Facebook and Google: How Online Ads Will Do In the Next Recession

Investors of Facebook (FB) and Google (GOOGL) have been worried about privacy issues and now anti-trust issues for the past couple years. I think an even more important question though, is how cyclical are their ad revenues?

Theoretically, advertising should be highly cyclical. Advertising acts as weapons in business wars for market shares gains, which primarily happens during an expansion.

Think about all the subscale/unprofitable startups that spend over 100% of revenue on sales/marketing. They’re willing to spend lavishly on advertising under the logic that market share begets more market share (network effect); that they will be able to retain customers, and therefore the lifetime value of those customers should exceed sales and marketing expenses. These companies are generally cash flow negative and depend on external funding. In a liquidity risk-off situation that funding get cut off, and their advertising spends on FB and GOOGL could come to a sudden halt.

In short, companies (especially SaaS guys) are treating advertising expenses as capital expenditures, and thus advertising revenues should reflect more cyclical characteristics.

This points to a downside scenario for Facebook and Google (disclosure: I’m long both). We will at some point hit a recession. Google and Facebooks’ revenues may not just stop growing but actually decline (volume and price both drop). Then you have operating deleveraging, lower margins, and earnings and free cash flow gets destroyed.

What Happened During 2008-2009

The easiest way to see how something might fare in a recession is check how well it did during 2008-2009. There the data is surprisingly good – online ads actually did well during the great recession. Google’s revenue growth slowed but never turned negative. 

Here’s a Harvard Business Review article from 2009 that discusses the strength of online advertising:

“Despite a deepening recession, marketers spent 14% more on online ads over the first three quarters of 2008 than they did over the same time frame in the previous year.”

There are structural and cyclical components to this phenomenon. The structural part is easy - digital ads were in its early innings during that time as people were still moving online.

The cyclical component is also intuitive. Companies become more budget conscious in a recession. They want more measurable ROI and more precise targeting as opposed to a scattershot approach. Both factors favor digital ads over broad based TV ads.

People also lean on their social networks more in a downturn, for job networking as well as emotional support. This resilient usage of social networks makes it a good media for ads. 

Differences Then vs Now, Conclusion

Granted, the current situation has important difference compared to 2008. Online ads are no longer in the early innings – 2019 will mark the first year digital ads make up over 50% share of advertising market. Furthermore, Google’s revenue was growing 50-60% a year heading into the 2007 downturn compared to the mid to high teens growth rate it now shows.

Both Facebook and Google have made strong moves into video advertising, where they still have the advantage of more quantifiable ROI and precise targeting against traditional TV. However, the competitive edge is less overwhelming nowadays with smart TVs like Roku that can also do targeted advertising.

The two companies will also have to fend off new entrant Amazon.

Despite these differences, some of the behavioral patterns should still hold. In a recession I think people will still network more, lean on their social circles more, and probably spend more time online searching for jobs and information.

Weighing these factors, my recession scenario for Google and Facebook is as follows:

• Ad prices should take a hit to maintain ad buyer ROI – (as conversion rates go down in a recession).
• Volume down 5-10%.
• Overall mid teens decline in revenue.
• Margins and earnings will take a hit and stock prices will get punished as well. 

Looking beyond though, there’s no question FB/GOOGL have the balance sheet to survive a recession. Once recession recovers, ad volumes would spike back up.

So for the long term investor, the question goes back to structural growth opportunities. Here the prospects are bright. Both companies continue to develop new products, both for consumers and for advertisers. Both have established strong presences in digital video advertising and taking the lead in trialing new formats to optimize advertiser ROI. From that perspective, they’re still in the early innings.

Notes from 1Q18

I was 100% allocated to stocks coming into 2018, had an ecstatic January, then a crappy February that led me to cut exposure. I continued to cut exposure in March and now sitting on more than 30% cash.

The low volatility regime has decisively ended, which means all sorts of previously overvalued stocks can come down on the slightest negative news. Fundamental based investors need to watch out for value traps! 

Often these negative news are laughable and are just an excuses to sell stocks. The problem is “value investors” have a tendency to confusing these negative news as the actual cause of stock drop, find reasons to say it doesn’t matter (which it doesn’t), then proceed to buy. But often the negative news is just a catalyst for a hibernating bear case to resurface, and NOT the real reason, so these investors are led completely off track.

The Facebook/Cambridge Analytica scandal is one such example. Bulls are attacking a non-existent bear case that people will somehow stop using Facebook. This line of reasoning dominates the market today, and will likely prop up the stock for now. But I do not think that is the reason for the price drop! There are other legit, but dormant bear cases that runs much deeper.  But I will save this for another time..

