Does US or Canadian Macro News Drive Canadian Bond Yields?

Introduction

Market participants pay close attention to monetary policy and macroeconomic announcements. Bond yields of various maturities move in response to macro and monetary policy news as market participants adjust their views about the state of the economy and the stance of monetary policy.

Despite all the attention paid by market participants, the existing literature generally finds that macro news explains only a small share of daily changes to bond yields. For example, Altavilla, Giannone and Modugno (2017) find that US macro news explains less than 10 per cent of the daily variations in the yield curve of US Treasury bonds.

These small daily changes, however, sum up to a significant portion (up to 35 per cent) of monthly/quarterly variations. These changes therefore show that macro news has a persistent impact on bond yields, whereas other shocks are transitory and dissipate over longer horizons.

This fact helps reconcile the evidence provided in event studies that use daily data and find a small role for macro news in driving long-term rates with the evidence from studies that use monthly/quarterly data and generally find a larger role for macro news.

We first replicate the analysis in Altavilla, Giannone and Modugno (2017) for Canada. In small open economies, market participants also pay close attention to foreign news, particularly from the United States. We find that US and Canadian macro news combined explain less than 15 per cent of daily variations in the yield curve of Canadian bonds, but contribute up to 35 per cent in monthly or quarterly variations. We measure the relative contribution of domestic versus US macro news in driving changes in interest rates in Canada.

In an earlier study, Gravelle and Moessner (2001) show that US macro news has a larger impact on changes in daily changes of Canadian bond yields than Canadian news does. More recently, Diez de los Rios and Shamloo (2017) show that most of the premium on Canadian long-term bonds is driven by global factors. We revisit this question using data from January 3, 2000, to July 20, 2016, and find that US macro news still accounts for a larger share of the daily variations in Canadian bond yields than Canadian macro news does.

Moreover, we find that Canadian macro news contributes to roughly 10 per cent of the quarterly variation of the Government of Canada (GoC) bond yield curve, while US macro news explains up to 25 per cent. In a forthcoming paper, Donfack, Feunou and Sekkel (forthcoming) show that US macro news plays a larger role than domestic macro news in explaining changes to government bond yields in several other small open economies, such as Australia, Switzerland and Sweden.

Effects of macro news on bond yields at high and low frequencies

Altavilla, Giannone and Modugno (2017) develop a two-step approach to investigate how much of the daily (high-frequency), monthly and quarterly (low-frequency) changes in US bond yields can be explained by macro news.

In the first step, they regress daily changes in bond yields on surprises about data releases of macro variables, such as gross domestic product, employment and inflation. They find that macro news explains only a small share (about 6 per cent) of the daily changes in 10-year yields.

In the second step, they measure how much of the monthly (quarterly), changes in bond yields can be explained by macro news by summing the daily regression results over each month (quarter). This aggregation substantially increases the explanatory power of macro news. At the quarterly horizon, macro news explains about 35 and 25 per cent of fluctuations in short- and long-term bonds, respectively. This is because macro news has a persistent impact on bond yields, while other factors tend to average out over longer horizons. We provide a brief but more technical description of the methodology in the appendix.

We replicate this exercise using macro news and interest rates in Canada. Table 1 shows our complete set of Canadian and US macro news. We also find that macro news explains a small share of the daily variation in bond yields in Canada: less than 15 per cent for 2-year bond yields and less than 7 per cent for 10-year bond yields, as shown in Chart 1. Macro news becomes much more important, though, over longer horizons. It explains up to 24 per cent of the monthly and 34 per cent of the quarterly variation in long-term GoC bond yields (Chart 2 and Chart 3). These numbers are in line with Altavilla, Giannone and Modugno’s (2017) results for the United States. However, our results for Canada include both domestic and US macro news. Given that Canada is a small open economy, there is reason to believe US macro news can affect Canadian bond yields, even though the reverse is very unlikely.

