Investor Understanding Of Human Capital Value Add And Stakeholder Value Add Can Be Substantially Improved

Even sophisticated activist investors focus on comparisons of median employee pay to infer the cost and benefit added by human capital. They would be better off comparing employee value add numbers. The ESG movement itself can profitably benefit from use of related stakeholder value add data.

On November 15, 2022, the activist investor TCI sent a letter to the CEO of Alphabet arguing that the company’s cost base is too high and their cost per employee is too high. The letter’s section on compensation states:

“Alphabet pays some of the highest salaries in Silicon Valley. As detailed in Schedule 14A filing, median compensation totaled $295,884 in 2021. An analysis by S&P Global illustrates that median compensation at Alphabet was 67% higher than at Microsoft and 153% higher than the 20 largest listed technology companies in the US. This is no justification for such this enormous disparity.

We acknowledge that Alphabet employs some of the most talented and brightest computer scientists and engineers, and these represent only a fraction of the employee base. Many employees are performing general sales, marketing and administrative jobs, who should be compensated in line with other technology companies.”

The letter goes on to depict a picture stating that the median compensation at the top 20 largest technology companies is $117,055 and is $176,858 for Microsoft.

Regrettably, a simple comparison of median compensation across competitors does not tell you much. At the very least, the analyst has to consider differences in productivity (e.g., net sales per employee), investor return, or total value added or any consideration of the employee share of total value added. That is exactly what Steve O’Byrne and I have measured in our piece in the Journal of Applied Corporate Finance.

Let us define a few terms to get started:

· Economic value added (EVA): Post tax profit minus the opportunity cost of capital used to generate that profit. Such opportunity cost of capital is calculated as company’s total debt and equity multiplied by the weighted average cost of debt and equity capital. As an example, consider the case of United Parcel Service (UPS). In 2020, UPS reported $11.0 billion of after-tax operating profit. And by subtracting from this $11 billion of operating profit a capital charge of $6.7 billion, one can see that UPS had $4.4 billion of Economic Value Added, or EVA.

· Employee value added (EmVA): This is the new idea introduced in the paper. We estimate the average and total compensation for every firm as best as we can with available data. Next, we subtract an estimate of the opportunity cost or the next best wage that an employee would have earned had she not worked for the firm.

In particular, we use aggregate labor market data from the Bureau of Labor Statistics (BLS) and employee pay data for comparable industries and competitors for a firm from their 10-Ks. We then estimate the average annual “market pay”—and hence the opportunity costs—for a company’s employees in the same year as EVA is calculated. We designate the remaining number [(employee compensation – market pay) as pretax employee value added, and assuming a 25% corporate tax rate, subtract 25% of such pretax employee value added to compute post tax employee value add.

Continuing the UPS example, the average total compensation, including the value of employee benefits, of UPS employees in 2020 was $86,000. This translates into total employee pay of about $44 billion. We estimated the average annual “market pay”—and hence the

opportunity costs—for UPS employees in 2020 to have been $67,000, or $19,000 less than their actual pay. With 519,000 employees earning $19,000 more than their market pay—and

giving that number a 25% haircut for corporate income taxes that makes it $14,000—we get an “employee value added” by UPS of $7.5 billion.

· Alignment of EVA and EmVA: Imagine plotting EmVa on the Y-axis and EVA on the X-axis. Fit a regression trendline to relate these two variables for a company.

o The slope of the regression trendline shows the sensitivity, or “leverage,” of employee value added to economic value added.

o The r-squared is a standardized measure of alignment that varies from zero, or no correlation, to 1.0, or perfect correlation.

o The intercept of the regression trendline gives us a measure of performance-adjusted cost that we call the “pay premium at zero total value added.”

o The last measure, relative risk, is calculated as the ratio of the slope to the correlation and shows the variability of employee value added in relation to the variability of total value added.

I do not use the last two measures in the argument below but lay them out in the interest of basic science to illustrate the immense analytical horsepower that underlies this framework.

The case of Alphabet and Microsoft

Steve was kind enough to run the numbers for Alphabet and Microsoft, the companies referenced in TCI’s letter. We considered data for five years 2017-2022 for both Alphabet and Microsoft. Our median wage figures are inflation adjusted back to March 2021.

Here is what we found:

Alphabet and Microsoft are both associated with very low alignment and very low leverage. In plain English, this means that employee value added and economic value added are not highly correlated, somewhat contrary to the idea that employees capture a fair share of shareholder value add in technology companies. But that is better addressed in a separate piece some other time.

It is worth noting that Alphabet and Microsoft are in different industries (Interactive Media & Services for Alphabet and Software for Microsoft) and average market pay is quite different in these two industries: roughly $120K for software relative to $375K for interactive media & service. We calculate average market pay as national average pay x (1 + industry pay premium) and use Compustat data to calculate the industry premiums. For the uninitiated, Compustat is the standard database of firms’ accounting data used by most quants and academic researchers.

The industry premiums we compute are employee weighted and measure the premium to the employee weighted average of all Compustat companies. Note that companies in the two industries we consider have very different number of employees. Hence, the need to employ employee weighted industry premiums. Note that we exclude each company from the calculation of its industry premium to avoid tautological inferences.

Microsoft pays $155K over market ($275K vs $120K) to an average employee and reports employee value added of $20.0 billion (average employee value add multiplied by the total number of Microsoft employees). Alphabet pays $36K over market ($409K vs $375K) and has employee value added of $3.7 billion. Note the radically different employee value add numbers.

If Alphabet indeed overpaid relative to Microsoft, as claimed by TCI, we should observe far larger employee value add numbers for Alphabet.

Employees at Microsoft provide 66% of total opportunity costs (for both shareholders and employees) and take 24% of total value added, while employees at Alphabet provide 78% of total opportunity costs and take 5% of total value added.

There is little evidence that Alphabet employees are less productive than Microsoft employees. Alphabet’s revenues per employee are $1,627K relative to $944K at Microsoft and Alphabet’s net sales per employee (i.e., [sales minus estimated vendor costs] per employee) are $818K vs $640K at Microsoft. We subtract estimates of vendor cost to address the criticism that technology companies use large number of contractors. Moreover, such contractor numbers are not included in the total number of employees reported by the firm in their 10-Ks.

In sum, investors cannot simply compare median pay per employee. Ideally, they need to estimate market pay, look at value added, share of value added and employee productivity (i.e., net sales per employee) to evaluate whether human capital is over or under compensated and the value that employees add to shareholders relative to the value that labor earns from the firm.

Extension to stakeholder value add

More intriguing, the framework we lay out can be conceptually extended to other stakeholders as well. For instance, can we compute a supplier value add, defined as what we pay the supplier for her input relative to her opportunity cost of supplying her input to the next best alternative? We could potentially compute consumer value add or the price the customer pays for the product relative to what she would have paid for the next best alternative product. Data to estimate these value-add numbers are of course a problem as of now but we should hopefully get access to better data in the future. More important, such an extension is a far more rigorous way of thinking of “stakeholder value add,” an idea that has eluded measurement for long.