One quick look at the Facebook's (FB) S1 filing is enough to dazzle a naive investor. 845M monthly active users, 2.7B daily likes and 100B friendships! Facebook is what many of us use day-in and day-out. Understandably, there was going to be a strong bias in its favor. At the time of IPO, noted academics like Professor Aswath Damodaran came out with their own valuations. A common critique of DCF based valuation is that it is subject to various assumptions. However, it is also true that it serves as a good benchmark to compare against multiples based techniques.
The intense debate that is happening on the issue of fair valuation for Facebook led me to create a DCF based valuation model of my own. My objective is to find the upper bound for Facebook's valuation. Therefore, this model makes assumptions that would result in a higher valuation. This approach applies to both revenue build-up as well as costs. My analysis leads to the conclusion that Facebook's upper bound for valuation is $60B. I whole-heartedly welcome any suggestions that will help us increase this figure. I welcome the readers to read through the detailed analysis below; however, here's a quick summary of my projections and assumptions.
- Aggressive user growth - Facebook will have 4.5 B MAU's worldwide by 2020. This will be 56% of projected world population of 8B.
- Aggressive revenue growth - Even though ARPU may actually decline as number of users grows, I have assumed continued 8% growth for next five years that is stabilized to 3% by year 10.
- Favorable COGS - While Google (GOOG) has COGS of 35%, I have assumed lower COGS of 23% throughout.
- Favorable costs - Assumed percentages that are either lower or same as Google, for sales and marketing, R&D, G&A, D&A and Capex.
- Favorable discount rate - Cost of capital is assumed to be 9% throughout.
- Favorable tax rate - Assumed a lower effective tax rate of 19% same as Google.
- Favorable terminal growth rate - 5% long term growth rate assumed.
Thus, it is evident that I have gone out of my way to provide the benefit of the doubt to Facebook. Yet, my analysis leads me to a figure of $60B. This is 42% lower than the IPO valuation of $104B. There are a few possible ways to explain this discrepancy.
- My assumptions are flawed or the model suffers from other mistakes.
- I have over-looked a major revenue stream that should have been considered.
- IPO valuation was not driven by fundamentals but inflated expectations or greed.
I am open to criticism on my approach and to suggestions for improvement. If my analysis makes sense, I don't know what drove Facebook team to go forward with a valuation that now looks outrageous.
Here's a statement from Facebook's S1. "Simply put: we don't build services to make money; we make money to build better services." The question is, then, if making money was not Facebook's goal, could it have settled for a more conservative valuation?
Detailed Analysis and Key Assumptions
Here is a link to the Google spreadsheet for your reference.
First, let's look at the assumptions for revenue build up.
- Based on the Monthly active users statistics from the S1, I have projected the MAU for the next ten years. In another ten years, based on these projections, FB would capture 80% of US and Canada population, and has a total of 4.5B MAU's worldwide, which is 56% of projected 8B of population by 2020.
- I increase the ARPU by 8% every year for every region for the next five years. Subsequently, I stabilize the ARPU growth to 3% for the next five years. In reality, as FB admits, the higher growth is going to come from regions like Asia and Africa that have much lower ARPU, so worldwide ARPU could actually decline as this happens.
During the second quarter of 2012, worldwide ARPU was $1.28, an increase of 2% from the second quarter of 2011. Over this period, ARPU increased by over 10% in the United States and Canada, Asia, and Rest of World, and by 8% in Europe. User growth was more rapid in geographies with relatively lower ARPU, such as Asia and Rest of World. These user growth dynamics resulted in worldwide ARPU increasing at a slower rate than the rate experienced in any geographic region. We expect that user growth in the future will continue to be higher in those regions where ARPU is relatively lower, such as Asia and Rest of World, such that worldwide ARPU may continue to increase at a slower rate relative to ARPU in any geographic region, or potentially decrease even if ARPU increases in each geographic region.[Source: Facebook's 10Q for period ended June 2012]
- I calculate revenue for each region (US & Canada, Europe, Asia, Rest of World) using the MAU and ARPU projections. I am assuming that this ARPU includes both advertising and other revenue sources such as sale of virtual goods etc. These assumptions are based on following statement.
