The most vital part of this paper focuses on the statistical analysis that compares large capitalisation companies with mid- and small capitalisation companies from the perspectives of both HCP Quant and HCP Focus strategies. I have collected data from 97 the equities and outlined the country, continent, revenue and enterprise value of the company as well as ESG related matters from 2012 to 2014. The data is ordered by the turnover from the largest to the smallest, and contains existing and former holdings of HCP Quant and HCP Focus Funds. I have added to the data large and very large companies by their enterprise value to balance and support the basis for comparison. For business reasons, the sources are published only by the results (*esg-analysis*) instead of relation to the companies’ data.

**The size of the company and ESG**

Intuitively speaking the larger the company, the higher pressures it has for the ESG transparency. The analysis uses Bloomberg ESG Score as the main metric. This score measures the ESG issues on transparency scale [0.01, 100], in which different factors have different weighting depending for example, the industry. If no data is available from the company (the company did not publicise any of ESG issues, or it is not part of the ESG Group ), its value cannot be given zero in this analysis, but a N/A, and it will not be taken into account, for example, when calculating of averages. This is because as by the first observational notation in this analysis was that the smaller companies do not have data available nearly as readily as the larger companies, and the purpose is to examine specifically how much the ESG Score has value when data is available and use the availability of the data to support the analysis, as a separate measure. In this case, it is easier to draw conclusions related to the company’s size, ESG Score quality, the resources and business strategies. All values are as of the last financial year end, in this case, the last values of 2014 unless otherwise stated.

Whilst the absolute costs of business ESG reporting increase in tandem with the size of the company and turnover, relative costs will reduce to a small amount. Reporting can therefore be dependent on the well planned business strategy when it comes to really large companies and the availability of the data or ESG Score ‘s value. If we compare the basic statistical indicators; Eθ_{S10} = 16 , 78 and 50 % with no data available for both Eθ_{S50} = 19 , 55 and 26 % with no data available. I use θ-variable in the analysis to describe the sample ESG Score: the sub-indices, as well as notes:

- S10 = 10 smallest companies sample
- S50 = 50 smallest companies sample
- L10 = 10 largest companies sample
- L50 = 50 largest companies sample
- CASE = sample of Asian companies
- EUROPE = sample of European companies
- N.AMERICA = sample of North American companies
- E = Expectation, in this case, the sample mean
- σ = standard deviation
- δ = The company’s turnover

When we added the additional 40 companies for the analysis sample of smallest companies, the sample average rose slightly, but the share of companies that do not have data available dropped sharply. Let’s consider the results of the largest of companies. Eθ_{L10} = 46, 14 and 20% which do not have available data. Eθ_{L50} = 36, 75 and 6% which do not have data available.

**Picture: ESG Score median and standard deviation by the company size**

*(Source: esg-analysis)*

It appears that the company’s size matters regarding the ESG Score: the larger companies have markedly higher variable value average compared to the smallest companies when analysing this material, and the larger companies have much more data available than the smaller companies. The expected value of the whole data: EΘ = 28 , 69.

Comparison of the standard deviation of the variables

σ_{S10} = 8, 58

σ_{S50} = 11, 92

σ_{L10} = 13, 36

σ_{L50} = 18, 77

In inspection of the above results, we find that the standard deviation increases. This wide dispersion of differences suggest that when the company size increases, the transparency of ESG factors are increasingly a strategic issue rather a question about resources. For this, we would assume that the costs related to the larger companies ESG reporting are similar to each other. σ_{Θ} = 18 , 50. The standard deviations of the largest 50 companies and the entire population are closer to each other when comparing to the smallest 50 companies’ standard deviation. The larger companies have greater differences between the ESG Scores, whilst smaller companies have more evenly smaller scores.

To ensure that the ESG Score difference between the small and large companies is statistically significant (because the standard deviation of the entire population is quite large), a Student’s T-test of the null hypothesis is applied H0: E_{θL50} − E_{θS50} = 0, that is, assuming that variables’ expected values do not differ from each other. This assumes the hypothesis that company size does not affect the ESG Score value. If the null hypothesis is true, the sample averages for the deviation between large and small companies can be explained by other factors, for example, by dispersion. In this case, this material would not provide sufficient information on whether the company’s size would have importance on ESG Score’s value. T-test therefore is used to examine, whether the samples have statistically significant deviation in the sample averages. If it is not, then the differences between the material’s ESG can be due to the standard deviation etc. or less probable population sample. The null hypothesis is a concrete and formal assumption under which the test is operated.

At the first, the F-test is applied whether these two variable variances are equal. As expected, the result (F – value of 2.25 and F- criterion of 1.61) was that the null hypothesis (equal variances) are disregarded and instead Student’s T-test, with unequal variances, is applied.

