Sauer: AMZN Is Just Getting Started


Courtesy of Matt Sauer, MFSWNB Investments:

A Day At The Beach

Imagine a beach a mile long with two competitors selling cool drinks, where should the two competitors locate their pushcarts? Hotelling’s seminal work suggests that they would place the carts right next to one another to maximize their potential customer pool. Letting the other cart encroach on a larger segment of the beach does not maximize sales. The bisected beach becomes the Nash equilibrium where neither party can improve their share of the market.

Now imagine a beach with two carts in the middle but the customers can remain in their original spot and use a phone to order their drinks. Assuming a smooth delivery by a near infinite number of servers, all but the customers seating next to the cart will begin to usethe new service. 

Hotelling’s original example of pushcarts on the beach describes two players that will split share but not every day. The share might be split 55/45 one day due to more beachgoers enjoying one side of the beach but then we would expect it to reverse back other days. The market share and profits are mean reverting, we expect that they bob around the mean and follow a normally distributed function. Statistics would dictate that it is a very low probability event that one cart gathers all the share while the other garners none. The odds of this are very low and are considered a “tail event” because of the traditional shape of a Gaussian distribution.

However, our modernized example is not mean reverting when it comes to share and profits, it is regressing to the tail. One of the properties of fat tail events is that every expected excess over a threshold increases, given that the threshold increases, the excess will increase. This is a description of the statement that the “winner takes all”, but it also makes the point that the wins are by larger and larger margins.

As the wins grow in magnitude there is not a business strategy capable of slowing down the momentum besides drastic price reduction. Markets that were once mean reverting compete on price and quality; our pushcart example is driven by distribution as the sole decision input by the consumer leaving the incumbents with no strategic moves.

Every smart phone user that feels that the value proposition is improved is a customer leaving the old profit pool and creating a new one. This new profit pool is ending the ability for the original pool to regress to the original mean (measured in dollars). The effect of disruption in an industry is to transfer the profit pools from mean regressing to tail regressing and that is not a small point for investors.

Why do we say it is regressing to the tail? Because as Bent Flyvbjerg writes about certain kinds of events such as Covid-19, there is a non-vanishing probability density function towards infinity. Or in other words, the tail becomes fatter and the probability of an extreme event increases while the events become more extreme. In our example, the probability that the share is lost to the smartphone app is constantly growing. There is no longer a reversion to the mean for the pushcart, it is regressing to what was once considered an improbable tail event.

When the underlying function is profitability, the disruptive event has easily drained the profit pool of the pushcarts. This is one specific and simple case but let’s move to a larger example. 

What Amazon Knows

Amazon is the disruptive force that fattens the tail in multiple industries and has been fattening the tail of the overall retail market in aggregate for over 20 years. Amazon’s market value suggests that the market is aware of the aggregate profit pool shift from most of the retail and is incorporating the probability of reaping the profits in the future. Add on some new growth businesses to value and the company is valued more on probability of outcomes than today’s price/earnings ratio.

As the definition of the regression to the tail suggests greater and greater tail events (or in this case profits flowing to Amazon), there is a stronger case to be made. As the other retailers become weaker, the probability that the profits are in what was once the fat tail increases and become larger. As Amazon exceeds a market share threshold, their expected gains are larger not smaller due to tail regression. Manyfinancial models incorporate share growth slowing and may underestimate the endgame for Amazon. Think about Tesla in this context as well.

The Wisdom of Crowds

The market impacts arising from this viewpoint are substantial. Value investing is predicated on the conviction that there is a reversion to the mean. Despite over 7 years of outperformance by growth, we see no path to highly levered value stocks outperforming FAANG (Facebook,Apple,Amazon,Netfix and Google) and Microsoft outside of a temporary trading bounce. As we have pointed out before in our description of the K-shaped market recovery, FAANG has business segments that have become a substitute product for travel, entertainment, and leisure. Perhaps the encroachment into those profit pools is small now, but if the reversion to the tail occurs look for larger and larger share gains.

Investors produce stock valuations by forecasting the discounted cash flow of the company and discounting such flows for the function of time. There are as many as 80-100 variables embedded in the model including share, price, interest rates and reallocation of capital. The models have traditionally been based on reversion to the mean while that assumption drives the illusion of a discount to “intrinsic value”. The market has been ahead of the analysts in understanding the impact of regressing to the tail. Invest accordingly.

--Matt Sauer, MWSWNB Investments. Originally published here.


Capital Markets