Essay backup: the value of economic classification systems

Statistically, my most popular essay on Medium. I have no idea who pushed it.

Understanding the intangible economy requires new economic methods. We need to get away from the standard neoclassical approach of treating a firm as a machine that takes resources as inputs and turns them into outputs. In this long essay, I cite four books that provide insights into the intangible economy. Instead of trying to imitate physics by looking for insights in equations, these economists are trying to imitate botanists or biologists or anthropologists by starting with what they find in the real world, figuring out what functions are served by observed behavior, and seeking out commonalities and differences. I call this approach “classification systems.”
Another term for a classification system would be a taxonomy. But “taxonomy” sounds too much like something that is about taxes and the economy. I wanted to avoid that potential confusion.
These four books derive their insights from observations of the complex and varied strategic challenges that business managers face in the real world. This stands in contrast to standard economic textbooks, which depict business managers as simple automatons, with nothing to do other than pick the best combination of inputs and choose the profit-maximizing level of output.
Amar Bhidé: Sorting out the business ecosystem
In The Origin and Evolution of New Businesses (2000), Amar Bhidé employed a two-factor matrix to classify new business ventures. One factor is the level of investment required to undertake the initiative. The other factor is the level of ambiguity, or irreducible uncertainty, in forecasting the outcome of the initiative.
When a semiconductor manufacturer undertakes to design a new microprocessor and to gear up for production, the investment required is high. But the firm can be fairly confident about its ability to forecast the volume of demand for the chip and its production costs. These sorts of high-investment, low-ambiguity projects are readily undertaken by large, incumbent firms, such as Intel.
When someone starts a small firm offering accounting, law, computer programming, or other professional services, both the investment and the ambiguity are low. The same holds true for the sorts of business that you find in strip malls — restaurants, hair and nail salons, martial arts studios, and so forth. This combination of low investment and low ambiguity draws individuals and families. In some cases the business owners have made the choice for non-financial reasons, as suggested by the phrase “lifestyle business.” When investment is low and ambiguity is low, profits are unspectacular. Bhidé refers to these as marginal businesses.
When ambiguity is high and investment is low, we see small teams of entrepreneurs gambling on their ideas. Think of someone composing a digital game to sell on the app store. Bhidé refers to these small, speculative ventures as promising start-ups.
When ambiguity is high and investment is high, venture capital enters the picture. Think of the investors backing Uber, which has blown through a lot of capital in order to obtain consumer acceptance but has yet to prove that its business is profitable.
Bhidé’s classification system helps to explain the variations in organizational structure and financing methods that we observe. Instead of treating every firm as solving an identical sort of problem, he offers us a picture of the business world as an ecosystem, in which there are separate niches for large incumbents, routine marginal businesses, speculative promising start-ups, and venture-backed start-ups.
Hal Varian and Carl Shapiro: Business Strategies for Digital Content
In Information Rules (1998), Hal Varian and Carl Shapiro anticipated the opportunities and challenges posed by goods and services that arrive in the form of bits rather than atoms. For producers to charge for what they create, they have to be able to exclude people from enjoying something without paying for it. But in the digital world, such exclusion is not natural.
Exclusion is relatively easy with a physical product, such as a loaf of bread. If you don’t pay the baker, you don’t get the bread. Excludability is the default characteristic for physical products.
With digital products, the default is non-excludability. Once I put this essay on the Internet, there is no natural barrier to anyone reading it. To prevent someone from reading it, I would have to set up an artificial barrier, such as a paywall. But paywalls set up a chicken-and-egg problem. Before you pay for my essay, you want to see it (to know what you are getting), but before I let you see it, I want you to pay for it.
Here is another way to think about the differences between physical goods and digital goods. When you consume a loaf of bread, it costs the baker something to make another loaf available to someone else. But when you read this essay, it costs me nothing to make the essay available to someone else.
Similarly, you, personally, impose no cost on producers when you download a song, use a software algorithm, or search a database. In short,
Information wants to be free. But creators need to get paid.
Varian and Shapiro offer a list of approaches for addressing this dilemma. Strategies include price discrimination, free samples, versioning, bundling, advertising, and creating network and lock-in effects.
With information goods, the same effort is involve whether you serve many customers or just one customer. Economists say that the fixed cost is high and the marginal cost is close to zero.
When the marginal cost is zero, then if someone is barely willing to pay for your product, you can benefit by charging a low price. But if someone is willing to pay a lot for you product, then you want to charge a high price. In other words, you want to charge different prices to different people, which means price discrimination.
