35 Years Of Mandatory Swiss Pension Fund Management 1985 to 2019

Part 1: Investable Universes And Performance Benchmarks

Chart-01 shows popular Swiss pension fund benchmark indices, the minimum payable rate, and three different but fixed compounding rates of return. All values have a base of 100 at 31.12.1984.

Initially, pension funds had more limited choices of permitted investments, with tighter exposure ceilings. Two a performance benchmarks dominated in succession, Pictet LPP-1985 and Pictet LPP-1993. With later revisions to federal investment guidelines, risk-tier benchmark indices were introduced, known by their equity weighting of 25%, 40%, and 60% respectively. The first of these was the Pictet ‚2000’ index family. In the Chart-01, these benchmarks have been linked to one another. Each new benchmark has been rebased to the value of its predecessor.

The same technique has been applied to subsequent index families, appending each new index to the value of its corresponding risk-tier equivalent. That process generates contiguous time series for each risk tier, but these share an identical history during 1985-1999. To avoid confusion with any of Pictet’s indices, the contiguous indices were given the letter ‚C’ as prefix.

There has been no reward for taking unmanaged risk. Any ‚buy-and-forget’ investment philosophy will have produced muted returns, regardless of the actual risk level. This simple, if uncomfortable, conclusion from 35 years of evidence is that risk-averse investors should never clone benchmarks. Data throughout this document will reinforce that same conclusion, again and again, but each time in different ways.

To give readers a more detailed insight into the investment climate during the time from 1985 to 1999, Table-01 shows profiles of monthly returns for Pictet 1993, and the original Pictet risk-tiers. The latter were not investable then, but they serve to illustrate patterns and changes therein. Pictet had published the full and detailed back-calculated data, ‚pro forma’, when the risk-tier benchmarks were introduced. This is the data leading to Table-01. While  not included in the table, the investable index Pictet LPP1993 and (then) not  investable Pictet 2000-25 have very similar profiles.

Particular attention is drawn to the metric  ‚Observed Risk’ across benchmarks, and to risk-adjusted returns calculated with it. The table shows that during the fifteen years 1985 to 1999, the more unmanaged risk an investor took, the higher the resulting return was. Nice indeed, when investing looks so easy.

When financial markets appear to reward recklessness, professionals become wary of that. It should be taken as harbinger of trouble, and evidence that excesses have begun to creep in. The market cycle of the late 1990s was publicly labelled ‚Irrational Exuberance’, by Alan Greenspan, as early as 1995, and  that label later became the title of Robert J. Schiller’s book, written in 1999, and first published in 2000. Equity prices  peaked at the end of March 2000. The actual collapse commenced in 2001. Any investor short-sighted enough to extrapolate a perceived ‚normality’ based on such the exceptional bull market from 1985 to 1999 was penalised severely, quite soon into the new millennium.

Further down in this document, the full extent of excess in equity prices will be illustrated. One of the illustrations is nearly identical to the approach used by Robert Shiller. It has indeed been borrowed from his work, and uses data compiled by him, made available, and kept up-to-date by a research team at Yale University.

A link to Robert Shiller’s original data is given at the end of this document. For the time being, readers are encouraged to cross-reference the data from Table-01 with that of Table-02,  which shows a totally different image.

Table-02 gives data for the same original Pictet benchmark indices (which have by now become investable for Swiss pension funds), but this time only for the twenty year long period from 2000 to 2019.

Merely comparing numbers can hardly reproduce the pain pension funds, indeed all investors, must have felt, caused by an allegedly unpredictable change towards ‚a normality’, that must appear outright hostile to any amateur investor. Even pension funds perceived the change like the expulsion from a proverbial Garden Eden. ‚Before’, exceptionally high nominal returns had been ‚earned’ with moderate to very low levels of risk.

One can put it in generalised terms: from an unselected, globally diversified, multi-asset investment universe(s), during 1985 to 1999, returns were approximately twice as high as the risk inherent in them. For the full twenty year period 2000 to 2019, returns passively generated by the same investment universe(s) dropped to a mere third of what they had been before, while risk remained unchanged. As a result, risk-adjusted returns were now deeply negative, except for a lower-risk asset mix.

