Performance Analysis Of Contiguous Swiss Pension Fund Benchmark Indices
Pension fund benchmark indices shown here are based on various benchmark index families calculated and published by Banque Pictet & Cie. But they should not be confused with Pictet’s indices. For the rationale behind Agathos’ contiguous benchmark indices, and a description of how and why they are being calculated, click here.
The performance analysis of benchmark indices serves two main purposes:
- Firstly, such benchmarks give a fair, if very simplified description of the investment environment from which pension funds need to find an optimum blend of risk and reward. An analysis of this environment gives insight into the specific challenges pension funds need to face, which may be comparatively easy during one cycle, but exceedingly difficult during another.
- Secondly, and only in view of the known and extraordinarily high correlation between actual pension fund performance data and such benchmark indices, the monitoring of relevant data with monthly frequency gives ‘live’ insights into the dynamics impacting the asset side of pension fund’s balance sheets. Monthly data, available with little or no delay and put into long-term context, generates an altogether different dimension of understanding, not possible with the annual data available from pension funds themselves, and after months of delay.
In Chart 01, two groups of risk tier benchmarks are shown, re-based to their values 18 months ago. Plotted in grey are the contiguous, C-type benchmarks, reflecting changes to investment restrictions over time, while the indices of the original Pictet benchmark ‚2000’ family are shown in red. Underneath each graph, cumulative monthly gains and losses are shown, giving more detail into the differences between indices of the same risk tier, and across risk tiers.
Chart-02 (below) shows year-to-date performance, in the context of seven preceding calendar years. While Chart-01 did emphasise recent volatility, this chart describes the general atmosphere of each year.
Going back to the most recent 120 months, Chart-03 (above) plots actual index values. All three benchmarks have an identical base value at the end of 1999. Having travelled on paths quite apart from one another for most of the time, they had converged in early 2018 and continue to travel as a group for the time being. During this observation period, the lower risk benchmark has nearly always been on top of the other two, with the higher risk-tier unable to recover sufficiently to from the previous loss.
In Table-01 (below), the same 120 months of benchmark history is examined on the basis of normalised rates of return, and risk, and the distribution of rates of change. With regard to returns, and over these ten years, higher risk has still produced a higher return, but not high enough to compensate for the massively higher risk. This shows in risk-adjusted rates of return, and in a lower ratio of gains to losses than the lower risk tier benchmark. There will be similar tables covering longer periods further down (Exhibits 09 and 10).
Chart-04 plots, across recent 120 months, the ratio of LPP-C60 over LPP-C25. This is an indicator of the relative performance of higher risk over lower risk. A 24 months moving average is included. The moving average has already been losing momentum, as the ratio had moved horizontally unstable, for most of 2018. Currently, there the smoothing travels sideways, without a clear trend.
Chart-05 depicts trailing 3-year normalised return, again for recent 120 months of history. Due to very poor data entering the trail, these values are now in rapid decline, only slowed by their smoothing. It has been fairly unusual that, with all swings up or down, normalised returns have remained positive for as long as they had. The bull cycle has been exceedingly mature. With the recent strong three months, all benchmarks normalised returns are rising again.
3-year normalised observed risk is seen in Chart-06. Risk is shown as a naturally negative value, which places low readings at the top of the chart. As an ex-post measure of risk, this is a contrary indicator. Perhaps more clearly than the preceding chart, Chart-06 illustrates the excessive nature of the last few years. Risk was far too low, and had converged for the three risk tiers, indicating that markets were not pricing in potentially bad news. For the last three months, and unusually, observed risk for the medium tier benchmark is greater than than risk for the higher risk tier benchmark.
Chart-07 shows risk-adjusted returns, obtained by deducting normalised risk (Chart-06) from normalised returns (Chart-05). For most of 2019, net values for the highest risk tier (LPP-C60) were travelling above the medium risk universe (LPP-C40). Observed risk is a very stringent measure of risk and it is rare for higher risk investments to show highly positive risk-adjusted return. Presently, all tiers show risk-adjusted return trails to be negative, if only just.
Table-02 shows index values at fixed points in time: 31.12.1999, ten, five, and three years ago, as well as 12 months, year-to-date and previous month. The right part of the table calculates corresponding rates of return, and compares recent 12 months with recent five years. This ratio indicates how rapidly medium term performance is impacted by recent events. Note, how across more than 20 years since 1999, LPP-C60 has returned less in nominal terms than LPP-C25 did, showing that unmanaged risk has not been rewarded. The beneficial impact of active asset allocation based on risk-management is illustrated here.
Table-03 compares the original Pictet benchmarks of the ‚2000’ series with their contiguous cousins. Please note that post 1999 all indices shown reflect investable universes.
Here, both groups of benchmark indices are shown in a grid of normalised risk (horizontal) and return (vertical). The dashed diagonal indicates risk/reward equilibrium. Current values are plotted in the scatter on the left. The scatter on the right gives values from one year earlier. In the last twelve months, all benchmarks have seen a deterioration in efficiency, with the medium risk-tier showing the poorest readings.
Table-04 is identical to Table-03 but reflects the full history back to the end of 1984 (when Pictet-2000 benchmarks had been calculated pro-forma only, and were not yet investable.
Raw data for the calculation of contiguous benchmark indices was sourced from: