No Wonder ECRI’s Weekly Leading Index Data is Free

February 17, 2012

37

This commentary reports the findings of a study that compares the S&P 500 with ECRI’s Weekly Leading Index (WLI), which is published every Friday on ECRI’s website.

My analysis shows that WLI is very highly correlated (0.95) with S&P 500. I also discovered that by applying the formula used to compute ECRI’s WLI growth index to S&P 500, we can establish an S&P 500 growth index — one that, interestingly enough, leads the WLIg in identifying turning points.

Recently there was a collaborative effort by Franz Lischka, Georg Vrba, Dwaine Van Vuuren, Doug Short and Kishor Bhatia (Refer: http://advisorperspectives.com/dshort/updates/ECRI-Weekly-Leading-Index.php) to determine the smoothing formula ECRI uses to calculate the WLI growth rate (WLIg). Georg Vrba discovered the actual formula in a 1999 article published by Anirvan Banerji, the chief research officer at ECRI. Refer to the above mentioned link for a reference to this article and the WLI growth formula. George Vrba tested the formula and compared the calculated WLIg with actual WLIg and found that these matched, hence validating the discovered formula.

For this study, I applied the referenced to S&P 500 and compared this calculated S&P500g with WLIg as published by ECRI. I used the daily closing values of the S&P 500 downloaded from www.yahoo.com. My study used the closing values of the S&P 500 that corresponded to the dates of the ECRI data. Using this method, I found that 249 data points out of 7056 (3.5% of data points for ECRI dates) did not have a corresponding S&P 500 closing value because ECRI dates can occur on market holidays: 12/25 (Christmas), 1/1 (New Years), 7/4 (4th of July) etc.. For these data points, I used the closest available previous closing value for S&P 500. For example, closing value of S&P 500 on 12/24 was matched to ECRI date of 12/25.

Appendix A has the data, correlation values, and charts. The charts show a comparison between S&P500 and WLI levels and S&P 500g and WLIg for each of recessions. The data was selected to capture values two years before and one year after each recession for each plot. The purpose of the charts is to identify turning points and leads. The left and right scales of the charts were selected to clearly show the curvature/turning points (tops and bottoms).

FINDINGS

The correlation between S&P 500 market index and WLI level is found to be 0.95 which is quite high. There are three possible reasons for this:
One:   S&P 500 has a high weight compared to all other components that make up WLI and hence dominates WLI movements.

Two:   WLI level is derived by applying some smoothing formula to S&P 500 to reduce noise. Reason two explains WLI’s lag to the stock market in capturing turning points.

Three:   It is known that the stock market is a discounting mechanism. Investors individually might be studying the same components that make up WLI in the hopes to get a lead and, as a result of their collective actions, might be pricing in effects of components that make up WLI into S&P 500 in approximately the same weights that ECRI chose for WLI. In other words, ECRI does manually what the market does by virtue of discounting collective wisdom

.

S&P 500 leads WLI and S&P 500g leads WLIg. S&P 500 does a better job of picking turning points in the business cycle (refer to charts in above and in the PDF attachment). However the market index more volatile (has a higher standard deviation) than corresponding WLI metric.

Correlation between S&P 500g and WLIg is 0.7. To check the lead, S&P 500 values were lagged by one week and the correlation was computed with WLIg values. With this technique, the correlation improved to 0.8, confirming that S&P 500g does lead WLIg.

CONCLUSION

The intent of this article is to show that the record of ECRI’s WLI is very similar to that of the stock market, represented by S&P 500 itself, in providing a lead on business-cycle turning points and forecasting recessions. From the three possibilities highlighted in the first finding above, the third possibility is the most intriguing and, in my opinion, is probably the reason for the high correlation. ECRI’s record is impressive, and this article does not intend to take anything away from their success. It merely shows the relation between one of its freely published indicators and the broad market itself. ECRI has indicated that they have other proprietary leading indicators that they rely heavily on in addition to the WLI in making their business cycle calls. This study underscores the assertion that ECRI uses much more than their publically available indexes in their forecasting methdology . The question then is whether investors gain any additional information from ECRI’s WLI and WLIg that they cannot get from the freely available information about movements of the stock market itself.