On Sunk Costs

A confession: I’ve become fascinated with sunk costs. The concept kept coming up in several books and articles I’ve read. So, I decided to write an essay about it connecting the ideas.

This essay tries to answer two questions:

  • What are sunk costs?
  • Why study sunk costs?

I reference various sources in this essay that I include links to at the bottom. I would love feedback on this should you want to comment.

What are sunk costs?

Sunk costs are costs that cannot be avoided regardless of the decision you make. Avoidable costs are just the opposite. Said another way, a sunk cost cannot be recouped. For example, if I choose to go to Mexico and make a non-refundable deposit, I have spent that that money whether I am going or not. That cost is now sunk. This is a personal example, but lets give a business related example. From the book, Economics of Strategy:

To illustrate the concept of sunk costs, take the case of an online merchandiser of laser printers. The merchandiser traditionally purchased large quantities of printers from the manufacturer so that it could satisfy rush orders. Increasingly, though, the merchandiser was carrying high inventories, including some lines that the manufacturer no longer produced and would not repurchase. A natural response to this problem would be to put the discontinued lines on sale and reduce inventory. However, the firm’s managers were reluctant to do this. They felt that even in the best of times the margins on their products barely covered overhead, and, by cutting the price, they would be unable to cover the costs of goods sold.”

Management’s reluctance to sell the laser printers in this case is wrong. The cost of buying the printers is now sunk. They cannot undo a decision they made in the past and the historical cost of the printers cannot be avoided. Anchoring on the historical costs of the printers is not helpful. The company should just cut its losses and sell the printers at the best price possible to minimize its losses. When you incur sunk costs, its crucial to make decisions at the margin and let go of the past. The time, money, and effort spent is no longer relevant to future decisions.

Now that we’ve introduced the concept, I want to highlight certain considerations around sunk costs.

A cost being sunk depends on 1) the decision being made and 2) the options at hand.

Using the printer example, the initial cost of the printers is sunk with respect to the current pricing decision. However, the cost would not have been sunk before they ordered the printers. In other words, the cost could have been avoided. Another personal example to drive this point home: many car leases have a “turn in” fee that is customarily waived as long as your next lease is of the same brand. If I make the decision to go with a new brand, that turn in fee is sunk. If I decide to stay with that brand, the turn in fee is avoided. The classification of a cost as sunk is therefore dynamic, not static.

Sunk costs and fixed costs are not necessarily the same.

For example, a rail car going from Norfolk to Santa Fe needs a locomotive and a crew whether it hauls one carload of freight or 20. The cost of the locomotive and the crew is fixed, but they may not be sunk. If the company abandons that route, the locomotive and crew can be redeployed to another route or the locomotive can be sold to competitor. A sunk cost can be fixed, but it doesn’t have to be. In many cases, these two concepts can be confused so it is important to keep them straight.

Sunk costs are both explicit and implicit.

Explicit costs are out of pocket costs that a company will make. Two examples include rent and employee wages. Implicit costs are opportunity costs and can take the form of time, effort, or other resources. Implicit costs may sound fuzzy, but they are just as important. For example, if I spend an hour researching a trip to Jamaica, but decide to go to Mexico, that hour is an implicit sunk cost. I am not getting it back. Time spent on something that doesn’t work out tends to be the ultimate sunk cost. This reminds me of a phrase I’ve heard several times before: “Time is money, and we don’t get a minute back.” Many people recognize explicit costs, but implicit costs are harder to see and may be more important.

Sunk costs are a source of cognitive bias for people, organizations, and governments.

Sunk costs can be really nefarious and interfere with reasoning. An example of this is the “sunk cost fallacy”. This describes a situation where individuals or organizations continue a behavior or project as a result of previously invested resources whether that be money, time, or effort. Economists argue these costs cannot be avoided and you should ignore them as they are not relevant to future decisions. These prior costs should be ignored and you should “cut your losses”. This is hard for people to do once they have set their mind on something, making us really sensitive to sunk costs.

