Technology development is highly concentrated in a small set of countries, often referred to in economics as “the frontier.” For example, the United States accounts for 25% of global research investment, while all of Africa and South Asia combined account for just 3.6%.
There are two broad views in the field of economics about the impact of these vast disparities on global inequality.
The first view is that technological progress is, in general, internationally transmittable and broadly applicable. As a result, the diffusion of new technology from the frontier reduces inequality and “lifts all boats,” especially in the long run.
The second contrasting view is that new technology is finely tuned to the specific characteristics and conditions of the high-income, research intensive countries that develop it. Innovators’ focus on developing locally appropriate technology severely inhibits its usefulness elsewhere in the world. Technological progress in the frontier, in this framework, is a major source of global inequality because it drastically increases productivity in some places while having little or no impact in others.
This second framework is often referred to as the “inappropriate technology hypothesis,” and while it has been a prominent idea in economic theory since the 1960s, we have a limited understanding of its real-world relevance or quantitative importance. Part of the challenge has been simply measuring for which contexts a given technology or set of technological advances is “appropriate” and for which contexts it is “inappropriate.”
In ongoing work, Karthik Sastry (MIT) and I empirically investigate the inappropriate technology hypothesis in the context of global agriculture. One outcome of this research— in the spirit of visualizing climate and loss—is that we are able to precisely map the parts of the world that are most lacking in appropriate agricultural technology because they are most ignored by global innovators (Figure 1, above).
Below we will explain in detail what determines the differentiation and shading in the map. For now, think of the map as indicating the “winners” and the “losers” in the current innovation system—light blue regions benefit from technological progress in agriculture while dark blue regions are systematically left behind.
The underlying mechanisms of the inappropriate technology hypothesis loom particularly large in the context of agriculture. First, there are massive global productivity differences, larger even than in manufacturing. Second, agricultural R&D is even more concentrated in high-income countries than R&D as a whole, with roughly 50% of private investment taking place in just the United States.
Finally, technology is often highly context-specific, in ways that are precise and measurable. In particular, a large share of modern agricultural research is focused on developing resistance to specific pests and pathogens; however, the distribution of these pests and pathogens across countries can be very different, leading researchers to focus substantially more on some than others.
Consider three major threats to corn production, displayed in Figure 2 below. The first and second pests—the European Maize Borer and Maize Rootworm—are dominant threats in the US and Western Europe, and as a result they are the subject of substantial research investment. Genetic modification for corn is focused explicitly on eliminating these threats. The rightmost pest, however—the Maize Stalk Borer—is a dominant threat to maize in many countries in sub-Saharan Africa, but not in Europe or the US. It is not the focus of research investment or targeted genetic modification research. Using patent data, we are able to identify 5,334 biotechnology patents directly related to the first two pests, and just five related to the third. Farmers in areas where the Maize Stalk Borer lives are left without technology to combat it.
We document in our research that these technology development patterns are systematic—innovation is disproportionately directed towards pest and pathogen threats in high-income countries, particularly the US, and tends to ignore threats in tropical and low-income contexts. To measure the global distribution of all pests and pathogens, as well as the specific crops that they affect, we rely on systematic pest and pathogen-level reports compiled by the Centre for Agriculture and Bioscience International (CABI) in its Crop Pest Compendium.
We then show that modern biotechnology for particular crops is less likely to diffuse to places with greater pest and pathogen mismatch from research intensive countries, and that these places are systematically less productive as a result. We measure pest and pathogen mismatch between all pairs of countries and for all crops using our data on (1) the full set of pests and pathogens in each country, (2) the specific set of crops that can serve as host plants to each pest or pathogen and (3) measurement techniques from population ecology for quantifying population similarity and divergence.
Since each crop within a country has a different level of pest and pathogen mismatch with technology producing countries, we can conduct all of our analysis focusing exclusively on these cross-crop differences within countries (or states) and holding all country-level characteristics—including political and institutional features—completely fixed. We also show that the findings are not driven by innate suitability or other features of geography that are unrelated to the appropriateness of technology.
