Wednesday, August 8, 2012

The S&P 500 versus Cyclical/Non-Cyclical Companies Revisted

I really cannot get over how well the price of the Morgan Stanley Cyclical Index relative to the Dow Jones Consumer Non-Cyclical Index has tracked the price and performance of the S&P 500 over the last decade (which I spoke about here and here). That said, I also still cannot get over the feeling there is some statistical aberration or other quality inherent in both indexes, that is not otherwise readily apparent, that results in this relationship.

To that end, I further expanded my research and expanded both the time frame and the available universe. First, I used monthly price and performance figures for a period including the end of 1991 through August 2012. In addition, I expanded the available universe to include all of the stocks (excluding shares of Berkshire Hathaway, as the high share price skews the results) in the Russell 3000, a broad universe of stocks covering everything from the mega-cap to small-cap stocks.

In this study, I defined cyclical and non-cyclical stocks using two different classifications. The first being the overall volatility in sales and the second being the volatility in the stock price, the later being used on the premise that the market will anticipate fundamental inflection points. Using the standard deviation of sales and stock prices, I broke the Russell 3000 into quartiles. Companies with either low volatility of sales or stocks prices were designated as non-cyclical stocks while I designated cyclical companies has having high volatility in either sales or stock prices. For any monthly period, I added the stock prices for cyclical and non-cyclical companies, calculated the relative value relationship, and compared this relative price to the S&P 500. The results are show below.

The first chart derives the relative cyclical/non-cyclical price performance based on the volatility of sales, delineating the Russell 3000 into two separate halves.
The second chart derives the relative cyclical/non-cyclical price performance based on the volatility of sales, with cyclical (non-cyclical) companies defined as the highest (lowest) volatility quartiles. (The field in the below chart is truncated due to the tech bubble pushing the relative relationship to extraordinary highs in late 1999.)

The third chart derives the relative cyclical/non-cyclical price performance based on the volatility of stock prices, delineating the Russell 3000 into two separate halves.

The last chart derives the relative cyclical/non-cyclical price performance based on the volatility of stock prices, with cyclical (non-cyclical) companies defined as the highest (lowest) volatility quartiles.

A few items worth noting.
- The structure of all four relative price charts are very similar.
- That said, the relative price of cyclical to non-cyclical stocks based on sales has more of a flat structure over time (some of this may be due to the scaling) as compared to the relative price chart derived from stock price data.
- I think the stock price derived models are probably more robust, as stock prices capture other important factors- over and above just sales- including earnings, income and balance sheet structures, cash flows, and most important forward-looking expectations. 
- With that said, my comments will be more focused on the stock price derived results.

You should immediately notice that, over the last 20+ years, the relative price of cyclical stocks relative to non-cyclical stocks tends to capture the directional momentum of the market very well, as the relative price tends to move up and down in tandem with the market. Interestingly, the relative price performance peaked ahead of the market tops in both the late 2000's and in late 2007. In addition, the relative price turned up ahead of the market bottom in early 2002. Most interestingly, you see that the cyclical stocks never recovered their value, relative to non-cyclical stocks, since the market bottomed in early 2009. The same dynamics, although less pronounced, is seen in the sales-derived models. One particular aspect I find noteworthy is that following the market bottom in 2009, the sales derived model that is based on the top and bottom quartiles of sales volatility showed cyclical stocks increasing in value relative to non-cyclical stocks from the market bottom in 2009 through February of 2011. Since February 2001, this model has not been confirming the market rally, but has fallen off. This corresponds with the results seen in the study using the Morgan Stanley Cyclical and the Dow Jones Consumer Non-Cyclical Indexes. It also fits the pattern seen in the weekly figures from the Economic Cycle Research Institute (ECRI), the firm who has repeatedly stated their call for a recession.

Getting back to the data, despite the market approaching highs again, cyclical stocks appear to have underperformed their non-cyclical counterparts. Is this a result of a Fed manipulated market where stock prices rise on the hope of monetary stimulus and not real economic growth. I think the answer is largely yes, but more evidence is probably needed.

No comments:

Post a Comment