Introduction

If a market participant, say a hedge fund, were to sell 1% of the total available supply of Government of Canada (GoC) bonds—worth about $11 billion as of December 2024—how much would GoC bond prices drop? Our analysis suggests they could drop by an average of 0.2% over one quarter, which translates to a 2–basis point rise in the yield for 10-year GoC bonds.

Market participants and policy-makers alike need to understand how trading flows—both sales and purchases—affect GoC bond prices. Such insights help make clear how large flows can move markets and how participants within the financial system might behave during periods of stress. For the Bank of Canada (the Bank), this knowledge is particularly useful to:

  • assess the market impact of its policy actions
  • understand how institutional investors’ trading affects market stability
  • monitor potential vulnerabilities in the financial system

We base our analysis on a basic insight about how prices respond to flows in bond markets. When investors want to sell bonds, prices must decline to attract other investors to purchase them. This price impact crucially depends on how willing other investors are to absorb these bond flows. This willingness can be measured by the concept of price elasticity of demand. Our estimates indicate that, at an individual investor level, if GoC bond prices decrease by 1%, investors would prefer to increase their GoC bond holdings by 3%, all else being equal. This elasticity represents adjustments toward or away from GoC bonds that investors would like to make to their portfolios in response to price changes. Of course, if the total supply of bonds remains fixed, prices must adjust sufficiently to induce some investors to hold the available supply.

Our analysis suggests that quarterly fluctuations in GoC bond prices are driven primarily by changes in demand from large institutional investors, such as foreign investors and Canadian pension funds. Over the period from 2000 to 2025, on average, 69% of the quarterly fluctuation in GoC bond prices can be attributed to demand-side investors. The supply side—new GoC bond issuances minus the Bank’s purchases of GoC bonds—accounts for the remaining 31% of quarterly price movements in the GoC bond market.

Key market players and their impact

On the demand side, the GoC bond market is dominated by a few large institutional investors with distinct investment strategies:

  • non-residents, such as foreign central banks, sovereign wealth funds and foreign hedge funds
  • Canadian trusteed pension funds
  • Canadian chartered banks
  • other institutional investors, such as mutual funds and insurers

As of early 2025, non-residents and Canadian pension funds together held 62% of publicly available GoC bonds. Canadian mutual funds, insurers and banks held another 20% (Chart 1).

Chart 1: Non-residents and Canadian pension funds hold more than half the share of Government of Canada bonds in the market

To isolate the impact of flows on prices, we need to separate investor-specific trading decisions from flows that are motivated by changes in economic conditions or shifts in monetary or fiscal policy. We do this by:

  • identifying GoC bond flows that are unique to specific investor groups
  • measuring how an aggregate of these investor-specific flows affects GoC bond prices
  • controlling for common factors that influence all investors simultaneously

This approach is based on Gabaix and Koijen (2024) and relies on investor-specific flows that are large enough to affect bond prices. The resulting price moves can then be used to estimate the price elasticities of private institutional investors (demand side), who must absorb shifts in the supply of GoC bonds issued by the Canadian government.1

We estimate a common price elasticity for the demand side (see Table B‑1 in Appendix B). While a pension fund may respond differently to prices than a mutual fund would, having a single elasticity across all private investors captures their average price sensitivity as a group.2

Conceptually, our estimated elasticities reflect the idea that GoC bond prices can adjust for several reasons:

  • When changes in the aggregate demand for GoC bonds are balanced with changes in their supply—If the government raises the total available supply by issuing new bonds, prices adjust to incentivize investors to increase their GoC bond holdings.
  • When the total available supply remains unchanged—This occurs if one investor simply wants to sell a GoC bond; another investor needs to be incentivized to buy it. GoC bond prices adjust to support this reallocation.
  • When there are no observable trading flows—For example, if weaker-than-expected gross domestic product (GDP) data lead all investors to increase their demand for GoC bonds, prices may rise even if no immediate trading occurs. In this case, all investors’ collective willingness to hold GoC bonds increases, resulting in a higher price.3

Specifically, our estimates suggest that, if the typical market participant sells 1% of the amount outstanding of GoC bonds for reasons unrelated to the macroeconomy and monetary or fiscal policy, GoC bond prices decline by 0.2% over a quarter, which translates to a 2–basis point rise in the yield for 10-year GoC bonds (Chart 2).

