Posts Tagged ‘yield spread’
Yield Curve Inverting
There has been a good bit said lately about the yield curve inverting. Historically, so they say, an inverted yield curve forecasts a recession. I decided to explore that concept a bit. Like all generalizations, this may have a grain of truth in it, but there is more than meets the eye.
By the way, this topic has been explored in greater depth, and with more granular data, by economists who actually focus on this topic. (Just as a reminder, my area is Real Estate Securities). If you are reading this with an eye to fleshing out some masters thesis somewhere, I’d suggest you keep looking for authorities. That said, the FED leadership is meeting in Jackson Hole this weekend, and you can bet this is on the agenda.
Speaking of the FED, I grabbed a bit of data from them — monthly 10 year treasuries and 6 month bills back to December, 1958. I actually explored some other time periods and even daily data, but this was the best pairing I could get in short order. Anyway, the “shape” of the yield curve is essentially the gap between these longer term rates and the shorter term ones. For a normal yield curve, the long bonds (10 year) should be about 2 to 4 percentage points above the short term yields. That makes some intuitive sense — in “good times”, borrowers are willing to pay more to borrow longer term, and investors are willing to accept less return for shorter term loans. When borrowers sense that there may be trouble ahead, they are less willing to borrow long-term, and hence the demand for long term money falls relative to short-term stuff. When things go really topsy-turvy, the short-term money is actually more expensive than the longer term, because borrowers simply don’t want to borrow long-term at all. The topic is w-a-a-a-a-y more complex than this, but hopefully you get the picture.
Speaking of pictures, I then took the difference between these two yields and graphed it. Along with that, I graphed the incidence of all of the recessions since 1958. Here’s what I got:
Data courtesy Federal Reserve, graphic (c) Greenfield Advisors, Inc.
Not EVERY inversion was followed immediately by a recession. although almost all were. The only exception was in 1966. That one is generally considered a “false positive” because it was triggered not by general economic trends but by a short-term reduction in Federal spending.
More interestingly, though, is the long-term bull market of the 1980’s, which was followed by the recession of 1991. That long-term market followed the double-dip recessions of 1980 and 81, which are often considered one long recession. (I know — I was there.) More to the point, the recession of 1991 was not following a yield curve inversion. Indeed, the yield curve spread, measured on a monthly basis, never got below 0.38%, in November, 1989. Also, notably, the behavior of the yield curve over the course of the 1980’s mimics what we’ve been seeing of late.
There is also an argument that rates are behaving more like the 1992-2000 period, and there is some rationale for that. If that’s the case, then the question is whether the yield curve recovers from here or takes a swan dive below zero. If the former, then we may have 2 years or so of continued positive GDP. If the latter, then we’re headed for a rough patch.
The yield curve spread ended July at 0.78%, and closed yesterday down at 0.59%. Again, this is the data the FED leadership is discussing in Jackson Hole. We’ll keep you posted.
Share this:
Predicting recessions
The onset of a recession is much like the onset of a bad cold, or perhaps the flu. You THINK you know you have it, and then the next morning you wake up feeling like you’ve been hit by a truck.
What if we had a tool that could actually PREDICT a recession a year or so in the future? Wouldn’t that be handy? In fact, two researchers at the New York Federal Reserve Bank actually developed something a few years ago (early 2006, to be precise) that seems to have some promise. They noted that the spread between 10-year Treasuries and 3-month Treasuries was a leading indicators of economic activity (which makes intuitive sense). Using historical data, they craft a formula which calculates the probability of a recession occurring in the coming 12 months:
…where “spread” is the difference, in percentage points, between the 10-year yield and the 3-month yield and F is the standard normal distribution (mean of 0, standard deviation of 1). Plotting their data over time, and comparing to the onset of a recession, they get Chart 2. While no model is 100% predictive, this one clearly indicates that a recession is highly probable whenever the spread-indicated probability gets much above 30%.
Of course, the statistical math can get a bit hairy, but there is a handy rule-of-thumb. As you can see from Chart 1, whenever the spread turns negative, a recession is fairly likely in the coming months. Why? Because a negative spread suggests two things: first, that borrowers have little demand for long-term money, and second that investors are looking for a safe place to tuck money away that they don’t think they’ll need for a while, and aren’t afraid of inflation. In short, the yield spread constitutes a fairly accurate survey of investor expectations about the economy.
Unfortunately, Estrella and Trubin end their research with July, 2006, which at the time was giving off a 27% recession signal for July, 2007. A great test of their model would be to see if it would have predicted the onset of the December, 2007, through March, 2009, recession. We used the same H.15 data from the Federal Reserve Board, and came up with the following:
Ironically, our analysis shows that the probability of a recession crossed the 30% barrier on July 31, 2006 — a month after the cut-off of their study, and 16 months before the official “onset” of the recession.
By the way, if you’re worried, the current probability is at 1.8%. We’ll keep you posted.
From a small northwestern observatory... 