2. International courier and small package service in Japan
The primary domain of our study is international courier and small package service in Japan. Twenty-seven member firms (see Table 1) are registered in the International Courier and Small Package Service Subgroup in the Japanese Air Cargo Forwarders Association (JAFA). Of these, 24 firms are actively competing in the Japanese market, which covers virtually all firms offering IC&SP service in Japan.

Generally, IC&SP service is comprised of five separate activities, which include three separate transportation segments (see Figure 1). A shipper contacts a freight forwarder whose responsibility is to coordinate a domestic truck picking up the package from the shipper and transporting it to an airport. The freight forwarder advances the freight through domestic customs and consolidates it for air transit. An international air carrier transports the consolidated package to a foreign airport. A foreign freight forwarder advances the freight through customs, separates the packages and coordinates delivery. Finally, a foreign truck delivers the package.
Empirically, we find that domestic and foreign freight forwarding activities in most cases are jointly owned or organized through some type of equity relations. Thus, we focus on the freight forwarder's organizational choice for the three different transportation segments-domestic trucking, international air; and foreign trucking-for which we observe substantial heterogeneity of organizational configurations. We investigate why freight forwarders vertically integrate into one or more of these transportation segments.
Although the phrase "integrated carrier" often is used by most industry participants, not all segments of the "integrated" firms are integrated. Integration can occur in any of three transportation segments. For instance, a variety of integration patterns are found for domestic trucking operations. Among 16 firms interviewed, only one firm, Fukuyama Transporting Co. Ltd., fully internalizes all pickup/delivery trucking operations. More typically, freight
forwarders vertically integrate some domestic trucking routes while contracting out for other trucking routes.
Similarly, only a few firms are integrated into international air transport out of Japan. While Federal Express, DHL, and UPS-Yamato can be classified as integrated , all other freight forwarders contract for international air carriage. Many of the firms such as Airborne Express, BAX Global, Emery Air Freight, and TNT Express Worldwide, do integrate into international air transport between some foreign cities or in the United States, but have not integrated into international air transport between Japan and foreign cities.
It is this multinational and multi-transportation segment characteristic of IC&SP service that yields the greatest organizational variety. Among the 24 firms Japanese competitors, 21 firms have at least one foreign freight forwarding and trucking operation subsidiary or parent company. For instance, Hankyu Express International, Kintetsu World Express, Nippon Express, Nishi-Nippon Railroad, Nissin, Overseas Courier Services, Proco Air Service, Seino Transportation, and Yusen Sea and Air Service all own foreign subsidiaries. Alternatively, firms like Maruzen Air Express International and World Courier are subsidiaries of foreign firms. Many of these firms have equity relations with freight forwarding and trucking operations in some countries but not other countries.

3. Prior literature
　To theoretically investigate the choice organizational in the Japanese IC&SP industry, we adopt the governance lens of transaction cost economics (Williamson 1975, 1985, 1996).
TCE argues that bounded rationality and opportunism make contracts inherently incomplete. While such incompleteness is of little consequence in a spot market where switching to an alternative trading partner is low cost, transactions characterized by specific assets, uncertainty, and frequency cause contract incompleteness to have important ramifications for organizing a transaction. Hierarchy, the organizational mode of last resort, offers transaction cost economizing advantages over markets when the transaction is characterized by specific assets, uncertainty, and frequency-under these conditions fiat is more advantageous than markets in reducing ex post exchange and maladaptation problems.
　This paper is not the first application of TCE to logistics industries. Palay (1984) explored determinants for market v. non-market governance mode choice in rail freight contracts. Specifically, he gathered information on formal contracts as well as informal agreements between rail carriers and their shippers, and found that a more "hierarchical" governance structure was chosen the greater the amount of transaction-specific capital in the exchange.
　Similarly, Pirrong (1993) observed that increases in "temporally" specific capital shifts bulk shipping contracts from the short-term duration found in spot markets toward long-term contracts. Unlike previous studies, he showed that even when capital is not sunk, i.e., not specific to a particular transaction in the long-run, the capital can still be a specific asset in the short-run. For example, the market for a shipping contract becomes "thin" for a ship visiting a remote harbor. As a result, spot contracts can be prohibitively costly because of the haggling between the shipper and the sea carrier.
　From a related perspective, Nickerson and Silverman (1997) argued that for-hire trucking carriers vertically integrate by owning the truck and hiring a company driver when the optimal tractor configuration for a particular haul incurs severe operating penalties should it be redeployed for non-optimal uses (e.g., independents own tractors with average configurations while carriers own tractors with configurations optimized for short-haul and light-weight freight or for long-haul and heavy freight). Also, Nickerson and Silverman (1997) argued that hub-and-spoke logistics (less-than-truckload) create coordination problems, which are reduced by employing company drivers instead of independents. Integration reduces the incentives independents have to deliver freight unreliably, which, because a small delay in the arrival of one truck can cause system-wide delays, could impose large operating costs and could damage carrier reputation. A large-sample empirical analysis of U.S. interstate for-hire trucking carriers provided support for their theory.
　Generally, TCE studies of logistics show that the contract hazards associated with specific physical assets, unavailability of timely shipping capacity, and coordination problems associated with hub-and-spoke logistics networks lead to vertical integration in logistics industries. While all of these contracting problems are likely to have ramifications for organization of IC&SP service, we focus, in part, on a different set of specific investments.

