Twitter's $2.9bn black hole

"We have a number in the accounts that doesn't reflect economic reality"

Podcast

This month’s podcast interviews Dan McCrum, the award winning FT journalist who exposed Wirecard as a fraud. Once a stockmarket darling valued at over €25bn, Wirecard crashed to worthless in June 2020, when its auditors could not confirm €1.9bn of cash. Dan pursued the story for six years and has now written a book about the saga. “Money Men” is an amazing account of how the company used every tactic possible to disguise its activities. This is an extraordinary tale and Dan, a brilliant raconteur, explains the highs and lows of his struggle to reveal the truth.

We also speak to market participants on the buy and sell side to get their perspective. Mark Hiley at The Analyst talks about the 40+ sell notes they published over the period while Freddy Brick of Muddy Waters Research recounts a phone call in which Wirecard tried to bribe Carson Block several million dollars not to publish on the company.

You really should buy the book. I would prefer you to go to a bookshop but if you must, you can find it on Amazon.co.uk or Amazon.com (affiliate links – I donate the proceeds to charity).

Stock Based Compensation

In a recent article I explained that tech investors were likely to see further dilution  of their investments because after significant share price falls, many companies are likely to issue new stock options to employees.  My original piece, entitled Stock Based Compensation – the Disappearing Expense, explained that stock based compensation (SBC) presents a problem for investment analysts in several ways:

1                     Almost every company adds back SBC in its calculation of adjusted earnings. Earnings are therefore overstated, as payroll costs would have had to be (much?) higher without this issue of stock to employees.

2                     Because the SBC expense is a non-cash item, it is added back in the calculation of operating cash flows. Cash flow multiples are therefore understated. Although buybacks are often deployed to offset the dilution from the issue of shares to employees, this cost is not reflected in Free Cash Flow (FCF) multiples. I explained a solution in that earlier article.

3                     Valuations should at least factor in the impact of the options outstanding on market cap, but analysts often forget to make this adjustment. Some analysts include it in the number of shares used to calculate earnings per share (EPS), but there is no consistency.

I then explained that accounting standards in respect of stock based comp leave a lot to be desired. Their intention is to give a measure of the opportunity cost at the time of grant – hence if there is a cost in year one of an option vesting over three years, the International Financial Reporting Standards (IFRS) say that the cost is spread over the three years. US GAAP (generally accepted accounting principles) is similar, as the two standards were introduced at the same time and the two standards setters were actually talking to each other.

The original charge is not varied if the share price quadruples in yeartwo, which is obviously a problem, as the cost to shareholders has increased significantly. Similarly, if the share price goes up, the real cost of the pool increases, but this is not reflected in the P&L.

This would be less of a problem if there was adequate disclosure and analysts could calculate the effective opportunity cost each year, which is what shareholders should be concerned about. Spoiler alert – the disclosure falls short.

So we have a number in the accounts which doesn’t reflect the economic reality. That’s not ideal but not that unusual. (IFRS 16 being a good example.) We can get round this. Last time, I explained how to adjust the cash flow and the FCF yield calculation. The other trick I use is to inflate the future share count by the historical growth in the number of shares in issue – I look at this before share buybacks so that I can estimate what level of dilution is likely in future. Of course, dilution may be higher or lower in subsequent years.

Real-life Examples of SBC

I originally illustrated the difference between the stock option expense and the real cost to shareholders by looking at the difference between:

-          The P&L charge for the option expense over a period

-          The number of shares issued each year to employees times the average share price in the year. This is only a rough estimate of the real cost to shareholders of issuing stock to employees, but it’s all we can do with the information we have.

I concluded that for some companies, the option expense in the P&L significantly understated the actual cost to shareholders.

Let’s look at some examples; For most of the group in the table, SBC is the principal adjustment to EBITDA (earnings before interest, taxes, depreciation and amortization. Less so for Netflix because Sentieo adjusts the adjusted EBITDA, which I shall explain in my next post on the company.

