Surveying Human Capital Disclosure Norms
In 2020, the SEC issued an amendment to Regulation S-K which required companies to disclose their human capital management strategies in their filings to the Commission. As we enter the 2021 10-K busy filing season, we wait with much excitement and anticipation to see how companies have evolved their Reg S-K Human Capital Disclosures since this time last year.
How many companies have increased their disclosures beyond merely disclosing the total number of employees to now include a detailed breakdown of employees, for instance, the number of full-time, part-time, salaried, or hourly employees?
To answer this question, you could source the data from a traditional data vendor (if they have it), or collect this data yourself. But, what about all the other Human Capital-related metrics like Average Hours of Employee Training Per Year and Employee Voluntary or Involuntary Turnover? Do the traditional data vendors already collect that data? How do you keep up with new Human Capital metrics as priorities and disclosure strategies of companies evolve over time in this rapidly changing landscape?
Taking on the burden yourself to manually collect the data, year after year, not to mention keeping abreast of new reporting trends and discovering new human capital metrics to collect, is a mammoth task and elusive for those with limited resources.
Alternate Visions for Collecting Unstructured Data
This task of collecting the data would not be so time and resource consuming, if, and it’s a big IF, these disclosures were already tagged by the companies themselves using XBRL, the global digital reporting standard. Once the qualitative and quantitative data are tagged using XBRL, the data can be electronically consumed and be ready for analysis. Although companies in the U.S. are not currently required to tag their Human Capital disclosures using XBRL, they have been required to tag their financial statements and accompanying footnotes since 2009.
Arguably, the ‘leap’ to tag additional disclosures using XBRL is not a large one. There is much discussion amongst leading regulatory and standard-setting bodies and stakeholders to adopt digital ESG reporting.
In addition, the VRF/SASB XBRL taxonomy lays the foundation for human capital disclosures and other related ESG disclosures to be tagged using XBRL.
While all the pieces are coming together to realize this ESG digital reporting utopia in the future, how do you solve the problem HERE and NOW?
Leverage XBRL Framework for Better ESG Datasets - Using Machines
One approach is to leverage the XBRL framework, human experts, and machine learning to collect and structure this data at scale.
Instead of having individuals spend days manually reading and combing through large volumes of documents just to then copy and paste relevant data into some spreadsheet for further analyses, train the computer to do this.
This data can be collected and structured in minutes by the computer, and the results can then be immediately reviewed and refined by the expert. In addition, this effort to train the computer need only be done once (with periodic updates to the linguistic models as disclosure patterns evolve or change over time). The manual repetition of searching, copying, and pasting relevant data points year after year can be eliminated.
Human and Machine
It’s a symbiotic relationship between the human and machine that produces the high-quality data set efficiently and at scale - humans bring the expertise and insights, computers bring their supercharged computing muscle.
Use Case Example - Employee Measurement and Policies
As an example, the human analysts seeded the idaciti machine learning model with commonly used language to disclose full-time employees, and the computer took 3 minutes to find, structure and tag the data, with full traceback coordinates back to the source location within the document.
Now, the 3 minutes is not for a single company for a given year…it’s for the entire Russell 3000 group of companies! Within minutes, the human expert can start reviewing the results that he/she/they have trained the computer to collect the data at scale.
Use Case Example - Voluntary Turnover Rates
In another example, we examine Employee Voluntary Turnover. In 2019, less than 1% of companies in the Russell 3000 group of companies disclosed the rate of employee voluntary turnover. Not surprisingly, the disclosure of this metric improved, and in 2020 to nearly 5% amongst this group of companies.
Already, for the 2021 10-K filings season, we are tracking this disclosure in real-time and have already identified 1.6% of companies in the Russell 3000 companies disclosing this metric. As this filing season unfolds, we will continue to monitor this (and many other) Human Capital metrics.
Once human experts set up the models to automate the searching and structuring of the metrics, they can now sit back, let the computers do the heavy lift, and review/analyze the returned results.
The bridge between the here and the now to that of the digital ESG reporting future exists - it just takes some creativity and innovation.
Check out this page for more information on the idaciti toolset and use-cases in ESG: See idaciti