The publicly-funded development of selected and cross-bred traditional varieties, and the emergence of the pri-vate sector
From the origins of agriculture to the late 19th century, improvements in agri-cultural productivity have been mostly driven by informal selection of varieties. Farmers “tinkering” (selecting traits among their cultures, picking up and crossing wild varieties, etc.), or plant prospectors importing improved crops from abroad have been responsible for “great advances in American agricultural productivity, […] before the modern scientific age” (Alston et al., 2010, Chap. 7). As claimed by Pardey et al. (2014), the beginning of scientific breeding of plant varieties can be dated back to the end of the 19th century, when Gregor Mendel’s works on heredity in plant reproduction were discovered again.1 This re-discovery has triggered the adoption of a scientific approach in selection and reproduction of plants, allowing breeders to isolate and maintain specific traits through generations of plants. Such activity has been essentially undertaken by publicly funded institutions, at least until the 1930s. Alotting research on plants to the public sector was largely justified by several market imperfections. In particular, in the absence of proper institutions, agricultural innovation gathers the characteristic of public goods. Indeed, improved plant varieties are, first, non-rival. The costs of collecting a few seeds or plant parts from existing cultivations of improved varieties is almost negligible. Innovation on plants, second, is non-excludable as well. Especially in the early 20th century, breeders could hardly trace their seeds and prevent a farmer from replanting, or giving away, the product of its cultivation. Thus, agricultural innovation is under-provided by a market of private actors and requires public intervention.
The reasons why agricultural innovation is a public good can be summarized as a lack of appropriability of the R&D outcome. Appropriability over plants is weak in general, but is heterogenous across plants, depending strongly on the way they propagate:
• Asexual reproduction involves one single plant, which, hence, clones itself (rather that reproduces in the common sense of this term). It may occur from either vegetative reproduction (in that case a whole new plant can arise from a piece of an existing plant) or from apomixis (the new plant germinates from a seed or a bulb produced by the plant itself and that does not requires any fertilization). Potatoes or oignons are examples of asexually reproduced plants. This type of plant is very stable through successive generations. It is thus obviously quite difficult for the breeder of an asexually reproduced plant to prevent farmers from replicating its invented variety (as long as they have access to one specimen of the plant), and hence to appropriate the benefit of its invention.
• Sexual reproduction requires the fertilization of an embryo (i.e. the gathering of a male gamete, the pollen, with a female gamete – be it pistil, cone, etc.) for propagation, which relies on the production of seeds. It can be subdivised into two categories:
– Self-pollination, in which most part of the pollen of a plant fertilizes the female gametes of the same individual. Wheat is an example of self-pollinated crop. Self-pollinated crops are thus rather stable (mutations occur during the formation of seeds, but to a moderate extent), and a single specimen of the variety can easily duplicate.
– Cross-pollination, in which most part of the pollen of a plant fertilizes the female gametes of another individual. Corn is an example of a cross-pollinated crop. Cross-pollinated crops are the least stable ones – how-ever, it is quite likely that, except for hybrids, the seeds from a variety planted in a field will have characteristics that are rather similar to its parent plants’ ones. Self-pollinating plants sometimes cross-pollinate and reciprocally, but the re-sulting seeds are a negligible proportion of produced seeds.
The lack of appropriability justified the intervention and the crucial role of pub-lic action in the development and distribution of seeds and crops. For example, the US Patent Office acknowledged, in the early 19th century, that intellectual prop-erty rights and market incentives for private research in the agricultural sector were not sufficient to encourage private firms to innovate and develop new varieties and animals. This conclusion caused the Patent Office itself to start importing seeds and breeding animals from abroad, to make up for the failure of national R&D to encourage the development of plants varieties and animal breeds in the US (Huff-man and Evenson, 1993). Hence, throughout the first half of the 19th century, the majority of productivity gains across the world had been obtained mostly by picking-up improved seeds from abroad (which, necessarily, provided varieties that proved to be poorly adapted to local conditions). Then, in the second half of the 19th century and until the late 1920s, plants have been improved, and seeds provided to farmers – generally for free – by governmental agencies. The relevant agencies adopted more scientific and systematic methods, and developed innovations that were more adapted to domestic conditions of production and demand. For example, in the United States, the second half of the 19th century saw the adoption of var-ious Acts aimed at financing agricultural universities and publicly supported R&D (Fernandez-Cornejo, 2004).
Highly productive varieties developed and com-mercialized by a concentrated private sector
Half of the agricultural R&D effort on varieties now focuses on biotechnologies and genetically engineered plants, and most of the potential for future innovation in plant varieties, as well, is concentrated in this technology. Research on GM crops is now essentially undertaken by the private sector, in a very concentrated market, which has various, and rather uncertain consequences on the market for innovations.
