Palmedo300x300This post presents data comparing branded antiretroviral medicine prices in countries which have entered into free trade agreements with the U.S. containing TRIPS-Plus intellectual property obligations, to the prices for the same drugs in other countries. According to publicly available data from the World Health Organization’s Global Price Reporting Mechanism (GPRM), prices of branded antiretrovirals negotiated by large institutional purchasers (like UNITAID and the Clinton Foundation) were more than twice as high between 2004 and 2014 when the sale took place in countries with U.S. FTAs. If one controls for per capita income, inclusion in international HIV/AIDS treatment guidelines, and the year of purchase, the average difference between the negotiated price of branded drugs in countries with and without FTAs in force is 57%. The price differences vary greatly by drug.

Price Comparisons

The World Health Organization’s GPRM reports the negotiated prices paid for large transactions between sellers and institutional buyers. Its website explains that the “main data providers of GPRM are the Global Fund, PEPFAR, UNITAID, and the procurement organizations working with them, such as the Clinton Foundation, Crown Agent, the Global Drug Facility (GDF), the International Dispensary Association (IDA HIV/AIDS), USAID/Deliver), Mission Pharma, Management Sciences for Health (MSH), the Partnership for Supply Chain Management (PFSCMS), the United Nations Development Programme (UNDP), the United Nations Children’s Fund (UNICEF), and the WHO/Contracting and Procurement Service (WHO/CPS).”

To see whether prices were higher in countries with FTAs, I compared their “unit price ex works” for transactions in countries that had FTA obligations at the time of the purchase with transactions in other countries.   “Unit price ex works” is the price the seller charges per unit (pill) in U.S. dollars which leaves the purchaser to assume transport and other transactional risks. This is the price used by the GPRM in its summary reports. The full GPRM database includes other gauges of prices, such as treatment cost per year. Comparisons and regressions with these alternate measures of price also show that FTA countries pay a premium for AIDS drugs (click here for raw Stata output).

The year in which FTAs were in force was taken from the U.S. Trade Representative’s page on trade agreements. I created the dummy variable fta_inforce, which equals one in country-years when an FTA was in force

On one hand, negotiated antiretroviral prices are not the best dataset to work with to show how trade agreements may be associated with higher drug prices; one would prefer to find a set of drugs priced by something more closely resembling “market pricing.” On the other hand, these are drugs relied upon by large numbers of people living with HIV/AIDS, so their prices are important.

The dataset includes prices for both branded and generic drugs over the period from 2004 to 2014, and this post focuses on the branded drugs. If one compares the prices of the branded drugs without controlling for other factors, you see that negotiated prices in countries with FTAs average 1.13 and negotiated prices in other countries average 0.49. Put differently, prices were 130% higher in countries with FTA obligations. (See the t-test results in Stata)

The difference in average prices for the drugs have varied vary, and tend to have been greater for drugs that have been in use the longest. This is illustrated in the table below, which shows prices for the medicines which had more than 1000 observations in the dataset. (There were also many observations for fixed dose combinations antiretrovirals such as stavudine + lamivudine + nevirapine, which were only available as generics, and are therefore not included in the table.)

Branded Antiretroviral No FTA FTA % Higher
Abacavir 0.49 0.68 0.39
Didanosine 0.59 1.71 1.90
Efavirenz 0.42 0.68 0.62
Lamivudine 0.11 0.22 1.00
Lopinavir + Ritonavir 0.28 0.49 0.75
Nevirapine 0.22 0.33 0.50
Stavudine 0.02 0.07 2.50
Tenofovir 0.72 0.75 0.04
Zidovudine + Lamivudine 0.35 0.68 0.94
Zidovudine 0.59 4.79 7.12

 

Controlling for other factors

Finally, I compared price levels while controlling for other determinants, namely national income, the therapeutic importance of the drugs in each transaction, time, and the price difference between drugs. I set up a heteroskedaticity-robust Dummy Variable Least Squares regression with the following variables:

  • FTA dummy variable, expected to be positive because prices should still be higher in countries with FTA obligations after the addition of controls.
  • Logged Gross National Income per capita in the country and year in which the transaction occurred, using data from the World Bank. Firms offer lower prices to low income countries, so the coefficient is expected to be negative.
  • Whether the drug was recommended by the World Health Organization for use in resource limited settings during the year of purchase. WHO’s recommendations can be found here (the recommendations have changed over the time period studied, and each of the different versions are available at the linked page). The effect is expected to be positive, because WHO-recommended medicines will be in greater demand.
  • The year is entered numerically. Since most antiretroviral prices have fallen over time, a negative coefficient is expected.
  • Differences between the price of each drug. Most of the variation among the prices in the dataset is due to the fact that it includes drugs that vary widely in price from each other, regardless of the country where the sale takes place. It includes older drugs produced by numerous generic firms (such as stavudine) as well as newer drugs subject to less competition (such as atazanavir). Price differences between sub class of antiretroviral can be quite significant too, as noted by MSF in their Untangling the Web series of price surveys. I use dummy variables to capture these differences.