Price discrimination is everywhere in the economy. When I used to teach high school economics, one of my catch-phrases was price discrimination explains everything. Why do sellers offer discount coupons? To maintain high prices for the people who are less price-sensitive and lower prices to the price-sensitive coupon-using shoppers. Why do airlines vary their ticket prices? Low prices help to fill planes with price-sensitive flyers, and high prices are targeted at flyers whose travel needs are relatively rigid, making them less sensitive to price.
With its Kindle e-books, Amazon has an opportunity for price discrimination. They can charge different prices at different points in time for the same book. They can charge different prices to consumers who opt for the “unlimited” option. In theory, they could anticipate which consumers are willing to pay more for a particular book and confront readers with different prices.
Free samples are a solution to the chicken-and-egg problem. You can read some articles for free on Medium.com or at the New York Times web site, but beyond that you have to pay.
Versioning means offering different information in different forms. For example, Spotify offers a free version of its service and a paid subscription version. The free version includes advertising that interrupts your listening, and it lacks some features of the paid version.
Bundling means combining goods together. Some firms will bundle a digital good with a physical good. For example, when you subscribe to the print version of a newspaper, you get unlimited access to the web version. Amazon Prime is a bundle that includes something physical (delivery services) and access to digital content.
Bundling is also a way to make the subscription model more appealing. You would not subscribe to a single label’s music offerings, but you might be willing to pay a monthly fee to Spotify or Pandora. In the 1990s, Microsoft was famous for bundling application software (word processing, spreadsheets, and most notoriously a web browser) with its operating system.
Advertising allows information providers, such as Google and Facebook, to earn revenue without charging users. Personally, I find Google’s advertising relatively unobtrusive, but I would rather have a subscription to Facebook than put up with its advertising.
One reason that Facebook is reluctant to use a subscription model is that its value depends on retaining as many users as possible. This is an illustration of network effects. As I said earlier, with physical goods, an additional user imposes an additional cost on the business, for which the producer expects to be paid. With network effects, an additional user provides a benefit to the business.
With network effects, the business strategy is to pay to get users in the short run, in order to be able to profit from consumers in the long run. For example, Uber has burned through investment capital to offer low prices, which is necessary in order to build a robust network of drivers and riders. Uber will be able to raise prices only after drivers develop confidence that they can find riders and riders develop confidence that they can find drivers.
Obviously, it does not do any good to subsidize users in the short run if they are going to desert you once you start charging enough to make a profit. This means you need lock-in effects in order to exploit network effects. Facebook needs to find ways to keep you locked into its service rather than joining another network. Uber faces a similar challenge.
Richard Bookstaber: Depicting the Financial Sector
In The End of Theory (2017), Richard Bookstaber depicts the financial sector using agent-based modeling. A key element in this process is developing a scheme for classifying the roles played by financial intermediaries. Three main components are maturity transformation, credit transformation, and risk transformation.
Maturity transformation means that the financial institution issues short-term liabilities while holding long-term assets. A bank deposit is a short-term liability, which can be withdrawn at any time. A commercial loan is a long-term asset, which will not come due for many months.
Credit transformation is a process by which a highly reputable firm effectively “rents” its good standing to less reputable firms. An example would be when AIG insurance wrote credit default swaps on mortgage-backed securities. Thanks to AIG’s backing, other institutions did not have to rely solely on their own credit strength to participate in mortgage security trading. Unfortunately for AIG and its counterparties, these arrangements were not robust when the mortgage securities market crashed in 2008.
Finally, there is risk transformation, which means changing the relationship between outcomes from uncertain investments and the returns to investors. For example, a mutual fund that holds a diversified portfolio of stocks can allow investors to obtain better risk-reward trade-offs than they could by simply holding one stock at a time.
The financial sector is further complicated by the way that these functions are performed by actual financial institutions. Examining the real-world financial sector, Bookstaber finds large banks/dealers, hedge funds, cash providers (including money market funds), securities lenders, and institutional investors. Each of these agents has its own set of objectives and operates using its own set of heuristics. Bookstaber writes (p. 137),
Each agent observes its environment and takes action accordingly. . .Each has a different business model, a different level of risk-taking, and different culture. Some of this will be spelled out in the governance and policies and procedures, some will be communicated to their investors. And during times of crisis, some of the heuristics are hard wired, without any ability for the agents to alter their course
If you have never studied economics, I imagine that this way of looking at actual financial institutions and the way that they behave seems only sensible. But most economists have no inclination whatsoever to even begin to study the actual workings of the financial system. To a typical economist, the financial sector is represented by a few abstract equations that pertain to interest rates. Actually trying to understand the financial instruments, trading strategies, and operating principles that have emerged in the real world is too daunting a prospect.
