Mar 26 2021 - Tesla: Why ARK's Use Of Wright's Law Is So Wrong

Summary

• ARK Invest has just published an updated valuation model and price target (of ,000 per share) for Tesla.
• A key driver of ARK's valuation is that Wright's Law will drive down Tesla's cost of production and drive up its margins.
• Even if you assume that Wright's Law is correct, the way that ARK uses this law is highly questionable.
• The fact that such a prominent Tesla bull is driven to these extremes of illogic to justify its position, is further confirmation that Tesla is a bubble of epic proportions.
• An analysis of ARK's historical returns, adjusted for their risk, shows that there is no reason to believe that ARK's opinions are market-beating.

ARK Invest (“ARK”), one of the most prominent Tesla (NASDAQ:TSLA) bulls, just published its revised valuation model and price target of ,000 per share for Tesla. The model, its assumptions, its math and structure, and even its data entry are being widely and rightly savaged, both in the pages of Seeking Alpha (here, here, here and here) and on FinTwit. And the stock market is joining the party. Tesla is currently over /share lower than where it was before ARK made its pronouncement and is sitting at only about 20% of ARK’s 2025 price target. If the market believed ARK’s number, that is a compounded annual return of almost 50%.

As tempting as it is to join the dogpile, I won’t. Instead, I will focus on ARK’s use of “Wright’s Law.” Followers of ARK will know that – along with “disruptive innovation,” “winner-takes-all” and “retail investors are the only ones who really get it” – this law figures very prominently in ARK investment writings and CNBC appearances.

This article follows, among other things, a lengthy Twitter exchange that the author had with Brett Winton, ARK's Director of Research. It also uses some research (the “Research Paper”) prepared by ARK in September 2019 entitled “Wright’s Law Predicted 109 Years of Auto Production Costs, and Now Tesla’s.”

The Research Paper presented a large amount of evidence in support of Wright’s Law and its applicability to the automotive industry and Tesla. I won’t question that. Instead, I hope to show that even if Wright’s Law is correct, ARK’s use of it, and the conclusions ARK draws from it, are highly questionable.

Why Does It Matter?

I believe that discounted cash flow (DCF), similar to the analysis incorporated in this SA article, is the best way to value a company, particularly one that is growing rapidly. However, regardless of your approach to valuation, there is one thing that is virtually universal: the importance of profit margins. Although today’s market is obsessed with growth and revenue, almost any valuation approach will tell you that growth without profitability does not enhance, and may in fact destroy, shareholder value.

A major driver of ARK’s valuation is the belief that Wright’s Law will drive down Tesla’s cost of production and therefore greatly increase Tesla’s margins. As the writer of the Research Paper puts it, “We believe that key to Tesla’s viability and value is the gross margin of the Model 3.”

This justifies a deep dive into ARK’s use of Wright's Law.

Fallacy Number One: A “Unit” is Not Necessarily a “Unit”

Here is the description of Wright’s Law provided in the Research Paper:

Pioneered by Theodore Wright in 1936, Wright’s Law provides a reliable framework for forecasting cost declines as a function of cumulative production. Specifically, it states that for every cumulative doubling of units produced, costs will fall by a constant percentage. Since it began taking shape in the early 1900’s, the auto industry has enjoyed an 85% learning curve – which has translated into a 15% cost decline with every cumulative doubling of units produced.

The question is: what is a “unit”? In other words, what qualifies as a fundamentally novel product which restarts the clock on cumulative production and calculations using Wright’s Law? And what merely constitutes a variant on an existing theme for which the clock does not restart?

ARK notes that there have been about 2.5 billion internal combustion engine (ICE) vehicles made over time. ARK states that Wright’s Law continues to apply to this production, but doubling 2.5 billion would take a long time and so cost reductions at this point are minimal. Not so for an electric vehicle (EV), of which only about 3.4 million (at the time of the Research Paper) had been built. So, concludes ARK, “the difference in time to achieve a cumulative doubling explains the difference in annual cost declines: the cost of the average ICE vehicle will edge down by -0.5% in the coming year while that of its EV counterpart costs (sic) will drop by -12%”

But, for the most part, an EV is merely a car with a different drivetrain. I can’t believe that ARK would claim, for example, that the inclusion of a diesel engine would have restarted the clock on petrol engine cars. Or, for a more recent example, that Toyota’s pioneering of hybrid drivetrains would restart the clock on all automotive manufacturing.

