Patents are key to the transition to de-agrarianization---extinction.

Risks of extreme weather events due to climate change, of pandemics due to anthropogenic land conversion, and of permanent displacement due to sea level rise are all growing (Hauer et al. 2019; IPCC 2019; Gibb et al. 2020). Many of these risks are likely to rise with continued agricultural development based on land conversion and imprecise application of biochemical inputs such as antibi- otics, fertilizers, and pesticides (Rohr et al. 2019). They are also correlated with conflict risk. Uninsured risk is a primary cause of poverty traps, so rising and shifting risks increase the likelihood of rural subpopulations collapsing into chronic poverty (Barnett, Bar- rett, and Skees 2008; Barrett and Santos 2014; Barrett et al. 2019).

Recent years have also brought a growing backlash against globalization, manifest in trade restrictions suddenly enacted by dozens of governments on both agricultural exports and imports. Combined with the upward pressure on prices generally over the past twenty years, this has translated into appreciably greater global agricultural commodity price volatility (figure 2).

Rising and shifting risks facing agri-food systems puts a premium on technological innovations to help manage risks better but also on solidarity mechanisms to transfer risk from the unlucky and vulnerable to the lucky and secure. Improved animal and plant genetics, soil and water management, and other on- farm innovations offer key means of risk reduction. Financial, institutional, and policy innovations to help reduce and distribute risk more efficiently will be equally crucial going forward. Furthermore, reliable and timely information is of increasing value in assessing and managing increasing risks, just as it is in capturing the growing demand for foods’ cre- dence attributes.

Social-Technical Bundles as Combinatorial Innovation

Breakthrough innovations are inevitably com- binatorial, clever means of combining prior discoveries in order to meet emergent societal needs (Arthur 2009). As I have already emphasized in discussing past successes, addressing food security challenges invariably requires a combination of both scientific advances that enhance agri-food system TFP and sociocultural change or policy measures that foster solidarity among people.7 Even as Herrero et al. (2020) enumerate scores of promising new scientific and engineering advances with the potential to transform agri- food systems, they especially emphasize eight social and institutional factors as “essential elements for accelerating the systemic trans- formation of food systems.” We must get the institutions and policies right in order for the science—and associated financing—to follow and enable technology diffusion to achieve real transformations at scale.

In this final section, I emphasize five key agri-food system innovation domains for the coming decades if we are to meet the global food security challenges that lie ahead. In doing so, I call attention to the central impor- tance of “socio-technical bundles”—the com- binations of scientific advances and acts of social solidarity—that are likely essential to progress in a wealthier, warmer, more urban, populous, land-scarce, shock-prone, and Afri- can world in which post-farmgate firms increasingly drive agri-food system behaviors and performance.

Agricultural R&D

Rising food demand growth must be met through TFP growth and increased capital investment. The world can afford neither a new round of price spikes that drive millions into poverty and food insecurity nor continued conversion of natural lands into agriculture, given the climate and extinction crises. It is unrealistic to expect an influx of labor into agriculture. So perhaps the most obvious imperative for the years ahead is to re- accelerate and re-orient agri-food systems R&D investment. Especially high priorities are for philanthropic and public funding to ensure open access to discoveries that focus more on Africa; on nutrient-rich fruits, nuts, and vegetables; and on post-farmgate pro- cesses. But science alone will not suffice.

The case for increased R&D investment is straightforward. Agricultural R&D already underway shows considerable promise to at least partly address some of the challenges ahead. There has been major progress over the past decade or so in single crop traits, such as biofortification (Bouis and Saltzman 2017), stress tolerance (Steinwand and Ronald 2020), transgenic pest resistance (Qaim 2016, 2020), and improved photosynthetic efficiency in plants (Simkin, López-Calcagno, and Raines 2019; López-Calcagno et al. 2020). Recent advances in genomic and synthetic biology tools (e.g., gene editing, biosensors, designed nucleic acid systems) promise to accelerate and broaden the scope of genetic advances in crops and livestock. In particular, these emergent methods and technologies show considerable promise in addressing the environmental and health shortcomings of Green Revolution era plant and livestock technologies (Qaim 2020). So there are rea- sons for optimism based on the scientific research already underway.