For example, most economic textbooks do not even include the term “repurchase agreement.” But the “repo” market is one of the most important components of the real-world financial sector, and in fact it was in this market that the most intense panic took place on Wall Street during the financial crisis. In addition to Bookstaber, see Gary Gorton and Andrew Metrick.
Jonathan Haskel and Stian Westlake: Intangible Assets
In Capitalism without Capital (2018), Jonathan Haskel and Stian Westlake highlight the gap between the market value of modern corporations and the value of their tangible assets. This ties back to the importance of the digital economy as depicted by Varian and Shapiro as well as to the emphasis on intangible value found in From Poverty to Prosperity (2009), reissued in 2011 as Invisible Wealth, which I wrote along with Nick Schulz.
There are many types of intangible assets. Patents and copyrights constitute formal intellectual property. Other forms of knowledge are protected informally by being embedded in corporate culture. Brand recognition is a an intangible asset. So are network and lock-in effects that have been captured by firms like Facebook. The talents, skills, and experience of a firms employees and managers are intangible assets. Concentration of talent in a location, which gives rise to the entertainment industry in Hollywood, the fashion industry in New York, or the computer industry in Silicon Valley, is an intangible asset. Standards that operate in society, such as a common language, are an intangible asset. Social norms that facilitate cooperation are an intangible asset. A legal and political system that promotes social cooperation and minimizes favoritism and corruption is an intangible asset.
Haskel and Westlake point out that intangible investments are characterized by four s’s. These are sunk cost, spillovers, scalability, and synergies.
If a pharmaceutical company invests in physical infrastructure, such as a factory or laboratory equipment, then if it no longer needs that physical capital it can recoup some of the cost by selling it. In contrast, if the company spends hundreds of millions of dollars on research to develop a new drug, and the drug does not work well enough to make it to market, then the entire research effort has to be written off. There is nothing that can be re-sold to another firm. All of the costs are sunk.
When I discussed Varian and Shapiro, I pointed out that ideas can be copied for free. This is what Haskel and Westlake refer to as a spillover. For the economy as a whole, spillovers are a bonus. But for an individual firm trying to profit from its ideas, spillovers are a problem. Imagine what the impact would be on drug development if a pharmaceutical company had to spend hundreds of millions to identify a new drug and prove its efficacy, but then it could not obtain a patent. Another drug company could produce the same drug, bring it to market at a low profit margin, and keep the first company from ever recovering its research costs. In short, spillovers illustrate that information wants to be free but creators need to get paid.
Scalability reflects the fact that intangible assets often are not subject to diminishing returns. If you want to manufacture more cars, then you have to build more manufacturing plants. But if you develop an app for smart phones, then as soon as you make it available you can provide it to as many customers as want to use it.
Synergies reflect the fact that ideas in combination may be much more valuable than ideas considered individually. The value of a smart phone is much greater than the value of each of its separate components.
Together, what the four s’s mean is that there is only a weak correlation between the value of intangible assets and the cost of acquiring them. Sunk costs can mean that the entire expense of building an intangible asset can be wasted. Spillovers can mean that the social benefit of ideas can be far greater than the benefit that accrues to any one individual or firm. Scalability and synergies mean that a firm’s ability to earn revenue can far exceed the book value of its investment.
All of this implies that the “neoclassical” approach to explaining the distribution of rewards is in peril. We need no longer find a close connection between the revenues from investment and the amount of capital invested. Instead, we find that some information-age companies that achieve spectacular returns by the standards of the industrial age.
Today, business competition does not consist of building bigger production facilities. It consists of trying to come up with the best strategies for capturing the value of ideas, including the value of spillovers and synergies that come from other people’s ideas. Economic textbooks continue to treat incomes as returns to “factors of production,” notably labor and capital. Meanwhile, in the real world, incomes at the highest levels are the outcome of management strategy, in mobilizing internal talent and in exploiting the opportunities to use synergy, spillovers, and scalability in the external environment.
Conclusion
Based on the methods that the foregoing authors use to gain their insights, let me offer some suggestions for economic research going forward.
Stare at the world, not at models. What models can teach is much over-rated. The complexity of the real world is important. The business strategies that have evolved in the real world can tell us a lot about the way the economy really functions. Abstracting away from those strategies dooms the economist to having limited understanding. The financial crisis powerfully illustrates this.
Work on classification systems, not hypothesis testing. The importance of intangible factors in the economy is increasing, which makes the economy increasingly opaque from a quantitative perspective. Better qualitative understanding will have increasing benefits, while quantitative methods will yield fewer insights.
Focus on the interaction of business strategies, not on resource endowments and allocations. As intangible factors increase in importance, strategy matters more and resource endowments matter less.

