The Columbia Dividend Income fund (I class: GSFTX) has been a favorite of investors over recent timeframes, with total assets in the fund hitting the $13.1 billion mark as of March 5th, 2019. Afterall, its performance has been in the top quartile of the Morningstar Large Value category for the trailing one-, three-, and five-year timeframes. However, could there be a better way for RIA’s to give their clients high dividend exposure? Raltin’s correlation matrix can provide insights into possibilities for capturing lower-cost exposure to the same dividend investing style offered by the GSFTX fund, for a fraction of the price.
By running a search with Raltin’s correlation matrix we find that the Wisdom Tree US Large Cap Dividend ETF (DLN) and Schwab US Large-Cap Value ETF (SCHV) top the list of correlated ETFs, both returning a correlation score of 0.92, and the Vanguard High Dividend Yield (VYN) coming in close behind with a correlation score of 0.91 with respect to GSFTX. All three of the alternatives produced by the search of Raltin’s database could have gotten a similar return stream to that of the Columbia Dividend Income fund for less than half the expense, as the GSFTX fund’s expense ratio was 0.71% compared to 0.28% for DLN, 0.04% for SCHV, and 0.06% for VYN.
The DLN ETF has a similar objective to that of the GSFTX fund, whereby it invests in dividend-paying US equities. The alternatives to GSFTX that Raltin’s database returns are also excellent sources of yield, with each surpassing GSFTX’s 1.99% should that be the investors’ main objective. The DLN yielded 2.73%, SCHV yielded 2.84%, and VYM yielded 3.20% (all figures as of March 5th). Continuing with our substitute DLN as an example, we can get a visualization of exactly how close the performance with the GSFTX fund actually was. By clicking on the hypertext of the ETF name in the “All Positively Correlated ETFs Sorted by 5 Year Correlation Score” table on the search results page, we can get the following two interactive data visualizations;
Additionally, Raltin’s database provides a regression analysis of the returns of DLN relative to those of GSFTX. Hovering the mouse over each datapoint shows the daily returns for GSFTX along with the corresponding return for DLN.
Finding Mutual Fund Substitutes Using Correlation Instead of Active Share
A popular measure for finding how “active” a fund is (or in other words how different it is from the benchmark) is the active share method, made public in a paper published by Yale professors Cremers and Petajisto in 2009. Conversely, active share is often used to measure how similar two funds are. The higher the figure the equation returns, the more different the two funds are, and vice versa. The relatively intuitive equation compares the portfolio’s exposure to a given index constituent relative to the exposure in the index, then sums the total in differences between the index and portfolio. The resulting figure is the fund’s active share. Investors often use active share to weed out closet benchmarkers, as well as to seek funds which have a strong probability of giving positive excess returns. Active share is not without its drawbacks, however, because in the above case of the substitute DLN, active share would not have picked up on this excellent substitution for GSFTX, since DLN holds so many “off-index” holdings relative to the Russell 1000 index. The active share figure would have been high, thus precluding it from any screen as an appropriate substitute. Correlation is in many ways a superior measure to find replacement funds across indexes and investing styles which may hold different securities but provide similar return streams.