September 5, 2006

Are Canadian fees excessive?
Paper calls Canadian funds most expensive by far

A preliminary research paper, titled Mutual Fund Fees Around the World co-authored by U.S. and British academics, is causing quite a ruckus. The draft paper concludes, among other things, that Canadian mutual funds are by far the most expensive in the developed world. Journalists are salivating over the paper, which has caused much trepidation in the Canadian fund industry. No doubt, Canadian fees are high. But, as usual, the devil is in the details.

The paper

Co-authors Ajay Khorana, Henri Servaes, and Peter Tufano assembled a database of fees and expenses as of the end of 2002 for nearly 47,000 mutual funds sold in 18 countries. The main goal was to explain differences in funds sold in many jurisdictions. They focused on three figures: a) management fees, b) total expense ratios (i.e. MERs), and c) total shareholder costs (i.e. MERs + annualized loads).

The researchers found that Canadian mutual funds sported an average MER of 2.68% per year. In particular, the paper devotes several pages to Canada's high fees, compared to its American neighbour. The paper pegs U.S. fund fees at an average of 1.42% per year as of the end of 2002.

These figures conflict somewhat with those cited by other sources I consider reliable. A June 2003 article by Mark Warywoda (then a Morningstar Canada quantitative analyst) stated that the average MER of Canadian mutual funds was then 2.44% - more than 20 basis points cheaper than the draft paper suggests. The 1.71% figure in the draft paper for stock funds sold in the U.S. is about 10 basis points higher than the figure cited by the U.S. Investment Company Institute (see page 3 of this paper ). In other words, the paper's fee figures differ somewhat from Morningstar and ICI data covering fund expense ratios in 2002.

But this issue is minor relative to some methodological concerns.

Methodology

The paper's measure of total shareholder costs (TSC) is calculated by taking the MER and adding an annualized estimate of front and deferred sales charges. Loads are estimated and based on a five-year holding period. The researchers excluded fees paid directly by investors to financial advisors and dealers. This is where the research hits some speed bumps.

First, since the authors did not have data on the actual amount of sales charges (a point they acknowledge), they could not work with actual numbers. I don't know how they estimated the level of front or deferred sales charges levied by each load fund. Nor is there any attempt to factor in the 10% annual "free" (i.e. no redemption fee) sale of fund units bought on a deferred sales charge - a cumulative feature in some cases.

Second, the authors admittedly used an arbitrary holding period that may or may not reflect actual investor behaviour. U.S. studies (such as this one published in the Journal of Financial Planning) have quoted holding periods for U.S. non-money-market mutual funds as 3 to 4 years depending on the period of measurement. My calculation of Canadian fund holding periods has fluctuated between 6 and 7 years - roughly double that of U.S. investors. A five-year holding period overstates sales charges for Canadian fund investors while understating them for U.S. investors.

Third, I'm puzzled by the amount of the annualized load calculated by the researchers. The draft paper shows annualized loads as 198 basis points annually for Canadian funds. I inferred this figure from the difference between the 2.68% average total expense ratio and the 4.66% average total shareholder cost figures. Assuming a five-year holding period, that's equal to a total sales charge of about 10%. That exceeds any front end or deferred sales charge. I figure a 10% sales charge is only possible if both front and deferred loads are added together for each fund and assuming a fairly punitive DSC schedule. In practice, only one of front or deferred load applies to any single fund transaction - not both.

Fourth, the researchers' database includes data from Morningstar Canada. Hence, for all load funds, lower fee classes used for wealthier clients (i.e. I class) and those participating in fee-based accounts (i.e. F class) were seemingly excluded. However, similar U.S. funds with advisor compensation stripped out are included since they are bought by both do-it-yourself and advised investors. The problem is that fees paid separately for advice are excluded.

Since most Canadians buy funds from advisors paid by commissions, the Canadian fee averages include most advisor compensation paid by fund investors. However, since the U.S. industry is more "unbundled" much of the compensation paid to advisors is left out of the final figures quoted by the researchers. This is significant since, according to the ICI, "80 percent of mutual fund investors seek professional advice when buying mutual fund shares outside of retirement plans at work".

Finally, funds held by U.S. investors through workplace savings plans are generally included in data found at www.morningstar.com - but the same sort of institutional funds are not included in Canadian databases. Hence, the lower fee versions of funds held by countless Canadian investors through group savings plans at work are not factored into the draft paper's Canadian MERs and total shareholder costs for Canadian fund investors. Such lower fee funds are included for the draft paper's U.S. figures, resulting in a potential overstatement of the relative TSC of Canadian funds.

The real deal

No doubt, management fees and other expenses charged by Canadian mutual funds are high. Further, roughly half of a fund's management fee (about 40% of the MER) is paid to financial advisors and their dealers. Conversely, U.S. funds are very cheap. If measured on a dollar-weighted basis, Canadian MERs are about double the sub-1% fees of U.S. funds (before controlling for fund type and advisory fees).

Omitting a portion of the Canadian fund universe and advisory fees paid directly by U.S. investors, however, casts doubt on the draft paper's calculations. Extend these issues to the markets with less robust data included in this draft paper and it becomes clear (to me, at least) that quantifying total shareholder costs (TSC) is a mammoth undertaking that will require a great deal more data than these researchers appear to have.

Moreover, the authors of the draft paper admittedly do not address if Canadian investors get good value for the higher fees they pay. While Canadian MERs are undoubtedly high, a measure of benefits received vs. costs incurred is key to determining whether Canadians are indeed being duped, as has been suggested by some journalists. Yes, we pay more than many other countries, but this draft paper is far from conclusive due to its preliminary nature and the weaknesses identified.

Indeed, one of the authors recently told the Toronto Star that they are "re-estimating" their results using a "value-weighted" approach. While this will show a dramatic difference in fees between the U.S. and Canada; I hope that some of the challenges mentioned above are also addressed.