25 Feb 2012 by Jim Fickett.
ECRI is standing behind their prediction that a new recession will begin in the US in the next few months. They have revealed a bit more about the data supporting their prediction. When I look at the data, I see a very mixed picture, and I doubt that ECRI really has any special magic that the rest of us lack (or, in fact, that anyone can reliably predict recessions).
In a recent CNNMoney interview Lakshman Achuthan of ECRI revealed some specifics behind their prediction of an upcoming recession:
Specifically, he identifies annual growth in industrial production, real personal income and spending, as well as the year-over-year change in gross domestic product, a broad measure of the nation's economic activity. That GDP reading has been stuck between 1.5% and 1.6% growth for the last three quarters, far less encouraging that the rising quarterly GDP, which is more widely reported. …
He said the so-called velocity of money, the number of times a dollar is spent by consumers, has fallen to a record low in recent months, which he sees as another indicator of underlying weakness.
“The world central banks are printing money as never before in history,” he said. “The money is not boosting economic growth, but [it's] still there and it goes into risk assets.”
So the four indicators mentioned are:
Let's look at these in turn.
Here is an updated graph on the year-over-year change in industrial production:
The year-over-year change remains very close to the long term average; nothing special there. And dips in the year-over-year change to values below average have often occurred in the past without any recession ensuing. It is difficult to see how the industrial production data provide any evidence for a near-term recession.
It is a similar situation with personal income and spending. Recent data don't look fantastic, but they don't look particularly bad, either. The most important measure, in my opinion, is organic income – household income less government transfers. This stalled for several months last year, but then started growing again:
Household spending was flat at the very end of last year, but it is unclear whether this is just month-to-month noise; the 2 year trend is normal growth:
If something about household income and spending indicates a recession, it is certainly not obvious.
GDI is a better indicator of the economic cycle than GDP, and in this area a recession prediction does gain some support. As I mentioned in the post, The latest reading on gross domestic income supports recession predictions, Q3 GDI figures were at least compatible with a recession outlook, though still quite ambiguous.
Many indicators last year were looking pretty bad in Q3, and then improved in Q4. We should have an estimate for Q4 GDI next week, which will help to clarify.
The velocity of money is usually measured as GNP divided by the money supply. Of course there are many different ways to measure the money supply, and it is not reported what measure Achuthan has in mind. For the argument implied by the above quote, namely that people are sitting on money rather than spending it, M2 would be an appropriate measure – this includes cash and most highly liquid accounts like savings, checking, CDs, and money market. The velocity of M2 money is indeed at an unusually low level:
(Graph courtesy of the Federal Reserve Bank of St Louis)
However it is unclear what to make of this indicator. Some recessions in the past began when the velocity of M2 was low and rising, some when it was steady, and some when it was high and falling. Where is the evidence that the velocity of M2 can help to predict recessions?
All in all, this revelation of ECRI's thinking increases my (already robust) skepticism that they have any special evidence.
Going a little further, I am with James Hamilton in thinking that, in fact, predicting recessions is not possible. This is not surprising because recessions are, in effect, events of mass psychology, and no one can predict the actions of the crowd. The best we can do is to keep on eye on indicators that tell us when the danger is higher than usual, and perhaps detect when a recession has already started.