One important facet of portfolio construction and portfolio management is diversification. By offsetting the risk associated with individual assets, diversification reduces the variability of portfolio returns.

To achieve diversification, the returns on the various assets in a portfolio should not be highly correlated. Adding securities whose returns are highly correlation does not reduce the variability of the portfolio return and does not contribute to diversification and the reduction of risk. Thus, the investor or portfolio manager needs to identify assets with returns that are low or even negatively correlated.

Exchange-traded funds (ETFs) have become a popular investment vehicle, since the investor acquires a basket of securities without having to select individual stocks or bonds. ETFs may be promoted as a means to achieve diversification and manage risk. Diversification, however, will not be achieved if returns on the various ETFs are highly correlated.

The following tables report the correlation coefficients for a variety of ETFs. Weekly returns are computed from the date of each ETF's inception. The resulting returns are used for the calculation of correlation coefficients. While some ETFs (e.g., the SPDRs based on the S&P 500 stock index) have existed for several years, many are relatively new, having been created only in the last few years. The relative newness of many ETFs limits the number of time periods that may be used to calculate the returns. While the results may be considered preliminary, they do suggest that the returns on many ETFs are highly correlated and hence their ability to contribution to diversification is limited.

Table 1 starts with three aggregate ETFs based on the S&P 500 (the SPDRs, the iShares and the ProShares Ultra S&P500). As would be expected, the returns on SPY, IYY and SSO are highly correlated. The subsequent parts of the table present the correlation coefficients relating SPY, IYY and SSO to ETFs specializing in specific sectors: health stocks, natural resources and real estate investment trusts (REITs), and ETFs based on diversified emerging markets, world stock markets and precious metals. If the investor adds these ETFs to a portfolio holding SPY, IYY or SSO, the expectation would be that risk is reduced through diversification.

With the exception of precious metals, the correlation coefficients in Table 1 are relatively high. The precious metal coefficients are consistently less than 0.3 and in some cases even negative, suggesting that adding precious metals ETFs to an aggregate stock portfolio reduces risk. For the remaining ETFs, the potential for risk reduction is small. With one exception (IGE) the coefficients exceed 0.5 and the majority exceeds 0.8. This may be surprising since even the emerging markets ETFs are highly correlated to the aggregate market ETFs.

Table 2 repeats the process using ETFs based on the Dow Jones US index and the Dow Jones industrial average (IYY and DDM). In this table, the ETFs are expanded to include currencies, U.S. Treasury securities, Asian markets, small cap stocks, as well as three sectors: financials, utilities and technology. The selection is also expanded to include ETFs that take short positions in U.S. Treasuries (TBT) and short positions in stocks (RWM for the Russell 2000 index, SKF for financials, SDP for utilities and REW for technology).

The coefficients in Table 2 duplicate the results from Table 1. The Asian markets, small caps and long sector ETFs are highly correlated to the aggregate market. The short ETFs (FPX, RWM, SDP and REW) are highly negatively correlated to the aggregate market; these ETFs virtually mirror the comparable long ETFs. For example, the correlation for the long ProShares ETF (UGY) and the aggregate market is 0.906. For the short ETF (SKF), the coefficient is -0.876. Adding the short ETFs should contribute to diversification.

Tables 3 though 5 report the coefficients between precious metals, real estate investment trusts, currencies and the other ETFs. While the precious metal ETFs are highly correlated among themselves, little correlation exists between the precious metal ETFs and the other ETFs. While the returns on the various REIT ETFs are positive (Table 4), the coefficients do suggest that adding a REIT ETF to an aggregate portfolio contributes to diversification. Even though the correlation coefficients between the REIT ETFs and the other ETFs generally exceed 0.5, they are lower than the coefficients between the aggregate markets and the various sectors.

The coefficients for currencies and the selected ETFs in Table 5 suggest that adding currencies also contributes to diversification. While the majority of the coefficients are positive, the numerical values are small (i.e., less than 0.4) and in some cases are negative. For example, one ETF based on the Japanese yen (FXY) is negatively correlated with the ETFs for small-cap stocks, utilities and technology stocks.

The results reported in Tables 1 through 5 are limited by the short period during which many ETFs have existed and by the return data being generated primarily during a period to declining stock prices. The results of the correlation coefficients relating returns on aggregate market ETFs and more specialized ETFs, however, do suggest the potential for diversification. The results also highlight that the investor needs to be selective in the choice of which ETFs to include if the objective is to diversify the portfolio.

Table 1: Correlation Coefficients Between S&P 500 and Selected ETFs

Previous Page

1  |  2  |  3  |  4

|  Full Article |Next Page