The Russell Reconstitution is a process that the Russell Indexes go through every year in June. This process involves rebalancing a wide variety of market indexes with the new indexes being quoted with their new components and weightings on the first trading day of July. The one Russell index that may get the most notice among traders is the Russell 2000. The Russell 2000 Index is comprised of a subset of the stocks that comprise the Russell 3000. The Russell 3000 is comprised of the 3000 largest US stocks by capitalization as of May 31 each year. If the top 1000 of the Russell 3000 stocks are removed from the 3000 you get the Russell 2000. The Russell 2000 is considered by analysts a great benchmark for small cap stocks and the performance of stocks that primarily do business in the United States. For traders the Russell 2000 is of interest due to an active futures and index option market based on the index.
By doing the majority of rebalancing of the indexes once a year, the Russell Index family is very unique. Many other benchmark indexes will change their composition periodically based on market conditions. There is a belief that reweighting their indexes only once a year saves managers that are trying to replicate a Russell index from having to rebalance their portfolio more than once a year. A criticism of this process is that it creates more volatility in the Russell Indexes (specifically the Russell 2000) than would occur without this rebalancing process. When I hear the word volatility I decide it is time to crunch some numbers. The numbers tell a story that does not match up to the impression that the Russell 2000 moves around more due to the reconstitution process.
I actually looked at both realized volatility and implied volatility for any seasonal anomalies that may occur around the reconstitution. First let’s take a look at implied volatility. CBOE disseminates volatility index values on a several market indexes. The CBOE Russell 2000 Volatility Index (RVX) uses the VIX methodology to determine 30 day implied volatility as indicated by Russell 2000 index option prices. I took the average RVX for June of each year since 2004 and compared that average with the average monthly RVX for the other 12 months of the year. The results appear in the table below.
Notice there really is no obvious pattern here. Five of the nine years the average RVX was actually lower than the average RVX for the other eleven months of the year. I decided to take things a step further and compared RVX to VIX. If the Russell 2000 was experiencing excess volatility due to the rebalancing process it possibly could be determined by looking at RVX versus VIX. The next table shows this comparison with the figures in the middle two columns indicating the spread between RVX and VIX by subtracting VIX from RVX. The Russell 2000 is a more volatile index than the S&P 500 so RVX is consistently at a premium to VIX.
Using relative volatility yielded basically the same results. Five of the nine years the spread between RVX and VIX was less than the average of other months out of the year. After running these two studies on RVX it appears that the Russell Reconstitution process does not result in higher implied volatility as indicated by Russell 2000 index option pricing.
The other type of volatility I wanted to check in on was realized volatility. I performed the same two exercises, but took the annualized volatility for the last five trading days of each month for each year since 2004. I wanted to narrow things down to the end of each month and see if the end of June for the Russell 2000 was any different than the average realized volatility for the last five days of the other eleven months of the year. This first table shows end of month realized volatility for June versus the average for the other eleven months of the year.
This was an interesting outcome as there were only three occurrences in the last nine years that the realized volatility for the Russell 2000 was greater than the average realized volatility at the end of the other months of the year. Note each of those three years the realized volatility was dramatically higher than other end of month periods. Finally, just to make sure market anomalies were not interfering with the outcome I looked at the Russell 2000 realized volatility at the end of the month versus the S&P 500. This final table shows these results as determined by subtracting S&P 500 realized volatility from Russell 2000 realized volatility.
As with the other three studies it appears Russell Reconstitution does not result in excess volatility. This lack of excess volatility shows up both in option pricing and implied volatility as well as short term realized volatility.
This Friday the final list and weightings for all Russell Indexes will be posted. Check the link below to find out everything you would want to know about the Russell Reconstitution process –