Time horizons
To begin with the time horizon, we had to define whether the research was going to pick a specific ‘snapshot’
from the past or more of a diary or collection of snapshots, we chose to proceed with the ‘snapshot’ approach called ‘cross-sectional’ time horizon. Most research papers, like ours and like many other academic research-ers, have limiting resources and, more specifically, time-constrained. Thus, we have limited our time horizon to 12 years, starting from 2009 and ending in the first quarter of 2021. Furthermore, cross-sectional studies often seek to ‘explain how factors are related to in different organisations at given point in time. For example, our research looks at the relationship ofc companies’ attitude towards ESG aspects, delving into the value cre-ation from ESG while looking at the effect on M&A transactions. All of this in the given snapshot of time horizon, 2011 to 2021 (Saunders, Lewis and Thornhill, 2019).
As some countries didn’t have been given an ESG score for all years, then the first data cleaning method was to assume that if no ESG score was given in the previous years, then the company had the same score as the latest given ESG score. However, as further studying the ESG scores, it was clear that there was not any trend of ESG scores being that close between the years. This resulted in the decision that a company will first be a part of the portfolio the year it was first given an ESG score, so stock data or a non-ESG score would not af-fect the calculations for the yearly portfolios and regression.
Figure 5: ESG Data representation
(Authors own creation, 2021).
The table above shows the average score each year across all the countries compared to the number of compa-nies that are brought in the portfolio calculations. It shows that in 2017 many compacompa-nies receive their first ESG score, which brings the average score of all three pillars down. However, it is worth noticing that the trend of each score from 2011 to 2016 is slightly increasing, and from 2017 to 2019, it is slightly increasing.
One assumption that has been made with the ESG score by the authors is in the year 2019, as at the time of extracting this data, several companies hadn’t received an ESG score in 2019, the assumption made here is that if the company have not got a score for the year 2019, then the score is equal to the ESG score in 2018.
To estimate the combined ESG score then there has not been made any assumption about if one pillar should be more weighted as the other. Refinitiv has made their own ESG template where environmental is 40%
weighted and social, and governance is 30% weighted. In this case, all the pillars are equally weighted in the calculated combined ESG score as following:
Equation 27
(𝐸𝑛𝑣𝑖𝑟𝑜𝑛𝑚𝑒𝑛𝑡 𝑃𝑖𝑙𝑙𝑎𝑟 𝑆𝑐𝑜𝑟𝑒 + 𝑆𝑜𝑐𝑖𝑎𝑙 𝑃𝑖𝑙𝑙𝑎𝑟 𝑆𝑐𝑜𝑟𝑒 + 𝐺𝑜𝑣𝑒𝑟𝑛𝑎𝑛𝑐𝑒 𝑃𝑖𝑙𝑙𝑎𝑟 𝑆𝑐𝑜𝑟𝑒 )
3 = 𝐸𝑆𝐺 𝑆𝑐𝑜𝑟𝑒
Model factors
In terms of calculating the model factors, all historical stock returns were gathered from Yahoo Finance on a monthly basis. Hereafter, the monthly return (𝑟𝑖𝑚) for each company was calculated as the adjusted closing price (𝑃𝑖𝑚) subtracted with the adjusted closing price the previous month (𝑃𝑖𝑚−1) and then divided with the adjusted closing price from the previous month:
Equation 28
𝑟𝑖𝑚=(𝑃𝑖𝑚− 𝑃𝑖𝑚−1) 𝑃𝑖𝑚−1
All companies in each month are equally weighted in terms of calculating the average monthly return for each of the portfolios.
