Estimating the cost of strategic entry delay in pharmaceuticals: The case of Ambien CR

Abstract

With the Hatch-Waxman Act of 1984, the FDA included an unchallengeable exclusivity period for newly approved drugs, independent of patents. This potentially generates an incentive for firms to strategically delay the introduction of new versions (reformulations) of drugs until just before patent expiration of the original drug. This way the reformulated drug competes mainly with newly introduced generics of the original drug. If instead, the reformulated drug was to be introduced well before the original drug’s patent expires, the reformulated drug would compete only with the original drug. While the pattern of strategic delay is well documented in the literature, its effects on consumers and firms are not. Reformulations may increase utility through improved efficacy and through fewer doses per day or a more even molecule decay rate. However, as suggested in the press and literature, it is also possible that the adoption of reformulated products is mostly the result of advertising rather than product-related benefits. Using detailed prescribing and pricing data from the prescription sleep aid market, I document significant adoption of the reformulation Ambien CR and show that it is not only driven by advertising. I use these estimates to evaluate two different policies designed to induce earlier entry of Ambien CR. I find that there are large potential gains in consumer surplus and in revenue.

Appendix: A: Including DTC and journal ad pages in demand model

In this section, I address the possibility that the omission of DTC or journal advertising may bias my results. Journal ad pages are generally a much smaller portion of pharmaceutical advertising. However, I also have collected product-quarter level data on this variable to include as a control. In practice, most of the within-product variation in these measures occurs at patent expiration, when all advertising drops close to zero. As such, it might be difficult to separately identify each. If that is the case, inclusion of DTC and journal advertising in my demand analysis should not bias the coefficients on other demand parameters.

I include DTC in the demand model as a control to see how much its omission biases the key demand parameters of interest. The results are presented in Table 5. I find no significant effect of DTC on product demand, nor do I find that the inclusion of DTC alters any of the main parameters of interest in a statistically significant way. While this may seem surprising, I note that the reader should interpret the DTC coefficient with caution without a strong identification argument. Similarly, I find a small positive point estimate on adpages, but it is not significant. In the IV specification, the instruments become very weak with the inclusion of these variables.

Table 5 Robustness - main results including DTC

I conclude from this analysis that detailing is likely a proxy for all firm advertising in my main specifications, and assuming as much will not bias the results of the estimation significantly.