Friday, 31 January 2020

Facebook - the bear case

Facebook is currently a stock beloved by both growth and value investors alike. It is an esteemed member of the FANG club - an aristocrat of the modern digital era. Everyone seems to own it. I don't like it. I think the stock is much riskier than most investors appreciate - particularly at this point in the cycle. (I am a day late in publishing this article, as the stock is down 6% to US$210 post its 4Q19 results - albeit this is small fry compared to the aggregate potential downside; I tweeted the WSJ Casper article referenced below over the weekend, and in related comments referenced the risks to Facebook, and began writing this post on Sunday. Unfortunately I was too slow to complete it).

The things to like about FB are trivially obvious. Unlike most of it media-industry brethren, its content comes from UGC (user generated content) and is therefore produced for free by its billions of active users, and the company is then able to sell ads against the significant traffic this UGC generates. The persistency of its UGC is also supported by the well-discussed (read 'excessively discussed') 'network effect'. Furthermore, the amount of money it can make from selling digital ads is being supported by rising digital advertising penetration, as well as the company's ability to target users, due to its trove of user data. The company has become a putative duopoly with Google in the online digital advertising space, and is seen as a reliable, incumbent FCF compounder. Investors own the stock as they think (1) the company can keep growing at high rates; and (2) the company's earnings and growth outlook is low risk, as it is supported by durable, structural drivers.

As is always the case in markets though, obvious merits are well known and well priced in. To make money in markets you have to prod for outcomes that are less likely but still possible, and are not being priced in, and I think there are many things to be legitimately concerned about for FB in this respect at the present time. As I have pointed out many times in prior posts, it does not matter that the below analysis might not be a base case likelihood. It only matters that it is a not-negligible potential scenario that is being excluded from investors base case return expectancies. That is enough to conclude the risk/reward associated with owning the company is poor.

Although FB is a large and important company, the core drivers of its value are actually extremely simple: (1) the number of users it has and the amount of time those users spend on its platforms (what I call 'eyeball-hours'); and (2) the degree to which it can monetise those eyeball-hours through selling digital ads, which is a combination of the ad load (number of ads per hour), and the average rate card (cost per ad). The ability of the company to successfully customise/target ads to specific demographics also contributes importantly to the latter. The company's cost discipline also matters, but is a somewhat less interesting and important issue on a stand-alone basis than the company's revenue outlook. If revenues grow a lot, costs matter less; if revenue disappoints, they matter a lot more.

With respect to #1, the company's global DAUs/MAUs continue to steadily grow. However, the company has already captured the vast bulk of the world's high-value users - those that can afford a smartphone and access to the internet, and can afford to spend a lot of money on products and services advertised on its online platforms, justifying the ad spend. However, just as important as the number of users is the amount of time they spend on the platform (the other important determinant of eyeball-hours), and transparency on this all-important metric is significantly lower.*

One of the problems/headwinds for the latter is something I have discussed in past blog articles - the increasing saturation of the 'attention market'. Although tech investors are accustomed to believing in limitless growth horizons, the reality is that there are only 24 hours in a day, 7 days in a week, and 52 weeks in a year, and so there is a finite amount of available attention for the world's online media companies (encapsulating social media, gaming, streaming, and other sinks for our attention) to compete over. Furthermore, the penetration of 3G/4G-enabled smartphones around the world is now high - particularly amongst affluent/high value users. The void of previously unused attention of people around the world waiting at bus stops, or queueing at the bank or airport, has now been filled. In other words, the available eyeball-hour market has now already become largely saturated.

One of the important consequences of this is that future growth in the attention market is going to become 'zero sum' in nature, as the amount of white space uncaptured attention disappears. This is an important change vis-a-vis the past, where all new online media companies could simultaneously grow by capturing some of this previously unutilised attention. If there is one thing that is virtually certain, it is that new and interesting online apps/ways to spend our time will continue to emerge over time. The recent rapid ascent in TikTok, for instance, is a case in point worth noting. Out of nowhere, it has come to capture a not-negligible portion of the attention market. That innovation will continue is a given, and has always been the case, but what has changed is that from here, the success of one new platform will become cannibalistic and come at the expense of the success of another, rather than being merely additive to the overall attention industry.