Bonds and Equities

So far, the focus was on representations of balanced, multi-asset universes, even if in distinct iterations (risk-tiers). At a macro-level, investment management must deal with the comparative assessment of competing asset classes (liquidity, money markets, long-term bonds, equities, real-estate, currencies, and commodities). Bonds and equities are the two, structurally most relevant. Jointly, they typically make up a very large portion of any pension fund’s total assets, regardless of currency mix, or, in the case of bonds, duration.

Chart-03 shows 5-year normalised returns, and risk, at year end. M5 Equities reflect blue-chip stock in each market: SMI (Switzerland), S&P 500 (USA), Nikkei 225 (Japan), EuroSTOXX50 (Europe), and FTSE100 (Great Britain). Bond returns are based on 10-Year government yields of the same list of countries, German Bunds are chosen to represent European bonds. 

Note the fluctuation in normalised return from equities, and more stable but diminishing returns from bonds.

The seemingly small M5 basket(s) make up a hefty chunk of the entire global investment universe for Swiss pension funds. Success here results in success of the entire global portfolio. Failure to perform here will depress performance just as thoroughly. Arguably, such a universe of the five most relevant markets would be more than large enough, more than liquid enough, and more than diverse enough, for every institutional investor to implement any desired investment strategy, at any time.

Shifting Performance of Equities and Bonds

In Chart-04, the relative performance (shown as ratio) of equities to bonds is plotted. A rising ratio shows equities outperforming. Typically, equities double relative to bonds before a reversal of relative performance sets in. From 2008 onward, one can detect a change in pattern. The cycle of relative performance of the two asset classes seem shorter and less pronounced than before. This is the manifestation of massive intervention by a few, major central banks in financial markets.

When bond prices rise, their yield drops, the income stream from equity dividends becomes more competitive. So equity prices will rise, lowering dividend yield. Lowered interest cost induce corporations to finance themselves more cheaply through debt than through equity capital: They use liquidity, even debt, to buy back their own shares. With fewer shares in circulation, even unchanged earnings will appear higher on a ‚per share’ basis. Lower bond yields justify higher p/e multiples, while the ‚e’ portion in p/e grows, be it from profits, or ‚financial engineering’.

Thus, after 2008/2009 (the financial crisis) onward, equities must be assumed to reflect poor quality of corporate profits. So it only seems appropriate that a changed pattern in the ratio of equities to bonds should reflect this. Chart-04 is a text-book illustration of the performance gains achievable through an active asset allocation process that identifies trends in major market forces and acts accordingly.

Visualising ‚Before’ and ‚After’

Chart-03 and Chart-04 have provided enough detail on asset classes to now better understand what happened within indices, or portfolios, that have kept asset allocation static, and why. Returning to the level of investment universe(s) Chart 05 shows and compares normalised data graphically. Risk (horizontally) and return (vertically) serve as coordinates. The scatter on the left covers the fifteen years 1985 to 1999. The one to the right does the same for the 20 years 2000 to 2019. A dashed diagonal line depicts equilibrium (return = risk). To truly compare like with like, Pictet’s 2000 benchmark family is used in both scatters. Pictet LPP1993 (in grey) is also shown, representing the investable universe at the time. Pictet discontinued that benchmark in 2013.

‚Before’ the change of millennium, fixed asset allocation has bestowed a false, unsustainable risk-efficiency, by ‚windfall’. All universes are above the diagonal line. Even the inevitable drop in efficiency that comes with higher risk is barely visible. Higher risk was associated with higher return, and these higher returns did fully compensate for the risk taken.

Such ‚evidence’ (data in the left scatter) removed from a relevant historical context, was aggressively used to bring about a massive paradigm shift among pension funds, toward greater risk tolerance. It was even propagated to leave such higher risk ‚unmanaged’. The scatter on the right clearly demonstrates how fatal a mistake that is. Higher risk strategies are financially more rewarding for financial service providers, enabling them to market entire catalogues of specific products and services. Thus, the paradigm shift in favour of increased risk must have been based on lack of knowledge, or lack of integrity.

