“It was the best of times, it was the worst of times, it was the age of wisdom, it was the age of foolishness, it was the epoch of belief, it was the epoch of incredulity, it was the season of Light, it was the season of Darkness, it was the spring of hope, it was the winter of despair, we had everything before us, we had nothing before us, we were all going direct to Heaven, we were all going direct the other way”
A Tale of Two Cities (1859)
This excerpt from Dickens probably is as good as any other to describe the different states of mind of investment managers during the past five years. In this fifth anniversary
issue of IPE we look back at the
investment managers’ expectations since the inception of the magazine in February 1997. At that time 49 investment managers submitted their six– 12-month views on the following 12 areas: US equities, euro-zone equities, Japanese equities, Asian equities, dollar-denominated bonds, euro-denominated bonds, yen-denominated bonds, sterling-denominated bonds, dollar/euro exchange rate, dollar/yen exchange rate and dollar/sterling exchange rate.
Figure 1 shows a level of participation of more than 100 managers well within the first year of publication. This number peaked at 111 in September 2000. Thereafter – mainly as a result of consolidations and acquisitions – a steady group of around 103 investment management firms from 14 countries participated in the monthly surveys.
What information can be derived from the 57 months of available IPE data (February 1997–October 2001)? A long list of topics comes to mind, for example:
q Does the number of expected rises and falls contain relevant information?
q On which horizon?
q What is the lead/lag structure?
q What are the signal strengths through time per indicator?
q What effect do major market events have on a manager’s behaviour?
The most interesting question is whether somewhere beneath the surface there was a subset of managers that showed significant skill in one or more areas. We analysed the available data from different angles. Here we present the outcome of a so-called value added analysis with the purpose of looking into the characteristics of the monthly contributions to the IPE Investment Managers’ Expectations Indicator when translated into a balanced portfolio for each contributor (see box).
Within the value added analysis, tactical asset allocation and geographical decisions are integrated. Because of the outperformance of equities vis-à-vis bonds in the first years of the sample, the tactical asset decision obviously surfaces as the most important decision.
The main finding is that most managers do not add value relative to the strategic benchmark. Figure 2 shows that the average manager attained a cumulative return of 51.5% over the total period from February 1997 until October 2001. This compares with a benchmark return of 52.3%, resulting in an average negative cumulative return of –0.7% for the group as a whole, a very small negative number a year. With an average realised annual tracking error of 0.8% this number translates into a negligible negative information ratio for the group as a whole.
The dispersion among the outcomes for the individual managers over time can also be seen in figure 2. The best manager in the group outperforms by some 10% whereas the worst underperforms by approximately the same percentage. More than half (54%) of the managers in the survey did not add value to the strategic benchmark over the sample period.
Looking at individual years, we can see that the best year was 1999 with a small outperformance (+0.1%) and the worst year was 2001, with an average underperformance of –0.4%. The year 2001 is an interesting case. Around year-end 2000, after a bad last quarter on world stock markets, a significant upgrading of the expectations for equities among managers was visible. The average relative weighting during 2001 was about +5%, as compared to an average of +2% in the preceding years. Apparently, as a group, managers expected a reversion of the negative results for the fourth quarter of 2000 results in 2001. As we all know, the reality of 2001 turned out to be different.
Top quartile managers bore the most risk. Figure 3 shows the relative results for four manager-quartiles. These quartiles were formed by ranking the cumulative results of all the managers.
The best quartile showed a cumulative outperformance of 2.8% over the sample period. The primary driver for this relative performance was the asset mix decision. On average, the overweight position in equities in this quartile was +3% – which equates to 30% of the maximum overweight position – accompanied by a compensating negative cash weighting and an almost neutral bond weighting. Naturally, the average overweight in equities comes with a higher average risk (10.0%) than that of the benchmark (9.5%). Per unit of risk, the return of the top quartile is fully comparable with the benchmark return. At the same time, the outperformance in the top quartile is realised with only a slightly higher tracking error (0.9%) than the average (0.8%). The information ratio for the top quartile comes out as +0.7.
Opposed to the winners, the losers in the fourth quartile had a negative relative weight in equities. Relative to the benchmark, on average they lost a cumulative –4.7%. With a tracking error of 1.0%, this translates into a realised information ratio of –1. The second quartile still ends with a slight positive, while the third quartile ends at –1.5%. Figure 4 shows the distribution of performances relative to the strategic benchmark over the total period.
Do winners remain winners over time and what happens to the losers?
In other words, can one predict which managers will outperform or identify persistent losers in order to avoid them? To analyse this ‘persistence’, all our results are sorted into four quartiles for each individual year – resulting in quartiles for each of the years 1997–2001. Thereafter, the performance of the different quartiles in the years after the formation period are calculated. The persistence of relative returns was high in the years 1997–99. Winners remained winners and continued to generate significant positive relative returns in the following years. Also, losers remained losers in this period. The persistence in these years is much higher than typically would be expected. In 2000 and 2001, the persistence fell apart, as is illustrated in figure 5. This graph follows the winners and losers from 1997 onwards (that is, the percentage of top quartile managers from 1997 that remain ‘top quartile’ in the following year, etc).
A partial explanation for the persistence lies in the stability of the relative weightings in equities in the once formed quartiles: in the best quartile, there is a high relative equity weighting, in the worst case there is a low relative equity weighting.
Figure 6 is another way of illustrating the higher-than-expected stability of the membership of the quartiles by showing which percentage of the members of the top quartile formed in each of the years would still be a member of the top quartile formed in the other years. For example, in 1998, 66% of the members of the first quartile of 1997 again were in quartile one. If there was no persistence at all, the expected percentage would be 25%.
The analysis shows that only two managers added value consistently in each year. At the other end of the spectrum there were six managers who consistently lost money each year vis-à-vis the benchmark.
The presented results give rise to a number of additional questions. Clearly, the persistence in the results is higher than expected and may require further analysis. In future issues we will zoom in on topics such as those alluded to at the beginning of this article. Furthermore, we will look into other aspects of the data such as the effect major market events such as 11 September, the NASDAQ collapse, LTCM and the Russian crisis have had on managers’ expectations.
Diederik van Buuren and Jaap van Dam are partners at VermogensGroep in Amsterdam
The value added analysis
The goal of the analysis is to ascertain in which way investment managers as a group, as a subset and individually have performed by way of their recommendations (expectations). By offsetting their performance with an assumed benchmark, VermogensGroep is able to determine the value added over time.
The following methodology was used for the analysis:
q The point of departure is a static balanced portfolio including cash with monthly rebalancing with asset class weights that correspond with those of the MSCI World Equity index and the JP Morgan Global Government Bond index.
q The results of the allocation strategy are calculated for each manager. Every month the manager’s expectation is translated into an active weight relative to the above portfolio.
q A positive expectation means a positive active weight and vice versa for negative expectations. If an expectation for a specific category is unavailable (not submitted) a neutral weighting is assumed.
q Consecutive positive or negative months mean the positive weight will grow or decline in increments of 5% to a maximum or minimum that is 25% above or below the category weight. If the total amount of positive expectations outweighs the total amount of negative expectations, the balance is drawn from cash.
q A manager had to have submitted at least 12 months of data to be included in the analysis.
Performance is calculated in euros.