In this writing, I dive into Lyft’s numbers again and value the firm as it battles the pandemic. Valuation is built on the expectation that after mobility levels recover, economies of scale will start taking effect in the mid-long term. Using rider based economics in my DCF valuation, I value existing active riders at ~$15 B (estimated rider lifetime value $1,062 x 14.4 M riders by the end of 2020), new and returning riders at ~$39 B, corporate drag at ~$40 B and I get an equity value of ~$15 B ($15 + $39 – $40 + Cash – Debt) making the firm fairly valued. Key to sustain the valuation levels is rider growth with Lyft’s push into expanding the use cases it serves and delivering on swiftly reducing non-direct operating expenses as it scales.
Recent Actions
Until last year, Lyft had maintained a focused narrative. Uber, on the other hand has been aggressively expanding globally into food delivery and logistics besides ride-hailing. Lyft seems to be broadening the narrative now albeit with restrain. While this section of the my blog focuses on the recent actions by Lyft and how they are differentiated from Uber, this article provides an in-depth analysis on how Lyft and Uber differ in their overall strategies (although, the article was written before the two companies hit the bourses, it is still relevant in understanding the difference in company culture, origin and business models).
In Oct’ 20, Lyft announced a partnership with GrubHub that allows Lyft’s loyalty-program members free food delivery from GrubHub restaurants. An excerpt from the 2020 Q3 earnings call, “what’s happening to restaurants in the time like this, when they sell food on a platform, like Uber Eats, they get charged 20% to 30%, they lose 20% to 30% of their revenue to that platform. And so what we’re hearing from these restaurants and retailers, especially during the pandemic is they want to partner not someone that’s going to be, you know, taking 20% to 30% but they want to just have the delivery capabilities, which obviously the 1 million plus drivers we have on the platform, we can provide. So not interested in a consumer platform, interested in kind of more of a B2B organization level approach, which we think is differentiated, and where we can say hey, we’re not going to step between you and your customer unlike other platforms.” Food delivery is a crowded business. Although, it’s hard to say who will be the winners, but I believe the scales will tip in favor of firms who have created an “ecosystem” with a halo effect for other services since restaurants will likely partner more with delivery platforms which have an army of loyal users. Good examples of such “ecosystems” are UberPass and Amazon Prime loyalty programs.
They are also making inroads in health care. In Oct’ 20, they announced integration with Epic, a leading electronic health record system that is used by a majority of the country’s top-ranked hospitals. They believe that “Through this integration, health system staff will be able to book Lyft rides for patients directly through their health records, helping to ensure that transportation is never a barrier to good healthcare. We view this as a significant opportunity, because of the many health systems in the US that use Epic, nearly 70% have not yet worked with Lyft for their non-emergency medical transportation programs. In aggregate, the non-emergency medical transportation market represents a multibillion-dollar opportunity“.
In their pursuit to expand the use cases they serve, Lyft has very recently started focusing on partnering with retailers by providing them with a white label system for delivery. Quoting Co-founder John Zimmer, this is “differentiated in that we are focused on being a partner with the retailer as opposed to taxing them 10%-13% on the package and giving them the muscle they need to compete in the digital economy”. It remains to be seen how this plan unfolds. We will model this use case explicitly as and when this happens and more clarity on specifics emerge.
I am not too overly optimistic about Lyft’s push into the autonomous vehicle (AV) segment. Uber has done way with their AV program. As per research by a leading consulting firm, AVs will drive out 45% of cost which are right now directly linked to drivers. I’d argue that although driver costs will certainly come down, but this will come with the need to maintain a fleet of such vehicles either by owning or leasing them, which in turn will need more reinvestments back into the firm. Infact, the foray into car rentals (SIXT) implies maintaining a vehicle fleet either way. This is a capital intensive business with low economies of scale. Our valuation of Lyft is built on the expectation that economies of scale will take effect in mid-long term.
Lyft is doing a lot of things to gradually inch towards its vision to build a transportation network that can handle every single one of their customers’ transportation needs. While I may be skeptical about a few of the things they are are pursuing, the overall story seems to hold promise.
In the next section, I dive into the numbers and value the company as it battles the pandemic.
Valuation
As in my earlier blog post, the valuation framework that I have used is the one you would use to value companies driving the shared economy. The framework has been pioneered by Valuation Guru Prof. Aswath Damodaran. In his paper, Prof. Damodaran has explained how to incorporate user economics in a DCF Valuation. The fundamental equation to value such companies that Prof. Damodaran gives is simple and intuitive:
Value of a user based company = Value of existing users + Value added by new users – Value eroded by corporate expenses
Please bear in mind that every valuation needs to have a fact based story, which is what we think of the company’s growth potential, margins, capital efficiency and user risk (i.e. would riders stick or ditch) supported by facts. In my valuation, I have taken some judgement calls at places where there is no historical data for extrapolation or historical data does not guide future performance (very true for young, growth and money losing companies).
