The infant formula rebate program of the Special Nutrition Program for Women, Infants, and Children (WIC) requires each state to hold an auction where the low-bidder among the three major manufacturers of infant formula became the sole provider of formula to the state, which issues WIC voucher to low-income WIC participants. Using these WIC vouchers the WIC consumers can get contract brand infant formula for free from participating grocery stores or directly from the state. The WIC agencies then reimburse the retailers for the full retail price of the formula purchased by WIC consumers. Since the rebate program started, the wholesale prices of infant formula have increased markedly. The infant formula industry is highly concentrated, with three firms accounting for more than 90% of market share. Since the start of the rebate program, Congress has been concerned with the program's effects on non-WIC consumers. Congress mandated several studies of this issue by the GAO and USDA. Despite these government and other academic studies, there is no consensus on a theory as to how the rebate program affects non-WIC consumers. Empirical studies have not resolved this issue due to a lack of data. The goal of this study is two-fold. First, in contrast to earlier studies that describe firms' behavior as price discrimination, we present a theoretical model based on spillover effects. We refer to increases in non-WIC demand for the WIC brand due to its WIC brand status as a spillover effect. A firm will submit an extremely low bid because the WIC contract winner brand gains substantial additional sales and becomes the dominant player in the non-WIC market. The resulting differential between the WIC and non-WIC price may exceed that of the pure price-discrimination model. Second, we use widely available scanner data to test our hypothesis and estimate the magnitude of the spillover effect. MethodologyOur theory is based on three important characteristics of this market. First, the wholesale prices do not change substantially before and after a firm wins the WIC contract in a state. Second, all three major firms set a national wholesales price annually. Third, the share of the WIC contract winner increases dramatically after a it takes over a contract. Hence, a price discrimination model where firms set wholesales prices to each retailer or state and adjust wholesale prices with every contract change is not consistent with these stylized facts. We solve for the oligopolistic equilibrium uniform prices (in the absence of a WIC rebate program) and compare those to the equilibrium prices under WIC if the firms could price discriminate or where they cannot price discriminate and there are spillover effects. Possible channel of spillover could be simply change in shelf space allocation in grocery stores. We show that with a large spillover effect, a firm is willing to bid more aggressively so as to be able to charge higher prices to the non-WIC market. As a result, the differential between the WIC and non-WIC price may exceed that of the price discrimination model. We estimate the magnitude of the spillover effects. It would be straightforward to investigate spillover effects if we had data that distinguishes WIC and non-WIC sales. However, such data are not available. Therefore to identify the spillover effect, we explore the variation over time in states where the WIC contract has changed between firms during our sample period. Our empirical strategy is based the following rationale. After the WIC contract change between brands, WIC participants, who receive the current WIC brand for free, immediately switch from the loser to the winner brand. Consequently, stores must provide more shelf-space to the contract winning brand. Non-WIC consumers may be influenced by shelf-space allocation and are more likely to choose the contract winning brand as a result. However, parents do not like switching brands for fear that their babies will reject the new brand. Thus, only new non-WIC consumers are influenced by the shelf-space allocation. Consequently, the non-WIC spillover effect occurs slowly over time as parents with new babies enter the market and those with older babies exit. Are idea, then, is to identify the non-WIC spillover effect through a gradual adjustment pattern following a contract change. The panel structure of our data also allows us to exploit the variation in share changes when different firms win or lose the contracts in various cities. Using IRI scanner data from year 1997 to year 1999 from 11 cities in 7 states where WIC contract change occurred during the period, we estimate a multinomial logit model with dependent variables being the shares of the WIC contract winner, loser and the remaining brands, and independent variable are the time elapsed since the contract change, state birth rates, and firm and state fixed effects. We confirm that there is an instantaneous switch in shares after the contract change followed by a spillover effect that gradually increase over time until it too is substantial. Potential for generating discussionThe WIC program provides formula to half of all U.S. infants. Despite the importance of this rebate program, no consensus exists on how this program affects non-WIC consumers, either theoretically or empirically. For example, the U.S. GAO (1998) argued based on simulations that spillover effects could not be substantial in the WIC infant formula market. Thus our theory and confirming empirical evidence contrast substantially with earlier studies. Reference:U.S. General Accounting Office. 1998. Food Assistance: Information on WIC Sole-Source Rebates and Infant Formula Prices, Report to the Chairman, Committee on the Budget, House of Representatives, GAO/RCED-98-146.
