Twenty odd years ago, I was a significant junior partner in a UK clone of US DVD rental company, Netflix. I persuaded easyJet founder Stelios, then seeking to expand easyCinema, to inject additional capital but the deal fell through. Back then, capital raising for early stage tech companies was more difficult. The business was eventually sold to deeper pocketed LoveFilm before its purchase by Amazon.
Founder and good friend Ben asks me still why didn’t we just buy Netflix stock. It would have been easier and had we held it, far more profitable. I have occasionally made money trading Netflix, but this has been a total failure relative to a buy and hold strategy. Making mistakes is part of the investor’s journey, and omissions like this are particularly galling.
Forgive me therefore if at times this article sounds a little pained or even prejudiced. I hope that this experience has not coloured my views but I should at least declare the possibility. I am a Netflix customer and a great admirer of Reed Hastings. But I have less respect for the company’s accounting and one of its policies is the subject of this blog.
We pose the question in this article, “Is Netflix Cooking the Books?”. We believe that Netflix has been aggressive in its accounting for some time. Consider the way it used to account for DVDs: Blockbuster treated them as stock, while Netflix (we believe having changed its policy) treated them as fixed assets. This allowed Netflix to report a higher level of earnings and also much higher operating cash flow: under this policy, DVD purchases were treated as an investing item and hence did not affect the operating cash metric. This was a much less conservative policy, in our view.
In this blog, we look at the way Netflix accounts for content. This is a complex area, and we explain below our approach, using techniques from our online course How to Read a Balance Sheet and our Forensic Accounting training course for professional investors, and justify our conclusion. First, let me summarise for those who find accounting dull.
the chart (repeated in larger size below) shows that the life of a Netflix film or series has doubled in the last few years. This means that its amortisation policy has been getting less conservative. An average life of 6 years is inconsistent with Netflix’s claim of 90% write down in the first four years.
nobody cares too much about Netflix’s earnings today – we are not trying to make any revelations about the share price; in our view, the rating appears rich, but is hard to assess and this means very little, particularly in a COVID-19 environment.
the new form of audit report in the US now conforms with other jurisdictions in that the auditor is required to identify key contentious areas. This is a huge step forward. It’s hardly a surprise that Netflix’s auditor should express concern about its accounting for content costs. (For US investors new to the concept, the auditor is obliged to note this area, becasue of its size and degree of subjectivity.)
Below, we discuss the audit report, review the life of content assets, test whether the numbers are consistent with the stated policy using different simulations, review the accounting policy wording and past changes, before concluding. Its technical but you don’t need to read all the 10-K extracts!
AUDIT REPORT 2019 10-K
Below is an extract from 2019 Netflix 10-K Audit Report. You can skip to Our Interpretation 10 lines below if you find accounting dull.
Description of the Matter
As disclosed in Note 1 to the consolidated financial statements “Organization and Summary of Significant Accounting Policies”, the Company acquires, licenses and produces content, including original programming (“Streaming Content”). The Company amortizes Streaming Content based on factors including historical and estimated viewing patterns. Auditing the amortization of the Company’s Streaming Content is complex and subjective due to the judgmental nature of amortization which is based on an estimate of future viewing patterns. Estimated viewing patterns are based on historical and forecasted viewing. If actual viewing patterns differ from these estimates, the pattern and/or period of amortization would be changed and could affect the timing of recognition of content amortization.
How We Addressed the Matter in Our Audit
We obtained an understanding, evaluated the design and tested the operating effectiveness of controls over the content amortization process. For example, we tested controls over management’s review of the content amortization method and the significant assumptions, including the historical and forecasted viewing hour consumption, used to develop estimated viewing patterns. We also tested management’s controls to determine that the data used in the model was complete and accurate.
To test content amortization, our audit procedures included, among others, evaluating the content amortization method, testing the significant assumptions used to develop the estimated viewing patterns and testing the completeness and accuracy of the underlying data. For example, we assessed management’s assumptions by comparing them to current viewing trends and current operating information including comparing previous estimates of viewing patterns to actual results. We also performed sensitivity analyses to evaluate the potential changes in the content amortization recorded that could result from changes in the assumptions.
Netflix’s auditor has acknowledged that this accounting policy is tricky and discloses that they have tested the policy and its sensitivity. They haven’t quantified anything, however. Presumably, House of Cards would have had an accelerated write-down, although Netflix claims in its accounting policy that it has not had such a write-down (see below). And the vast sums that Netflix has been spending on new series by definition cannot be tested as Netflix has not been around for long enough in its present form to determine their true “shelf life”, or digital equivalent. It’s highly likely that the auditors have no idea of the correct policy and it’s perfectly possible that the management don’t either.