1. Value of Existing Riders (or customer lifetime value)
I estimate the value per rider using the below simple equations:
Contribution from each Rider = Revenue per Rider - Cost of Serving (Cost of Revenue + Ops & Support) each Rider
Value per Rider = After tax Contribution from each Rider - Re-investments
Now, I will list the facts and frame my story for the parameters in these two equations.
a. Revenue per Rider
Fact – # of Riders and Revenue per Rider: Lyft had 22.9M active riders by the end of 2019. This represents 17% rider share. # of riders declined to 12.5M in 2020-Q3. Management expects revenue / rider to rise by 11-15% in 2020-Q4
2019 Q1 | 2019 Q2 | 2019 Q3 | 2019 Q4 | 2019 | 2020 Q1 | 2020 Q2 | 2020 Q3 | |
# of Rider (M) | 20.5 | 21.8 | 22.3 | 22.9 | 22.9 | 21.2 | 8.7 | 12.5 |
Revenue / Rider | $ 37.9 | $ 39.8 | $ 42.8 | $ 44.4 | $ 164.8 | $ 45.1 | $ 39.1 | $ 39.9 |
Story – # of Riders and Revenue per Rider: We extrapolate the same growth rate to # of riders. This translates to overall ~$165 revenue per rider and over 14M riders by end of 2020.
Now, to project out revenue per rider in the future, we need the addressable market size and also need what % of that market will be captured by Lyft.
Fact – Addressable Market Size: As per the following excerpt from Lyft’s Q3-2020 earnings call, Lyft considers that their addressable market is $1.2T. (they don’t really provide the year and detail on how they got to this number) “we are going after and have been going after since day one the $1.2T consumer transportation market. And we have been laser focused on that. I think all of the same structural elements that have been shifting this market from one that’s been based on car ownership to transportation as a service are still as true today as they were pre-pandemic, and we think all of these forces will continue to be in play as we see the recovery play out. And just to talk about that for a second. Today’s car ownership ecosystem is extremely fractured. A typical consumer has to interact with 10 different companies just to keep up with the basics of owning a car. And at each step, you’re paying full retail prices. So you have a bad disjointed experience with very high prices. And our vision is to build a transportation network that can handle every single one of our customers’ transportation needs. So imagine taking those 10 different companies down to one. And we can create a completely frictionless customer experience and leverage the scale of our network to deliver incredible value to our customers. So that’s our vision. That’s the Total Addressable Market (TAM). I think it’s unchanged. And I think all of the secular trends at force, moving people from ownership to transportation as a service are still at play. If you look at what happened to DVDs and CDs as the world moves to streaming, when you can deliver something as a service at a lower cost with a better experience, that ends up being a really, really powerful combination.”
Although, Lyft has been expanding use cases by foraying into car rentals, bikes and scooters, I believe that this TAM is rather too high. Morningstar estimates Lyft’s total addressable market, including taxis and ride-sharing as well as bike- and scooter-sharing world-wide, to be around $500B by 2023 and growing at 24% a year from 2018. This translates to over $200B market size in 2019. Other research reports peg the shared mobility market size at $164B and $79B in 2019 with growth rates of 25% and 15% over the next 5-10 years.
Story – Addressable Market Size: With this huge variance in market sizes, I resort to the $120B market size in the year 2019 used by Prof. Damodaran. Prof. goes on to say “That is, of course, well below the size of the transportation market, but the $1.2 trillion that Lyft provides for that market includes what people spend on acquiring cars and does not reflect that they would pay for just transportation services“. Moreover, following Prof.’s league, I make the assumption that the market size will double in 10 years. This translates to a CAGR of 7.2% (remember rule of 72?). I won’t blame you if you accuse me of being conservative since almost all research firms have predicted growth rates in excess of 15% over the next 5-10 years.
Fact – Addressable Market Share: In 2019, Lyft had revenue of $3.6B representing a market share of 15% considering Lyft collects 20% of what it bills to Riders.
