原文信息： - 标题：End of Mean Reversion? - 作者：John Huber - 发表时间：2021-01-04 - 链接：[Saber Capital](https://sabercapitalmgt.com/end-of-mean-reversion/) 翻译信息： - 参考翻译：无 - 翻译：Ponge --- [Aaron Edelheit of Mindset Capital wrote a post](https://mindsetvalue.substack.com/p/power-law-dynamics) about a talk he had with my friend and fellow investor Fred Liu. One of the concepts discussed was the Power Law Distribution. Applied to the market, this means most of the value created comes from a small minority of companies. This is the 80/20 Rule. Baseball geeks like me reference the term slugging percentage. Mindset Capital 的 Aaron Edelheit 写过[一篇文章](https://mindsetvalue.substack.com/p/power-law-dynamics)，讲的是他和我的朋友、投资伙伴 Fred Liu 的一次谈话，讨论的内容之一是幂律分布。这个规律在资本市场上表现为少数公司创造了整个市场大部分的价值。这就是 80/20 法则。对于像我这样的棒球爱好者来说，用“长打率”这个概念可能更好理解。 # End of Mean Reversion? 均值回归的终结？ Aaron’s main point in his post is that the Power Law appears to be impacting one of the most reliable laws of capitalism: mean reversion. 在该文章中，Aaron 的主要观点是，幂律法则似乎正在冲击资本主义最可靠的法则之一：均值回归。 I’ve been tinkering with this idea for years, trying to better understand why so many great companies seem to defy the laws of gravity when it comes to returns on capital, profitability, and size generally. 多年以来，我一直在思考这个问题，想要更好地理解为什么那么多伟大的公司在资本回报率、盈利能力和规模方面似乎无视万有引力定律。 On one hand, the internet has lowered barriers to entry, which is good for competition. Starting a business has never been easier. Big tech companies collect a tax on this burgeoning small business ecosystem, but in return they provide valuable services. Instagram, YouTube and TikTok help you build an audience, Shopify might provide the backend infrastructure to help you sell online, Stripe handles payments, AWS lets you scale computing power and storage needs without big capital investments, and Google and Facebook help you find new customers. If you’re in retail, you might deal with just one tax collector and have Amazon handle all of the above. These services can become costly as your business grows, but none require a big up front investment. 一方面，互联网降低了进入壁垒，这对竞争是有利的。创业从未如此简单。大型科技公司对这个新兴的小企业生态系统征税，但作为回报，大公司也提供了有价值的服务。Instagram、YouTube 和 TikTok 可以帮助你建立受众群体，Shopify 可以提供后端基础设施来帮助你在线销售，Stripe 可以处理支付环节，AWS 可以让你在不需要大量资本投资的情况下扩展计算能力和存储需求，Google 和 Facebook 可以帮你找到新客户。如果你从事零售行业，你可能只需要和一个税务员打交道，然后让 Amazon 来处理上述所有事宜。虽然随着业务的增长，这些服务会变得昂贵，但没有一项服务需要大量的前期投资。 A few years ago I wrote about how digital upstarts are able to leverage social media and self-serve advertising on Google and Facebook to [build businesses that can make even the largest incumbents vulnerable](https://sabercapitalmgt.com/wp-content/uploads/2013/03/Saber-Capital-2017-06-13-Investor-Note-Most-Important-Moat-1.pdf). The fact that Coca-Cola and Proctor & Gamble are seeing cracks in their once-impenetrable moats is a testament to the competition that big tech has ironically helped foster (given the antitrust scrutiny they are under). 几年前，我曾写过一篇文章，讲述了数字新贵们如何利用社交媒体、并在Google和Facebook上投放自助广告来开展自己的业务，[这些业务甚至会冲击现存龙头企业的市场地位](https://sabercapitalmgt.com/wp-content/uploads/2013/03/Saber-Capital-2017-06-13-Investor-Note-Most-Important-Moat-1.pdf)。可口可乐和宝洁公司曾经坚不可摧的护城河也出现了裂痕，这一事实证明了竞争的存在。具有讽刺意味的是，大型科技公司促进了行业竞争的出现，而他们自身却正在接受反托拉斯审查。 