Monday, March 5, 2012

Can you really have a recession if houses and cars (and stocks and bonds and money supply and a bunch of other stuff ) won't play?

- by New Deal democrat

A recession in economic terms isn't synonymous with a period of hard times. Rather, it is a contraction in the economy. Can one really happen if leading sectors don't decline, and with no warning from the usual indicators?

Take housing. Here is a graph of housing permits, an acknowledged long leading indicator, for the last half a century:



Not once during that time has housing failed to turn down in advance of a recession. The weakest decline was from 1.742 million units annualized in December 1998 to 1.542 million annualized in July 2000, a decline of 200,000. The closest comparison in our situation is the decline of 154,000 from June 2010's 688,000 annualized permits to February 2011's 534,000. Since then permits have risen back to 682,000 in January of this year.

There is simply no post-WW2 for a recession happening without housing construction declining first. The only possible precedent is the slight increase in houses built in 1938 vs. 1937, but the statistics are only annual so we have no way of knowing what kind of quarterly or monthly declines may have occurred.

Records of vehicles sold have not been published on that long a basis, but the records we do have suggest that vehicle sales too have almost always declined in advance of a recession:



The only exception here is the Volcker induced recession due to the Fed hiking interest rates to 20% to break an inflationary cycle.

Another statistic getting a lot of attention recently is initial jobless claims. These have continued to fall to new lows:



Never since these statistics began to be kept almost 50 years ago has a recession occurred without initial claims turning up first. The minimum period of time from the bottom to the onset of recession was 2 1/2 months.

Another long leading indicator is bond prices. Since WW2 there has never been a recession without bond prices declining (i.e., bond yields increasing). Here is a graph showing inverted yields of BAA corporate bonds (i.e., prices):



While the above are weekly, monthly BAA bond prices go all the way back to 1919. Here is the graph of monthly prices from then until the 1960's"



Only twice did BAA bond prices not decrease before the onset of recession: in 1927 and in the 1945 demobilization.

Real money supply is also generally thought to be a long leading indicator. Again, since modern records have been published, at no time has a recession occurred without Real M1 turning negative first:



During the deflationary 1920's and Great Depression era, negative real money supply was at very least coincident with the onset of recession.

Stock prices are also at least a short leading indicator. While famously stocks continued to rise for three months after the economic downturn began in 1929, in our era their record continues to be good if not perfect:



Stocks only failed to turn down in advance of the 1980 and 1990 geopolitical Oil shocks.

Average hours worked in manufacturing has long been considered a leading indicator as well, and these records are available since WW2.



Again, there has only been one exception, in this case 2007 right before the "Great Recession."

I have found only two leading indicators that might support the case for an imminent contraction. Nondefense durable goods orders ex-transportation show clear signs of rolling over:



Further, if one believes that Real M2 can signal a recession even after more than a year of improvement, then the precedent of 2007 exists, although just like the 1920s and 1930s, its turning negative was coincident with the onset of the December 2007 "Great Recession."



The increase in Oil prices now, and the relative weakness of wages, are not unprecedented. Yet several of the above leading indicators admit of no exceptions; others only one or two. For a recession to have already started, or even to start now or in a month or two, would represent an unprecedented constellation of exceptions to a host of indicators that typically lead directional changes in output, sales, and jobs by several months to over a year. This is why I am proverbially scratching my head at ECRI's certitude that a recession will begin by the end of June, and their defense of that call by use of a lagging construction of coincident indicators.