Determinants of EDI Adoption and Integration by U.S. and Japanese Automobile Suppliers

Abstract

This article examines determinants of Electronic Data Interchange (EDI) adoption and integration in 76 U.S. and 93 Japanese automobile suppliers. The article constructs several hypotheses based on the transaction-cost and resource-dependence approaches and tests these hypotheses by data from these suppliers. Our study showed: (1) U.S. firms were more EDI-integrated with their customers, while Japanese firms adopted more EDI with their suppliers; (2) the resource-dependence approach seemed more effective in explaining EDI adoption, while the transaction-cost approach seemed more effective in explaining EDI integration; (3) the transaction-cost approach seemed more suited to the U.S. context, while the resource-dependence approach seemed more suited to the Japanese context; (4) EDI adoption had a positive impact on EDI performance in the U.S., suggesting the higher validity of our framework in the U.S.

Keywords:

• automobile industry
• automobile suppliers
• business-to-business electronic commerce
• Electronic Data Interchange (EDI)
• technology adoption

Notes

¹We define a first-tier supplier as a company that supplies its products directly to Original Equipment Manufacturers (OEMs; i.e., final assemblers), while a second-tier supplier is a company that supplies its products directly to first-tier suppliers and does not supply its products directly to OEMs.

²Furthermore, it was rumored that suppliers who cannot comply with EDI may lose business with the Big Three [64].

³These are based on our interviews with 20 IT managers in the automobile industry in the U.S. and Japan.

⁴Based on interviews with four managers at Toyota (April 2001) and e-mail communications with these managers thereafter.

⁵We will use “buyers” mainly as first-tier suppliers and “sellers” mainly as second-tier suppliers, henceforth.

⁶Reekers and Smithson [22] claim that “network” approach is also useful for explaining such buyer-seller relationships.

⁷Hart and Saunders [30] also agree with our argument.

⁸Some studies, however, have employed broader definitions of EDI Integration. For example, the Iacovou et al. [2] measured EDI Integration internally and externally. Namely, “internal integration” refers to the variety of applications interconnected through EDI, while “external integration” refers to the number of trading partners with which the firm can transact business through EDI.

⁹Here, we argue that EDI systems are generally very specific to specific buyers. According to our preliminary interviews with IT managers at Denso (a Japanese first-tier supplier), suppliers face the “translation hell,” because final assemblers and first-tier suppliers are likely to force their suppliers to use their proprietary EDI systems (April 2001).

¹¹However, there is a possibility for the few suppliers to behave opportunistically. The few suppliers also need to be assured that the first-tier supplier will commit to their business relation in the long-run [65].

¹⁰O'Callagham, et al. [46] found a reverse causal relationship—i.e., a seller's EDI adoption is likely to increase the seller's dependence on a buyer.

¹²The number of competitors can be considered as a horizontal competitive force from Porter's perspective. However, in constructing Hypothesis 5, we emphasized more buyers' “diluted” power for EDI promotion than Porter's competitive force.

¹³To our knowledge, no study has examined this point empirically.

¹⁴There are also merits stemming from the seamless integration—i.e., “scale economy” and “network externalities.” “Network externalities” can be defined as higher utility due to use with a larger number of other related firms (see Shapiro and Varian [66]).

¹⁵There are also merits stemming from “scale economy” and “network externalities.”

¹⁶We used Herfindahl-Hirschman Index (HHI) to measure a level of concentration. HHI is frequently used to measure a level of concentration in an industry by using firms' market shares, especially in the IO (Industrial Organization) literature (Scherer 1980: 56–59). However, the index has also been used as a mathematical calculation for concentration—e.g., import/export concentration (Ludema and Mayda, 2005) or journal concentration (Swanson, Wolfe and Zardkoohi, 2006). In this article, we used HHI, not to measure concentration in a market or industry, but to measure “a firm-level perceived pressure” from suppliers' or buyers' forces.

¹⁷Mukhopadhyay, et al. [40] claimed that EDI can reduce operational costs, $62.26 per car, by introducing EDI in the automobile suppliers. However, it is generally difficult to measure the performance of EDI in such financial terms.

¹⁸O'Callagham et al. [46] and Ring et al. [29] argued that “EDI promoters” can gain more operational benefits from EDI than “EDI adopters”—i.e., those adopting EDI by the pressures from the EDI promoters—can.

