The rally in US/global equities since the March 23rd lows has been impressive, but consider for a moment that the long dated (+20 year) US Treasury bond is also marginally higher since then. A few numbers to frame the discussion:
- The S&P 500 is +26.2% from March 23rd
- The TLT ETF (+20 year Treasuries) is +1.9% since March 23rd
- And, of course, over a long timeframe TLT has it over SPY:
YTD: +24.9% vs. -12.6%
1-year: +37.6% vs. -2.82%
Back on March 10th we looked at rolling 30-day returns on TLT and came to the conclusion that the long end of the curve was pretty tapped out, and unless one was hedging a new equity position it was best to avoid the longer dated Treasuries. That call worked out reasonably well: TLT closed on 3/10 at $162.51 and meandered for the next month at or below those levels. But… it has started to rally again in the last week as long rates have declined.
Today we will expand the conversation and look at 90-day rolling returns to capture what history has to say about where Treasuries may go over the next 4-5 calendar months (we use trading days in our work). A few points here and an accompanying graph above:
- Since 2002, the average 90-day price return on TLT has been 1.4% in terms of price movement (we’re excluding coupon payments and focusing on price return only).
- The standard deviation of that mean is 7.5 points, which means any 90-day return between -21.1% and +23.9% is solidly in the meaty part of the curve.
- The graph above supports that math: there are only six instances where 90-day returns touched/went through +20% and only one -20% occurrence.
As of today’s close, we’re at a 22.0% 90-day return, which given the math we have just presented is pretty remarkable:
- It is very close indeed to a 3-standard deviation return.
- The only notably higher readings this year was for the 90 days ending March 9th (24.8%) and March 23rd (23.0%), the day of the lows for equities.
- You have to go back to December 2011 to find higher trailing 90-day Treasury bond returns than we’ve just experienced in March/April 2020.
Here’s what has happened when we’ve been in a similar spot before:
History shows 4 instances since 2002 when Treasuries got close to/exceeded a 20% 90-day return for just a brief period. Here are those dates and forward 30- and 90- day price returns
- August 26, 2010: 30-day forward return: -3.9%. 90-day return: -13.7%
- July 26, 2010: 30-day forward return: -5.3%. 90-day return: -4.2%
- January 28, 2015: 30-day forward return: -7.1%. 90-day return: -14.1%
- August 28, 2019: 30-day forward return: -3.8%. 90-day return: -6.9%
- Averages: 30-day forward return of -4.0%, 90-day return of -9.7%
History also has 2 examples when 90-day Treasury returns blasted through 20% rather than stopping there:
- December 2008: on 12/1, trailing 90-day returns hit 20.2% and bonds continued to rally another 11.8% through 12/19. Trailing 90-day returns peaked on that day at 33.4%.
- August – November 2011: after first hitting a 20.1% trailing return on 8/17, long dated Treasuries rallied another 8.3% over the next 30 days. 90 days later after the first +20% day, however, Treasuries were 20.0% lower.
The message from this: to hold long dated Treasuries here you have to be absolutely certain the COVID-19 Crisis is going to have another wave of impacts on the US/global economy and investor confidence beyond what we’ve seen to date. In that environment we could, as in December 2008 (Financial Crisis) or late 2011 (Greek Debt Crisis) see further gains. But history says you need that sort of damage to get a follow-on move from here.
One out of the box thought to wrap up: the New York Fed recently published a blog post titled “How the Fed Managed the Treasury Yield Curve in the 1940s”. During World War II the US central bank played a critical role in managing both long- and short-term rates as the government issued large amounts of debt to fight the Axis powers. By 1942 rates were fixed at 0.375% on the short end and 2.5% on the long end. Could the Fed do something like this again, essentially manufacturing a positively sloped curve with a +1.0% long end? For holders of long dated Treasuries, it is a question worth considering. These are unusual times….