Data is at the heart of EVE Score.
In order for this list to be activating the change we hope for the workplace, we want to be transparent with the approach and the principles that are leading the methodology.
Below are some of the main questions and responses we thought you might have as well as a technical appendix.
As everything we do, this list of questions is alive and will evolve as we are adding on features, data points, and your suggestions. It will be completed by regular posts in our blog section (“Voices”) when we feel like emphasizing on a specific choice or finding of the score.
We think the first step is to know “where we are” today, that is what the corporate world currently looks like when it comes to gender parity.
Absolutely and that is a great step forward. However, individual opinions may (and should!) vary when assessing the commitment of a given employer to gender parity or when describing the culture of a given firm. This can be a source of confusion for a company outsider, and practically hampers the valuable efforts of many existing mission-driven initiatives that rely exclusively on crowdsourced aggregated opinions.
EVE List believes it is possible to reach a broad consensus based on reliable data and actual quantifiable results achieved by an employer. For that reason, we have prioritized the collection of publicly available employer-level data, to establish an objective diagnosis on the state of gender inequality within each specific firm.
This being said, as our community grows it is our intention to use crowdsourced data (specifically individuals’ perceptions or experiences) and integrate it in our analysis, recognizing it as a very powerful tool to measure (and eventually address) the difference between reality and insiders’ perception. We will do this transparently and highlight the respective contributions of public official data and crowdsourced data when presenting our educated view on the state of gender parity within a company.
Like many social enterprises, EVE List will be standing on the shoulders of giants: the data we seek to use has already been patiently collected and stored by firms themselves, by government organizations, by non-profits and social businesses, by research institutions or by professional data providers. Our added value is to bring it all in one place and our aim is to then present the most accurate and educated view to the EVE List community, thus assisting those who wish to make well-informed decisions about their career.
We want our data to be “alive”, by which we mean continuously and organically enhanced. For example, there will be instances where our data only pertains to a specific region of the world, because of differences in corporate governance standards and regulation: we will strive for the “highest common denominator”, encouraging firms and/or governments in other regions to match the level of transparency prevailing in leading regions. Additionally, the EVE List community will comprise both users and reviewers: the former will candidly suggest ideas on how to enrich our datasets to further inform women on their career choices, while the latter will help us fix possible inaccuracies in our data.
Armed with public as well as crowdsourced data, we intend to inform women and empower them to choose the employer and the career path that best meet their aspirations. More specifically, we use the data collected to produce an aggregate assessment – we call it the EVE Score™ – of the gender gap in each company.
Armed with public as well as crowdsourced data, we intend to inform women and empower them to choose the employer and the career path that best meet their aspirations. More specifically, we use the data collected to produce an aggregate assessment – we call it the EVE Score™ – of the gender gap in each company.
The EVE Score™ is
- multifaceted: several components contribute to the score, reflecting multiple dimensions of gender parity and thus attempting to capture the complexity and nuances of each company’s situation.
- factual, not judgmental: the score relies on objective public data and statistics (describing results and achievements) and is complemented by crowdsourced data (describing individuals’ personal perception and experience).
- transparent: components of the score and their respective weights are clearly described
- sometimes incomplete: we will disclose cases where not all required data is available for a given company while it is available for a significant number of peers; missing data will eventually penalize a company’s EVE Score™ (in proportion to the relevant component weight) but we will first apply a “cure period” with no penalty to give the company the opportunity to provide the required data.
- sometimes region-specific: we will use region-specific data when global data is not available and when the same region-specific data is available for all peers (to avoid biased comparisons); this could make regional idiosyncrasies “look global” but we would rather run this risk than not use the data at all; we would also expect such idiosyncrasies to equally impact industry peers and therefore not impact the score on a relative basis
- iteratively revised and enhanced: growing and maturing datasets will allow us to enrich the EVE Score™ methodology; perhaps more importantly we learn by doing, we are open-minded and humble, and we can count on the feedback of our community, all of which will help us improve the EVE Score™ methodology whenever necessary and possible.
We currently use Version 0.
For a given company the EVE Score ™ takes a value between 0 and 100 (where 100 denotes the best score in terms of gender parity) and is computed as the weighted sum of several component scores (each between 0 and 100).
In Version 0 the component scores and their respective weight are as follows:
- Hourly Pay Gap (weight = 20%)
The component score is driven by the difference between the average male hourly pay and the average female hourly pay. The larger the difference (in both directions) the lower the score. The top score (100) is attributed if no difference is observed. - Uniform Gender Representation – High vs Low Earners (weight = 25%)
The component score captures differences in gender representation between two specific groups: the highest earners and the lowest earners. If the same gender composition (%female vs %male) is observed amongst the highest and the lowest earners a score of 100 is given.