We calculate our revenue by user geography based on our estimate of the geography in which ad impressions are delivered or virtual goods are purchased. We define average revenue per user (ARPU) as our total revenue in a given geography during a given period, divided by the average of the number of MAUs in the geography at the beginning and end of the period. Our revenue and ARPU in markets such as the United States, Canada, and Europe are relatively higher due to the size and maturity of those advertising markets as well as our greater sales presence and the number of payment methods that we make available to advertisers and users.
It is also interesting to note below statement in Facebook's S1. This creates a doubt as to whether payments based revenue accounted for over-the-top valuation, and perhaps this is a link that led to Facebook's fall with Zynga (ZNGA). However, this is just a speculation on my part.
Our Market Opportunity for Payments: When users purchase virtual and digital goods from our Platform developers using our Payments infrastructure, we receive fees that represent a portion of the transaction value. Currently, substantially all of the Payments transactions between our users and Platform developers are for virtual goods used in social games. According to an industry source, the worldwide revenue generated from the sale of virtual goods increased from $2 billion in 2007 to $7 billion in 2010, and is forecasted to increase to $15 billion by 2014. We currently require Payments integration in games on Facebook, and we may seek to extend the use of Payments to other types of apps in the future.
Let's now talk about building the cost side. At every step, I have given benefits to FB.
- Cost of revenue - I use 23% for all ten years. This is based on 2011 number. Yes, possibly the cost of revenue could further decline.
We anticipate that cost of revenue will increase in dollar amount for the foreseeable future as we expand our data center capacity to support user growth, increased user engagement, and the delivery of new products. We expect costs will also rise for payment processing as we increase Payments volumes and add new payment methods. The expected increase in cost of revenue may be partially mitigated if we are able to realize improvements in server performance and the efficiency of our technical operations. We expect cost of revenue as a percentage of revenue to decline modestly over time to the extent we are successful in meeting our objective of efficiently increasing revenue.
However, let's not forget that Google which generates almost 96% of its revenue from advertisements, has 35% cost of revenue. (Source: Google's 10Q for period ended June 2012.)
- Sales and marketing - FB has this expense at 12% of revenue in 2011. Even though FB has significantly increased this expense off-late, I still keep it at 12% for all ten years. Google too has this at 12%.
- R&D - 10% of revenue based on 2011. Google too has it at 10%.
- G&A - 8% which is same as Google. Never mind, that FB has had higher expense.
- D&A - 6% of revenue based on median long-term trends for Google.
- Cash cycle - Account receivable and payables are likely to follow the same trends as that of last few years. AR is 15% of sales, while AP is 27% of COGS in 2011. Google's AR is 14% of revenue and AP is 3% of COGS. I have again given benefit to FB by assuming a higher AP at 27% of COGS.
- Capex - While FB has mentioned a capex of $1600 M in 2012, over the long-term, it is expected to come closer to 10%, same as Google.
I have used Google as a close comparable business. However, whenever in doubt, I have used numbers such that Facebook will always get a benefit that should result in a higher valuation.
We need few more assumptions to arrive at the final valuation.
- Effective tax rate - FB has mentioned that effective tax rate for the last two years has been about 40%, which is higher than statuary tax rate of 35%. Instead of using 40% as tax rate, I want to use a much lower number of 19%, which is based on Google's effective tax rate. Underlying assumption is that with time, FB will also learn intelligent tax-savvy techniques from Google. (Such as setting up offices in Ireland, Bahamas etc.)
- Cost of capital - I use Google's beta to find unlevered beta for FB. Online advertising is the core business for both, and subject to similar market risk. Assuming 30 year risk-free of 2.65% and market risk premium of 6%, I get a discount rate of 9%. This is again lower than what some experts have calculated for FB.
- Finally, a perpetual growth rate of 5% is assumed after 10 years. Typically, this should be around 3%, same as world GDP growth rate but what the heck, we'll give it to you FB!
Thus, I arrive at the $60 B figure, which in my view defines the upper bound for Facebook's valuation based on current business model. Will be glad to review and change any of the assumptions made above, especially anything that will help us get closer to $100+ B!