But while competition has never been fiercer, the internet has created certain business models that — once a certain tipping point is reached in size — get more dominant as they grow. 另一方面，尽管竞争从未如此激烈，但互联网创造了某些商业模式，一旦规模达到某个临界点，这些商业模式会随着规模的增长而占据更大的市场主导地位。 Why is the Power Law so strong in the economy today? Why do excess returns in some businesses seem more likely to expand than to mean-revert? 为什么幂律法则在当今的经济中如此强大？为什么有些企业的超额收益有可能越来越大，而不是均值回归？ There are many reasons, but I wrote down four to share here: 原因有很多，我这里分享四点理由： **1. Digital vs Physical Stuff** **1、数字内容与实物** I once read that all the electrons that make up the internet weigh less than a single strawberry, or about 50 grams. In a way, this is a one reason why mean reversion has lost part of its bite. 我曾经看到过一种说法，说构成互联网的所有电子的重量比一颗草莓还轻，也就是 50 克左右。从某种意义上讲，这是均值回归失去部分作用的原因之一。 When compared to digital stuff, physical stuff is heavier, it takes up more space, it’s harder to transport, and it takes longer to create. Weight, space, shipping, and time all tie up capital. 与数字内容相比，实体的东西更重，占用空间更大，运输更难，制造时间更长。重量、空间、运输和时间都和资本紧密相关。 Digital stuff can be created instantly and can be scaled rapidly. Each new customer can be served immediately for virtually no cost outside of the payments processing fee. Facebook has spent billions on computing power to support the foundation of its business, but 2020’s demand for 300 million new users across its networks can be met almost immediately with an infinite amount of digital supply. 数字内容可以被快速创造，并迅速扩散。每一个新用户都可以立即享受服务，除了支付时的手续费之外几乎没有任何成本。Facebook 在算力上花了数十亿美元来夯实业务基础，相比之下，无限数量的数字内容几乎可以立即满足其网络中 2020 年的 3 亿新用户的需求（而不需要额外的成本）。 Zoom is a great case study. Needless to say, demand for Zoom skyrocketed this year. Physical investments were required to meet that growth (Oracle and Amazon needed to provide more server capacity, Zoom needed more employees, and had to make various security investments), but despite the growing pains, the massive new demand was accommodated with newly created supply. Zoom’s business grew 4-fold. On the other hand, masks, toilet paper, milk, lumber, and any other physical good that saw demand spikes ended up in a shortage, which puts a ceiling on providers’ sales growth potential. Zoom 是一个很好的案例。毫无疑问，今年 Zoom 的需求量猛增。为了满足这种增长，需要进行实物投资（甲骨文和 Amazon 需要扩充服务器容量，Zoom 需要更多的员工，以及在安保方面各种必要的投入），尽管存在这种成长中的阵痛，但新创造的供应还是满足了大量的新需求。Zoom 的业务增长了 4 倍。相比之下，口罩、卫生纸、牛奶、木材和任何其他需求激增的实物商品最终都出现了短缺，这就限制了供应商的销售增长潜力。 When growth requires bits instead of atoms, there is no such ceiling. Demand can scale very fast, but supply can also scale in lock step, and **those two conditions working in concert is what allows certain businesses grow so fast and become so valuable**. 当增长需要比特而不是原子时，这种增长就没有这样的上限。需求扩张的速度可能非常快，但与此同时，供给也可以同步扩张，**这两个条件共同作用，才使得某些企业发展得如此之快，变得如此有价值。** Buffett noticed this early when he discovered that every time someone clicked on a Geico ad, Google got paid. The toll road Google built cost a huge amount of money in data centers and engineering talent, but every additional click above that fixed cost base was pure profit. 巴菲特很早就注意到了这一点，因为他发现每当有人点击 Geico 广告时，Google 都会获得报酬。Google 建立的“收费公路”在数据中心和工程技术人才方面花费了巨额资金，但在这个固定成本基础之上，每一次额外点击都是纯利润。 **2. “Data Economics Shared”** **2、“数据经济共享”** I believe it was Nick Sleep who coined the term “scale economics shared”. As Costco grew, it decided to forego higher profit margins that could have been achieved from its growing sales over a fixed cost base. Costco instead kept margins stable and lowered prices. In other words, Costco chose to give customers the value that accrued from economies of scale, instead of keeping the value for itself… a short term investment for long-term gain. This was great strategy, and it created more growth, which gave Costco more economies of scale that could be passed down to customers in the form of better prices still… a classic feedback loop. 我相信是 Nick Sleep 创造了“规模经济共享（scale economics shared）”这个词。随着 Costco 的成长，它本可以在固定成本的基础上，通过快速增长的销售额实现更高的利润率，但 Costco 决定放弃这部分利润。Costco 依然维持之前的利润率水平，并进一步降低了商品的价格。换句话说，Costco 选择将规模经济所产生的价值还给客户，而不是将价值据为己有——这是一种短期投资换取长期收益的做法。这是一个伟大的战略，它创造了更多的增长，给 Costco 带来了更多的规模经济，并进一步以更优惠价格的形式让利给顾客——一个经典的反馈循环。 Many of today’s great companies operate in a similar way. But today, the value shared with customers often comes not in the form of better prices but in better products. 当今许多伟大的公司都以类似的方式运作。但时至今日，与客户分享的价值往往不是以更优惠的价格，而是以更好的产品来体现。 The more people use Stripe, the more data they collect, and the better their product gets, which brings in still more users, more data, better products, and so on. Stripe’s customers benefit from these “data economies of scale” because each new customer helps Stripe create new and better products for all existing customers. 使用Stripe的人越多，他们收集的数据就越多，他们的产品也就越好，从而进一步带来更多的用户、更多的数据、更好的产品，如此循环往复。Stripe的客户从这些 "数据规模经济 "中受益，因为每一个新客户都能帮助Stripe为所有现有客户创造新的、更好的产品。 [Packy McCormick had a great post](https://notboring.substack.com/p/apis-all-the-way-down-audio) describing these data feedback loops that Stripe and many other “API First Companies“ possess. The more customers Stripe gets, the more it can spread R&D costs over a larger customer base, resulting in a payments product that is much better (as well as much cheaper) than any customer could build themselves. Like Costco, Stripe’s “data economics shared” will likely strengthen its moat as the business grows, extending any early lead over competing products. Packy McCormick 有一篇[很棒的文章](https://notboring.substack.com/p/apis-all-the-way-down-audio)，描述了 Stripe 和许多其他“API 优先”公司拥有的这些数据反馈循环。Stripe 获得的客户越多，就越能将研发成本分摊到更大的客户群中，从而保证自己的产品比任何客户自己打造的支付产品都要好得多（同时也便宜得多）。像 Costco 一样，随着业务的增长，Stripe 的 "数据经济共享 "策略很可能会加强其护城河，扩大其相比于竞品的早期领先优势。 Google’s algorithm famously worked this way. Many now have a cynical view of advertising business models, but in the early days Google was very much a “data economics shared” business. Google search is free, so the benefit of scale came not in the form of lower prices but in terms of a better search experience: each search gave Google data they used to improve the product, which brought in more users, more improvements, etc… Google used the insights it got from its ever growing volume of data to create better products, sharing those innovations with users. Google 的算法以这种方式闻名。尽管许多人现在对广告的商业模式持愤世嫉俗的看法，但在早期，Google 在很大程度上是一个“数据经济共享”企业。Google 搜索是免费的，所以规模带来的好处不是价格降低，而是更好的搜索体验：每一次搜索都给 Google 提供了他们用来改进产品的数据，从而带来更多的用户，更多的改进等等。Google 利用从不断增长的数据量中获得的洞察力创造出更好的产品，并与用户分享这些创新。 