Take the real-world example of the COVID-19 crisis in the first quarter of 2020. Our model shows the Bank’s purchase of $19 billion in GoC bonds would have caused GoC bond prices to rise by an estimated 1.1%, which translates to an 11–basis point decline in the 10-year yield. Our model's predictions are broadly aligned with previous research findings based on alternative methodologies (Arora et al. 2021; Diez de los Rios 2024; Azizova, Witmer and Zhang 2024).4, 5

Chart 2: Estimated impact of trading flows on the Government of Canada bond market

What drives Government of Canada bond prices?

Our model attributes quarterly GoC bond price movements to three distinct drivers:

  • Investor-specific factors—Each investor group responds to factors specific to their own circumstances, such as changes in their investment mandates.
  • Common factors that influence all investors—Investors react to common economic factors, such as growth in GDP. However, different investor groups may respond with varying intensity.
  • Size of investors’ GoC bond holdings—The influence of investor-specific and common factors on prices depends on the size of the investor group: larger investors have a greater impact on prices when they trade simply because their trades are bigger in absolute terms.

By examining these three drivers together, we can measure how much they contribute to the overall quarterly price movements in the GoC bond market.6

The clear pattern from our analysis is the importance of investor-specific factors in driving quarterly GoC bond prices (Chart 3). Bond supply changes account for 31% of quarterly price fluctuations. Almost all of this variation reflects unique factors specific to the government’s funding needs.

For the demand side, large institutional investors make up another 39%, with demand from non-residents being responsible for 24% and demand from Canadian pension funds for 15% of the overall quarterly price fluctuation. Within the model we use, these numbers reflect how prices must adjust to induce other investors and the supply side to accommodate changes in GoC bond holdings of a given investor group.

Chart 3: Quarterly fluctuations in Government of Canada bond prices are mainly driven by investor-specific factors

Important considerations

Although our analysis provides valuable insights, several factors that are not explicitly considered here can influence the actual price impact of trading flows:

  • Market conditions—Heightened uncertainty in stressed markets can make market participants less willing to take on risk. Investors may need larger price adjustments in stressed markets than in normal market conditions to compensate for the additional volatility.
  • Bond maturity—Long-term bonds carry more interest rate risk than short-term bonds, reflecting the greater uncertainty about future interest rates over longer time horizons. Investors may require larger price adjustments to accommodate the flows of long-term bonds.
  • Persistence of flows—Flows that revert quickly have lower price impacts than permanent shifts in bond holdings do. Market participants are therefore more willing to accommodate temporary imbalances, knowing they will soon reverse.

Conclusion

Understanding how flows affect GoC bond prices helps the Bank of Canada monitor and assess:

  • financing conditions for the government
  • the transmission of monetary policy
  • stability risks of large asset sales by the private sector

We find that large flows can impact GoC bond prices significantly, even in normal market conditions. The size of this impact depends on market participants’ willingness to accommodate the corresponding flows, as measured by their price elasticities.

Appendix A: Estimating the demand system for Government of Canada bonds

We base our estimates on the methodologies of Gabaix and Koijen (2021; 2024).

The demand and supply of bonds

We model the quarterly percentage change of sector \(\displaystyle i\)’s Government of Canada (GoC) bond holdings as

\(\displaystyle\, \Delta q_{it}\) \(\displaystyle=\, \alpha_i\) \(\displaystyle-\, ζ^d \, \Delta p_t\) \(\displaystyle+\, λ_{i}^{'} \, η_t \) \(\displaystyle+\, u_{it} \) \(\displaystyle,\)

where \( \Delta q_{it}\) \(=\, \frac{Q_{it}-{Q_{it-1}}} {Q_{it-1}} ,\) \( ζ^d \) is the demand elasticity, \( \Delta p_t\) is the log change of GoC bond price, \( η_t \) is a vector of common shocks, \( λ_{i}^{'} \) is a vector of factor loadings and \( u_{it} \) is an idiosyncratic shock.

