 So, good afternoon. I'm glad that so many people are here despite the good weather, although on second thought maybe it is because of the hot weather and the fact that we have air conditioning here. So, either way, I welcome you to today's seminar, which is on intellectual property rights and access to innovation in the case of pharmaceuticals and the evidence that we have from the implementation of the TRIPS agreement. And by way of background, I think it's quite unambiguous to say that one of the elements of the TRIPS agreement that really brought policy change in the member countries of the WTO was the introduction of patent protection for pharmaceutical products, at least in those countries that previously did not protect pharmaceutical products. And that has prompted many economists to look at what the effect of that policy change has been on pharmaceutical markets, various aspects of the performance of pharmaceutical markets. And also personally, I believe as time goes by, these types of studies will become even more promising because as probably all of you know, the TRIPS agreement came with a number of transition periods and not only the transition periods, but also it takes a long time for pharmaceutical patents that are being filed to have an effect on pharmaceutical markets given the delay between patenting and the introduction of pharmaceutical products. So, as time goes by, these types of studies are becoming ever more promising. Now, we are very fortunate today to have Professor Margaret Keil with us, who is an expert in this field. She's a professor at the Toulouse School of Economics and just by way of background, the Toulouse School of Economics is probably one of the premier schools in Europe, if not in the world, on industrial economics. There are a number of really well-known scholars there and this is probably one of the best places in the world to do research on industrial economics and innovation. And Professor Keil has done a lot of interesting work generally on innovation and in particular on the pharmaceutical industry. She previously was a professor at London Business School, at Duke University and at Carnegie Mellon University and she received her PhD from the Massachusetts Institute of Technology. So, we're very fortunate to have you with us today. We discussed that Margaret would take questions throughout the presentation, but if those questions could be mainly of the clarifying type and the broader questions on the paper, I suggest we leave for the questions and answer period. So, Margaret, welcome. So, thank you very much. It's really a privilege to be here. I'm always happy to have an excuse to visit Geneva and particularly to present this work to people who are really experts on some of the details here. So, I'd like to emphasize that this is preliminary work and I'm here in the hopes that I can learn something from you. If I've missed something important, if I have the detail wrong, please do let me know. That's extremely important. As was just stated, please feel free to interrupt me. I become a bit nervous that everyone has gone to sleep if I don't get questions with some regularity. I promise that I will try to leave time for the end so that we can have a full discussion at the end. And one other detail I should note is that the data that I'm using in this paper was made available to me from Pfizer. So, I did not receive payment directly from Pfizer. I just received the data, but it's important to make this disclosure upfront. And last bit, which is obvious from the title slide, this is joint work with Yi Chen at Northwestern and Huizhi, I'm hoping I say his name right, at University of Illinois at Chicago. Okay, so I'm sure that this is unnecessary for this particular audience, but in general, as we know, there's a trade-off that we face in setting policy for intellectual property rights. There's a need to balance dynamic and static efficiency, or another way of putting that is that we need to balance the incentives for innovation that are created through intellectual property rights from the short-run costs that are associated with the fact that intellectual property rights provide some market power and allow firms potentially to increase prices, which has the unfortunate effect of reducing access to their innovations. I think in the economics profession, there's general agreement that the optimal balance of dynamic and static efficiency probably varies across countries and across industries. In other words, we probably think that the appropriate patent term in pharmaceuticals is not the same as the appropriate, the ideal patent term in something like electronics. But we're dealing with the realities of setting policy. We can't easily set different patent terms for different industries and expect that to be the same over time. I think there's also general agreement among economists that different countries will see it as in their interest to implement intellectual property rights at different times, in particular at different stages of development, and depending on what their own domestic needs are. Historically, in fact, that's been the case that when a country hit a certain level of development and had some domestic firms interested in protecting their innovations, it's at that point that they would implement patent protection. Having said that, as we're all aware, over the last few years or a few decades, there's been a move towards harmonizing intellectual property rights across countries. So even though we might think that the optimal balance varies across countries in practice, if you are compliant with the TRIPS agreement, if you're a WTO member, you're expected to have at least some minimum level of protection from patents, copyrights, trademarks, etc., which is the same everywhere. So there's, of course, still some variation across countries, but some effort to harmonize as well. And I think there's, this is probably not a very controversial statement that over time we've seen a general increase in the duration of patent protection, in the enforcement of patent protection, and in the scope of patents, as in the definition of what can be patented. Now, there have been some important decisions in the U.S. Supreme Court recently that maybe is starting to pare that back a little bit, and that's certainly an area where there's lots of differences across countries, but as a general statement, I think IPRs have become more prominent over the last couple of decades. In the case of pharmaceuticals, this has been a very contentious issue. So I have here some quotes from the two extreme positions, I would say. So one from an NGO, Metzels on Frontières, or Doctors Without Borders, which in general objects to intellectual property rights, seeing them as a barrier to treatment for people in a lot of poorer countries. On the other side, you have, as an example, the U.S. Trade Representative here saying, in fact, stronger IPRs are in the interests of developing countries. The belief is that stronger patent protection will increase the willingness of firms to make their innovations available in poor countries and increase the availability of medicines. So it's been difficult to find some points of agreement between these two extreme positions. There was a lot of controversy in what I'm going to focus on here. The TRIPS agreement was being negotiated, and I think that there remains a lot of controversy over this question. So I'm going to focus in this paper specifically on the TRIPS agreement because this was a change that required many developing countries to make an important adjustment to their patent systems. So in particular, they were required to introduce product patents on pharmaceuticals, which not all of them had prior to that point. And so that was a substantial change in patent law. That was especially important in this particular sector. So I should