Sunday, 18 June 2017

Valuing Disruption, Tesla, and other stuff I will never understand

On the subject of never understanding stuff...

I watched game 6 of the Nashville/Pittsburgh series last Sunday night, and I will "NEVER" understand how the play was blown dead in the first few minutes of the second period on a goal that should have counted, by a referee who's view of the puck was obstructed (at least temporarily).

If you haven't seen this play, you can watch it below and judge for yourself:





Every single sport out there allows reviews of mission critical plays, and to say this play was not mission critical is the understatement of the century. This was the Stanley Cup Final! As far as ridiculous calls go, this has to rank up there in the top 10 of all time. If the goal had counted, the tone of game 6 would have been completely different. I'm not saying that Pittsburgh wouldn't have eventually won, but come on guys, call a fair game.

On the subject of disruption...

I believe that there is a noticeable dichotomy between participants strictly governed by a grounding in conventional (i.e., Graham-esque) valuation theory and participants latching onto the theme of disruption. This dichotomy has probably always existed, I've just never really thought about it until recently.

I also believe that there is a theoretical intersection between the two approaches somewhere, and I guess that as long as participants at both ends of the spectrum generate alpha for themselves and their clients, who really cares, right?

The main difference, in my mind anyway, is that strict adherents of conventional Graham-esque valuation principles probably scoff at the at the idea of paying lip service to the notion of valuing disruption, and yet here we are, collectively paying seemingly outrageous multiples for ownership stakes in disruptive businesses, while casting aside businesses which appear to be victims of disruption. Need proof? Look no further than juxtaposing the performance of anything supplanted by Amazon as the road to Amazon $1,000 has been littered with the corpses of all things bricks and mortar (at least up until recently anyway).

By my calculations, approximately 10% of the S&P500's current capitalization is attributable to Google/Alphabet, Apple, Amazon, Facebook, Tesla, and Netflix, the collective epitome of disruption, so indeed, someone out there loves paying for disruption and the broad indices reflects this fact almost daily. Passive do-it yourself indexers be wary, 10% of your future retirement fortunes are locked into the fortunes of these 6 companies.

And the weighting towards large cap tech doesn't stop at just the S&P500. Even Harvard educated, 30 years in the business, institutional fund managers are heavy on the tech disruption bus. Don't believe me? Here's Fidelity's Will Danoff's top holdings by weight, who, incidentally, was profiled earlier this year by the Globe in the following article...

"Star stock picker Will Danoff now accessible to Canadian investors"...

Fidelity has even run a tear jerking commercial spot tapping into a cascade of emotions surrounding entrusting your family's future with the legend himself, by touting units of the Fidelity Insights Class under Mr. Danoff's careful jurisprudence.























Here's a link to the full holdings, and here's a previous incarnation of said tear jerking commercial:





My question: why would a Harvard educated, 30 year tenured fund manager, introduced to Canada as "legendary" want to simply mirror the QQQ's? I will leave the question open and move on.

On the subject of Tesla...

Not a day goes by that Tesla isn't front page news, so I thought, why not try a simplistic DCF valuation. After all, I'm a simple muppet, so here goes (I'm going to preface my attempt with a deep take this with a grain of salt/caveat).

Before I get to my results, a conclusion up front: I must have approached the exercise incorrectly from the get go, why? Because I attempted to use linear logic to run my DCF.

In my first attempt, I used the historical financial results from 2012 through current in order to model out a number of assumptions, gathering info from various sources on Global Plug-in/EV forecasts out to 2025, which seem to peg 2025 EV units at between 25% and 35% of estimated deliveries globally. This translates into around $750B of total global EV sales (assuming global auto sales of $2.5T in 2025, and taking the mid range forecast of EV share at 30%).

I also had to make assumptions pertaining to Tesla's expected market share in 2025. Currently around 4% of global EV's, I'm going to optimistically assume a 10% market share as the company moves towards mass marketing and maturity in 2025 and beyond. My sources for global EV sales (in $'s) is a combination of reports from both the International Energy Agency, link here, and Statista, link here.

By my estimates, Tesla's market share in terms of dollars worked out to around 4% of total EV sales in 2016 (i.e., $7B in sales vs. $164B in sales globally). Assuming EV sales represent 30% of global auto sales in 2025, this works out to 30% x $2.5T x 10% Tesla market share = around $75B in sales, a staggering tenfold increase in sales vs. 2016.

Taking industry wide ratios for mature auto manufacturers (i.e., GM, Ford, Daimler AG BMW, etc.), I used average EBIT of 7.5% to model expected operating profit and an average reinvestment ratio of -1.14 in order to estimate reinvestment (net capx + working capital), courtesy of Professor Aswath Damodaran at NYU. I also estimated a future tax rate of around 25% (assuming the company uses up it's $4B and counting of loss carryforwards before 2025).

