ARKK And Wright’s Law: Buy The Law, Not The Fund (And Buy XBI Instead)

Overview and thesis

Making predictions is not easy, especially about the future. That is one major reason why many investors are drawn to Wright’s law and the ARK Innovation ETF (NYSEARCA:ARKK). The former has been demonstrated on many of the disruptive technologies (in the past though), and the latter uses the former as a pivotal theme in its selection of holdings.

However, the thesis of this article is:

• Even though many real-world problems have been demonstrated to follow the Wright law (ranging from then disruptive technologies such as airplanes and automobiles), application of the law to ongoing problems always entails considerable risks and uncertainties.
• For many of the investments in ARKK, two of the top risks and uncertainties are limited data and a very uncertain timeframe.
• To me, these uncertainties and risks are so large that they make any effort for active selection a futile attempt.
• For investors who want exposure to high-risk high-potential investments in the biotechnology subsector, the SPDR S&P Biotech ETF (NYSEARCA:XBI) provides a sounder alternative. Its passive and equal-weight approach makes better sense to me when it comes to highly uncertain bets.
• Furthermore, other positives for XBI include A) many of XBI holdings are targets for acquisitions, B) XBI is cash and earning positive both at the fund level and holding level for many of its holdings, and C) XBI’s valuation is currently at a secular bottom.

What is Wright’s Law?

I first read about Theodore Paul Wright as an aeronautical engineer and educator (when I was studying engineering in graduate school). Only years later did I learn about his most important contribution, which is not in aeronautics (though it is rooted in aerospace engineering). While studying airplane manufacturing, Wright determined that for every doubling of cumulative airplane production the labor requirement was reduced by 10-15%. Now known as “Wright’s Law”. The law has been so effective in predicting the cost of a wide range of products beyond airplanes, including the automobile production costs in the past 100 years, as shown in the chart below.

Wright determined that for every doubling of airplane cumulative production, the labor requirement was reduced by 10-15% on a per-unit basis. Specifically, it states that for every cumulative doubling of units produced, average unit costs will fall by a constant percentage between 10-15%. Take the lower bound of 10% as an example. If we have so far produced 1000 units of something (say microwave oven) and the average cost per unit is $100. Then Wright predicts that when cumulative production reaches 2000 units, the average unit cost will be only $90. When cumulative production reaches 4,000, then the average unit cost will only be $81, and so on. And if you plot such a relationship on a log-log plot (average unit cost vs cumulative units produced), you would see a straight line – which is precisely what the following chart shows.

And if you measure the slope of this line in the chart, you would see that it is about 15%. The traditional auto industry has enjoyed a 15% cost reduction per cumulative production doubling – near the upper limit of Wright’s law.

Impressive? Indeed. It only gets more impressive when you consider:

• It is hard to talk about Wright’s law without mentioning Moore’s law. However, Moore’s law has a limit – eventually. It applies to miniaturization – and it cannot go on indefinitely as atoms and molecules have finite sizes. But there is no apparent limit to Wright’s law – as it relies on human innovations and our accumulation of experiences and knowledge.
• Traditional industries (such as airplane and auto) have never been scalable, but the new technologies (e.g., those ARKK heavily bets on) are scalable – at least we think so. And such scalability SHOULD make Wright’s even “more correct” and more potent, which leads us to the ARKK fund itself next.

ARKK fund and Wright’s Law

The ARKK fund aims to provide broad exposure to disruptive innovation centered around artificial intelligence, robotics, energy storage, DNA sequencing, blockchain technology, and so on. And a pivotal theme under many of the stocks selected for the fund is Wright’s law. There are good reasons for this approach if you consider:

• Traditional industries (such as airplane and auto) has never been scalable. Meaning they cannot produce an additional unit without additional cost (even that cost diminishes).
• But the new technologies (e.g., those ARKK heavily bets on) are scalable. The software industry is the best example of a scalable business. Selling an additional copy of the software adds almost no additional cost at all. And many of the stocks that ARKK invests in are scalable – at least to a good degree. And such scalability SHOULD make Wright’s even “more correct” and more potent.
• Even if we put aside the more soft-oriented areas such as artificial intelligence and fintech, and just consider the more hardware and manufacturing-oriented areas such as EV and robotics, the prospect of scalability is still very appealing. The current technology landscape is completely different from what the traditional auto and airplane industry has gone through over the past 100 years. Disruptive technologies like robotics, the fusion of software and hardware, autonomous vehicles and machines, have unpredictable potential to cutting production costs.

ARKK – Why I buy the law but not the fund?