Similarly, we model the quarterly change in the supply of GoC bonds to the private sector, that is, net of government holdings, as

\(\displaystyle\, \Delta S_{t}\) \(\displaystyle=\, \alpha^s\) \(\displaystyle+\, ζ^s \, \Delta p_t\) \(\displaystyle+\, \phi^{'} \, η_t \) \(\displaystyle+\, \epsilon_{t} \) \(\displaystyle,\)

where \( ζ^s \) is the supply elasticity, \( \phi^{'} \) is a vector of the supply sector’s factor loading and \( \epsilon_{t} \) is an idiosyncratic supply shock.

The GoC bond price must adjust to make sure that the aggregate change in investor demand for GoC bonds is equal to the change in the supply of bonds to the private sector. Let \( w_{it-1}\) \(=\, \frac{Q_{it-1}} {\sum_{i} Q_{it-1}} \) denote sector \(\displaystyle i\)’s holding share of GoC bonds. Given the above expressions for demand and supply, market clearing imposes that \( \Delta S_{t}\) \(=\, \sum_{i} \Delta q_{it}\). This implies that

\(\displaystyle\, \Delta p_{t}\) \(\displaystyle=\, \left(\frac{1} {ζ^d + ζ^s}\right)\) \((\sum_{i} w_{it-1} \alpha_{i} \) \(\displaystyle+\, u ̌_t\) \(+\, (λ^{ ̌ ' }-ϕ^{'} ) η_t\) \(\displaystyle-\, \alpha^{s} \) \(\displaystyle-\, \epsilon_{t} ) \) \(\displaystyle,\)

where \( u ̌_t\) \(=\sum_{i} w_{it-1} u_{it} \) and \( λ^{ ̌ ' } = \sum_{i} w_{it-1} λ^{'}_i \).

Granular instrumental variables

We follow Gabaix and Koijen (2021) and construct the granular instrumental variable (GIV). First, we obtain quarterly changes in GoC bond holdings for different sectors \( i\) from Statistics Canada’s National Balance Sheet Accounts and Financial Flow Accounts. Then we obtain the residuals, \( \Delta q ̌_{it}\), of a weighted regression of \( \Delta q_{it}\):

\( \Delta q_{it}\) \(=\alpha_{i} \) \(+ \alpha_{t} \) \(+ λ^{'}_i \,GDP\,growth_{t} \) \(+ \delta_{i} t \) \(+ \Delta q ̌_{it}\) \(,\)

where \(GDP\,growth\) is quarterly real gross domestic product growth, and a proxy for common shock, \(\delta_{i} t\), allows for a different time trend. Some sectors grow faster than others, and the weights are equal to the standard deviation of \( \Delta q_{it}\).7 We extract the principal components of \(E_{i}^{\frac{1} {2}} \Delta q ̌_{it}\) and denote the estimated vector of those components by \(η_t\). We construct the GIV instrument as \(GIV = \sum_{i} w_{it-1} \Delta q ̌_{it} \) and use it to estimate both demand elasticity, \(ζ^{d}\), and supply elasticity, \(ζ^{s}\).

Appendix B: Price elasticity estimates for the demand and supply sides

Table B-1 reports the estimated average sensitivity of quarterly changes in investor holdings of Government of Canada (GoC) bonds to quarterly returns of the GoC bond market. This estimate is based on the model described in Appendix A. Estimates are calculated over the sample period, from the second quarter of 2000 to the first quarter of 2025. Standard errors are reported in parentheses.

Table B-1: Sensitivity of investor flows to Government of Canada bond market returns

Table B-1: Sensitivity of investor flows to Government of Canada bond market returns
Sector Average elasticity of trading flows to returns 95% confidence interval
Private institutional investors (demand) -3.26%***
(0.76)
[-4.74,-1.78]
Combined Government of Canada and Bank of Canada (supply) 1.34%*
(0.81)
[-0.25,2.94]

Note: *** and * refer to statistical significance at the 1% and 10% levels, respectively.
Sources: Statistics Canada and Bank of Canada calculations

References

Arora, R., S. Gungor, J. Nesrallah, G. O. Leblanc and J. Witmer. 2021. “The Impact of the Bank of Canada’s Government Bond Purchase Program.” Bank of Canada Staff Analytical Note No. 2021-23. https://doi.org/10.34989/san-2021-23.

