Links Are Not Free

The Nature of the Firm, Page 89-92:

The factors that limit the size of firms—and imply a second quantization treshold—have been studied extensively in a branch of the academic literature known as transaction cost theory of new institutional economics. Additionally, the factors that limit the size of the networks human form—whether firms or not—have been studied extensively by sociologists, political scientists, and economists working on social capital and social networks. Since this is an extensive literature, I will review the basics of the new institutional economics in this chapter and leave the discussion of social theories for the next chapter.

Transaction cost theory, or new institutional economics, is the branch of economics that studies the cost of transactions and the institutions that people develop to govern them. In simpler terms, it is the brach studying the cost of economic links and the ways in which people organize to deal with commercial transactions.

The origins of transaction cost theory can be traced back to a 1937 paper by Ronald Coase, "The Nature of the Firm". As a young scholar, Coase realized that the descriptions of the economy that were prevalent at the time tended to overlook one aspect of the economy that seemed obvious to him: the fact that economic transactions are costly. As a student at the London School of Economics, Coase attended a seminar by Arnold Plant, who had been recently appointed as a professor of commerce. It was there that Coase heard a description of the economy that contradicted his intuition and would accompany his thoughts throughout his life. This was a quote from Sir Arthur Salter: " The normal economic system works itself out."

Paraphrasing Einstein, we can say that Coase saw in Salter's quote a description of the economy that was not simple, but simpler. In his landmark 1937 paper Coase noted that economies involve plenty of planning that is not coordinated by the price system and takes place primarely within the boundaries of a firm. He noted that descriptions of the economy overlooked obvious aspects such as the fact that workers who relocate from one department to another within a company are responding not to the price system but to the orders of a manager, or that drafting and executing contracts often involves a lot of work. Coase noted that economic transactions were not easy, and that the economy was not as fluid as many of his colleagues liked to assume.

In Coase's view, the economy was not a collection of fluid and frictionless market transactions, but a set of islands of conscious power, shielded from each other and from the dynamics of the price mechanism, Coase emphasized, and the interactions between a firm's workers are often political. So in Coase's view, hiring a worker was a form of contract in which a person was hired to do a task that had not yet been specified, since what a worker will be asked to do a few months down the road is rarely known when she is hired. Coase dedicated much of his academic career to explaining the existence and boundaries of these islands of power. His answer became known as the transaction cost theory of the firm.

Coase's explanation of the boundaries of a firm was brilliant and simple. It was based on the idea that economic transactions are costly and not as fluid as the cheerleaders of the price mechanism religiously believed. Often, market transactions require negotiations, drafting of contracts, setting up inspections, settling disputes, and so on. These transaction costs can help us understand the boundary of the firm, since accoirding to Coase, a parsimonious way of understanding the islands of central planning that we know as firms is to search for a point at which the cost of transactions taking place internally within the firm equals the cost of market transactions. When the external transactions become less costly than the internal transactions, firms stop growing, since it is better for them to buy things from the market than to produce them internally.

The minimalist version of Coase's theory that I present here help us understand that there are fundamental forces that limit the size of the network we know as firms, and hence that there is a limit to the knowledge and knowhow these networks can accunmulate. Moreover, it also tells us that there is a fundamental relationship between the cost of the links and the size of these networks: the cheaper the link, the larger the network.

More importantly, the limited size of firms tells us that volumes of knowledge that are larger than the networks we can form need to be shared across networks of firms. Hence, Coase helps us explain why plants like River Rouge have not taken over the world.