The negotiated prices of branded antiretroviral are, on average, 57% higher in countries with free trade agreements than they are in other countries. The coefficient is significant at the 99% level (meaning there is a 1% chance the variation between the mean price in the FTA group and mean price in the non-FTA group is merely statistical noise). A summary of the regression results is shown below. For the full results, see the Stata output.

Independent Variable Coefficient Standard Error
FTA In Force 0.574 0.059
Logged GNI per capita 0.055 0.004
WHO-recommended – 0.077 0.019
Year – 0.034 0.003
N = 20,472; Adj. R2 = 0.769; F-statistic= 88.48

 

All of the control variables are significant at the 99% level. Logged GNI per capita is positively correlated with price, and the time trend is negatively correlated, as expected.

The coefficient on the dummy variable indicating whether or not a drug is recommended by the World Health Organization is significant but negative – which is unexpected. This may be because many of the drugs recommended and purchased are older drugs that may be off patent or non-patented in developing countries (such as stavudine, lamivudine, nevirapine, and combinations of these). I run another regression this variable dropped, and the results for the remaining variables are very close to the same. Negotiated prices are 57% higher in countries with FTAs, and the remaining controls still have the expected, significant relationship (see stata output).

The overall fit of the model is good (adjusted R-squared = 0.77, F statistic = 88), though most of the variation is due to heterogeneous differences between the prices of drugs. If you drop the dummy variables for each drug, the adjusted R-squared falls dramatically, though the coefficient on FTA changes little (see stata output).

Discussion

This blog is meant to quickly illustrate that countries with free trade agreements pay more for these medicines. It simply reports differences between prices in countries with and without TRIPS-Plus obligations from trade agreements with the United States. This should not be surprising. Intellectual property is a temporary monopoly granted to inventors by the state in order to reward inventors with monopoly rents, and this means higher prices for consumers.

Nonetheless, staffers from the U.S. Trade Representative’s office have reportedly downplayed the relevance of trade agreements with TRIPS-Plus IP obligations on drug prices when privately discussing the Trans Pacific Partnership with Members of Congress. Policymakers should understand that, of course, stronger IP for medicines is associated higher prices for medicines.

There is a lot of room for a deeper dive into the underlying causes of these price differences.

A country that has entered into a trade agreement with TRIPS-Plus obligations may or may not have fully implemented them. Countries that have not signed a trade agreement may have been otherwise persuaded to enact TRIPS-Plus intellectual property protections (i.e. – through Special 301 pressures or in-country lobbying by multinationals). Furthermore, there are different TRIPS-Plus intellectual property rules that countries have adopted, and some may have more of an effect on prices than others. It would be helpful to have a large set of data showing which countries implemented which TRIPS-Plus provisions in their intellectual property law and at what particular time. As far as I know, there is no such review – but if you know of one, please point me to it in the comments or offline.

For a good review of studies which have examined the impact of particular TRIPS-Plus provisions in particular countries, see The Effects of TRIPS-Plus IP Provisions on Access to Affordable Medicines, by Jennifer Reid. It includes abstracts and excerpts from studies showing that:

  • Secondary patents have led to “effective patent term extensions” in Chile delaying generic entry (Abud, et. al., 2015); have raised the public sector price of the important HIV/AIDS drug Abacavir in Malaysia to 8 times the price paid in neighboring countries (MSF, 2014); and have added add 6.3 to 7.4 years to the effective patent life of drugs in the U.S. (Kapczynski et. al., 2012).
  • Data exclusivity requirements have led to higher prices and $396 million additional expenses for Colombia’s public health system (Cortés, et. al., 2012); have blocked generic versions of off-patent medicines from the Guatemalan market (Shaffer and Brenner, 2009); and have delayed the introduction of cheaper generics into the Jordanian market in 79% of medicines (Malpani, 2009).
  • In the U.S. one particular off-patent drug had its price raised from nine cents to $4.85 per pill after data exclusivity was applied (Kesselheim and Solomon, 2010).
  • “Linkage” between patent authorities and health regulators in Canada has been extending the patent life of “weakly inventive products” in Canada, keeping generics off the market (Bouchard, 2010).
  • Stricter rules regarding the seizures of goods in transit led European customs officials to seize numerous shipments of generic drugs being sent from India to Brazil, even though the medicines were legal in both the point of origin and the destination, which led to a “disruption in the supply chain of legal generic drugs.” (Zarocostas, 2010).

The regression described above shows the average difference in prices of each drug between countries, rather than the difference in prices between drugs. There has been other work on the sources of drug price differences, such as Meiners et. al. (2011), Waning et al (2009) and Chien (2007). In general, the patent status, the presence of competing drugs in therapeutic classes, and bulk purchasing have been found to affect price.

As noted above, the data used here is for antiretroviral prices negotiated by large institutional purchasers. Other sources of data may provide prices closer to a “free market” price. Health Action International conducts surveys of over-the-counter pharmacy prices that are publicly accessible. IMS Health sells international price data comparable across time and countries. These and other data sources will be good to utilize in future price comparisons.