Azizova, C., J. Witmer and X. Zhang. 2024. “Assessing the Impact of the Bank of Canada’s Government Bond Purchases.” Bank of Canada Staff Discussion Paper No. 2024-5. https://doi.org/10.34989/sdp-2024-5.

Diez de los Rios, A. 2024. “Evaluating the Portfolio Balance Effects of the Government of Canada Bond Purchase Program on the Canadian Yield Curve.” Bank of Canada Staff Analytical Note No. 2024-22. https://doi.org/10.34989/san-2024-22.

Gabaix, X. and R. S. J. Koijen. 2021. “In Search of the Origins of Financial Fluctuations: The Inelastic Markets Hypothesis.” National Bureau of Economic Research Working Paper No. w28967. DOI 10.3386/w28967.

Gabaix X. and R. S. J. Koijen. 2024. “Granular Instrumental Variables.” Journal of Political Economy 132 (7): 2274–2303. https://doi.org/10.1086/728743.

  1. 1. Under our consolidated government perspective, changes in the supply of GoC bonds are the result of new bond issuance by the Government of Canada, net of maturing bonds and bond purchases by the Bank of Canada. Bond sales by the Bank would increase the supply. The Bank’s bond market activities include, for example, both deliberate policy measures, such as quantitative easing, and routine operations to manage its balance sheet. See “About the Bank of Canada’s balance sheet” for more details.[]
  2. 2. We allow our consolidated Canadian government grouping (the GoC and the Bank) to be reactive to GoC bond prices to better capture how supply-side factors interact with market conditions over time. We find the supply elasticity is substantially smaller than the demand elasticity. And if we remove the 2008–09 global financial crisis period, when issuance expanded while interest rates fell, then the supply elasticity would be statistically indistinguishable from 0. If we assume a supply elasticity of 0, meaning the consolidated government is completely unresponsive to prices, then our main conclusions are unchanged.[]
  3. 3. Our model captures shifts in macroeconomic expectations (e.g., GDP) through common factors, which influence all investors’ demand simultaneously. See Appendix A for more details.[]
  4. 4. A 1% sale of the available supply of GoC bonds impacts the value of the GoC bond market. This impact is the inverse of the sum of demand and supply elasticities: -0.2% = 1/(3.26 + 1.34)*(-1). This -0.2% return is converted into the approximate change in the 10-year GoC zero-coupon bond yield by dividing by the bond’s duration and multiplying by 100 to change from percent to basis points: (-0.2%/-10)*100 = 2 basis points.[]
  5. 5. Our model focuses on the portfolio channel of the Bank’s quantity-based policies and does not model any impact these policies may have by the Bank signalling a change of its policy stance to the market.[]
  6. 6. Allowing price elasticities to vary across investors groups could affect this decomposition.[]
  7. 7. Specifically, we construct pseudo-equal value weights, \(E_{i} = min \left(\frac{1.5} {N} , ξ \frac{σ_{i}^{-2}} {\sum_{i} σ_{i}^{-2}} \right) \), where \(σ_{i} = σ(\Delta q_{it})\) and \(ξ\) is chosen such that \( ∑_{i} E_{i} = 1\).[]

Acknowledgements

We thank Jean-Philippe Dion, Miguel Molico, Stéphane Lavoie, Antonio Diez de los Rios, Jean-Sébastien Fontaine, Toni Gravelle, Grahame Johnson and Stephen Murchison for helpful comments and suggestions. We are also grateful to Maren Hansen and Meredith Fraser-Ohman for editorial assistance, to Julie Porlier and Maxime Beaudet for translation assistance, and to Himawan Sudarso and Mike Dalziel for publishing this note on the web.

Disclaimer

Bank of Canada staff analytical notes are short articles that focus on topical issues relevant to the current economic and financial context, produced independently from the Bank’s Governing Council. This work may support or challenge prevailing policy orthodoxy. Therefore, the views expressed in this note are solely those of the authors and may differ from official Bank of Canada views. No responsibility for them should be attributed to the Bank.

DOI: https://doi.org/10.34989/san-2025-20

On this page
Table of contents