An example of a produvt that is produced by a network of firms, rather than a single company, is the personal computer. While personal computers tend to have an identifiable brand, different firms design and manufacture different parts of a finished computer. Even Apple's devices, which are proudly designed in California, contain parts—such as their displays—that are designed and manufactured by others, including Apple's nemesis, Samsung. In fact, soon after Steve Jobs returned to Apple he begun to outsource the manufacturing of devices, relying heavily on technology from other firms. The iPod was made possible by a small hard drive that was invented by Toshiba. The Gorilla Glass screen of the iPhone was the brainchild of Corning, a glass manufacturer in upstate New York. What is true for Apple products is also true for many other modern devices. In fact, no matter what brand your computer is, it is probably a salad of electronics: powered by a chip from Intel or AMD, a hard drive made by Quantum, Samsung, Seagate, or Fujitsu; a memory made by Kingston, Corsair or PNY; and a network card made by D-Link, P-Link, or Netgear. All of these brands and companies are likely to be different from the one that slapped a logo on your machine, showing that computers are constructed by networks of firms rather than firms.

Yet the network of firms involved in a computer is much larger than we have described so far. While some computers carry the same brand as their operating system (namely, Macs), the application software that makes computers fun and useful comes from a variety of other firms, from large companies such as Adobe or small indie game studios such as the ones that create cool games like Blek, Machinarium, or the World of Goo. Finally, many use computers to access the Internet, so at the end of the day this salad of software and hardware becomes a simple prerequisite for visiting social networking such as Facebook or Twitter, reading articles in the New York Times, or playing massively multiplayer online games such as World of Warcraft.

Oliver Williamson's classification of transactions, Page 93-95:

Coase's intuition tells us that the ability of networks of firms to hold knowledge and kbowhow will depend on the cost of links. That is, when making and sustaining links is inexpensive, creating large networks of firms will be easier, and accumulating vast volumes of knowledge and knowhow will be easier, too. When links are expensive, on the other hand, it will be harder to connect firms, and so it will be harder to create the network of firms and people needed to accumulate vast volumes of knowledge and knowhow. In short. when links are costly, our world becomes fragmented. SO to answer the question of whether networks of firms facilitate or hinder the accumulation of large volumes of knowledge and knowhow, we need to learn more about the cost of firm-to-firm links.

The problem we face in this line of argument is that there are many ways in which two firms can interact. Hence talking about firm interaction in general is an oversimplification. Some firm-to-firm interactions are simple, such as ordering ink cartridges from a catalog, while others are incredibly complex, such as developinga partnership for the construction of a new manufacturing plant. Moreover, many firm interactions are embedded in social networks, which is a fact we will consider in the next chapter. So talking about the cost of links is not simple, and it makes sense only when we define links narrowly enough.

Oliver Williamson, a student of Ronald Coase, understood that commercial links come in different sizes. He wrote extensively about the connection between the cost of firm-to-firm interactions and the institutions people develop to manage these links.

Williamson's classification of links is based on two axes. On the first, he separated transactions by frequency, into recurrent and occasional. On the second, he separated transactions by specificity, from nonspecific to idiosyncratic.

To understand Williamson's parsing of the world, think about the amount of paperwork and people needed to establish a commercial link.For example, think of buying a latte at your local coffee shop. This rquires very little paperwork—a receipt—and just a few seconds from the cashier to the barista. No intermediaries are required to get the task done. In Williamson's language, buying coffee is a nonspecific recurrent transaction. Now consider buying a house. This is a transaction that requires much more paperwork. A house is a relative specific buy, so interactions will likely be catalyzed by institutions that are external to the buyer and the seller, such as the bank is issuing the mortgage, the home inspector, the real estate agents, and the real estate lawyers. In Williamson's language, the purchase of a home is an occational and specific transaction, meaning that it is an interaction that needs to be chaperoned by additional institutions. Finally, consider a long-term but highly specific deal. For instance, a garment manufacturer wishes to establish a long-term collaboration with a button manufacturer that produces very specific buttons—maybe tiny pieapple-shaped buttons in gold and silver. In Williamson's language, this is a specific and recurrent interaction. In such interactions, developing a relationship with the supplier is more important than the participation of external institutions.

Williamson used his classification scheme to connect economic transactions with the governance structures best suited for their management. Yet in our case we are interested in classifying transactions by type, not because we want to explain the institutions involved but because we want to know the costs of the links as these affect the formation of the networks we need to accumulate knowledge and knowhow.

