Updated: Jul 1, 2022
Data from 19 industries across more than a decade do not support this assertion. Why, and where do we go from here?
In February 2021, Warren Buffet published his annual letter to shareholders for the year 2020. This is one of his more notable letters in the past decade, and, as always, full of useful information. Towards the end of the report, Mr. Buffet reveals an interesting data point about his company, Berkshire Hathaway:
“Recently, I learned a fact about our company that I had never suspected: Berkshire owns American-based property, plant and equipment – the sort of assets that make up the “business infrastructure” of our country – with a GAAP valuation exceeding the amount owned by any other U.S. company. Berkshire’s depreciated cost of these domestic “fixed assets” is $154 billion. Next in line on this list is AT&T, with property, plant and equipment of $127 billion.”
He goes on to say:
“Our leadership in fixed-asset ownership, I should add, does not, in itself, signal an investment triumph. The best results occur at companies that require minimal assets to conduct high-margin businesses – and offer goods or services that will expand their sales volume with only minor needs for additional capital." (bold emphasis added here).
Many people have attributed global supply chain challenges of the past 18 months to supply chains simply being “too lean.” Article after article has proclaimed that decades of cost cutting have made supply chains vulnerable, indeed fragile to volatility. In this interpretation, “lean” equals minimal levels of inventory, capacity, and labor, with very little room for error. But how does one know what minimal means? And how do we make this assertion that supply chains are too “lean?” Are Warren Buffet’s $154B in domestic fixed assets too low? (For reference, Berkshire Hathaway had a total of $187B in property, plant, and equipment (PP&E, net of depreciation) on its balance sheet at the end of 2020 (the $154B number he quotes is the USA portion).
It seems intuitive that companies would strive for “leaner” operating models, and that over time they would, as Warren Buffet notes, create models that “expand sales volume with only minor needs for additional capital.” Furthermore, decade-long digitalization efforts would presumably show that software is replacing physical assets in alignment with the thesis that “software is eating the world.” Furthermore, assets are made up of metals, petroleum products, and other commodities and consume a lot of resources to create, something we should be looking to reduce to have less of an impact on the planet.
But more than a decade worth of data doesn’t support any of this. Industries require just as much or more fixed assets per dollar of revenue today and operate with fewer inventory turns than they did a decade ago. How can this be if the prevailing orthodoxy is that supply chains have become too lean? Let’s investigate, try to explain what is going on, and then offer some suggestions on what should be next for companies and industries.
Property, Plant, and Equipment – the Assets Needed to Make, Move, and Store Product
Let’s first examine PP&E, or as Warren Buffet calls it “business infrastructure.” The chart below is a list of the top 20 public companies globally based on the net value of PP&E on their balance sheets at the end of 2020. (Note: this analysis includes operating leases as part of PP&E). Berkshire Hathaway is indeed number 1 for US companies, but comes in at number six globally, behind four oil and gas companies and a utility. (Note: this looks at all industries; normally we would not include Berkshire Hathaway in a supply chain analysis since they are considered in the insurance industry).
As you can see, the list is dominated by oil and gas, utilities, and telecommunications companies. It’s also interesting to note that there are two retailers on the list: a brick-and-mortar retailer (Walmart) and an internet retailer (Amazon). Amazon is on the list not just because of its warehouses, but because of the data centers it needs to run its own operations as well as those of other companies through Amazon Web Services. Amazon added $50B in PP&E in 2020 and another $50B in the first three quarters of 2021, effectively doubling its fixed asset base in a little over 20 months (through September of 2021, its net PP&E stood at $199.3B).