make clear I'm only looking at one sector, and I wouldn't necessarily expect these results to apply in other sectors. Because this was controversial, because this was something that a lot of developing countries did not see as in their immediate interests, they negotiated a transition period. So TRIPS compliance was not required for most developing countries until 2000, 2005, and for the least developed countries, it's been pushed out until 2016. And those deadlines have been renegotiated at various points. So as I mentioned at the beginning, we face this trade-off between the dynamic effects of intellectual property rights and the static effects. So just a little bit more background on the dynamic effects, there are people who would argue we don't need patents. So I think this is not the mainstream position for most of the economics profession, but it is a position. And I think it's a fair position in certain contexts. So alternative mechanisms to provide incentives for innovations, such as prizes or advanced market commitments may be the most appropriate in certain situations. But the evidence overall is that across all industries, the chemicals industry or pharmaceutical industry is the one that identifies patents as being extremely important for them to make money on innovation. And the reason is that lacking patent protection, ex-post imitation of a successful product is pretty easy in pharmaceuticals relative to a lot of other kinds of products. And there's a large upfront cost of investment that the innovators need to somehow make back. The empirical evidence on the dynamic effects of patents is somewhat mixed. So my co-author has a paper that looked at the introduction of IPRs mainly prior to the trips period. She found that they were associated with an increase in domestic innovation, but this was largely isolated to relatively rich countries or relatively developed countries. In this particular case, that in a very poor country, we should expect that the introduction of patents is going to cause an immediate increase in the number of innovative firms and the number of patent applications coming from those firms in that country. Lanyell and Poeburn in 2001 looked at the effect of the introduction of IPRs in pharmaceuticals in these developing countries on changes in R&D investment. And they found that there wasn't much not when they could pick up as of 2001, which was prior to the adoption of patent rights for most developing countries, product patents at least. They didn't find a big increase in R&D directed at diseases that are most prevalent in developing countries. So I looked at this again using more recent data. So the hope was that we've had some time to elapse now. Maybe there's hope of seeing something. So with a co-author, we looked at drug development efforts at the disease level, and what we were trying to do was exploit the fact that disease burden varies across countries. So malaria is much more prevalent in Africa than it is in the United States, obviously. The disease burden of HIV varies around the world. The disease burden of various cancers varies, et cetera. So we were looking at one country that had a particularly high disease burden in this specific disease. When it changed its patent law, we were not willing to invest in new drugs to treat that disease. So now that there's patent protection in Africa, does that increase the R&D effort directed at malaria? So we found overall, if you look across all diseases, that there was an increase, but that that effect was largely for global diseases, as in the ones that have a market in relatively rich countries. And the effect of introducing IPRs to the income level of a country. So we would see a much bigger effect of introducing patents in a relatively rich developing country than one of the very poor ones. And so the introduction of patents in low-income countries did not seem to have shifted R&D development efforts either for global diseases or for diseases with a high local burden. So when I mentioned that alternative mechanisms for inducing innovation are appropriate in certain contexts, neglected diseases, the historically neglected diseases is an example of such a context. That introducing intellectual property rights is probably not going to be enough to get firms over that fixed cost burden, because even with a patent the price that a firm could charge from malaria treatment in many poor countries is just not going to allow it to cover those fixed costs. So we concluded that the dynamic effects of IPRs in certain countries for developing countries were fairly limited. Now, that's still early, right? We could see that change over time. So first developing countries are developing. So as they get richer and richer the effect of intellectual property rights in those countries may be more pronounced. And, you know, it's still early. So maybe firms are just waiting to see how these markets develop, etc. But at least so far we didn't see much in the way of dynamic effects. Yes. You referred to developing countries. What is the concept that you're using? This is the great majority of the least developed countries. So I'm using the World Bank definition of upper middle income, lower middle income and least developed countries. I'm referring to upper middle income, lower middle income and least developed. So in fact most of my sample is going to be from the middle income group, not the LDCs because they haven't implemented patent protection yet and because I don't have good data on them. So in terms of the static effects economic theory would predict that actually we can solve a lot of the problems associated with the market power derived from IPRs by having price discrimination or differential pricing. So this is a setting in which there's general agreement that for most products the marginal cost of producing a drug are very low relative to the fixed cost of developing a drug. And so long as a firm can cover the marginal cost of producing that extra pill it should be willing to sell at the marginal cost or anything above marginal cost price in any country around the world. In practice we haven't seen that happen. And there are various reasons that firms and others have offered for why we don't see more price discrimination, more differential pricing, more tiered pricing. There's various terms used for it. Some firms would point to the threat of parallel trade. So if we launch at a very low price in let's say Tunisia there's a worry that that product is going to come back and cut into our market in Europe for example. Now that kind of gray market trade is not actually legal but there are it is legal within the European Union and there are other situations where it could be legal. They also will point to the use of external reference pricing or international reference pricing. So if we set a very low price in one market there's a risk that when we go to a developed country and we try and negotiate the price of that same product in this developed country the regulator there will say well I see that you're offering this product at one third that price in this other country. So I don't want to pay more than that. That kind of external reference pricing is widespread within Europe and among some other countries there's a question of what are the countries to which this reference pricing refers but these are the arguments offered for why a firm might not want to price discriminate or use very low prices in poor countries because that's going to have consequences for their ability to set high prices in richer countries. In practice the static welfare losses are also addressed very directly by the use of price controls. So almost all developed countries and now many developing countries regulate drug prices. So if for example in France if you want to sell a drug in France you have to negotiate with a committee, persuade this