Historically, the returns to agricultural R&D have been exceptionally high, indicating systematic underinvestment relative to alter- native uses of scarce funds. Pardey et al. (2016a) estimate mean (median) internal rates of return, 1975–2014, of 42 (35)% and 62 (38)% in SSA and the rest of the world, respectively, a clear sign of underspending. The food systems impacts of R&D are clearly reflected by Hertel, Baldos, and Fug- lie’s (2020) finding that from 1991–2011 direct R&D investments in SSA were the main driver of lower food prices in Africa, not trade and not the adoption/adaptation of technolo- gies developed via R&D elsewhere. If afford- able, healthy diets are a first-order objective in meeting the coming global food security challenge, then increased agri-food R&D is plainly an instrument of primary importance, especially for Africa. Combined with borrow- ing costs that presently sit at historic lows, the world is awash in capital seeking attractive returns of the sort that agricultural R&D has long delivered.

Indeed, the returns to agri-food research increase further when we recognize their role in insuring against temporal uncertainty (Cai, Golub, and Hertel 2017). Given the long lags between R&D expenditures and agri-food market impacts at scale, as well as the consid- erable temporal uncertainty in climate, demand, and supply projections in the absence of technological advance, R&D holds option value. This is especially valuable because food price series are positively skewed due to their storability (Williams and Wright 1991). Exces- sive stinginess in agri-food R&D investment therefore increases the risk of future food price spikes, with potentially calamitous impacts on food security. Cai, Golub, and Her- tel (2017) show that this uncertainty implies a need to sharply accelerate public R&D spend- ing relative to estimates that ignore uncer- tainty, especially for places such as Africa where future uncertainty—especially around population and income growth rates—is espe- cially great.

But we cannot merely replicate agricultural R&D patterns of the past. First, agricultural R&D increasingly needs to aim to mitigate the adverse effects of climate change on productivity growth (Hertel and de Lima 2020). Anthropogenic climate change has been a strong headwind in the face of technological change, reducing global agricultural TFP growth by an estimated 21% since 1961, equiv- alent to losing roughly a decade’s productivity growth (Ortiz-Bobea et al. 2020). Climate adaptation research is required not only to adapt to warming temperatures, but equally to increased risk of drought, flooding (especially with sea water), and to pathogens and pests whose ranges are shifting, that is, to abiotic and biotic shock resistance and stress tolerance. Even in temperate zones where warmer tem- peratures should physiologically stimulate more rapid crop growth, because warming also changes pathogen and pest pressures, climate change increasingly compels defensive (e.g., pesticide, weeding labor) investments by small farmers to protect crops, generating costly tradeoffs with productivity-enhancing inputs such as inorganic fertilizer or improved seeds among liquidity constrained small farmers (Jagnani et al. Forthcoming).

Driving down the real and relative prices of healthful foods such as fruits, legumes, nuts, and vegetables will require significant expan- sion of R&D directed towards these products, not just doubling down on traditional focal cereal, livestock, root, and tuber commodities (Nelson et al. 2018; Mason-D’Croz et al. 2019; Haddad 2020). Expanding the per capita sup- ply of healthful foods will also require signifi- cant reallocation of lands from starchy staples and feed to more nutrient-dense commodities (Sanchez 2020). We must stop treating horti- cultural crops as “specialty crops,” as if they are not absolutely central to delivering a healthful food supply.

