Examining the relationship between the Gini coefficient and drug prices
An excerpt from our monthly HEOR Newsletter – Alliance Aligned.
“Today you have one global price for innovators so as a result there are whole countries without access to medicines. And within some countries, there are whole sections with no access. Differential pricing would solve this. But how do you set different prices? I don’t know.”
– Severin Schwan, CEO Roche Group
As opposed to setting one global price for all markets, differential pricing is the practice of setting different prices different markets to potentially create larger profits for pharmaceutical manufactures and possibly greater access to drugs for patients who might not be able to afford a homogenous, global price, which may result from reference pricing.
In Garau et. al., the authors discuss the potential benefits differentiating prices in drug markets could reap for consumers and pharmaceutical manufacturers. Though the economic theory is sound, the practical of issue of how to define the rule to set drug prices in different markets remains. Under plausible assumptions, the authors expect that variation in national market demand for drugs due to price changes will vary with country income per capita and conclude that higher income countries should pay higher prices whereas lower income countries should pay lower prices (Garau et. al. 4).
However, in Danzon, the author points out that often times the distribution of income in many low- and middle- income countries is skewed (178). Consequently, manufacturers may aim to price medicines at the small, high-income subgroups. This is contrary to public health goals where pragmatism and fairness would dictate that it must be available to the majority of the population in need, not just the privileged. The goal of this article is to understand how fairness and equity currently play out in drug markets.
To examine fairness within markets necessitates some measure of income distribution within a country such as the Gini index. To see how Gini indices of countries relate to drug prices, Spearman’s partial correlation between the Pharmacy Purchasing Prices and Gini indices were calculated controlling for the effects of two confounding variables, total population and purchasing power parity adjusted GDP.
These were selected a priori because it was suspected that they could be associated with Pharmacy Purchasing Prices. Several pharmaceutical products were collected in various countries around the world for various therapeutic areas. To be comparable, prices were converted to US dollars using the nominal exchange rates at the time the drug prices were current. In situations where there are several packs of the same presentation, the result for the pack that was available in the most number of countries was presented. In Table 1, we present the calculated Spearman’s partial correlations for the pharmaceutical products examined.
For 20 out of the 29 drugs considered (69%), the partial correlations were negative. Interpretively, this means that if population and purchasing power parity adjusted GDP were the same in a group of countries, those with higher inequality in income distributions would have slightly lower Pharmacy Purchasing Prices on average.
This agrees with what would be expected if prices are set based on public health need and equity. To be effective, drugs must be available to the whole population as opposed to a select few who can afford them. As the Gini index gets higher, the wealth becomes more concentrated in a minority subset of the population. Consequently, prices need to be lower in order to be accessible to the majority of the population, thus explaining the negative value of Spearman’s partial correlations of PPP with the Gini index.
For the exceptions, further research is still needed. A possible hypothesis could be that drugs intended for conditions that are non-fatal, palliative and merely meant to improve quality of life behave as luxury goods.
Thus only people with disposable income would be willing to spend money on them. In societies where there is high inequality in the income distribution (higher Gini indices), such drugs would only be marketed to those where the wealth is concentrated thus have higher prices. Hence these drug prices would have a positive Spearman’s partial correlation with the Gini index.
The same hypothesis might also hold for the method in which the drug is delivered. Certain delivery methods may be preferred, but elective, over others due to various risks, discomforts caused by the method or side-effects avoided. For instance, pills, syrups or inhalers are less invasive than injections. Patients wishing to have a more desirable deliverer method would be subject to paying a premium.
In conclusion, after adjusting for the effect of population and purchasing power parity adjusted GDP, Pharmacy Purchase Prices tend to be lower in countries that have higher income inequality. This supports the notion that the current drug pricing practices are conducive to accessibility for the majority of patient who need them.
However, to be feasible differential pricing should be profitable to pharmaceutical manufacturers. Consequently, a company’s profit maximizing rule might not necessarily yield prices based on fairness. Lastly, the exact rule relating to income inequality under a differential pricing scheme needs further examination.
Garau, Martina, Towse, Adrian, and Danzon, Patrica M. “Pharmaceutical pricing in Europe: Is differential pricing a win-win solution?” Office of Health Economics, The Wharton School – The University of Pennsylvania. Feb. 2011. Print.
Danzon, Patricia M. “At What Price?” Nature 449 Sept. 2007: 176-179. Print.
 The Gini index for a country is a number between 0 and 1 that reports how skewed the income distribution is within that country. A value of 0 indicates that the wealth is equally distributed within the country (everyone’s income equals per capita income) whereas a value of 1 indicates that the wealth is concentrated in one individual.
 Spearman’s partial correlation measures the strength of the monotonic relationship between two variables of interests after controlling for the effects of confounding variables.