After the positives, now we move on to the negative issues. The top two risks and uncertainties are limited data and a very uncertain timeframe for many of the ARKK holdings. To me, these uncertainties and risks are so large that they make any effort for active selection a futile attempt.

First, uncertainties. The nature of Wright’s law requires a long history of data to make any meaningful interpretation. One of the key operative words in Wright’s is cumulative (for every doubling of CUMULATIVE production, the labor requirement was reduced by 10-15% on a per-unit basis). As a result, data points from recent history (like recent quarters and even recent years) are insufficient. Wright’s Law requires data throughout a long historical period to determine the cost declines associated with every cumulative doubling of production.

Take the cost of industrial robots (one of the ARKK focus areas) as an example. As you can see from the chart below, if you only use the data points between 2005 and 2014 (which is already 10 years’ worth of data), you would predict a very different cost reduction curve than that obtained by using the entire set of data available since 1995.

This is precisely the first major issue for many of ARKK’s holdings – there are not that many data points available (yet) and the uncertainties can be so large even if they do indeed follow Wright’s law.

Second, Wright’s law is what I call a steady-state law. Meaning it tells us what EVENTUALLY would happen (which cost will be reduced by 10 to 15% per doubling of cumulative production). However, it does not tell us HOW QUICKLY such reduction would happen.

In investing, many of us have heard the cliche that the market can stay irrational longer than you can stay solvent. To paraphrase, it can take longer for Wright’s law to kick in than for the fund to stay alive.

Let’s look at another ARKK focus area (EV), as represented by its flagship holding Tesla (TSLA). The chart below shows the average cost per vehicle for TSLA as a function of its cumulative vehicle deliveries since 2013 (the year TSLA first started delivering vehicles). The chart is directly taken from my earlier article here and details won’t be repeated here. I will just directly get to the key points:

• Let me first admit the uncertainties in my analysis shown in the chart. The slope of the Wright’s Law line depends on many factors – like breaking it down to models, splitting labor costs from material costs, Shanghai factory vs other factories, et al. Not all these inputs are available (at least not publicly available). But my feeling is that my results here are on the generous side (i.e., exaggerating the cost reduction TSLA’s has achieved so far) as I lumped all costs in the analysis.
• Now even under the above generous treatment, its cost reduction has not been following Wright’s law so far – more than 8 years since it started delivering. It has not even followed the lower bound of 10% (as shown by the orange line).
• Looking forward, there are good reasons why TSLA can converge to the law as mentioned in my earlier article. However, the timeframe remains very uncertain.

XBI provides a workaround for the bio subsector

Now if I sounded too negative in the previous section, let me clarify:

• There is no easy workaround when it comes to predicting the future, especially in the realm of highly innovative and disruptive technologies.
• I stay away from the ARKK fund not because Wright’s law is incorrect or Cathie Wood and her team applied the law incorrectly.
• I stay away because the risks mentioned above (limited data and uncertain timeframe) are so large that they make any effort for active selection a futile attempt to me.

And this leads me to the SPDR S&P Biotech ETF (XBI) provides a sounder alternative -at least for investors who want exposure to the biotechnology subsector. Its passive and equal-weight approach makes better sense to me when it comes to highly uncertain bets.

The next chart provides a recap of its performance since inception in 2007, compared to the overall market (represented by SPY) and the large cap-oriented IBB. As can be seen, XBI returned more than 610%, significantly higher than the overall market and also the large-cap-oriented IBB. I do not think the past performance was an accident; and for the reasons to be discussed further below, I believe such a trend will continue into the future also.

As illustrated more by the next two charts, comparing the top 10 holdings XBI and IBB. As can be seen, XBI holds some of the most innovative small-cap biotech companies at an equal weight (almost). The top 10 holdings occupy only about 8.6% of the total assets. In contrast, IBB is indexed by market capitalization, and as result is top-heavy. As you can see, the top 10 holdings are all well-established large-cap businesses and occupy more than 50% of the total asset. Therefore, in a sense, I view investment in XBI as a venture capital investment on biotech startups with equal bets – an approach I fundamentally subscribe to.

XBI – Why an equal-weight approach is better?

The answer, in the end, lies in what risk really is – or what you believe it is. The bottom line is that, with venture capital type investment (like many of the holdings in both ARKK and XBI), an equal weight bet approach is the optimal approach. As the saying goes, the stakes are so high that we cannot afford not to be in all of them.