committee that you have an effective new product argue about what the price should be and the French government because it pays a lot of the cost associated with providing this treatment to the French population is of course interested in keeping the price fairly low. Now these policies can also make launch in a country less attractive or delay access for the same reasons as listed above for international reference pricing and parallel training. Yes. Thank you Margaret. I'm Peter Bayer from the World Health Organization. I'm wondering do you have data on that price controls delaying access to medicines across developed and developing countries? So I do in the context of developed countries. So I wrote a paper about this. I think it was published 2007 looking particularly at developed countries or OECD countries. And countries that use very stringent price controls and countries within Europe that are likely to be sources of parallel exports to the rest of Europe see fewer products launched or products launched with a longer delay. So there is some evidence that that does take place in practice. Now how important that is within the developing world I think there's far less evidence on. So one last point about the static effects in theory. It's not always the case that patents are the only barrier to access. So in many countries the populations may be too poor to even cover marginal cost. So it's not the existence of a patent that's preventing the drug from being launched it's that there still is no market. Particularly in poor countries where you don't have a government providing health insurance or health coverage to its population if people are paying out of pocket the market is just going to be very small for some products in particular. And there are other aspects of access that might be important. So it might be good to have strong infrastructure, a cold chain, distribution channels, complements like diagnostics etc and if you lack all of that then even if you don't have a patent it's not clear that there's this large set of generic firms that are willing to make a product available. In terms of empirical evidence Lanyal a few now several years ago has a paper showing that IPRs were associated with faster launch of new drugs in rich countries but the effect in poor countries was ambiguous. At very low income levels as she pointed out the ability to pay might not have exceeded marginal cost and patents in that case are not the only barrier to access and she also noted the existence of these regulatory spillovers through reference pricing etc. There's another very important paper in the economics literature which looked at a single class of drugs in India so a class of antibiotics prior to the introduction of patents in India so this was a structural model that said this is how competition takes place in this class of antibiotics if we did a counterfactual simulation and said we now introduce patents and remove generic competitors from the market what would we expect to see they found that not surprisingly the patent holder would increase the price which has the negative effect on access welfare etc but there was a follow on effect which is that if one drug now no longer faces generic competition it increases its price the price of competing therapies even if they still have generic competitors will also tend to go up right so in econ speak prices are strategic compliments the price of your competitor goes up your best response is also to increase the price of your product so there was an additional welfare loss associated with that so what we're trying to do in this paper is take advantage of the fact that now some time has elapsed so we can observe after the introduction of IPRs what's actually happened in a lot of these developing countries so we're going to look at the speed of launch the price at which a product is made available and its level of sales we're looking at a large number of companies and across a large number of drugs so we're not looking so this earlier paper was focused on a single class of drugs in a single country so we're going to try and look a bit broader than that to see if that makes any difference and again I want to emphasize that these are preliminary results but what we have so far is that patents do appear to be associated with faster launch not surprisingly patent to drugs also but these the patent premium is not as high in developing countries as it is in developed countries what's most surprising to us so far is that it doesn't appear that the prices of products with post-trips patents so these are the ones that should be the strongest they don't appear to have changed very much so it hasn't been the case that we've seen huge price increases on drugs that received patent protection under the post-trips regimes in developing countries so at least so far there's some evidence that there could be some beneficial static effects for developing countries but there are a number of important caveats that I will try to return to at the end and that I'm sure people here can think of some others so the approach that we're going to take is we're going to look at equilibrium roughly speaking equilibrium outcomes for launch delay prices and quantities with and without IPRs so we're going to exploit the fact that we can compare countries with IPRs to those that do not have IPRs and I should say not just IPRs I should be more specific we're really going to focus on pharmaceutical product patents we can compare the same country with and without IPRs for those that change their patent laws as a result of becoming compliant with the trips agreement and we can compare the same drug and its outcome in different countries depending on whether it has IPR okay so there's a number of challenges associated with doing this in practice the first challenge we face is that when a firm is deciding whether to introduce a product on the market it's factoring in its expected price its expected quantities that it would sell when it makes the launch decision it's also going to be accounting probably for things like external reference pricing or the threat of parallel trade these are simultaneous choices in practice the way that we're going to deal with this is not entirely satisfactory but we're going to say we're going to look at launch conditional on launch occurring we'll look at price conditional on price we'll say what is the quantity sold so it's not a full structural model by any means a more serious challenge for us is that the introduction of patents is an endogenous policy choice so there's a risk that if you just look before and after the introduction of patent protection if you see an increase in price for example you don't necessarily know that increase in price is due just to the adoption of intellectual property rights because a country that adopts intellectual property rights may be getting richer and richer and richer countries have higher prices so if you look before and after you might be misled and finally there are a number of other policies that some countries have probably adopted or have adopted in some cases that we know about and have probably adopted in cases where I'm unaware that specifically try to address some of this potential static costs associated with IPRs so I'll come back to those again at the end and again this is an area where from people here it would be very valuable to me Margaret Yes One question and maybe the question is a bit too early confused about if you wish your counterfactual in a sense that if you ask yourself the question what impact does the trips or might the trips agreement have on prices I would actually argue there should be no impact in a sense that for those drugs that were on the market before trips induced patent protection kicked in those were in some sense up for grabs and you had generic competition and the trips obligations only