 So I'm very excited to be here and have the chance to present this paper which I wrote with my former PhD classmate Karthik Sastry inappropriate technology evidence from global agriculture So the starting point is that global research and development innovation is highly concentrated in a handful of countries To give you one example by one estimate 25 percent of total Investment happens in just the US as compared to 3.6 percent in all of Africa and South Asia combined however In growth economics there are two highly Contrasting interpretations of the impact of these vast disparities in research investment on disparities in productivity One perspective suggests that in general ideas are broadly applicable and can spread around the world from innovating countries So even if innovation only happens in a handful of predominantly rich countries technological progress in those rich countries reduce global disparities in the long run as technologies diffuse a Second contrasting interpretation However is that technology is in general highly context specific designed to match the specific characteristics and conditions of the places that develop it And as a result is often inappropriate in places dissimilar from those frontier innovative countries in this perspective technological progress in the frontier in the specific focus of innovation can actually generate and underlie persistent disparities in productivity around the world Now the second Inappropriate technology hypothesis has been in the literature for many decades getting back to the 60s and 70s our understanding of its global incidents and Quantitative importance is relatively poorly understood So what this paper does is that it tends to investigate each pillar of this inappropriate technology hypothesis Empirically in the context of global agriculture a sector where all of its underlying forces loom potentially particularly large And to give you an example of exactly what we mean I'm gonna start with this story about basically three pests So these three pests are you know might look similar the three dominant threats to corn production Around the world But there are important things that separate them the first two the European Mace born their maze root worm are dominant threats in the US and Europe and in the US respectively as a result They've been the subject of massive amounts of Scientific research R&D investment so much so that the maze root worm is now referred to as the billion dollar bug And one accomplishment of this research is the development of a genetically modified variety that is designed to specifically target These two varieties and make corn able to withstand these threats in the US and Europe Okay, the third pest however is the Mace stock bore It's a dominant threat to corn production in large parts of sub-Saharan Africa by one estimate Decimates 10% of corn production in Kenya each year However, perhaps as a result of where it is and the types of farmers it affects It's not been the subject of much R&D investment and the new technologies developed to combat these first two It turns out don't work as well against the third and it remains a dominant threat to production in large parts of the world So this is the inappropriate technology story in a nutshell But beyond a handful of case studies like this We don't really know the extent to which this type of narrative explains the global diffusion of technology And moreover, we don't know the extent to which stories like this can explain Productivity differences around the world in agriculture. So that's what this paper really tries to do So the first thing we do I'm gonna give you the outline of the whole paper on this one slide As our the main thing we need to measure is Inappropriateness so how do we measure the extent to which technology developed in one environment is appropriate in another and to do that building exactly on that Example I just told you we combine we measure at the crop by country pair level dissimilarity in their local pest and pathogen environment using comprehensive data We've collected on the global distribution of all known crop affecting pests and pathogens bugs viruses Fungi protus etc as well as which specific crops they can attack And we can we combine this with data first on variety development and diffusion to think about technology development and diffusion and second on agricultural production to think about productivity Next we're going to present two main results that frontier technologies Inappropriateness as measured in this, you know dissimilarity of pest and pathogen environment first inhibits international biotechnology transfer and adoption So really affects which places around the world get modern technology and second lowers crop specific output by shifting production away from crops that are For which frontier technology is less appropriate. So reduce the technology diffusion and lowers output and You know thinking about these reduced form estimates combined with a model We argue that this mechanism just focuses at focusing on differences driven by pest and pathogen dissimilarity You can explain about 10 to 15 percent of productivity gaps around the world and moreover are important for understanding and interpreting ongoing Changes in the geography of R&D and environmental change around the world So for example it affects how we think about the rise of new technology leaders and the importance of having R&D in different parts of The world and the impact for example of the rise of China and also the impact of climate change which is going to shift Ecology and geography in different parts of the world and as a result which technology works. Well, where? Okay So the first thing I'm going to tell you about is the measurement Which the key goal here is exactly to measure this Inappropriateness thing that the whole paper is about so to do that we combine data on crop pest and pathogen Which I'm going to refer to as CPP just because makes it less of a mouthful So we combine CPP level information sheets from this cabbie center for agricultural biosciences international crop pest Compendium and they basically do a literature search of you know sources ranging from the World Bank to the food and agriculture Organizations of the USDA to try to be as comprehensive as possible And this has really become the gold standard for measurement in the biological sciences of where these crop pest and pathogens are And we use two key pieces of information that I'm showing you in this map The first is the global distribution of each one of these roughly 5,000 CPP so this is the African maize stock bore that was in the photos at the beginning and it's showing you which countries it It lives in The second key pieces of information is which crops each of these things can affect so we also know Which of any of the 132 host plants each each pest and pathogen can attack So for example for the maize stock bore that's maize, sorghum, rice, sugar, cane, etc And the key piece of information that we get from all this is the identity of all CPPs in a given country or in some cases region within a country That damages a particular crop Using that we construct what as our main measure of Inappropriateness or technologies in appropriateness at the crop by country pair level which I'm going to refer to as CPP mismatch at the crop by origin by destination level which is one minus the number of CPPs those two environments share have in common Normalized by the number of CPPs in each location to the one-half power so you can think of that as like one minus the correlation between the two and it's turns out as part of a standard class of Divergence measures that that ecologists and population Population biologists use and you know we can change this a number of ways that turns out the functional form doesn't only matter so much And there are other kind of robustness tests we can do to this measure thinking about purging any variation from invasive species Which we have other strategies to measure and also measuring only the variation that emerges directly from like fixed ecological and geographic Characteristic but for the purpose of this talk, I'm going to just talk about this CPP mismatch measure as our main source of Inappropriateness So the first thing you want to verify is this sort of premise of the inappropriate technology hypothesis Which is that research is really focused on these problems that exist in rich countries in this case super concrete on the pests and pathogens that live In rich countries So I'm just going to show you this kind of one way to look at this The first thing to notice that researchers focus on their local problems. So like Let's just showing you that relative to CPPs that exist in other countries Innovators are 17 times more likely to take out a patent related to a crop pest and pathogens in the country in which they work That fact combined with the fact that most R&D happens in a handful of rich countries leads to a major skew in the focus of research Toward these pests and pathogens that live for example in the US as compared to ones that live in not the US and To give you one particularly striking cut of the data over there I'm showing you the number of patents related to each crop pest and pathogens that exist only in Brazil or only in India Or only in the US and the differences are staggering 1 1.9 and then 42 so huge disparities in the focus of research So the question is how does this matter first for technology transfer technology diffusion? Okay, so last piece of data I need to present in this talk before getting to the actual results We need to we want to measure technology diffusion around the world to do that We compile seed certificates, which are basically like patents for seeds for all countries with any protection for for for seed variety So that's the green countries in this map This is we get this data from the Union for the protection of new varieties Which is the international body tasked with Harmonizing this information and standardizing it across countries and the key thing we have in this database Which is actually in this organization's Constitution to compile this information is the unique Identity of a variety that is first protected in one country and then subsequently in another so we can track Individual seed varieties where they're first developed and everywhere else are subsequently protected To do that we can construct our main outcome variable for this Part of the analysis on technology diffusion, which is the number of varieties for a given crop first developed in a country L prime and subsequently transferred to country L in a fixed cross-section since 2000 in the paper We think more about dynamics, but for here now. I'm just going to show you Static results So our first estimating equation is looking at exactly the relationship between Inappropriateness as measured by CPP mismatch and technology transfer at the crop by origin by destination level Conditional on all two-way fixed effects. So country pair fixed effects capturing things like trade and geographic distance and cultural distance and things Like that origin by destination fixed effects capturing like overall R and D at the origin at the sending country And destination by crop fixed effects capturing things like you know market size Characteristics of producers overall levels of education technology use etc. All of that we can absorb in these fixed effects And our hypothesis if this inappropriate technology hypothesis is true is that this beta is going to be negative Okay, that inappropriateness is a barrier to the diffusion of technology and that's exactly what we find across specifications either focusing on the combination of the intensive and extensive margins or looking at them both separately and These results suggest that for the median crop by country pair in our sample this mechanism reduces the amount of technology you get by about 30% But now so this is a sample of all countries and all crops But the story that kind of I've tried to motivate the talk with was really about this kind of Concentration of R&D and a handful of countries and the fact that being dissimilar or being inappropriate Relative to that those handful of innovation intensive countries is particularly costly But to investigate that hypothesis I take the same regression great rest regression include an interaction term between whether the origin country the sending country It was what I'm going to call a frontier innovator for a given crop And I'm just going to define that as one of the countries that develops the highest number of varieties for that crop in the world So if you make the most currencies for a particular crop your country innovates the most currencies I'm going to