The development of standards, Page 95-100:

Following Coase's intuition, we know that the cheaper links give rise to larger networks, and the personbyte theory tells us that we need larger networks to accumulate more knowledge and knowhow. Now let's add Williamson's intuition to the mix.

First, let's consider the simplest links, market interactions. These are the links that Williamson call unspecific, such as buying a coffee, a spatula, a light bulb, or a single sheet of Plexiglas. During the last decades the cost of market transactions has fallen due to, among other things, changes in transportation and communication technologies, so we should expect the networks composed of markets to be more fluid and dense. For instance, the inflation-adjusted cost of moving goods fell by 90 percent during the twentieth century, while the change in long-distance communications in the three last centuries took us from a world in which telegraphs were mechanical contraptions used by the french elite to pass on messages to one in which videoconferencing became a teenager's pastime. With falling costs the number of long-distance market links can be expected to increase, and in turn, this should increase our ability to accumulate knowledge in networks of market interactions.

The cost of interactions was reduced directly by reductions in tariffs and by improvements in shipping and communications, but other forces, such as the emergence of standards, have also reduced the cost of market interactions. Examples of standards in the computer industry include the VGA port, Wi-Fi, and the USB port. These standards allow manufacturers to create products that connect seamlessly without need for any coordination between manufacturers. In fact, the USB port was invented in collaboration between Intel, Compaq, DEC, IBM, Microsoft, NEC, and Nortel and was made available through an extremely inexpensive license, because these manufcturers knew that the ecosystem benefits of a standard were larger than private gains they could harvest from an expensive license that would restrict access to the technology and generate a platform war.

Standards are prevalent in our modern world because they reduce the costsof interactions among the firms and people that subscribe to them. Hence, it is not unexpected to see standards coevolve with markets. Many people are surprised to learn that only a few centuries ago simple measures of mesures of weight and volume, such as the pound and the pint, were not standard. Even though the same word was used in different towns, the weight of a pound varied from town to town—sometimes as much asa factor of four. But as cities began to trade with one anotherand governments began to impose rules over larger areas, the use of standards grew. The coevolution of standards and markets is easy to understand, since anyone buying a bushel of corn from a vendor in another town would want the bushel to mean the same in both towns. So the possibility of trade created an incentive for standardization, and helped the expansion of governments that were keen on the use of standards.

Another example of an ancient standard that helps reduce the costs of interactions is language. Language allows people to weave networks by empowering them with the ability to communicate complex ideas, coordinate their actions, and establish commersial links. Language is the quintessential standard.

Consequences of difficult links, Page 102-105:

Yert, as Coase and Williamson noted, not all interactions are this easy. Now it is time to talk about the consequences of the linka that are difficult to make. These are the links involving long-term collaboration or large projects—links that often require a large amount of paperwork and people's time to be established.

Consider two large organizations wanting to establish a research collaboration. The interaction might be ignited by the interest of two researchers who fell in love with each other's work and would like to explore the fertility of their intellectual love. The "first date", however, would not involve a cup of coffee or dinner and a movie, but signing a nondisclosure agreement. This will be followed up by the creation of a contract that will need to be approved by department heads and the legal departments of both organizations, a process that often takes months. Most likely the legal departments of each organization will have conflicting policies for intellectual property—each wants all the intellectual property for its own organization, and none for the other—and this will delay the process even further. As a result, the intellectual romance may be lost in the negotiation of an interaction "prenup" that is more concewrned about the potential divorce than about the fertility of the interaction. Using our shortcut to Williamson's theory, the large number of people and huge amount of paperwork involved in this interaction tell us that it is extremely costly to make, and hence we should not expect networks that involve such highly bureaucatized interactions to hold too much knowledge and knowhow.

