Netflix: An Updated Valuation
What do I have to believe to expect a average long-term return? And what's my base case suggest about forward returns?
The single greatest error I have observed among investment professionals is the failure to distinguish between knowledge of a company’s fundamentals and the expectations implied by the company’s stock price. - Mauboussin
I posted some mostly qualitative thoughts about Netflix after its first-quarter results (Netflix: To the Victor Go the Spoils). But however competitively advantaged Netflix is, however long their growth runway may be, whatever their long-term margin and cash flow profile might look like can be irrelevant if our expectations are already reflected in the stock. If that’s the case, the company can perform as well as we expect but we still might only earn average returns over time. Or if even higher expectations than ours are already baked in, the company could meet our expectations but the stock can still underperform. That’s the disaster we’re trying to avoid.
Instead, we want to ensure a modest or, ideally, an unrealistically negative outlook is baked in at the current stock price. And a more realistic outlook would suggest the stock is worth a multiple of where it is trading. That way there is plenty of room for the future to miss our base case while still earning an acceptable return—our margin of safety. And with any luck, the future turns out better, in which case we’re buying a boat. In other words, price is a key lever we use to skew the odds in our favor.
It should go without saying that regardless of how good a company’s future prospects look, the stock’s future prospects depend on the price. Everyone knows that’s true. But it’s often either forgotten, overlooked, or we convince ourselves our expectations were too conservative and revise them ever higher. Sometimes that is justified, but it’s also easy to take it too far because we often want it to be true. So it is important to remain as objective as possible and not succumb to endowment bias—where we overvalue what we already own—and confirmation bias—where we seek out information we already agree with and dismiss contrary evidence. And if you’re imagining a love story between you and your favorite stock, it’s in your mind alone. A stock doesn’t know you own it and won’t hesitate to put a hole in your portfolio if you’re holding the bag when investors realize the baked-in expectations were far too high. I think the goal is to try to continuously and dispassionately assess the long-term odds and be willing to cut ties if the circumstances justify it.
When we do the work to figure out what sort of expectations are baked in at any given price, it becomes apparent that the implied expectations can differ wildly between one price and another. That seems underappreciated to me. It seems like a more a common alternative approach is to not actually value the business at all but assume accelerating growth deserves a higher multiple and decelerating growth deserves a lower multiple. That sort of heuristic is easy to apply and may even make sense on average but is incredibly prone to error. For example, what would you pay for the stock market equivalent of a beautiful oceanfront rental property that generates high and accelerating rents but is 100 feet from a cliff that has been eroding 5-15 feet per year? That could look appealing in the near-term but deserves a low valuation to the extent one expects the erosion to continue.
High Level Valuation Thoughts
Given the value of a business is the present value of all future cash flows, I focus on how the key value drivers could play out over the long term. Obviously, any forecast is certain to be wrong to some extent, which explains why I underwrite several possible scenarios that we can all ponder. Importantly, with this process the value drivers are explicit and can be debated and continuously revised with new information. All the cards are face up on the table. I prefer that to picking a multiple where the value drivers are implicit, hidden from scrutiny, and frankly, we have no idea what value drivers we’re betting on.
For example, what does a given multiple imply about Netflix’s subs, ARM, content spending, and operating leverage looking out 5 or 10 years? Someone out there may have a thoughtful answer but I certainly wouldn’t—not without first laying out the value drivers and discounting the cash flows. Only then could I say, “My base case values it at X, which happens to be a Y multiple. But the multiple is an output, not an input, of the valuation process. The inputs are the long-term value drivers.”
At a high level, Netflix’s revenue can be forecast using two variables for each of the four geographic segments. We simply forecast 1) paid memberships and 2) average revenue per member (“ARM”) for each of UCAN (U.S. and Canada), EMEA (Europe, Middle East, and Africa), LATAM (Latin America), and APAC (Asia-Pacific).
While UCAN’s fastest days of paid member growth are certainly behind it, I think investors are too quick to assume UCAN is a no-growth segment. There are 131.2 million households (2022 Census data) in the U.S. and another 15.1 million in Canada (2021 data) for a total of 146.5 million UCAN households. That’s going to grow over time and the already high broadband penetration rates are likely to approach 100%.
The Joint Center for Housing Studies of Harvard University forecasts U.S. households increase to 139.8 million by 2028 and 149.4 million by 2038. The Canadian Housing and Mortgage Corporation (CMHC) forecasts Canadian households increase to 16.2 million by 2026, 17.0 million by 2031, and 17.7 million by 2036. Using those two forecasts, the UCAN segment as a whole should have something like 160.7 million households in a decade and 172.0 million a decade beyond that.