In other words, Amazon increased its physical asset base in 20 months by the same amount it had accumulated in its first 25 years in business. Furthermore, it had $57B in capital expenditures over the past twelve months, which is 40% of the total capital expenditures of the 180 global retail companies we track. In a year in which the word “unprecedented” has been used too much, this is a stunning pace of investment – of Rockefeller and Carnegie-like scale. Consider that this is roughly six percent of the $1T infrastructure bill just passed by the US Congress, spent by one company in the span of one year. From the time that Warren Buffet penned his 2020 letter to the end of the third quarter of 2021, Amazon passed Berkshire as the US leader in fixed assets. Its growth will likely slow, but it won’t be long before Amazon has the largest physical asset infrastructure of any company in the world.
It might also come as a surprise that some software-intensive companies have large physical asset infrastructures. For example, at the end of 2020 Alphabet (Google) had $97B in PP&E, which is 25% more than General Motors. Likewise, Microsoft and Meta Platforms (Facebook) had $71B and $55B in PP&E, respectively. Both Alphabet and Microsoft grew their physical infrastructures twice as fast as sales in 2020. This is due to the high cost of data centers and the high level of competition in the market for cloud services.
The absolute level of physical assets tells only part of the story. We like to look at the “asset intensity” of a business, defined as PP&E (net, including operating leases) divided by revenue. This is an expression of the amount of money a company must invest in physical assets to generate a dollar of revenue and is critical to understanding if we are getting leaner. For example, a lot of Mr. Buffet’s physical assets are tied up in two very asset-intensive industries: railroads and energy. Railroads and energy companies of the type owned by Buffet typically require more than $2 worth of assets for every $1 of revenue (and in many cases much more). Furthermore, they must continue to invest in these physical assets at a rate of 15% or more of revenue every year (this is expressed as capital expenditures).
Figure 2 shows the asset intensity of 19 industries with supply chains for the past twelve years. The chart is based on data from 2,375 global public companies, representing close to $31T in revenue. Asset intensity is calculated by adding up all the net PP&E for all companies in an industry and then dividing by all the revenue for all companies in an industry.
This chart shows clearly that companies have not become “leaner” regarding fixed assets, at least not since 2010. In fact, if you look at trailing twelve month (TTM), 2019 and 2018 numbers (considering 2020 is a pandemic-induced revenue trough year), asset intensity is up in most industries when compared to 2010. These data clearly do not support the often quoted “cost cutting has caused supply chains to become too lean.”
What About Inventory (Materials)?
Global inventory turns are about 5.9 on a trailing twelve-month basis. This means that the world is sitting on about two months of materials in various forms (raw materials, WIP, finished goods) at any given point in time. Is this too lean? Should it be three months?
How have inventory turns evolved over the course of the past ten years? Just as asset intensity has gone up in the past ten years, inventory turns have gone down in most industries. Figure 3 shows aggregate inventory turns for 19 industries with supply chains from 2010 to today. Aggregate inventory turns are calculated by adding up all the cost-of-goods sold for all companies in an industry and dividing by all inventory for all companies in an industry.
Figure 3 shows that inventory turns have been relatively flat but have gone down marginally in most industries when comparing TTM results and 2019 results with 2010. These data clearly do not support the often quoted “cost cutting has caused supply chains to become too lean.”
What About Labor?
It’s clear there’s a labor shortage across today’s global supply chains, particularly in critical need areas such as trucking. The following table shows the amount of revenue per person in each industry, calculated by adding up all the revenue for all the companies in an industry and then dividing by the sum of all the people working at those companies (again, this chart is based on 2,375 public companies representing $31T in revenue on a trailing twelve-month (TTM) basis.
This is a measure of productivity, since revenue is a measure of output. However, it does not reveal if headcount was grown by offshoring functions to low-cost countries, thereby lowering the cost per head. In any event, on the surface, it reveals that people productivity is relatively flat across the decade, which is consistent with what economists have been saying. Interestingly, it does show a significant uptick in productivity on a TTM basis, which is also consistent with what economists have been saying.
Nonetheless, these data clearly do not support the often quoted “cost cutting has caused supply chains to become too lean.”
What’s the Story, Jerry?