Stock Based Compensation

Source: Behind the Balance Sheet from Sentieo data

On SBC to revenue, Netflix is not as prodigious a user of stock based comp. The main users here are Twitter, Salesforce and Meta Platforms. For Twitter, the options expense exceeds its reported EBITDA and it’s almost as large as Salesforce’s reported EBITDA. And for IHS Markit and for Meta it’s significant at nearly 20% of reported EBITDA.

Quantifying the Issue

In my original article on this subject, I pretended that at the start of 2014, you, dear reader, were a successful venture capitalist and you owned 5% of Twitter and 5% of Facebook/Meta (OK, a very successful VC). You would have owned the following:

Source: Behind the Balance Sheet from Sentieo data

You would have done much better with your Meta stock, as we know, which would have trebled (previously, it had nearly quintupled) as opposed to Twitter which actually declined, and that’s after the Musk bid! But let’s look at what happened in the eight years from 2014 to 2021 in stock based compensation (all data from Sentieo):

Twitter’s charges for SBC in those eight years amounted to $4.2bn; for Meta Platforms the total was $36.4bn. Large numbers and equivalent to 11% of the opening market capitalization of Twitter and 26% for Meta Platforms. Had you known this at the start of 2014, that might have coloured your view of the shares, but let’s consider the relative significance of SBC for the two companies.

The conventional way of assessing this might be to compare these charges with the respective P&L statements which is quite easy to do, as illustrated in the above table. More important, and more difficult, is to assess the economic impact of the issue of this stock to employees.

Source: Behind the Balance Sheet from Sentieo data

The table shows the SBC expense from the accounts which for Twitter over the eight years amounted to 18% of revenues, 63% of adjusted EBITDA, and the company reported a cumulative net loss. For Meta Platforms, the same ratios are 9% of revenues, 15% of adjusted EBITDA and 4 times net income.

To understand the real economic effect, however, we have to leave aside the accountants’ notional value of the options, and look at the number of shares against which options were issued, how much cash was received (not much usually) and what the company would have needed to spend on buybacks in order to offset the dilution; this last measure is a fair economic representation of the true cost to shareholders at the time.

In the case of Twitter, Sentieo’s standardised cash flow helpfully gives you the cash received from the exercise of options  – just $75m in eight years. By analysing the Consolidated Statements of Stockholders Equity, we can assess the number of shares issued each year on the exercise of stock options and in relation to the vesting of Restricted Stock Units.

There is a difference here in that I am calculating the stock actually issued, rather than the options granted, which forms the input to the SBC expense line. But in practice, the real economic cost to the business is the stock actually issued – this is what actually dilutes shareholders, not the notional options charge in the P&L.

Over the eight years, 230m shares were issued by Twitter to its employees and with an average calendar year share price which fluctuated between $16 and $54, we estimate the value of the stock issued was $7.2bn or $7.1bn after taking cash receipts into account.

Source: Behind the Balance Sheet from Sentieo data

The SBC expense in Twitter’s P&L totalled $4.2bn, versus the actual economic cost of $7.1bn net – a huge difference. This suggests that using the actual stock issued in the year at the average share price looks likely to be a better indicator of the true cost to shareholders.

Doing the same comparison for Meta Platforms, we get 352m shares issued in the period and a total value (at average prices in each year) of $60bn, which compares with $36bn of SBC in the same period. For Twitter, our estimate is 72% higher than the total reported in the accounts and for Meta Platforms our estimate is 67% higher – these are significant differences for two tech stocks pulled at random.

US companies are incentivised by the tax legislation to use share options because the options are a deductible for tax purposes. For some companies, the use of stock options is likely more significant than the examples portrayed here. But in all cases, the current practice of in effect ignoring the cost and taking the benefit of the expense in a reduced tax rate, is sloppy analysis. This is a double whammy benefit to the companies’ apparent valuations.

This week, free subscribers have received the same letter as paying subscribers who will be getting some additional content in the coming weeks. This is a commercial exercise for me, so I would encourage you to subscribe. If you can’t afford it, please stay for the free edition; and remember, each article costs less than a latte.

But just so they don’t miss out, for paying subscribers I have added my review of Dan’s book below as well as one of the obvious accounting red flags.