applied to those patents that were applied for after the trips changes came into effect and those products would then be introduced at a later point in time but those would be completely new products and I wonder what the basis for comparison would be to comparing them to any type of pre-trip scenario Right, so that's a really important point that another problem with just looking at the post trips period is that in fact products that were already on the market would generally not qualify for patent protection if the product was already on the market in India in 2002 it's not like it suddenly gets to get a patent in 2005 when India becomes trips compliant so we wouldn't expect to see a change in price for products that didn't qualify for the post-trips patent. So I have the measure of patent protection at the product level and I'm going to do my best to try and control for the difference between having any patent or having a patent that was granted post-trips So I'm not sure that I'll entirely address all of your concerns but at least I think this is the best I can do to estimate the effect of having a post-trips patent So the endogeneity of IPRs so as I mentioned it's not like patent rights are randomly assigned across countries So historically patent protection, IPRs are things that countries have chosen to adopt when they reached a certain level of development and saw it as in the interest of that country So for example the US was very happy to rip off the more developed countries at the time when it was still an emerging market itself and it's only after there were a lot of US inventors that wanted protection that the US would strengthen IPRs Canada, so I'm going to just not pick on developing countries I want to make the point that developed countries have also used IPRs very strategically in the past. Canada used to use compulsory licensing of pharmaceuticals to benefit its domestic generic industry up until the early 1980s which is when they signed the NAFTA agreement and the US forced a change in that particular policy Having said that by 1995, most developed countries had already voluntarily adopted IPRs Now there was going to be some variation in the strength of that protection and how it was actually implemented but as a general statement countries had pharmaceutical product patents as of 1995 but protection was weaker in most developing countries So as an empirical study the problem with just doing the pre-post comparison between countries that adopted IPRs and looking at how much prices changed in those countries versus countries where there was no change is that the set of countries that changed their IPRs are probably not adopted, right? In general they would adopt patent protection when they saw it as in their interest and so it's not really a natural experiment Now we kind of argue that TRIPS forced the adoption of IPRs in a lot of these countries that did not appear to be very willing to do so initially but nevertheless that's a problem with most historical studies that have looked at the effect of IPRs there's a selection into treatment is a selection into being a treated country So there's one approach that one could take to try to deal with that and that's to use instrumental variables and say well we're going to use WTO required compliance dates it's not a perfect instrument so for those who are immersed in studies of instrumental variables this is not a perfect instrument because the required compliance dates were still correlated with income level and correlated with the incentive that a country would have to adopt IPRs on its own what we're going to do is to focus on this difference between drugs that qualified for a POSTRIPS patent versus those that had a priority date just before they would have qualified for that so we're going to say that the priority date, the initial application date of a drugs patent would determine its selection into being a treated product and that that is exogenous to the policy change so in practice how does that work that means that a drug that had a priority date in 1999 would not have qualified for a product patent in Malaysia or in India if it had a priority date in 2003 it could qualify for a product patent in Malaysia but not India although it could try for a mailbox patent in India and in 2006 at that point India is also supposed to be TRIPS compliant and India in principle should be granting a product patent on that product so we'll run regressions of this form the dependent variable is going to be either launch price conditional on launch or quantity conditional on having launched and setting a price we're going to have a dummy variable for whether the country is a TRIPS treated country as in it changed its pharmaceutical patent law as a result of TRIPS we'll interact that with a dummy variable for the post TRIPS period we'll look at the effect of a drug having a patent interact that with whether this is a TRIPS country and I have a triple interaction between this is a country that changed its patent laws to comply with TRIPS this is a patented drug and this is a post TRIPS patent on that drug so we're basically trying to pull out the fixed effect of being different as a TRIPS treated country and any change in that set of countries that is not associated change in prices etc that might be associated just with a general increase in development etc from the effect of having a pharmaceutical product patent specifically on this product so what we're really going to focus on is that that triple interaction one question your TRIPS country variable is not a time variable? no so we'll have the interaction with whether this is a post TRIPS year or quarter for that TRIPS treated country but we're trying to get at the fact that the set of countries that were required by TRIPS to change their patent laws are going to be a fundamentally different group than the countries that were already compliant by 95 but how do you get at the fact that different countries implemented the TRIPS obligations at different times? the post TRIPS period is going to vary by country and year or quarter and that is according to the deadline of the TRIPS agreement or according to when countries implemented the law? we're going to use the actual compliance date we've run some robustness checks to make sure that it doesn't matter whether we use too much whether we use the actual compliance date or the required compliance date so the data that I'm using comes from IMS Health so they have a database that they call Midas which tracks prices and sales of products across a large set of countries the data comes to us at the package level so the same drug can appear in many different presentations we're going to but the presentations vary a lot across countries so we're going to aggregate that up to the product level and the data is quarterly we have this from 2000 to 2011 there's a total of 60 countries and for the set of products in that data we also have their first launch date anywhere in the world their global launch date and their local launch date so we convert the sales which come to us in local currency so we adjust that for local inflation, convert it to 2011 US dollars just so that we can compare prices across countries we're going to exclude certain classes of drugs for which we didn't have very complete data across all the countries so that's going to include diagnostics, hospital solutions and injectables and we're going to focus on drugs that were first launched after 1990 and in at least two markets the reason for that limitation is the drugs first launched since 1990 are the ones for which patent protection was relevant at some point during this period in at least two markets the reason for that is that China and India in this data China and India both have a large alternative medicine or herbal therapy market of products that probably wouldn't qualify for patent