Are Treasuries Oversold Yet?

By in
Are Treasuries Oversold Yet?

Two “Data” items today:

#1: Are long-term US Treasuries oversold enough to consider buying? To answer that question we:

  • Pulled the price data for TLT (iShares +20-year Treasury ETF) back to 2005.
  • Calculated rolling 50-day (about 1 calendar quarter) returns.
  • Note: unlike similar analyses we’ve done recently for Emerging Markets or the Russell 2000, this is not a relative return calculation. It is also not meant to be a “should I own Treasuries to diversify a portfolio” analysis. We’re only looking at actual price returns to see if there is an absolute return investment opportunity here.

This chart shows the data:

Four points on this:

  • The average 50-day price return for the TLT ETF is +0.8 percent, which makes sense given that 2005 – 2021 has been a period of generally declining interest rates.
  • The standard deviation of those returns is 5.9 points, which means we should expect to see a band of returns that most commonly ranges between -11 percent and +13 percent. That is the 2 standard deviation band around the mean, and when you eyeball the chart those numbers line up with typical extremes in 50-day long-dated Treasury price volatility.
  • Long-dated Treasuries are more likely to become overbought (exhibit +2 standard deviation positive returns) than oversold (2-standard deviation negative returns). In fact, since 2005 TLT has shown a +3 standard deviation upside return (+19 percent) on 3 occasions: December 2008/January 2009 (Financial Crisis), September/October 2011 (Greek Debt Crisis), and March 2020 (Pandemic Crisis). There have been no 3-sigma downside moves.
  • Buying long-dated Treasuries/TLT when they/it are down more than 11 percent over the prior 50 days or shorting them when they are +13 percent over the same period is a historically proven money-making strategy. The only exception, as noted in the prior point, is when there is an existential market crisis and then one must wait for a +19 percent upside move.

As for whether the math signals that long-dated Treasuries are over-sold, the short answer is “absolutely not”. As of today, the trailing 50-day return on TLT is only -2.2 percent. This is not even 1 standard deviation from the mean (5 percent), let alone 2 sigmas.

The broader point this raises is that Treasury yields can continue to rise in the near term without that move being statistically unusual. Remember: the analysis here shows that speedy moves in Treasury prices (worse than -11 percent or better than +13 percent) mean-revert. As long as long-dated Treasury price volatility remains within those guardrails, there’s no statistical reason to expect they will suddenly shift direction.

Takeaway: all this fits with our expectation that Treasury yields will slowly increase over Q4 2021. As dramatic as the move over 1.5 percent on the 10-year may have seemed, there’s not much real volatility in this market just now. Certainly not enough to call for a change in underlying direction …

#2: What’s the latest on the Atlanta Fed’s GDPNow estimate for Q3?

The short answer is that this regional Fed’s algorithmic model is continuing to cut its expectations for US economic growth. Here is their graph with the trajectory of their estimates (in green) as compared to the Blue Chip economists consensus (in blue). The most recent GDPNow estimate is 3.2 percent.

As for what has caused the downward revisions since the model’s 5 percent readout in late August, the table below outlines the specific economic inputs that have taken almost 2 points of growth out of the estimate. We read this as an “all of the above” explanation – weak labor market growth, slack vehicle sales, and constrained housing markets.

Worth noting: GDPNow’s next release is tomorrow (Friday), so there will be a new data point to assess then. The trend is clearly not our friend, however.

Takeaway: GDPNow is flashing a warning sign about a slowing US economy, constrained by labor shortages and supply chain/inventory issues. Assuming a 3 percent growth rate for Q3 is accurate, there’s two ways to consider that number:

  • One is positive: since systematic constraints should ease in the next year, we should have a runway to slow but steady (and positive) growth well into 2022.
  • The other is negative: after 2 quarters of 6-7 percent GDP growth, Q3 is just 3 percent so what might Q4 2021 and Q1 2022 bring? Put another way, if the US economy grinds to a halt later this year, then hiring and investment might also slow considerably and only exacerbate the situation.

So far, capital markets are pricing in the optimistic case, which we still think is the more likely scenario. But the pessimistic case has some merit, especially if equity markets lose their footing as we start Q4 2021. Rising stock prices have done a lot to both support consumer spending through the wealth effect and buoy business/CEO confidence which helps hiring and investment. All this is a long way of saying we continue to expect more equity market volatility in the month ahead; the scales are too-finely balanced just now to think anything else.


Atlanta Fed GDPNow: