Wednesday, May 18, 2022

Calcbench has a released an improved version of our Excel add-in.  Performance is drastically improved.

Download the new version @

If you have questions or difficulties installing the add-in contact us.

Tuesday, May 17, 2022

If you’re a hardcore, process-oriented analyst, this post is for you. Chris Petrescu is the founder and CEO of CP Capital. As a data strategist and a former corporate finance manager, he has a long-standing relationship with financial data. We talked with Petrescu about Calcbench’s point-in-time data and how you can leverage it for a competitive edge.

Q. Tell us about Calcbench’s point-in-time data. What does it do? A. Calcbench’s point-in-time data allows you to take a snapshot of a company’s past. Analysts use point-in-time analysis to assess the evolution of a particular data point.

Q. How does point-in-time analysis help you achieve more accurate financial analysis? A. Point-in-time data isn’t altered by reclassifications or restatements. For historical market analysis, using point-in-time data allows you to remove certain biases that can undermine your research.

Point-in-time data helps you avoid “survivorship bias.” Some datasets will remove companies that have gone bankrupt, or overwrite companies that have been acquired. For systematic firms, this is an issue. Your historical analysis will look rosier because you’re only analyzing companies that have “survived” into the present time.

In addition, point-in-time data helps you avoid “look-ahead bias.” Vendors can store data delivery information incorrectly. For example, a vendor might time-stamp data delivered at 9:30 a.m., when it may have been known to the public markets, but it was actually not delivered until 11:00 a.m. If you’re calibrating a model to trade at the market open, you need to build a model that is trading on data known at that time.

Point-in-time data controls for these biases by accurately storing all company data, regardless of current-day status, and accurately time-stamps the data.  It sounds simple, but some of the largest data vendors in the world don’t grasp these concepts, or don’t care to.

Q. What pitfalls should analysts watch out for? A. Certain hedge funds are more prone to point-in-time data issues than others.  The farther back in time you look, the greater potential for point-in-time data issues.  So, a discretionary investor analyzing a handful of names over a few quarters is far less likely to uncover these issues, while a systematic firm analyzing 3,000 stocks over 5 years most certainly could detect such issues.

Q. What advice would you give an investor looking to integrate point-in-time data into their financial analysis? A. For starters, ask your financial data company how they handle their historical data. Also run high-level analytics to check if your data vendor is in fact accurately storing its data correctly.  It’s extremely important to know this about a dataset early on in the research process, so you can give yourself the opportunity to evaluate alternatives.

Q. How can you get point-in-time data from Calcbench? A. Calcbench subscribers get point-in-time standardized data by downloading the Calcbench API client Python package, and then calling the standardized method. (See screenshot below.) For a demo, contact

Thursday, May 12, 2022

The war in Ukraine is a humanitarian tragedy, and even worse, it shows no signs of ending any time soon. Now corporations need to confront a tricky financial reporting question: What should you disclose to investors about how the war in Ukraine is affecting your business?

Calcbench first wrote about Russia’s invasion of Ukraine at the beginning of March, when the war first started and nobody had any idea how events would unfold. Even then, a few firms began making some disclosures about the war’s effect in their operations, typically saying something in their Management Discussion & Analysis.

The issue is on our minds again now for two reasons. First, multiple companies have started citing the war as reason for a non-GAAP adjustment to net income. Second, the Securities and Exchange Commission just published a sample comment letter asking about the war’s effect on corporate reporting — a letter that told companies to tread carefully when putting a Ukraine non-GAAP adjustment in the earnings release.

Let’s start with the disclosures themselves.

Some of the disclosures are straightforward. Owens Corning ($OC), for example, filed its latest earnings release on April 27. Included in the release was this note:

In late March, Owens Corning made the decision to exit Russia through a transfer or sale of its facilities and, earlier this month, halted all future investments in Russia. The company is working to expedite its exit, while remaining committed to the safety and security of its employees in the country. 2021 net sales in Russia were approximately $100 million, or about 1 percent of the company’s consolidated net sales.

That’s all Owens Corning had to say about things. While the company did include various non-GAAP adjustments to net income (like just about every other firm out there), an adjustment for lost business in Ukraine and Russia was not among them. But providing a sense of net sales in Russia in 2021 does give investors some sense of proportion.

More interesting is Philip Morris International ($PM). The tobacco giant filed an earnings release on April 24 and did include adjustments for lost revenue in Russia and Ukraine. See Figure 1, below.

Philip Morris then provided an updated forecast for 2022, where the company did include that $0.10 adjustment for first-quarter 2022, and then assumed no other revenue from either country for the rest of the year. See Figure 2, below.

Further down, Philip Morris also discloses that Ukraine accounted for less than 2 percent of total revenue in 2021 (Philip Morris revenues were $31.4 billion last year, so Ukraine was somewhere less than $628.1 million) and the company’s Ukrainian assets were worth about $400 million. Meanwhile, total Russia revenue in 2021 was about $1.9 billion (6 percent of revenue) and assets in country were valued at $1.4 billion.

A final example comes from heavy-industry manufacturer Cummins ($CMI), which filed an earnings release on May 3 that includes detailed breakdowns of which divisions suffered how much costs due to Russia disruptions. Total costs were $158 million, which Cummins first disclosed across its five operating segments. See Figure 3, below.

Separately, Cummins also included a narrative explanation of its Russia charges. They included inventory write-downs, asset impairments, and accounts receivables that presumably will never arrive. Within that narrative note Cummins also included another table, shown below, that breaks down the $158 million another way.

Meanwhile, the SEC Says…

We still have that sample comment letter from the SEC, published at the beginning of May. The letter raised two points about non-GAAP adjustments related to Ukraine.

First is the idea of adjusting for lost revenue due to Russia’s invasion of Ukraine. The SEC’s terse directive: “Please remove these adjustments.” Why? Here’s the full explanation in the comment letter:

We note your adjustment to add an estimate of lost revenue due to [Russia’s invasion of Ukraine and/or supply chain disruptions]. Recognizing revenue that was not earned during the period presented results in the use of an individually tailored revenue recognition and measurement method which may not be in accordance with Rule 100(b) of Regulation G. Please remove these adjustments.

Basically, if you include a non-GAAP adjustment for revenue that never actually, ya know, existed — that’s a no-no under SEC reporting rules.

Second, the SEC is also on the lookout for adjustments that a company claims are related to Ukraine, but perhaps are just normal and routine costs of business. In that case, the SEC will be looking for more fulsome disclosure about whether a company’s one-time Ukraine adjustments really are as one-time as it claims:

We note your adjustment for certain expenses [such as compensation expense or bad debt expense] incurred related to your operations in Russia, Belarus, and/or Ukraine that appear to be normal and recurring to your business. Please tell us the nature of these expenses. Explain how you have considered Question 100.01 of the Division’s C&DI for Non-GAAP Financial Measures and why you believe that the expenses excluded from your non-GAAP measures do not represent normal, recurring operating expenses.

The rest of the comment letter addresses other disclosure issues, such as what goes into Management Discussion & Analysis, cybersecurity risks, and internal control over financial reporting. All of the comment letter is food for thought for analysts; if the SEC is wary of what companies are reporting about Ukraine issues, those issues are worth keeping on your radar screen too.

Monday, May 9, 2022

You may have noticed that the stock market has been going down the tubes lately. Here at Calcbench, however, we wanted to press further. Just how far down the tubes have share prices gone, really?

To answer that question, an analyst could look at the price-to-earnings (P/E) multiple for one or more firms you follow — but that might not be the most accurate metric, because the earnings part of the equation is net income. Since scads of companies adjust net income and report non-GAAP earnings, that implies that they don’t believe “pure” net income is the most accurate measure of business performance. That, in turn, calls into question whether P/E ratios based on net income are the most reliable metric of value.

We decided to answer the question by examining the trailing 12 months of net income for 353 non-financial firms in the S&P 500, and then adding back the amortization of intangible assets to net income. Why? Because those amortization costs are the most common non-GAAP adjustment that large firms make. If companies believe that amortization costs are worth adjusting for net income, then by the same logic those costs are also worth adjusting for P/E ratios.

Once we had those adjusted earnings numbers, we then divided them into companies’ market capitalization at the end of March and at the close of business on May 5. Table 1, below, shows the results for all 353 firms we studied, tallied up as a whole.

As you can see, including the amortization adjustment produces a small but significant difference. We also calculated the earnings yield, which essentially is the inverse of the P/E ratio. It looks backwards at prior earnings, rather than anticipates future earnings; and earnings yield rises as share price falls. Again, we see a small but significant difference when amortization adjustments are included in the calculations.

What to Do Next

For the record, the long-term historical average P/E ratio for S&P 500 firms falls somewhere between 13 and 15, according to Investopedia, which suggests that the market still has a few more tubes to go down before it reaches bottom. We also found a nifty chart from that tracks P/E averages back to 1926 if you’re that much of a history buff.

In more practical terms, you could also use insights like this to sharpen your own assessments of firms’ value. For example, you could compare the adjusted P/E of firms you follow against our adjusted group averages; or you could research adjusted P/E averages of specific sample groups — say, all tech stocks. You could even add other adjustments if you like, such as for stock-based compensation.

One would do all these calculations from our Multi-Company page. There, you can select the companies you want to study and start by finding their net income. Then you can add other non-GAAP adjustments by entering them into the search field on the left-side of your screen above the company results. Calcbench tracks all the most common non-GAAP adjustments, so you can pull them up with a few keystrokes.

That gives you all the data you need to start adjusting earnings, then adjusting P/E ratios, and then estimating whether a firm’s share price still has further down the tubes to go or whether it’s time to buy.

Thursday, May 5, 2022

Calcbench is devoting the month of May to non-GAAP financial disclosures — and as fate would have it, a great example fell into our laps right away. Uber filed its first-quarter earnings releases this week, with interesting non-GAAP disclosures galore.

Let’s start at the top. Uber ($UBER) filed its Q1 2022 earnings release on May 4, and according to GAAP rules the quarter looks horrible: a net loss of $5.9 billion (yes, billion) on revenue of $6.85 billion.

Except, after a number of non-GAAP maneuvers, Uber also reported an adjusted EBITDA profit of $168 million. How did that come about? Figure 1, below, is Uber’s reconciliation of non-GAAP reporting, and it tells the tale.

Almost all of Uber’s $5.9 billion loss comes from a $5.56 billion charge Uber recorded to write down investments it made in overseas ride-hailing services Grab, Aurora, and Didi. Elsewhere in the earnings release, Uber broke down the write-downs by specific investments:

  • $1.9 billion unrealized loss on the Grab investment;
  • $1.7 billion unrealized loss on the Aurora investments;
  • $1.4 billion unrealized loss on the Didi investment;
  • $462 million unrealized loss on its Zomato investment. (Zomato is a food delivery service in India.)

Uber’s Non-GAAP Through History

Uber’s non-GAAP reporting for this quarter also caught our eye because we also wrote about Uber last year. For first-quarter 2021, Uber reported an adjusted EBITDA that was actually lower than the GAAP-approved net income number. You don’t see that too often.

At the time, Uber had sold its self-driving car unit to Aurora (a tech startup that hopes to deliver self-driving cars some day in the future) for $1.68 billion, and made a $400 equity investment in Aurora to boot.

That sale, along with a few other small adjustments, led to Uber reporting a $1.71 billion gain on Other Income for the quarter. Uber had to report that as a non-GAAP adjustment, but it was an adjustment downward because you can’t always assume that other income will be a gain. Hence Uber’s adjusted EBITDA ended up being considerably lower than its GAAP net loss.

Now things have come full circle: after selling its self-driving technology to Aurora and booking a profit, which pushed its adjusted EBITDA down; Uber is now booking a loss on the investment it made in Aurora at the same time, which pushes its adjusted EBITA up.

Such a wonderful time to be alive and a financial analyst, right?

Without question, technology stocks have taken a beating in the last several weeks. Trying to discern growth, risk, and opportunity amid the flood of data those companies have reported is no easy task.

Case in point: Apple ($AAPL), one of the giants.

Apple is often praised for the rigor of its financial reporting. The challenge is simply that there’s so much of it, analysts can have a hard time discerning trends or identifying important nuggets.

For example, when Apple filed its first-quarter 2022 report on April 29, it included this breakdown of revenue by product:

To be clear, we love Apple’s voluminous disclosures; it’s one of the reasons we mention the company so often on this blog. From an analyst’s perspective, however, you need more context than what we see above to understand how Apple’s product sales might be evolving over time.

To get a better sense of that context, we used Calcbench’s ability to export historical segment disclosures to Excel. (The same ability we mentioned a few weeks ago while looking at Delta Air Lines and its passenger revenue per available seat mile.) Just hold your cursor over one of the hyperlinked numbers, then hit “Export History” when the feature box opens. 

We did that for Apple, and got a nifty spreadsheet with quarterly disclosures for each revenue segment going back to Q1 2018:

From there, you can start charting the performance of individual product lines over time. You could also overlay those performance graphs against similar segments for other firms — say, tablets or phones made by Samsung, or desktop computers made by HP. Or you could just study revenue growth within Apple’s several product lines, to see which ones are growing fastest.

We took the Excel data and quickly cooked up a chart for Mac, iPad, and wearables product lines. (iPhone revenues are so much higher they’d make the chart difficult to read.)

So, 20 minutes after we thought,”Hmmm, how is Apple doing these days,” we already had this chart done:

Whatever you’re looking for, Calcbench has the data.

Monday, April 25, 2022

Yes, we know — everyone in the universe is paying attention today to Elon Musk and his $44 billion deal to take Twitter ($TWTR) private. Calcbench, however, just noticed an intriguing disclosure from one of those other social media companies out there: Snap ($SNAP), parent company of Snapchat.

Snap filed its Q1 2022 earnings report late last week, filled with all the usual  numbers. Revenue rose 38 percent from the year-earlier period, but the pioneer in nude selfie technology still posted an operating loss of $271.5 million and a net loss after taxes of $359.6 million.

The number that most caught our eye, though, was in the Statement of Cashflows.  In the first quarter of 2022, SNAP bought stock worth a little over $1.3 Billion.  

Realizing that they didn’t pay cash for the stock, we wanted to learn more and noticed the increase in Convertible Senior Notes on the balance sheet.

Going through Snap’s footnotes, we stopped short at THE disclosure about convertible notes. In February 2022, Snap closed a private placement of $1.5 billion in convertible notes due in 2028, with an interest rate of 0.125 percent per year.

That part alone is impressive; Snap secured $1.5 billion in cash with an interest rate so low it’s essentially free money. Then came this statement:

The 2028 Notes are convertible into cash, shares of our Class A common stock, or a combination of cash and shares of our Class A common stock, at our election, at an initial conversion rate of 17.7494 shares of Class A common stock per $1,000 principal amount of 2028 Notes, which is equivalent to an initial conversion price of approximately $56.34 per share of our Class A common stock.

So Snap can convert those notes into shares — at management’s sole discretion, mind you —  with a strike price of $56.34. That is considerably higher than Snap’s closing price lately, around $30 per share.

We have so many questions. First, who did this deal? Who believes that Snap, a company that has never printed the bottom line of the income statement in black ink, is on an upward trajectory? When Snap says it has an “initial” conversion rate of $56.34 per share, does that mean the deal has other possible conversion rates? If so, what are those trigger events?

Snap’s disclosures about the deal don’t say much. The company did include this paragraph explaining when Snap might convert some of the notes into cash, rather than shares:

We may redeem for cash all or any portion of the 2028 Notes, at our option, on or after March 5, 2025 if the last reported sale price of our Class A common stock has been at least 130 percent of the conversion price then in effect for at least 20 trading days at a redemption price equal to 100 percent of the principal amount of the 2028 Notes to be redeemed, plus accrued and unpaid interest, if any.

For the record, 130 percent of a $56.34 conversion price is $73.24 per share. Again, however, notice that the disclosure refers to the conversion rate then in effect. So our question about other conversion rates, and possible trigger events for that, remains valid.

We should also note that Snap’s share price was $32.54 at the end of January and $39.94 at the end of February; so these mystery investors were buying into a rising stock. Plus, Russia’s invasion of Ukraine happened on Feb. 24. It’s possible that the investors bought these convertible notes earlier in February, when most people believed Russia wouldn’t invade Ukraine. Then it did, and the markets have been in turmoil ever since. You can’t blame the investors for that external event and its bad timing.

Snap did file a template agreement for the convertible notes on Feb. 11, but it’s only a sample agreement that includes no buyers’ names and lots of blank spaces, including blanks for the exact date of the transaction. Interesting, but not that helpful.

Figure 1, below, shows Snap’s share price history over the last five years. The company did see a remarkable run-up in price from late 2020 into late 2021 (was there a nude selfie craze we missed?), but then the price dropped precipitously. So a share price in the mid-50s isn’t unheard of — but Snap does need a lot of enthusiasm over future earnings to achieve it.  

Then again, as this convertible notes placement suggests, at least some people believe in the company. We’ll see.

Delta Air Lines filed an optimistic earnings release this week. Calcbench tools, however, show a clear picture of how much more recovery needs to happen.

Delta Air Lines’ latest earnings statement had an on-time arrival Wednesday morning, and overall the numbers looked impressive enough for a business battered by the Covid-19 pandemic — but dig a little deeper into the data, and you can see that Delta still has a long-haul flight to full recovery.

First, the headline numbers. Delta ($DAL) reported $9.35 billion in revenue and an operating loss of $783 million, compared to $4.15 billion in revenue and a $1.4 billion operating loss in the year-earlier period. Except, of course, the year-earlier period was a miserable one in the throes of a gigantic covid surge. So that’s not really comparable.

In its earnings release, Delta instead compared Q1 2022 to Q1 2019, the last time we had a first quarter that one could call normal. That’s fair, and in which case the comparables look like this:

The research crew at Calcbench, however, was more interested in the non-GAAP performance metrics reported further down the earnings release. We found a telling one there on Page 9: passenger revenue per available seat mile.

“PRASM” is a popular efficiency metric in the airline world. It is measured in cents, and gives a sense of how much revenue — airfares, in-flight meals, change fees, baggage fees, and everything else — an airline can wring from a person with his or her rear end on a seat. Delta’s PRASM for Q1 2022 was 13.33 cents, compared to 14.83 cents in Q1 2019.

Calcbench, however, lets users dig deeper into that number with a few keystrokes. If you hold your cursor over that 13.33 number, an option appears allowing you to Export History. Click on that, and you see this interface.

Our Export History feature lets you download prior-period disclosures for whatever number you’re staring at, including non-GAAP metrics. We decided to download the last 12 quarters’ worth of PRASM disclosures. Once that Excel sheet popped into view, we converted those numbers into a chart — and we saw this.

So as impressive as Delta’s recent pace of recovery might be, the airline still has a long way to go simply because it fell into such a steep nosedive in 2020 and 2021. Not until the end of 2021 did Delta come even within sight of pre-pandemic PRASM levels.

Calcbench subscribers can do their own analysis along similar lines, for Delta or any other company. (The research we did on Delta, from seeing the PRASM number to generating the Excel chart above, took us less than two minutes.) Just use our Interactive Disclosures page to find the filing you want to research, and then scroll through to the specific disclosure you want to research. Hold your cursor over that number, look for the Export History window to appear, and then do what we did above.

Happy flying!

Thursday, April 7, 2022

Now that proxy season is upon us and investors will start asking about executive compensation again, we want to remind Calcbench users that you can always pull together a quick summary of exec compensation from our databases with a few keystrokes.

To begin, start on our Executive Compensation page. You can reach that page by moving your cursor over the My Products menu at the top of your screen, and then scrolling down to the option that says, you guessed it, “Executive Compensation.”

The page itself is pretty straightforward. Simply enter the company or companies that you want to research in the search field, and hit enter. You’ll then get two links: one for executive compensation (that is, the CEO and other named executive officers) and another for board director compensation.

Click on either option, and the results will download onto your computer as an Excel file. That’s all there is to it.

For example, we entered “Dow Jones” in the search field, which quickly suggested “Dow Jones Industrial Average” as the group to analyze. That led us to Figure 1, below.

Then we selected the Executive Compensation option. Figure 2, below, shows the results.

Notice that we offer a lot of executive compensation data — a row for each named executive, for each year we have one file; and then columns that contain each executive’s base salary, bonus, options, equity pay, pension benefits, and other compensation. Plus a nifty total number at the end of each row, and the option to trace those numbers back to the actual filing, too.

Quick, easy, and informative! That’s how we like to serve up the data here at Calcbench, executive pay included.

Wednesday, April 6, 2022

Analysts can use Calcbench to research numerous profitability, liquidity, and solvency ratios, for individual firms or to compare ratios across groups.

So there we were the other day, skimming the Wall Street Journal, when we came across an interesting article about Kontoor Brands ($KTB) the maker of Lee and Wrangler jeans spun off from VF Corp. ($VFC) two years ago.

The article interviewed Kontoor’s CFO, Rustin Welton, who talked about how the firm has been reducing net debt in the last several years by focusing on its cash conversion cycle. That cycle tracks how quickly a company turns over inventory and collects payments; the more quickly your cash conversion cycle runs, the more quickly you generate free cash that can be used for tasks such as paying down debt.

In Kontoor’s case, focusing on the cash cycle has allowed the company to cut net debt from $922 million when it spun off from VF in 2019 to $606 million at the end of 2021.

So what does any of that have to do with Calcbench? We have such data! The cash cycle is one of many performance ratios we track and calculate as a matter of course for all publicly traded firms.

To find those ratios in Calcbench, one place to start is our Bulk Data Query Page. Scroll to the bottom of the page and you’ll see the many ratios we track. Figure 1, below, tells the tale. (The formal name we use for the cash cycle ratio is “Cash to Cash Cycle,” but it’s substantively the same thing.)

First, select whatever firm or group of firms you want to analyze at the top of the Bulk Data Query Page, along with whatever period of time you’re researching. (You can get data for one specific period or a range across multiple years.) Then scroll to the bottom, select the ratios you want to analyze, and press “Export to Excel.” You’ll get all the data in one nifty spreadsheet.

Alternatively, you can also use the Multi-Company Page. Again, start by selecting one or more firms that you want to study. Then enter “cash to cash cycle” in the standardized metrics text field on the left side of your screen, and the results will emerge. Figure 2, below, is an example. We included Kontoors, VF Corp., and Under Armor ($UA) to give a sense of how Kontoors compares to a few peers, and tracked the data back to 2018.

You can research any of the other ratios listed in Figure 1, the Bulk Query Page, in the same way; they are all standardized metrics one can find on the Multi-Company page too.

That concludes today’s tour of performance ratios. Pick the one that matters to you, fire up the database, and enjoy!

Thursday, March 31, 2022

In February we told you the tale of Amazon’s two main segments, where we looked at the Amazon (AMZN) retail segment (national and international) and AWS. In this piece we want to see how those two segments compare to similar businesses. We compare Amazon’s retail segments to Walmart (WMT) and AWS to Microsoft’s (MSFT) Intelligent Cloud.

Let’s start with the retail side. As we compare Amazon to Walmart, remember that Amazon FYE is Dec. 31, 2021, while Walmart FYE is Jan. 31, 2022. Walmart’s Sam’s Club segment is included in Walmart’s U.S. operations, and all amounts are in millions.

As you can see from the above table, Walmart’s operating profit margin is much higher than Amazon’s; so is Walmart’s return on operating assets (ROOA). Even Walmart’s asset turnover is higher for U.S. operations, and slightly lower for international operations.

We had seen before in our previous analysis that Amazon’s retail segments do not perform as well as AWS. Let’s see now how AWS stacks up to Microsoft’s Intelligent Cloud (IC).

As we compare Amazon to Microsoft, again remember that Amazon’s FYE is Dec. 31, 2021 while Microsoft’s FYE is June 30, 2021 (and again, all amounts in millions). Also, Microsoft does not allocate assets to segments for reporting purposes so ROOA and Asset Turnover cannot be calculated for Intelligent Cloud.

As you can see in the chart above, again Amazon does not compare favorably to Microsoft. Although AWS’s revenue is somewhat higher than Intelligent Cloud, Intelligent Cloud’s operating income is much higher.

This analysis was conducted using the companies’ segment disclosures. Want more? See our attached spreadsheet or go to Calcbench Disclosures Query and find more companies of interest.  

The month of March has been all about share repurchase programs here at Calcbench, with numerous posts and pieces of research dissecting how buyback programs work and how analysts can evaluate those efforts.

Today we offer a recap of all that material, so that Calcbench subscribers can understand everything available to them to analyze buyback programs and how you can get started analyzing the firms you follow.

First is a research note we published in early March, reviewing how much money firms have spent on share buybacks in the last decade and which firms have been the biggest spenders. Apple ($AAPL) tops the list by far and away, spending $479.55 billion on share repurchases from 2012 through 2021 — but corporations spent more than $6.52 trillion on buybacks overall in that period. Download our research note to see our full findings.

Second, we also have two pieces from columnist Jason Voss unpacking all the details of how share buyback programs work:

Third, Calcbench built an Excel template that subscribers can use to perform the same analysis Voss did for firms beyond Nvidia. You can download the template and then use it for your own analysis by entering the ticker symbols for firms you follow. (Please note, that interactive feature will only work for Calcbench subscribers. If you’re not a subscriber and want a trial of the service, just contact us at

And as you might have guessed by now, we love this stuff. If you have special requests related to share buyback research, we’re happy to hear your needs and see how we can help with one-off projects or research requests; again, just drop us an email to get the conversation started. We have millions of pieces of financial data in our archives, so whatever you’re hoping to do, we have the data to do it somewhere in there.

By Jason Apollo Voss

Today we continue our two-part dive into share repurchase programs, trying to answer an important question for financial analysts. Considering the trillions of dollars that firms have spent over the years buying back shares — how can one determine whether that’s the best use of a company’s capital?

The information to answer that question is tucked away in a company’s Consolidated Statement of Shareholders’ Equity. In my previous column we reviewed how the Statement of Shareholders’ Equity is constructed using microchip technology company Nvidia (NVDA) as an example.

Now let’s study the data that Nvidia has disclosed on that statement in recent years, to analyze Nvidia’s own share repurchase spending.

First, we should note that Nvidia’s disclosure on its stock-based compensation program is excellent; the company even attributes issuances to different parts of its organization. That, in turn, allows us to do an interesting analysis to answer the question: Is Nvidia getting a good bang for the buck from its research and development department stock options program?

Starting the Analysis

To begin, think of Nvidia’s options grants as an investment. Like any investment, it should generate a return on capital — and like all returns on capital, calculations are done over an investment time horizon.

In its Note 4, Nvidia says that its vesting period for options is 2.5 years. So if its options program is structured well, then we should see an increase in revenues about 2.5 years after an investment in R&D. Given that Nvidia is competing in an industry where the technology turnover is rapid, it’s reasonable to think that two to three years out, its R&D ought to drive a large proportion of new revenues.

So how did Nvidia actually perform?

In 2019 Nvidia expensed $336 million in its stock options grant (according to Note 4) to its R&D team. Total R&D expense for that year was $2,376 million, meaning stock option incentives were approximately 16.5 percent of R&D expense.

How much did revenues grow by two years later? By $4,959 million ($16,675 million in 2021 – $11,716 million from 2019).

This increase may not entirely be attributed to new products; some of it doubtless came from sales to new customers and new regions. Nonetheless, Nvidia managed to have an incremental revenue increase relative to its R&D expense from two years prior of 208.7 percent. (That is, $4,959 million in incremental revenues after t + 2 years ÷ $2,376 million in R&D expense in time t.) Not bad.

Because of Nvidia’s excellent disclosures, we can improve on this blunt force calculation even more (and we will, below). Nonetheless, this evaluation of R&D can be done for other companies, too.

Next up: let’s answer some of the basic questions about share repurchase programs. The following summary table created by Calcbench is helpful in answering our questions. (Calcbench subscribers proficient in Excel and XBRL can do this themselves; or Calcbench can provide help upon request. Also, those DIV/0 errors in 2020 and 2021 happen because Nvidia stopped buying back shares in those years.)

Are these programs just offsetting share issuance to execs and employees?

If the answer to this question is “yes,” then it might be that the company is managing its earnings per share via its share repurchase program. In Nvidia’s case, however, based on this analysis, the answer at first glance is likely no, because its ratio of shares repurchased to issuance is less than 1.0. This means that the company’s share buyback program is not fully offsetting share issuance.

When coupled with the analysis above about the incremental revenue achieved by the company relative to its R&D spend (including options issuance), then we can conclude that Nvidia’s repurchases are not just offsetting share issuance to execs and employees, since the repurchases are also earning a return on their R&D. In other words, Nvidia’s incentives seem to be driving better business outcomes.

Are these programs simply a way of managing earnings per share?

In Nvidia’s case, the answer seems to be that it isn’t massaging earnings per share via its share repurchase program. If a company were trying to massage EPS through share repurchases, then its ratio of shares repurchased to issuance would be above 1.0 — meaning, it would be buying back more shares than it was issuing. As we just noted above, Nvidia isn’t doing that.

Are these programs a better use of capital than other types of investments, such as in R&D?

That’s a difficult question to answer in Nvidia’s case, because we can’t  ascertain from its disclosures something equivalent to a “same store revenue growth” estimate.

That is, what portion of sales growth each year is attributable to new products and new markets versus old customers buying more of what they have to offer? Knowing the answer to this would allow us to identify new revenues that are a result of R&D, versus a return from employees working harder or old customers buying more. A review of Nvidia’s Management’s Discussion and Analysis and its discussion about revenues doesn’t help either,  because Nvidia doesn’t  provide a “same store revenue growth” type figure.

One possible way to estimate the effect of this is to look at Note 4 again and notice the proportion of issuance attributable to different parts of the company. (Again, this is only possible because of Nvidia’s excellent disclosures.)

That information for Nvidia looks like this:

We can treat as a hypothesis that Nvidia knows what it is doing when it incentivizes different employees within the company. Said differently, the above averages are likely good proxies for what proportion of new revenues can be attributed to its sales force (likely represented by the “cost of revenue” stock-based compensation), R&D scientists and their new inventions, and other employees (likely represented by the SG&A line, above).

Knowing this, we can make an estimate that 59.2 percent of new revenues stem from R&D. That comes directly from the table above, in the “Average” line.

That, in turn, allows us to re-evaluate the possible returns on R&D that Nvidia experiences. Incremental revenues of $4,959 million (from earlier) x 59.2 percent (assumed contribution attributable to R&D efforts from our table above) = $2,936 million of incremental revenues due to R&D. R&D expense from 2 years prior was $2,376 million. Meaning the total return in two years is 23.6 percent.

By contrast, let’s look at the amount of value-add of the share repurchase program. In 2019, Nvidia bought back 9 million shares of stock according to its Consolidated Statements of Shareholder’s Equity, for a price of $1,579 million. The average price per share of these shares was therefore: $1,579 million ÷ 9 million shares = $175.44. As of two years later at the end of fiscal 2021 Nvidia stock was trading at $129.90 (closing price on 29 January 2021), or a two year total return of: -26 percent.

Ouch! Their share repurchases seem to have lost them money. But wait, there’s more!

Recall that Nvidia said its options vest over 2.5 years? So we should look at the three-year period, too. The closing price three years later was $251.04, or a three year total return of: 43.09 percent.

I know what you are thinking, however: What about its stock price at the 2.5-year mark, to match the options vesting period with outcomes? Nvidia’s closing stock price at the end of July 2021 was $197, for a total return of 12.3 percent.

What this highlights is that research is hard; lots of our work is as much art as it is science. Two of the three periods examined above (the 2-year and the 2.5-year periods) actually look as if Nvidia’s share buybacks are not as beneficial as a straight investment in R&D.

But our analysis is not complete because there was also a boost to earnings per share due to Nvidia’s buybacks. The value-add is measured as:

  • Earnings per share as reported 2019: $4,141 million in net income ÷ 608 million shares outstanding = $6.81
  • Earnings per share, without 9 million shares bought back = $4,141 million ÷ 617 million shares = $6.71
  • So, an additional $0.10 per share in Nvidia EPS cost $1,579 million, versus total net income of $4,141 million and EPS without the buyback of $6.71.

So far our analyses seem to suggest that share buybacks are not earning as high a return for Nvidia as its R&D program is earning. For completeness, though, perhaps the company is buying back shares because of the price per share effect? To do this analysis we:

  • Multiply that $0.10 increase in EPS by its P/E ratio of 19.1 (which is $129.90 price per share ÷ $6.81 EPS). This gives us a possible increase in market cap of $1,907 million.

Again, the company spent $1,579 million to get a possible valuation boost of $1,907 million. What if we look at the 2.5-year period? To calculate the P/E ratio at the 2.5-year mark we need the EPS figure from Nvidia’s third-quarter 2020 through to the end of second-quarter 2021.

This is a bit tricky to calculate, but using the figures from annual and quarterly filings for the periods in question, we can figure out that Nvidia’s EPS for that period was $8.11. Its stock price at Aug. 1, 2021 was $197, which gives us a P/E ratio of 24.3. Our estimate for the boost to market cap is therefore $2,429 million versus a capital outlay of just $1,579, or a return on investment of 53.8 percent. By this calculation, it looks as if the buyback is inexpensive and actually a creation of capital for shareholders.

What is fascinating, of course, are the possible feedback effects of the buybacks. That is, the company buys back its shares and earns a valuation bump. That, in turn, leads to enthusiasm for the company’s share price performance, which results in an even higher multiple. Something to ponder as you await that next earnings release.


Our two columns on share buybacks provide the basis for any financial analyst to look more critically at a firm’s share repurchase programs. We do have two final points worth noting.

First, Calcbench has designed a template to let you import as-filed Statement of Shareholders Equity from any firm; we encourage you to download it. Data diehards can also use our API if you’d prefer. Drop us a line at and we’ll get you squared away.

Second, after we wrote these two columns examining Nvidia’s 2017 to 2021 annual reports, the company filed its 2022 report. Those numbers look a bit different for several reasons, including a would-be merger that ultimately was called off and a 4-to-1 stock split. Don’t let that throw you; the analytical techniques discussed here, plus the Excel templates Calcbench has cooked up, are a solid foundation for you to conduct your own analysis of the firms you follow.

This is an occasional column written by Jason Apollo Voss — investment manager, financial analyst, and these days CEO of Deception and Truth Analysis, a financial analytics firm. You can find his previous columns on the Calcbench blog archives, usually running every other month.

By Jason Apollo Voss

Companies issue and buy back their own shares all the time. In the last 10 years, as Calcbench research shows, companies have repurchased more than $6 trillion worth of shares.

This column is the first in a two-part series to expand on that research — to examine share repurchase programs in detail, and try to determine whether they’re executed to offset share offerings to executives, to manage EPS, or simply because the company believes its stock is undervalued. We also evaluate whether these buyback programs are a good use of the company’s capital.

In this first part, we’ll walk through the financial statement that tells you what you want to know about share buybacks: Consolidated Statement of Shareholders’ Equity. To perform this analysis we selected Nvidia (NVDA) as an example, and we’ll have numerous excerpts from its disclosures.

Next week, in Part II, we’ll consider how you can analyze those disclosures to understand whether buybacks are the best use of a company’s capital.

Also, throughout this column and the next, you’ll see many spreadsheets with Nvidia data. Calcbench has designed a template to let subscribers import as-filed Statement of Shareholders Equity from any firm; we encourage you to download it. Data diehards can also use our API if you’d prefer. Drop us a line at and we’ll get you squared away.

U.S. companies have executed more than $6.5 trillion in share repurchases over the last 10 years. Given the gigantic amounts of capital spent on buybacks, all savvy investors would do well to question these corporate capital allocation decisions. For example:

  1. Does the company believe that their shares are undervalued and is simply returning that capital to shareholders?
  2. Are these programs simply offsetting shares given to executives and employees as compensation?
  3. Are these programs simply a way of managing earnings per share?
  4. Are these programs a better use of capital than other types of investments, such as in R&D?

Where to Begin Your Analysis

In the United States, companies are required to provide details of the activities of equity shareholders in the “forgotten'' financial statement: the Consolidated Statement of Shareholders’ Equity. We say “forgotten” because most courses and books in financial statement analysis provide scant instruction on how to assess this statement — even though it is among the major ones. It contains the information necessary to answer our questions above, and is typically reported just after the balance sheet, which it supplements.

To aid in your understanding, we’re going to start with Nvidia’s Consolidated Statement of Shareholders’ Equity as reported for 2017. See below:

You likely notice that the presentation above differs from financial reporting such as the income statement. This is because the Statement of Shareholders’ Equity shows two types of data:

  • the effect of corporate activity on both shares outstanding; and
  • the dollar amounts associated with these decisions.

Hence the information can only be shown in a matrix, rather than in the more traditional linear style.

Let’s start our work by looking at line 1, above. It lists the basic (not diluted) shares outstanding at the beginning of Nvidia’s fiscal year: 539 million shares. Moving horizontally across the first line of the table, each figure that we encounter is a summary of the stock-based activities that have affected the shares outstanding of 539 million shares throughout the history of the company.

The next figure encountered (in Column 2) is the par value of the 539 million shares, labeled “Amount” by Nvidia. You might already know that  par value is typically around $0.01 per share (or less), which accounts for the low value (roughly $1 million) assigned to such a large number of shares.

Moving to the right, we next encounter  “Additional Paid-In Capital (APIC)” of $4.170 billion. Importantly, this value is not the market value of these shares at the end of the reporting period. Instead, it is the market value on the date these shares were issued and that is over the par value of $1 million.

Combining these two figures of $1 in the par value column with $4,170 in the APIC column allows us to compute the first number we want to know:  the average price per share at issuance. For Nvidia this amount is roughly $7.73, or ($1 + $4,170 million) ÷ 539 million shares.

The next number we encounter, $4,048 million, is  “Treasury Stock.” This is the total value of all shares repurchased by the company throughout its history.

As you can see above, the figure is shown as a subtraction from the value of the equity of the company. This makes sense because the business spent cash to purchase these shares in the secondary market, and now the shares are no longer in circulation. Notice that as 2017 concluded Nvidia had, over time, spent almost as much on share buybacks as the amount raised by selling its equity to the public. Wow!

Specifically, Treasury Stock ÷ Shares Issued and Outstanding is: $4,048 ÷ $4,171 = 97.05 percent.

Continuing across Nvidia’s first line you can see the accumulated “Accumulated Other Comprehensive Income (Loss)” of $4 million. This figure is the summation of the miscellaneous things that affect income to shareholders after net income is calculated.

According to Nvidia’s disclosures, these transactions include items such as available for sale debt securities and cash flow hedges (among many other things). Given the small size of these amounts relative to the total value of their shareholders’ equity, there’s not a need to scrutinize this account in detail.

Our last figure is the accumulated profits of Nvidia throughout its history, or “retained earnings.” Again, note the large size of its stock buybacks in proportion to its profits through time.

Moving on Down

Continuing down the Consolidated Statement of Shareholders’ Equity we encounter the important accounts (highlighted in red) of “Issuance of common stock from stock plans,” “Share repurchase,” and “Stock-based compensation.” Before we assess these in detail, let’s briefly discuss the other accounts listed:

  • “Retained earnings adjustment” due to new accounting policies. The effect of this exclusively affects (as the title suggests) Nvidia’s retained earnings.
  • “Other comprehensive income (loss)” for a single year is added to the total from all prior years.
  • “Net income” for the year 2017 is added to retained earnings, as per usual.
  • “Issuance of common stock in exchange for warrants” issued. Note: there is both a share effect of 44 million shares and an APIC effect.
  • The “conversion of convertible debt” that also has a share effect of 23 million shares, and an Additional Paid-In Capital (APIC) effect. Note: the share effect seems to have been hedged by Nvidia (see below).
  • “Tax withholding related to vesting of restricted stock units.” Again, there is a share count effect and an APIC effect.
  • “Exercise of convertible note hedges.” These were probably options that Nvidia used to manage the effect on its common shares outstanding from its issuance of a convertible note.
  • “Cash dividends declared and paid” by the company to its common stockholders. Notice the very low percentage that dividends are relative to share repurchases: $261 million versus $739 million, or 26.1 percent of monies returned to equity holders.
  • Last, “Reclassification of convertible debt conversion obligation,” which appears to be an accounting adjustment.

Analysis of Share Issuances and Buybacks

With an understanding of the types of information that appear on the Consolidated Statements of Shareholders’ Equity, we can now conduct our analyses of Nvidia’s share issuances and buybacks. Below is the full set of Nvidia’s statements 2017 through 2021. (Our 2022 data is incomplete because Nvidia has not yet published its 2022 10-K. That should happen sometime soon, and analysts can reperform all this work at home with that fresh data then, if you’d like.)

First, let’s get one of the categories colored in red out of the way: stock-based compensation. You probably know that many businesses like to incentivize their employees by issuing stock options; this line item accounts for those activities, in part.

How so? GAAP accounting requires that companies expense the value of their options issuance on their income statements, which lowers net income. Since this is a non-cash expense, however, there must be an offset. For stock-based compensation, it’s an increase in Nvidia’s additional paid-in capital account.

We can confirm this by looking at the company’s footnote disclosure (Note 4 – Stock-Based Compensation; although in some years Nvidia puts this information in Note 3). It confirms the amount of Nvidia’s non-cash expense for its options program exactly offsets each year the amount of stock-based compensation shown under the APIC column.

Phew! We have covered a lot of ground already. The material above lets an analyst understand how the Consolidated Statement of Shareholders’ Equity is reported, and what that information means.

Next week we’ll continue with Part II. How should one analyze all that information properly, to understand whether buybacks are the best use of a company’s capital?

(Again: Calcbench has designed a template to let you import as-filed Statement of Shareholders Equity from any firm; we encourage you to download it. Data diehards can also use our API if you’d prefer. Drop us a line at and we’ll get you squared away.)

This is an occasional column written by Jason Apollo Voss — investment manager, financial analyst, and these days CEO of Deception and Truth Analysis, a financial analytics firm. You can find his previous columns on the Calcbench blog archives, usually running every other month.

We examined corporate filings for the week of March 8-15 to see what was being said about Russia. There were 291 firms mentioning Russia in their recent filings.

Where it was mentioned:

These mentions have a variety of stories behind them. Focusing on risk factors, some companies may have a general warning like Kronos Worldwide (KRO) who mention the Russia-Ukraine conflict in the context of “risks associated with global and regional economic, political and regulatory environments”.

Whereas others may be more specific, like Salesforce (CRM) which was concerned that “retaliatory cyber attacks stemming from Russia’s recent invasion of Ukraine, including ransomware and distributed denial-of-service attacks”, or IDT which has “significant number of R&D personnel in Belarus”, and are concerned about the impact on personnel and their work, or Century Casinos Inc. (CNTY) which were concerned about the impact of the conflict with Russia on “discretionary consumer spending”.


Examining the potential economic effects of the Russia-Ukraine conflict, we examined how many companies reported to have subsidiaries in those countries (listed in Exhibit 21 of the 10-K). In the annual 2021 filings, 192 firms listed at least one subsidiary in Russia, and 99 firms listed at least one subsidiary in Ukraine.


In addition to examining subsidiaries, the potential impact on companies could be examined by studying the segment data coming from the segments footnotes, and looking for companies that mention a segment that includes Russia. For example, Pepsico (PEP) reported almost $3.5B in revenue from Russia or ArcelorMittal (MT) who reported $4.1B in revenue from Ukraine.

Calcbench has all this information, and much more, right at your fingertips, waiting to be discovered. If you want to know more, please let us know at

Tuesday, March 8, 2022

This month Calcbench and our esteemed contributor Jason Voss will be taking a deep look at share repurchase programs, and how analysts can evaluate whether such programs are the best use of a company’s capital.

Today we kick off that project with our latest research note, which examines just how much cash companies (all public companies, regardless of size) have spent buying back shares over the last 10 years. The report can be downloaded from our Research page, free to all.

The headline findings:

  • Total dollars spent on share repurchase programs from the start of 2012 through end of 2021: $6.52 trillion.
  • Average amount spent on repurchases per “firm quarter” (that is, every incident of a company buying back shares): $155.3 million.
  • Number of quarterly share repurchases that exceeded $1 billion: 1,296.

By the way, those 1,296 super-sized repurchases were only 3.1 percent of all repurchases in the decade we studied, but accounted for 55 percent of all dollars spent on repurchases.

In other words, only a small fraction of repurchases from some very large firms accounted for a majority of all money spent on repurchases across the entire decade we examined.

Another way to frame the trend is to look at the companies spending the most money on share repurchases. We found seven firms that each spent more than $100 billion buying back shares, led by Apple ($AAPL) and its $479.5 billion. See Table 1, below.

Altogether, those seven companies spent $1.247 trillion on share repurchases. That’s 19.1 percent of the $6.52 trillion that all companies spent on repurchases in the last decade — just from those seven big spenders, and particularly Apple.

We also charted median spending on repurchases versus average spending, which helps to see how outliers like our seven firms above affect overall trends. See Figure 1, below.

The outlier effect is clear and enormous, and it accelerated through the course of the decade — especially in 2021, when the stock market and corporate profits were both going gangbusters.

The rest of our research note looks at the number of companies buying back shares in any given quarter compared to dollars spent that quarter; a distribution chart of how many companies were spending how much money on repurchases (spoiler: most weren’t spending that much); and of course, a list of big spenders beyond the seven we named above.

So download our note, give it a read, and look for more content soon explaining how an analyst might decipher whether repurchase programs really are the best use of company capital.

Tuesday, March 1, 2022

Russia’s invasion of Ukraine will have enormous implications for financial markets. Calcbench can help people unravel those implications in several ways.

Russia’s invasion of Ukraine is first and foremost a moral and humanitarian nightmare. Before anybody even begins to contemplate the financial repercussions of Vladimir Putin’s actions, we first urge people to remember that fact.

That said, the West's response to Putin’s invasion is already having significant ramifications in the financial markets. Businesses and financial analysts will need to anticipate those ramifications, since resilient economic performance here will be crucial to defeat Putin there. So today we have a quick rundown of how analysts can use Calcbench to better understand companies’ exposure to Russian interests and to analyze disclosures about Russia risk.

Geographic Segment Disclosures

One place to start is the Segments, Rollforwards, and Breakouts page. This page allows you to search companies’ various segment disclosures — including geographic segments, and the companies that list Russia as one such segment.

You can do this by selecting a group of companies to study and then entering “Russia” in the Segment column. This will return a list of companies that disclose financial data associated with Russia, and what that data is. See Figure 1, below.

Be warned that relatively few companies make segment disclosures about Russia specifically (we found only seven in the S&P 500 for 2021), because most report Russia operations as part of some larger geographic segment such as Europea, EMEA, or Asia. You can always search those terms too, and you’ll get results; but they won’t necessarily provide much insight into Russia itself.

When you do find Russia disclosures, you can also use our Trace feature to go from the listed number to the specific footnote disclosure. For example, you can see in Fig. 1 that Cummins ($CMI) reported $334 million in distribution sales to Russia last year. We traced that back to the original disclosure and found this, Figure 2:

OK, Cummins had 4.3 percent of distribution sales in 2021 come from Russia — and that $7.74 billion is only a fraction of Cummins total revenue, which was $24.02 billion last year. So its Russia revenue really isn’t material, although clearly Russia was a growth country for Cummins; sales more than doubled in two years. Plus, as we said, there is the moral consideration of whether to keep doing business in Russia at all; and there may be regulatory questions soon enough if the Biden Administration enacts a total export control ban against Russia.

Anyway, that’s one example of how to use Calcbench to dig through Russia disclosures, to sharpen your questions for company CFOs and CEOs on the next earnings call.

Subsidiary Disclosures

Analysts can also use our Interactive Disclosures page to dig up footnote disclosures for what corporations are saying about foreign subsidiaries with ties to Russia.

Here, you need to first define your search group (say, the S&P 500), and then select “Subsidiaries” from the Choose Footnote/Disclosure Type pull-down menu on the left. You’ll then need to enter “Russia” in the text search field and check the “restrict to specified disclosure type” box immediately under the text search field. See Figure 3, below. (The “Subsidiaries” disclosure choice is near the top, in the section for 10K/Q disclosures.)

We did all that, and found 97 companies in the S&P 500 that listed Russia subsidiaries for 2021. That compares to 98 such companies in 2020, so the number had been holding steady through last year.

What do these results actually tell you? Foremost, the name of the subsidiary operating in Russia. For example, Cooper Companies ($COO) has a subsidiary named CooperVision RUS Corp. Akamai Technologies ($AKAM) lists a subsidiary operating under the same name in Russia. Citigroup ($C) has a Russian subsidiary called AO Citibank, and also reports that AO Citibank exists five levels beneath the parent company.

You won’t necessarily know any more about those subsidiaries, such as revenue, operating profit, or PP&E value, unless the company reports those numbers elsewhere — but you will have the subsidiary name, which gives you a start to ask more probing questions.

Other Disclosures

You can also use the Interactive Disclosure page to search for any disclosure about Russia. (That is, don’t restrict your search to mentions of subsidiaries or anything else.) That might be quite valid right now, in the immediate and chaotic environment, as companies start making observations about Russia in their Management Discussion & Analysis disclosures.

We might also see companies making 8-K filings in the near future as they announce divestitures from Russia projects or declare impairments for Russian assets they will no longer carry on the books. Citigroup, for example, coincidentally filed its 2021 10-K just yesterday, and disclosed a total of $10 billion it expects to write off in Russia. That’s not much for a bank with $2.3 trillion in assets, but given the political risks around Russia those details are still notable.

So we recommend that you check our Recent Filings page often, to find the latest from firms you follow.

Tuesday, February 15, 2022

SUMMARY: A handful of tech giants are spending staggering amounts of money on research and development. Calcbench tried to put that spending into context.

