Wednesday, March 4, 2015

Updating two mid-cycle indicators

 - by New Deal democrat

Ultimately I would like to have a set of indicators to describe an entire business cycle: long leading, short leading, coincident, short lagging, long lagging, and mid-cycle.  A mid-cycle indicator is one that turns roughly in the middle of an economic expansion.  It is a way to focus attention.  Are we in the first part of an expansion, or are we at a point where we should begin to watch for a turn down in long leading indicators?

In this post I want to update 2 series which appear to make good mid-cycle indicators: retail sales vs. personal consumption expenditures, and the personal savings rate minus the inflation rate.

First, three years ago, I identified a consistent pattern whereby retail sales grew faster than the broader category of personal consumption expenditures early in an expansion, but slower later in an expansion.  Retail sales constitute about 50% of PCE's.  Note, however, that real retail sales are much more volatile. And, as this graph below (subtracting YoY PCE growth from YoY real retail sales growth through 1997) shows, in a very specific and non-random way:

Retail sales minus PCE's are always negative before the economy ever tips into recession. That's 11 of 11 times. Further, in 10 of those 11 times (1957 being the noteworthy exception), the number was not just negative, but was continuing to decline for a significant period before we tipped into recession.

Essentially these graphs tell us that, in the later part of a business cycle, consumers cut back on discretionary purchases to preserve other spending. Until they do, consumer spending does not support any claim that a recession has begun or is even imminent.

I updated this graph a few times last year, noting that it looked like it was turning.

So, what does it show now?  Here is a 25 year history of the relationship, measured YoY:

Here is a close-up since the end of the last recession:

Two trends are apparent.  The first is that PCE growth has generally been approaching,equal to, or slightly greater than real retail sales growth for the past five quarters.  The second is that both are increasing, just as they did in the second half of the 1990s. This supports the notion that the midpoint has been reached, and probably passed, but that a downturn in nowhere near.

The second mid-cycle indicator I want to update is the real personal savings rate. This is, essentially, a measure of economic confidence. How much of their paychecks do consumers feel they need to save over and above the rate of inflation? Here's the answer, going back over 50 years:

Note that in every single economic expansion except the 1980-81 double-dip, at some point from about the middle to 3/4 mark, there is a steep decline in the real personal savings rate from its peak of about 5%.  Further note that in every single recession, the real personal savings rate increases as consumers seek to buttress their balance sheets.  Generally speaking, as an economic expansion goes on, consumers expose themselves to too much risk, either due to overconfidence, or the need to stretch their finances to keep up.

I do not think we have seen the relevant 5% decline yet, even though it has happened twice during this expansion.  The December 2012 - January 13 decline is an artifact of income shifting due to the fiscal cliff.  The early 2010 decline is too early, and is the result of the big run-up in gas prices after the end of the Great Recession (inflation went up, so the real savings rate went down).  

Similarly, I am not concerned by the recent rise in this metric, which is also an artifact of the plunge in gas prices. To show you why, let's compare our current situation with the 1986 and 2006 steep declines in the price of oil and gasoline (red in the graphs below).  First, here's 1986:

and here's 2006:

In each case, the initial reaction by consumers was to pocket the savings (for 12 months in 1986, for 3 months and declining thereafter in 2006). That appears to be what has happened in the last few months as well.  

To summarize, one mid-cycle indicator looks like it has been reached, the other has not.