The relative effects of US and Canadian macro news

Next, we describe the relative importance of US and Canadian macro news in explaining high- and low-frequency changes to Canadian bond yields. Chart 1, Chart 2 and Chart 3 show that US macro news explains a larger share of bond yield variation in Canada than domestic macro news does. This is true at all the sampling frequencies and at all bond yield maturities, except for daily changes to very short-term bonds. Moreover, the relative importance of US macro news increases with sampling frequency and bond maturity. At daily frequency (Chart 1), both US and Canadian news explain about 5 per cent of the variation of bond yields. Chart 3 shows that, at quarterly frequency, US macro news alone can explain more than 25 per cent of the variation in long-term bond yields, while Canadian macro news accounts for less than 10 per cent of this variation. This shows that US macro news has a more persistent effect on Canadian yields than Canadian news does.

The relative effects of US and Canadian macro news on real interest rates and break-even inflation rates

Nominal yields sum real interest rates and break-even inflation rates (BEIR). A natural step therefore is to analyze the role of these two components in explaining the low-frequency movements in nominal interest rates. BEIR are widely used by market commentators to gauge the expectations of investors regarding the outlook for inflation. Thus they are often interpreted as a measure of a central bank’s credibility.

Chart 4 shows how much of the quarterly variation in 10‑year Canadian real rates and BEIR can be explained by US and Canadian macro news. The difference is quite striking. Whereas Canadian macro news barely explains any fluctuation in 10‑year BEIR, it explains about 25 per cent of the quarterly variation of 10‑year real yields. In contrast, US macro news is responsible for about 35 per cent of the quarterly variation of 10‑year BEIR, but only 5 per cent of the 10‑year real rates.

Conceptually, the 10‑year BEIR can be thought of as the sum of two components:

  1. investors’ best forecast about what inflation will average over the next 10 years; and
  2. the inflation risk premium—the extra yield over that horizon that investors demand for bearing the inflation risk.

It is widely accepted that a key benefit of an inflation-targeting framework, such as the one in Canada since 1991, is that it provides a good anchor for inflation expectations (Ehrmann 2015). The inflation risk premium is then the likely channel through which US macro news affects long-term Canadian yields.

Conclusion

We find a strong relation between US macro news and movements in long-term Canadian bond yields. While the impact of Canadian macro news is important for short-maturity yields, its importance declines rapidly for longer maturities. The inflation risk premium appears to be the most likely channel through which US macro news affects Canadian long-term nominal rates.

Taken together, our findings show that nominal long-term rates in small open economies are largely driven by global factors, though domestic macro news still exerts some influence. However, movements of real rates appear to be more related to domestic macro news. In a forthcoming paper, we extend the results shown for Canada to other small open economies.

Finally, changes in nominal rates (due to macro news) might also affect exchange rates. How is the CAD$/US$ exchange rate affected by this flow of domestic and international macro news? We leave that for future consideration.

Appendix

Methodology

The empirical analysis starts by evaluating the daily reaction of bond yields to macro news during the period between January 3, 2000, and July 20, 2016. This is measured using a simple regression that relates daily changes in bond yields (\(Δy_{t}^τ\)) to the surprise component of news releases on a given day (\(news_{i,t}\)):

\(Δy_{t}^τ=c\) \(+\,\displaystyle\sum_{i=1}^{n}β_{i}^τ news_{i,t} \) \(+\,ε_{t}^τ \qquad (1)\)

These surprise components are determined by looking at the difference between headline news announcements and market participants’ expectations of those announcements. If variable \(i\) is not released at time \(t\), then \(news_{i,t}=0\). Then, we cumulate these changes in bond yields monthly or quarterly to get an estimate of the changes in asset prices over those periods due to news. Formally, a daily news index denoted by \(nix_{t}^{1,τ}\) is defined as the fitted value from Eq. (1), where the superscript “1” indicates that the frequency used to obtain the index is one day. By summing the daily news indexes, longer-horizon news indexes at daily frequencies are obtained as follows:

\(nix_{t}^{h,τ}=\displaystyle\sum_{j=0}^{h-1} nix_{t-j}^{1,τ} \)

After that, using a second stage regression,

\(y_{t}^τ-y_{t-h}^τ=γ^{h,τ} nix_{t}^{h,τ}\) \(+\,v_{t}^{h,τ}, \qquad (2)\)

we determine how much of the monthly or quarterly change in the bond yields of different maturities can be explained by macro news in Canada and the United States.