Bond Markets In Five Year Bins

Previous exhibits have shown 5-year normalised data at year end. In Chart-06, Chart-07, and Chart-08, history is split into bins of five years at the time, but normalised values for the full history are also shown. Note the exceptionally high returns for bonds from 1985 through to 1999 in Chart-06. Returns of this magnitude, close to 10% annually, are more typical of equity markets, not bonds. This pattern persisted for fifteen years. Towards the end of that (1995-99), risk in bonds even fell, while returns remained at exceptionally high levels. After that, the environment for bonds began to resemble typical cycles. There are limits to how far yields can drop. It is normal that bond returns diminish as a (bull) cycle matures, while risk goes up, simply because yields have come close to a natural limit, and lower coupons do not stabilise as much as early in a cycle.

2005-2009 looked even as if the bond cycle had come to an end. It most likely would have done, had it not been for the ‚financial crisis’. It prompted the US Federal Reserve to deploy a new tool: ‚Quantitative Easing’, the purchase of government bonds at a massive scale, driving yields down, and prices up. For the time being, it has, and continues to keep bond yields at artificially low levels. Directly, and indirectly, QE has created a genuine abomination, negative bond yields. How institutional investors, or indeed any investor, can rationalise buying, or holding bonds at negative yields will be the subject of future studies of irrationality. It is a bet on something that ought not exist to continue, and even to become more pronounced. 2015-19 promptly saw bond returns to be lower than the risk inherent in them. An unusual record, yet, under the circumstances, also a perfectly plausible one.

Equity Markets In Five Year Bins

The ride for equities was not as smooth as the one for bonds, as can be seen in Chart-07. Two exceptionally rewarding periods, 1985-89, and 1995-99 were interrupted (1990-94) by five ‚lean years’. These were not so much outright painful. They were simply less rewarding. The asset class level disguises what happens in major national equity markets, and these were somewhat ‚out-of-sync’ with one another. That is most valid for Japan. But it is perfectly normal that individual markets generate different returns, even when they move in the same general direction.

In the chart, 1985-89 looks more excessive than 1995-99. Again, that is the result of averaging markets, and does not reflect the full picture. This appearance is also the result of how the 5-year window relates to actual peak, and troughs. As will be shown further down, the 5-year box from 1995 to 1999 covers the most excessive equity bubble in equities in recorded history, particularly in the USA.

The sharpest contrast of 5-year intervals is seen between two neighbouring periods: 1995-99 (very high return, extremely low risk), and 2000-04 (painful losses, very high risk). As it happens, this coincides with the liberalisation of investment restrictions for Swiss pension. That is unfortunate, but does not excuse the apparent lack of differentiation between permitted exposure ceilings, and timeliness to exhaust them. 2005-14 as a 5-Year bin, belies the acute severity of the financial crisis hitting equities. The reader is reminded of comments to Chart-02. The year 2009 was almost as good as 2008 was bad, all because of the sheer weight of money pumped into markets by quantitative easing.

Part 2: Placing The 35 Years Into Historical Context

The Most Excessive Equity Bubble In Modern History

Preceding illustrations have shown most favourable readings for returns, and risk, or both, during the fifteen years from 1985 to 1999. Just how excessive and unsustainable markets were in the late 1990s can only be assessed by placing that time in a much wider, even multi-generational historic context.

History so far back is sparsely documented for most markets. Prof. Robert J. Shiller (of Yale University) has compiled monthly data for USA back to 1871, in preparation for his book ‚Irrational Exuberance’. That raw data is publicly available, making the slightly modified illustrations shown here possible. A link to Shiller’s database is given at the end of the document. Shiller calculates ‚CAPE’ using inflation-adjusted data. Here, nominal statistics have been used.

Chart-08 (A) shows 5-year normalised risk and return for the Standard & Poor’s 500 Index. Apart from the length of record, this is the same method of illustration applied in several of the Charts further above.

