Appendix B
Networks of Small Firms

The question arises as to whether more reciprocity will be obtained in a system that has a network of firms versus a system that consists of just one firm, hierarchically organized. One could argue that the successful single firm that has brought within it most of the components needed will have a high degree of reciprocity among the divisions, departments, and work groups. But such a setup is unlikely to have multiple suppliers and each component has only one customer, so there is a loss in resiliency. On the other hand, one might say that the competition among a variety of networked firms might inhibit reciprocity. However, the literature stresses the sense of common fate and sense of community in these networks, especially if the units are in the same geographical region. In contrast to the literature on networks, the literature on the problems of organizing in a single firm stresses the lack of reciprocity. In the principal-agent literature that is so dominant today, the emphasis is on the agent “shirking” responsibilities and the necessity for authoritative leadership and surveillance to counteract this shirking. The classic statement is from Oliver Williamson (1975) and challenged by myself. (Perrow 1981; Perrow 1986) The literature on networks of small firms, on the other hand, stresses the degree of reciprocity.

The reliability of networks of small firms is more difficult to asses, since there is no convenient metric. But students of small-firm networks attest to their robustness, even in the face of attempted consolidations by large organizations. Saxenian effectively contrasts the decline of the nonnetworked group of high-technology firms around Boston’s Route 128, when federal funding declined and Japanese mass-production techniques matured, with the networks of small firms in California’s Silicon Valley, who charged forward with new innovations for new markets. (Saxenian 1996) Despite predictions of their imminent demise (Harrison 1994; Harrison 1999), dating back to the 1980s when they were first discovered and theorized, the small-firm networks of Northern Italy have survived and dominate a variety of low- and high-tech industries. In the United States, the highly networked biotech firms are prospering, for the time, despite their linkages with the huge pharmaceutical firms. (Powell, Koput, and Smith-Doerr 1996) Particular firms in small-firm networks come and go, but the employees and their skills stay, moving from one firm to another as technologies, products, and markets change.

Some economists argue that networks of small firms are inefficient because of increased transaction costs—everyone is buying or selling with everyone else—but elsewhere I have argued that transactions are not necessarily increased in small-firm networks, they are just external to each firm rather than occurring within it. Problems of self-interest with guile are reduced since such networks generate trust, and they also have lower surveillance and monitoring costs. (Perrow 1981; Perrow 1986; Perrow 1992) The favored design of large, multidivisional firms with forward and backward integration need not reduce dependencies. The dependency is merely transferred from the environment to the system, and alternative customers or suppliers are less available, reducing safety. Adaptive multiuse systems and decentralized networked systems are probably the most resilient to natural, industrial, and deliberate disasters. But our government, our economic ideology, our economists, and the economic interests of the large players do not favor such systems. Consequently, vulnerabilities are increased.

Increasing the size of an organization increases hierarchy and centralized control. In nineteenth-century America, for example, economic output expanded greatly without mass production, and without vertical integration (incorporating supply and distribution services, legal services, construction, and so on into the organization rather than purchasing it from others). Instead, a firm added more shops, or work groups, which replicated those already there. According to the standard functionalist argument, the innovation that mass production produced was to separate out the distinctive tasks in each shop (e.g., welding), merge them into specialized functions, and link them together in a production sequence, ideally into an assembly line. This also required that the goods and services heretofore purchased in the market be produced within the firm in order to ensure complete control and complete reliability of these goods and services. Vertical integration was driven by efficiency, in this literature.

I have argued elsewhere that the source of vertical integration that occurred suddenly at the end of the nineteenth century was only in small part efficiency; much of it, and the most fateful aspects of vertical integration, was a search for market control and the concentration of economic and political power. The bulk of economic enterprises could have remained decentralized and small or medium-sized without significant economic inefficiencies, and have produced substantial social efficiencies. (Perrow 2002) There were undoubted increases in profits with mass production, and these were socially acceptable because the social externalities that came with market control, while appearing quickly and alarmingly in some industries such as the railroads and petroleum and steel, were slow to appear in the rest of the economy. There was also less interdependency in the economy and society, making mass production in a few industries less consequential. Now that has changed, and scale appears to have many undesirable features, some of them unanticipated until the very end of the twentieth century.

The multidivisional form was in part an attempt at decentralization. But multidivisional firms often fail at buffering their divisions from each other because they seek coupling that is too tight—for example, imposing the same organizational structure or accounting system or personnel policies on units with diverse technologies that need diverse structures. The successful firms are loosely linked stand-alone systems that allow different technologies to have different organizational structures. There may be five or ten divisions, some centralized, some decentralized, having a variety of “cultures,” but with a common headquarters unit. Better yet, if the technologies and environments are truly diverse, they will be linked only by equity ownership. (For the classic statement on forms of linkages see Powell 1990.)

Not only do we have large vertically integrated corporations and multidivisional firms that still remains centralized, but the global economy has produced another form of concentration. It threatens the networks of small firms that were doing so well in the late twentieth century in northern Italy, northern Europe, Japan, and even parts of the United States. First, many of our consumer goods came to be produced in offshore factories, but initially these were relatively small. However, the last twenty years has seen the emergence of giant retailers such as Wal-Mart and, even above them in the commodity chain, the rise of giant transnational contractors based primarily in Hong Kong, Taiwan, South Korea, and China. They operate large factories in the less-developed world and provide a potential counterweight to the growing power of retailers. (Applebaum 2005) The disaster threat, other than to our economy, is not apparent, but the economic concentration in consumer goods could spread to products more critical to our critical infrastructure. For example, economic concentration in the management of our ports has led to the ability of a foreign nation that was once quite hospitable to terrorists, and still the nexus of the arms trade, to almost take over the principal port management system in the United States. The dangers of this tight integration and dependency on single suppliers is amply demonstrated by Barry Lynn. (Lynn 2005)