The result of the T-test is that the null hypothesis is rejected. The difference between the samples’ 50 smallest and 50 largest companies ESG Scores is statistically significant at 5 % significance level: P-value = 0.00001. P-value therefore means that with a 0.001% probability the sample lays within the null hypothesis framework, i.e. that the real differences between the sample averages are zero (the company’s size is irrelevant). The probability of this is so small (less than 5 %) that we can reject the null hypothesis and conclude that the company’s size matters.

NOTE. As a rule in statistical tests, we use material which has more than 30 observations (θ_{ASIA} is the only sample with less than 30 observations), so it is logical to use tests without assuming that the variables would be approximately normally distributed.

Analysis of correlation coefficients

Corr(Θ,δ)=0,2112

Corr(Θ, γ) = 0, 2147

The correlation coefficient of the sample’s ESG Score and company’s turnover (Revenue) is almost equal to the correlation coefficient of the sample’s ESG Score and company’s value (Enterprise value). The company’s turnover and value are evidently related to each other, but are two separate measures of the company’s size, which is clearly positively correlated to the ESG Score value. The correlation is not very high due to the whole sample’s large standard deviation, nevertheless, a clear positive correlation exists.

**Regional differences**

Analysis of regional relevancy, in this in case, the continents. We test the hypotheses:

Ho :Eθ_{ASIA} − EΘ = 0

Ho :Eθ_{EUROPE} − EΘ = 0

Ho : Eθ_{N.AMERICA} − EΘ = 0

P-values of two tailed T-tests:

Asia = 0,06424 (df = 117)

Europa = 0,0453 (df = 129)

N.America = 0,49112 (df = 97)

We reject the null hypothesis in the case of European companies at 5% significance level. The Asian P-value is on the borderline, whilst North American companies comply with the test most clearest. The company’s continental location has therefore importance, but in respect of the Europe’s expected value of deviation of ESG Scores from the whole sample’s expected value is statistically significant at the 5 % significance level.

Comparison of the expected values

Eθ_{ASIA} = 24,19

Eθ_{EUROPE} = 39,74

Eθ_{N.AMERICA} = 22, 59

Asian companies have higher expected ESG Score compared to North American companies. Notable is that 41% of the Asian companies did not have data available, while the North American data was not available for only 8 % of companies. By comparing the variables, standard deviations, as well as the average Revenues:

σ_{ASIA} = 18,74

σ_{EUROPE} = 19,21

σ_{N.AMERICA} = 14,34

Eδ_{ASIA} = 127789167, 31

Eδ_{EUROPE} = 74522, 18

Eδ_{N.AMERICA} = 58232, 66,

We find that the Asian companies are significantly larger than the European or North American counterparties, and the standard deviation is higher with North American companies. Although the Asian companies’ ESG Score is slightly higher than North American, so the size of companies and the relative availability of data should be considerably higher if this correlation was to comply with the entire line of data.

**Growing Trend?**

Responsible business, as a growing trend, has been around for years and is accompanied by a strong corporate transparency. When compared to the data available two years ago, we cannot say that there would have been an increase in transparency of ESG factors.

Θ_{2012} = 30,33

Θ_{2013} = 30,70

Θ_{2014} = 28,69

The data’s ESG Score growth since 2012 has been negative at -5.43%.

**Picture: The growth in the number of PRI- Principles of Responsible Investing (PRI) signatories. ****PRI – Principles of Responsible Investing**

**The conclusion of the statistical analysis**

A clear conclusion is that the company’s larger size will bring greater pressure to publicise businesses’ ESG issues because the value of the variable grows (to publicise wider scale of factors), and for these companies, the data was much better available compared to the smallest companies.

The location of the company is also important. The ESG data is clearly less available for Asian companies, even if the data set contains many large Asian companies, while the data for North American companies is readily available. If you compare the magnitude of European and North American companies, and the ESG Score, one can conclude that European companies publicise ESG issues far more than North American (greater ESG Score) when the data is available. The data was not available for 20% of the European companies, while the corresponding figure in North America was 8%. The expected value of European companies vary from the whole data set with a 5% significance level. Stating that the European companies’ ESG Score is strongly fragmented compared to, for example, North American companies with their ESG Score expected value closer to the whole data set. One has to note however that the data holds more North American companies than others and this has some effect as the ESG Score standard deviation depending on the area is quite high. The difference in the number of European and North American company observations is not so large that this could alone explain their deviation of expected values from the expected value of the data.

ESG Score is relatively poor indicator to help in measuring any reasonable responsible investments for specifically smaller companies (SME) and therefore is ill-suited HCP Quant strategy. When considering larger companies the standard deviation is high and suggests that the transparency of ESG issues is more a strategic issue when compared to SMEs. When you look at HCP Focus strategy, which primarily invest only in large-cap companies, the companies ESG data is also much more accessible and the score is significantly higher on average. Since the standard deviation is also considerably higher, one can conclude that it is difficult to come up with any sensible value of what the score should be if you wanted to invest responsibly on the basis of available Bloomberg data.

