Gold Funds: The Select American Gold and Precious Metals
fund were very strong at the start of the year, but you did not buy them. Why not?
Those of you who have read some of my articles or heard me speak
already know that I eliminate these two funds from consideration.
I checked my literature, and it does not discuss this issue since
it already has no lack of text. I have tested with and without
incorporating these two funds, and the results are better without
these funds. It may be possible to develop a system similar to
mine that uses the two gold funds, but I decided not to try. I
view the gold funds not as investing in major sectors of the economy
like the other Selects do, but more as a commodity. (In fact,
gold stocks tend to be more volatile than the price of the metal.
However, a gold stock fund is less volatile than a gold future
since the fund is not leveraged and the future is to a high degree.)
Thus, I felt the inclusion of the two gold funds was philosophically
inconsistent with the Select Switching Strategy.
This leads to the question of why the results were
worse when the gold stock funds were included. This is due to
two factors. First, in my test period going back to 1987, the
gold funds have been the worst overall performers of the Select
funds. Gold in general has been going down since its peak price
in 1980. It is up from the lows of a few years ago, but gold prices
and stocks have not done much over the past nine years. In contrast,
the broad stock market has done very well over this period. It
is possible that had my test period included a period of rising
gold prices, that I would have included the funds. However, given
the number of Select funds in existence before 1986 and the fact
that the first one started in 1981, such a test is impossible.
The second factor is that the two gold funds are
the most volatile of the Select funds. That means they will rise
to the top of my rankings fairly often. However, this volatility
combined with the overall poor performance of the funds means
that trades in the two funds according to my system are quite
likely to be losers.
Let's see how trades in those funds would have worked
this year. If the two funds had been included in my system, the
parameters I use almost certainly would have been different from
those I actually use, but for the sake of simplicity, I will apply
the same ones I use now. Precious Metals could have been bought
on January 8, 15, 22, or 29, and it would have been sold on March
11. Based on the closing prices those days, the results would
have been +4.2%, +4.6%, -0.6%, and -2.5%, respectively. American
Gold could have been purchased on February 5 and sold on March
25, which would have been a loss of 0.9%. Actual trades made in
client accounts on the five Mondays from January 8 through February
5, some of which appear in the table of typical trades above,
had outcomes of +7.2%, +8.5%, +2.1%, +0.5%, and -3.9%. The actual
trades were not made at the closing prices, so the comparisons
are not perfect, but they do not make me wish I could have used
the gold funds in this instance.
The above does not mean that I object to owning gold stock mutual funds. I own some (not Fidelity's) for my personal accounts to serve as hyperinflation insurance. My strategy is to keep about 5-10% of my equity investment in these funds. I do not try to time them at all. Each quarter, I review and evaluate my holdings and then, if necessary, buy or sell gold stock fund shares to bring my holdings into the indicated range.
Rate of Return: You say that the rate of return calculations
shown in the quarterly summary follow the guidelines of the Association
for Investment Management and Research (AIMR). Are the rates shown
in our quarterly reports calculated using AIMR? What are these
rates; are they the same as the internal rate of return (IRR)?I use the AIMR method for calculating the rates shown in your
quarterly reports, except that my management fee is not included
in that report. Most of the time the method of calculation is
irrelevant since the only time the method makes a difference is
when there has been an addition or withdrawal during the quarter.
In such cases, the AIMR return may be different from the IRR.
The AIMR method is designed to evaluate the performance of the
portfolio manager, and the IRR is a rate of return that brings
all the cash flows and the ending amount into balance. I will
illustrate the AIMR method and the difference from the IRR with
a contrived, exaggerated example.
Suppose a portfolio begins with $1,000 in it. After
a period of time (the exact period is not crucial to the example,
but let's say it is nine months), the value has doubled to $2,000.
At that point, $8,000 is added bringing the total value to $10,000.
In the subsequent period (again not crucial, but for simplicity
we will say it is three months), the values falls to $9,000, a
loss of 10%. The AIMR calculation proceeds by taking the ratios
of the ending to starting values of the two periods and multiplying
them together. In the example, the ratio for the first period
is 2, and the ratio for the second period is 0.9. The product
of the ratios is 2 x 0.9 = 1.8, which translates to a total return
of 80%. Since the total time was one year, this is also the annualized
rate of return. In contrast, $9,000 was invested ($1,000 at the
beginning and another $8,000 after nine months), and the ending
value is also $9,000, so the IRR is 0%, which is a far cry from
80%!
Which is correct? Is this another example of "figures
don't lie, but liars figure"? The answer to which is correct
is that it depends on what we want to measure. From an investment
standpoint, there has been no increase or decrease from the total
amount invested, so the return is 0%, the IRR. However, when evaluating
the portfolio manager, is it fair to "penalize" him
or her because additional funds were added just before the poorer
performance period? Most would say it is not fair, and that is
why the AIMR method is used. It factors out the cash flows to
focus on investment performance. A manager who can double the
value in nine months and only give back 10% in the next three
months is probably doing a good job. (Since we don't know what
the market conditions were, this opinion has to be hedged.) The
exception would be if the manager's "system" called
for the addition of the $8,000 at the end of the first period.
Since all additions and withdrawals are determined only by my
clients, I think the AIMR calculation method is appropriate for
evaluating Select Switching performance. If you have not made
any additions or withdrawals since opening your account, then
the AIMR and IRR rates will be the same. If any such cash flows
have been relatively small compared to the overall size of your
account, then the differences, if any, should be small. However,
I will calculate the IRR for your account upon request.
Each quarter, I do a calculation similar to the one
described for each client using the account values on the dates
of the cash flows to determine the ratios. For most, this means
dividing the ending amount by the beginning amount since there
were no additions or withdrawals. Accounts that have not been
open for the entire quarter are excluded from the calculation
of the values shown in the table at the beginning of the newsletter.
The individual rates of return are reduced by 0.5% to account
for my management fee. Next I need to average these rates of return
to get a composite rate that is representative of all the accounts
open for the entire quarter. I use two methods. In the first,
which is recommended by AIMR, the rates for each client are weighted
by the amount under management at the beginning of the quarter.
For the second calculation, which is encouraged by AIMR as additional
information to present, each client rate of return is given the
same weight, which means it is a simple (i.e. normal) average:
add up the numbers and divide by how many numbers are included.
Since I manage all accounts somewhat similarly, the two values
should be close, and so far they have been. A manager who applied
much different strategies to different accounts might have weighted
and unweighted calculations that were not close. Finally, I use
the quarterly composite rates to determine the annual, year-to-date,
and total rates of return shown in the table.
Not all portfolio managers and investment advisors follow the AIMR guidelines when reporting rates of return. While most welcome the de facto standard that can provide a fair basis of comparison, for some the computational complexity can be onerous. An advisor with many accounts that have frequent cash flows could find the type of calculations described above quite a burden. Imagine the number of calculations involved if there were an average of two cash flows per month for many accounts. The problem would be compounded if the advisor had been in business for a long time and had not computerized his bookkeeping in a way that facilitates AIMR calculations. Fortunately, my accounts have infrequent cash flows, and I have been set up from the start to do AIMR type calculations.