An argument can be made that Tesla’s EV drivetrain is sufficiently novel – at least if we are willing to ignore the roughly 20 million hybrids, each of which has an EV drivetrain, that have been produced since 1997 – to warrant starting at ground zero for Wright’s Law calculations. The problem is that ARK thinks that the drivetrain is only about 20% of the total value of a car.

But if the only part of an EV that is subject to the steep learning curve of Wright's Law is the drivetrain, and if the drivetrain only comprises 20% of the total cost, then why isn’t the total savings 12% of 20%, or 2.4% in total?

In other words, why should increasing the production of Tesla Model 3s reduce the cost of their metal stampings? Or their tires? Or their seats? Or their stereos? Or of anything contained in the 80% of costs that do not correspond to the EV drivetrain?

If you believe in Wright’s Law, then the logical way to apply it would be to break down a car into its components and then apply the law separately to each of them. In fact, one Twitter commentator, with apparent expertise in the matter, said that this is exactly the way it should be done.

ARK did not do this. Instead, ARK just assumed that EV manufacturers would apply the same “cost model” as ICE manufacturers, so that the drivetrain would continue to equal 20% of the total cost of the vehicle. In other words, when Wright’s Law drove down the cost of Tesla’s EV powertrain by 25%, then magically Tesla would be able to buy tires, seats, steering wheels, steel, aluminum, suspension parts, stereos, etc., for 25% less than before. Or maybe Tesla would start supplying the car with only 3 tires, as the following tweets from Winton suggest?

In other words, ARK assumes that total cost of EV production falls at the same rate as Wright’s Law dictates for the cost of the EV powertrain because Tesla will follow the same “cost-model” by building the rest of the car to a Ford Fiesta standard instead of a Toyota Camry one. Apparently customers won’t notice.

Here is another way to look at this: Volkswagen is just starting to produce EVs and therefore has very little cumulative production of these vehicles. These EVs share many parts and processes in common with the ICE vehicles VW has been producing since the 1930s. Yet, if we apply ARK’s logic, then the cost to VW of producing the entirety of these EVs should be on a steeply downward slope, whereas their poor ICE cousins trundle along with tiny efficiency gains, held back by their huge cumulative production.

I believe that there is no logic to ARK's use of Wright's Law, but it is still nothing compared to….

Fallacy Number Two: “Cost” Is Not the Same Thing as “Margin”

If you read the Research Paper, you will quickly find that ARK uses the words “production costs” and “margin” almost interchangeably. The problem is that they aren’t.

ARK’s poster child for the workings of Wright’s Law in the automotive industry is Ford Motor. ARK spends a lot of time showing how Ford’s production costs have adhered to the law. But this leaves the obvious question of why Ford today is not an enormously profitable company with huge gross margins. The answer, according to the Research Paper, is this:

Since selling just under 2,000 Model As for an inflation-adjusted ≈,000 in 1903, Ford…has been on a constant cost decline….Wright’s Law suggests that it would be able to manufacture the Model A today for ≈,500 today. The original Model A had 8 horsepower (HP) and a top speed of 28mph….If auto buyers were willing to pay ,000 for an 8 hp car in the US today, then Ford’s gross margins would approach 90%+.

I guess that we will never know the answer to the question of whether US auto buyers would be willing to pay ≈,000 today for an 8 hp car if they had no alternative. But the whole point is that they do have alternatives, which is precisely why Ford does not have 90%+ gross margins today and why Tesla will likely never have the gross margins that ARK projects.

The thing about a law is that, by definition, it applies to everyone. And it’s a basic rule in economics that a cost saving which applies to every producer, in any kind of reasonably competitive market (and the automobile industry is nothing if not competitive), will eventually be passed to the consumers in the form of lower prices (or, which is the equivalent, a much better product). This is precisely what has happened with Ford and the automobile industry. And this is precisely why the Research Report has to qualify its predictions with words like this:

Now using Wright’s Law to forecast, and assuming the Model 3’s average selling price (ASP) remains at ,000 (emphasis added), we forecast its gross margins at more than 30% by the end of 2020.

Here is a chart from Wolfgang Pipperger (@WPipperger on Twitter) showing how well Tesla is adhering to this assumption of an unchanging ASP.