On some metrics, the effect this is having on old-hands like Facebook is already becoming tangible - at least with respect to the company's core Facebook platform (i.e. excluding Instagram and WhatsApp & Messenger). The below infographic is something I shared on Twitter a month or so ago, and is interesting on many levels. It highlights that FB's core Facebook property has already entered into a fairly rapid decline in usage in recent years, and an important contributor to this has likely been the increasing saturation of the attention market, and cannibalistic effect of competing social media and other mobile application platforms beginning to be felt. This is a problem not just for Facebook's all-important eyeball hour share, but also for its network effect flywheel, as it depends on UGC. Less visits equals less UGC, which leads to less frequent visits (and scrolling dwell-time), as there is less UGC to consume, leading to even less visits, etc.

Luckily for FB, they were smart enough to buy Instagram for a song ($1bn), and Instagram is still on the up and up, keeping FB's combined user metrics on a positive incline. However, the rapid recent decline in the core Facebook platform's visit frequency nevertheless serves to highlight that now that the attention market has been saturated, permanent incumbency in the social media space/attention market - let alone permanent growth - should not be automatically assumed. New platforms will always be launched - the Instagram, Snapchat and TikTok's of tomorrow - and the market will continue to mutate, evolve, and fragment, and at present the popularity of the core Facebook platform is clearly trending in the wrong direction - particularly because history suggests that once a platform's popularity is past its peak, comebacks are rare.**

The bulls may acknowledge that the attention market is becoming increasingly saturated, but they will counter by arguing that FB's attention is still significantly under-monetised. A major component of this argument is pointing out that the company's US ARPU is still substantially above the ARPU generated in international markets. However, what if the actual truth was not that FB is under-monestising abroad, but rather that it is actually over-monetising its US user base?

Relevant in this regard was an excellent recent WSJ article about VC-backed online mattress company Caspers (see here - paywall). Like a large number of hyped, VC-backed Unicorns, Caspers has been growing rapidly, but chalking up considerable losses. The article notes that one of the challenges the company has faced is that almost all of its key business functions are outsourced (most notably all mattress procurement/manufacturing), while the company is sourcing customers by paying for online ads. All of these business functions are easily replicable, and hence the company has seen the emergence of hundreds of immitators sourcing products from the same suppliers, and competing for the same Google keywords and Facebook demographics. This had driven marketing and CACs through the roof, hammering profitability.

All these companies are losing money, as FB and GOOG are capturing all of the value. It makes sense - if you can buy and have delivered a mattress from a third-party manufacturer for say $200, and sell it online for $400, it makes sense not just to pay FB and GOOG $20 to acquire a customer, but also $50, $100, $150, and ultimately $200. Indeed, it even makes sense to pay $250 if you are being valued on revenue growth rather than profitability, as most VC-backed Unicorns are. Growth at all costs is their mantra.

Market and economic history is replete with examples of companies and industries that grew rapidly but incinerated capital, because if supply grows faster than demand, and companies lack any sustainable competitive advantage, it doesn't matter how fast demand grows - everyone loses money. This has happened in everything from railways in the 19th Century; automobiles in the 1920s; fibre-optic cable in the late 1990s; wireless networks in the 2010s (at least with respect to new investments to upgrade networks, which have not been monetised); solar panels in the early 2010s; shale oil & gas in the 2010s; and more recently, in cannabis and lithium. Stocks typically surge as investors anticipate a boom in demand, only to collapse later as supply grows faster than demand and losses mount.

Companies like Caspers are no different, and their large losses also reflect their extremely low level of organisational competencies. Caspers is not a mattress company. They have no manufacturing; no mattress technology; nothing. All they do is get manufacturers to to produce mattresses for them; set up an online storefront; and then spend a tonne of money branding and marketing those mattresses through online channels. They are simply marketing businesses. Because they have very little value-add; no unique organisational competences; and no competitive advantages, they don't have the necessary ingredients for robust economic returns - namely, being able to do something other people/organisations can't easily do. They are facing lots of competition from other well-funded start-ups willing to lose money, and it is therefore no surprise they are losing money. Anyone can pay to buy ads online, and brand and sell products manufactured by real companies with real capabilities (designing and manufacturing mattresses, which requires some skill and installed capital).