The fallacy operates at many levels. The development of the Concorde is an illustration of this. The original program to develop a supersonic jet was estimated to cost 70 million pounds, but ended up costing 1.3 billion pounds(!) that both the British and French governments had to pay.

This is tough to get over psychologically. The more you put in, the harder it is to give up. The phrases: “We’ve come too far to stop now” or “If we just work a little harder, we’ve invested so much”. These phrases or some variant of them should give you pause and help you realize you may be in a sunk cost situation.

So far, we explained sunk costs with some examples and given several relevant nuances related to them.

Now, I’d like to explain why sunk costs are worth your attention.

Why study sunk costs?

Sunk costs are worth studying for several reasons:

  • Intangible investment tends to be sunk and makes up the majority of investment for the most valuable companies in the world.
  • Understanding sunk costs can provide important insights into strategy and competitive advantage.
  • Sunk costs can help explain market structure.
  • Due to their sunken nature, intangible investments are hard to value and accounting standards do not help in reflecting that value.

Intangible investment tends to be sunk and makes up the majority of investment for the most valuable companies in the world.

From, Capitalism without Capital: The Rise of the Intangible Economy:

The basic economic properties of intangibles make an intangible‐​rich economy behave differently from a tangible‐​rich one. Why? First, intangible investments tend to represent sunk costs; second, they generate spillovers; third, they are likely to be scalable; and fourth, intangible investments tend to possess synergies, or complementarities, whereby they are more valuable together in the right combinations.

In this essay, we will focus on the sunken nature of intangible investment. Here are some common examples of intangible investment:

  • Research and development costs to develop a new drug or technology.
  • Advertising costs to build and develop a brand.
  • Design and development costs to develop a new product and get it to market.
  • Training costs to develop codified procedures to ensure the consistency of some product or service.

This is by no means an exhaustive list, but rather to illustrate some common intangible investments.

Intangible assets tend to be sunk because these costs are not easily recouped nor are they easy to sell once established.

Once a company begins to invest in a research and development project or a brand it is not easy to undo those costs. If you cannot recoup the costs the investment is likely sunk. Many intangible investments are dependent on labor. Research involves paying scientists and software involves paying employees or developers. Firms cannot ask their employees for their wages back if the project does not work out.

Intangible assets (that result from intangible investment) are not easily sold. Tangible assets are much easier to sell as many are mass produced and standardized, allowing secondary markets to develop and determine value. Buildings, land, and machines are useful to many types of businesses so they are easier to sell. Tangible assets are physical making them easier to finance with debt. If a company cannot pay its debts, the lender can seize the physical asset. Many lenders are not comfortable lending against intangible assets. Spoiler alert: there is likely an opportunity here.

This brings us to an important point: if an asset is more specialized, the harder it is to sell. If the asset is harder to sell, that implies the cost (or a portion thereof) is more likely to be sunk. Assets range from general purpose (i.e. land, machines, and buildings) to intangible (i.e. software, brands), with the former being easier to sell and the latter harder to sell. Specialized assets tends to lie somewhere in the middle and it is best to consider these three classifications a continuum through which to view “asset salability” and potential sunk cost implications.

Related to the idea intangible assets are difficult to sell is the idea they are highly complementary to other assets, both intangible and tangible. On one hand, these combinations can create value in non-linear ways. On the other hand, they are very difficult to disentangle or sell. Consider airlines: they represent a bundle of tangible assets such as planes and equipment and intangible assets such as a brand, a network of routes, and a loyalty program.

United chose to securitize its loyalty program probably because it is more feasible than selling it or spinning it off. These assets are all highly complementary to one another making any one of them hard to sell and thus some of the investment sunk. To use another example, consider Toyota and its lean production system: the factories they use to produce cars can be sold, but the lean production techniques developed over many years cannot be sold. However, it is hard to argue the combination of the two do not create immense value.

