Accounting for Change

Leslie Robinson and Phillip Stocken’s creative use of closely-held accounting data shows a long-term trend to more autonomy for U.S. overseas subsidiaries.

Globalization has been among the most salient business trends of the current generation, but researchers have had a difficult time measuring parent-subsidiary management relationships. Now, using data maintained by the Commerce Department’s Bureau of Economic Analysis, Tuck professors Leslie Robinson and Phillip Stocken have discovered evidence that helps sort out conflicting expectations of managerial centralization versus autonomy for foreign subsidiaries of U.S. firms.

Organizational theorists would generally expect more decentralization because of increased business complexity. On the other hand, “greater capability to coordinate activities around the world through information technology and the growing prevalence of English might lead us to anticipate a trend toward increased centralization,” says Stocken. “But that’s not consistent with the data.”

The key to solving this conundrum is FAS-52, a provision of the Financial Accounting Standards that requires each U.S.-based multinational to report its performance and the performance of its foreign subsidiaries to shareholders in a common currency. In preparing these audited reports, companies are required to distinguish between foreign subsidiaries that are integrated with their domestic operations and those that are self-contained. The result is what previous researchers in this area have lacked: a very large and uniform set of data.¨

By analyzing filings from 4,451 U.S.-based multinational firms and 49,070 affiliates, the researchers created a measure of the extent of integration that exists among parent-subsidiary relationships at U.S. multinationals. The method provides an elegant solution to the most problematic limitations of previous research in this area: small sample sizes and a reliance on imprecise measures such as surveys, email analysis, and conference-call rosters. Applying their own equation to this large dataset, Robinson and Stocken are able to objectively show that in 1982, 30 percent of U.S. parent companies considered all of their foreign affiliates to be decentralized; in 2004 that figure had grown to 50 percent.

The reason for increased autonomy seems to be performance. Affiliates whose organizational structures match the operational needs of their business environment tend to be more profitable. Robinson suspects that “autonomous subsidiaries are better able to compete in increasingly competitive and complex markets by being more sensitive and responsive to their local environments.”

“Different circumstances call for different organizational structures, and if you can match the organizational structure to the environment in which you operate, you should be a more profitable enterprise,” Stocken says. “While that makes sense on the theoretical level, showing that empirically is not easy.”

The data also quantified the continuing trend toward an increasingly global economy. When the FAS-52 standard came into effect in 1982, the median U.S.-based multinational firm generated 15 percent of total sales abroad. By 2004 that figure had nearly doubled, to 28 percent. While a trend toward decentralization is clear, the research shows that most companies avoid change when business is good. The likelihood that a firm will move to correct an organizational mismatch increases when firms perform poorly. “We believe that our measure also opens the door to questions about how information is shared and communicated within a firm,” says Robinson, “a start to research that seeks to understand how information flow within the firm is affected by globalization.”

While Robinson and Stocken limited their research to U.S.-based firms included in the BEA data, their method could be applied to foreign-based multinationals that report under similar accounting standards, assuming governments make those records available. But gaining access to data with sufficient detail and at a sufficiently disaggregated level may ultimately prove to be the highest hurdle for researchers. Because of the sensitivity of much BEA data, Robinson was allowed to view it only in person in Washington, D.C., after being granted access through the BEA’s research-proposal process.

Robinson L A and Stocken P C, “Organizational Structure and Performance of Multinational Firms,” forthcoming