Normalised returns for 1995-99 tower far above all other extremes, even more extreme than the time before the infamous 1929 crash. Equally ‚off’ were the readings for normalised risk (a contrary indicator). It has never been as low, as in the five years 1995-99, a clear sign of an extremely overheating bull market. The notion, that such a boom could have been sustained, or would re-occur any time soon, is outright delusional. It goes against all evidence to the contrary. The collapse of global equity markets early in the 21st century was inevitable, and any half-reasonable analytical approach did show that. Chart-08 (A) is essentially a momentum analysis.

Another, this time fundamental approach, leads to exactly the same conclusion. Chart-08 (B) depicts P/E ratios on historic earnings. In the upper portion, a red line represents the S&P500 index divided by earnings over 120 months, the black line uses 60 months to calculate a p/e multiple. A thin, red horizontal line gives the median 10-year based multiple for the entire history (1880-2019), at 16.2x. If the period 1996 to 1999 were excluded, such a ‚cleansed median’ would show only 15.2 times earnings. The grey band is calculated using one standard deviation around a smoothed trail of the 5-year based p/e multiple.

The lower portion of the graph illustrates percentage over-, or undervaluation versus either the all time median value (in red), or the smoothed, more opportunistic/flexible 5-year trail.

The dominance of red bars, high above ‚neutral’ in that oscillator clearly shows just how excessively priced US equities were: 3x more overvalued than during any previous bubble in recorded history. Worse, by this measure, after 1990, they have never been less expensive than ‚neutral’, not even at the low point around the financial crisis. If the more opportunistic 5-year reference is used, then they have only been undervalued briefly, in 2003, and again in 2009.

The much lamented, ‚inexplicably poor’ performance of equities in the last 20 years is, if anything, somewhat better than fundamentals justify. It is only supported by an epic central bank intervention suspending the laws of gravity. In the course of 2018, even the opportunistic interpretation of this data has begun to flash warning signs. Arguably, a proper correction in the classic sense has not been allowed to happen.

A Secular Decline In Bond Yields

For bonds, the long-term backdrop is just as revealing. Chart-09 shows yields on US-Treasuries, 1875 to 2019. From under 2% in 1941, yields (broadly lagging CPI Inflation) rose to a peak of over 15% by 1981. By the late 1990s, yields had plenty of room to drop further, they were still at the upper end of the historic range. But ‚now’ at the end of 2019, some 40years after that all time high in yields, they stand near a record low (establishing new lows in 2020). Even with epic interventions on the part of the Federal Reserve, or the European Central Bank, to name but two, the process of falling yields must be in a final stage. At these levels, there is little to no income paying for the risk of holding bonds but prices will, at best remain stable, and with increasing probability fall, as eventually a new interest rate cycle will bring back higher rates.

There are similarities between bonds in 2019, and equities in 1999. While Charts-08 and 09 show US data only, they broadly mirror other developed countries. Prior to another long term ascend of yields, an oddity already seen in the Treasury bonds of Switzerland, Japan, and many Eurozone countries, may well occur in the US too: negative yields. That will not make the interest rate cycle any more immune to a long overdue and sustained reversal, nor will it bestow a halo of rationality on yet another, completely irrational excess, perhaps worse than the ‚Dot.Com“ bubble in equities 1995-1999.

Pension funds with exposure to bonds at negative yields manifest the irrational conviction that a situation that should not exist in the first place will, a.) continue for considerable length of time, and b.) become even more pronounced. With a typical mix, equities are held in weightings of 30% – 40%. Bond exposure is usually much higher than that. A fully blown bear market in bonds, almost guaranteed to happen, could hit pension funds harder than a typical bear market in equities. Bonds held at negative yields are an investment with a guaranteed loss. How any professional can argue that a guaranteed loss complies with the Swiss legal requirement for ‘Müdelsicherheit’ remains to be seen.

Part 3: Pension Fund Performance From 2000 To 2019

Against the background established above, it is now possible to evaluate how Swiss pension funds have fared in the environment described.  Chart-10 gives a first impression. Sadly, comprehensive performance data on Swiss pension funds only became available at the end of 1999, when Credit Suisse was the first custodian bank to calculate and publish aggregate pension fund indices. Several other have followed since. Unlike pension fund data for individual pension funds, aggregates such as the Credit Suisse Pension Fund Index, or the family of UBS Pension Fund Indices, are available easily, and with monthly frequency. By and large, these aggregates are highly representative of the Swiss pension fund industry, and/or specific sectors thereof.