What we can do, however, is to apply a couple of forensic accounting tests:
ascertain if the company looks like it is accounting for content in a conservative fashion
check if the company has made any changes to its accounting policy. We shall cover that later, after we have reviewed the policy.
CONSERVATIVE AMORTISATION TEST
Our investor training courses spend significant time on intangible asssets, and the first test we recommend when looking at capitalised intangible assets is to determine the average life by comparing the amortisation charge with the average asset. Our calculation here takes the average long term asset (the life would be even longer if current assets were included), calculated from the cash flow statement lines and adding back the cumulative amortisation. The trend for Netflix is shown in the chart below, which clearly shows that Netflix has become less conservative in its content life assumptions, significantly so:
The chart’s trend is steepened by the fact that Netflix has had an increasing proportion of its content asset in production – these assets are naturally not depreciated. Even adjusting for this, however, the trend has been of lengthening asset life, with the average life now being longer than would be consistent with our reading of the policy. Netflix appears open and helpful about its content accounting policy, even producing a video on the subject. In a helpful presentation, Overview of Content Accounting in January 2018 (since updated) indicated that you should not apply this test, and gave the following justification:
1 the content library is presented net of amortization, not on a gross basis
2 content is amortized on an accelerated basis
3 amortization in any given period is also affected by the mix of content as different categories of content are amortized on different schedules (based on historical and projected viewing patterns)
We review these points in turn below:
(1) As the content is presented net, our calculation of the average life adds back the cumulative amortisation to date. It’s possible that some assets have been fully amortised and should be eliminated from the calculation. Therefore, we may have included in our calculation the amortisation of an asset which has been fully amortised. This would lengthen the life in the chart, but would not be sufficient to create this trend, as Netflix has been increasing its content spend significantly – the fully amortised assets would almost certainly be a low proportion of the accumulated spend.
Netflix Content Spend
(2) The use of an accelerated depreciation policy could impact the trend in the life. Spending more on big new releases would accelerate the depreciation and we would expect the life to shorten, not lengthen.
(3) Mix will clearly affect the data, but we would not expect the impact seen.
Netflix could have argued that content assets are created as cash is spent, and often production schedules are such that cash is increasingly invested longer in advance of release date. That would be reflected in a build-up of an asset, ahead of amortisation. This would be a logical explanation for a slight change in the life, but hardly to the degree in the chart.
Here is the methodology as explained by Netflix in its IR’s Overview of Content Accounting.
There are two key takeaways from this:
Content is amortised over a title’s window or 10 years at most
Content should be 90% amortised after 4 years
These are tested below.
TESTING THE NETFLIX POLICY USING DIFFERENT ASSUMPTIONS
We initially ran a few tests using the 10 year life and compared our estimates with the Netflix accounts. We ran several scenarios using simple depreciation assumptions, with higher depreciation in the early years, trying to see which would come closest to the actual outcome, and we show just three of those below:
A chart of these three scenarios, compared with the actual charges, is shown below:
Case 1 results in estimated amortisation which is significantly higher than the actual in the later years, although closer in the earlier years. Writing off 50% in year 1 seemed a logical number and in fact probably as low as that number could be.
Increasing the first year percentage only increases the gap between expected and actual. Case 3 is much closer to the actual result today – but we are only writing off 30% in year 1 and are only 50% depreciated by the end of year 2, leaving another 8 years of even depreciation. We used straight line to simplify the test, so that we could understand the impact of varying the parameters. This is close enough to the actual result that it seems a not unreasonable representation of what Netflix is doing.
This is quite different from Netflix’s claim that it amortises 90% of the cost within the first four years.
Before we examine the results of our further testing, we explore the accounting policy note for Streaming Content, and highlight six points noted (1) to (6) in the text with our comments in italics. For those of you that find accounting dull (we know there are one or two), skip to Our Interpretation 15 lines below.
ACCOUNTING POLICY (2019 10-K)
STREAMING CONTENT POLICY
The Company acquires, licenses and produces content, including original programming, in order to offer members unlimited viewing of TV series and films. The content licenses are for a fixed fee and specific windows of availability. Payment terms for certain content licenses and the production of content require more upfront cash payments relative to the amortization expense. (1) Payments for content, including additions to streaming assets and the changes in related liabilities, are classified within “Net cash used in operating activities” on the Consolidated Statements of Cash Flows.