Stripe and Google didn’t lower prices like Costco did; but they used the benefits of scale to give their customers better value like Costco did. Stripe 和 Google 并没有像 Costco 那样降价；但他们利用规模优势，像 Costco 那样为客户提供更好的价值。 All three of these companies had feedback loops, but technology speeds the rate at which the feedback loop spins. 这三家公司都具有反馈循环，但技术加快了反馈循环的旋转速度。 **3. Technology has Enhanced the Power of the Network Effect** **3、技术增强了网络效应的力量** Some of the best businesses in the world are two-sided marketplaces with strong network effects. They are so valuable because they are the only business model I’ve seen where the customer acquisition cost actually declines as the business grows. The larger the network, the more value there is to the next new user, and thus the easier it is to acquire that next user, at least up to a point. 一些世界上最好的企业在双边市场上都具有强大的网络效应。它们之所以如此有价值，是因为它们的商业模式在我所见过的里面非常独特：它们的客户获取成本实际上随着业务增长而下降。业务的网络越大，对新用户的价值就越大，因此在一定程度上，获得下一个用户就越容易。 Facebook is the poster child here. Each new user adds more value to the network overall. Facebook 是这类公司的样本。每一个新用户都会给网络整体增加更多的价值。 But the strongest moats I’ve seen created over the past decade have combined physical investments with the zero-marginal cost economics of the internet. 但在过去十年中，将实体投资与互联网的零边际成本经济相结合的企业，构筑了我见过最强大的护城河。 Copart owns a large percentage of the auto salvage auction market. If you total your car in an accident, the insurance company sends the car to one of Copart’s lots to be sold in an online auction. Copart 公司在汽车报废拍卖市场上占据很大市场份额。如果你的车在事故中全损，保险公司就会把车送到 Copart 的某个拍卖场，进行网上拍卖。 Copart’s network effect is very strong. Insurance companies send their totaled vehicles to Copart because they know Copart has the largest network of dismantlers, dealers, and other buyers; and those buyers all come because Copart has the largest selection of cars for sale. This liquid market provides the highest selection for buyers and ensures the sellers they’ll get fair market value. Copart 公司拥有非常强的网络效应。保险公司将他们报废的车辆运到 Copart 公司，因为他们知道 Copart 公司拥有最大的拆卸商、经销商和其他买家网络；而那些买家也会选择来 Copart 公司，因为在 Copart 公司可选待售的汽车最多。这个流动的市场为买家提供了最多的选择，并确保卖家能获得公平的市场价值。 Copart invested a lot of money over the years to own land which it uses to store these vehicles until they are sold, but Copart doesn’t own the inventory; it just takes a commission connecting buyers and sellers. The auctions are all done online, so each new sale on its network is very high margin. 多年来，Copart 公司投入了大量资金购置土地，用来存放这些车辆，直到它们被出售，但 Copart 公司并没有库存；它只是在连接买家和卖家时收取佣金。拍卖都是在网上进行的，因此其网络上的每笔新交易都有很高的利润率。 It would be very hard to recreate the combination of land assets, technology IP, and industry expertise that Copart has developed over the years. Those fixed costs do present a barrier to entry. But Copart’s growth is so valuable because each new sale above a certain fixed cost level doesn’t cost anything, thanks to the zero-marginal cost nature of online commerce. Copart 公司的重置成本非常高，它包括了多年来开发的土地资产、技术知识产权和行业专业知识。这些固定成本构成了真实的进入壁垒。但 Copart 公司的成长之所以如此有价值，是因为超过某一固定成本水平后，每一单新交易都不需要任何成本，这得益于在线零售的零边际成本特性。 Etsy’s CEO Josh Silverman says that two-sided marketplaces have low barriers to entry, but high barriers to success. They’re very hard to get off the ground, but once established they’re very hard to kill. Etsy 的首席执行官 Josh Silverman 表示，双边市场的进入壁垒很低，但成功壁垒很高。