The best explanation of the above approach that I’ve ever read myself is Nassim Nicholas Taleb’s book entitled Fooled By Randomness. A great book to really understand what risk really is and what it is not. Here I will use two simple examples to illustrate.

Example 1. Consider the following two independent bets:

Bet A: an investment (e.g., Apple in my view) that has a 50% of chance of losing 20% and 50% chance of gaining 40%. You can easily confirm bet A has an expected return of +10%.

Bet B: an investment (e.g., Treasury bond in my view) that has a 50% chance of losing 20% and 50% chance of gaining 20%. You can easily confirm bet B has an expected return of 0%.

How would you bet? For me, it is a no-brainer – I would bet on A with heavy concentration (which is what I actually do – my real portfolio currently bets much more heavily on Apple-like equities than on treasury bonds).

Example 2. Now consider the following two independent bets:

Bet C: a 45% of chance of losing everything and a 55% chance of doubling your money. You can easily confirm bet C has an expected return of +10%.

Bet D: a 40% chance of losing everything and 60% chance of doubling. You can easily confirm bet D has an expected return of +20%.

Now, how would you bet? Would you bet heavily (or exclusively) on D? Statistics 101 teaches you to bet all your money on C because it offers a higher expected return.

But in real life, I would never do that, and I suggest you do not do that either. Always bet on A and B in some proportion. In real life, you do not want to maximize your expected value. You want to maximize your ODDS OF SURVIVAL. In other words, you can only afford to optimize return until you’ve ensured enough odds for survival.

The key difference between examples 1 and 2 here is precisely the issue of survival. In example 1, I can actively optimize my returns because neither bet is likely to cause a survival issue. But in example 2, either bet has large odds of causing survival issues. Betting on both of them, therefore, may reduce my expected return, but provides much better odds for survival (and BTW, betting equally on C and D roughly cuts my odds of ruin, i.e., total loss, by a half).

This is why I subscribe to the betting strategies of XBI but not ARKK. To me, both XBI and ARKK are in the category of example 2. And XBI bets in a sounder approach – an approach that emphasizes survival before expected return.

XBI – Three more positives

Besides the sounder approach, XBI offers three more positives:

• XBI focuses on smaller biotech companies, as can be seen below, it holds 187 companies with a median market cap of only $1B and a weighted average market cap of about $10B. As a result, many of its holdings are targets for acquisitions and many have been acquired at a considerable premium. In contrast, the weighted avg market cap in ARKK is $113 B.
• Many holdings in XBI are cash positive, and the fund itself is cash positive and earning positive too. Its price to cash flow multiples is about 13.5x and PE ratio is about 14x. in contrast, most of the holdings in ARKK are cash negative. Only 13 of its 35 holdings have positive cash flow. According to Yahoo Finance, the fund has a price to cash flow ratio of about 30.7x overall (and TSLA contributed most of the cash flow), and the fund has no positive earnings.
• XBI’s valuation is currently at a secular bottom as you can see from the third chart below. We have been holding and tracking this fund for years. The following chart shows my personal record of its price/cash flow ratio I’ve recorded at some of the critical market junctures from the fund’s website. As can be seen, its long-term average is about 16.9x. And now it’s only valued at 13.5x. It is not only discounted from the historical average, it is discounted by more than 1 standard deviation.

Summary and conclusion

I stay away from the ARKK fund not because Wright’s law is incorrect or Cathie Wood and her team applied the law incorrectly. I stay away because of the risks of limited data and uncertain timeframe. More specifically,

• I have no doubt the core areas ARKK identified will eventually benefit from Wright’s law. However, just like the market can stay irrational longer than you can stay solvent, the law can kick in more slowly than the fund can stay alive. An uncertain timeframe and lack of positive cash flow is a risky combination.
• There is no easy workaround for investors who want exposure to highly disruptive and speculative stocks. After all, making predictions is hard, especially about the future. Although at least for the biotechnology subsector, XBI provides a sounder alternative.
• I am more comfortable with XBI’s approach – an approach that emphasizes survival before expected return. Other considerations include A) many of XBI holdings are targets for acquisitions, B) XBI is cash and earning positive both at the fund level and holding level for many of its holdings, and C) XBI’s valuation is currently at a secular bottom.

Although XBI has its own risks. Investment in small-cap companies in their early stage is highly speculative. And the biotech sector certainly has earned its notorious reputation for featuring an almost boom-or-bust binary outcome. As a result, the fund can suffer large short-term volatility risks. As you can see from the following table, XBI has the most volatility on a daily, monthly, or annualized volatility as highlighted in the orange box. Its volatility almost doubles that of the overall market.