For context, Netflix had 74.4 million paid memberships in UCAN at the end of March. So that would be 51% household penetration today and 43% household penetration looking out a couple decades. Both of these figures seem meaningfully below where they will be at maturity. For one thing, there’s 30 million households in UCAN sharing a paying subscriber’s password right now. Some of them will convert to paying subs. Getting half of those would drive up the long-term penetration to 52%. That would be a start.
One interesting data point is pay-TV penetration reached 91% of U.S. households in 2010 before cord cutting began. While pay-TV had live sports and news, it has also cost a multiple of a Netflix subscription. If Netflix could eventually penetrate 90% of 172 million households, the UCAN segment could reach 155 million households, more than double it’s current level.
My base case is more conservative than that—actually so is my super bull case—but it’s not impossible. Keep in mind Netflix is now starting to target a wider range of demographic segments with the lower-priced ad tier. Management has also strongly suggested it is also developing a free ad-supported tier. As YouTube or Facebook’s roughly three billion users show, there are few limiters to adoption when a compelling global service is made free to the user. And Netflix management has also proven itself to be completely willing to adapt and evolve. Live sports and news are the two genres missing from Netflix compared to linear, but we could see live sports on Netflix one day to the extent it can monetize ads better than others and therefore economically justify outbidding the next highest bidder. News is not out of the question either. There’s no reason Netflix will not account for much more of our viewing than it meets today on a long enough timeline, which would have interesting monetization implications—especially because of the fixed cost nature of Netflix’s content. YouTube shares just over 50% of its ad revenue with the content creators, a variable cost. In contrast, Netflix’s fixed cost content plus a compelling free-to-the-user global offering helping attract billions of viewers should drive far higher margins at maturity.
As for ARM, I think there is meaningful room for better monetization. For starters, expanding to lower income demographics with ad tiers not only increases the TAM but ad tiers also monetize better. On the last call, management said the $6.99 Basic with Ads (BWA) in the U.S. already monetizes better than the $15.49 per month Standard tier after only a few months, which suggests to me it will almost certainly monetize better than the $19.99 Premium tier as Netflix improves ad targeting, measurement, verification, and eventually brings the ad tech in-house.
Priced In/Reverse DCF
The scenario I build to back into the current stock price includes the following key assumptions:
382 million subscriptions monetizing at $16.48 per month in 20 years. That’s an anemic 2.6% subscriber growth rate, a 1.7% ARM growth rate, and a 4.4% revenue growth rate over that period.
Netflix household penetration rates across UCAN, EMEA, LATAM, and APAC reach 53%, 21%, 25%, and 15%, respectively. Those are very modest 20-year improvements compared to 51%, 13%, 19%, and 5%, respectively, given broadband penetration and connected TV penetration increasing, a large increase in Netflix’s overall screen time share given the death of linear, a doubling of Netflix’s annual content spending, the existence of low cost and even free ad-supported tiers all around the world, and the precedent that pay-TV penetration reached 91% in the U.S. despite costing far more than Netflix.
Content amortization grows from $14 billion last year to $28 billion in 20 years. That’s a huge increase in content spending that does not result in much revenue growth. As a percent of revenue, content amortization drops from 44.6% to 37.9%. That is a very modest and, arguably, unrealistically low level of operating leverage on content spending over that time span.
Operating margins expand to 28% in a decade and 35% a decade beyond that due to fixed cost leverage. That’s still comparable to HBO’s operating margin in 2017, as I discussed two weeks ago in Netflix: To the Victor Go the Spoils, despite Netflix having massive economic advantages over HBO in 2017.
An effective tax rate rising from mid-teens today to 28% in 20 years
A 10% discount rate and 5% terminal year free cash flow yield
Given I think that scenario is priced in, I think we’d earn average returns in NFLX even if something that disappointing played out. To the extent Netflix’s future is better than that, I think we’d earn better than average returns.
Base Case
In my base case, I assume UCAN subs increase from 74.4 million last quarter to 97.1 million in a decade and 114.1 million a decade beyond that. That’s still only reaching 66% of UCAN households, which I think leaves a lot of room for upside beyond my base case given Netflix will have locked down password sharing, continues to pump out more must-watch content across genres and geographies, competitive streaming value propositions stagnate or worse, and Netflix’s price no longer becomes a barrier to anyone around the world with lower-priced and free ad-supported tiers.
As for ARM, my base case assumes UCAN ARM compounds at a 4.0% average annual rate over time. That could be very conservative. First, it’s meaningfully lower than the 9.7% average annual rate it has grown ARM over the last five years and prior to that. Second, the growing mix of ad-tier subscribers should be a meaningful tailwind to ARM. Third, a 4.0% rate isn’t that much faster than a typical rate of inflation. And finally, Netflix’s engagement is almost certainly going to increase over time from its ~10% share of screen time in the U.S. and U.K. and lower in other markets as streaming continues to take share from linear. The more people watch Netflix, the more pricing power it will justify.