它们很难顺利开始运转，但一旦建立起来就很难被杀死。 The result is a business with returns on capital that defy capitalism’s law of gravity: 这种强网络效应的结果是，一家企业的资本回报率可以无视资本主义的万有引力定律： ![](../Elements/Mean_Reversion/image-1024x116.png) _Copart Returns on Capital (source: Tikr)_ *Copart 公司的资本回报率（来源：Tikr）* Copart also reinvests its earnings into new land to support a larger supply of vehicles, which attracts still more buyers, which means Copart’s network effect only gets stronger as it grows. Copart 公司还将其收益再投资到新的土地上，以支持更多的车辆供应，从而吸引更多的买家，这意味着随着 Copart 公司的发展，其网络效应只会越来越强。 **4. Growth from Others’ Capex** **4、利用其他人的资本支出发展** A recent article on Ebay and Adyen got to thinking about the payments business. Merchants pay to make stuff and have to fund their own inventory, shippers build the distribution networks and buy the trucks, but the companies processing the payments earn most of the profits that come from that aggregate invested capital. Adyen is now larger than Spotify, a big early customer, which is itself more valuable than many of the labels which own the content that Spotify distributes. You can see who reaps most of the value in that ecosystem, and who is funding the ecosystem’s growth. The returns on invested capital are not distributed proportionately. Similarly, PayPal was once a minnow in the Ebay’s ocean, but PYPL now has a market cap nearly 8x the size of its former parent. 最近一篇关于 Ebay 和 Adyen 的文章让我开始思考支付业务。商人出钱制造商品，还要为自己的存货垫付资金，托运人建立配送网络，还要购买卡车，但支付公司赚取了大部分利润，这些利润来自于这些总投入资本。Spotify 是 Adyen 的一个早期大客户，它本身就比很多在 Spotify 上分发音乐内容的厂牌更有价值，但现在 Adyen 已经比 Spotify 规模更大了。你可以看到谁在这个生态系统中捕获了最多的价值，谁在为生态系统的发展提供资金。投入资本的回报并不按投入比例分配。同样，PayPal 曾经是 Ebay 海洋中的一条小鱼，但 PayPal 现在的市值几乎是其前母公司的 8 倍。 A big reason why PayPal became so valuable is it could grow using other people’s money. [I wrote about this point last year when I highlighted an old email that Warren Buffett sent to a Microsoft executive](https://sabercapitalmgt.com/warren-buffett-1997-email-exchange-on-microsoft/). PayPal 之所以变得如此有价值，一个重要原因是它可以用别人的钱来发展。去年，当我在[讨论一封沃伦·巴菲特发给微软高管的旧邮件](https://sabercapitalmgt.com/warren-buffett-1997-email-exchange-on-microsoft/)时，提到了这一点。 Microsoft’s growth was funded by IBM’s investments. Facebook’s growth was funded in part by telecoms’ broadband investments. 微软的发展是由 IBM 的投资资助的。Facebook 的发展一定程度上是由电信公司的宽带投资资助的。 I’ve also noticed a pattern emerging where the current big tech companies (which all were early beneficiaries of the investments made by others) are using their huge free cash flows to reinvest back into their businesses, creating a new crop of companies that are benefiting from someone else’s capex. Like Facebook, Apple has earned enormous returns on the capex budgets of Verizon and AT&T, but Twilio is now benefiting from Apple’s R&D and Facebook’s investments in data centers that were needed to make WhatsApp a global communication platform (smart phone usage generally has fueled Twilio, and WhatsApp is a large customer). 我还注意到一种正在出现的情形，即当前的大型科技公司（它们都是其他公司投资的早期受益者）正在利用其巨大的自由现金流再投资于自身业务，创造出一批受益的新公司，这些公司正从别人的资本支出中获益。