protection anywhere but also are specific to those countries so they're not widely available and it's hard for us to to do any kind of comparison across countries for products like that and then finally we need to match these drugs to patent information so that leaves us with a total of 516 products so we also have country level controls from the World Bank the World Development Indicators data set what is very important for us is how we measure IPRs and I'm sure that people in this audience will have a lot of issues with some of the shortcuts that we've taken we looked at first trips required compliance so according to the WTO when should this country have been compliant with trips some of them complied early, some of them delayed compliance a little bit so we ultimately moved to actual trips compliance and for this Intan was extremely helpful since she had done prior work trying to identify specifically when developing countries had become trips compliant we used data from her and then our patent information at the drug level comes from another IMS data source called patent focus that gives us the number of patent applications at the country level and the type of patent applications at the country level for each drug so that will give us the initial global patent application date which more or less determines eligibility for protection so I'm working under the assumption that whatever the priority year is is going to be the priority year everywhere so I apply for a product patent in 1999 I have to apply for protection in all other countries that I want protection in by 2000 so people here will probably know this but usually this requires a lot of explanation to people who are not down in the weeds of patent data sometimes in the economics literature the difference between the pre-trips and post-trips period is a little bit oversimplified it's not in general the case that compliance with the trips agreement all of a sudden introduced a patent system for most of these countries there was some kind of patent system prior to its perfect compliance with trips but that patent system might not have provided the same level of patent protection so there are patents that exist in countries that were not actually trips compliant in any given year but we can't ignore the fact that there's patent protection of some sort all we can try and do is say there's something fundamentally different about those patents versus those that were granted after trips compliance it's also the case that most drugs have multiple types of patents so the most important is typically considered the product patent but they can continue to apply for process patents on patents for new uses in some case etc so it's not like there's a one to one correspondence between drug and patent it's generally a one to many correspondence there are also cases where drugs get patent extensions because they spent a long time being reviewed at the patent office or supplementary protection certificates sometimes these follow on patents can be effective that's part of the debate in the US and Europe now these follow on patents and whether firms are able to ever green their products as a result of using these follow on patents and there can also be data exclusivity in place even when there's not a patent so for this version of the paper when we say a patent to drug what we mean is there's a product patent enforced in that country in that quarter so which the grant date of the patent and prior to the expiration date okay again just as a bit of background how pharmaceutical firms have used IPRs and well patents in particular varies a lot across product etc so here's just a few examples if you look at Lipitor which was until recently the most the highest selling drug in the world I think this is a Pfizer product they patented it in 87 countries the first patent application was in 1986 they were still applying for patents on Lipitor up through 2011 so obviously those later patents can't provide or shouldn't be providing the same level of protection as that initial patent on the molecule but nevertheless Pfizer sees it as in its interest to continue trying to patent aspects of this product Glevec which is a cancer treatment has been patented in 66 countries again you see a long period after the initial patent application during which follow-on patents are submitted and an HIV medication Norvier patented in 53 countries again a big difference between the last patent application and the initial patent there are a lot of situations where we see a drug without a patent in a country and what's hard for us to figure out is there no patent in that country the innovator didn't think this was going to generate any profits even if they had a patent is it that the innovator didn't bother applying for the patent because they didn't expect it to be granted or didn't think it was worth having a patent because it wouldn't be enforced or because it just wasn't necessary to have it there so in other words the incentive to patent is also going to be varying over time and we don't have a great way of controlling for that other than throwing in country controls etc certainly an issue that we face so this is work in progress there's a bunch of data issues that we haven't yet resolved that we're still trying to make progress on so one issue is that under-trips compulsory licensing of pharmaceutical product patents is permitted under certain circumstances so if there's a public health emergency a country can declare that this patent is going to be not exactly invalidated but that it will be compulsory licensed out we have some information on the compulsory licenses that were actually issued in practice some additional information on when there was a threat made to issue a compulsory license so we'd like to try and control for that now we can measure a threat if it showed up in the report of some sort but even the unstated threat of compulsory licensing is potentially important here and again we'll return to that issue later there are some other changes that have occurred during this time period so in particular there are other kinds of trade agreements that have been signed and some of them require what have been termed trips plus changes so this is something that in particular the US trade representative has been pushing for so these are your stronger data exclusivity terms stronger patent protection in general so we're collecting that kind of information to control for as well so these are preliminary findings so subject to change, these additional controls might change some of our results but I'm going to present the equation by equation triple diff estimates of launch price and quantity where we don't try and directly deal for the endogeneity of IPR adoption using instrumental variables what we are assuming is that after a country is compliant with trips that there's no endogeneity in patent grants okay and this is actually suspect and again I'll come back to that at the end okay so our sample includes a total of 60 countries about half of which are high income countries and the remaining are lower middle income or upper middle income and we don't have any of the least developed countries in this data set unfortunately we don't have great representation from sub-Saharan Africa so that's a limitation so again these results are only subject to the caveat that we don't have everything okay so just this subset of countries the trips compliance deadlines for most high income countries was at the time they joined the WTO in 1995 so they had to be compliant as of January 1st 1996 for most developing countries that was pushed out to 2000 for countries that did not grant pharmaceutical product patents at the time they joined they got an additional extension for that till 2005 and so you can see the distribution there we're going to end in practice we'll use the year of actual trips