Effects of wholesaler concentration on retailers by format: evidence from Japanese brand-level price data

Abstract

Using unique brand-level wholesale price data, this article examines how purchase prices of commodities for retailers are influenced by wholesale concentration. Compared with studies focusing on retail prices, little empirical work has measured the impact of wholesale concentration on wholesale prices. Economic theory suggests that the sensitivity of purchase price in response to degree of wholesale concentration is smaller for large retail formats than for small retail formats in the presence of intra-format retail competition. To test the validity of the theoretical implication, I measure the impact of the concentration on purchasing prices for retailers by format. Empirical findings support the hypothesis that the large retail format tends to draw advantageous purchase prices when wholesalers become increasingly oligopolistic.

Acknowledgements

The author is grateful to Kenn Ariga, Tatsuhiko Nariu, Hiroyuki Odagiri, Akio Torii, Masayoshi Maruyama, Koki Arai, Yoshihiro Nakata, Tomohiro Machikita, Shuzo Abe, Takuya Satomura, Miyuri Shirai, other participants at annual research conference of Japan Society of Marketing and Distribution and anonymous referees for helpful comments. The research for this article is supported by a Grant-in-Aid for Scientific Research (B) from Japan Society for the Promotion of Science. The author is solely responsible for any remaining errors.

Notes

¹ Although hypermarkets and other emerging large retail formats contribute to high retail concentrations in advanced countries, the retail industry in Japan still exhibits a lower degree of concentration. As a consequence, the average scale of a retail outlet is lower, and the number of retail outlets per person is higher, relative to many other advanced economies. See Ariga et al. (1991), Ariga (1995) and Matsui et al. (2005).

² Although statistics on the distribution sector in many developed countries are more difficult to obtain than those for the manufacturing sector, Ariga (1993) argues that compared with other advanced economies, the Japanese statistics allow a more thorough examination of distribution systems.

³ In the National Price Survey, goods such as beer or disposable diapers are defined as ‘items’, whereas brands, such as ‘Pampers’ or ‘Ultra Moony’ disposable diapers, are defined as ‘specifications’.

⁴ As shown in Table 1, wholesalers and retailers are classified into several types of formats. The survey defines wholesalers as follows: (1) Producer's branch: the branch office of a producer that sells and delivers its own products to wholesalers and retailers. (2) Sales Company established by producer: a subsidiary or affiliated company that was established by the producer for the exclusive sale and delivery of its own products. (3) Other wholesaler: a single unit or association of agricultural cooperatives and livelihood cooperatives that engage in selling to retailers or wholesalers. Purchasing departments of chain stores, namely voluntary chains and franchise chains, are also categorized here.

⁵ A rationale for using the HF index on the basis of the number of workers instead of on the basis of the amount of sales is that wholesale formats in Japan are less diversified compared with other countries such as the United States. See Nishimura (2004) for details. If every wholesaler has similarities including format, the correlation between the sales amount and the number of workers across wholesalers will be high. This correlation allows us to approximate the sales-based HF index with one calculated on the basis of the number of employees. By contrast, the wholesale formats in the United States are considerably more diversified, including drop-shippers, cash-and-carry, wagon distributors and so on. Under these circumstances, the correlation between the number of workers and the sales amount of each wholesaler would be weaker.

⁶ Full manufacture reimbursement for returns is frequently adopted for the distribution of shopping goods, reflecting a sort of power from department stores.

⁷ The author thanks the referee for the helpful suggestion.

⁸ Equation 4 includes three independent variables, ln HF, (Department store dummy) × (ln HF) and (Supermarket dummy) × (ln HF), all of which have the possibility of correlation with the error term. However, Equation 5 for separate estimation according to retail format includes only one variable, ln HF, that has such a possibility.

⁹ Torii and Nariu (2004) demonstrate that length of marketing channels and degree of wholesale concentration are influenced by characteristics of products, and argue that the Japanese distribution systems particularly have such a tendency.