Agri-food R&D must also increasingly attend to farm- and landscape-scale cropping systems change rather than just to single spe- cies. Most animal and plant genetic improvement efforts focus on improving indi- vidual plants through some form of biomimi- cry, finding and replicating or adapting solutions found elsewhere in nature. Wei- ner (2019) makes a powerful argument that the most promising opportunities to promote higher yielding plant genotypes will rely not on selection and ecological engineering for improved plant-level productivity but, rather, for greater population performance that comes expressly at the cost of individual plant fitness. He argues that the semi-dwarf varieties at the heart of Green Revolution advances were “due to inadvertent group selection” as pure stands of shorter stature varieties could thrive and generate higher plot-scale yields but would have been selected out had they been interspersed with taller varieties that would outcompete them for solar energy. Especially with monocultures’ increasing vul- nerability to pests and diseases (Fones et al. 2020), R&D needs to work more on poly- cultural mosaics that preserve, or even advance, agrobiodiversity. For example, crops bred to “waste” scarce energy by fixing atmo- spheric nitrogen and accessing water deeper underground through long root systems can generate significant positive spillovers for other plants and the planet, at the cost of stalk-level edible biomass in those specific commodities. As I discuss below, Wei- ner’s (2019) point about targeting population- level performance, not just individual perfor- mance, applies more broadly to policy and the institutions that govern community life, not only to promoting animal and plant productiv- ity through R&D.

R&D for affordable, healthy diets will also increasingly require advances in off-farm products and processes. Many of these can be enabled by lower cost renewable energy and digital technologies, points to which I return to below. As energy and information costs continue to fall, for example, indoor, vertical farming of horticultural crops using aero/ aqua/hydro-ponic methods in urban/peri- urban settings inevitably becomes more com- mercially viable, magnifying the need for R&D to tailor varieties to these unnatural environments (Pinstrup-Andersen 2018; Beacham, Vickers, and Monaghan 2019). The same goes for improved food preserva- tion and cold chains to facilitate farm-to- market intermediation of fresh and whole foods, for example, solar-powered cooling for value chains to reduce post-harvest nutrient loss and safeguard food safety. More progress can be made to improve diet quality by invest- ing in technologies that help lower both the real and relative prices of healthy foods than by inherently zero-sum price policies or by consumer behavioral change communications efforts.

We also need significant shifts in the geogra- phy of agricultural R&D, which has lagged badly over the past generation in both Africa (and other LICs) and HICs, while growing sig- nificantly in middle-income Brazil, China and India. Public R&D in MICs has now over- taken that in HICs and both groups’ R&D dwarfs research spending in LICs, especially in Africa (Pardey et al. 2016b). China now out- spends the US in both public and private agri- food R&D (Chai et al. 2019). Agroecosystems exhibit huge heterogeneity, so place-specific R&D is essential. Given that most growth in food demand will take place in Africa, and we can only rely so much on trade to meet expanding demand, a growing relative share of expanding agri-food R&D needs to occur in SSA.

Agricultural R&D has also increasingly shifted to the private sector, especially but not exclusively in HICs (Pardey et al. 2016b). Private R&D now accounts for more than two-thirds of total agricultural R&D spending in both China and the US (Chai et al. 2019). Many of the most promising innovations on which future agri-food systems will depend— especially in digital innovations and de- agrarianized food production systems discussed below—rely overwhelmingly on pri- vate finance. If private financing ties up discov- eries in complex intellectual property (IP) issues, it will retard diffusion and diminish the impact potential of technological advances.

This last point underscores how simply pushing more money at yawning R&D needs is unlikely to deliver much. Consider the cau- tionary tale of “golden rice,” the transgenic variety that biosynthesizes beta carotene, a remarkable scientific discovery aimed at addressing the prevalence of vitamin A defi- ciency among poor rice-eating populations worldwide.8 The US Patent and Trademark Office recognized golden rice with its 2015 Pat- ent for Humanity Award. But the discovery of this breakthrough was first published in 2000 (Ye et al. 2000). Twenty years later, despite great fanfare and scores (if not hundreds) of millions of dollars in follow-on investment, golden rice still has not been approved for commercial cultivation and sale in a develop- ing country. It has faced myriad hurdles associ- ated with IP licensing, national biosafety and food safety regulations, environmental activ- ists mobilized in opposition to this and other genetically modified organisms, and so on. Juxtapose the golden rice case with that of the IR8 and IR64 varieties originated in 1966 and 1985, respectively, by the Interna- tional Rice Research Institute. These were arguably far less noteworthy scientific advances than golden rice, yet the semi-dwarf IR8 was the first “miracle rice” of the Green Revolution, and the third generation IR64 became purportedly the most widely diffused cereal seed variety in history. The difference in impacts between IR8/64 and golden rice arose less from scientific differences than due to social ones. In the face of broad popular dis- trust of genetic engineering, and active— sometimes violent—opposition from environ- mental groups, and faced with a dense thicket of patents and diverse national regulations to navigate, golden rice diffusion has been far slower and more costly than for the earlier rice varieties developed using conventional plant breeding methods and public R&D and exten- sion funding in an environment more trusting of science and less reliant on private finance and IP protections. Advances in genetics (or any other scientific field) are insufficient to ignite systemic change. Social innovations must accompany R&D to make investments in technological advances impactful.