Throughout this report, the proxy used for Swiss pension funds is a composite index, calculated from the Credit Suisse Pension Fund Index and the UBS Pension Fund Index Family. The rationale for calculating derivative indices (LPP-C types) has been already been explained. Too few observers consider the need to use contiguous indices and are thus unaware of their performance details. As some other illustrations have shown both groups of benchmarks, Chart-10 plots Pictet-2000-40 alongside it’s contiguous equivalent, LPP-C40. Even at such a rudimentary level as a simple plot of values over time, the performance of pension funds in comparison to benchmarks (proxies for unselected investment universes) looks disappointing. Where there should a noticeable excess over ‚unselected’, there is a considerable, and steadily widening negative differential. Whatever pension funds pay in investment fees to their professionally ‚selected’ managers, is clearly not justified as the long-term results are worse than random.

Table-03 gives profiles of monthly rates of change, including Swiss pension fund performance. Across the twenty years since the end of 1999, the lowest risk benchmark has seen the highest increase. Performance of the upper risk tier has been weakest, although the return difference to the other tiers is not particularly pronounced. Clearly, during these twenty years, running uncontrolled risk has not been rewarded, and pension fund performance reflects that.

Pension fund performance is below any of the three benchmarks. Expressed as  difference in annualised returns, pension funds lag the medium risk benchmark by 0.4% (3.1% versus 3.5% p.a.). Some will argue that this is due to management costs. Only at first this may seem a plausible explanation.

The truth is probably more complex, and worse. Given the distortions that arise when comparing static allocation (i.e.: synthetic) benchmarks with ‚natural’ portfolios, even an otherwise exact replica would appear to generate superior performance, provided the cloning portfolio would re-balance it’s asset mix less frequently than the benchmark index it clones. Estimates of the magnitude of this distortion indicate the range from 0.25% to 0.50% per annum. The effect is to understate benchmark performance.

Many a pension fund is allegedly run by active managers. Yet, countless minor and temporary allegedly ‚tactical’ deviations from a chosen benchmark are random. If tactical deviations were based on functional methodologies, then,  no matter how slowly, over time they would generate a positive performance differential. Clearly, that is not the case. They serve only to give superficial observers the false impression that management is ‚active’, research-based, and worthy of a fee. Pension funds could easily clone benchmarks far cheaper, in-house, circumventing fees altogether.

Chart-11 shows 20 years of Swiss pension fund performance, relative to the medium risk tier benchmark LPP-C40. Pensions funds in aggregate fail to generate performance over and above such a benchmark. Up and until 2012, there were periods of pronounced differences, both better and worse. But any relative gains, even where achieved, were always lost again fairly soon. In the most recent 12 years, since the ‚financial crisis’, there has not been a single meaningful period during which pension funds have managed to outperform. The graph suggests that relative performance is deteriorating.

Benchmarks are subjective representations of a particular investment universe, free from attempts to generate value. There is neither a need, nor an obligation to mirror the benchmark composition, except that legally set ceilings must obviously be complied with. The regulator does not stipulate minimum exposure to risky assets. Benchmarks are ‚what-if’ illustrations and should be treated as such. They are not the ‚gold-standard’ for investment acumen.

Pension funds enjoy a great deal of investment discretion. Since a synthetic benchmark is free from risk-management, to mirror such a benchmark means to forfeit the control of risk, and with that, the optimisation of risk and return. Either pension funds will use risk-management to comply with their obligation for risk-aversion, or they can mirror benchmark indices. The two ‚philosophies’ are mutually exclusive. One deals with risk to asset value, the other with decision risk only. The public is made to believe that pension funds deploy elaborate manager selection processes, in order to find the most competent investment managers for their assets. That only puts the performance data into an even worse light. If the selection processes is functioning, then how can the result be worse than random?

