Suppliers Performance and Parts Transactions with Customers

This paper investigates the relationship between suppliers’ performance and their parts transactions with customers in Japanese automobile industry. We generate a hypothesis focusing on supplier’s learning processes, and test it using Japanese auto parts transactions data set during the period 1993-99. Results show that suppliers who prove important to the main customer at the same time maintaining business relationship with numbers of other customers, tend to surpass other suppliers in their performance. In other words, we find how important it is for suppliers to build stable business tie with the main customer and broaden relationship with other various customers.

From an economic point of view, it does seem contradictory that suppliers are "broadening the scope of customers," at the same time, "strengthening business ties with specific automakers." However, when we take particular note on the suppliers' learning process, both efforts are consistent and desirable to be practiced simultaneously. Hereafter, we will generate a hypothesis and conduct limited test of it.

Regarded Important from Main Customer
Definitely, information, which is passed on from automakers who are customers to their supplier, play a major role in the suppliers' learning process. In fact, it is commonly observed to have automakers supervise suppliers on product management and quality control. On the other hand, it is often seen among suppliers to receive technical experts from brotherly automakers or send their own men to customer automakers for training. Moreover, it is ever crucial for suppliers to gain information, However, even when automaker B is the main customer for supplier A, this does not simply allow supplier A to draw significantly valuable information from automaker B. Among Japanese manufacturers, each auto part is usually procured from more than one supplier. Automakers procure same parts from multiple suppliers, and in doing thus they refer to a list in which suppliers are minutely rated by competence and background. As for automakers, especially in co-development projects where highly confidential information needs to be shared, tendency is stronger to partner-up with suppliers whose main customers are themselves. This is because suppliers whose main customer is another automaker might have the potential to lay priority on the other company and, may it not be intentional, leek confidential information to rival companies. In fact, in our interviews, automakers would tie up with their main supplier whose main customers are themselves as well in co-development projects of new technology. Therefore, as long as supplier A is not treated as important supplier from company B, A will hardly gain valuable information from B: Advantages gained through learning from main customer will be reserved.
On the contrary, when tight relationship between makers and suppliers are maintained, in which both companies regard each other as main partner, trust grows and opportunistic breach of information become unlikely: Information of higher confidentiality can be exchanged. Thus, suppliers regarded important from main customers take advantage of learning process compared with suppliers that are not.

Customer Scope
Another thing that is import in suppliers' learning process is to broaden the scope of customers. First, because dealing with many numbers of automakers would allow suppliers access to many different kinds of information, compared to suppliers who limit the number of customers. Additionally, developing and producing virtually same parts for different automakers will engage more numbers of test run and improvement activities during product design and production design procedure, thus quite possibly increase learning effect and experience effect of the supplier. Thus, we believe that suppliers' learning process is improved through business relations with variety of automakers.

Mutually Complementing Relationship between "Regarded Important from Main Customer" and "Customer Scope"
We further expect that, from discussions in Sections 2.1 and 2.2, suppliers' importance in main customers' eyes and scope of customers are mutually complementing: meeting either one alone would be insufficient. In other words, suppliers are able to merit through learning process only when both priorities are high in main customer's eyes, and business relation with many customers are maintained.
We consider that enhanced learning processes improve suppliers' ability. Thus, from above discussion, we can derive a hypothesis: Suppliers who remain highly important to main customer automaker at the same time maintaining business relationship with numbers of other customers, tend to surpass other suppliers in their performance.

Sample and Data
This paper will test the hypothesis presented in Section 2, though quite restrictively, on Japanese auto parts transactions during the period 1993-99, utilizing logistic regression analysis method.
We here treat parts transaction as the unit of analysis. That is, we analyze each delivery of parts from supplier to automaker as individual transaction.
We dropped from sample any kind of auto parts not outsourced, partially or totally. This is because we cannot compare indexes correctly between parts which is totally outsourced and those that are not.
Besides, other types of parts were excluded; for example, partial data flaw, not procured by every company, lack of data in either time period, or part category changed. As a result, we obtained 90 types of parts with satisfactory data.
In addition, for data on each supplier company unit, we referred to Nihon no Jidosha Buhin Kogyo: 1992Kogyo: -1993 Numbers of sample is mentioned later.

