Exploratory compositional data analysis using the R-package robCompositions Compositional data Aspects of ECDA for compositional data : outlier detection and compositional biplots

Abstract

Extending Dyer (1996), the author tries to understand how Nissan and Toyota manage to have lower transaction costs and higher asset specifity compared to the three US manufacturers. A related question is how come GM has higher transaction costs than Ford and Chrysler. After a round of interviews, he proposes five concurrent factors as the solution to the puzzle: 1. Repeated transactions with the same group of suppliers: i.e. the average relation in Japan lasts longer; 2. Econmies of scale and scope in transactions, i.e. the same supplier get more business, on average from a smaller number of clients. This means that fairness can be achieved in the long term, as opposed to for each transaction. 3. Higher information sharing, which reduces opportunism. 4. Use of self-enforcing (vs. contractual) safeguards. 5. Investment in relation-specific assets, which increase the value of the transaction (as opposed to reducing the transaction costs). He suggests that efficiency arguments should move from the minimisation of transaction costs to the maximisation of transaction value.