by Charles Schwartz, Professor Emeritus, University of California, Berkeley
email@example.com November 4, 2002
>> This series is available on the Internet at
Lies, Damn Lies and Statistics
We expose an attempt by some of
the University's top officials to manipulate the statistics on UC's past
investment performance and thus create a distorted picture of the choice
between internal and external management of the pension
At the September 18, 2002 meeting of the UC Regents' Committee on Investments I watched a very cunning show. Let me tell you about it. Regent Gerald Parsky, Chair of the Committee, introduced the subject as follows.
At that September meeting Mr. Russ gave a detailed presentation to the regents, with many pages of data on performance (returns) of the investments, over various time periods and with detailed comparison to the benchmarks that have been adopted. Here is a most interesting portion of that presentation, with additional comments by Parsky and also by Steve Nesbitt, the University's investment consultant from Wilshire Associates.
Parsky: "Just to pause on that. The benchmark that David has used is the benchmark that was in place at the time. So it was the S & P for the period when it was the S & P and it shifted to the Russell 3000 as we shifted. So, this is a fairly dramatic chart when you take a look at it over a ten year period."
Russ: "It is relative to the historic benchmark that has been reported to you in the past."
Nesbitt: "I might also mention that again this is annualized. So over a ten-year period, while it shows minus 1 percent, that's on an annualized basis. Over the total period it is 10 percentage points relative to the benchmark."
President Richard Atkinson: "That's why this term 'cumulative' is a little confusing."
Others: "Yes." "It is." "Yes."
Nesbitt: "I would just also add that If you extend this back to the early 1980s when performance records for equity were first kept at the Treasurer's Office, this pattern is roughly the same through the eighties. So the cumulative deficit over 10 years is 10 percent, the cumulative deficit going back to 1983 is a little more than 15 percentage points. So it is kind of a slow and steady underperformance."
The graph shown by Russ had data on the difference between the returns of the UCRP Equity portfolio and the benchmark, averaged results over various time-periods within the last 10 years, ending up at June 2002. This is the basis for Nesbitt's statement that "the cumulative deficit over 10 years is 10 percent," and that sounds like a big loss attributed to the internal management of equity investments. The data in the graph, however, show a lot of fluctuation from year to year. For comparison, if I ask the same question about the performance over the period up to June 2000 (the time when Wilshire's new investment plan for UC went into effect), the answer is that UCRP lagged behind the benchmark by only 4 percent (cumulative) over the previous ten-year period. So you can get quite different conclusions (10% deficit, or 4% deficit, or ...) depending on which data you select to emphasize.
But there is actually something much more dishonest going on here. The data presented by Russ and emphasized by Parsky and by Nesbitt was only concerned with comparative Returns. It made no reference at all to comparative Risks (and this despite Parsky's earlier remarks about looking at both returns and risks.)
It is a fundamental idea that some investments or types of investments carry a greater risk and therefore must provide a greater return on the average. Lower risk investments offer, on average, lower rates of return. When comparing two performances, such as your portfolio and some market index, it makes sense to look simply at the difference of returns only if both have the same risk. In fact, here they do not. The accepted way to measure risk is to calculate the statistical quantity "standard deviation", which measures fluctuations of the returns over some extensive period of time. Let us see now how the picture changes when we examine both sets of data.
In April 2001, I
detailed data on these questions from the UC Treasurer's Office. Mr. DeWitt
Bowman, the Interim Treasurer at that time, provided me with authoritative
numbers which are displayed in Tables 1a (Returns) and 1b (Risk). This
is data for UCRP and for the benchmark S&P 500, looking back over
periods of time (5-, 10-, 15- and 20- years) from the end of fiscal year
|5 years||10 years||15 years||20 years|
|UCRP - common stocks||22.2%||17.4%||17.1%||16.8%|
|5 years||10 years||15 years||20 years|
|UCRP - common stocks||17.6%||14.7%||16.5%||16.1%|
These Tables show consistently lower returns and also consistently lower risk for the UCRP common stock portfolio, compared with the S&P 500 stockmarket index. It is not a trivial matter to say what conclusion one should draw from this comparison that involves both returns and risks. (A different example: comparing the data on UCRP common stock investments and the Russell 3000 index, over the 15- and 20-year periods up to 6/30/2000, we saw that UCRP had a larger return and a smaller risk. In such a situation one can say that the UCRP performance was unquestionably better.)