Dependent Variable and Independent Variable
We will then explain the variables applied in logistic regression analysis. See Table 1 for calculation methods of each variable.
In this paper, we applied Continuity of Parts Transactions (CPT) as suppliers' performance index, which is dependent variable. This is a discrete

Customer Segment
In testing above hypotheses, we divide all parts transactions in two groups, that is, transaction with main customer (Main Maker Transaction) and transaction with other customers (Other Maker Transaction). We conducted logistic regression analysis on the latter group alone because "main customer importance" having continuous positive effect on "main maker transaction" is in some sense unquestionable. Thus we took the other samples for testing hypotheses.
Furthermore, 319 supplier companies delivered at least one of 90 parts dealt in this paper to either 9 car companies in 1993. Among them, 13 were 100% Parts market HF Sum of square values of suppliers' share in volume of delivery to either of 9 automakers in parts market divided by 100 (Herfindahl Index).
Market rate Transition of total parts procurement of 9 automakers through 1993-99, divided by total parts procurement of 9 automakers in 1993.

Importance to Customer
Suppliers' share in total procurement made by customer automaker for each auto parts in 1993.
Automaker dummy Dummy value given to 9 automakers except Isuzu concerning which maker transaction continued or did not continue in each parts transaction.

Parts category dummy
Dummy value given to 5 of 6 categories of auto parts: engine part, engine electrical component part, car body part, exterior trim part, internal trim part: car body electrical component part excluded.
Main customer dummy Dummy value given to 9 automakers except Isuzu concerning who is the supplier's main customer company with each parts transaction unit.

Corporate level control variable
Degree of dispersion Ratio of numbers of categories supplied by each supplier in above 6 categories in 1993. foreign capital supplier and we excluded them from sample because "regarded important from main customer" and "customer scope" in mother country must be more important. Also for 76 suppliers, company unit data could not be attained, thus suppliers data used in this paper is 230. Then 79 have main customer transaction alone, which leaves 151 companies and 1119 parts transactions with customers other than main as our test sample.

Degree of concentration
As for control variables, we applied many parts level variables and corporate level variables. Table 3 shows the result of logistic regression analysis. In model 1 RIMC and CS is the independent variable and in model 2 interaction term Total volume of parts transactions 5
is added besides RIMC and CS as independent variable. In each model (a) is parts level variables only and (b) include corporate level variables.
First, in models 1 and 2, RIMC has positive effect on CPT. Besides, model 1 (b) is statistically significant at 5% while others are statistically significant at 1%. Therefore, hypothesis 1 is supported. Second, in models 1 and 2, CS has positive effect on CPT and all are significant at 1%. Therefore, hypothesis 2 is supported. Third, in model 2, RIMC*CS has positive effect on CPT and all are significant at 10%. Therefore, hypothesis 3 is supported.
This paper will conclude from above analysis Recently, news media loudly report the "collapse of 'Keiretsu' relationship," nevertheless, discussions in this paper suggest that such point of view is superficial.
Surely, many suppliers are increasing the number of customers, and as a result, competition among suppliers has become intent than ever.
However, on the other hand, as clarified in this paper, it is nonetheless important for suppliers to maintain and develop intimate relations with main customers, and as for car manufacturers, it is meaningful to maintain and develop relationships with highly competent core suppliers. Thus, at least between automakers and suppliers who consist Japanese auto parts supplier system, "long term, continuing and collaborative maker-supplier relation" and "competition among suppliers" should be encouraged simultaneously. In other words, "Keiretsu" is not collapsing but its principle characteristic, that is "co-existence of competition and collaboration," would be purified and further developed.