Charts and tables

Chart 1: Daily variation in Canadian bond yields explained by macro news

Sources: Bloomberg and Bank of Canada calculationsLast observation: May 2018

Chart 2: Monthly variation in Canadian bond yields explained by macro news

Sources: Bloomberg and Bank of Canada calculationsLast observation: May 2018

Chart 3: Quarterly variation in Canadian bond yields explained by macro news

Sources: Bloomberg and Bank of Canada calculationsLast observation: May 2018

Chart 4: Quarterly variation in the 10-year Canadian BEIR and real bond yields explained by macro news

Note: BEIR is the break-even inflation rate, typically measured as the difference between a long-run yield and an inflation-linked yield of the same maturity.
Sources: Bloomberg and Bank of Canada calculationsLast observation: May 2018

Table 1: Canada and United States macroeconomic news effects on bond yields at high frequency

Releases 2-year 5-year 10-year
Canada macroeconomic releases
Bank of Canada rate decision 0.545 0.148 -0.092
Building permits (MoM) 0.117 0.085 0.057
Business outlook future sales 0.144 0.162 0.137
Capacity utilization rate 0.129 0.107 0.097
Consumer price index -0.064 -0.044 -0.036
CPI core (MoM) -0.166 -0.061 -0.041
CPI core SA (MoM) -0.108 -0.092 -0.074
CPI core (YoY) 0.476 0.454 0.542
CPI NSA (MoM) -0.136 -0.166 -0.181
CPI SA (MoM) -0.096 -0.073 -0.076
CPI (YoY) 0.476 0.298 0.160
Current account balance -0.020 -0.082 -0.091
GDP (MoM) 0.554 0.500 0.588
GDP (YoY) -0.029 0.041 0.045
Housing starts -0.015 -0.048 -0.066
Industrial product price (MoM) 0.055 0.056 0.053
International merchandise trade 0.069 0.036 0.046
International securities transactions -0.081 -0.124 -0.168
Ivey purchasing managers index SA 0.068 0.039 0.014
Labour productivity (QoQ) 0.011 0.018 0.007
Manufacturing sales (MoM) 0.018 -0.053 -0.088
Net change in employment 0.564 0.492 0.509
New housing price index (MoM) -0.057 -0.057 -0.052
New housing price index (YoY) 0.006 -0.017 -0.026
Participation rate -0.095 0.005 0.003
Quarterly GDP annualized 0.058 0.021 -0.036
Raw materials price index (MoM) -0.115 -0.099 0.060
Retail sales ex. auto (MoM) 0.286 0.166 0.129
Retail sales (MoM) 0.112 0.128 0.071
Unemployment rate -0.202 -0.139 -0.078
Wholesale trade sales (MoM) 0.100 0.107 0.075
United States macroeconomic releases
Advance retail sales 0.204 0.250 0.215
Business inventories 0.012 0.028 0.014
Capacity utilization 0.240 0.216 0.198
Change in non-farm payrolls 0.769 0.779 0.578
Consumer confidence 0.135 0.152 0.116
Consumer credit 0.006 -0.004 -0.021
Consumer price index (MoM) 0.019 -0.023 -0.033
CPI Ex. food & energy (MoM) 0.060 0.070 0.047
Domestic vehicle sales 0.131 0.128 0.103
Durable goods orders 0.139 0.118 0.105
Employment cost index -0.049 -0.022 0.007
Factory orders 0.063 0.053 0.047
Housing starts 0.027 0.008 0.014
Imports price index (MoM) -0.033 -0.061 -0.090
Industrial production -0.062 -0.041 -0.085
Initial jobless claims -0.471 -0.454 -0.576
ISM manufacturing 0.552 0.544 0.545
ISM non-manufacturing composite 0.136 0.161 0.168
Leading indicators 0.015 0.039 0.046
Monetary policy surprise 0.240 0.218 0.158
New home sales 0.110 0.055 0.041
Personal income 0.014 -0.009 -0.007
Personal spending 0.063 0.040 0.017
Philadelphia Fed 0.201 0.210 0.194
PPI Ex. food & energy (MoM) 0.142 0.176 0.165
Producer price index (MoM) 0.016 0.012 0.007
Retail sales less autos -0.098 -0.131 -0.126
Trade balance 0.026 0.033 0.050
Unemployment rate -0.277 -0.068 -0.032
Wholesale investors -0.010 0.013 0.027
GDP annualized QoQ A 0.252 0.205 0.155
GDP annualized QoQ S 0.006 0.015 0.001
GDP annualized QoQ T 0.023 0.062 0.070
GDP price index A 0.024 0.052 0.070
GDP price index S 0.080 0.095 0.069
GDP price index T -0.119 -0.155 -0.116
Non-farm productivity P -0.120 -0.086 -0.081
Non-farm productivity F 0.119 0.136 0.148
Unit labour costs P -0.014 0.004 -0.001
Unit labour costs F 0.019 0.022 0.019
University of Michigan confidence P 0.221 0.260 0.214
University of Michigan confidence F -0.019 0.019 0.045
R2 0.133 0.103 0.071

