Update 8/14/2019: The 10/2 spread inverted now, very slightly. Am I worried now? Certainly more worried than in April when I wrote this piece. The ISM-PMI index is around 51 – that’s also weaker but not weak enough to worry (everything above 50 is still called expansionary). Unemployment claims are still very low, which is a good sign. So, this is still “only” a mixed bag. Consistent with a false alarm a la 1998. But the probability of worse things to come has certainly gone up!
Well, there you have it: The Yield Curve inverted last month. Finally! Starting on March 22 and throughout much of last week, short-term interest rates (e.g., the 3 months bills) yielded slightly more than the bond market bellwether, the 10-year Treasury bond.
People in finance and economics view this with some concern because history has told us that an inverted yield curve is a pretty reliable recession indicator. And I made this point in my post in February 2018: The yield curve shape, especially the slope between longer-term yields (10 years) and the short end (e.g., 2-year yields) is one of my three favorite macro indicators:
Also notice that I usually look at the 10-year vs. 2-year yield rather than 3-month spread and that made a bit of a difference recently, more on a little bit that later. But in any case, since I went on the record about the importance of the yield curve and now got several reader requests to comment on this issue, here’s an update: in a nutshell, I’m not yet worried and here are eight reasons why…
1: The 10-2 yield spread never inverted
Keep in mind that in the post last year I had used the spread between the 10-year and 2-year Treasury yields. That spread never went below zero (at least up until 4/2/2019 when finalizing this post). Am I just cherry-picking here? Imagine back in 2018 I had proposed using the 10y-3m spread (the one that recently inverted) and now I said: “oh, don’t pay attention to what I said last year, because the 10-2 spread is still positive!” Well, that would look like shifting the goalpost. But I’m consistent here: the 10-2 is what I proposed back then and that’s what I still look at. In the past, the 10-2 had been a more timely and consistent indicator than the 10y-3m spread and, therefore, I’ve always preferred the 10-2 version of the yield spread. There are a few more reasons that would all justify their own separate blog post – remember I used to do economics and finance for a living – but I don’t want to bore you with too much econ-babble. 🙂 All I want to stress here is that despite the big hubbub in the media last month, I’m not that worried until the 10/2 spread inverts!
2: The Yield Curve “un-inverted” again this week
Well, the whole episode didn’t last too long. The 10-year yield dipped below the 3-month yield for a few days between March 22 to March 28 but jumped up again after that. What do I make of that? Again, together with the fact that the 10/2 spread didn’t even invert, I’d not consider the event last week a full-blown significant macro signal.
Of course, critics would argue that in most previous events, the yield curve “un-inverted” again before the start of the recession, see the time-series chart above in the intro. To which I’d respond that not just was the duration of the yield curve inversion quite short but also the intensity was much lower than in the previous events, which brings us to the next point:
3: The 10y-3m and 10y-1y yield spreads only barely dipped below zero
Not only did the two spreads revert again but they never even fell far below the zero line. That’s in stark contrast to prior events, see the table below. In every prior yield curve event that preceded a recession, the short-end yield not just touched and slightly passed the 10-year yield but the magnitude of the yield curve inversion was much more significant than in 2019!
Well, the one occasion where we saw only a very shallow inversion was in 1998. But do you notice something? That was the one yield curve inversion that wasn’t immediately followed by a recession. Which brings us to the next point…
4: There’s been one “false alarm” that looks strikingly similar to the March 2019 Yield Curve inversion!
Every recession in recent history was preceded by a yield curve inversion. But in and of itself, that’s not all that impressive. Every recession in the past has been preceded by a Wednesday. That doesn’t mean that every Wednesday we’ll have a recession. A good macro indicator should (almost) never miss an actual event (type 2 error) but it shouldn’t have too many false alarms (false positive, type 1 errors) either; think of Paul Samuelson’s famous quote “The Stock Market Has Predicted Nine Of The Past Five Recessions.”
The yield curve signal did produce one false alarm in 1998. In other words, the curve inverted back then but that was way too early! There wasn’t a recession for about 3 years after the 1998 event. Rather, the yield curve normalized again before it eventually inverted in early 2000 (and with a vengeance, see the chart below!), about a year before the March 2001 business cycle peak.