The Company recognizes content assets (licensed and produced) as “Non-current content assets, net” on the Consolidated Balance Sheets. For licenses, the Company capitalizes the fee per title and records a corresponding liability at the gross amount of the liability when the license period begins, the cost of the title is known and the title is accepted and available for streaming. For productions, the Company capitalizes costs associated with the production, including development costs, direct costs and production overhead. Participations and residuals are expensed in line with the amortization of production costs.
Based on factors including historical and estimated viewing patterns, the Company amortizes the content assets (licensed and produced) in “Cost of revenues” on the Consolidated Statements of Operations over the shorter of each title’s contractual window of availability or estimated period of use or ten years, beginning with the month of first availability. (2) The amortization is on an accelerated basis, as the Company typically expects more upfront viewing, for instance due to additional merchandising and marketing efforts and film amortization is more accelerated than TV series amortization.
On average, over 90% of a licensed or produced streaming content asset is expected to be amortized within four years after its month of first availability. (3) The Company reviews factors impacting the amortization of the content assets on an ongoing basis. The Company’s estimates related to these factors require considerable management judgment. (4)
The Company’s business model is subscription-based as opposed to a model generating revenues at a specific title level. Content assets (licensed and produced) are predominantly monetized as a group and therefore are reviewed in aggregate at a group level when an event or change in circumstances indicates a change in the expected usefulness of the content or that the fair value may be less than unamortized cost. (5) To date, the Company has not identified any such event or changes in circumstances. (6) If such changes are identified in the future, these aggregated content assets will be stated at the lower of unamortized cost or fair value. In addition, unamortized costs for assets that have been, or are expected to be, abandoned are written off.
(1) “Payment terms for certain content licenses and the production of content require more upfront cash payments relative to the amortization expense.” This would be a legitimate reason why the rate of amortisation of the asset may look low for Netflix and we have tested for this, as we explain below.
(2) “Company amortizes the content assets (licensed and produced) in “Cost of revenues” on the Consolidated Statements of Operations over the shorter of each title’s contractual window of availability or estimated period of use or ten years, beginning with the month of first availability. The amortization is on an accelerated basis.” Accelerated amortisation is obviously the correct route, but as we showed above, the actual results do not confirm heavy acceleration, but more on this below.
(3) “On average, over 90% of a licensed or produced streaming content asset is expected to be amortized within four years after its month of first availability.” Again, this does not appear to be supported by our tests above.
(4) “The Company’s estimates related to these factors require considerable management judgment.” You can say that again! Netflix is at least being open and honest here. When we see this sort of statement in an accounting policy, it’s a red flag, as it means that there is scope for companies to manipulate.
(5) Content assets (licensed and produced) are predominantly monetized as a group and therefore are reviewed in aggregate at a group level when an event or change in circumstances indicates a change in the expected usefulness of the content or that the fair value may be less than unamortized cost. This seems a peculiar policy. They have a lot of shows, but we would expect them to use data on individual shows’ viewings to determine accelerated write-offs. This is a concern.
(6) To date, the Company has not identified any such event or changes in circumstances. This is a surprising comment – presumably House of Cards would have required a write-down, post the Kevin Spacey controversy, unless it was already in the books at zero, which is not consistent with their policy nor the overall depreciation rates. This is a concern.
FURTHER ANALYSIS OF THE AMORTISATION RATE
We reran the numbers on Netflix’s amortisation using the reducing balance method. For those unfamiliar with the term, it simply describes a method where you generally use a faster rate but depreciate the net balance. You may have a car which loses 40% in its first year so you depreciate at 40% using the reducing balance method:
The car costs $10,000 and is worth $6,000 after the first year – in year 2 the depreciation is 40% of $6,000 and the car is worth $3600 at the end of year 2 and so forth. We ran the numbers for Netflix’s content spend initially at a 50% rate (remember the 90% write-off after 4 years) but this gave far too high a result. As you can see in the chart, the closest fit was at a rate of just 25%, which means that after 4 years the asset is not quite 70% depreciated, well short of the 90% advertised by Netflix. Again, we see a difference between our projections and the reported results, with the accounts data not fitting well with the stated policies.
Our standard research process to check that the numbers stack up with the policy does not work for Netflix, and we do not know why. In other circumstances, we would conclude that the company is inflating earnings.
