Big Data and Impact Investing

By Billie Trinder

The financial industry is into both big data and impact investing. The enormous volume of big data created every day, and associated innovations in analytics are revolutionizing effectiveness and measurement of impact. Crucially, big data can be used to more accurately and comprehensively assess companies’ social and environmental impact. This allows investors to be confident about the actual effects their investments are having, and consider non-financial factors without necessarily compromising on returns. At present, the contribution of big data is limited for reasons like a lack of data, and concerns about privacy and correct data use. 

Proponents of impact investing argue it is a win-win investment style, offering both return on investment and generating beneficial environmental or social impact. Some studies have shown that impact can be achieved without compromising on returns. In fact, though they may deliver lower returns during risk-off periods, impact investments may exceed expectations overall due to resilience to downturns (BlackRock 2018, p2). Over time, compounding environmental risks and attraction of greater capital to impact investing could amplify this effect. Through continuing failure to consider social and environmental factors, firms may expose themselves to risks they have a fiduciary responsibility to disclose (Sullivan et al 2015, p43). 

While the idea of impact investing is not new, the sector has surged in popularity in the last few years. Most notably, this new interest is generated by millennial investors, an especially critical fact in light of the $59 trillion USD in assets expected to move from baby boomers to millennials by 2030 (Shirer and Demers 2018, p7). As of 2017, 86% of millennial investors identified themselves as being interested in impact investing, which was reflected in their being twice as likely to take social or environmental goals into account when making investment decisions (Morgan Stanley 2017). 

Despite enthusiasm on the part of wealth holders, actual movement of capital investment has been hampered by reluctance on the part of professional money managers. Of Ultra High Net Worth families (UHNW), 70% were in favour of investing with social and/or environmental factors being taken into consideration, while just 43% of their investment managers agreed (Shirer and Demers 2018, p2). The managers’ concerns reveal a common theme: the lack of data. In 2017, 64% of asset owners identified a “lack of robust data” as the greatest barrier to making impact investments a more significant part of their current portfolio (BNP Paribas 2017, p17).

Traditional Measurement and Assessment

Presently, methods of assessing companies’ impact are subject to significant limitations. First, convenience and cost drive information collection methods, leading to significant use of survey approaches. This data collection compromises the quality of the data collected, due to issues of subjectivity, bias and qualitative assessment. The limited scope of survey questions also means that assessment of impact investing criteria can be based on incomplete data, with significant detriment to the conclusions drawn from it. 

Second, understanding companies’ true impact is complicated by what Niki Gilbert, co-founder of investment firm Matarin Capital refers to as a “cacophony of measurement” (Fry 2018). A lack of industry standards for assessing social and environmental impact has led to disparate rating frameworks and data sets, rendering companies’ true impact opaque and making comparison difficult. In some cases, self-professed compliance with a rating agency’s criteria is enough for a company to be listed as being a sustainable investment. The resulting confusion is evident in extreme divergence between ratings as reported by different agencies. Analysis conducted at MIT Sloan demonstrates that where the ratings published by any two of the top five agencies are compared, there is likely to be a correlation of between just 10-15%. In about 5-10% of cases, a firm in the top 5% as according one agency is in the bottom 20% according to the other (2013). 

Big Data for Assessing and Measuring Impact

An opportunity arises with the colossal volume of data being constantly produced as exhaust of peoples’ engagement with the “Internet of Things”, including through banking, social media use and ICTs (information and communications technologies). Big data analytics can provide detailed and comprehensive information in real time, making accurate and specific assessment of an investment’s impact far more attainable. This data provision helps investors strike the balance between purpose and profit, and enables companies to better understand how they can most effectively work towards social and environmental goals. 

Social Physics 

The transformational impact of big data innovations for understanding the social landscape is evident in the emergence of what is heralded as a new science: “social physics”, which uses mathematical laws and big data sets to understand and predict the behaviour of human crowds (Pentland 2015). While impact assessment must look to understand social factors (which do not traditionally lend themselves to objective, quantitative assessment) social physics takes advantage of the scale and detail of available data to accurately map human behaviour on a massive scale. From this, it is possible to identify patterns and thus make inferences that would have been invisible to traditional modes of assessment. 

International organisations (IO) including the United Nations, recognize the potential social physics presents to development projects. IOs have worked on a number of development related research projects which employ big data. One such project, conducted in partnership with Consumer Insights company Crimson Hexagon employed data gleaned from Twitter use to detect crisis-related stress in Indonesia (UN Global Pulse 2011). In this instance, Crimson developed a taxonomy of stress-indicative keywords and used them to divide posts into categories indicating levels of concern such as “afford”, “expensive” and “substantiation” (p7). Relevant messages were then quantified and compared with official statistics and ‘significant events’, and a strong correlation between the number of tweets about rice prices and actual inflation was revealed, demonstrating the significant power of social media to provide real time information about the “immediate worries, fears and concerns of populations” (p1). Another project conducted by UN Global Pulse in partnership with SAS was able to provide a three-month early indicator of an unemployment spike in Ireland using data from social media conversations only (UN Global Pulse 2011, p5). 