Intangible assets tend to be sunk given cost cannot be recouped and they are hard to sell. You may ask: what of it? Well, in the last 40+ years, the nature of business investment has changed from that of tangible assets such as factories, machines, and trucks to intangible assets such as knowledge, ideas, brands, software, networks, and training.

This graphic summarizes the growth in and size of intangible investments and intangible assets. In 2018, business investment was estimated to be $25 trillion, of which $21 trillion was intangible. Understanding both the nature of intangible investment and intangible assets will be the key to understanding growth, profitability, and value creation going forward. This involves studying their sunken nature.

Understanding sunk costs can provide important insights into strategy and competitive advantage.

From Economics of Strategy:

Sunk costs are important for the study of strategy, particularly in analyzing rivalry among firms, entry and exit decisions from markets, and decisions to adopt new technologies.”

Industry rivalry is one of Michael Porter’s 5 Forces. These forces help provide insight into the profitability of an industry, with strong forces implying low profit potential and weak forces implying the opposite. If an industry has a few large firms and the industry has large sunk investments, the rivalry could be intense. If rivalry is intense, this likely will lead to pricing wars, fights for market share, and limited profit potential. Sunk costs also drive exit barriers. Higher sunk costs imply a firm will be less likely to exit the industry. These costs can be specialized equipment, purchase commitments for raw materials, or collective bargaining obligations. Anything that keeps a competitor from exiting an industry will limit profit potential. Sunk costs impact both rivalry and a firm’s decision to exit the market.

Now, lets consider a firm who wants to enter a new market. Entry can be thought of an investment an entrant must make to enter a new market. This investment is somewhat sunk as costs cannot be fully recovered if the company exits. Entry is in part a decision that must consider sunk costs. The hope is that post entry profits will exceed sunk entry costs. In others words, a firm hopes the entry decision is NPV positive. There are many types of sunk entry cost in creating or entering a market. These costs could include developing a brand, investing in research, building out a sales force, or buying specialized equipment.

Just as sunk costs can result in rival and exit barriers limiting industry profit, they can serve to erect entry barriers protecting profit if sufficiently high. Profits invite entry and sunk costs may be able to prevent entry. The incumbent may have an advantage relative to the entrant if it has some asymmetry in its favor. A certain type of asymmetry could be sunk costs that the incumbent has borne and the entrant has not.

Consider Boeing and Airbus, who have made large sunk cost investments in facilities, tools, and training. This would represent an incremental cost to an entrant but is sunk to them. Sunk costs are not just these explicit investments but can also be implicit. For example, the time spent establishing customer and supplier relationships can serve as an entry barrier. Finally, consider branding investments. Branding provides an entry barrier given the cost, time, and uncertainty involved in brand building.

It is notable that several well-known entry barriers (or moats) may result from sunk costs. Moats such as Scale Economies, Network Economies, Branding, Switching Costs, and Process Power are all Powers that result in part from sunk costs. For example, right now I am learning to use Roam Research to organize my note taking. The time spent in learning the intricacies of the software and its shortcuts increase my Switching Costs of moving to another note taking software. The time spent is a sunk cost to me. Understanding sunk costs is understanding the nature of competitive advantage and likely how durable it is.

Decisions to adopt new technology can be influenced by sunk costs. The “sunk cost effect” occurs when there is an asymmetry between a firm that has committed to a particular technology and one that has not. The firm that made its technology decision has invested the firm’s resources and capabilities that it won’t get back if it switches to another technology. These costs are sunk. Ideally the firm that has chosen the technology should ignore those costs and choose the best technology going forward. In practice, companies usually don’t. On the other hand, the firm who has not yet chosen its technology is able to more clearly compare the benefits and costs of all alternative technologies to make a better decision.