The absence of enabling sociopolitical inno- vations helps explain why TFP gains per real dollar of agricultural R&D have been steadily declining, by an average annual rate of 3.7% from 1970 to 2007 (Bloom et al. 2020). One likely partial explanation is that prior research tackled many of the lowest cost high-return options, that is, there may be intrinsically dimin- ishing returns to R&D over time. But the declining cost effectiveness of agricultural R&D also clearly underscores the need to address sociocultural and political obstacles to both inducing appropriate scientific research and to enabling promising innovations to dif- fuse and scale. R&D investments must be matched with institutional and policy change to unlock the potential of scientific advances to meet the global food security challenges ahead.

One problem is the complex political econ- omy of agricultural biotechnology, which has deep roots in scientific illiteracy and resulting political activism in the broader population (Herring and Paarlberg 2016; Fernbach et al. 2019).9 The result has been overly bur- densome regulation, especially in Europe, most of SSA, and parts of Asia (notably, India), that impedes field deployment and upscaling of advances arising from agricultural biotechnology (Qaim 2016, 2020; Steinwald and Ronald 2020). A notable consequence is that biosafety and food safety concerns differ- entially retard genetic improvement and refor- mulation of agrichemicals and foods relative to advances in digital and mechanical technol- ogies, with the predictable result that the latter innovations attract wildly disproportionate private investment in agricultural R&D today.

A further complication arises from IP regimes that loom increasingly large in influencing the pace and directions of research as the share of private agri-food R&D increases. To be sure, private R&D is desirable, indeed essential, and IP protections are necessary to induce more private investment. But patent systems are increasingly broken, discouraging the sorts of innovation society needs (Jaffe and Lerner 2011). In the agri-food space, changes in IP law have substantially compli- cated R&D, most notably following the 1980 Chakrabarty decision by the US Supreme Court that for the first time permitted patenting a living organism. Strategic hold-ups and patent thickets drive up costs and slow progress as private R&D becomes more dominant (Schimmelpfennig 2004; Jaffe and Ler- ner 2011). Solutions are available, as manifest in the Open COVID Pledge launched in April 2020 to enable biomedical innovators to make their IP widely available free of charge, following a model based on those used for open- source software; the pledge covered more than 250,000 patents worldwide by the end of July 2020 (Contreras, Eisen, and Peters 2020). Innovations to patent law could likewise prove fruitful to induce greater R&D investment targeting LMIC challenges for which the gap between enormous societal gains and limited commer- cial market potential limits financing.

Economies of scale and scope in advanced agricultural biotechnology, combined with IP and regulatory issues, naturally lead to compe- tition and anti-trust enforcement issues because patents confer state-sanctioned monopoly power on the IP holder and regula- tions often create barriers to entry. The past five years have seen unprecedented consolida- tion among massive global seed and agrichem- ical corporations, raising important concerns about how industry concentration shapes agri- cultural R&D investment and TFP growth, a topic about which the evidence base remains exceedingly thin (MacDonald 2019). The worry is that consolidation can discourage innovation, either through catch-and-kill acquisitions of innovative competitors or if there exist internal diseconomies but external economies of scale and/or scope to R&D.