Standalone ESG Score does not tell anything other than the fact that how much public information on environmental, social and administrative issues is available by the business, and does not, for example, tell what the company has done and is trying to do to avoid or prevent problems. My own view is that the data is meaningful when used on individual cases and to support the comprehensive responsible investment strategy explicitly only in large companies whilst using different measurement tool to analyse the SMEs’ responsibility. If you use this database for a strategy that invests in SMEs, one should give this metric a lower weighting comparing when you would use it for a strategy that invests in large companies.

**End Notes**

*“The workshops revealed that there are many misconceptions between companies and investors on ESG factors and their financial materiality. Companies found that they have unique expertise on how and why ESG factors are material and core to their business—they understand their business best. Meanwhile, asset managers have not gained access to this information through current ESG questionnaires and desk research, and tend to focus on reputational issues. Many asset managers generally find the information contained in sustainability reports difficult to use for the purposes of valuing a company.” (Report from an international workshop series of the WBCSD and UNEP FI, 2010)*

ESG transparency involves a lot of difficulties when it comes to using them as a public communication tool between the company and the investor. The sole ESG Score does not directly indicate where there might be misconceptions about what kind of ESG factors would be significant between the investor and the company. The information derived from Bloomberg is added to this equation, a third party, whose priorities are necessarily fully representative neither of the company’s nor the investor’s point of view. While the above quote is already five years old, I have come across the same problem in connection with the responsibility reporting, that is, the abyss in an agreement among the company producing reports and public documents, as well as stakeholders who read those reports and outsiders. The gap has been reduced, but it still exists. This is probably due to the fact that the communication language is still young and is seeking to be formalised. This will last for many years if not decades.

*“ESG activities may increase various cost or investment cost for companies which are active on ESG issues. Investment cost may positively affect corporate profits in the long run although ROA (return on assets) may be lowered in the short run. If shareholders are not able to understand this point of view because they are shortsighted, companies’ ESG investment is merely regarded as cost for shareholders and is not reflected in Tobin’ q as a factor which may increase corporate profits. Especially environmental measures are regarded as increase in investment cost. Although this may be a problem for shareholders, it is differently evaluated from the standpoint of ESG. For those companies and management that consider contribution to not only to shareholders but also to various other stakeholders with respect to ESG issues, contribution to ESG is not regarded as cost but meaningful investment and should therefore be evaluated from broader perspective.” (ESG Factors in Corporate Valuation, SAAJ, 2010).*

**Photo: Growth in the use ESG of institutional investors in the USA in 2013-2014**

*(Source: https://workingcapitalreview.com/wp-content/uploads/2014/12/esg-graph-e1417643188944.png)*

The numerical value of ESG factors is really a multi-dimensional issue. In the context of publication of them, one should keep in mind that if two companies’ ESG Score differ dramatically from one another, one cannot directly infer which company would be more ethical and make for a more responsible investment. On my view regarding large companies with respect to ESG Scores, using of the negative screening rather than a positive screening is likely to be a better strategy. If two equally large companies that possess equal pressures to publicise businesses’ ESG issues were placed against each other and the second had no ESG data available, the decision not to invest in the company without it is more justified than investing with the company that provides this data. This is the theory, but in practice the situation is far more complex, and no decisions can be made only on the basis of data availability.

There is no factual information on how the company’s size factors in the costs of publication of ESG issues in this analysis. I.e. ESG Score: how much of the ESG Score difference between large and small companies is explained by the pressures to bring ESG factors in public and by the resources available. Only indicative information was available on this matter. While smaller companies due to their weaker market presence have lower pressures on publication of ESG issues, how great is their desire for it on average? Or does this desire diminish due to the lack of resources? This is a broad question to be analysed.

**Note on data**

The data used in the analysis is based on Bloomberg ESG Score. Bloomberg uses an algorithm to create a score. The calculation of the ESG Score takes 291 raw data points from which is a sample of 100 most appropriate is used based on the nature of each industry and the company’s operations. The final score consists of the number of 100 selected data points the company has reported. For example, the average of the 10 largest businesses is around 46 which means that companies report an average 46 out of 100 possible ethical criterion. Using this data has a few problems. As Bloomberg does not publish the methodology, you cannot know what data points are used. Due to company data fluctuations, ESG Scores are also not fully comparable. In the end, this methodology can be used on large companies on negative screening basis, that is, excluding those large companies which do not report at all. On this basis, we primarily use this scoring system with investing in large-cap companies for the HCP Focus Fund. For the HCP Quant Fund’s small and medium-sized company investing, this methodology is not appropriate.