call you a leader country for corn same for wheat, etc And now if this beta 2 is negative that means that it's particularly costly in terms of the amount of technology You get to be dissimilar ecologically to these leader countries and again, that's exactly what we find Either defining the leader as a top variety developer top two or top three in the top three case It's basically only reduces the amount of technology You get relative to that set of leader countries the rest of the countries essentially don't matter from a statistical point of view So countries and we're country by crop pairs local Producers for which frontier technology is inappropriate really don't get access to modern technology But in the remaining minutes I want to show that that doesn't just matter for access to technology It actually affects production and productivity around the world. So to investigate that we turn to the second estimating equation We're now the outcome variables output for a given crop in a given country I'm on the right hand side our measure of inappropriateness now is that CPV mismatch relative to the frontier countries the one two or three countries I showed you before the results I'm going to consider there being sort of two frontier countries, but you can do it with any number up to three Now conditional just on crop and country fixed effects and some controls that I'm not going to explain in detail I'm happy to answer questions about we really want to make sure that we're purging any variation driven by differences in innate Suitability around the world that might be driving output differences So we have a number of strategies to try to account for any differences in the fact that land differences around the world Just make certain areas more or less productive for certain crops in an innate perspective absent technology And so now beta being negative means that inappropriate as effects not just technology transfer, but also output And that's what we're showing in this table. So for the This negative coefficient means that the inappropriate as a frontier technology reduces substantially the amount to which that crop is produced in the country One way to think about the magnitude is that a one standard deviation Increase in your in appropriate this measure reduces output by about point five standard deviations We do a number of robustness checks I'm again happy to talk about these in question and answer one thing that I think is maybe most interesting here is we can exploit Natural experiments in which the geography of R&D changed around the world where people are doing research The first is the green evolution of the 60s and 70s And the second is the recent kind of massive rise of us biotechnology And they can show that those changes in where research is getting done and hence changes in the relevant notion of Inappropriateness relative to the frontier Leads the changes in where production happens around the world and as a result the pattern of productivity and productivity disparities in different parts of the world Okay So in the remaining two minutes I just want to talk to you a little bit about how we think about the aggregate consequences here and Basically what we do in a nutshell is we combine sort of these exact reduced form estimates with a model that allows us to do two things The first is adjust for the fact that those reduced form estimates combine the fact that Inappropriateness is making each sort of unit of land more or less productive, but also affecting Sort of where you might grow one crop or another So it's affecting the underlying productivity distribution only some of which is kind of comes out in that in that regret or so as a result The reduced form coefficient combines both the direct productivity effects and the selection effects We want to kind of take that seriously and second the fact that prices adjust And so doing that we can think about the aggregate consequences of inappropriate in this here I'm showing you the loss in productivity due to inappropriateness in a histogram That's cotton encoded and we estimate that this mechanism reduces global productivity by about 58% and here I'm showing you the relationship between that country that will last in productivity and productivity today Strongly negatively correlated suggesting that the countries that are losing most as a result of this mechanism Are the ones that are also least productive in the modern cross-section the result of which means that Inappropriateness explains about 14 to 15% of global disparities in productivity around the world so a pretty large share of Differences in productivity and productivity gaps like every one one minute left or okay So the last thing we do is we think about okay, that's kind of a not such a real-world counterfactual What does it mean to remove inappropriate and that's not something we can do with policy? But there are things you can do with policy So one thing we can do with the model is to say well if we really think this mechanism is important It tells us something about where we think research should be targeted to have them Most benefit around the world we can answer questions like that We can also answer, you know what the impact might be of climate change which shifts Where these different crop pest and pathogens can live around the world and hence that network of Inappropriateness across countries and what impact that might have it turns out that might actually coordinate Global research toward a more common set of threats which could mitigate part of the negative direct effects of climate change and finally we can estimate the consequences of the rise of new technology leaders like brick countries Brazil Russia India and China which it Turns out actually span a larger share of global ecology and you know according to our results and model will Sort of reduces inappropriate technology problem by endogenously developing technology that is more appropriate for larger parts of the world So you know in conclusion hopefully I've convinced you that it's inappropriate technology mechanism is important Explains patterns of technology diffusion and productivity around the world I think we're interested in a range of future questions that I'll just leave up here about policy about other sectors about climate change But you know rather than talk about that I'll open it up to questions Thanks