4 thoughts on “Essay backup: the value of economic classification systems

  1. I have congestion management as the key to price discrimination.

    The firm wants to keep a customer flow and that means a bit of congested demand. The goal of the firm is to establish a minimal flow o customers, then determine profit and loss. Absent customer flow they have no way to estimate results at all, a far worst outcome than a marginal loss.
    Finance and banks systems all operate on congestion management. Yield curves slope upward normally because of congestion being distributed.
    Projects with large ambiguity take the most advantage of price discrimination because it reduces ambiguity first, then viability can be determined second. No customer flow then no way to calculate ambiguity.
    So, yes, I still see a one dimensional force in economics, we do not like waiting in line, and determining our willingness to wait is called pricing.

  2. “One reason that Facebook is reluctant to use a subscription model is that its value depends on retaining as many users as possible. This is an illustration of network effects.”

    No, although there surely are network effect here, Facebook’s reluctance to use a subscription model cannot be taken as evidence for such effects.

    Suppose to the contrary that there are no network effects. A platform must still decide how best to monetize its content, whether by subscription fees, or advertising space, or a combination of both. Consumers dislike both fees and ad clutter; the question is which they dislike more.

    Put differently, how does the elasticity of demand (or site traffic) with respect to price compare to the elasticity with respect to advertising intensity? There may be an interior optimum in which the elasticities are equal and the platform adopts a mixed business model. Or, there can be a corner solution either way. If demand is more elastic with respect to price than to advertising, the platform will avoid subscription fees and rely solely on ad support.

    Do network effects made an ad-only outcome more likely? Not at all. The issue remains precisely the same—the key issue is consumers’ relative tolerance for price versus ad clutter. In a competitive platform environment, network effects may push toward lower “net” pricing, but that says nothing about whether the reduced net price takes the form of lower subscription fees or lower ad intensity.

  3. Classification systems are great.
    But aren’t colleges becoming against such systems?
    Like, it’s no longer functional to classify folk as men or women?

    When math “truth” is no longer considered true, trying to get other classification systems going seems utterly useless, and possibly job-destroying.

Comments are closed.