If you have worked in large organizations, whether in the private sector, public sector, NGO sector, or academia, you know exactly what I am talking about and will probably agree that unnecessaily bureaucratic interactions are pervasive. The simplest interactions between organizations are hardly as fluid as picking an apple from an orchard or choosing a new printer from a catalog. In the most extreme cases, examplified by the United nations or many public sector organizations, external interactions are constrained by long administrative processes and require an army of administrators, extensive paperwork, and a long sequence of approvals. This means that interactions that should be simply market transactions—such as simple forms of service procurement—wind up subject to a regime of governance that makes them far more complicated than they ought to be. As a result, the bureaucratic burden involved in establishing links is comparable to the effort required to execute the task being contracted, and the organizations effectively exclude themselves from the market they are trying to participate in. A symptom of this is the subpar provision of services in countries that have plenty of capacity; a good example is the conspicuous difference in quality and cost between the webpages of the US government and those created in Silicon Valley.

Extreme levels of inefficience can only be supported by organizations whose revenue stream does not depend on their interactions with others, for if it did, they would have gone broke. Chief examples are organizations whose revenue comes from the collection of taxes, such as governments, or organizations that receive funds in a more or less unconditional way, such as the United Nations. More important, these costs affect not only the number of links but also whom these links connect to. Lengthy and cumbersome bureaucratic requirements provide an advantage to incumbents familiar with the paperwork and the people involved in their approval (known informally as "beltway bandits" in the context of the US Federal government). The people who are going through the paperwork, however, may not necessarily be the best to provide the service requested.

Extremw bureaucracy generates large networks containing many people but few personbytes. Most of the personbytes available are consumed by internal procedures, uch as lengthy and politicized approval processes. As a result, such networks are weighed down by their own links, and although large, they become unable to generate or accumulate much knowhow and knowledge.

We can illustrate the costs of overbureaucratized networks by looking at a few cases, such as the healt care sector in the United States. A recent survey of private US health care facilities estimated that the support staff of hospital physicians spends nineteen hours a week interacting with insurance providers in prior authorizations, while clerical staff spend thirty-six hours a week filing claims. The cost of interactions bewteen private health care providers and private insurance providers was estimated to be $68000 per physician per year, totaling a whopping $31 billion per year—equivalent to the GDP of the Dominican Republic in 2005. The interaction costs in 1999 for the entire health care system, including private and public, were estimated on the low end to be $31 billion and on the high end to be $294 billion—which is comparable to the present day GDP of Singapore or Chile. Moreover, between 1969 and 1999 the administrative cost of health care in the United States grew from 18.2 percent to 27.3 percent of the health labor force. So the costs of interaction between the health care and insurance sectors are not only high but rising.

The goal here is not to open up a discussion on the administrative costs of the US health care system but simply to highlight that interactions between different firms or organizations can be substantial. Ultimately, in a world in which differences in the costs of links translate into differences in connectivity of the networks of firms and organizations that beget them, the ability of the networks to hold vast volumes of knowledge and knowhow will be the ones that suffer.

* * *

We started this chapter by looking at River Rouge as a quintessential example of a personbyte cathedral. The Rouge amassed large volumes of knowledge and knowhow and produced a large number of vehicles, but it did not become the dominant model of production for the twentieth century. As the twentieth century continued to roar, networks of firms became increasingly the new norm. In the case of manufacturing, the ability of firms to interact was helped by reductions in the costs of tariffs, transportation, and communication, but also by the emergence of industry standards and the coordination enabled by hub languages. The reduction in the cost of commercial links allowed the dismemberment of manufacturing into networks of firms, as illustrated by the Barbie doll example. But in other cases, such as computers, this was a condition that was also required by the limited capacity of firms to hold knowledge and knowhow. That is, the complexity of computers, which encompasses not just the hardware but software and online services as well, requires the existence of networks of firms.