How Is Wright’s Law Doing So Far?

Here is a graph of Tesla’s reported automotive gross margin (excluding regulatory credit sales so that we can focus on the factors which Wright's Law is supposedly driving) since the Research Report was published in 4Q 2019. (Unfortunately, Tesla does not provide the information necessary to show a chart only for the Model 3, but for much of this period the Model 3 has been the dominant vehicle for Tesla.)

Needless to say, this does not show the smoothly increasing margin that would be predicted if Wright’s Law were the only factor here. Similarly, the prediction contained in the Research Report that Tesla’s gross margin would reach 30% by the end of 2020 was not exactly prescient.

(The above chart takes Tesla at its word about its gross margin, notwithstanding that there are ample reasons – as pointed out here – to doubt this.)

The Tesla bulls, including ARK, might respond that Tesla has chosen to “spend” its constantly growing gross margin in price cuts and an expansion of the market. Leaving aside the question of whether a company that allegedly faces an unquenchable demand for its products should or would be cutting prices, this comment still does not bite in the context of the ARK valuation. The ARK model shows both an enormous increase in sales and a huge increase in margins. There is no room in this valuation for “spending” margin to buy market penetration. You need both.

Stock Price, Bro!

There are doubtlessly many Tesla bulls out there who think that this is just jealous pettifoggery when put up against the extraordinary returns generated by Cathie Wood’s ARK.

Although, judging from the comments to my articles, there are a great many bulls who believe that “stock price, bro!” is a dispositive argument, there are two possible explanations for ARK’s returns:

• One, the 65-year-old Cathie Wood is an investment genius and visionary who lay undiscovered for 30 to 40 years until ARK’s funds went vertical starting in 4Q 2019, or
• Two, ARK is riding a lucky bubble.

The best way to answer this question is, of course, to observe ARK’s performance over a long period of time, through many market cycles. If ARK consistently outperforms, then at some point the answer has to be “one.” Or “two” if the opposite happens.

Unfortunately, running this experiment is time consuming and you may only have results when it is too late to put your money with ARK for an early retirement. But there is a way to get some quicker insight and that is to benchmark ARK’s returns to a reference portfolio with similar characteristics. That is the approach taken by this article. You should read it, but the punchline is contained in this graph showing that a leveraged investment in the NASDAQ 100 (via the Invesco QQQ Trust), with the amount of leverage chosen to replicate the volatility of the ARKK fund, would have done as well as ARK.

(A separate analysis using a “factor model” approach shows the same thing: ARKK is nothing exceptional.)

But even this comparison is too favorable to ARK. Unlike the NASDAQ 100, which is composed of extremely liquid stocks, AARK has heavy concentrations in illiquid stocks. The chart above shows that investors in ARKK are not being paid for taking on this dramatically higher liquidity risk. That’s not smart.

Conclusions

Warren Buffett is fond of saying: “If you have been in a poker game for a while, and you still don’t know who the patsy is, you’re the patsy.”

If being a Tesla bull requires the type of analysis done by ARK, then I am very confident that I know who the patsy is. Of course, we live in a time of utter madness – Gamestock, Hertz issuing stock out of bankruptcy, Bitcoin, record margin debt, Robinhoodies, SPACs, Dogecoin, stimmies, etc. – so the Tesla bulls will probably dismiss this as yet another case of an embittered short crying wolf. But the important thing to remember about the fable is that, in the end, the wolf actually does show up.

Tesla's stock price has been divorced from reality for a very long time. It appears, however, that a reconciliation and remarriage may be forthcoming. Tesla is down about 0/share versus its all-time high. In addition, the "Law of Diminishing Marginal (Stock Pump) Returns" appears to be setting in - witness the "meh" reaction to ARK's latest price target or today's reaction to this customary quarter-end, company-wide exhortation to his troops with material non-public information that Musk knows fully well will be "leaked."

Finally, also witness the bids that the shares of other EV makers - including VW - have caught. Unless this is like Lake Wobegon, where every child is above average, surely they can't all be priced for total market domination?

There could be an interesting short opportunity emerging. Given the irrationality of the stock, however, I am only doing this with a small part of my portfolio and in the form of long-dated puts or outright shorts with reasonably tight "stop" orders.

Trade carefully. The cult may not be dead yet, although it is certainly looking more sickly.