In EMEA, LATAM, and APAC, I do the same forecasting with households, factoring in increasing broadband penetration, and assume household penetration can reach 30%-34% looking out a couple decades. Today, these rates are about 5%-20% depending on the region. I also assume ARM compounds at 3.4%-4.1% average annual rates over time, depending on the region. Frankly, I think this leaves a lot of upside on the table.
Big picture, this assumes Netflix has 441 million subs and a $16.25 global ARM in a decade. And 706 million subs and $21.98 a decade beyond that. These numbers sound big but there are already a billion or so broadband households, which will grow over time. And 20 years is a really, really long time. This is 5.8% average annual subscriber growth and 3.2% average annual ARM growth, which sounds entirely plausible to me. That works out to $183 billion of revenue in 2042.
Like I talked about in Netflix: To the Victor Go the Spoils, how much content spending that will require is an open question. How much could its current offering support? Well, I’d say somewhere between its current revenue of $32 billion and $46 billion, which is an estimate that assumes full monetization of password sharers. But there’s also the increasing trends of broadband penetration and connected TV penetration. And again, lower-priced or free ad tiers, which have the potential to take sub growth to another level. These things should bring in much more revenue without necessarily requiring that more content spending.
My base case assumes about $25 billion of content amortization when Netflix hits $100 billion of revenue in 2034 (25% of revenue) and about $34 billion of content amortization on $183 billion of revenue in 2042 (~18% of revenue). For context, content amortization was 45% of revenue last year. But frankly, it’s possible Netflix might be spending less—maybe a lot less—than $34 billion on content annually in 2042. For one thing, the competitive environment has never looked more favorable—and that is likely to only continue given scale drives everything in this business. The idea that another streaming service is going to outspend Netflix was always fantasy to me, but that view now seems to be eliciting less skepticism. Second, the more scale Netflix gets, the more bargaining power they have with content providers. Is Netflix really paying market rates for Drive to Survive or Full Swing or Headspace Guide to Meditation or the upcoming FIFA docuseries? Or are the true beneficiaries of that content being available on-demand to one billion or more people meaningfully funding them? This is on top of the fact that the best acting and showrunner talent prefers their works to be seen on the largest stage. Look at what being on a hit Netflix series does for a previously no-name actor’s Instagram followers. Gaten Matarazzo went from unknown to 18.3 million IG followers on the back of one show. Night Agent has been out for barely over a month and previously unknown Gabriel Basso suddenly has 300k IG followers. Put that show on another streaming service and he’d have fewer followers.
As for all other non-content amortization expenses—other cost of revenue, marketing, tech and development, and G&A—I have it falling from 37% of revenue this year to 24% of revenue in a couple decades. A lot of these are fixed or semi-fixed costs that should scale nicely as they have started to already.
Put it all together and Netflix’s revenues would 6x over the coming two decades and operating margins would expand from ~20% to approaching 60%. Believe it or not, I think that leaves a good amount of upside on the table. For example, one assumption behind that is ~30% international household penetration looking out two decades even when there’s almost certainly going to be a free ad-supported offering. Imagine free services like Facebook or YouTube topping out at 30% international penetration. Facebook already has 3 billion MAUs, which is almost half of the global population ex-China. What’s even more impressive is 35%-40% of the world’s population isn’t even online today. So you could say Facebook’s penetration of the global internet-connected population seems closer to 70%. So my base case is nowhere near as good as it could get for Netflix.
Here’s the usual scenario analysis and valuation PDF file with all the value drivers laid out:
And here’s the downloadable Excel file to play around with for those of you subscribed to IE with Models:
And finally, the valuation summary that starts with the valuations derived from the file above:
The first thing I’d do is examine the assumptions underlying the Priced In scenario, which I discussed above. That tries to answer the question, “What’s priced in at $320 per share?” If that scenario plays out, I’d expect average shareholder returns from here. If you don’t think those assumptions are achievable or think it’s too hard to handicap, then Netflix is an clear pass.
Aside from that, you may notice my valuations have moved up from February. My February valuation baked in an 11% discount rate and 6% terminal year free cash flow yield. Netflix is the only company I model that I was still burdening with those versus my usual 10% and 5%, respectively. Frankly, I think the reason I did that was I thought the valuations I derived were just too damn high relative to the stock price using 10% and 5%, respectively. And I was very happy with my other assumptions and had no interest in getting any more conservative on those.
Is that a good reason to use 11% and 6%? No. I’ve changed them back to 10% and 5%. That alone explains 84% of the valuation increase from February to May. And the passage of time accounted for some as well. So the underlying value drivers didn’t change much.
Disclosure: Long NFLX
Disclaimer: This post is for entertainment purposes only and is not a recommendation to buy or sell any security. Everything I write could be completely wrong and the stock I’m writing about could go to $0. Rely entirely on your own research and investment judgement.