This table lists for each Canadian and US macroeconomic release the value of the coefficients (in basis points) estimated from Eq. (1) for the bond yields with maturities \(τ\) at 2, 5 and 10 years. The values in bold are significantly different than zero at 5% confidence level (t-stat values are based on Newey-West HAC standard errors). The sample period goes from January 3, 2000, to July 20, 2016.

Table 2: Canada and United Sates macroeconomic news effects on bond yields at low frequency

2-year 5-year 10-year
Canada data: macroeconomic news effects on bond yields at low frequency
R2
1-day (h=1) 0.075 0.047 0.027
1-month (h=22) 0.105 0.081 0.075
1-quarter (h=66) 0.147 0.097 0.096
\(γ\)
1-month (h=22) 1.080 1.081 1.198
(0.38) (0.37) (0.36)
1-quarter (h=66) 1.206 0.997 0.992
(0.45) (0.35) (0.33)
Canada and United States data: macroeconomic news effects on bond yields at low frequency
R2
1-day (h=1) 0.133 0.103 0.071
1-month (h=22) 0.242 0.213 0.163
1-quarter (h=66) 0.343 0.316 0.261
\(γ\)
1-month (h=22) 1.259 1.256 1.211
(0.24) (0.20) (0.19)
1-quarter (h=66) 1.389 1.304 1.178
(0.34) (0.23) (0.19)

This table lists the R2 and the fitted \(γ\) values obtained from Eq. (2), for h=22 (monthly frequency) and h=66 (quarterly frequency) for the Canadian yields of bonds with maturities \(τ\) at 2, 5 and 10 years. Standard deviation (heteroskedasticity and autocorrelation consistent HAC) for \(γ\)* are reported in parenthesis. The sample period goes from January 3, 2000, to July 20, 2016.

References

  1. Altavilla, C., D. Giannone and M. Modugno. 2017. “Low Frequency Effects of Macroeconomic News on Government Bond Yields.” Journal of Monetary Economics 92: 31–46.
  2. Diez de los Rios, A. and M. Shamloo. 2017. “Quantitative Easing and Long‐Term Yields in Small Open Economies.” Bank of Canada Staff Working Paper No. 2017-26. Available at https://www.bankofcanada.ca/2017/07/staff-working-paper-2017-26/
  3. Donfack, M., B. Feunou and R. Sekkel. forthcoming. “Do Domestic or U.S. Macro News Drive Interest Rates in Small Open Economies?” Bank of Canada Staff Working Paper.
  4. Ehrmann, M. 2015. “Targeting Inflation from Below: How Do Inflation Expectations Behave?” International Journal of Central Banking. 11(S1): 213–249.
  5. Gravelle, T. and R. Moessner. 2001. “Reactions of Canadian Interest Rates to Macroeconomic Announcements: Implications for Monetary Policy Transparency.” Bank of Canada Staff Working Paper No. 2001-5. Available at https://www.bankofcanada.ca/2001/04/working-paper-2001-5/

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.

Follow the Bank