In a recent article, "Financial Fundamentals Are Quite Healthy", Seeking Alpha contributor Calafia Beach Pundit cited a chart of two-year swap spreads and suggested this, among many other indicators CBP monitors, implied that the economy is on a solid footing. I rarely disagree with CBP's articles and, in fact, I agree with his basic premise. I have followed CBP for years, and his writing is a welcome breath of fresh air in an online atmosphere squawking about pending financial Armageddon.
An example is the constant Economic Cycle Research Institute's (ECRI) warning that a recession is imminent. Like Godot, it has yet to show up. Like Godot, its Broadway show should have ended a long time ago.
However, as a statistician, I am always on alert for misleading interpretations of data, and especially time series data presented in graphical form. The chart below is from Calafia's recent article:
The Pundit defends his use of two-year swap spreads as follows:
"Swap spreads have a record -- albeit not a very long record -- of anticipating changes in the health of the economy. They rose in advance of the past three recessions, and fell in advance of the past three recoveries...."
But this chart shows no such thing. To see why, you must look closely, and be consistent in your interpretation of the data. The shaded areas represent recessions. Therefore:
- Premise: the rise in swap spreads in late 1990 presaged the 1990/91 recession. The increase was 20 basis points.
Ok, in plain English: when swap spreads rise, recession follows.
But if this indicates an imminent recession, why didn't the 20 basis point surges in (count them... Here we go...) 1992, 1995, 1998, 1999, and 2000 predict imminent recessions? Some of these surges were even larger than their 1992 cousin. A 1-5 record is nothing to brag about.
Much the same can be said for the 2003-2006 period. The trend was upward, and there were several spikes (significant at the time... I see two). But there was no recession. So we drop to 1-7.
By the time the 2001/2002 recession arrived, for those of you who are fans of that greatest statistics excuse -- that there is a lag between data and events -- swap data had already been falling steadily for well over a year.
So the second premise of the article, that falling swap spreads precede a recovery, is also flawed. In fact, in both the early 1990s and 2001 recessions, spreads fell years before, during, and after the recession was over! Now we are 1-9!
Or perhaps we have discovered a miracle: an indicator that is so prescient that it forecasts recovery years before an intervening recession even takes place? I doubt it.
Let us accept for now that the huge spike in late 2007 was a timely forecast for the severe recession that followed less than a year later. Also, the rapid descent suggested a recovery would soon be underway. Before you jump for joy at the 3-9 record, I want to point out that the last two recent spikes haven't led to a recession. Yet? How much of a lag might you claim these things have? None, like in 1991? Three years, like 1998? One year, like 2007? For the moment, those last two spikes have not led to any recession. Nor does Pundit suggest they will, to his credit. Two more misses.
Maybe football fans in Kansas City might feel good about a 3-11 record in the near future. But such poor performance can be pretty damaging to your portfolio.
Disclosure: I am long XLV.