Pension funds are included when a history (shown at the top left of each exhibit) of at least five years is publicly available.
In the first column (left) the frequency of absolute positive and negative yearly returns is shown. The other columns show performance differentials between the pension fund and each of the three synthetic benchmark indices.
In large type, the simple year count is given. Highlighted with grey background is a the percentage equivalent of that count in view of the length of history. Above these frequencies measures, median return differential are shown: in green above the frequency data the median positive differential, and in red below the median negative differential.
It is not always known, which benchmark index a pension fund seeks to clone, if any.
Thus, dependencies were calculated for all three but the benchmark to which the pension fund shows the highest dependency (value for ‘R-squared’) is highlighted by red background.
R-squared can range from 0% (indicating no dependence whatsoever) to a maximum of 100% (proof of a perfect clone).
The higher the reading for this dependency, the lower is the impact of ‘investment management’ decisions, regardless wether this management refers to a macro level (i.e.: asset allocation), or a micro level (stock picking, etc.), or both.
Flipping a coin has a ’50:50′ chance of yielding ‘head’ or ‘tails’, each time the task is performed. Results are totally random. But the smaller the number of coin flips, the more the frequency ratio can deviate from 50:50, yet the random nature is upheld. With three rounds, the result cannot possibly be 50:50, due to the uneven number of rounds. Head (or Tails) will appear twice as frequent as the opposite but the 50:50 law remains valid.
In the analysis of the success of pension fund’s investment management relative to a synthetic benchmark, each year is comparable to one coin flip.
Hence, the number of ’rounds’ in the sample ranges from five (pension funds with data records of less than five years were excluded), to a maximum of 19.
Not a single pension fund in the sample appears to generate investment performance different from pure chance (50:50), not even when this observation is restricted to that benchmark to which each fund shows the highest congruency (R-squared).
Put differently, the data shown here undeniably suggests, that Swiss pension fund’s performance relative to their benchmark could be replicated by implementing ‘overweight’ or ‘underweight’ ‘decisions’ based not on (alleged) expert assessment, but simple coin flips every once in a while.
This conclusion should provide food for thought for absolutely everyone involved: the legislator, pension funds, and the entire financial services industry, and perhaps above all, a faithful and trusting public who are even less likely aware of this data, than pension fund trustees themselves.