welfare by intervening above an asset’s fair value, which encourages dealers to purchase even larger quantities.

However, buying assets above fair value may contradict other central bank objectives. Central banks may face costs from factors such as:

• moral hazard—the concern that the central bank’s actions encourage risk taking by market participants
• the financial risks from maintaining a large inventory

Because our model is linear, it provides useful input for thinking about the size of a central bank’s interventions, but not about whether the central bank should intervene at all. Our model shows that the costs of these interventions may limit their size. This is particularly true when many competitive, small dealers are involved and when asset risk is high, suggesting that central banks may be most willing to intervene in large-scale, highly concentrated, low-risk markets.

Finally, we draw conclusions on the differences in the impact on market liquidity between central bank loans and asset purchases. With no further constraints, asset purchases and loans serve similar functions in improving market liquidity in our model. However, in practice, central bank loans may increase the leverage of dealers. Increasing leverage brings banks closer to internal and regulatory constraints. For example, Canada’s banking regulator, the Office of the Superintendent of Financial Institutions (OSFI), imposes a strict limit on bank leverage, known as the “leverage ratio.”

At the onset of the market turmoil in March 2020 brought on by the COVID‑19 pandemic, the Bank of Canada conducted a combination of repo agreements and asset purchase programs. At the same time, OSFI exempted central bank reserves from banks’ leverage ratios.4 Our model suggests that central bank lending may have been less effective at improving conditions in fixed-income markets if OSFI had not taken this action.

Endnotes

References

1. Bagehot, W. 1873. Lombard Street: A Description of the Money Market. London: Henry S. King and Co.
2. Cimon, D. A. and A. Walton. 2022 “Central Bank Liquidity Facilities and Market Making.” Bank of Canada Staff Working Paper No. 2022-9.
3. 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.
4. Kargar, M., B. Lester, D. Lindsay, S. Liu, P.-O. Weill and D. Zúñiga. 2020. “Corporate Bond Liquidity During the COVID‑19 Crisis.” National Bureau of Economic Research Working Paper No. 27355.
5. Madigan, B. F. 2009. “Bagehot’s Dictum in Practice: Formulating and Implementing Policies to Combat the Financial Crisis.” Speech to the Federal Reserve Bank of Kansas City’s Annual Economic Symposium, Jackson Hole, Wyoming, August 29.