2  Introduction

2.1 Perimeter of the thesis

During my master course: Data Analytics for Politics, Society and Complex Organizations (DAPSCO [https://dapsco.unimi.it/]) at the University of Milan, I had the opportunity to undertake an internship at ESGeo [https://esgeo.eu/], a sustainability start-up that guides companies through their sustainability reporting process – from the collection and validation of non-financial data, to sustainability KPIs calculation and report creation. The ESGeo team developed a cloud software application designed to go beyond the traditional sustainability reporting approach while enabling the management of the whole value chain of Environmental, Social and Governance factors. This tool allows organizations to keep track of their sustainability identity by measuring their material ESG issues: keeping track of ESG KPIs, comparing their current ESG positioning with benchmarks and peers and finally disclosing their ESG achievements with external stakeholders through downloadable reports.

After six months of internship I have been proposed a full-time open-ended contract and I am currently working in the ESGeo functional team. During my time here, from the requests of certain partners and as a potential additional service for our clients, the need of a way to forecast a company’s ESG score rating to channel investments has emerged. ESG ratings, following the footsteps of the long tradition of financial ratings, evaluate the level of commitment of a company towards its Environmental, Social and Governance (ESG) issues and gives a score to how said company deals and manages the sustainability matters that are most relevant to its business. Both small and medium-sized enterprises and big enterprises disclose their non-financial reports and initiate the ESG rating procedure in order to be more attractive in the eyes of investors and as a commitment towards the Planet.

The ESG ratings matter, that come as a tool to identify virtuous ESG initiative for investment, opens an extremely wide chapter that is still being investigated both by private organizations and researchers. In this section of the thesis, we will introduce the topic concerning ESG ratings and investments in order to give a perimeter for the following analysis.

2.2 The ESG investing scenario

When talking about ESG investing, we refer to responsible and sustainable investments that take into account environmental, social governance matters to guide investors’ portfolio decisions. (Matos, 2020) Investors sensitive to this matter, avoid investments that are exposed to high ESG risks and try to divert their investments towards more ESG-friendily companies in order to generate positive impacts for society and the Planet.
In the last years, ESG investing is a growing sector in the overall capital market, according to the Global Sustainable Investment Alliance in 2019 more than US$30 trillion were managed according to sustainable criteria. Still, it is hard to really have proof of how and what ESG factors are implemented into investment decisions, in other words: what are the actual drivers for ESG investing? and how a company becomes a highly appetible ESG investment?

The main focus in ESG investing lies in the possibility of doing good in terms of being sustainable toward humans and nature while doing well in terms of enhancing financial returns. Economists that have studied this phenomenon, affirm that the starting point of this change of direction in investments is due to the awareness that climate change is a real threat that humanity cannot avoid and that this crisis is a long-term one.
Larry Frink, CEO of BlackRock, in a client letter (2020) affirms that “individuals, companies, investors, governments must prepare for a significant relocation of capital” underlining the fact that this shift must be a joint effort of society as a whole.
In line with Frink’s thought, also Christensen, Hail, and Leuz (2019), authors of a comprehensive literature review of accounting and finance academic work on ESG projects and investments disclosures, feel that the combination of private initiative (keeping track of ones ESG data and publicly disclosing them) and public enforcement (government policy tools supporting sustainable initiatives) is the strongest driver for ESG investing.

Although this seems the generally acknowledged position on the strongest drivers in ESG investing, two matters arise:

1. the regulatory environment that would set sustainability standards for private companies is not internationally settled nor standardized;

2. the quantity and quality of Corporate Social Responsibility data that would help identify ESG investment opportunities currently has many gaps.

We will discuss these two points in the following sections of the chapter.

2.2.1 1. Regulatory environment in ESG investing

Governments around the worlds have activated to find government policy tools such as taxes and subsidies to support sustainable initiatives while penalizing non sustainable behaviors but these tools have not been internationally standardized.
Different regions around the world are proceeding each at different pace on ESG regulation, says Matos in his critical review on “ESG and Responsible institutional investing around the world” (2020).

In 2017, the Japan’s Government Pension Investment Fund (GPIF), which is the largest pension fund, revised its investment principles in order take into account ESG parameters and make the capital markets more sustainable.

The Norges Bank Investment Management (NBIM, the institution managing Norway’s sovereign wealth fund), in 2012, started an investment transition revising sustainable goals of its portfolio firms.

The European Union ha currently the most ambitious agenda in terms of regulatory effort to support the transition towards a low-carbon economy. In 2018 the Action plan has been released and it comprised several policy initiatives oriented towards the shift of private capital flow into sustainable projects in line with the United Nations 2030 targets.
This includes:

- a taxonomy classification for sustainability activities, standards and labels to establish clearer ESG benchmarks for firms and products;

- the obligation for institutional investors to disclose the ESG factors that can be considered drivers for their investment choices.