Though more slowly, investment firms are also beginning to use big data analytics to assess impact. BlackRock, an investment management firm which considers environmental, social and governance factors in investments across all asset classes, is increasingly using big data analytics for screening (Fry 2018). The potential offered by big data analytics is also evident in other areas of finance. Microfinance has effectively used analytics to make information about social and financial impacts into investable insight for some time. Assessment of non-traditional social and financial information contributes feedback for improvement of lending models as well as risk (Byrd 2016). For example, data collected by microfinance non-profit Opportunity International in Uganda demonstrates that within three years, a $10,000 improvement loan for schools with 250 students will typically grow the student body by 22.5% (or an additional 54 students), and that schools with access to finance were able to improve literacy levels by 60% more than schools without access. The organisation was able to use information from collected data to adjust loan requirements, including when analytics revealed that schools operating for under two years were almost twice as likely to default on repayments. 

Barriers

While it seems the constant production of massive amounts of real time data, and associated innovation should be enough to encourage use of analytics for impact assessment, investors face challenges in this area. It is important to note big data does not equate to open data. Open data is free from copyright and can thus be shared in the public domain. This is not necessarily a feature of big data, which can be privately owned or have restrictions on availability (UN Global Pulse 2013 p2). The Case Foundation, which aims to increase information sharing to encourage greater capital investment in ESG, notes that basic information about companies is often missing from existing databases (Case 2017). One such example is a company’s “impact geography” – the location in which the company focusses its ESG efforts. Presently, only 58% of companies can be accounted for in this respect, which could have important implications where investors seek to foster change in specific regions. Other information such as company legal structures is notably unavailable. Better availability of this type of data requires changed practises on the part of companies, who should aim for greater transparency, and a culture change that encourages reporting. 

Where privacy or sensitivity of data is a concern, encryption is suggested. Though in the long run it will not likely be enough. While legislation struggles to catch up with individuals’ concerns about use of their data it is important that companies take the initiative. To prevent misuse of data without creating problems for important ESG analysis, it is suggested that the principles of Pentland’s OPAL (Open Algorithms) architecture is adopted (2018). OPAL requires that algorithms are sent to existing databases. The analysis is conducted behind a firewall and only the encrypted results are shared, helping to ensure the safety and integrity of datasets. The advent of Endor.com, a program based on Pentland’s social physics and OPAL framework, which allows analytics to be performed on encrypted data, means that this is already possible. 

In addition to information gaps from companies’ non-disclosure, the gap that exists between the information rich and poor can translate into data blind spots.  Some of the world’s most vulnerable people would be overlooked by studies (UN Secretary-General’s Independent Expert Advisory Group on the Data Revolution for Sustainable Development 2014, p4) and subsequently by companies, investors and rating agencies associated with the impact assessment process. While much of the concern surrounding the big data revolution centres on privacy, and too muchpersonal information being available to companies and other actors, “data poverty” (Walji 2015) must also be considered. It is therefore important to assess and recognise the completeness of data sets when actionable insight is being drawn from them. 

For example, mobile phone data obtained from telecommunications companies have been useful in tracking and predicting outbreaks of infectious diseases (Kangbai et al 2018, p19). One such example is where big data has been used to understand and predict the spread of Ebola. In fact, the 2014 West African outbreak was first picked up by open software tool HealthMap. However, although ICTs are currently prevalent enough even in the poorest communities in Africa that some insight from associated data, mobile phone ownership represents a skewed sample of affected populations. Subsequently, when using mobile phone data to understand the spread of Ebola in West Aftrica, certain groups including children (who represent the majority of at risk group) are likely to be underrepresented or completely excluded from the data set (Kangbai et al 2018, p20). In Sierra Leone women are 43% less likely to own a mobile phone. Therefore, when using telecommunications data to understand an outbreak, this could translate to what appears to be higher instances of sickness and death in men due to incompleteness of the data set (Kangbai 2016). In these cases it is especially important that issues with datasets are taken into account where insights are being drawn from them. 

Rating Framework Harmonisation 

Once data has actually been collected, the issue of metrics and frameworks for assessment comes to the fore. Currently, the Impact Investing and Reporting Standards (IRIS) catalogue as developed by GIIN (Global Impact Investing Network) is the most popular option amongst impact investors. 65% use the framework for assessment (Reisman et al 2018, p.390). IRIS aggregates metrics from impact investment, development and non-profit organisations, and harmonises impact measurement across the board. That IRIS encourages this input and incorporates such a large number of impact metrics is a strength. For this reason, investors are able to customise the assessment process, applying impact values and priorities as they fit, potentially encouraging investors with areas of specific social or environmental interest. For example, an investor with a particular interest in clean energy could use the metric to compare how many households had gained clean energy access as a result of the companies’ activities. If the investor was to prioritise rural access, she could identify investee companies primarily serving rural areas without prior access to electricity (Global Impact Investing Network 2019).  