Corporate executives always say that spending on research and development is a high priority for their firm. Still, when you look at the numbers actually reported — it’s clear that a few select firms are responsible for a staggering amount of R&D spending, that dwarfs what many other firms spend on, well, anything.

We recently pulled the R&D numbers reported by 157 firms in the S&P 500 that have already published their fiscal 2021 financial statements. Altogether, those companies reported spending $301 billion.

The top spenders, shown below in Table 1, should be no surprise.

The five tech giants above collectively spent $114 billion in R&D costs, or 37.9 percent of all R&D spending for our entire sample of 157 firms.

Moreover, these R&D budgets alone are larger than scores of other S&P 500 companies’ whole operations. For example, Google’s R&D spending of $31.56 billion exceeds the revenue of 128 firms. Apple’s spending of $21.91 billion exceeds the revenue of 111 firms. Even Intel’s R&D spending, the cheapskate of the group with $15.19 billion, exceeds the revenue of more than 90 firms.

Looking at R&D Another Way

But wait, you say! Is R&D spending measured in absolute dollars really such a useful metric? Wouldn’t R&D spending as a percentage of revenue be more informative, since that gives us a sense of how much the company is trying to innovate new products and services?

Fair point. So we recalculated the numbers to express that percentage — and got a very different set of results. See Table 2, below.

The results were so different that we added the tech giants in a second tier to give a sense of perspective; none of them come anywhere near this set of firms that devote significant portions of revenue to R&D. The results here do make a certain amount of sense; you’d expect pharma companies like Incyte and Vertex to spend boatloads on research, and likewise for semiconductor manufacturer Xilinx (which, by the way, was just acquired by AMD).

R&D Spending Per Employee

Back to the tech giants. Another way to understand their gobs of R&D spending is to express it per employee. So we did that, and found these results in Table 3.

Another interesting set of results, because Facebook spends more on R&D per person than any rival tech giant by far. Something to think about next time you see one of those commercials for its Metaverse and wonder what that’s all about; clearly the company is devoting a lot of money to that future, uh, whatever it is.

What About Amazon?

Astute observation! Nowhere on this list do we see that other tech giant,

That’s because Amazon doesn’t break out and report its R&D spending like other firms. Instead, the company categorizes its R&D spending as “technology and content.” Why? Here’s how Amazon explained things to the SEC in a comment letter on the subject back in 2017.

Our business model encourages the simultaneous research, design, development, and maintenance of both new and existing products and services. For example, our teams are constantly working to build new Alexa skills and simultaneously maintain current skills, and these activities … are not easily distinguishable operationally. We focus on innovation and customer obsession and manage the total investment in our employees and infrastructure in order to improve the customer experience. We believe our technology and content presentation is most representative of the investments we make on behalf of our customers.

Um, whatever, Amazon; this post is already long enough so we’ll let your analysis speak for itself. For the record, however, Amazon did spend $56.05 billion on technology and content in fiscal 2021. That was 11.9 percent of its $469.82 billion in revenue, and only about $35,000 per employee. (Amazon has 1.6 million employees, reflective of its vast commercial operations.)

Thursday, February 10, 2022 is often on our mind here at Calcbench because it’s such a fascinating source of financial reporting nuance. In 2018 we told you about an SEC comment letter Amazon received, prodding the company about disclosing the number of Prime members. In 2020 we compared AWS (Amazon Web Services) to Microsoft’s Intelligent Cloud.

Today we want to look at Amazon’s ($AMZN) latest 10-K filing, which arrived on Feb. 4. Specifically, we want to examine the company’s segment footnote.

Amazon recognizes three operating segments: North America and International, which are both retail segments; plus AWS, its cloud-hosting segment. So for practical purposes Amazon really operates as two halves: its retail operations and AWS.

Let’s start by examining the trends in both revenue and operating income for the segments. As you can in the chart below, revenues for all three segments seem to be increasing substantially in the last three years.

At first glance this might look quite promising, but if you squint at the operating income numbers you see another story. Operating income has declined for both the North America and International segments (it’s actually negative for the International segment), and has increased for AWS.

This made us wonder: How significant is AWS for Amazon, in revenue and operating income? Very significant, apparently. As you can see from the charts below, although AWS accounts for only 13 percent of Amazon’s revenue, it accounts for 69 percent of the company’s operating income.

Fair enough, but more questions come next. Generating all that income requires assets, so how much Property, Plant, and Equipment (PP&E) do each of Amazon’s operating segments have? Well, Figure 3, below, shows that the North America segment has 52 percent of all PP&E, while AWS has only 27 percent.

Using operating income as a proxy for net income, and using PP&E as a proxy for operating assets, we can calculate the return on operating assets (ROOA) for each segment and the sales generated by those assets— a metric known as asset turnover.

As you can see in Table 1, below, AWS has an extremely high ROOA, but much lower asset turnover. That’s because AWS needs lots of hardware to support all that computing power.

So there you have it: the best of times and the worst of times for Amazon. It begs the question: Is retail a tale of the past and computing a tale of the future?

Summary: SEC comment letters are not immediately released to the public, but a study of long-term trends shows that the delay in disclosure of those letters has fallen dramatically. 

In our previous post we examined the number of comment letters the SEC sent companies in 2020 and 2021, as well as how many replies companies sent back. An interesting point is that comment letters sent to or from the SEC do not immediately become public. Rather, there is a delay of at least 20 business days (per SEC rules), and sometimes the delay is longer.

So what is the average delay in publication of comment letters? Calcbench decided to investigate.

When one searches the EDGAR database of SEC filings, you can find SEC comment letters to a company and all the ensuing correspondence — but you can only find when the letters were filed, not when the letters became public. Fear not, however! Calcbench has you covered.

Calcbench our database notes when comment letters are released by the SEC, and records both the date in which the letters were filed and the date those letters were released. The delay between when a letter is filed and when it was released is interesting.

We examine the trend in the average delay between when a letter is filed to when it is released to the public over the last 10 years. As you can see in Figure 1, below, that delay has declined markedly — from roughly 140 days in 2011 to only 60 days in 2021. (Keep in mind that some comment letters filed in later years have not yet become public, which will increase the delay, but the trend seems to be very telling.)

We can also see that the delay in releasing SEC comment letters is slightly higher than the delay in releasing companies’ correspondence.

Are these patterns true for both large and small companies? We decided to answer that too, so we ran the same analysis for S&P 500 companies specifically.

As you can see from Figure 2, below, the same meaningful decline holds true for large companies. When comparing the delays, we note that the average delay is actually shorter for the S&P 500 than it is for all companies. The average delay we observe for large firms in 2021 is roughly 40 days, compared to 60 we saw for all companies.

Many factors can affect the delay in the SEC releasing the letters to the public, such as:

  • The time it took the SEC to complete its review;
  • The time it took the company to respond to the SEC;
  • The complexity of the topic reviewed;
  • The efficiency in the review or the resources spent on it.

Our analysis doesn’t examine the reasons for the factors that affect the delay, but for those of you interested, Calcbench has the data. Visit Calcbench’s filings page to see when filings are files and released. If you want to search the comment letters, you can do that in the disclosures page.

FREE Calcbench Premium
Two Week Trial

Research Financial & Accounting Data Like Never Before. More features and try our Excel add-in. Sign up now to try the Premium Suite.