In an era of unprecedentedly loose capital market conditions in the VC space, there are many, many companies like Caspers out there - essentially online marketing businesses that while growing fast, are losing a tonne of money, because they have low quality business models that are easily replicated, and there is an excess supply of capital in their industry chasing growth. This is unsustainable, and when the VC funding spigot is turned off - which will happen when the very high return expectations of their end investors are inevitably disappointed, dampening the appetite for more (particularly from investors such as large pensions/endowments that can ill-afford the losses) - a large fraction of these businesses will fail, and massive consolidation will occur (much as occurred in 2000-03). Indeed, there are some signs we may already be reaching this point, as the WeWork blow-up has severely damaged Masayoshi Son's credibility, which is reducing his ability to raise additional capital.

This time is seldom different. Every single growth boom in history that has featured a vast number of new entrants supported by a wave of capital all vying for a slice of the pie, have eventually seen the pace of supply growth outpace demand; mounting losses; and an eventual severe downturn that leads to mass bankruptcies and eventual consolidation. It happens every time, without exception.

The reason I discuss the Caspers of this world here, however, is that the risks relate not just to Caspers, but also to the likes of Facebook, as it has been an important marketing and customer-acquisition channel for the said loss-making start-ups. How much of FB's ad revenue is driven by wastefully-unsustainable marketing dollars being thrown around by VC-backed startups desperate for growth without any regard for profitability is unknown, but it could well be substantial. Caspers alone is spending more than $100m a year on online marketing. Facebook's revenues are the other side of the unsustainable marketing and CAC expense line items of the said loss-making VC-backed Unicorns. 

What this means is that in a severe retrenchment in VC start-up land, a very considerable amount of online advertising spend could suddenly disappear, as many of Facebook's key customers suddenly substantially reduce their online ad spend. Furthermore, when end markets eventually consolidate (e.g. if Caspers ends up the winner by raising IPO capital and buying out all of its competitors for close to zero), there will also be less bidding competition for key words (relevant also for Google) and key ad slots, which could reduce prices as well as bidding volume.

How much of FB's US ARPU is being driven by unsustainable CAC spending is unknown, but there is a notable gap between FB's monetisation in Europe and the US & Canada ($13.21/quarter vs. US$41.41/quarter, respectively, in 4Q19), despite Europe also being a relatively high-income region. The bulls have argued this is due to FB being under-monetised in Europe and the RoW (FB's global average was US$8.52/quarter in 4Q19), but the gap could also reflect FB being heavily over-monetised in the US, reflecting rampant Valley financing of worthless tech start-ups willing to spend billions of dollars on Facebook and Google ads to acquire customers and sustain rapid top-line growth at any cost.

When you think about it, US$41.41 per user in the US & Canada - a user base that includes kids with limited financial resources - is a staggering amount of money. It would have to influence several thousand dollars of incremental annual product and services spend a year, for every single user (including kids, and the many very light users of FB), for that to be sustainable and economically justified. That's a tall order to sustain, let alone grow at rapid rates, and it is entirely possible inflated Unicorn CAC spend is a key reason the metric is so high.

If in a major tech bust, FB's US ARPU was to fall to European levels, the company's revenues and pre-tax earnings earnings could fall catastrophically, by as much as US$20bn, reducing pre-tax income from US$30bn to US$10bn. Impossible you say? FB's revenue was only US$56bn (vs. US$71bn in 2019) as recently as 2018, and US$40bn (US$30bn less than current levels) as recently as 2017! US$10bn is also about as much money as the company was making in 2016, and the company's cost base has meaningfully inflated since then, as compliance/content moderation spending has grown (although there may be some downside flexibility in a bust scenario given how much costs have risen).

One of the major mistakes investors made in the late-1990s tech bubble was to ignore the cyclicality of IT capital spending. It was argued and believed that the secular growth trends driving rising internet adoption and networking would be so strong that it would overwhelm any conceivable cyclical factors. People were wrong. There was a collapse in corporate IT spending for several years; fiber networks were so significantly overbuilt that excess capacity remained for as much as 10-20 years post bust, wiping out many leveraged long-distance fiber companies and forcing consolidation; and online advertising spend (less important/systemic at the time) also collapsed.

