by Charles Schwartz, Professor Emeritus, University of California, Berkeley
email@example.com January 26, 2003
>> This series is available on the
Internet at http://socrates.berkeley.edu/~schwrtz
The Treasurer Responds; Some Criticisms Get Stronger
A new report from the UC Treasurer
purports to answer my earlier criticisms of his calculations of past UCRP
investment performance. While the simplest issue has now been explained,
we find that the main import is to substantiate the two stronger criticisms
-- about arbitrary selection of historical benchmarks, and about the total
neglect of considering risk-adjusted returns.
At the January 7, 2003 meeting of the regents' Committee on Investments, UC Treasurer David Russ presented a 55-page report, titled, "Long Term Performance Calculations for UCRP." He said that this detailed report was compiled in response to questions that had been publicly raised about where his data came from, and his Executive Summary states,
Clearly, this new report was a response to my criticisms, published in Part 13 and Part 14 of this series, which I can summarize as follows:
# 1: The Treasurer looked only at the bare Returns (on the equity investments of UCRP over the past decade) and did not consider Risk-Adjusted Returns, which is the proper way to evaluate the past performance of a portfolio.
# 2: I found that the 10-year annualized return on equities as reported by the Treasurer (10.3%) did not match the result I calculated (10.7%) using data collected from past annual reports published by his office.
# 3: I questioned whether the benchmark he used (the S&P 500 Index) was actually the benchmark in place at that earlier time, or an ex post facto imposition.
The primary point of Mr. Russ' new report was to respond to my criticism #2, the least important of them all. Two footnotes on page 6 of his report (the table on Returns of Active Equities in UCRP) explain how he had chosen to revise the data published in the U.C. Treasurer's Annual Reports over the past decade. Beyond that, the massive amount of stuff he provides in this new report serves to verify that I had copied the data faithfully from those earlier official documents. If Mr. Russ had simply answered my email inquiry of December 9, all the tedious staff work and bother that went into this report could have been avoided.
But there is more to be learned from Treasurer Russ' new report, when we ask what he had to say about my other two criticisms. On my questioning (#3) about what was the "established benchmark" for those earlier years, he says nothing explicitly. In my paper Part 14, I wrote, "Maybe there is some dusty document that I have never seen that could justify their position" that the S&P 500 Index was the true "historical benchmark" for UCRP equity investments. Mr. Russ offers no such documentation (and I assume that he searched to see if any such existed.)
What is more, the detailed pages from those earlier Annual Reports, which Russ has copied into the massive appendix of his latest report, show explicitly how he selected the "S&P 500" comparison data (which appear in only five of the earlier references) while ignoring the other comparison data - "CRA(SEI) Equity Only Median" - which appears in all of the earlier references. As I noted in Part 14, adopting the CRA comparison as the true historical benchmark, turns the final result upside down! If Mr. Russ has nothing further to say about this, he has demonstrated a major act of dishonesty.
My #1 criticism was about ignoring the combined consideration of Return and Risk in any proper evaluation of a portfolio's past performance. Mr Russ has now acknowledged this explicitly. The last paragraph of his Executive Summary states,
For the regents, I can mention that at the October 29 meeting of the Committee on Investments you heard from an invited expert, Stanford University Professor Joe Grundfest (a former member of the U.S. Securities and Exchange Commission); and here is something that was recorded in the Minutes of that session:
Since it appears that many regents (and maybe also their Treasurer) do not have a good understanding of what "risk-adjusted returns" are all about, I will presume to present here a brief tutorial on the subject.
It is essential to understand that Risk and
are two measurable variables and one must evaluate any portfolio, or any
benchmark, in term of both of these variables considered together -- not
two considerations taken separately! This is conventionally done with a
two-dimensional graph, in which one plots Returns along the vertical (y-)
axis and Risks along the horizontal (x-) axis. See Figure 1, where I have
plotted two hypothetical points, one representing the market (M), also
called the benchmark or the index, and the other point representing your
portfolio (P). I have not put any numbers on this graph because, for this
discussion, we are only interested in understanding the relative
evaluation of P and M.