Thus, pest and pathogen mismatch with the frontier is a strong predictor of which places are left out of global technological progress, leaving them systematically worse off.
Now we have all the tools to understand Figure 1, which displays a global map of pest and pathogen mismatch with the “frontier.” In Figure 1, we assume that the US is the one center of technology development and display each location’s pest and pathogen mismatch with the US. In the paper, we also use comprehensive data on technology development to determine the frontier countries for each crop based on how much crop-specific technology they develop. While this data-driven strategy incorporates a broader set of technology developing countries, including France, Germany, Italy, and the Netherlands, the map and associated findings remain very similar since technology producing countries tend to be ecologically similar; the more transparent, US-only version is displayed here for simplicity.
To construct the map, we first determine from the CABI data which pest and pathogen threats exist in each country, and which specific host crops each can consume or infect. We then use data from EarthStat, which compiles fine-grained global data on agricultural production, to determine which specific crop is cultivated on each parcel of agricultural land around the world. Combining these two sources of data, we determine the crop-specific pest and pathogen mismatch between each pixel and the US.
Darker shades of blue correspond to locations with greater pest and pathogen mismatch with the US. These are the places where modern technology is predicted to be least “useful” and where, even if producers adopted the world’s “best” technologies, productivity would still remain low since modern technology is not appropriate for the local context. For example, these are places where the Maize Stalk Borer and other ignored pests dominate—even if farmers adopted the most advanced genetically modified seeds, they would remain unproductive because their threats are different from those that the seeds were designed to combat.
There is substantial variation in pest and pathogen mismatch across countries. Large swaths of sub-Saharan Africa are dark blue in the figure, and we show in our research that the lack of appropriate technology is a major cause of the low average modern technology use on African farms. Other low-income areas, however, like India and parts of Brazil, are a much lighter shade of blue in the map, suggesting potential benefits for these countries from innovation by US firms. There are also large differences across regions and crops within countries, and we show in our work that pest and pathogen mismatch is not only relevant for explaining cross-country differences in productivity, but is also a strong determinant of which sub-national regions are most (or least) productive and which crops they specialize in.
This map also provides a guide for where future technology development could have the largest global effects on productivity. It allows us to identify the locations where crop-specific breeding and research investment could have the largest potential effect on reducing global disparities in productivity. In particular, the network of pest and pathogen mismatch across countries and crops allows us to identify the places that are most “ecologically similar” to highest number of dark-blue locations on the map, and hence the places from which new technology could diffuse and increase productivity by the most. We find that the largest benefit to the currently least-productive countries would come from developing technology to match the ecological characteristics of India and Nigeria, and to a lesser extent Ghana, Zimbabwe, Tanzania, and Democratic Republic of Congo, Pakistan, and Thailand.
While the map illustrates the stark global inequalities generated by the “inappropriate technology problem,” there is also some reason for optimism. The Green Revolution, a period of institutional and philanthropic investment in developing seed varieties more appropriate for tropical countries, shifted the direction of innovation toward technology that was more appropriate for low-income contexts. This historical episode suggests that seeding appropriate technology is possible, and our findings above on the “optimal” locations for future research could serve as a guide for where funds for a “Second Green Revolution” may be directed.
Moreover, in recent years, the “BRIC” countries (Brazil, Russia, India, and China) have contributed a growing share of global agricultural R&D. If this trend continues, it could offset some of the inappropriate technology problem since the BRIC countries are, on average, more ecologically similar to the dark parts of the map than the current centers of global R&D. In practice, however, the global impact of BRIC-developed agricultural technology will likely be determined as much by geopolitics as by ecological mismatch.
Finally, a chorus of voices is drawing growing attention to the stark inequities in global technology investment, and the fact that even philanthropic and public-sector dollars flow disproportionately to rich countries. For example, the Gates Foundation has come under criticism for the fact that its agricultural research grants, which are explicitly designed to fund technology development that would benefit the world’s poor, rarely fund research in low-income countries themselves. Our findings highlight the costs of a highly concentrated research ecosystem and map the parts of the world that the current system leaves behind. Technological progress in the frontier alone underpins existing global disparities, some of which can only be reined in by appropriate technology.