Chart-12 positions the aggregate Swiss pension fund data in a grid of risk (horizontal) and return (vertical), together with the three risk-tier benchmarks. The date reflects normalised values across the twenty years for which pension fund data is available. The dashed diagonal line shows risk/reward equilibrium.

If LPP-C40 or LPP-C60 were used as reference, then pension funds, in aggregate, appear to have achieved a risk reduction (see Table-03, above) and that may be used to legitimise a return deficiency. But the seeming risk advantage is so small that it is likely due to the benchmark effect alone. Once more. The effect understates benchmark gains, and overstates benchmark losses.

Higher unmanaged risk was not rewarded in this period. Even if it is, then never for long. The supposed exception is indeed the rule. Markets behaved in line with their excessive valuation, the sky is NOT the limit, as demonstrated in Chart-08 (B).

Pension fund love to argue, that lower returns (generally, or relative to indices) are legitimised by an alleged policy of risk aversion. As the chart shows, this is plainly wrong. The facts show the opposite. Across the time for which data is available, lower unmanaged risk has, on balance, generated higher returns than higher unmanaged risk has. That alone uncovers the insincerity of that defence. Seen from a ‚bird’s eye’ perspective, it is most likely that pension funds have earned lower returns on equal, or higher amounts of risk, than the universe represented by LPP-C25. No matter how one wants to look at it, the blue sphere in Chart-12 (representing pension funds) should either be further to the left (less risk), or higher up (more return). Considering that the investment climate shifted several time during the period lumped together in Chart-12, the full explanation of poor performance is probably more complex still.

Chart-13 shows yearly pension fund performance since the end of 1999, together with LPP-C40. A black line plots the accruing differential between pension funds and that index. Below the actual chart, a table supplements that graphic display with a choice of quantitative and qualitative metrics.

In general, and if we discard fractional differentials (2004, 2015) the data shows that pension funds typically outperform bear markets and underperform bull markets. Markets tend to rise more often than fall. If one were truly long-term minded, this is the opposite pattern to the ‚winning formula’ that may bring about the desired outcome.

Even if pension funds suffer less in bear markets, they clearly have losses far too great, meaning more than is justifiable, and more than what they can afford: It is counterproductive having to reduce risk exactly when risk has turned cheap again, just because one can no longer afford to run rewarding risk. In part, this pattern reflects the benchmark effect, but not in full. Conceivably, pension funds often do the opposite of what would be wise.

It is perfectly legitimate for pension funds to underperform a mature bull market. Their risk aversion should cause them to raise cash, as and when excesses become pronounced. But likewise, they should accrue a surplus over markets with every cycle, because they are able to re-deploy liquidity near market bottoms, investing in ‚cheap’ risks. Instead, there is a lingering suspicion, that pension funds tend to do whatever would have been right during the preceding years, holding on to expensive risks without ever getting into proper sync. The data suggests a severe lack of functional decision making methodologies.

Chart-14 shows the ‚drawdowns’ from all time high. It is only natural, that a fund’s value drops below such a most demanding reference. The blue horizontal lines represent the normalised annual risk for the three benchmarks, while red columns depict loss from peak value. If the choice of benchmark were to serve as proxy for risk-tolerance, then this chart demonstrates that pension funds fail to act, even when markets enter a dramatic shift of direction. Forward looking risk is equal to the difference price and value. That is the risk that matters most at critical junctures.

If no preventive action (raising liquidity) has been taken before, then a pre-determined level of risk (value erosion) must trigger swift action to contain loss within affordable limits. In the chart, such conceivable limits are represented by the normalised risk of the three benchmarks.

Regularly, pension funds drawdown is far worse than any common sense degree of tolerance. Chart-14 gives overwhelming evidence of a near total absence of risk control and risk limitation The Chart may not be able to explain why that is so, but it clearly illustrates the fact itself.

This is data on balanced, multi-asset portfolios.  Imagine the virtual price implosions at stock level this contains. When their total value drops by 15%, while the largest asset class (e.g. bonds) remains stable in value, or even increases, then another portion (equities) with 40% exposure, must have dropped far more in value worse. Again, readers are encouraged to consider Chart-08, and 09. It is legitimate to remain invested, preferably less so as an overvaluation runs wild. But once the music stops, swift action is crucial.

