ادغام شرکت ها موقع رقابت شرکت ها با همدیگر بر سر قیمت و ارتقاء کیفیت

We analyze the bias from predicting merger effects using structural models of price competition when firms actually compete using both price and promotion. We extend the standard merger simulation framework to allow for competition over both price and promotion and ask what happens if we ignore promotional competition. This model is applied to the super-premium ice cream industry, where a merger between Nestlé and Dreyer's was challenged by the Federal Trade Commission. We find that ignoring promotional competition significantly biases the predicted price effects of a merger to monopoly (5% instead of 12%). About three-fourths of the difference can be attributed to estimation bias (estimated demand is too elastic), with the remainder due to extrapolation bias from assuming post-merger promotional activity stays constant (instead it declines by 31%).

Antitrust laws prohibit mergers that substantially lessen competition, and most often this means mergers that raise price. Enforcement agencies have a relatively good understanding of price competition, and how to measure the loss of price competition caused by merger. Less well understood are the effects of mergers when firms compete in multiple dimensions. In this paper we study mergers in industries where firms compete by setting both price and promotion, and ask what happens if we mistakenly assume that price is the only dimension of competition. To answer the question, we build a structural merger model where firms compete using both price and promotion. We find two sources of potential bias from ignoring promotional competition. The first is what we call “estimation bias,” a type of omitted variables bias. If promotion is correlated with price, then observed price changes will proxy for unobserved (or ignored) changes in promotional activity. As a consequence, price elasticity estimates will be biased. Bias in estimated own-price elasticities affects the post-merger price prediction because a merged firm facing a more elastic demand would not raise price as high as a merged firm facing a less elastic demand, all else equal. The second source of potential bias is what we call “extrapolation bias.” Following a merger, we would expect the merged firm to internalize both price and promotional competition among its commonly owned products. In price-only merger models, promotional activity is implicitly held constant at pre-merger levels when the post-merger equilibrium is calculated. This leads to extrapolation bias when optimal price depends on the level of promotional activity. Interestingly, both types of bias depend, in part, on whether promotional activity makes demand more or less elastic. Our analysis suggests a simple heuristic: if promotional activity makes demand more (less) elastic, price-only merger models will under- (over-) estimate the magnitude of the post-merger price increase. Only in the special case where optimal price and promotion are independent of each other will a price-only model provide an accurate prediction of post-merger prices. To empirically assess the magnitude of this bias, we extend the standard merger simulation framework by allowing firms to compete over both price and promotional activity. This framework is applied to the super-premium ice cream industry, where the Federal Trade Commission (FTC) challenged a 2003 merger between Nestlé and Dreyer's.1 The results from our price-plus-promotion merger simulation framework are compared to counterpart estimates from a price-only model. The structural model which ignores promotion substantially under-predicts the price effects of a merger when firms actually compete via both price and promotion. This comparison demonstrates that the standard merger simulation framework can be extended to accommodate situations where firms compete in multiple dimensions, and that doing so may significantly alter merger predictions.

Models necessarily simplify the real world. What we want to know is whether the simplifications bias the model predictions. In this paper, we analyze mergers in industries where firms compete by setting both price and promotion. Our investigation identifies two types of bias in merger models that ignore promotional competition: estimation bias from omitting promotional activity in the demand specification, and extrapolation bias from mistakenly holding promotion fixed at the pre-merger equilibrium. Extrapolation and estimation biases arise when optimal price varies with promotional activity (and vice versa). This is the case in the super-premium ice cream industry, the focus of our empirical analysis. Pricing and promotion decisions are jointly determined in this industry, as is evident by the fact that most price variations occur simultaneously with variation in promotional activity (i.e., products typically have price reductions in those weeks where they are on promotion via a feature advertisement or an in-store display). The results from our case study confirm that a price-only model performs poorly when price and promotional activity are jointly determined. Our analysis demonstrates the potential for biased predictions in price-only merger models, and shows how multi-dimensional competition can easily be incorporated into the standard merger simulation framework. Additional research is needed to determine whether models that explicitly account for non-price competition can better predict the effect of real world events, such as consummated mergers. For mergers that were blocked by the antitrust authorities, however, testing the sensitivity of models to alternative assumptions of how firms compete, like the exercise in this paper, may be the next best alternative.