If companies haven’t gotten any leaner, then what have they been doing for the past decade or more? One might look at the previous charts for fixed assets, inventory, and people and conclude that companies haven’t done anything for the past decade (at least when aggregated at an industry level). This couldn’t be further from the truth. What companies have been doing – particularly leading companies – is accruing more and more choices to consumers and providing better, faster, and more technologically sophisticated products and services (particularly in delivery). This is the strongest of all long-term trends and has had the greatest impact on supply chains in the past two decades. Amazon added $100B in fixed assets in the past 20 months not just to continue this trend, but to accelerate it. The time between when companies introduce nice-to-have features and services and when such features become expectations has collapsed to zero. This trend was temporarily upended by the pandemic, but now continues unabated.
As a consumer, think back ten years and compare what you can get today versus what you could get then. The amount of consumer choice in product and service today is staggering and is likely to continue to grow, adding to the complexity of supply chains. I address this topic in more detail in my article Consumer Choice, Productivity, and Managing Supply Chains.
The customer has come to expect myriad choices in SKU counts, products, and product variants, including product content, packaging and size, to name a few. The customer has likewise come to expect myriad choices in delivery techniques, including in-store, buy online and pickup in store, curbside service, and delivery to home or wherever you happen to be at a given point in time. All for free. Therefore, for roughly the same amount of fixed assets, inventory, and labor as ten years ago, we are getting much more choice. In that sense, assets, inventory, and labor on a per-SKU basis have gone down.
Furthermore, while companies were in the process of providing more products, SKUs, and product variants, along with more choice in delivery (buy online, buy in store, buy online and pickup in store, curbside service, et. al.), they were also globalizing their supply chains to deliver such choice at low prices. These two forces – on the front end and the back end – collided in the face of a persistent and ubiquitous volatility event called the pandemic.
Thirty years of growing complexity on the front end and growing complexity on the back end has given us the system we have today. The system works reasonably well when variability is limited to a manageable range, and when volatility events are isolated to parts of the system that can be cordoned off. However, as we know, variability and volatility have been increasing at a steady pace for the past twenty years (See the world uncertainty index). As the system has increased in complexity and as the world has come more and more to rely on it, the impact of volatility events has become more widespread and much more visible. For a discussion on variability and volatility see “Supply Chain Volatility?”
And then came the pandemic, the mother of all volatility events – both persistent, deep, and ubiquitous. It has shined a bright light on how the end-to-end system works – ordering from retailers and consumers to factories 10,000 miles away followed by production and all the myriad handoffs and modes to get the product into the hands of the end consumer.
Therefore, supply chains have evolved towards higher and higher levels of complexity while attempting to deliver higher and higher levels of precision. When a monkey wrench is thrown between these two forces, the entire system can oscillate like the Tacoma Narrows bridge. The pandemic, with its simultaneous and global effect on both demand and supply is the mother of all monkey wrenches.
So, while supply chains have not gotten leaner, they have certainly gotten more complex. This is both on the demand side, where the products and services provided have gotten substantially more diverse and precise, but also on the supply side where the number of handoffs, relationships, and organizations involved has grown substantially.
The failure rate of a such a complex system with components that depend on each other is a multiplicative factor of the failure rate of each of its components. For example, in a system with ten subsystems that each have a process capability of 98%, the failure rate of the overall system is roughly 20%. In the pandemic environment, with process capabilities oscillating all over the place it is little wonder that it was difficult to predict output.
So, What’s the Answer?
If data reveal that companies are not becoming leaner, then what’s the answer to the apparent fragility of our existing supply chains? Do we already have enough fixed assets, inventory, and people for the revenue we are generating? Maybe, maybe not.
One answer is to simultaneously raise the level of fixed assets, inventory, and labor everywhere. This is unlikely to happen and nonetheless impractical. It goes not just against the economic motives of individual companies but is at odds with a planet that needs us to become more efficient with our asset usage, not less.