compliance which has spread out a bit more because some countries chose to adopt trips compliance systems early on some delayed a little bit to the extent that we can that we can capture that well one issue is maybe officially they're granting pharmaceutical product patents but are they enforcing I don't have a great way of picking up it's a matter of law the law provides for pharmaceutical so it's a matter of law so that's what we have has the law actually been changed okay so the data set is going to end up being structured as a drug country quarter observation so when we look at launch we're going to consider a drug at risk for launch in any quarter after it's been launched somewhere in the world up until the point where it's launched in the specific country in this quarter and at that point it drops out of the data so because a lot of launch does not occur there's going to be a very high number of country of drug country quarter observations with launch as being a very rare event so it's not that it was I mean I'll show you a probably more useful measure of of a drug's launch in a couple of slides but the way that the data set is structured launch has a mean of only .01 so I point that out because when we look at the results from the regressions the coefficients are going to look very small but their marginal changes off a very small number off a very small dependent variable the trips country dummy is equal to one for about 61% of the observations here in part that's because if you are a trips treated country and launch takes longer those drug country quarter observations are going to be in the data set longer so I don't want to go into too many of the details it's just to explain how the data set is structured when we get to estimating price or quantity now we only have an observation after the drug has been launched in a country and then we have it until 2011 so the drug will end up with a price observation in a drug country quarter after it's been launched and stay there until 2011 and so that's why now the trips country dummy observations are a smaller fraction of the data okay so before we get into the regressions just some simple statistics on launches etc so if we look at the set of drugs that were launched sometime after 1990 and break it down by income group you can see that high income OACD countries get about half of that population of drugs at some point during our time period here they have about 3.7 patents okay now some of those will expire during that time period but at some point during that time period they had 3.7 patents on average and there's a very small rate of what I call illegal generics illegal in that I don't really know whether they're fully legal or not but generic competition in the presence of a product patent that is still active in that country for poor countries as you go down the income level you get lower levels of launch so as in there's only 27% of this set of new drugs drugs that were new at some point during the 1990 to 2011 period only 27% of them make it to low income countries low middle income countries and there are fewer patents in poor countries in part because they might not have been granting patents in part because firms might not have seen it as worthwhile to even bother applying for patents if you look at just the effect of having a product patent you see that having a product patent is slightly more likely to make a product available in a country but not that much more product patents are also linked with a higher number of other patents on this particular drug if you look at how long it takes for a drug to be made available in a country we broke this down by how long it takes the originator, the patent holder to make a product available the originator or somebody to whom it is licensed versus generics so again you see originators launch in rich countries first and poor countries last so that's not particularly surprising I think they should be willing to launch at any price that covers marginal cost and there's a patent clock ticking they should want to launch as quickly as possible everywhere but there are longer delays getting to lower middle income markets if you look at how long it takes generics to get into the market you can see in high income countries where there's more likely to be a product patent there's more likely to be other kinds of patents protecting a drug it takes generics about seven and a half years to get in patents are 20 years but the drug development process is very long so it is not unusual for a drug to spend 10 years in clinical trials before it gets released anywhere in the world so the period of patent protection for pharmaceuticals effectively if you look at from the initial application date it's 20 years plus maybe a little bit of extensions but in practice they're not getting 20 years of product patent protection on the market generics get to market a little bit quicker in lower middle income countries or poorer countries more generally again there's fewer patents acting as a barrier to generic entry but there's still a pretty long delay even when there is no patent if we look at the subset of products that were first launched sometime after 2000 so now we see that launch is actually even more widespread among rich countries it hasn't changed that much for the lower middle middle income group and patenting seems to have picked up a little bit again you see a higher number of patents per drug now it's up to 4.2 instead of 3.7 in relatively rich countries so there was a report a couple of years ago from DG Comp on the use of patent thickets in pharmaceuticals within the European Union there's some worry that patents are acting patent thickets are acting as a barrier to entry for generics it is the case that the number of patents per product has been increasing in rich countries for how quickly that set of products becomes available again launch is becoming quicker in the high income countries it has gotten quicker in the lower middle income countries too although there's still 2 or 3 years delay compared with rich countries looking at the years to generic entry it's somewhat misleading because there's not that many products that were launched after 2000 for which patents have expired so that generics can get into the market so there's not much information in that first column okay so if I finally get to regression results so with this first set what we're using as our definition of patent for the purpose of this patented drug dummy and the interactions what we're looking at is the existence of an active product patent so in this set of regressions we have year fixed effects and drug fixed effects but not country fixed effects so the trips country is picking up the difference between the set of countries that change their patent law as a result of becoming compliant with trips versus everybody else which is roughly going to be the high income versus developing country breakdown so there what you can see is that actually the trips country dummy doesn't seem to have a very big effect for launch but and actually is associated with slightly lower slower launch post trips patented drugs are associated with quicker launch but there's not much going on there with the interactions so no big changes post trips and no big changes comparing developed countries to trips treated countries with price not surprisingly trips treated countries have lower prices post trips there is a positive but statistically insignificant coefficient on that first interaction patented drugs have about 20% higher prices on average and again the interactions with trips country and patented drugs and the triple diff there are generally insignificant quantities are lower in these trips treated countries patented drugs sell in higher quantities which is a little bit surprising and slightly higher quantities in the post trips period for trips