Federal Bond Durations, the Bank of Canada, and other Random Thoughts

If that headline doesn’t attract readers, I don’t know what will.

TL;DR version: The Federal Government has issued a whole lot of short-to-medium term debt. Surprisingly, the interest payments the feds will have to pay on that debt when it rolls over could go down. While commentators are advising the government “lock in” interest rates by using long-duration debt, the federal government is currently paying billions in extra interest costs thanks to past lock-ins.

There’s been a fair bit of discussion in the media and on Twitter on whether or not the federal government should be “locking in” rates by issuing more long-duration debt, and less short-duration debt. In an attempt to help people answer that question, I thought it would be helpful to add some data onto the discussion.

The Bank of Canada recently released their end of November data on bond issues and bond holdings. You can find them here:

• Bank of Canada Bond Holdings — Nov 30, 2020
• Government of Canada Treasury Bills and Domestic Marketable Bonds Outstanding — Nov 30, 2020

This piece will just focus on the bond issues, but from those two data sets we see that the Federal Government has 265.9 billion in T-Bills outstanding, which within the next 365 days, with the Bank of Canada owning 60.5 billion.

I combined the bond information from the two datasets to make them easier to analyze. You can access the combined dataset on Google Sheets.

Here’s the data in graphical form, showing how much bond debt matures each year, up to 2051, which is all the bonds outstanding from the federal government, outside of a 4.75B issuance that matures in 2061:

A few notes:

1. 63% of the nearly 800B of bonds outstanding mature in the next five years. This does not include the 265.9B of outstanding T-Bills.
2. Half of all bonds held by the Bank of Canada (270.3B) mature in the next three years. If the Bank of Canada needed to reduce the size of their balance sheet, the mechanics here are really simple — they let the bonds mature and reduce their purchases of new ones. Given that the Bank of Canada is projecting interest rates to be at the zero lower bound until 2023, their bond portfolio is more likely to grow than shrink over this period.
3. That spike in 2030 is due to a high number of 10-year bonds issued this year. Similarly, the 2051 spike is due to 30.8B of 30-year bonds being recently issued.

I thought it would also be enlightening to map out the average (weighted) coupon rate for all of these bonds. Here’s the data for 2020–30, when the vast majority of these bonds mature:

In 2022 there will be a large quantity of 2-year bonds, issued this year, that will mature, that have very low coupon rates. In all other years, coupon rates are above 1%, many significantly so. Given the current yield curve is sub-1 until you get out about 20 years or so:

Source: Bank of Canada

There is a very good chance when these bonds roll-over, they will be issued with lower coupons than they have today, even if they are issued at longer-maturities.

Here’s what the coupon data looks like out to 2051:

I know a lot of commentators will see all the debt that is maturing in the next 5 years and argue that the federal government should issue more long-term debt. I suspect that the government will agree with this viewpoint, and issue more long-term debt over the next three years.

At the risk of being contrarian (who, me?), I’ll suggest that this would be a mistake. Issue long-bonds has been a losing bet for provincial and federal governments over the last 25 years because:

• The yield curve is generally upward sloping, so short debt has lower interest rates than long AND
• We’ve experienced a 25-year secular decline in interest rates

Notice those interest rate spikes in 2033 and 2037? Here’s a few examples of the federal government issuing long-bonds and it turning out very badly:

• CA135087XG49, 12 billion of 30-year bonds, maturing in 2033. Coupon rate: 5.75%
• CA135087XW98, 11.7 billion of 30-year bonds, maturing in 2037. Coupon rate: 5%.
• CA135087UT96, 2.5 billion of 30-year bonds, maturing in 2023. Coupon rate: 8%.
• CA135087TZ75, 567 million of 30-year bonds, maturing March 15, 2021. Coupon rate: 10.5%.

The federal government is paying, 5–10.5% on these bonds when interest rates are sub 1. Let’s consider the interest payments on just two of these bond series, relative to today’s 1%.

• CA135087XG49: Yearly interest payment: $689 million. Payments if coupon were 1%: $120 million. Difference: $569 million.
• CA135087XW98: Yearly interest payment: $586 million. Payments if coupon were 1%: $117 million. Difference: $469 million.

That’s an extra billion dollars, from just those two series, that the federal government will be paying this year, thanks to them locking in rates.

[Edited to add: It is also useful to see how much coupon interest the federal government will be paying on these bonds. As of November 30, 2020, the government will be paying approximately $13.1 billion in interest on these bonds, with $4.4 billion going to the Bank of Canada (assuming their current portfolio stays constant) and $8.7 billion going to non-Bank of Canada entities. Here’s how it breaks down by maturity date of those bonds:

Next year, the federal government will be paying out $3 billion to non-Bank of Canada entities on bonds that mature in 2033 or later. This represents 34% of all interest payments they’ll make to non-BoC bondholders, despite these long-bonds representing less than 18% of all federal government issued bonds. ]

“Locking in” interest rates is a form of insurance against rates going up. But like any form of insurance, it carries a premium, in governments having to pay billions more in interest than they would have if they had stuck to issuing short instead.

It’s good to be cautious, but we need to recognize that caution can come with a hefty price-tag.