Values needed to generate the weighted portfolios was extracted from Standard & Poor’s Capital IQ. All the company’s market capitalization was extracted through their company-specific ID, which has the same func-tion as a company ticker on Standard & Poor’s Capital IQ platform. The extracted historical market capitaliza-tion was in DKK and was on a monthly basis, whereas the monthly median was calculated to weigh the com-panies in either “Big’ or “Small’ sized. The median (𝑥̃) was each month only calculated on those comcom-panies who had ESG or stock data available for the same months, and was calculated as followed:
Equation 29
𝐹𝑜𝑟 𝑠𝑖𝑧𝑒 𝑜𝑟𝑑𝑒𝑟𝑒𝑑 𝑀𝑎𝑟𝑘𝑒𝑡 𝐶𝑎𝑝𝑖𝑡𝑎𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛 𝑑𝑎𝑡𝑎 𝑙𝑖𝑠𝑡: {𝑥1, 𝑥2, … , 𝑥𝑛}:
𝑀𝑒𝑑𝑖𝑎𝑛 = 𝑥̃ = { 𝑥𝑛+1
2
𝑖𝑓 𝑛 𝑖𝑠 𝑜𝑑𝑑 𝑥𝑛
2+ 𝑥 𝑛
2+1
2 𝑖𝑓 𝑛 𝑖𝑠 𝑒𝑣𝑒𝑛
Each monthly median value decides how the company is weighted. If the company’s market capitalization is above the monthly median value, then the company is categorized as “Big’, and if their market capitalization is below the monthly median value, then they are categorized as “Small’.
There are several ways to calculate the book-to-market value of a company, either through the book value per share or multiplying the market capitalization with the price to equity ratio. To keep these variables most simi-lar, the authors decided to choose the value for each month through Standard & Poor’s Capital IQ platform as well, then the calculations for the book value and market value of equity is from the same platform. By divid-ing the book value of equity to the market capitalization for the same month, the 𝐵/𝑀𝑖𝑚 ratio can be derived:
Equation 30
𝐵/𝑀𝑖𝑚 = 𝐵𝑜𝑜𝑘 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝐸𝑞𝑢𝑖𝑡𝑦𝑖𝑚 𝑀𝑎𝑟𝑘𝑒𝑡 𝐶𝑎𝑝𝑖𝑡𝑎𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛𝑖𝑚
Regarding other financial information for each company then they are all extracted from the Standard &
Poor’s Capital IQ platform in DKK. For the supporting analysis, the costs of revenue, revenue, return on in-vested capital in percentage, and total enterprise value has been extracted for each company. To make the cost of revenue more comparable across the companies then a cost margin has been calculated as:
Equation 31
𝐶𝑜𝑠𝑡 𝑀𝑎𝑟𝑔𝑖𝑛𝑡 =𝐶𝑜𝑠𝑡 𝑜𝑓 𝑟𝑒𝑣𝑒𝑛𝑢𝑒𝑡 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑡
Where 𝑡 indicating the year of the financial information. In this part of the analysis, the data is gathered on a yearly basis as the financial information were not available monthly.
Benchmark Data
The index benchmark selected for this portfolio was the STOXX Europe 600 who represents small-, mid-, and large-capitalization companies across Europe. Choosing this index lies in the belief as it the best representative market when comparing data and the effect across the Nordic countries, as the laws and regulations lie within each other by the European Union.
Risk-free Rate
Being one of the key variables in CAPM and the other multi-factor models, then choosing the most relatable risk-free rate is crucial for the analysis and the results. Investors chose this to calculate the market risk pre-mium, which is the benchmark of investors willing to hold market risk (Berk & DeMarzo, 2020).
The risk-free rate corresponds to the risk-free rate that investors can borrow and save to. Generally, the yield of government bonds is being used as the risk-free rate. However, it is unusual in practice that investors can
borrow at the same rate as governments. In addition to this, it is important to determine the maturity, whereas, for short-term investments, it is usual to pick a short maturity and, for long term investments, a long-term ma-turity (Berk & DeMarzo, 2020).
There are two approaches to gather the risk-free rate. The fundamental approach by calculating itself by solv-ing for the discount rate that is consistent with the current level of the index. This approach is highly inaccu-rate for an individual firm, and by assuming constant growth then more suitable for the overall market. The second approach is by looking at the historical risk premium. The authors have chosen the historical approach, as this is a historical analysis and not trying to calculate future cash flows. Therefore, assuming these are long-term investments, the historical rate on the ten-year government bond for Denmark, Finland, Norway, and Sweden been located. This rate has been calculated to an average rate where each rate is equally weighted. To make the risk-free rate adjusted to months, it has been divided by 12 on each observation.
Equation 32
𝑟𝑓𝑚= 𝑟𝑓𝑡 12