The complexity and finesse of modern manufactures are evidence of the increase in the ability of manufacturing networks to amass increasingly mlarge volumes of knowledge and knowhow. Only a few centuries ago, many products were strongly associated with a few regions of the world: Champagne with France, clocks with Switzerland, Parmesan with Parma. Even though legal restrictions on the names can enforce the continuation of such associations, the point is that these associations are no longer true for many modern products. Are iPhones Californian, Chinese, or Korean? With the dismemberment of production, the nationality of products no longer makes much sense.

Not all links are like, Page 105-108:

In addition to large volumes of knowledge and knowhow, modern manufacturing is the result of large international networks that are made of links involving the exchange of intermediate products, such as smartphone displays or the plastic pellets used to make a Barbie doll. Yet as manufacturing grew, other networks began to fold under their own weight. The cost of interactions in some sectors with large administrative burdens, such as education and health care, is high and increasing, and it is hard to arguethat the increase in administrative complexity we are observing will translate into an increase in the ability to embody large volumes of knowledge and knowhow. In the case of helath care, the inverse relationship between the administrative burden and the quality of care has been documented, suggesting an excess of management and administration is not helping these networks improve their ability to put knowledge and knowhow to good use.

But does this mean we should attempt to make all links market links? If Coase and Williamson are right, and they are right about a few things, the diversity we observe in economic links are not a matter of choice, but a fundamental characteristic of the nature of economies. The market links we use to buy toothpaste are fundamentally different from the ones involved in long-term contracts, which, as Coase highlighted, often involve the purchase of something that is not specified at the time of the original contract. So it would be naive to think that it is possible to make all links like those we use to transact well-defined crystals of imagination, such as wineglasses and paper towels. As we will see later, adding the social dimension to these interactions will illuminate why a diversity of links is needed to form the networks that accumulate knowledge and knowhow.

Fortunately, we only need to take a few lessons from this long discussion. After all, we want to focus not on transaction costs per se but on the ability of the economy to generate structures that can hold the knowledge and knowhow required to crystallize imagination. With this goal in mind, one lesson to remember is that there is a relationship between a productive network and the volume of knowledge and knowhow it can embody, with larger networks being able to embody larger volumes of knowledge and knowhow, all else being equal. Second, our ability to weave large networks depends on the cost of establishing links, with cheaper links favoring the creation of the large networks needed to amass large volumes of knowledge and knowhow.

The third lesson is that there are fundamental breaks, or transition points, in the structures of the networks that we use to accumulate knowledge and knowhow at the collective level. To amass large volumes of knowledge and knowhow we need to quantize both knowledge and knowhow, and two fundamental quanta of knowledge and knowhow were introduced for that purpose: the personbyte, which emerge from the finite capacity of humans, and the firmbyte which emerge from the finite capacity of firms.Certainly neither the personbyte nor the firmbyte should be taken in their strict numeric sense. Both limits represent conceptual boundariesthat point to important transctions in the structure of the networks needed to accumulate knowledge and knowhow, not narrowly defined tresholds. Both personbytes and firmbytes show that our ability to accumulate large volumes of knowledge and knowhow is packaged in a nested structure in which what we consider to be a network at one scale becomes a node in the next. Networks of neurons become nodes when we abstract them as people, and networks of people become nodes when we abstract them as networks of firms.

The bottom line is that accumulating large volumes of knowledge and knowhow is difficult because itvrequires evolving the networks that embody that knowledge and knowhow. We can think of knowledge and knowhow as continous, but since the networks that hold them are not continous, knowledge and knowhow must be quantized, and not just in theory but also in practice.

So the quantization of knowledge and knowhow, which is brought about in part by the cost of links, helps us answer the question of why it is difficult to accumulate increasing volumes of knowledge and knowhow. The answer is that accumulating knowledge and knowhow is difficult because creating the networks required to embody both knowledge and knowhow is difficult. But there's an important caveat. So far our explanation of the cost of links has leaned too strongly on the economic arguments such as those advanced by Coase and Williamson. There are also important sociological and cultural processes that effect the structure of social networks. So in the next chapter I will review other strands of literature connecting the structure of social networks with economic outcomes.