A major claim to fame of economists over the past decades is a set of arithmetic procedures for making useful comparisons in the non-trivial situation. The common phrase is to compare "risk-adjusted returns", some way of combining both return and risk numbers for a useful comparison. There are several methods described in the textbooks.
In one of my earlier papers in this series (Part 7) I reported calculations using one of those formulas, the Modigliani-Modigliani method. Combining the data sets in Tables 1a and 1b in that way, UCRP came out with a better risk-adjusted return than the S&P benchmark over the 10-year, 15-year and 20-year periods.
A different calculational procedure gives something called "alpha", which is not the same as the quantity called "alpha" in the data presented to the regents by Mr. Russ. The textbook definition is the difference in annualized returns (this is the part given by Russ), corrected for the relative risk of the portfolio compared to the benchmark. I do not have the resources to do that calculation properly. However, I have been able to estimate the correction and I find that the true alpha (the risk-adjusted difference in returns) puts the UCRP equity portfolio ahead of the S&P500 index, completely contradicting Russ, Nesbitt and Parsky. I hope some professional can do this calculation correctly and tell us the exact result.
Still a third method, much favored in textbooks, is to calculate what is called Sharpe's Ratio, which involves again both return and risk data. I have found some authoritative data on Sharpe's Ratio, calculated for UC's General Endowment Pool Equity portfolio and for the relevant benchmark. This data is reported in the Quarterly Performance Reviews for December 2000 through September 2001. In two reports the UC portfolio had the same Sharpe's ratio as the benchmark, and in the other two reports the UC portfolio had a higher (better) Sharpe's ratio.
Thus, on the basis of this information, I conclude that either:
But it gets even worse. Here is another speech by Regent Parsky, near the end of the session.
It now seems clear to me that these guys are playing a most dishonest game. (Heads I win and Tails you lose.). They are skewing the data to make the past UC investment performance look bad.
But what is the purpose of this hatchet job? Read some comments by a couple of other regents, from earlier in the session when Mr. Russ presented data on recent performance of the investments.
Parsky: "As we go forward in analyzing what should be done it's important to break these out and look at relative performance."
Regent Richard Blum: "Gerry, Just to follow
on Judy's remarks. I think the number you probably want to look at -
can have a good quarter or a bad quarter - really is to look at your
numbers and your two[ten?]-year numbers of the money that's managed
and you will see that the five-year number is 2.3%, and the ten-year number
is 10.3%. You would have done substantially better ... The fixed income
portfolio did substantially better than the equity portfolio over a
period and about the same over a ten-year period. And obviously in equities
you take a substantially greater risk or in principle you shouild be. And
so I think it calls into question, certainly in my mind, as to whether
really you can put together an internal group of managers that would
with the better professionals that manage money in companies that do that.
And I think one of the disadvantages the University clearly has is the
ability to attract people because of compensation issues. It is very hard
I think for the University to compete with the, you know, the better money
Following adoption of the first Wilshire recommendations in 2000, the regents moved $11 Billion in UC's investment assets out of house, and into index funds managed by private investment companies. Here they come for another bite.
At the August 6 meeting
of the Investment Advisory Committee there was a Closed Session Item for
Discussion, "Possible Addition of Active External Equity Managers"; and
at the October 29 joint meeting of the Regents' Committee on Investments
and the IAC there was a Closed Session Item for Action, "Implementation
of Multiple Active Investment Management Programs." I filed formal protests
about excessive secrecy in both cases, but got no response.
The Players - Update
In Part 5 I gave some information about several regents who were key players in the first-round privatization of UC's investments, noting their connections to private sector investment management.
The Investment Advisory Committee was established by the Regents a couple of years ago to provide inputs from some private sector investment experts. Three regents - Parsky, Hopkinson and Blum - also sit on that advisory committee and serve to connect that body to the Board of Regents. Some background on Regents Parsky and Hopkinson was given earlier.
Richard C. Blum is the newest member of the UC Board of Regents, having been appointed by Governor Davis earlier this year. The following background comes from Hoover's ("The Business Information Authority") online resource.
|Regent||Campaign Contribution to the Governor|
|Hopkinson||$ 50,000 ($175,000)*|
|Lansing||$ 7,500 ($15,000)*|