A significant part of economic theory focuses on the assumption of a representative consumer buying a homogeneous good. For example, think of the standard Bertrand and Cournot models of oligopoly. Consumers only care about the prices in the market and decide how much of a good to buy and from which firm in order to maximize their utility (given a budget constraint). We know that price competition is fiercer than quantity competition and this result is described by the so called Bertrand Paradox (i.e. in an oligopolistic market for homogeneous goods price competition leads to the zero profits outcome). In the real world however goods might differ not just in prices, but also in other characteristics such as variety and quality. As pointed out by Waterson (1989) and Cremer et al. (1991), “product differentiation is one of the pervasive features of modern economies” and “most of the real world markets are characterised by product differentiation”. Goods, even if they satisfy identical needs, are not always identical, homogenous. At the same time, consumers are not identical either: they can have different willingness to pay (or income) and different preferences regarding some product characteristics. The literature distinguishes between horizontal and vertical product differentiation. Following Lancaster (1979) characteristics approach, a good can be thought as a bundle of characteristics. Two product varieties are horizontally differentiated when they contain different amounts of each characteristic and two consumers with different tastes for such characteristics would not unambiguously choose the same good. Two product varieties are vertically differentiated instead when a variety contains more of all characteristics than the other, so that if goods are sold at the same price all consumers would prefer the former. Examples of goods horizontally differentiated can be two cars: one faster but less safe than the other (consumers who value speed more than safety will prefer the first car and the rest will prefer the second). If instead a car is faster and safer than the other, then these two cars are vertically differentiated (and all consumers unambiguously prefer the faster and safer model). In Figure 1 may be a car faster but less safe then the others, while another could be slower but safer ( in the figure). Cars and are horizontally differentiated. Some consumers prefer the former, others the latter. If a car has the highest level of all characteristics, all consumers will prefer it ceteris paribus. In Figure 1, and are vertically differentiated since is faster and safer than .
Figure 1: Characteristics space for cars
Another important distinction considered in the literature is the one between address and non-address approaches to product differentiation. The non-address approach, not considered in these notes, assumes that each product is a generalized substitute for all other (a finite number) varieties in the market, as in Chamberlain (1933) and Dixit and Stiglitz (1977) monopolistic competition model. The address approach (the one we are going to focus on here) assumes that each good is located somewhere in the characteristic space and has specific neighbours. Consumers purchase only one good, making an either/or choice among product varieties. Seminal examples (just to mention a few) from the literature on product differentiation under the address approach are Hotelling (1929), d’Aspremont et al. (1979), Salop (1979) when products are horizontally differentiated; Gabszewicz and Thisse (1986), Shaked and Sutton (1982), (1983), (1987) represent the counterpart when products are assumed to be vertically differentiated. An important question that could come to mind should be: why might firms decide strategically to differentiate their products (for example deciding to produce a different variety or a different level of...