($3.6B/20%)/$120B = 15%
Bear in mind that Lyft does not reveal revenue % of gross billing. Our estimate of 20% is based of Uber Mobility, which has this in the 20-25% range
Story – Addressable Market Share: With the expanding use cases following the foray into car rentals, bike and scooters and the fact that ride hailing has revolutionized the way we commute, I believe they will be a able to capture 30% of the US and Canada market by 2025 (from 15% in 2019) and increase it to 40% share by 2030. As more players enter the scene, I believe their share will decline to 35% by 2035. Essentially, I am laying the groundwork for my DCF by assuming 3 phases – In the first phase, mobility levels recover by 2022 and then pick up momentum to double market share from pre-COVID levels. Growth continues in the second phase which is followed by a phase of stability or moderating growth in which competition becomes more pronounced resulting in market share decline to 35% by 2035.
Using the above facts and my stories, I can now project out revenue per rider over the next 15 years from $164 at the end of 2020 to ~2x by 2025 and 3x by 2035.
b. Cost of Servicing (Cost of Revenue + Ops and Support) each rider
Fact: Cost of Revenue is direct operating expense and Ops and Support is non-direct operating expense. Infact, R&D, Sales and Marketing and General Admin expense also make up non-direct operating expenses. Operating profit (or EBIT) is the profit left after removing both direct and non-direct expenses from Revenue. For any young growth and money losing firm, you live with the hope that eventually economies of scale will kick and result in faster declining non-direct operating expenses (in % terms) with scale.
Uber mobility reported an adjusted EBITDA of 24.4% by Q4 of 2019. Uber has long term target of reaching 45% EBITDA margin in its mobility business. Although, Lyft has not given any targets for operating profitability, they have been constantly vocal about become op. profitable on an adjusted basis by Q4 of 2021. Adjusted income does not include stock based compensation and historical insurance payouts (companies are known to stretch “adjusted” to suit their needs).
Story: In my valuation, I do not stretch “adjusted” and restrict it to exclude only stock based compensation (I don’t ignore them in my valuation. Just hang on for a bit). This reflects in my assumption of a positive annual adjusted operating margin beginning 2023 (and not 2022 as guided by Lyft). In my bull, base and bear cases, I hit an adjusted EBIT margin of 25%, 20% and 15% by 2025. In setting both short term (2025) and long term (2030) margin targets, I have taken Uber mobility as my guiding light. I build in my hope that economics of scale kick in resulting in faster declining non-direct than direct operating expenses as the firm scales. The cost structure for the three cases projections have been presented below:
Although, I present the entire cost structure projections in the exhibit, I only consider Cost of Revenue and Ops and Support expense as Cost to Serve. The rest of the cost heads are factored in valuing new riders and estimating corporate drag as we will see in subsequent sections of this post.
c. Re-investments required for each rider
Reinvestments required = Change in Revenue / Sales to Invested Capital, where Sales to Invested Capital is also called Capital Efficiency
Invested Capital = Working Capital + PPE + goodwill + intangibles + Research Asset (obtained by capitalizing R&D)
On capitalizing historical R&D expenses, I computed a Sales to Invested Capital ratio of 2.9 for Lyft based on 2019 numbers using the above two equations
To project out Reinvestments, I need the Sales to Invested Capital ratio (and in turn the Invested Capital) for future years.
I capitalize the projected R&D numbers to estimate the value of the research asset for each year 15 years out. I extrapolate the research asset growth rate to estimate the total Invested Capital for each year 15 years out.
I then project out Sales to Invested Capital ratios and use them to estimate Reinvestments.
d. Annual Renewal Probability
Story: There is no data on annual retention. As per popular belief by 2023, COVID should be completely behind us. 2021 and 2022 should be a phase when mobility levels recover – active 2019 riders who stayed at home will return by 2021 and 2022. In 2021, I believe that almost all riders that used Lyft amidst the pandemic (in 2020) will stay active on Lyft. In 2022, as recovery picks up momentum, I assume that 95% of the riders from prior year will stay active on Lyft.
Now, assuming an annual renewal probability of 95% (each year from 2023 to 2035), 54% (=95%^12) of active riders from 2023 will continue using Lyft in 2035. Assuming an annual renewal probability of 90%, only 28% (=90%^12) of active riders from 2023 will continue using Lyft in 2035. I take a judgment call. I believe that the value should be in the middle i.e. 40% of active riders from 2023 will be actively using Lyft in 2035. The 40% translates to an annual retention probability of ~92.7% (=40%^1/12) from 2023 to 2035.
I use the annual renewal probability to discount value of existing riders. This bakes in the wisdom that we don’t assume that all active riders today will be still riding on Lyft 10-15 years out.