countries now again that's not that additional quantity is just picking up probably the fact that these trips treated countries are growing and the markets for all drugs are getting bigger that's not the effect of having a patent on this specific drug that would be picked up by that last triple interaction if you include country fixed effects so now we're not going to be able to separately identify the trips country dummy because that's going to be the same for any given country across all time periods so we'll just focus on these interactions launch seems to have gotten slightly slower for all drugs patented drugs again are associated with faster launch the interactions are insignificant for launch for price again prices are slightly lower in trips countries patented drugs are more expensive but now if you look at this last interaction the set of products that have qualified for post trips patents in trips countries actually have slightly lower prices in order to get the net effect you need to add in the patent dummy and all of these interactions so you'll end up with about 3% higher prices as compared with if it's a pre-trips patented drug it would be almost 9% so that was very surprising to us anyway now a lot of that is being driven off a fairly small set of products because in order for a product to qualify for a post trips patent in a lot of developing countries it means that it has to have a patent an initial patent application date after 2000 or after 2005 for some of these countries after 2005 we have data through 2011 that would be a very short period of clinical development before it gets launched so there's just not that many products that have a post 2005 application date that are on the market as of 2011 so that's an important caveat here we have a lot of products to identify this extra difference for the countries that became compliant in the later years for the ones that changed over in 2000 we have more evidence we have more products to use if instead of just using product patents I look at the existence of any patent and I've also run regressions where we look at the number of patents on this particular product in this country in this quarter to what I shared you before so I won't go through the coefficients in detail, product patents are a little bit more important which isn't too surprising I mean that's really the important patent you don't see huge differences with these other sets the summary of what we found so far it looks like patents overall increase the speed of launch launch incentives probably matter more for originators in general than they do for generics now that's not that is less true for these trips treated countries but patents overall are associated with faster launch which is consistent with some of the earlier work that's out there patents also are associated with higher prices what was surprising to us is that post trips that price premium is actually a little bit smaller conditional on launch and price patented products sell in lower quantities in trips treated countries which is also somewhat surprising to me so I thought once the originator has a patent in this trips treated country and it's not worried about sharing the market with generics it should have incentives to make country specific investments market the drug, invest in infrastructure and really push out the demand curve so I expected actually to see higher quantity sold which I'm not seeing okay so then for us really the challenge is what do we make of this so what's the right way to interpret these results so I think one point to make is that the lack of a patent doesn't guarantee entry by generics in most countries so that's not as evident from the regressions but if you look at some of the appendix tables where we break down whether the originator was the first to market or a generic was first to market India other countries that have big generic sectors will see generic entry prior to an originator entry if there's no patent but for the most part generics follow originators so even when there's no patent generic firms are going to follow the originator and maybe the reason is that originators are in a better position to provide the regulatory information about safety and efficacy than a generic would be it depends on whether local regulators allow a public applicant to rely on the regulatory dossier in other countries or if they insist on seeing specific clinical trials we also find that patents are associated with higher prices but a lower premium in developing countries after trips and we're wondering if that's because all of a sudden pharmaceutical firms have changed their policies on differential pricing or if they're doing more voluntary licensing which is lowering prices in some of these developing countries or if that's a response to other policies as in a country puts in patent rights or product patents but at the same time they adopt stricter price controls to knock the price down so even though you now have a patent you can prevent generic competition it's not like you can charge the monopoly price that you might want you're going to be negotiating with the government and if the government puts in stricter price controls that price is going to be a little bit lower it could be the threat of compulsory licensing the relationship between patents and tonal quantity sold is again less clear are there policies that favor generics is it lack of marketing effort by originators we don't really have a good sense of that yet again our results are preliminary and there's lots of factors that we haven't controlled for that we would like to be able to address more directly in particular when I say the effective IPRs on price and access etc it's not really the effective IPRs it's the effect of trips as implemented so far so trips plus all other associated policies so it could be that in the absence of these other policy changes which I'll start going through in a second we would have seen bigger effects on price or bigger effects on launch but at least anecdotally I know that there have been a number of other policies in developing countries that are largely a response to concerns about reduced access as a result of IPRs so in some cases that's changing patent office rules in some cases that's threatening compulsory licensing and in some cases it's price controls and I should note also that we haven't included here the potential interaction with NGO activities or political pressures on price a lot of that has been focused mainly on HIV malaria and tuberculosis but that could be spilling over to how pharmaceutical firms are thinking about pricing their other products in some of these countries so to go through those in a little bit more detail different standards of patentability so there was a decision rather recently in India on a Novartis patent on this cancer drug Gleevec the first application for this drug I should note was prior to India's compliance with official compliance with trips but essentially what Novartis did was they had an initial patent and then they applied for another patent on a slightly different version of the same chemical and I think they had improved things like the manufacturing process or the cost of actually producing the product etc and they received that extra patent in a lot of other countries India rejected it and this was something that Novartis fought for for many years and finally there was the Supreme Court decision on that a couple months ago that's not the only situation for which that's occurred so there was also a case of an HIV drug for which the Indian Patent Office rejected an additional patent application on I think it was the heat stable version of this particular HIV drug and the argument that the Indian Patent Office has used is that there's just not an inventive step there you don't qualify for an extra patent okay so that's a different patent