Actually, the 1998 event is a bit reminiscent of the one in March this year: A very short and shallow yield curve inversion. And in 2019, the 10y-2y slope didn’t even invert! We can’t know for sure how the future will turn out. Maybe we will face the first big recession with only a very mild curve inversion. But I wouldn’t assign a very high probability to that!
5: Even if this yield curve inversion signals an upcoming recession it could still be years(!!!) away
There isn’t too much comfort in this one but I should still point out the obvious. In the past, the yield curve inversion would sometimes occur way before the start of a recession, see the table below. About nine to ten months at the minimum and most often more than a year in advance (not even mentioning the 1,040 days or almost three years of the 1998 false alarm).
Well, maybe the performance of the economy is not on your mind all the time. But I’m sure the stock market is, right?! Which brings me to the next issue…
6: In past instances of yield curve inversions, equities didn’t even perform that badly in the near-term
This is obviously related to the point above, but it’s important enough stress it again. Not only is the recession potentially a long time away, but so might the equity market peak! In other words, even if you take the yield curve inversion seriously, it’s not exactly an equity-sell signal. There have been instances in the past where the inversion occurred when the bull market was still very much intact, with months and even years, to run. Below is a table with a chart with the S&P500 index performance (nominal dollars, dividends reinvested) after the 10y-3m spread went negative. There was only one precedent where you lost money over the next year! Not exactly an urgent sell signal!
Thus, even if we take this yield curve inversion really seriously (and I still have some serious doubts!), right now I’d be on alert rather than panicked. This is not the time to run for the hills and sell stocks!
7: The Federal Reserve isn’t as aggressive as in the previous tightening cycles
A lot of the previous recessions coincided with the Federal Reserve raising short-term interest rates aggressively. It always goes like this: The Fed tries to fight inflation (perceived or actual) by raising interest rates to cool down the economy. The short end of the yield curve goes up but in expectation of the inevitable economic slowdown, longer-term interest rates already price in the prospect of a future recession and lower rates. The Fed will eventually go overboard with rate hikes and sink the economy. And right before that the yield curve inverted.
The current tightening cycle seems very different, though. Rate hikes started from a very low level, the rate hikes were very gradual and the final resting point (a 2.25%-2.50% range as of today) is much lower than in prior events. In fact, today’s (tentative) endpoint of the rate hikes is lower the lowest rate during the entire 1990s, which stood at 3.00%!
8: Other macro indicators still look quite solid
In the original post last year I mentioned that the yield curve signal is only one single indicator. Nobody in their right mind would propose forecasting an economic slowdown with just one single indicator. In my professional life before FIRE, I used to monitor hundreds, even close to a thousand global economic and financial indicators. No worries, I found a way to automate this and let the computer do all the work for me to sort out the noise vs. signal, thanks to some innovative methods I developed. But you know what, the Pareto Principle works too: 80% of the outcome is due to 20% of the effort. In this case, it may even be that 0.3% of the effort, i.e., checking 3 indicators out a thousand gives you almost everything you need to know.
In any case, the other two macro indicators I like to follow all look really good until now. Let’s take a look…
8.1: ISM-PMI Index
The ISM-PMI indicator stood at 55.3 as of 3/31/2019 (data released on 4/1/2019). That’s still a very solid reading: 50+ means expansion and even a tempdrop below 50 is common during economic expansions. I’d be on alert (not yet panicked) if the PMI drops below 50 and I’d be pretty panicked only if the PMI drops below 45. We’re still far away from that!
8.2: Unemployment Claims
Nothing new here either. There was a small bounce in late 2018/early 2019, likely related to the government shutdown but we’d need to see a much larger move. Remember, unemployment claims are still very close to multi-decade lows! This doesn’t look like a bona fide macroeconomic slowdown! Quite the opposite, the labor market is still rocking!
So, strictly speaking, out of my 3 indicators, none of them indicate a slowdown yet if I use the 10/2 yield spread rather than the 10y-3m spread! Nothing to see here folks, let’s move on, there doesn’t seem to be a recession around the corner!