Raising rivals’ costs or improving efficiency? An exploratory study of managers’ views on backward integration in the grocery market

Abstract

Large retail grocery chains’ backward integration into distribution, procurement and production is controversial, and has received a lot of attention by both policy makers and market players. If a large retail chain for instance takes over scale intensive distribution activities to its own outlets from some suppliers, direct distribution from these suppliers to other retail chains might become more expensive (and could even initiate costly industry-wide backward integration). An interesting question is thus whether large retailers undertake backward integration mainly for efficiency reasons or whether they do so in order to gain a competitive advantage through raising the costs of the smaller rivals. Theory and econometric analyses are inconclusive. The current study uses semi-structured interviews to investigate managers’ views on this issue, and does not formally test different theories. However, the results clearly indicate that large retail chains gain a competitive advantage if they choose to backward integrate, but that their main motivation for choosing this strategy is to increase channel efficiency.

Appendix

Proof of Proposition 2

From Eqs. (1) to (5) in “Backward integration might be profitable even if own costs increase” section we find:

$$\pi_{1}^{\text{B}} - \pi_{1}^{\text{N}} = \beta \left( { - \frac{1}{4} + \Delta_{1} \left( {\frac{1}{6}\frac{1}{3 - 2\beta } + \frac{1}{36}\Delta_{1} \frac{15 - 8\beta }{{t(3 - 2\beta )^{2} }}} \right)} \right)$$

(6)

First, note that if there is no size effect, β = 0, firm 1 is indifferent between backward integration or not. In this case, if firm 1 chooses backward integration, this gives rise to a competitive advantage over firm 2 as long as Δ₁ > 0. However, under the spatial demand set-up, firm 2 would reduce its price accordingly to nullify the profit effect. Obviously, this neutrality result does not survive under alternative demand specification, but it allows us to scrutinize the specific properties of the size effects of the present model.

Therefore, let us concentrate on β > 0. We now observe that (6) is negative if Δ₁ = 0. Furthermore, we directly observe (6) is increasing in Δ₁ as long as β > 0. The upper bound of Δ₁ is given by the value that forecloses firm 2 from the market; i.e. Δ₁ = 3 − 2β. Inserting for Δ₁ = 3 − 2β into (6) we have

$$(\pi_{1}^{\text{B}} - \pi_{1}^{\text{N}} )_{{\Delta_{1} = 3 - 2\beta }} = \frac{\beta }{9}(3 - 2\beta ) > 0.$$

Consequently, there exists a critical value of Δ₁, such that if $\Delta_{1} \ge \Delta_{1}^{\text{crit}} > 0,$ then $\pi_{1}^{\text{B}} - \pi_{1}^{\text{N}} \ge 0,$ and firm 1 prefers backward integration. QED.