M = Market or Benchmark
P = Your Portfolio
In Figure 1 we see that this portfolio P has a lower risk and also a lower return compared to the market M. In Figure 2, I show four different portfolios to compare with the same market M. Portfolio P2 has a lower risk and a higher return than M; this certainly means that P2 did better than the market M. Portfolio P4, on the other hand, has a lower return and a higher risk compared to M; definitely a worse performance.
Portfolios P1 and P3 in Figure 2 leave us with
an apparently ambiguous situation (as did point P in Figure 1). We
the general idea that greater return should be associated with greater
risk. But the central question we now face is: How do you determine
how much additional risk is expected to yield how much additional
M = Market or Benchmark
P1 - P4 = Various Portfolios
P2 is better than M
P4 is worse than M
P1 and P3 are ??
A standard answer to this question is presented
in Figure 3. You bring in the concept of a "risk-free" investment that
produces some known return and mark this point (F) on the graph. In
this is usually taken to be the rate of return on short term U.S. Treasury
Bills. Then you draw straight lines from P and from M to this risk-free
point. The line with the steeper slope is the "winner". (This is Sharpe's
Ratio.) The M2 procedure is to slide the point P along its line
until it has the same risk as the point M, then compare this
return (P') to M. In the case shown in Figure 3, the portfolio P is seen
to have beat the market by either of these two comparison methods.
M = Market or Benchmark
P = Your Portfolio
Lines to the risk-free return F
show that P is better than M
There is another risk-adjustment procedure that is more sophisticated; it requires calculating something called beta, which involves the covariance of the Portfolio returns and the Market returns. I have been trying to get the data from Treasurer Russ so that I can do all these calculations for UCRP and find out the correct result. For some unexplained reason he has been balking at providing the data, although he has acknowledged that he has it available in his office computer files.
There are a few more comments to make about these two-dimensional graphs. Firstly, the data that one plots are typically annualized (averaged) risks and returns taken over some interval of time -- ten years, for example. The accepted definition of risk is the statistical quantity "Standard Deviation" of the returns.
Secondly, if you look in the Quarterly Performance Reviews, which the UC Treasurer has presented to the regents' Investment Committee for many years, you will find some examples of these two-dimensional plots up through September 2001 and going back to at least June 2000. [See the collection posted at www.ucop.edu/treasurer/qpr/welcome.html ] That period overlaps the tenure of former UC Treasurer Patricia Small and also of Interim Treasurer DeWitt Bowman. In the subsequent Quarterly Performance Reviews, produced under our current Treasurer, those plots have disappeared.
Thirdly, it appears that Mr. Russ' office is slowly coming up to speed. He recently hired a new staff member with the title of Director of Investment Risk Management, and this new expert, Jesse Phillips, gave a presentation at the last meeting of the regents' Investment Committee, on January 7, 2003. His slides included two that were this type of Risk-Return 2-dimensional plots. Unfortunately, that discussion did not manage to say anything about risk-adjusted returns.
Perhaps Mr. Russ will eventually learn about risk-adjusted returns and use them in making proper evaluations of his office's performance in presentation to the regents and the public. What bothers me terribly is that his half-baked analysis has already led the regents to make a precipitous decision -- firing the whole equity investment staff and dumping $15 billion in stocks into the index fund.
The next step, promised by Mr. Russ, is to hand this bulk of money out to a bunch of private investment companies, in the hope that they can do a better job of stock-picking. My best advice to the regents is: Whoa, don't go there until you are sure that you know what you are doing. In the next paper (Part 16) I will present a number of issues and raise questions about the wisdom of that new Multiple Manager approach, and ask whether the regents' advisers should be trusted.
Throughout this paper I have focused my critical
remarks on UC Treasurer David Russ. He has been the point man in this
story but I do not hold him solely responsible for the dubious practices
that I point out. Certainly the consultant from Wilshire Associates (Steve
Nesbitt) has played a major role in (mis)guiding the regents; and I also
suspect that a few members of the Board are knowingly involved in the