Agency relationship and technology diffusion: Evidence from post-patent pharmaceuticals

Presenter: Toshiaki Iizuka, Keio University

Abstract

In many countries, increasing the utilization of generic pharmaceuticals is an important policy agenda. Prescription drug expenditures are one of the fastest growing segments of medical expenditures. However, the cheaper alternative is not always used. For example, in countries such as Belgium, France, Italy, Japan, and Spain, the quantity share of generic drugs is below 20% as of 2004.

Regardless of the importance, still relatively little is known about what influences the choice between generic and brand-name drugs. This is in part due to the difficulty to observe financial incentives that doctors and pharmacists face, which would affect generic substitution. Dynamic nature of the prescription process also makes the analysis difficult. Prescription choice is likely to be influenced by the prescription history of the patient and doctor, but panel data that can follow doctor and patient behavior over time is rarely available.

This paper studies the determinants of generic substitution, using large patient-level panel data from the Japanese market. The data set covers more than 380,000 prescriptions between August 2003 and December 2005 and includes 52 drugs that faced generic entry after 1995. A unique feature of the Japanese market is that doctors often both prescribe and dispense drugs and can earn the price-cost markup by purchasing and selling the drug to the patient. Importantly, I can observe these markups from data and thus can examine whether, and to what extent, financial incentive affects generic substitution. The panel data also allows me to examine how patient-doctor preferences and state dependence affect generic substitution. I estimate dynamic probit models that take into account these factors.

I find that financial incentive, both observed and unobserved heterogeneity, and state dependence play important roles in generic substitution. Several findings are noteworthy. First, doctors who both prescribe and dispense drugs are more likely to dispense generics, suggesting that financial incentives play an important role in generic substitution. Interestingly, this effect is present only in small clinics that are generally operated and owned by a doctor but not in larger hospitals. This may be because, unlike the owners of small clinics, doctors in large hospitals have no personal incentives to profit themselves by prescribing generics. Second, more detailed analysis that focuses on small clinics revealed that, when doctors both prescribe and dispense drugs, doctors are sensitive to both the price-cost markup they pocket and the out-of-pocket expenses of the patient. However, these markup and price sensitivities disappear when doctors prescribe drugs but do not dispense them. Third, estimated parameter values suggest a strong state dependence in this market: a generic prescription yesterday increases the probability of getting the generic version today by 64%. Finally, the results indicate that accounting for both patient and doctor “preferences” are important: the doctor preference term is highly significant and is an important determinant of generic substitution. Also, the variables that condition for unobserved patient heterogeneity also enter the regression significantly.