- Bonus Pay Gap (weight = 5%)
The component score is driven by the difference between the average bonus paid to male employees and the average bonus paid to female employees. The larger the difference (in both directions) the lower the score. A score of 100 is attributed if no difference is observed. - Bonus Attribution Gap (weight = 5%)
The component score is driven by the difference between the proportion of male employees receiving a bonus and the proportion of female employees receiving a bonus the same year. The larger the difference (in both directions) the lower the score. A score of 100 is attributed if no difference is observed.
- Gender Balanced Population (weight = 10%)
The component score is driven by the extent to which the employee population deviates from a 50/50 split between male and female. The larger the deviation (in both directions) the lower the score. A score of 100 is attributed if male and female employees each represent 50% of the overall population. - Uniform Gender Representation – Senior Management vs Employee Population (weight = 25%)
The component score captures differences in gender representation between two specific groups: senior management and the overall employee population. If the same gender composition (%female vs %male) is observed amongst senior management and the overall employee population a score of 100 is given. - ESG Rating – Absolute (weight = 5%)
The component score is driven by the ESG Rating of the company, as published by MSCI on their public website (please see terms of use: https://www.msci.com/terms-of-use). - ESG Rating – Relative to Industry (weight = 5%)
The component score is driven by the ESG Rating of the company relative to industry peers, as published by MSCI on their public website (please see terms of use: Terms of use).
We define PayGap% = 1 – [average female hourly pay / average male hourly pay].
A score of 100 corresponds to PayGap% = 0, a score of 0 is given if PayGap% is outside of the [-30%, 30%] range, and the score is linearly interpolated as PayGap% varies between 0% and +/-30%.
For example, if PayGap% = 10% (or -10%) the component score will be 66.7
We define HighLowEarnersRatio% = [% of female amongst the highest earners / % of female amongst the lowest earners].
A score of 100 corresponds to HighLowEarnersRatio% = 100%, a score of 0 is given if HighLowEarnersRatio% is equal to 0% or above 200%, and the score is linearly interpolated as HighLowEarnersRatio% varies between 100% and 0% / 200%.
For example, if HighLowEarnersRatio% = 70% (or 130%) the component score will be 70.
For Small and Medium-Sized Enterprises (SMEs), we do not penalize the component score when the HighLowEarnersRatio% is greater than 100%. This is to account for SMEs’ smaller datasets which may be prone to outlier effects.
We define BonusPayGap% = 1 – [average female bonus pay / average male bonus pay].
A score of 100 corresponds to BonusPayGap% = 0, a score of 0 is given if BonusPayGap% is outside of the [-50%, 50%] range, and the score is linearly interpolated as BonusPayGap% varies between 0% and +/-50%.
For example, if BonusPayGap% = 10% (or -10%) the component score will be 80.
We define BonusAttributionGap% = [%proportion of male receiving bonus – %proportion of female receiving bonus].
A score of 100 corresponds to BonusAttributionGap% = 0, a score of 0 is given if BonusAttributionGap% is outside of the [-20%, 20%] range, and the score is linearly interpolated as BonusAttributionGap% varies between 0% and +/-20%.
For example, if BonusAttributionGap% = 10% (or -10%) the component score will be 50.
We define FemaleEmployee% = [# of female employees / # of employees].
A score of 100 corresponds to FemaleEmployee% = 50%, a score of 0 is given if FemaleEmployee% is equal to 0% or 100%, and the score is linearly interpolated as FemaleEmployee% varies between 50% and 0%/100%.
For example, if FemaleEmployee% = 35% (or 65%) the component score will be 70.
We define SeniorMgmtPopulationRatio% = [% of female amongst senior management / FemaleEmployee%].
A score of 100 corresponds to SeniorMgmtPopulationRatio% = 100%, a score of 0 is given if SeniorMgmtPopulationRatio% is equal to 0% or above 200%, and the score is linearly interpolated as SeniorMgmtPopulationRatio% varies between 100% and 0% / 200%.
For example, if SeniorMgmtPopulationRatio% = 70% (or 130%) the component score will be 70.
For Small and Medium-Sized Enterprises (SMEs), we do not penalize the component score when the SeniorMgmtPopulationRatio% is greater than 100%. This is to account for SMEs’ smaller datasets which may be prone to outlier effects.
The MSCI ESG Rating (please see terms of use: https://www.msci.com/terms-of-use) comprises 7 levels from AAA to CCC. The following correspondence table is used to determine the component score:
ESG Rating | Component Score |
AAA | 100 |
AA | 80 |
A | 60 |
BBB | 40 |
BB | 20 |
B | 0 |
CCC | 0 |
For a given company, the MSCI ESG Rating Distribution (please see terms of use: https://www.msci.com/terms-of-use) measures the proportion – expressed in % – of industry peers rated lower than the company. The component score is obtained by multiplying such proportion by 100. For example, if 60% of industry peers are rated lower than a given company the component score will be 60.