We now have all the inputs we need to forecast the life time value of a rider. I use a terminal growth rate of 2% (with the expectation that US T bond rates will increase from 0.9% now) to estimate terminal value and discount the future cash flows to the present using a cost of capital of ~7.5%
2. Value of New Riders
Value added by a New Rider = Value of a Rider - Cost of acquiring a Rider
Value of a Rider in the base year is the life time rider value we estimated in section 1 above. Every year the rider value is modeled to increase with the inflation rate.
For estimating cost of acquiring a new rider, we make a simplistic assumption that all the sales and marketing expenses is done on acquiring more riders. We arrive at the per rider acquisition cost by dividing the projected sales and marketing expense each year by the new riders added each year (sales and marketing expense projection present in the cost structure exhibit in section 1. Remember this is a non-direct op expense and their decline is key to achieving economies of scale).
New riders added in year n = Total riders in year n - (annual retention probability * total riders in year n-1)
We already have the annual retention probability. To estimate the total riders in each year, we need to define the addressable market for the number of riders as well. We assume that 37% of the US and Canada population is addressable (rationale behind this assumption presented in appendix). Although, the derivation is for US, but we extrapolate the addressable % to the 37M population of Canada as well). Lyft had 22.9M active riders by the end of 2019. This represents 17% rider share. We further assume rider share increases from 17% (in 2019) to 24% in 2025 and 30% in 2030. With this assumption, we arrive at 34M riders in 2025 and 44M in 2030 from 22.9M in 2019. We have baked in the COVID impact, by assuming that riders return to the platform gradually from 14.4M (63% of the riders in 2019) in 2020 to 20M in 2021 (which is below the 2019 ridership)
We now have the inputs for the variables we need to estimate the value added by new riders each year. We discount the future values to the present using a higher cost of capital 8.5% reflecting higher risk owing to cash flows from riders not on the platform yet.
3. Value of Corporate Drag
Expenses which do not directly relate to the existing and new users to extent possible fall under corporate drag. We have classified general and admin and stock compensation expenses as corporate drag. These are non direct operating expense and hence as we said earlier we expect them to decline in % terms over time as the firm scales.
We project out these expenses into the future and discount them to the present to estimate the value of corporate drag. (projections present in cost structure exhibit in section 1)
Putting it all together
In the bull, base and bare cases, I have used diff margin targets, but I have stuck to the same story presented in this writing above.
My sense of the value in the firm is centered on the base case scenario, which makes Lyft fairly valued as of this writing (12/13/2020). Lyft traded at $46.87 at close on 12/11/2020.
In the three cases, there is little variance in user lifetime value. Major difference in equity value is due to the value by new riders and the corporate drag, both of which are primarily dependant upon non-direct operating expenses. For any young growth and money losing firm, it’s hard to rely on limited history and make projections. The business model thrives on the hope that as the firm grows it will realize economies of scale resulting in declining non-direct operating expenses (as a % of revenue; in absolute terms it will rise). This valuation is built around that hope. I have presented facts supporting my story to the extent possible. I have taken quite a few judgement calls based on my story for Lyft. Your assessment can be very well different from mine. The value that Lyft offers depends upon your story.
Thank you for reading.
Appendix
Rider addressable market size
In M | Population | Addressable riders | % of Pop which is addressable | Remarks |
# of People <18 of Age | 74 | 19 | 25% | Assumption |
# of People >65 of Age | 56 | 3 | 5% | Assumption |
# of People between 18 and 65 | 203 | 101 | 50% | Assumption |
Total | 333 | 123 | 37% | Derived |
Rider addressable market share
Year -> | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | 2026 | 2027 | 2028 | 2029 | 2030 | Growth Rate |
Lyft riders (in M) | 18.6 | 22.9 | 14.4 | 20.0 | 25.0 | 28.0 | 31.0 | 34.0 | 36.0 | 37.9 | 39.9 | 42.0 | 44.0 | |
US Pop (in M) | 328 | 330 | 332.6 | 334.8 | 337.0 | 339.2 | 341.4 | 343.7 | 345.9 | 348.2 | 350.5 | 352.8 | 355.1 | 0.66% |
Can Pop (in M) | 37 | 37.0 | 37.5 | 38.0 | 38.6 | 39.1 | 39.7 | 40.2 | 40.8 | 41.4 | 41.9 | 42.5 | 43.1 | 1.40% |
Addessable (in M) | 134 | 135 | 136 | 137 | 138 | 139 | 140 | 141 | 142 | 144 | 145 | 146 | 147 | |
% Lyft share | 17% | 24% | 30% |
Common assumptions