与 Facebook 一样，苹果也从 Verizon 和 AT&T 的资本开支预算中获得了巨大的回报，但 Twilio 现在正受益于苹果的研发和 Facebook 对数据中心的投资，这些投资也使得 WhatsApp 成为一家全球通信平台（智能手机的使用总的来说推动了 Twilio 的发展，而 WhatsApp 是 Twilio 的一个大客户）。 Twilio is basically collecting a royalty on messaging between businesses and their customers; and thanks to their recent acquisition of Segment; it’s also now benefiting from the growing amounts of data, stored in someone else’s servers that were paid for with someone else’s capital. Twilio also benefits from the “data economics shared” concept, as more customers create more data which leads to product improvements and better value for customers. Twilio 基本上是在企业与客户之间的信息传递上收取使用费；由于他们最近收购了 Segment，Twilio 现在还受益于数据量的不断增长，而这些数据是存储在别人服务器上的，并且这些服务器是用别人的资金来支付的。Twilio 还受益于“数据经济共享”的概念，因为更多的客户创造了更多的数据，从而为客户带来产品的改进和更好的价值。 Twilio is not in Saber’s portfolio but it is on Saber’s shopping list, and it is one of a number of companies I’m following that can grow largely on the capital investments made by other companies before their time, which means that if demand is there, growth can happen extremely fast. And if growth is combined with one or two of the other advantages outlined above such as network effects or data feedback loops, an early lead can end up being lethal to competitors, and hugely valuable for shareholders. Twilio 并不在 Saber 的投资组合中，但它在 Saber 的“待购清单”上，它是我关注的众多公司之一，这些公司的成长很大程度上依赖于其他公司在此之前的资本投资，这意味着如果有需求，增长可能会非常快。而且，如果其发展结合了上面提到的一项或多项优势（比如网络效应或数据反馈循环），早期的领先优势最终将对竞争对手造成致命威胁，并为股东带来巨大价值。 As Buffett says; the best business is a royalty on the growth of others. 正如巴菲特所言：最好的生意是他人成长的提成费。 # Kodak is Toast “柯达要完蛋了” The Power Law is getting stronger, and if you’re a stock picker, it’s imperative to think critically about how this will impact your investments. 幂律法则正变得越来越强，如果你是一个精选个股的投资者，则必须认真考虑这将如何影响你的投资。 In 1991, more than a decade before Kodak became an obvious value trap, it looked like a good old-fashioned value stock, and Bill Ruane was pitching it to a group that included Warren Buffett and Bill Gates. Gates told Ruane: [“Kodak is toast”](https://twitter.com/JohnHuber72/status/1345775133982674946), which seems obvious at this point but was prescient in 1991. 1991 年，在柯达公司成为一个明显的价值陷阱前十多年，它看起来像一只很好的老式价值股，比尔·鲁恩正在向包括沃伦·巴菲特和比尔·盖茨在内的一群人推销它。盖茨对鲁恩说：“[柯达要完蛋了](https://twitter.com/JohnHuber72/status/1345775133982674946)。”这句话在现在看来是显而易见的，但在 1991 年却有先见之明。 Investing is a balancing act of picking winners and avoiding losers. Both are hard, but I think understanding the reasons behind the strengthening Power Law will help us avoid businesses that are “toast”, while being more open-minded about the upside potential for certain companies that a prior paradigm would not have predicted. 投资既要选择赢家，也要避免输家。这两点都很难，但我认为，理解幂律法则在什么情况下会得到加强，将有助于我们避免那些“要完蛋”的企业，同时对某些企业的上升潜力持更开放的态度，而这些潜力是之前的分析框架无法预测的。 [I wrote in June why I think this is a Stock Picker’s Market](https://sabercapitalmgt.com/stock-pickers-market/). This does not mean most stocks are cheap. Most are not, although a few are very undervalued. But the Power Law has created lots of interesting opportunities to study. Curious minds should find the next decade a lot of fun. 我在 6 月份写过[为什么我认为这是一个适合选股者的市场](https://sabercapitalmgt.com/stock-pickers-market/)。这并不意味着大多数股票都很便宜。尽管有少数股票的价值被非常低估，大多数股票并不便宜。但是幂律法则创造了许多有趣的学习机会。拥有好奇心的人应该会在接下来的十年中发现很多乐趣。