The Observed Economy
I'm an amateur economist with fairly extensive formal training as a social scientist. I have a long-standing interest in modeling and simulation, and a nearly lifelong fascination with probability and statistics. Professionally I've done P&L forecasting for a Fortune 500 company and built models of customer behavior for a credit union.
Boom Wave theory - 11 September 2011
A key component to the industrial cycle model I published here last August is what I call boom waves. These appear as regular patterns of growth kicked off (or at least signaled) by certain inflection points in the labor market.Boom wave: A six year pattern of industrial activity that has tended to follow periods of moderate unemployment (5.5% to 6.1% range) in the postwar era.
The model was constructed by looking at average DJIA returns over eight prior postwar booms. Please note that most of what is said here could also apply to varying rates of GDP growth. It's often said that the stock market is not the economy. While this is true, the research at hand seeks to measure cyclical business activity. Those cycles are picked up with great sensitivity by certain stock market averages.
A fairly consistent six year pattern emerges:
First year: up (increased cyclical activity; rising DJIA)
Second & third years: variable but often flat to modest increases
Fourth & fifth years: up
Sixth year: down (decreased cyclical activity; falling DJIA)
Intercorrelations among annual returns for the eight boom wave periods average about .50 (Pearson's r). Notably, though, five of the eight correlate with the others at very high levels (>.60) while the other three are not significantly intercorrelated (although no negative correlations are observed).
The two year period spanning the fourth and fifth years of the boom wave corresponds, in theory, to 2012-2013. The latest boom wave kicked off as the unemployment rate rose past 5.5% in 2008. This two year period is the most consistent observed in the model, featuring DJIA gains ranging from 18% (mid 1990s) to 56% (mid 1970s). Two year gains have averaged 37.5% across eight boom waves. However, please read the caveats below if this finding makes you want to contact your broker!
A hypothesis:
Is there a period of accurate labor matching that takes place in times of moderate unemployment? When the labor market is neither too tight nor too loose, perhaps people (or at least a certain group of people) make job changes for the right reasons, get into projects that best suit their talents and at first through their personal gains and investments and later through the products of their labors drive the economy to higher levels.
A real-life example:
An acquaintance of mine then employed as a web developer changed jobs during 2008, with the unemployment rate at about 5.5%. He left an old-line tech company, moving to a social media startup. Fast forward three years: his company recently announced plans for an IPO and my acquaintance has been promoted to management. His company appears poised to vault to new heights. At the same time, social media (Facebook et al.) along with mobile devices and the software that runs them have emerged as economic drivers to an extent that seems surprising, but perhaps in retrospect should not.
This chart illustrates the current boom wave in blue (Dow Jones Industrial Average) against a background of historical averages in red. 2009 was the initial year, 2010 and 2011 are midpoint consolidation years and 2012 and 2013 should, in theory, be a long boom period as projects come to fruition and are monetized, followed by a correction in 2014.
Elements of the theory certainly have some appeal. The idea of a "natural rate" of unemployment is widely held, yet has thus far found little real-world application. This theory places an optimal range of unemployment firmly at the center of a dynamic model of labor activity. The booms observed - five years of expansion followed by a downturn - align well with our historical knowledge of expansionary periods. Folk wisdom in the business community also regards the length of a business cycle as roughly five or six years. For the investor the potential to explain the dominant cyclical activity of the market is compelling (the model correlates best with highly cyclical indexes like the NASDAQ and Hang Seng).
All the usual caveats apply, of course. I completed the basics of this theory in 2009 and it is still firmly in the validation phase. While I think the next two years should be telling, reality often diverges from even sound theory and frankly I can't be certain at this point whether the theory is sound.
The safe way to double your money is to fold it over once and put it in your pocket - Frank Hubbard
Market Note - 10 June 2011
The Morgan Stanley Cyclical Index closed below its 200 day moving average.Market Note - 17 February 2011
The Morgan Stanley Cyclical Index closed at an all-time high today. The previous high close was in July of 2007.A Forecast of US Industrial Cycles - 26 August 2010
The first product I'd like to share is some output from a forecasting model of the US industrial cycle. It's a very simple model. The well-known four-year political cycle (sometimes called the "presidential cycle") is represented by a fixed progression, and modified by inflection points in the labor market. The labor market is generally thought of as a lagging indicator, but if you subscribe to the idea of recurring cycles there is no reason it can't also be a leading indicator.
The model forecasts slow industrial activity this year and perhaps into early 2011, followed by three rather strong years for the major industrials.
The Dow data points are year-to-year change based on the index's initial opening prices each October. The implication is that there is a "real" industrial year that runs from Q4 through Q3. The backtested model is strongly correlated with the Dow (.621 Pearson's r).