For what concerns the United States, instead, the regulatory environment is still in the making and the inclusion of ESG factors as drivers in investment choices is in active debate.
ESG factor integration in the US mainly depends on whether it can be proved that they have a positive effect on the investment portfolio performance and therefore on the financial return for beneficiaries.  According to the Employee Retirement Income Security Act of 1974, investors should always put the economic return of stakeholders before the promotion of ESG goals, in particular in regard of pension plan investments chosen by the US Department of Labor for pension plan participants. A greater sensibility towards sustainable investment was promoted during the Obama administration in 2015 but this approach was again downgraded during Trump administration and economic interests again put in the foreground compared to ESG issues.

This misalignment between regulations on ESG matters around the world create a lack of universal standard on where to concentrate sustainability efforts for global companies. Still, Matos (2020) does not see in this an insurmountable problem: in a globalized world, having a region that is more ambitious in regulatory efforts means that it is most likely that it will affect and drive manager and investor choices in other regions towards the growth of sustainable investment.

2.2.2 2. Data in ESG Ratings

The aspect instead, that might be more problematic is how to assure investors that a company is sustainable. Since the regulatory alignment is not yet standardized at governmental level, private rating companies that were historically engaged in financial ratings, started to also offer sustainability ratings that should be applied globally.
Sustainability ratings or ESG ratings are a tool to evaluate the level of commitment of a company towards its Environmental, Social and Governance (ESG) issues and gives a score to how said company deals and manages the sustainability matters that are most relevant to its business. However there are many aspects that are still to be settled around the matter of ESG ratings that are becoming more and more the backbone for responsible investing.

The aspect that one should deal with when using ESG rating scores to drive investments, regards the quality of data. Surely, the last years have seen a rapid growth of availability of ESG data, however there is no assurance on the quality of these information. Firstly, data published in sustainability reports and disclosures are still difficult to compare and often turn out to be inconsistent across firms because of missing data, for example less availability of ESG information for companies operating in emerging markets. (Yang, 2019)
But even if we assume an empirical case in which there is no missing data: two companies operating in the same sector and region and that disclose the same ESG data in their non-financial annual report, another problem would arise. In fact, some ESG factors cannot be objectively measured but instead require subjective decisions, such as a tick-the-box approach from assessors. Think about the factor of environmental harm of energy sources, this aspect can be debated and described depending on the interpretation of the phenomenon and therefore on the quantitative weights chosen for the calculation of impacts.

Moreover, to complicate even more the matter, as shown by Gibson, Krueger, Riand, and Schmidt (2019), ESG ratings have an extremely high level of divergence, in fact, average correlation between overall ESG scores among different rating providers is less that 50%.
This divergence among rating scores is a symptom of different views of the concept of what it means for a company to be social responsible and therefore of which metrics are best to measure it.

2.2.3 What has been done and what will be done in this thesis

With such a high number of open points and vague interpretations of what it means to be sustainable, unsurprisingly, scholars such as Drempetic, in his paper named “The Influence of Firm Size on the ESG Score: Corporate Sustainability Ratings Under Review”, call for an urgent and global debate on “what sustainability rating agencies measure with ESG scores, what exactly needs to be measured, and if the sustainable finance community can reach their self-imposed objectives with this measurement”. (2020)

Starting from Drempetic statement, which condenses in itself all the main concerns regarding ESG investments and therefore the reliability of ESG rating scores that currently are the backbone of sustainable investments, this thesis will try to unpack an ESG rating to understand which are the most influential factors concurring to the company final ESG score.

On the matter, ESGeo has already conducted a preliminary internal analysis to forecast companies ESG score rating from a set of variables coming from an anonymous database of a non-specified ESG rating company. The approach used for this analysis was an Algorithmic Modeling one and it has pointed:

- Sector (in which the company operates);

- Geographical Area (in which the company operates);

- Revenues (of the company for 2018 and 2019)

as predicting factors for the final ESG score.

What I wanted to do was to run the same analysis with a different method being more focused on all the methodological steps used to conduct it. For this reason I chose a methodology that:

- on one hand is more similar to the classical statistical analysis (the Data Modeling approach previously described) where the black box, where the association between variables happen, is to be more specifically analyzed and discovered;

- while on the other hand, has a strong operational and contextual drive based on standardized algorithms and the appropriate software.

I choose the Qualitative Comparative Analysis, probably one of the best known set-theoretic methodology, being well prepared to the possibility of not achieving results because of the lack of a precise data-set codebook and of clear theories that could guide configurational choices.