Community Research and Development Information Service - CORDIS

Final Activity Report Summary - SASWCO (Search, Switching Costs and the Design of Optimal Competition Policy)

This project made significant progress on a number of important lines of research. One line has been on deepening our understanding about the effects of search and switching costs in markets. We found out that search costs need not make markets less competitive per se, but perhaps the opposite. This depends on whether prices are observable or not, and on the strategies available to the firms, such as advertising that persuades consumers to visit firms in a given order. Another important finding has been that switching costs need not result in higher prices for consumers. Another line has been on recognising that search and switching costs generate consumer heterogeneity in price and product information.

We have first studied a 'reduced-form' model of this heterogeneity and encountered that some consumers may benefit and some may lose from increased competition. Moreover, the gains from competition can be asymmetrically distributed across buyers. The project also made progress in making such heterogeneity endogenous. A third line has been on understanding how market transparency on the consumer side effects incentives to merge, collude and locate. The project showed that when search costs are significant the short-run gains from merging are negative, therefore eliminating the incentives for firms to merge in environments where firms compete in prices and products are differentiated. In the long-run, however, we proved that the merging firms can reorganise their business so as to generate economies of search thereby making a merger profitable and, in some cases, even welfare improving.

The project also studied how cartel stability is affected by search costs and found that cartels are more stable in markets with high search costs. This result should be useful for fine-tuning the targeting efforts of competition authorities. Finally, this line of effort has investigated the reasons for homogenous product firms to cluster in malls and found that search costs provide a rationale for the typically observed market structure where some firms remain isolated and others move to a mall. In a fourth contribution, the project made substantial progress on developing new empirical techniques to estimate models with limited choice sets arising from search, switching costs, or firm heterogeneity in product availability. This effort has included both work on the identification and on the estimation of search and switching costs. Moreover, effort has been put on both markets for homogeneous products and in markets for differentiated products. These techniques were firstly applied to online markets for memory chips. An important finding is that consumers have heterogeneous search costs and that this heterogeneity is somewhat limited. Secondly, we applied the techniques to the automobile industry.

The project confirmed that search costs are related to demographic aspects such as seniority, household composition and income. A very important finding is that search costs are relevant in this market, making it less competitive than initially thought. Thirdly, the project studied the estimation of demand systems for differentiated products allowing for firm heterogeneity in product availability to enter consumer utility. Fourthly, there has been measurable progress in modelling the role of search costs in the demand for saving deposits and the role of switching costs in natural gas and electricity contracts. Finally, the research team has dedicated a significant amount of time to study the performance of new estimators, among others the conditional empirical likelihood estimators (CEML). These estimators are useful in that they allow for the estimation of a conditional moments model without the need of using instruments.

This is relevant for the topics of the project because demand systems for differentiated products can also be formulated as a conditional moments model, and its estimation by the method of instrumental variables is known to be problematic. The results suggest that when the instruments are weak, the conditional empirical likelihood estimators perform poorly, however, when the instruments are strong, they perform very well.

Reported by

See on map
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top