Some notes on a few stocks I’ve been involved in recently:

Digital Realty Trust (DLR) – bought at around $105/share
Very simple thesis: 1) Positive industry trends, 2) valuation/dividend coverage/leverage all check out fine, and 3) timely technicals

• Positive industry trends. 1) Datacenters have plenty of runway from ever-expanding cloud adoption. 2) The race toward over-the-top video means companies will want their content physically closer to consumer locations, so the Google, Facebook, and Netflix of the world will need the physical proximities that DLR can offer. 3) Edge computing is the future (required by internet of things, autonomous cars, 5G…etc), and again that raises the value of physical proximity to end consumers

• Numbers check out fine. About $18.5bn of market cap versus ~$1bn of cash flows from operations. That CFFO is growing rapidly. $1bn of CFFO easily covers ~700mm of dividends. Debt to EBITDA is high at > 6x , but comparing debt to value of investment properties shows a manageable ~50% loan to value ratio.

• Timely technicals. 10 year U.S. treasury yields got close to 3% resistance and backing off. DLR stock seem to bounce of $100 support level.

Arena Pharmaceuticals (ARNA) – bought around $41.5/share
I bought ARNA right after their March 21st secondary offering. ARNA has market cap of about $2bn, but it has two phase 3 assets that are potentially best in class, targeting multi-billion dollar markets. You really don’t need to be that smart to see this is good risk and reward here.

The 2 drugs are different mechanism targeting different indications. So success probabilities are uncorrelated. Etrasimod is an S1P modulator that is targeting ulcerative colitis, a $4-5bn market. Ralinepag is an IP receptor agonist going for the pulmonary arterial hypertension (PAH) indication, a $10bn market.

Phase 3 trials have about 50% chance of success, so that means only about 25% chance of failing both. I think the distribution of outcomes are follows:

• 50% chance they have 1 indication with peak sales 1-1.5bn; at 3x revenue, this thing is worth 3-4.5bn enterprise value; stock doubles 

• 25% chance that both Etrasimod & Ralinepag scores, then peak sales would be like $2-3bn. Again at 3x EV/revenue this thing is worth 6-9bn. Stock triples or more.

• 25% chance that both fail, so stock trades down to cash value, or down something like 80-90%.. But realistically they still have other early stage pipelines and that’s worth something. 

There’s actually more upside than this. First, another asset, APD-371, has its phase 2 readout coming in 2Q18. If successful we have not two, but THREE phase 3 assets. 

Second, Etrasimod’s real value is that it’s an S1P modulator, which has applications beyond ulcerative colitis. Novartis’ Gilenya and Celgene’s ozanimod are both S1Ps and can be used to treat rheumatoid arthritis, a much bigger market. 

The risk is time to market (will take at least a couple years) and cash burn before we get there. But ARNA just raised a ton of cash so we're somewhat de-risked.


A big loss in Melinta (MLNT), and lessons

I’m taking a big loss here. It was never a big position, but a 55% loss on a ~2% position still hurts.

I bought at about $17.2/share. Originally, my thesis was that at that price, MLNT has ~$500-550mm market cap. They have 4 FDA approved drugs hitting the market, and I estimated $250mm peak sales. So this is a great bargain given established pharma regularly trade at 4-5x EV/revenue.

So what went wrong? 1) too slow to adjust fundamental outlook, 2) failure to see that future fund raises are a form of leverage.

First, I was too slow to adjust my fundamental outlook. I watched happily as MLNT stock collapsed. Small cap biotech/pharma is notoriously volatile, so there’s nothing out of the ordinary. The company has a big pile of cash, so net debt is low, and I’m not worried. “This is the sort of stuff I can double down on!” Management talked about $1bn peak sales, so my $250mm peak sales is quite conservative. I cheered as the stock went down so I can buy more at lower prices.

It’s pure negligence. If I never believed management’s $1bn peak sales numbers, why should I think my much lower estimate is conservative? That's straight up anchoring bias. Only after the latest earning call did I revise my estimates. I dug deeper into competitor revenues and downgraded my estimate of peak sales to $200mm, then $150mm. Now I’m not even so sure about that.

Second, I failed to realize that future fund raises are a form of leverage. I falsely thought of MLNT as a low net debt kind of company, the sort I can double down on. The truth is that MLNT is burning cash, meaning at some point they will need to raise equity or debt. This means MLNT is way more leveraged, and stock valuation is way more sensitive to peak sales than I realized. I underestimated modeling error, and that led to overconfidence and negligence.