Table-04 shows values for r-squared, a statistical measure for the congruency between two sets of data. Rates of change of Swiss pension funds are always the y-value (‚dependent’) with every benchmark index as x-value (‚independent’). R-squared has a range between 0% (suggesting no dependence of ‚y’ from ‚x’) and 100% (y being fully dependent on x). In Table-04, the highest dependence is shown in bold red, the lowest in bold black. The full history is shown with three times, with distinct calculations using months, quarters, and years. Additionally, the most recent five years (60 months) and most recent ten years (120 months) are shown.

Lastly, the ratio between most recent five years and full history is shown. Ratios below 1.0 suggest that recently dependence has fallen, above 1.0 indicates dependency has risen.

In spite of minor differences between the two index families, Table-04 speaks loudly and clearly. With so readings generally as high as they are, benchmark-cloning, and/or pseudo-active management (closet-indexing) must be assumed to be the dominant religion among pension funds.

Indeed, the comparison between monthly, quarterly, and yearly data hints at ‚pseudo-active’ management: tactical deviations cause lower congruency across monthly increments.

Regardless of the index family used, the correlation with the higher risk tier is always considerably greater than that with the lower risk tier. One can also see that the original three tier benchmarks (Pictet-2000) gradually become much less relevant. Correlations to them are declining. But correlations with the contiguous benchmarks show clearly that risk appetite among pension funds appears to have grown, just as the market cycle has matured, as evident from the 60/240 month ratios. The increased market risk (late cycle) is once more unmanaged, exposing pension fund asset values to substantial drawdowns, in excess of what they can afford.

In Chart-15, Swiss pension funds congruency with LPP-C40 is shown as R-squared values for monthly rates of return. Bins of five years are used to illustrate how benchmark congruency has evolved. Due to a lack of public data, it is impossible to track benchmark congruency prior to the year 2000. Interesting is the temporary drop in congruency in the period 2005-2009.

So soon after investment guidelines were liberalised, a value of 96% looks surprisingly high for 2000-04. Pension funds must have been eager to embark on what was the winning approach from 1985 to 1999. But then the bubble burst set in, caused by unbelievably excessive valuations, but triggered by the 9/11 attacks. Pension funds were caught on the wrong foot at the wrong time. Most probably, they attempted to reverse direction (again) either permitting equity weightings to drop simply by value erosion, and/or by actively selling the stocks they had aggressively bought shortly before. R-squared dropped from 96% to 94.5%.

Since then, the congruency with this, and similar benchmarks, has steadily moved higher, suggesting increasingly ‚passive approaches’ used in Swiss pension fund investing. One can literally feel in the data how the paramount concern has become to minimise decision risk, and seek shelter, and salvation, in the ever closer shadowing of benchmark indices. As a result, r-squared stands at a record 98.2% for the five years 2015-19. With pension funds’ fortunes quite literally tied to the fate of financial markets, and no-one left, willing to take genuine investment decisions, the next bear market will hit pension funds every bit as hard as all previous ones have, perhaps worse. There is one fairly reliable constant in financial markets: over-priced assets will decline in value, undervalued ones will provide opportunities. Risk monitoring and risk management are the most vital parts of professional investing. But such extremely levels of correlation leave no room for risk management.

Ignoring the changeable nature of risk has lethal consequences. Forward looking risk can only be quantified by using a sound methodology. A cynic might say that forward looking risk is always equal to 100% of invested capital. A more practical definition is to see investment risk as equal, or at least similar, to the gap between price and value. Then there is ex-post risk, which seems easier to describe because it can calculated, seemingly exactly. But only the portion of risk that has actually materialised is measured. Ex-post risk is mainly used to judge ascertain quality of returns: between two equal rates of return, the one achieved with less risk is of higher quality.