CHANGES IN ACCOUNTING POLICY
In our Forensic Accounting Course, we suggest that when doing this sort of deep dive into the accounts, it’s helpful to check whether companies have changed the wording of the accounting policies; sometimes there is a hint here as to whether they are cheating. We reviewed Netflix policies for streaming content, from 2019, 2018 and 2017. In 2019, they appeared to stop considering content as current assets: Added “The Company recognizes content assets (licensed and produced) as “Non-current content assets, net” on the Consolidated Balance Sheets.” This is in accordance with accounting standard ASU 2019-02, per Netflix IR.
Netflix also added the comment in (3) above that “On average, over 90% of a licensed or produced streaming content asset is expected to be amortized within four years after its month of first availability”; this is not a change in itself as they had previously made the comment in the investor relations documentation as we saw in the earlier slide. But by including this in the accounting policy note, that makes it subject to audit. Therefore, either this is correct (and by implication, that our analysis is wrong – it’s important for analysts to recognise this possibility) or the auditor is not doing its job properly (also not unknown).
In the prior year, after “The amortization is on an accelerated basis, as the Company typically expects more upfront viewing, for instance due to additional merchandising and marketing efforts” Netflix added “and film amortization is more accelerated than TV series amortization”. We have not investigated the relative spend, but the occasional TV series can have an incredibly long shelf life (think Dad’s Army or Sgt Bilko or Friends”, but we find this surprising as on average, a great movie would likely last longer than a great TV series.
Given the change in the current content assets, we wondered if part of the gap between estimated and actual amortisation was due to larger payments being made up front and a deferral of amortisation. We therefore reran a couple of scenarios assuming that only half the spend resulted in output in the current year.
There clearly is a delay between spend and amortisation and this is a factor we should have considered in the earlier tests. We estimate that about 40% of spend last year was for future content, up from just over 10% (or 6 weeks) 3 years ago. If we assume that all the spend is deferred by 6 months, the results are illustrated in the chart. Again our estimates of amortisation are much higher than the actual in spite of an overly generous timing assumption.
Overall, there does not appear to be anything sinister to emerge from this check.
DOES THIS MATTER AND SHOULD YOU SELL NETFLIX?
Netflix bulls will argue that today’s earnings are irrelevant to the share price and Netflix could write off content on day one, report huge losses and the share price would be no different. As the chart shows, Netflix has been trading at eye-watering multiples on traditional metrics – we cut the scale on the chart between 0 and 200x and Netflix has spent large parts of its existence outside these parameters. The market is looking beyond this. Clearly, it would not matter to the stock if Netflix was reporting much lower profitability today.
Moreover, Ben Thompson in the Stratechery blog pointed out that conventional methods of accounting for content in the case of a company like Netflix rather miss the point, because content serves more than one purpose, as in the case of Bird Box which “does triple duty for Netflix:
For current customers, Bird Box provides two hours of entertainment and a pass into popular culture. It is a cost of goods expense.
For prospective customers, Bird Box makes Netflix more attractive for the same price. Or, to look at it another way, it lowers Netflix’s customer acquisition cost. It is a marketing expense.
For marginal customers, Bird Box is a reason to stay on the platform. It lowers Netflix’s customer retention cost. It is an operating expense.”
Of course Ben Thompson was not really talking about accounting, but he was making a great point about the inadequacy of conventional accounting techniques for many of the current tech giants (for more on this subject read Harvard Business Review’s articles here and here). This is why I think Netflix would be better to present its numbers with a more aggressive amortisation of content cost. It would not affect the share price and would reassure cynics like us that they weren’t trying to be cute with their accounting.
We are comfortable with our arithmetic and readers can easily replicate our calculations, but we are confused about motive. Usually, companies have a reason for accounting gymnastics and that appears to be missing here.
But lets say that we are correct and that Netflix is being aggressive in its content accounting and that its accounts should reflect a much higher amortisation if its policy were being correctly implemented. Indeed, as we mentioned at the start of this article, looking back at its accounting for DVD purchases, we believe that Netflix has been aggressive in its accounting for some time.
If we are correct, then there is a more important issue, and that is one of trust. Netflix tried to represent itself as more profitable and more cash generative than reality in the DVD era. What if that continued today? That would be puzzling but it could undermine investor faith. I stress that I am not accusing Netflix of cheating, as that would make no sense. But we think their accounting merits a better explanation and will send this blog to Reed Hastings and his team. We shall report back if/when we receive a response.