4. Hypothesis
　Our primary investigation focuses on the relationship between investments in information networks and organizational choice. Information technology has had a dramatic organizational impact on logistics and on IC&SP service in particular. Information technology is critical for tracking freight, trucks, and airplanes and providing real-time information to carriers and customers. Such information is valuable to a wide variety of customers that need to know the precise location of their packages while in transit. Moreover, information technology is indispensable for the fastest international courier service-collecting parcels from all over the world, sorting in a few hours, and delivering them on the next day depends critically on accurate and real-time information about each package. In addition to internal needs for information, carriers choosing to provide such information to customers need to collect and transmit data from each transportation segment: domestic trucking, international air, and foreign trucking.
　Presently, information technology and the corresponding computer network in most instances are unique to each firm and represent specific investments in both hardware and software. Several firms invest in unique portable electronic devices to record and code data about packages on both pick-up and delivery. UPS, for example, provides its 125,000 drivers with firm-specific hand-held computer systems to capture and transmit signatures and package data. Additionally, general purpose computers become specific to the firm once installed because of rapid price declines in computing equipment, which makes selling used equipment much less valuable than their continued use. More importantly, software is either partially or completely customized for each firm and for each piece of information a carrier tracks. Specific investments in the information network must be made in each transportation segment the freight forwarder wants to collect information from. These investments have little value if redeployed in an alternative, second-best use. For example, software written for UPS's information system is unlikely to work on Federal Express' information system without substantial and costly modification.
　Moreover, information technology investments can be substantial. UPS and Federal Express together employ 4,000 information technology staff and spent approximately $2.5 bn in 1995 and 1996 to maintain and enhance their information networks. UPS alone is approaching annual expenditures of $1.5 bn on information technology (Bicknell 1996). Our interviews disclosed that other Japanese freight forwarders actively invest in proprietary information networks, though the investment levels vary in terms of percentage of total cost (0.3% to 5%).
　Large and firm-specific investments in information technology give rise to contracting hazards (i.e., hold-up and maladaptation hazards). Firm-specific investments by one trading partner could be exploited by the other partner in ex post negotiations or in an opportunistic response to changing conditions, or could impose substantial costs should the partners fail to adapt in a coordinated way in response to changing circumstances (Williamson 1985). In IC&SP service, information technology investments could be made in each transportation segment: domestic trucking, international air, and foreign trucking. Transaction cost economics maintains that hierarchy, although a costly form of organization, offers efficiency advantages of markets for minimizing these contracting hazards. Conflicts can be resolved by fiat instead of appealing to the courts. Therefore, even if courier service requires few idiosyncratic physical assets such as containers, trucks, or airplanes, the information network is transaction-specific, which, according to TCE, calls for vertical integration to safeguard against opportunistic actions that could expropriate or devalue specific investments. Thus, we predict that vertical integration in each transportation segment is a response to specific investments in information technology in each segment:

Hypothesis: The greater the investment in information technology in any particular segment of IC&SP transportation the greater the likelihood of integration between freight forwarding and that segment of IC&SP transportation.