Firms can also use commitments that involve sunk costs to reason how a competitor may respond to them. Nucor made a decision to adopt a new thin-slab casting technology by inferring its competitor U.S. Steel would not respond with a similar commitment. Nucor correctly reasoned that U.S. Steel incurred sunk costs in modernizing its integrated steel mills in the mid 1980’s and would have no appetite to adopt the thin-slab casting technology. Observing a competitor’s investment in sunk costs can be a useful strategic forecasting tool.

Sunk costs can also affect decisions about whether to adopt new business models. In a prior post, we discussed Counter-Positioning and how Netflix had Power over Blockbuster with its subscription based, DVD by mail business model relative to Blockbuster’s brick and mortar model. The Power and the barrier of Counter-Positioning stems from the leader’s concerns over collateral damage to its existing business model and profits. For Blockbuster, those profits were supported by some level of sunk costs in its stores, people, and organizational resources and capabilities built up over a long period of time. These concerns over collateral damage in Counter-Positioning come from a mix of opportunity costs, sunk costs, and concerns of a firm’s capability to successfully adopt a new business model. Sunk costs can be a driver of a firm’s reluctance to alter its business model.

Sunk costs help explain market structure.

When I refer to market structure, I am using the economic term that defines the underlying competitive nature of the market. This ranges from perfectly competitive markets to monopolies. In a perfectly competitive market, there are many firms who are price takers and tend to price at marginal costs wringing most profit out of the industry. On the other hand, a monopoly will price above its marginal costs and enjoy high profits.

Why do market structures differ and what can explain it? We can start by using some basic microeconomics to explain the differences. In an earlier post, we discussed the concept of minimum efficient scale (or “MES”). This concept explains that most industries have cost curves that exhibit a “U” shape so there is a specific level of output at which costs have been minimized. The MES may be larger when there are larger upfront sunk costs of establishing production, especially relative to the ongoing costs of operation. If we generalize further, a lot of market structures may be explained by the ratio of the total addressable market (or “TAM”) to MES. In other words, the number of competitors would roughly be that ratio. This ratio certainly explains some market structures, but not all of them.

For instance, how does it explain a common market structure with a few large players and many smaller niche competitors? Enter John Sutton, who wrote Sunk Costs and Market Structure. He reasoned that production costs and economies of scale are important, but only tell part of the story. He also pointed out that many consumer industries such as soft drinks, canned foods, and breakfast cereals consist of a market structure with 2-3 large firms with many smaller niche competitors. Clearly, economies of scale in production are important, but do not tell the whole story.

Sutton pointed out many consumers gravitate towards brands. Creating and maintaining these brands required significant sunk cost investments which raises the MES for branded products. This explains many consumer markets with a handful of large brands and many smaller niche players. It should be noted the advent of broadcast television helped raise this MES. Initially, brands were built out through individual sales people which was inefficient and did not scale well. Once television became a popular medium, building a brand became a function of the ability to spend allowing players with an initial advantage to become large.

Moreover, the size of these branding investments is not determined by some technology like in production. The size of investment is determined by management making them “endogenous sunk costs”. In addition to brand advertising, Sutton identified research and development as another potential endogenous sunk cost.

This is significant as both research and development and advertising are two forms of sunk cost intangible investment. These costs represent “industry table stakes” and raise the MES, discouraging smaller firms from taking on established firms. Sutton’s theory argues in the future industries will be more concentrated given the rise of intangible investments. The level of concentration will depend on some mix of how much of the industry cost structure is sunk and the ultimate diversity of consumer tastes, spawning opportunities for niche differentiation. We are already seeing industries becoming more concentrated given the success of big tech and some of that has to do with sunk costs.

Due to their sunken nature, intangible investments are hard to value and accounting standards do not help in reflecting that value.

it’s really easy to reason about the value of tangible capital like that espresso machine. It’s really hard to reason about the value of intangible capital, like Stripe Radar

John Collison, Invest Like The Best Podcast

As explained earlier, intangible assets that arise from sunk costs are hard to sell and finance, making them hard to value. They are also highly specialized and can be complementary to other assets making them of limited value outside of the company who employs them. In addition to all of this, the value of intangible assets is really uncertain until a lot of money has been spent. Tying these ideas together, the value of intangibles tends to be binary: either the intangible asset is very valuable or its worthless. This may be exaggeration but in certain cases its true.