office policy or rule than what for example the US Patent Office has used compulsory licensing so this again this is the threat of compulsory licensing might be really important in practice even if you don't observe that many compulsory licenses issued there's a recent paper that has looked at compulsory licensing you see I would say that's not huge numbers of compulsory licenses in terms of the total number of drugs for which a compulsory license might be issued not necessarily restricted just to developing countries so high income countries occasionally discuss compulsory licensing as well so for example the US when there was the anthrax scare in 2001 there was a single treatment for anthrax which was a drug made by Bayer and there were several senators were very concerned about having a monopolist owning the rights to this treatment for anthrax and so it turned out that anthrax was a big outbreak but there were US senators who were saying we need to look into compulsory licensing of this drug and the same thing has happened with flu treatments at various points it's not just developing countries who decided to threaten this every once in a while political pressures on the pharmaceutical industry I think have substantially increased over recent years so in particular after with the HIV the AIDS epidemic in Africa I think there were some studies in the 1990s that showed that prices of some of these products were actually higher in poor countries than in rich countries higher than the US price and that was horrifying to a lot of people out there and there's been a lot of pressure put on drug companies since then to make their products more affordable in poor countries so this is just an example of an index out there that's monitoring these various companies and scoring them on how well they do on these various dimensions of making their products accessible in poor countries so Gilead here is at the top on pricing Gilead actually has a policy of issuing voluntary licenses for some of its HIV treatments in developing countries but it could be that this change in pricing that we observe on these trips is a response to these kinds of political pressures so yeah now maybe they have a patent but it's not like they're able to exploit the patent as much as maybe they once could have and then there's also the possibility that price controls have been introduced so this is a news article saying announcing that price controls reduce the price of medicines entering the Brazilian market we would wonder what kind of price control policy that was but anyway to the extent that these price controls are introduced in response to a concern about higher prices associated with patents that's going to counteract the market power that originators have now it still could be profitable for them to have this patent and prevent generic competition they still get 100% share of whatever they make available but they may not be able to charge prices that are as high as they would like there's also the possibility that firms have changed their pricing strategies optimally that they've decided differential pricing even not in response to political pressure but just differential pricing makes sense from a profit maximization standpoint so this is an example of a recent announcement on cancer vaccines and of course they're not going to lose the opportunity to make this public that they're cutting prices in poor countries but to the extent that pharmaceutical firms are now less concerned about parallel trade or less concerned about international reference pricing because they've successfully persuaded governments in developed countries that they should not be referencing the price in Brazil or should not be referencing the price in South Africa and that makes them more comfortable making products available at very low prices in those countries that could also be an explanation for what we observe now obviously I'd like to be able to say more about that right now I'm giving you a series of anecdotes what I'd really like to do is say well it's definitely this story so one last one voluntary licenses so this is another possibility it could be that the originator decides they don't really want to mess with this developing country they don't bother figuring out how to market there and making the big investments but they issue voluntary licenses so that there is effectively generic competition in some of these countries Gilead is still making some money they're still getting a license fee but the product is perhaps more accessible because of this competition between the generic firms okay so let me wrap up so that we still have time for discussion these policy questions around IPRs are obviously of extremely high interest not just in this auditorium but in general particularly as we see these bilateral agreements being negotiated which often include these trips plus obligations we should have a good understanding of what exactly that's going to mean it's also important for the debates on the appropriate use of compulsory licenses or the appropriate use of price controls so should we be more precise about under what circumstances compulsory licensing is acceptable should it be used more broadly a lot of pharmaceutical firms complain about the use of price controls in developed countries in addition to developing countries that's a huge issue in the US right now which doesn't formally have price controls but once the government starts picking up a larger share of the pharmaceutical tab maybe there's going to be a move in that direction and we think that a lot of the debate here ends up being very polarized in practice I don't think that there's such a clear cut story so at least from what we've been able to find so far we do find that IPRs are associated with higher prices but maybe it's not as bad as was feared for the post-trips period IPRs are also associated with faster launch maybe the effect of IPRs on access isn't as bad as a lot of people were worried about now I should say that I mentioned an earlier paper that I had on the dynamic effects and I found that putting in patent protection in developing countries did not shift R&D incentives very substantially well this is actually consistent with that I would be most worried if we found no change in R&D but we found huge increases in price that would be the worst of all worlds if prices haven't changed that much and profits haven't changed that much then it's not surprising that R&D also hasn't adjusted this is basically consistent with implementing IPRs in a lot of these developing countries hasn't shifted profits that much and therefore hasn't shifted R&D incentives hasn't shifted prices that much now again I want to emphasize that in particular for the second statement on prices this is very preliminary work so I think there's a lot of these important controls that we need to have better information on so voluntary licenses compulsory licenses better information on the use of price controls better understanding of what's going on at these different patent offices we think that probably there's also a very different story in different countries IPRs might be hugely important in certain countries and irrelevant in others so for example in India putting in patent protection and shutting down generic production of a lot of drugs I mean they're not going to shut down generic production of stuff that's already being produced but it's going to certainly limit the size of the market for generic producers for some time there IPRs are going to be really important but in other countries without that generic sector maybe the introduction of IPRs doesn't matter as much so the optimal extent level of IPRs as I mentioned at the beginning is probably going to vary across countries and their own policy choices are going to reflect those different interests so I'm happy to move to discussion now