Chart-16 illustrates the changeable nature of ex-post risk. It plots observed risk (a fairly stringent definition of risk) for the three risk-tier benchmarks. The plot is shown in ‚windows’, bands of 5% each, representing choices of risk tolerance. Due to the fluctuating nature of risk, risk of any preferred universe (benchmark) can transgress into either a higher, or lower band than what an investors has opted to tolerate. Obviously, one may choose to set different limits for such bands. But their purpose is to illustrate how risk-management is incomplete when all it involves is the choice of a benchmark.

But even ex-post risk metrics can give clues to future market direction. Then they  are contrary indicators. High readings describe excessive pessimism and low readings a bubble. Consider the exceedingly low levels of ‚high risk’ (red line) in 1999, 2007, and 2016-2018. Each time, a massive market correction set in. The reverse is also true. By 2002-2003 risk had risen to extremely high levels (describing the 60 months before). Markets were oversold, or cheap. By the end of 2019 risks seem dangerously low, suggesting excessive optimism.

Returning to the topic of risk management, or rather, the absence thereof, Chart-17 plots drawdown for each of the LPP-C benchmarks. Unlike Chart-14 which displays pension funds’ drawdown, here the same technique is applied to the various investment universes. To accommodate the rather different amplitudes, in each graph the value scale has been set individually. But the normalised risk for LPP-C60 (the high risk benchmark) is drawn into all three graphs, having a value of 7.9%.

Not only does the drawdown for the higher risk benchmark travel extensively far below its own normalised level of risk. The statement even applies for the medium, and lower risk universe. In fact the cyclical variation of manifest risk (which is really what ‚drawdown’ means), is so great, that  the benchmarks with supposedly less risk not only exceed their own typical level, but even that of the highest risk benchmark.

This is yet another illustration of the dangers inherent in a paradigm that assumes with the choice of a (any) benchmark, the obligation to manage investment risk has been fulfilled. This is a most expensive error.

It has brought the entire Swiss pension fund industry within inches of a collective  bankruptcy twice, just in the most recent 20 years, not something any one would have predicted, considering Switzerland’s reputation for safety and diligence.

Indeed if such a worrying development was possible, and has even become the norm, then it only serves to show just how far the pension fund industry has moved away from ‚Mündelsicherheit’.

Part 4: What Professional Investment Management Should Deliver

Deploying Benchmarks To Articulate Legitimate Performance Expectations

Benchmark indices, in particular ‚synthetic’ ones with a static asset allocation, are no reference of quality. They are neither more, nor less than a simplified description of what happens within a particular investment universe.

Professional investment management must justify its existence, contracted services, and associated costs, by adding value, over and above ‚windfall’. Such value-added may be achieved trough choices: discriminating against risk, or towards return, or a sensible combination of both. That is not reflected in a benchmark index, nor can it be.

Formulating Legitimate Expectations

To calculate thresholds of fair, in a given environment, achievable levels of value-added through investment management, the annual rates of change of an applicable index can be systematically modified, by using separate factors for gains and losses. Chart-18 does just that, using a constant differential (0.4x) across different types and levels of value-creation. Level ‚A’ determines an expectation where emphasis is placed on risk reduction alone. Level ‚E’ shows a conceivable target for a highly risk tolerant investor. Given that investment managers are free to use all stages of a selection process, subject only to specific ceilings, the factors, and the 0.4 gap used for gains and losses, are perfectly reasonable. This is what professionals must be able to deliver.

Considering how poor actual pension fund performance looks against any unselected investment universe (as represented by benchmarks), it is hardly surprising that the industry seeks to deflect any genuine calls for quality and ‘value-added’, as shown in Chart-18, or Table-05 above.

Yet, such demands are quite moderate, considering the impact of even basic forms of risk management and the deployment of one, or several functioning asset allocation methodologies. That is amply illustrated by the long term performance record of many genuine investment professionals worldwide.  It is also made tangible by the illustrations found on this site: Improving Returns Through Asset Allocation.

The trouble is not that adding value is not possible, because it is. The real problem is that to add value over benchmarks, one must differ from them.

Clearly, where it comes to Swiss pension fund management, there is massive scope for improvement, once a culture of legitimising lethargic investment behaviour can be addressed.

Contact Author

Related Pages

Improving Returns Trough Asset Allocation

Performance of Swiss Pension Fund Aggregates