While the ability to customise the specific areas of social or environmental impact which are assessed is a strength, the actual assessment process by which a company’s performance against these metrics is gauged focusses primarily on output rather than outcomes, whereby output refers to what is produced by a company, and outcomes refers to the difference this product makes. This in an approach that may oversimplify assessment of impact – social impact especially (Dodd 2017). For example, a company’s educational impact could be measured by the number of students who are enrolled in school as a result of said company’s activities. However this would not necessarily indicate an overall improvement if the school was overstretched as a result. Here, the measurement of students enrolled in school is a measurement of output. In addition to limiting the accuracy of the information provided to investors, focus on output rather than outcomes may also direct companies to focus on “meeting quotas” rather than the best way of working toward the social or environmental impact they seek. As more data becomes available into the future, employment of big data analytics may provide for a more thorough and sophisticated assessment process. To use the above hypothetical, if outcomes were the focus of assessment, educational impact could be measured in terms of actual ability. For example, social media posts could be analysed for signs of changes to literacy levels. 

Mainstreaming Impact Investing 

Big data analytics presents an opportunity for impact investment as it allows for new and greater understanding of social and environmental impact. There are of course, issues with quality and availability of information and its measurement. As Managing Director of Blackrock Heather Loomis Tighe emphasises, the currently available information and frameworks are not perfect, but they are fast improving. It’s important investors do not let “the perfect get in the way of the good” (Fry 2018). In the future, the good that can be done with big data will be even greater, as companies contribute to available information. It is vital that analysts recognise data set limitations and respect data integrity throughout collection and analysis. With these improvements, it is likely that impact will become a core aspect of mainstream investing. This transformation would likely have a profound effect in helping to mitigate climate change, foster human health and provide access to education amongst other things.

Bibliography

Aggregate Confusion(2013) MIT Sloan School of Management. Accessed: http://mitsloan.mit.edu/sustainability/aggregateconfusion

Case, J. (2017) Data: The Next Frontier for Impact Investing.Stanford Social Innovation Review. Accessed: https://ssir.org/articles/entry/data_the_next_frontier_for_impact_investing#

Dodd, R (2017) What to Know About Standardized Social Impact Metrics. Sopac. Accessed: https://www.sopact.com/perspectives/standardized-social-impact-metrics

Fry, E (2018) Does Impact Investing Have A Data Problem?

Global Impact Investing Network (2019) Getting Started with IRIS. Accessed: https://iris.thegiin.org/guide/getting-started-guide

Great Expectations for ESG – What’s next for asset owners and managers?(2017) BNP Paribas Accessed: https://bvai.de/fileadmin/PDFs/DE/Themenschwerpunkte/ESG_12_18_1__9/Survey_ESG_Great-Expectations_EN.pdf

Institute for Sustainable Investing. (2017) Millennials Drive Growth in Sustainable Investing. Morgan Stanley Accessed: https://www.morganstanley.com/ideas/sustainable-socially-responsible-investing-millennials-drive-growth

Kangbai, J (2016) Social network analysis and modelling of cellphone-based syndromic surveillance data for Ebola in Sierra Leone. Asian Pacific Journal of Tropical Medicine. 9(9). Accessed: https://www.sciencedirect.com/science/article/pii/S1995764516301420

Kangbai et al (2018) Tracking Ebola through cellphone, internet of things and Blockchain technology. Current Research in Integrated Medicine. 3(1). Accessed: https://www.pulsus.com/scholarly-articles/tracking-ebola-through-cellphone-internet-of-things-and-blockchain-technology.pdf

Pentland, A (2015) Social Physics: How Social Networks Can Make Us Smarter. Penguin: London. 

Pentland, A (2018) Big data meets the markets: The future of data-dominated financial systems. MIT Connection Science. 

Reisman, J (2018) Putting the “Impact” in Impact Investing: The Rising Demand for Data and Evidence of Social Outcomes. American Journal of Evaluation. 39/3. Accessed: https://journals-sagepub-com.ezproxy1.library.usyd.edu.au/doi/pdf/10.1177/1098214018779141

Shirer and Demers (2018) Data Driven Impact: A new era of impact measurement.Boundless Impact Investing and Distilled Analytics. Accessed: https://distilledanalytics.com/wp-content/uploads/2018/04/Data-Driven-Impact-Excerpt-April-2018.pdf

Sullivan et al (2015) Fiduciary Duty in the 21^stCentury. Accessed: https://www.unpri.org/download?ac=1378

United Nations Global Pulse (2011) Twitter and Perceptions of Crisis-Related Stress: Methodological White Paper. Accessed: https://www.unglobalpulse.org/projects/twitter-and-perceptions-crisis-related-stress