Tyler9000 wrote: ↑Fri May 27, 2022 3:16 pm
McQ wrote: ↑Thu May 26, 2022 10:15 pm
I'll share results once achieved (not next week or next month
I just wanted to second Siamond's interest in your work on this. I look forward to seeing what you're able to share.
Thanks Tyler, I do well with encouragement
As a down payment, in case I get distracted and this effort takes a while, here is a periodization for “estimating the historical costs of owning a broad stock market index before the era of the index fund,” with a few notes on sources discovered thus far.
My motivation: before today’s era of ETF index funds you couldn’t own the stock market or any good proxy for it without incurring significant costs. Today you pay single digit basis points annually for an index ETF; as recently as 1990 you paid 20-25 bp; and when the Vanguard 500 index fund began (first full year was 1977) you paid about 60 bp.
Accordingly, any equity premium or long-term stock investment return estimated using the Stocks, Bonds, Bills & Inflation yearbook—or the Simba spreadsheet before 1977—must be overstated, relative to the returns that could have been realized relative to expenses. And as John Bogle showed on many an occasion, having to subtract 50 – 100 bp from an annualized return of about 10% has a baleful effect on wealth accumulation over longer periods.
A typical retort might be that all the other historical asset data out there is also estimated without costs, therefore the omission cancels out when it comes to calculation of the equity premium or other asset comparison. But that’s not correct: it has always been possible to purchase and hold a government debt instrument at very low cost. The overstatement is concentrated in historical stock market returns (and corporate or total bond returns).
Periods
Again, it is not an issue after December 31, 1976: once simply uses the VFIAX returns in the Simba spreadsheet to get the after-expense return from owning a broad portfolio of stocks.
The more distant past can be divvied up as follows:
1940s to 1976: Wiesenberger, Johnson and similar Morningstar predecessors table expense ratios and fund returns. These are the records I’ll be examining shortly. I’m hoping to get tracking error as well as average expense ratios.
1927 to 1940s: There is an SEC report for the earlier part of the period. It is remarkably modern in tone, i.e., compares fund performance to an unmanaged index (S&P 90). There is a later SEC report from 1962 generally accounted to have kicked off the modern era of academic mutual fund research; I am working through it now.
1918 to 1927: There is only one open-end fund in this era, and closed-end funds may not be tabled. I’m expecting to take the 1927-1940 data and extrapolate back to my 1918 beginning (that’s when long Treasury bonds of the modern sort became available, and is the terminus of my current project). The Jones paper cited earlier will be helpful in making that extrapolation.
1917 and before: I probably won’t attempt. It would all be guesswork (although Jones’ cost estimates do go back to 1900).
An important finding from Jones: expenses have not been constant over time. There can be no simple extrapolation backwards from, say, the 1977-1981 expense ratio on VFIAX.
Here is an interesting table from Bogle’s blog, which I found courtesy of sycamor. A few comments and then a regression analysis follow.
1) In the 1970s and early 1980s, Vanguard could not offer an index fund for a substantially lower cost than its existing stable of actively managed funds. The overall Vanguard expense ratio of 60 – 66 bp is about the same as the tracking error for VFIAX in those years (see my earlier post).
2) The expenses on everything drop steadily over time. Bogle had every reason to hammer the rest of the industry on costs: the Vanguard record shows what a firm committed to passing on as much of the asset return as possible to investors could accomplish.
3) I ran a simple regression on selected columns, with expense ratio expressed in bp as the dependent variable.
Findings
a) The passage of time explains most of the drop in expense ratios: R-square = 92.2%. The beta on calendar year is -1.3 bp per year.
b) Next I added assets under management (e.g., economies of scale). This slightly improves prediction (R-square 94.5%).
c) Then I added assets in index funds. This doesn’t improve R-square much (95.4%), but the beta is significant, and it does amp up the calendar year beta to -2 bp per year.
Conclusion
The passage of time (=developments in the markets) explains most of the steady drop in Vanguard fund expense.
Even as recently as four decades ago, it was much more costly to offer ownership of a broad stock fund to investors, with an annual drag over 50 bp. Nor was it much cheaper to index than to actively manage a fund.
Not sure what I will find in the years before. Jones argued that costs peaked in the 1960s and early 1970s, so costs in the 1930s, 1940s, and 1950s may be lower than in the first years of VFIAX. Bogle, in another book, has Mass Investors Trust, the oldest open-ended fund, with an expense < 20 bp by 1960.
I’ll let you know what I find. In the meantime, a query for anyone: Almost all the early open-ended funds were load funds, typically 8.5%. If I include the load, after-cost historical returns are going to be that much lower still. Is that kosher, or thumb on the scale?
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.