¹⁹ Application advice (824) is used to notify trading partners that there are changes in previous orders. Customers transmit planning schedules (830) to suppliers to inform them of anticipated component requirements. Advance shipping notices (856) let customers know the arrival time and the quantity of products to be delivered to the customer site. Receiving advice (861) is used to inform suppliers of discrepancies between an advance shipping notice and the actual shipment received. Shipping schedules (862) are used to assist trading partners in planning and executing their shipments. A functional acknowledgment (997) is commonly used in the U.S. to confirm the information received. According to our interviews with Toyota's IT managers (April 2001), however, it is not well used in Japan, simply because of its redundancy in the Japanese business practices. Hart and Saunders [30], and Mukhopadhyay et al. [40] also used such measurements.

²⁰We also check the original data. Both U.S. and Japanese data generally indicated “a hierarchical pattern”—i.e., those who use (2) planning schedules tended to use (1) application advice, as well, those who use (3) advance shipping notices tended to use (2) planning schedules, and (1) application advice, as well, and so on.

²¹We also measured “annual sales” in the previous year. Because there was a high correlation between the two indices—i.e., 0.708 at the 1% level—we used only number of employees.

²²Reflective scales are used when the observed variables are manifestations of the underlying constructs, while formative scales are exogenously determined, thus cannot be explained by correlations among indicators [65].

²³Endogenous variables are factors in a causal model whose value is dependent on the states of other variables in the system, while exogenous variables are factors whose value is independent from the states of other variables in the system [74]. For example, the transaction cost was measured by the three formative scales—i.e., asset specificity, uncertainty, and frequency—because (1) the transaction cost is conceived as explanatory combinations of the three scales; and (2) the three scales cover different facets of the construct [73], [67].

²⁴Accordingly, in the following regression analyses, we will analyze factors affecting EDI Adoption by using the total samples in the U.S. and Japan, and analyze factors affecting EDI Integration by using only those who adopted EDI with both their sellers and buyers. Table 2a, 2b, and 2c include those suppliers who adopted EDI with their sellers and buyers, and those who adopted EDI only with their buyers.

²⁵Although not reported here because of the limited space of this article, we conducted correlation analyses by using those firms that adopted EDI with both sellers and buyers in the two countries. The results indicated that correlations among the defined independent variables and control variables were not very high, suggesting that multicollinearity is not a serious concern in the following multiple regression analyses.

²⁶It should be noted here that Tier and Size were related neither with Buyer Concentration nor with Seller Concentration.

²⁷It should be noted that Size was negatively correlated with both Buyer Concentration and Seller Concentration, while Tier was positively correlated with only Buyer Concentration. Although not reported here because of the limited space of this article, we conducted correlation analyses by using those firms that adopted EDI with both sellers and buyers in the two countries. The results indicated that correlations among the defined independent variables and control variables were not very high, suggesting that multicollinearity is not a serious concern in the following multiple regression analyses.

²⁸In general, logistic regression is more appropriate for observational studies, whereas profit analysis is appropriate for designed experiments. See Pindyck and Rubinfeld [68].

²⁹We checked the variation inflation factors (VIF) associated with each coefficient in the six regression analyses. All the VIFs were below 2.2, suggesting no serious problems with multicollinearity in our analyses.

³⁰Our finding that Hypothesis 5 (Number of Competitors) was supported on EDI Adoption by the total sample confirms our assumption that “buyers' pressure” is more important than Porter's horizontal pressure from competitors. Vertical pressures [69]—i.e., customers' requests and, to a lesser extent, suppliers' requests—seem to be more important than horizontal pressures from competitors in encouraging EDI adoption and integration in the automobile industry.

³¹Here, we considered our finding that Japanese data conflicted with Hypothesis 3 (Frequency).

³²Because our focus is on firms' EDI adoption with their suppliers, we do not report our regression analyses by using those firms that adopted EDI only with their buyers.

³³Our F-test analysis on EDI Performance showed that variances in the two countries were statistically different at the 5% level. (However, there is no statistically significant difference in standard deviations of EDI Adoption between the two countries.)

³⁴Our interviews also confirmed this data. For example, one manager in a Japanese transplant in the U.S. said, “U.S. auto suppliers are something like a small Cessna flying freely without a radar, while Japanese suppliers are something like a jet with a radar. We can't fly freely because we must watch carefully our customers and suppliers with our radar.”

³⁵In order to mitigate this problem, the authors plan to collect data from the same companies in the near future to create panel data sets.