A good way analogy is that costs such as research and development as well as branding and advertising are subject to economic tournaments with uncertain outcomes (Hobart, 2020). These sunk costs can build value and create intangible assets if nurtured properly, but you need to “win”. Intangible investments that don’t work out are likely worthless and ones that do become valuable cannot or won’t be sold.

Consider a company that builds a factory: it invests money to build and develop it. These costs act as gravity for the factory’s value. Sure, sometimes a company will take a loss to sell an asset, but most reasonable people would agree the cost to build and develop the factory is a basis upon which to asses value. Moreover, the value of this asset is capitalized and depreciated on the company’s balance sheet. An investor who reviews the balance sheet sees there is a factory and likely assumes its an asset to the business.

Now we come to the problem. Accounting standards do not always reflect value creation properly especially in the case of intangible assets. To understand value creation, you have to understand both the size of the investment and the returns on those investments.

Firms that make tangible investments have them show up on the balance sheet either as assets from capital expenditures or increases in working capital. As a result, the income statement usually looks good and the balance sheet may look bad (Mauboussin, 2020).

On the other hand, most intangible investments are expensed on the income statement immediately and do not show up on the balance sheet. In many cases these investments confer a benefit of more than one year. Therefore, companies that invest a lot in intangibles will have poor or even negative profits, but these investments could be creating a lot of value. The investments do not show up in the financial statements. There is a discrepancy between economic reality and accounting reality. Or, in other words, accounting depreciation and economic depreciation.

I really like how Michael Mauboussin ties this all up:

Saying this differently, two companies can have the same level of investment and return on investment but very different financial statements based on where accountants record investments. Free cash flow, the number we care about, may be the same but the path to get there is different.

The practice of expensing intangible investments immediately is conservative, but that doesn’t mean its accurate. This treatment has come under fire and has been well covered. If you are interested in learning more about this, I encourage you to read this piece by Adam Keesling here (paywall).

I submit this conservative treatment is in part due to the sunken nature of intangible investments. Better to expense this now than try to estimate its value on an ongoing basis. As discussed in Adam’s article, accountants nor management teams have much incentive to do this. The problem is investors and other constituents need to.

As the nature of investment shifts towards intangible assets, we need a way to monitor this accounting discrepancy and make adjustments to financial statements to reflect that reality. You could take the approach of capitalizing and amortizing these intangible assets just like you would a factory. Hewitt Heisman takes this approach in It’s Earnings That Count if you want one approach.

The key assumption is the underlying difference between economic and accounting depreciation. Accounting today assumes the depreciation of these investments is one year by expensing them immediately. If you want some guidance: “Literature and surveys suggest: high economic depreciation (around 33 percent per year) for software, design, marketing, and training; medium rates for R&D (around 15 percent per year); and rather longer for entertainment and artistic originals and mineral exploration” (Haskel & Westlake, 2018).

These adjustments won’t be 100% accurate, but they will be better than not considering the investment at all. Additionally, you can ask yourself more qualitative questions to force you to think about what a company is doing such as: “Where is this company investing?; “How much are they investing?”; “If successful, what is the potential value of the asset they could be building?”. Thoughtfully answering these questions along with quantitative analysis can provide a clearer economic picture of the underlying intangible assets and value. These qualitative questions may help guide you in dealing with the sunken nature of these intangible investments to assess value creation.


After reading this, I hope to have grounded you in the concept of a sunk costs. In addition, I hope to have highlighted how sunk costs tie into many strategic situations and its connection with intangible investment.

Please leave any comments you have either expanding on this or pointing out flaws. I’ve thought about this a lot but also have tons to learn. I’d enjoy your feedback.