How does minimum variance perform without "Big3"?

Three large companies dominate the Swiss equity market: Nestlé, Novartis and Roche. In terms of market capitalization, the "Big3" account for around 50% of the total market over the past 15 years. This concentration is not without its problems and raises a number of questions, particularly with regard to diversification. Nevertheless, for many investors the size of the three stocks alone gives them a special status and systematic exclusion seems almost unthinkable for most. But what does it really look like? How dependent is a portfolio on the "Big3"? How big is their contribution to the risk of the portfolio? What impact does exclusion have on portfolio performance? And last but not least: Is it permissible to go so far as to completely dispense with these three companies?

Below we show how a systematic exclusion of the three large SPI stocks affects the OLZ Minimum Variance portfolio. We calculate two variants of our Swiss equity model: one based on the normal SPI universe ("OLZ Eq CH Opt") and the other excluding Nestlé, Novartis and Roche ("OLZ Eq CH Opt excl Big3").[1]  The benchmark is both the SPI and a rescaled variation excluding the three large stocks ("SPI excl Big3").[2]  The results for the period 2003 to 2019 are shown in the following two tables[3]:

The following observations can be derived from this:

• Even excluding the three major stocks, the OLZ Minimum Variance Strategy clearly outperforms the SPI.
• The OLZ optimization is significantly less dependent on the three large stocks than the index. This is reflected in the smaller yield differences of the OLZ strategies with and without exclusion.
• Surprisingly, the exclusion of the "Big3" also has only a small impact on the SPI over the long term ("SPI excl Big3"). This is even more astonishing as these together account for more than half of the market capitalization in the index.
• The exclusion of the "Big3" in both the OLZ strategy and the SPI leads to a higher standard deviation and a significantly larger maximum drawdown.
• The exclusion of the "Big3" leads to a slightly higher turnover in the OLZ portfolio.

The shown simulations were created according to all rules of the art: they use only the information available at the time (out-of-sample) and consider realistic costs. Nevertheless, we would like to check the robustness of the results shown by comparing them with the track record actually achieved by our "Equity Switzerland Optimized ESG" fund.[4] For this purpose, we analyze the portfolios separately for the period 2011 - 2019. The results can be found in the following two charts:

As expected, the results hardly differ in comparison with the longer period. The following points can also be noted:

• The exclusion of the three large stocks changes the risk-return profile of the SPI significantly more compared to the minimum variance portfolios.
• For this subperiod, too, the risk-based OLZ strategies with lower risk outperform the capital-weighted benchmarks by a significant margin (after costs).
• The live track record of our fund differs only slightly from the simulations. This, too, outperformed significantly for the period under review and also boasted the lowest maximum drawdown.

In summary, the questions asked at the beginning can be answered as follows: Excluding the "Big3" from the SPI universe does not dramatically affect the risk-return characteristics of the portfolios examined. In particular, the risk-optimized OLZ strategy is largely resistant to the exclusion of individual stocks and is able to substitute the affected stocks with alternatives with similar risk characteristics and thus absorb the restrictions on degrees of freedom.

[1] In contrast to OLZ's usual practice, we deliberately avoid ESG restrictions in both models, as their application would be ruled out by Novartis from the outset due to controversies.

[2] The two OLZ portfolios are based on our minimum variance approach and are regrouped quarterly. "SPI excl Big3" is a self-calculated index with the same rebalancing frequency as the two OLZ portfolios. At the respective times, we generate an index portfolio with the capital weighting of the SPI, but exclude the "Big3" and rescale the weights.

[3] The simulation runs from June 2003 to March 2019. For the OLZ portfolios, annual costs of 70 basis points and transaction costs of 25 basis points on the corresponding turnover were taken into account.

[4] The returns of the "Equity Switzerland Optimized ESG" fund are after deduction of all costs (TER). Since April 2018, we have excluded Novartis from the investment universe due to ESG controversies.