One of the great benefits of scientific inquiry is the capacity for presumptions and prejudices from experience to be challenged and explained in ways that are counter-intuitive. This is true even for disciplines like economics, where the complexity of the variables involved (and the inability for most studies to be experimentally manipulated for pragmatic or ethical reasons) increase the necessity to avoid "just so" explanations for the observed data. Recently a paper by Cockburn, Lanjouw, and Schankerman, "Patent and the Global Diffusion of New Drugs," American Economic Review 106: 136-64 (2016), challenges the conventional wisdom that patents impede and price regulations promote access to new drugs around the world and particularly in developing countries.
The study evaluated the time to launch of 642 new drugs in 76 countries between 1983-2002. The data in broad strokes showed that both patent protection and price regulations "strongly affect" how quickly new drugs become commercially available. But contrary to expectations, the data showed that launch is delayed by price regulation, whereas "[l]onger and more extensive patent rights" accelerate new drug launch. In addition to these results, the authors noted that other factors (health policies, demographics, economic factors) that make drug markets more profitable also speed up introduction of new drugs, using data controlled for policy regimes country-by-country and otherwise.
The authors recognize the competing interests and factors: on the one hand, providing adequate incentives for new drug development in the first place, and on the other hand having those drugs priced to be "affordable" for a sufficient percentage of the citizens who need or could benefit from the drug to avoid political consequences. The authors also recognize that governments encourage the first with patents and the second with price controls.
Much of the literature on these competing interests has been focused on the GATT/TRIPS agreement and its semi-compulsory harmonization of patent protection globally, particularly in developing countries. There, the "tradeoff" seems particularly inequitable, because in addition to the "deadweight loss" of higher prices due to patents (these are economists, after all) there is the prospect of significant portions of such populations being unable to afford access to the drugs. Some have also argued that the benefits of patenting in these countries is far less, and thus the risks to innovation of price regulation are sufficiently low as to minimize the need for patenting.
These authors looked at another metric: the effects of patents and price controls on the rate of diffusion (i.e., new drug introduction) in countries outside the U.S. This is because, according to the authors, "[t]he public health benefits of new drugs depend, first, on how quickly (if at all) drugs are launched in different countries and, second, on how widely they are adopted within a country, once launched." The calculus they used ignores the R&D costs of the drugs, which by definition are "sunk costs" by the time the drug begins to "diffuse" into countries other than where it was developed. Even so, drug makers are sensitive to the balance between anticipated profits and costs of launching in each country. Such costs include those incurred to obtain regulatory approval (such as clinical trials) as well as setting up distribution networks, "detailing," and reimbursement mechanisms. These are country-by-country costs, and thus the authors could meaningfully investigate how different patent and price control regimes affected drug launch in each country.
Traditional economic theory assumes these costs are negligible and thus introduction in each country depends on idiosyncratic variables in the demand for each drug in each country. These authors posit that the costs are not negligible, and that the policies in each country and their effect on expected profits matter in determining whether and how quickly new drugs are introduced into each country.
The countries chosen for study "span all levels of economic development and exhibit a wide variety of patent regimes." The study also differentiates between process patents (protecting methods for manufacturing drugs, which the authors characterize as "weak" and use India as an example of a country that permitted only process patents as a way to promote domestic competitive entry, i.e., generics) and product patents, which they say "confer stronger rights . . . and allow[] for more effective appropriation of rents."
They discuss four principal empirical findings:
• There is limited scope and slow pace of global diffusion of new drugs (with there being long (> 10 years) lags in many countries and many drugs never being launched outside a small number of developed countries).
• Patent policies adopted by different governments "strongly affect" the rate of diffusion, with "[l]onger, and stronger, patent protection" increasing the rate of adoption (reducing the lag by 55% when comparing countries with no product patents with those with "long" product patent terms). Interestingly, which process patents also have an effect they report to have a much lower impact, and product patents having "short" terms have no effect at decreasing the lag in drug introduction rates.
• Price controls have the opposite effect, with countries having "strong" controls having "significantly longer launch lags" (i.e., increasing by about 25% on average and, in some circumstances, by as much as 80%).
• New drugs are launched more quickly in countries having institutions that promote drug availability and distribution (citing adoption of the WHO's Essential Drug List as one factor).
There was also an effect of market size, consistent with earlier studies showing, inter alia, that "market size is associated with greater pharmaceutical innovation and nongeneric entry."