Will supply chains look a lot different ten years from now? Will companies have a lot more fixed assets, inventory, and people? If so, will the shifts at the company level be enough to show up in industry and economy level numbers?
As Willy Shih of Harvard Business School points out in his recent WSJ article, company economic motives will have a lot to say about the breadth and depth of changes that are made. So will consumers. When push comes to shove, both companies and consumers do not necessarily put their money where their mouths are.
In other words, companies will opt for the most profitable path and consumers will opt for the lower cost item. Game theory comes into play here as companies ask the question – am I making myself less competitive; what is the competition doing? Consumers ask similar questions – am I paying more than others?
Will consumers pay a higher price for a product that is made in a lower CO2, regional supply chain? If the product had a CO2 label, would they choose the product with the lower value if it cost more? Or will regulators apply taxes to equalize the prices? All of these are open questions. So far, outside of a small cohort, consumers have not been willing to shift from the lower price product. A lot of this is tied up in incentive and value systems and the need for large parts of populations to have low prices to live a reasonable life.
Largescale reshoring or even near shoring, along with holding extra buffer capacity (fixed assets), inventory and people to withstand shocks is unlikely, at least to the extent that it will show up in macro, aggregate industry numbers five or ten years from now. Largescale reshoring requires skilled labor at reasonable costs, which is a limiting factor for mature economies. As Figure 5 shows, the world’s population is skewed heavily towards Asia.
That said, it is likely that companies will seek surgical strategies, better understanding their risks and making educated bets on where to mitigate those risks with a combination of capacity, inventory, and people. Many are already doing so. This includes dual or multiple sources of supply for critical components and materials.
Furthermore, there is a possibility that these shifts will be “nudged” along by an alignment of factors: 1) regulatory actions; 2) a need for climate change initiatives; and 3) a desire to reduce supply chain risk at both government and company levels. Carbon reduction alone should be a strong reason for companies to shorten their supply chains. Even a moderately complex assembled product today has tens of thousands of frequent flyer miles attached to it, along with the associated carbon footprint.
Technology to the Rescue?
In many ways, it’s innovation and technology that got us into this problem of fragility, starting with the invention of the shipping container sixty years ago. As techno-optimists, we also think technology is the long-term way out of it.
Sure, many will say let’s reduce the complexity of supply chains through SKU reduction, less consumption and structural change. All of this will be helpful. However, a more practical and forward-thinking approach is to think of how technology can get us out of this problem, by better managing complexity. Consider the fact that we have only started to scratch the surface on connected devices and consider the fact that 5G has only recently started to roll out.
The bottom line is that if we are going to continue to provide higher and higher levels of precision to customers and at the same time reduce fragility, then everything in supply chains will have to be instrumented and connected for visibility, synchronization, and decision making. Many elements across the supply chain are simply not connected or are connected manually with considerable levels of latency and error.
A practical, doable approach to greater resilience will include a combination of the following:
Surgical reshoring, nearshoring, and regionalization (probably amounting to less than 10% of overall fixed assets)
Capacity increases at critical, common bottleneck points (e.g. ports) through automation and other means
Dual and multiple sources of supply, including China +X strategies
Continuous supply chain design allowing for rapid adaptation of supply chain structure and policies
Rapid growth of instrumentation and real-time connection to assets, inventory, and people and integration of these connections to tactical and operational software
Visibility platforms to connect assets, inventory, and people and operationalize the connections and integrate them to other systems for insights and decision making
Rapid and intelligent tactical and operational decision-making, enabling agility and synchronization of operations
Widespread adoption of scenarios for tactical decision making and risk management
Risk-based planning as a standard part of monthly tactical planning
Widespread adoption of collaboration between parties along the chain
Government regulations regarding carbon footprint, acting as a catalyst for shorter supply chains and demand shifts
Government incentives and regulations to reduce geopolitical risk to certain supply chains (pharmaceuticals, semiconductors, healthcare equipment, rare metals, et. al.) resulting in regionalization and geo-politically aligned supply chains.