Ody-Brasier, A., and F. Vermeulen

2014 “The price you pay: Price-setting as a response to norm violations in the market for Champagne grapes.“ Administrative Science Quarterly, 59: 109–144. (Original DOI: 10.1177/0001839214523002)

On page 120, before the subsection Foreign subsidiaries, the following text should be inserted as a new paragraph:

“Rayon Boissons is a publication dedicated to “buyers specializing in drinks for retail, supermarket directors, drinks manufacturers, floor and aisle managers, wine traders.” These are not data that grape growers would normally have access to directly, as they are archived only at the French National Library in Paris and only available to researchers. Although information of who is supplying the supermarkets can find its way within Champagne through other, informal channels, it is possible that growers make errors in judging who engages in it and not. They might suspect a house that does in fact not supply supermarkets, and they could potentially also not suspect a house although it does engage in the practice. It could also be that these errors are not randomly distributed among houses but are more prevalent for certain actors than others. We believe both types of errors, however, would lead to a conservative bias in our estimates, rather than overrepresent the effect. Nevertheless, for the aforementioned reasons, we have to be careful interpreting the findings on this variable.”

On page 131, the second paragraph reads as follows:

“Supermarket brands. We also predicted that the more supermarket brands a house supplies, the higher the price it is charged for grapes. All three model specifications using random and fixed effects—displayed in table 4—support this hypothesis. The random-effects models, as displayed in the columns labeled model 1 and model 4, show that a house that supplies a supermarket brand pays almost 2.24 euros more per kilogram of grapes than a house that supplies no supermarket brands. The effects of the fixed-effects estimator display even higher values. These fixed-effects models make us confident that there are no obvious endogeneity problems, for instance, pertaining to issues of local integration.”

The paragraph should read as follows:

“Supermarket brands. The three model specifications using random and fixed effects—displayed in table 4—obtain positive effects as predicted. The random-effects model, as displayed in the column labeled model 1, suggests that a house that supplies a supermarket brand pays 2.24 euros more per kilogram of grapes than a house that supplies no supermarket brands. However, using a two-tailed test, the coefficient is only fully significant in model 1; in model 2a, it is only marginally significant (p < .10), and in model 3 not significant (p > .10). This is caused by the higher variance on the variable’s estimate in these models, rather than the coefficient itself decreasing. This increase in variance is likely the result of teasing out the time-invariant house effects, either through the fixed-effects estimator (model 2) or the Mundlak transformation (model 3), while most variance on this variable occurs between houses. The difficult access of the supermarket data to grape growers, as noted earlier, could also play a role, because the information on firms changing their actions might not disseminate instantaneously within Champagne.”