The econometric analysis presented in the paper is beyond the scope of this discussion here; to illustrate, the authors assert that:
The probability that the drug is launched in country j at time t, given it has not been launched before (the hazard rate of launch), is
h(t|Zijt) = Pr(ωijt > ω*ijt| ωij1 < ωij1, . . . , ωijt-1 < ω*ijt-1)
= Pr(ωijt > ω*ijt| ωij, t-1 < ω*ijt-1)
where Zijt ≡ (xij,Tj,t,σij,ηi,μj) is assumed known to the firm, and the second equality follows from the AR(1) assumption on ω. This implies that the hazard rate is a decreasing function of factors that raise the threshold ω*.
'Nuff said.
The dataset used was based on the entry date of the drugs in each country, branded or generic, as compiled by IMS Health Inc., to avoid the vagaries in public health and regulatory records in each country. For India (which was not monitored by IMS during the relevant time period), data from an Indian market research company (ORG/MARG) was used. The 642 drugs chosen were based, in part, on the authors' ability to (relatively) unambiguously identify the drugs in each country based on generic/non-proprietary naming conventions that they recognize could vary. Also, all forms of the drugs (salts, ester, etc. of a given "active moiety") were treated as equivalents.
The patent regime in each country was characterized by four qualities: "duration of patent term, coverage of pharmaceutical products, coverage of chemical manufacturing processes, and an index of the strength of patent protection that reflects the degree to which patent law provisions favor patent holders versus potential infringers," forming a "propatent index" that ranged from zero to 1. Term was further split into "short" = 0-12 years, "medium" = 13-17 years, and "long" = > 18 years; assuming delay between patent protect and market approval is on average 10 years, the "short" category was equivalent to no patent protection in a country. For product patents, the distribution of these different terms in he countries studied were 6.4, 16.5, and 58.2 percent, and for process patents they were 10.8, 22,3, and 60 percent. Similarly, although the authors acknowledge the variety and complexity of different price control schemes in the countries studied, they separated the countries into ones with "no," "some," and "extensive" price controls, where an extensive price control schema was one where "most or all drugs are regulated." Arranged this way, 22% of the countries had no price controls, 31% had some, and 47% had extensive price controls.
Finally, the "other" variables considered include 1) "whether a country had adopted a national formulary" (65% of all countries studied); 2) "whether a country had adopted the Essential Drug List (EDL) promulgated by the World Health Organization" (41%); and 3) "whether a country has a formal 'national drug policy,' i.e., an effort to coordinate industrial policy and domestic regulation to promote access to safe and effective pharmaceuticals" (63%).
Overall, these researchers found drug distribution to be slow:
In the entire sample of new drugs, 39 percent were launched in ten or fewer countries during the sample period, and only 41 percent were launched in more than 25 countries. The mean number of countries experiencing launch is 22.4 (median of 18) out of a possible 76. Even among the wealthier countries with the most developed health care systems, not all drugs became available during the sample period: e.g., only about 60 percent were launched in the United States, Germany, or the United Kingdom.
And they acknowledged that there were a variety of factors other than patent and drug price policies that can affect these results. But they found that "higher quality" drugs (ones subject to FDA approval) showed the widest extent of diffusion, with more than half being launched in more than 25 countries.
These authors report three "main" findings from their data:
• "First, pooling over all drugs and countries, even after ten years only 41 percent of drug-country opportunities for a launch were taken up. Even after 20 years or more, less than 50 percent of possible launches had taken place, and as a practical matter, many of these drugs may never be launched in large numbers of countries."
• "Second, the pace and extent of diffusion is strongly associated with a country's patent and price regulation regimes."
• Third, "launch delays are strongly related to market size."
The remainder of the paper provides and explanation of how the data were analyzed and validated, which is also outside the scope of this discussion.
The authors conclude by noting that, while their data show that strong patent protection and no to little price controls increase the rate of new drug diffusion:
[T]he same policies that promote faster launch—stronger patent rights and the absence of price regulation—are also those that raise prices. This highlights the basic trade-off countries face between making new drug therapies available and making them affordable. Finding ways to mitigate the adverse effects of this trade-off remains a major challenge. One possible approach would be to introduce multi-lateral recognition of drug trials and regulatory approval, lowering launch costs and speeding up global drug diffusion. Finally, our findings highlight the broader point, not limited to pharmaceuticals, that patent rights can have an important impact on the diffusion of new innovations as well as on the rate at which new innovations are created.
These conclusions underscore the significance of the message that can be drawn from the data: things aren't always as they seem. Such findings are important if policymakers are to address disparities in access to medicines for the poor in countries developed and developing, while not causing the contrary effect of having those policies impede rather than promote these lofty policy goals by being based on assumptions that are contrary to the reality that these results illuminate.