The ongoing geographic mismatch between where agri-food R&D is most needed versus where it is growing, and heightened reliance on scientific methods and data with variable and uncertain regulation, underscore why successful agri-food R&D requires not just investment in new science but policy and institutional change, and acts of solidarity with marginalized populations. If we get the right regulatory and market signals in place, adequate R&D financ- ing and breakthrough science will almost surely follow. These sorts of socio-technical bundles will be essential to develop the next generation of improved animal and plant genetic material; chemical, mechanical, and digital technologies; and business processes that will make producing healthier, environmentally sustainable foods more remunerative.

De-Agrarianization

The prior section made the point that agri-food TFP growth is essential and requires renewed (and changed) investment in R&D. But TFP growth is also insufficient if we are to address the climate, and especially the extinction and pandemic crises. We must move land out of agriculture and into production of environmen- tal and energy services, all while increasing food production so as to drive down real food prices, and especially for healthier foods. This implies a need to de-agrarianize food production, that is, to decouple from the land, replacing it with capital.10 Although there exist opportunities to expand renewable energy gen- eration with agricultural landscapes, the need to conserve more land for biodiversity will necessitate accelerating the de-agrarianization already underway in HICs.

This change is coming. At least four distinct, rapidly advancing classes of innovation are poised to facilitate de-agrarianization. The first involves the emergence of livestock feeds based on algae, insects (e.g., black solider fly larvae), and so forth as substitutes for land- intensive cereals and oilseeds-based proteins and fishmeal. These novel “circular feed” alternatives currently are costlier than soy- meal, maize, hay or fishmeal, but costs are fall- ing rapidly, firms (especially in Europe) are beginning to tap these feed sources, and research increasingly shows that these feeds are scalable, yield ASFs of similar quality and safety as those based on conventional feeds, and potentially offer added health benefits (Caporgno and Mathys 2018; Altmann et al. 2020; Cottrell et al. 2020).

The second are tissue engineering-based methods that culture cells to grow animal tissue outside the body, without the environmen- tal, ethical, or financial costs of raising and slaughtering live animals. These “clean” or “cellular” meats have gained considerable attention through start-ups like Memphis Meats and Finless Foods, prompting legisla- tive and regulatory battles over product label- ing, that is, what constitutes “meat.” Although these products remain expensive, unit costs are falling fast, in Memphis Meats’ case by more than an order of magnitude in just two years (2016–2018) and are predicted to fall to the level of conventional ground beef by 2026 (Tubb and Seba 2019).

The third group is controlled environment agriculture (CEA)—so-called “indoor” or “vertical” farming—much of it based on aero-, aqua- or hydroponic methods. This is already taking off across urban areas of HICs and Asia, where concentrated demand is booming, especially for year-round produc- tion of high-quality, high-value, short cycle horticultural products (WWF 2020). The main disadvantage of CEA is that producers forego free sunshine for expensive electricity to stim- ulate plant photosynthesis. The higher cost of labor in urban markets is also a disadvantage, further encouraging greater capital intensity in CEA than in conventional production methods. Yet commercial outfits such as France-based Nature Urbaine or India-based Kisano are already breaking even or turning a profit. And as electricity costs fall, especially with advances in small-scale renewables (e.g., rooftop solar films and panels), the price competitiveness of CEA improves (Pinstrup- Andersen 2018).

The fourth group of innovations use microbes to produce novel foods through a process broadly known as “precision fermen- tation.” Fermentation is a simple, centuries- old process familiar in all food systems glob- ally. It is used to make beer, cheese, and so on in virtually every LMIC in the world. Recent advances in synthetic biology now enable labs to design micro-organisms, such as bacteria or yeasts, that produce more com- plex proteins from inexpensive plant-based feedstocks. This is the technology behind, for example, Beyond Meat and Impossible Foods. These firms’ rapid growth11 suggests the com- mercial viability of at least precision fermentation in the years ahead. Independent assessments confirm that precision fermenta- tion methods, in particular, can scale at costs below those of conventional systems for pro- ducing animal-source foods (Buckler and Rooney 2019; Tubb and Seba 2019).