On this basis, I came up with expected sales of $75B in 2025, EBIT of $5.6B, and total units of 1.2M at an average sales price of $60K per unit (assuming average sales price per unit decrease incrementally each year commencing in 2018 post Model 3 mass production rollout).

Logical and linear, right?

Here's the problem. In my opinion, linear logic gets tossed out the window when considering the ongoing net reinvestment required to produce $75B in sales in 2025 (at least from a logical valuation perspective). At $75B in sales (up from my own estimates of $63B in sales in 2024), and assuming industry average net reinvestment of -1.14 x delta sales, this level of sales would seem to suggest net reinvestment of around $10B, see below:










Here's the historical net reinvestment between 2012 & 2016, using published financial statements (2017 is ttm):











So certainly not outside the realm of possible to accept that sales growth should be accompanied by ongoing material net reinvestment. The issue thus far is that the company has been spending more on capital in advance of the Model 3 rollout than would normally be required by a mature industry participant already mass producing. Therefore, in modelling future production, I have assumed that the sales/capital ratio will increase from .24 to .74 commencing in 2018, and as the company reaches economies of scale post Model 3 rollout, I have added 10 bps per year commencing in 2019 in order to reduce estimated net reinvestment, until I get to the industry average of 1.14.

This matters (from a DCF modelling perspective) because the growth in sales has to be supported by net reinvestment in capital. And as net reinvesment grows, free cash falls until net reinvestment levels off. And I'd hazard a guess that net reinvesment levels off or even declines in the case of non-capital intensive businesses (due to to scale-ability), which doesn't seem to be the case with Tesla.

Assuming EBIT is $5.6B in 2025 and tax is 25%, I get NOPAT of $4.2B, less net reinvestment of -$10B, which results in negative free cash flow of -$5.B. And this was my issue all the way through my first attempt. I never came up with positive free cash flow!

The argument therefore, is that I can't really use a DCF model here unless I make some very aggressive generalizations as to what EBIT and net reinvestment should be once the company becomes a mass producer operating efficiently near capacity. At some point net reinvestment should be small relative to what the company is spending now to build out, but I have no idea when this will be. My best guess is that when the company is mature, it will spend around what other manufacturers spend, or around sales / -1.14.

Maybe, the best approach here is to try to derive what cash flows the market is estimating in terms of current valuation, and work backwards from there in order to determine how reasonable these assumptions are. So here goes (and forgive me if this looks like an extreme case of curve fitting, but that's exactly what it is):

I started my second attempt with the following question:

What combination of sales, EBIT margin, reinvestment, and two-stage discount rates is the market pricing in today in order to arrive at a valuation of $65B, assuming 2025 as the terminal cash flow year?

Before I get to a "possible" answer, I want to make mention of Marketwatch running the following piece on Baron Capital regarding Tesla last week. Apparently, Mr. Baron believes that Tesla will get to $70B in sales and operating profits of $10B in 2020 (i.e., an EBIT margin of 14%, or double the industry average), a full five years before my estimated 2025 $75B in sales. This implies 1,000,000 in units sold in 2020 (vs. my originally estimated 497K units). And the source for this 1,000,000 unit target is the company itself.









So, rejigging my DCF model to incorporate Mr. Baron's estimates, in order to justify a valuation of close to $65B, the market must be expecting a terminal reinvestment rate of close to sales/-2x (vs. an industry average reinvestment rate of sales/-1.14x), a terminal EBIT margin of 20% (vs. an average industry EBIT margin of 7.5%), and imputed two stage discount rates of 4.9% in the near term and 6.9% terminally on 1M units in 2020, and $70B in sales, as follows:


























The growth curve isn't linear at all: it's exponential in terms of the jump from current production to mass marketing the Model 3. For this model, I've assumed sales triple between 2018 & 2020, and then decrease incrementally at the rate of 90% x the previous year's growth rate between 2021 and 2025. Number of units sold is a function of the incremental decrease in avg unit price between 2020 & 2025. EBIT margins inch up from 7.5% in 2019 to 20% in 2020 (a possible function of economies of scale and lower lithium input costs over time).

Concluding Thoughts

So are these expectations reasonable? I honestly don't know. My gut tells me no, and yet what do I really know? Maybe the company growth curve is still in its infancy, and non Graham-esque participants dismissing the disruptive merits here are missing out on the potential for a 10 bagger. Maybe, just maybe, the cult of Tesla followers have vision, and I don't.

What I do know is this: There is a huge amount of uncertainty surrounding future cash flows. And when there is huge uncertainty combined with an admittedly outrageous starting valuation predicated on future production targets which may or may not transpire, everything, and I mean everything has to go right from now until 2025 in order to justify the current valuation. And this begs the following question: If the market is happy to pay $65B (fair value) for this story at 4.9% near term, and 6.9% terminally, what happens if we double the discount rates (for silly argument's sake):

Here's what happens:










Caveat emptor indeed.