Case study by Bank J. Safra Sarasin

The aim of our Sustainable Investing Laboratory is to exploit the opportunities offered by financial data science, the intersection of three fields: statistical analysis of big data, investment management and computational science. In contrast to financial economics approaches which presuppose which performance indicators are theoretically key and analyse only those, financial data science analyses all performance indicators to identify any financially material signals.

Given the topicality of smart beta and sustainable investing, our Sustainable Investing Laboratory explored the possibility of a Sustainable Smart Beta strategy. As environmental and social indicators are often industry-specific, we focused initially on governance indicators, exploring 96 indicators for financially material signals.

The process

Our data analysis process involves five phases:

  • Import the governance data plus the relevant financial market data and check data quality.
  • Define performance measures that describe the downside protection and upside opportunities we aim to identify.
  • Analyse the investment performance of tens of thousands of hypothetical portfolios of 30+ stocks.
  • Test robustness to ensure that a risk reduction opportunity does not come at the expense of a constrained return or other undesirable features.
  • Repeat the analysis with out-of-sample data to ensure that the governance indicators identified as financially material are broadly viable performance drivers.

Out of scope framework analysis

After importing and checking the data and defining the performance measures, in the third phase, the tens of thousands of annually updated portfolios of larger than 30 stocks are formed from a global investible equity universe according to the governance indicators over a recent sample period of up to eight years. Not all of these are practically implementable. Some of the appealing results might come at the cost of a constrained alpha; others might be affected by non-comparable risk levels due to differences in portfolio diversification despite these being not too substantial beyond 30 stocks. Hence the fourth stage involves a range of robustness tests for these risk reduction opportunities to check for potential upside constraints. As a result, we identify more than half a dozen robust, financially material investment signals based on governance indicators.

With the fifth and final phase, we use the last two years of data available as an out-of-sample period. We define a few portfolio construction strategies based on the half a dozen robust, financially material investment signals identified in the fourth stage, and then repeat the full data analysis process for these strategies in the out-of-sample period to understand the financial materiality implications.

As shown in Figure 1, the strategy with the best balance between firms classified as good governance (N=2149) and those classified as poor governance (N=2097) displays 75% winning months for the good firms between January 2013 and December 2014, and sees the good governance portfolio delivering hundreds of basis points value added. Due to the equivalent balance between firms classified as good and bad, and the jurisdiction-specific nature of corporate governance, the country exposure of the strategy is tilted towards the US, although its industry tilts are inconsequential.

The insights gained from the five stage process flow into our Sustainable Smart Beta strategy, affecting several steps in the investment process.

They are:

  • an integral part of our ESG analysis, where they are influential in deriving our sustainable investment universe by selecting and weighting the relevant aspects of our ESG assessment of companies;
  • factors in the portfolio construction process used to reduce governance risk and protect the downside portfolio risk of our smart beta strategy without limiting upside alpha opportunities.

In practice: examining executive compensation

One of the half a dozen financially material governance indicators is related to long-term incentives for CEOs. A company displays good governance if there are effective stock ownership guidelines for the CEO (in cases where a CEO’s ownership is worth more than five times the CEO’s annual pay). Testing the robustness of this factor globally, we find it to significantly reduce risk across all developed markets and two of three emerging markets (figure 2).


Semi-standard deviations (in bps) of investment strategies based on CEO equity policy across regions

Download the full report

  • Download report

    A practical guide to ESG integration for equity investing

    September 2016