Analogously, today investors are assuming the secular trends towards higher online activity and digital marketing are so strong that cyclical factors can be ignored, and yet (1) advertising has always been a notoriously cyclical industry, as marketing spend is one of the first things companies cut in a recession when they need to shore up cash flow; and (2) digital marketing has now already grown to a relatively large share of total global advertising spend, which means the extent to which secular growth drivers (digital spend rising as a percentage of total advertising spend) can overwhelm cyclical factors (the aggregate rise and fall of marketing spend through all channels) is also naturally declining.

If we go through a severe tech VC/Unicorn bust, which I think is virtually inevitable and have called for multiple times on this blog, I think investors could easily be surprised by the extent to which this dents FB and GOOG's marketing revenues (the trends discussed in this article are also relevant to GOOG's marketing revenue, but unlike FB, it has a more durable long term core business monopoly, and also has a valuable and rapidly growing cloud business (supported by its AI capabilities), YouTube, etc, which make it less risky than FB long term, but still very vulnerable short term).

If FB's ad revenues were to unexpectedly drop by say 20% - particularly if FB is unable to get its cost base under control, as it continues to invest heavily in 'content moderation' to comply with increasingly strident calls for online content moderation by regulators - the company's earnings could easily halve (increasing the stock's P/E to 50x), and the stock could be absolutely massacred. If it fell to 15x reduced earnings that would be a 70% decline. Sound impossible? A 20% drop in the company's ad revenue would only take its top line back to levels seen in 2018, and a fall in the multiple will be amplified by the current level of crowding into tech names, and the false expectations many holders have (that FB is a 'low risk' structural grower immune from cyclical risks). Falling prices also kick into gear various self-reinforcing psychological and momentum/flows forces, which is why historically, the gravity of the bust is usually inversely proportional to the gravity of the preceding boom. And all this is before considering any fears that may emerge about FB losing market share to other competing social media platforms such as TikTok.

At this point, I will inevitably be asked, does that mean I'm short FB? No it doesn't. I don't like shorting stocks for reasons I have discussed in past blog articles, and more importantly, I never have 100% confidence that any of my views are right. That's not the way I think/invest; I don't form a fixed view of what I think is going to happen in the future, and then invest on the assumption that I am right, which is what most investors do. The future is radically indeterminate, and trying to predict the shape of future online ad spend is extremely difficult, and it would be foolish to try to do so with any degree of certainty - particularly for a generalist, non-specialist such as myself, that has not dedicated any serious research time to the issue. I have spent a small fraction of one percent of my research time thinking about these issues, as I am focused on uncovering promising opportunities in unusual corners of global markets, and don't waste my time on areas that are already well picked over/expensive/crowded.

I instead consider a variety of potential futures, to form a probability distribution of sorts, and then look at what's priced in/generally believed. In the case of FB, I see a situation where everyone is long and has signed up to the view that this is a bulletproof structural growth story, and that this will inevitably remain the case for many years. They might be right. But I am much less sure, and have concluded from my analysis that there are important risk factors that appear to be under-appreciated by investors. That's all I need to know to know the stock is not a buy on an expectancy basis - I don't need to know for sure what happens. However, if I short the stock, I do need to know for sure what's going to happen, and also need to be sure the stock doesn't go parabolic in the interim - something that happened in the bubble and wiped out many short sellers who were forced to cover (and is currently happening with Tesla), even though they were ultimately right in their fundamental views.

So I won't be going short, but I definitely won't be going long.


*Also notable here is that time spent on messenging apps is of considerably lower value than on Facebook and Instagram, as it is very difficult to effectively monetise the former. 

**Incidentally, the other notable collapse in web traffic has been from Baidu - another popular stock owned by many value investors, which appeared cheap relative to Google, but which never succeeded in making and distributing any cash of note, as it squandered its prosperity on many loss-making upstart ventures - perhaps because it recognized vulnerabilities in its core business that most investors overlooked, which are now becoming manifest. Unlike Facebook's share price, however, which has continued to rise and rise, Baidu's has sharply declined.