Don't Look Now, But The Yield Curve Is Flattening Again

by: Trading Places Research

Summary

The first signs of bearishness in the early summer came with the flattening yield curve.

As summer turned to fall, the curve steepened and spreads widened, leading everyone to refocus on rising long term rates and trade concerns.

But in the last couple of weeks, the long end has been dropping again, leading to a flattening of the curve. This continues a 2018 trend.

The action on spreads has been primarily driven by the long end, the rates least under the control of the Fed. This is troubling.

Our regression model currently predicts that the 10-year minus 2-year yield spread will flatten in the early spring 2019 and invert from there. The Fed's yield-curve-based probability model currently predicts about a 15% chance of going into recession near-term.

Fear The Yield Curve

If you watch CNBC or Bloomberg or read the financial press, this summer you got an earful about the yield curve. Under normal circumstances, the long end of the curve is higher than the short end. But as we head into a recession, investors discount the future interest rates of the long end, leading to an inverted yield curve, where the long end is lower than the short end. An inverted yield curve has preceded the beginning of the last six recessions by 6-24 months.

But these fears and discussions receded in September/October with the Fed rate hike and the 10-year Treasury spiking to 3.23% on October 5. The fear trade became about trade, long term rates rising, and the Fed over-tightening. But while the talking heads of the business networks continued to focus on the long end rising too quickly, something funny happened: the long end began to drop again and spreads dropped, continuing the 2018 trend.

2018 Is The Year Of The Yield Curve

Since the beginning of the year, I have been charting Treasury rates every day at close, watching for the trends in rates and spreads. I chart the progression of two spreads: the 10-year minus 2-year and the 10-year minus 6-month. The 10-minus-2 is the most common, but Jerome Powell has indicated the Fed spends a lot of time also looking at the 10-minus-6-month, so I look at that too.

First let's look at the overall trends is the spreads YTD.

Source: Chart mine. Data from US Department of Treasury

As you can see, even though there's been been a lot of daily volatility here, the R2 of the linear regressions is pretty high, indicating a good fit to the line. Overall they continue to predict a slow downward trend in both spreads (about 0.20 bps per day as of Friday). Let's look at the progression of this predicted growth trend.

Source: Chart mine. Data from US Department of Treasury

The curves here are the change in the predicted rate of change of yield spreads. This trend has been negative since April, and since the early summer, the model has predicted roughly in the range of 0.15 - 0.30 bps per day decline. The more data we add, the more accurate the model will get and it seems to be settling in around -0.20 bps per day for both spreads.

Eyeballing The Curve

Another way to look at it is to chart the curve itself, eyeball the shape and slope, and also measure the trendline which will give us an equation for the slope. Here's the 6-month through 10-year:

Source: Chart mine. Data from US Department of Treasury

The thick grey line is Friday's close. The green line is the steepest day of the year (2/9/18) and the red is the flattest day (8/20/18). Just eyeballing, we can see that the slope of Friday's curve sits between the red and green extremes, but that its slope is much closer to the flat red slope than the steep green slope.

Another way of looking at it: look how much the 6-month has moved since February, and how little the 10-year has.

Looking at the regression equations, the important number is the X coefficient, which tells us the slope of the fitted line. A positive number is normal, zero is flat and negative is inverted. The numbers confirm our eyeball test, and also that in the 6-month through 10-year range we are much closer to our flattest day than our steepest day.

Another result of the eyeball test is the "elbow" that has developed at the 2-year note. The slope of the 6-month through 2-year curve stays pretty stable, and the 2 through 10 year seems to be where all the movement is.

So lets look at the the 2-10 year curves by themselves.

Source: Chart mine. Data from US Department of Treasury

That's a super tight fit on the regression lines. As you can see, the X coefficient (slope) of the grey line from Friday is even closer to the flattest day in 2018 than in the full 6-month through 10-year.

That's A Lot of Fancy Words There; Break It Down

The main conclusion here is that the yield curve continues to flatten and that the action in yield curve spreads is happening primarily at the long end, which are the rates least in the Fed's control. Despite the panic you may see on CNBC at times, the long end is not responding to Fed policy nearly as much as the short end is.

The market for US debt is having a hard time deciding the future value of long-term interest rates (though generally trending downward in relation to the short end) and which we can see in the unusually high volatility in spreads lately. This is the interest rate equivalent of the currently high VIX index.

Contrary to popular belief, the Fed does not set these rates -- they only set the Fed Funds rate, which becomes the baseline very-short-term rate. All the rates discussed here are determined through market auctions and daily trading in Treasuries and there are many variables that go into the price that have nothing to do with the Fed. The farther out the curve you go, the less control the Fed has. Generally speaking, it's worrisome that the narrowing-spread action is where the Fed has the least control.

Model Predictions

Currently, our regression model predicts the 10 minus 2 year yield curve will flatten completely in 121 trading days, about 5.5 months (early spring) and the 10-year minus 6-month will flatten in 292 trading days, 13 months from now (early winter 2019-20). The 2-10 inverting is a 6-24 month warning to the onset of recession, putting it around Q4 2019 at the earliest, and Q2 2021 at the latest, which is in line with the range of current economic forecasts.

These charts show the progression of the the predicted date of flattening (I used an estimate of 22-trading-day months, which means the dates are a little rough). As data is added to the model, the predictions become better, and we can see that the volatility in the predictions is beginning to settle down. Right now, I'm pretty confident in the current prediction, but by year's end, the model should be much stronger.

Source: Chart mine. Data from US Department of Treasury

I began the chart in June to get rid of the early volatility in the model's predictions as it added data. As you can see, the date has been trending forward since the summer, but is in-line with early June predictions.

Fed Says What?

Back in March, the Fed released a paper that discussed the yield curve issue. They set up a 1-variable model to predict the current probability of entering recession based solely on the yield curve (data is only through Feb, 2018). Their conclusion is that at flat, there is a 24% chance of recession in the near-term horizon, rising sharply with the inverted curve (the blue line is the important part here; you can ignore the other lines which are add-on models). As you can see, at about 30-35% probability is where the recession typically begins in the last three cycles.

Source: San Francisco Fed

Their model predicted an 11% chance of recession back at the end of February. I don't have access to their model, but just to hazard a conservative guess based on the previous cycles in the graph, I think that number is 15% or higher now.

Yield Curve Bullets

Too long? Didn't read? Here's some handy bullets:

  • The trend in yield curve spreads continue to be negative, even as the rates rise.
  • The changes in spreads are being driven by the long end, the rates least in the Fed's control.
  • Models based around the yield curve are starting to predict recessions coming sooner rather than later.

I will try and update this article monthly with new data. Thanks for reading. Comments/insults welcome.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.