The Full Story:
Thank you to those of our readers who made it to the end of last week’s market missive. I know who you are because I received multiple references to my obvious frustration with Government dysfunction, debt denial, and surging Treasury yields. After typing so aggressively last week, I have regained composure this week (to my keyboard’s delight), having devoted significant thought to the question of where yields will peak. Clearly, this is an inexact science in the short run, given the vagaries of the marketplace and manic geopolitical factors, but longer-term equilibrium forces provide gravity in economics, and US Treasury yields are not massless.
Calculating 10-Year Treasury Yields
When investors purchase 10-year Treasury bonds, they consciously or subconsciously forecast three macroeconomic variables. First, they must forecast an expected GDP growth rate for the US economy over the next ten years. Second, they must forecast an average inflation rate for the US economy over the next ten years. Third, they must forecast the probability that the US Treasury will default on its obligations. Historically, investors haven’t required a yield premium for default given the perfect payment record of the US Treasury. Holing the odds of default near zero, three variables become two: GDP growth and inflation. Fortunately, the Government provides us with a measure that combines the two variables known as Nominal GDP. To test our theory, nominal GDP growth has averaged 6% since 1960. The 10-year Treasury has also averaged 6% since 1960. Correlation proven. But let’s dig into this relationship a little deeper for assurance:
Note that the near-perfect long-term relationship between Treasury yields and Nominal GDP weakened over the past 20 years. What could have caused that distortion?
Over the past 20 years, the US Federal Reserve bought $8 trillion worth of bonds to intentionally suppress longer-term interest rates… to intentionally boost nominal GDP. This explains the gap between nominal GDP and Treasury rates. It was by design! That effort has now reversed, with the Fed selling $1 trillion worth of bonds over the past year to intentionally boost longer-term interest rates… to intentionally suppress nominal GDP. It is, therefore, no coincidence that long-term interest rates have doubled since the Fed’s bond sales began. It’s by design! For this reason, one could argue that, unlike the last 20 years when nominal GDP well exceeded the 10-year Treasury yield, the inverse could now occur where 10-year Treasury yields well exceed Nominal GDP. Maybe. But it’s highly unlikely that the Fed will continue to sell bonds for the next 20 years. Most observers expect the Fed to continue shrinking its balance sheet for another year or two, totaling between $1 trillion to $2 trillion in sales. With that being the case, the more traditional parity relationship between nominal GDP and the 10-year Treasury bond should reappear. Therefore, to correctly forecast long-term interest rates, we must properly forecast nominal GDP.
Forecasting longer-term economic growth isn’t actually that difficult. Economic growth depends on labor force growth and labor productivity growth. Labor force growth is easy to forecast as tomorrow’s workforce has already been born. Productivity forecasts provide a greater challenge but tend to average between 1-2% depending on innovation cycles. Based on the CBO’s estimate, here is what we can expect from inflation-adjusted GDP growth over the next ten-plus years:
The CBO’s 1.8% GDP growth estimate over the next decade could prove low if immigration levels rise and/or technological advances (such as AI) boost productivity. As an optimist and a believer in the power of AI, I will adjust this figure up to 2.2% for our purposes.
To finish our nominal GDP forecast we must properly estimate inflation. Here is a long-term look at CPI inflation trends. Note the historic spikes in inflation correlate with the fiscal and monetary policy mistakes of the 1970s and the unprecedented stimulus measures deployed during COVID:
Removing the inflation of the 1970’s and COVID era, long-term consumer price inflation since 1965 averages about 2.5%. The modern US Federal Reserve has adopted a 2% inflation target that it rebroadcasts every chance it gets. Based on their own forecast, inflation should return to 2% by 2026. According to the futures markets, inflation will run slightly higher, averaging its more traditional 2.5% over the next ten years. We tend to agree more with the markets than govies and will therefore add 2.5% inflation to our 2.2% GDP figure, deriving a nominal GDP estimate of 4.7% over the next ten years.
Using 4.7% as our baseline for nominal GDP growth allows us to baseline our 10-year treasury yield expectation at 4.7% as well. For those who want to raise that yield estimate for Fed bond sales over the next couple of years, tack on another .5%. This takes you to 5.2%. For those who want to raise the yield premium further for the potential of a ratings downgrade or Fed default over the next couple of years, tack on another .25%. This takes you to 5.5%. Barring some market catastrophe, 5.5% feels like a sensible upward bound. However, this 10-year forward analysis doesn’t help us forecast rates over the next quarter… but history offers near-term comfort here as well.
Last quarter (Q2), nominal GDP grew 3.8%. The 10-Year Treasury exited the quarter at 3.8% as well. The 10-Year Treasury currently yields 5%, suggesting a 5% nominal GDP growth rate for Q3. This is certainly possible as economic momentum has increased, but it’s highly unlikely to maintain that pace for subsequent quarters and represents a clear counter-trend:
The recent record on the path for interest rates after quarters where they exceeded nominal GDP growth rates is consistent. Either GDP rates rise (unlikely) or interest rates fall (likely). Therefore, as nominal GDP stabilizes under 5%, we expect interest rates will stabilize there as well.
Have a great week!
David S. Waddell
CEO, Chief Investment Strategist