posted on September 21, 2020 10:47
This month marks the 50th anniversary of the publication of Milton Friedman’s seminal article in the New York Times, “The Social Responsibility of Business Is to Increase Its Profits.” Therefore, I thought this week would be a good time to share the work that I have been working on with teams of students from my graduate-level business analytics class at The George Washington University.
Through our analysis of social impact and portfolio returns, we have found the premise of Friedman’s thesis is negated in that:
- Social impact can be measured through the utilization of big data analytics
- There need not be a tradeoff between investment returns and greater social impact
- Specifically, in the Healthcare Sector, higher social impact often leads to higher financial returns
Over the course of the past year, we have been applying the current evidence-based ESG concepts to the development of a new methodology that utilizes the combined power of Natural Language Processing (NLP) and Machine Learning (ML) to:
- Value firms by applying advanced ML techniques to fundamental financial analysis
- Quantify firms’ current and potential social impact.
We call this methodology “Impact Alpha”.
The Impact Alpha methodology was designed to empower investors and consumers to outline a series of social impact objectives and tailor their investment portfolio or corporate support accordingly. Given the general alignment of social impact and the healthcare sector, we decided to focus on developing an AI-based portfolio selection model to evaluate publicly traded healthcare firms selecting an analytically optimal portfolio of investments with both the highest estimated returns and social impact.
Although we decided to focus on the healthcare sector prior to the outbreak of COVID-19, as the pandemic spread and the performance of the global economy declined, it became increasingly obvious that this methodology could not only be utilized to select an investment portfolio but to surveil global big data to evaluate products and firms that will ultimately support “flattening the curve”.
Impact Alpha applies a score between -1 and 1 to every publicly-traded healthcare firm where the absolute value measures the level of the impact and a positive score implies a relative net benefit to society and a negative score implies relative harm to the society. Furthermore, each firm has an Impact Alpha score in each of the six categories:
- Saving Lives
- Curing & Treating Life Threatening Disease
- Improving the Lives of the Elderly
- Lowering Costs
- Enabling Improved Coordination
- Embracing Value-Based Care
These broad categories were borrowed from research compiled by EntryPoint Capital. More detail on these categories can be found in their white paper.
In backtests, our model calculated the Impact Alpha for each month between 2010 and 2020 with publicly available unstructured data available prior to each period, such as news stories and journal articles. At a high-level, the steps of the model include:
- News Scraping: Evaluating millions or articles and disambiguating and extracting firm name and selecting articles about US-based healthcare firms.
- News Classification: Classifying topics within the articles into one or more of the six social impact categories.
- Sentiment Score: Calculating the sentiment score for each topic identified within each article.
- Calculate Final Score: Users of the methodology can weight each of the six Impact Alpha categories for their personal objectives.
In comparing the Pre-COVID era with where we are now, the Impact Alpha model has largely reallocated the entire “portfolio” to the prevention, care, and treatment of COVID-19-related firms. For example:
- Doubled its allocation in drug companies, with a focus on the firms that have been in the news regarding vaccine developments
- Doubled both its COVID-related bio-medical and medical services allocations
- Removed its dental supplier and nursing home allocations
The purpose of this discussion is not to recommend specific stocks or sub-sectors but to discuss the concept of an innate alignment of social responsibility, corporate earnings, and financial returns. We are continuing to surveil and report on the data and refine the methodology. Please reach out if you are interested in discussing the results and/or contributing to the ideation.