R&D on plants varieties has allowed the development and adoption of highly productive varieties
Research on genetical engineering in agriculture has made possible the identifi-cation, and introduction in existing plants, of genes that develop specific, valuable traits. For a long time, breeding of improved crops has been based on crossing varieties, which has allowed significant increases in plants productivity. Built on both “conventional breeding” and, since the 1980s, on genetical engineering as well, modern varieties of plants have driven significant increases in cultures productiv-ity. Fuglie et al. (1996) found that genetic improvement of varieties have largely contributed to productivity gains in the US. For instance, 50% of the 1.13% av-erage annual yield increase in wheat cultivation over the 1975-1992 period can be attributed to newly developed varieties – the remainder is due to improvements in processes and innovations on other inputs (chemicals, irrigation systems, machinery, etc.). More generally, Fischer and Edmeades (2010) estimated that improved vari-eties have driven an average increase in cereal yields between 0.5% and 1% per year since the 1980s – which is consistent with the findings of Duvick (2004). Evenson and Gollin (2003b), summarized in Evenson and Gollin (2003a), studied the pat-tern of agricultural yields in developing countries. They showed that the diffusion of modern varieties (MVs) with enhanced characteristics has been a continuous process from the early 1960s to the late 1990s. International agricultural research centers (IARCs) have played a major role in this process (almost 50% of MVs have been ei-ther developed in IARCs directly, or involve a parent line developed in such centers). During the “early Green Revolution”, between 1961 and 1980, 21% of the growth in yields (which averaged 2.5% per year across all developing countries), and 17% of the growth in output growth may be attributed to MVs. During the “late Green Revolution”, between 1981 and 2000, the contribution of MVs accounted for almost 50% of the 1.8% average annual increase in yields across developing countries, and 40% of output growth.
Genetically modified crops deserve a specific attention because they represent a large share of both investments in R&D on varieties and outcomes of this R&D. According to RoAPE (1998), genetically engineered seeds can be classified into three generations:
• First generation: crops with traits modifying inputs use. These crops allow farmers to modify their mix of inputs (pesticides, herbicides, or fertilizers), as the traits belonging to this generation give special features to the plant itself. Herbicide tolerant and pest resistant crops are the most widely cultivated crops in this generation. Herbicide tolerant (HT) crops allow farmers to spread wide-range, non selective herbicides over their cultivated areas, without having to target weeds precisely as required with conventional crops to avoid killing them with the weeds. The herbicides for which tolerance traits have been developed are glyphosate (commercially known as Monsanto’s Roundup) or glufosinate (commercialized under several brands as Basta or Liberty). Pest resistant crops produce, without farmers intervention, one or several pesticides that are toxic to some of the crops’ natural predators. The most common insect resistance is based on the emission of different forms of Bacillus thuringeiensis (Bt), which is lethal to various insects – among them, the European corn borer, several stemborers and the corn rootworm (Romeis et al., 2008). Belonging to this first generation, virus resistance traits have been developed as well, although their commercial success so far has been weaker than HT and Bt. Most of these traits are still in their development phase (Halford, 2006). for instance, drought tolerance can be obtained by genetical engineering, as for Monsanto’s MON87460 maize. The technology has not fully developed its potential yet, however. Crops tolerant to other abiotic stresses (heat, salt, etc.) are also under early stages of development and are likely to be available in the next decade (Qaim, 2009).
• Second generation: crops with specific output traits. In contrast to first gener-ation ones, second generation traits give specific characteristics to the product of the plant, and not to the plant itself. The first GM crop that was (unsuc-cessfully) commercialized actually belonged to the second generation: Calgene had introduced in its Flavr Savr tomato a gene delaying the fruit’s decay – this was supposed to allow farmers to leave the tomatoes on their plant longer, considerably improving their taste. After the introduction of the Flavr Savr tomato in the early 1990s, few significant attempts to commercialize second generation traits took place over the next 15 years. However, second gen-eration of GMOs regained attention recently, with the very interesting and promising introduction of enhanced nutritious capacities in agricultural out-put, named “biofortification”. A very well known biofortified GM crop is the “golden rice”, a variety of rice enriched in vitamin A developed in the early 2000s for non-profit diffusion by the Philippines’ International Rice Research Institute. For several reasons (lack of public acceptance, cautious regulation of GMOs, etc.), the golden rice is still in its development phase, and was still not commercially available in 2016 (Philpott, 2016). Several second gen-eration GMOs are to be released in the coming years, including “functional foods” (i.e. “foods that provide health benefits beyond basic nutrition”) with enhanced fatty acid profile, increased vitamin or mineral contents, etc. (Pew Initiative on Food and Biotechnology, 2007).
• Third generation: crops producing output whose main use is neither food nor fiber. This last generation of GMO traits are those that make the plant pro-duce a valuable output (in general, a molecule) that it does not produce at all naturally, without the GM trait. The plant is thus used as a “molecu-lar farm” (Moschini, 2006). Such output can be either plant-made pharma-ceuticals (PMPs) or plant-made industrial products (PIMPs). Some specific proteins are already produced using GM plants – for instance avidin, which has been the first PMP to be produced, in the late 1990s (Hood et al., 1997), aprotinin and trypsin (Howard, 2005). However, only a very limited share of third generation genetical engineered crops are exploited, and many of them are currently “in the pipeline” of R&D firms (Qaim, 2009). PMPs production, often referred to as “pharming” or “biopharming” is the major share of third generation GM crops R&D currently undergoing, and, among them, the de-velopment of antibodies and vaccines is particularly advanced and promising on short to middle term. Pharming provides an opportunity to produce, at a much lower cost and/or with much weaker side effects, pharmaceutics for which laboratories used to rely on either natural processes, or host systems other than plants (microbes, yeasts or animals). PMIPs, despite a less ad-vanced development than PMPs, benefit from a broad spectrum of potential applications: GM crops can produce enzymes used in the production of paper and textile, proteins such as collagen, and biodegradable plastics (Moschini, 2006).
Concentration in the private sector of R&D on plant varieties
A striking feature of the sector of plant varieties is its concentration: few firms control most of the market for GM traits. The movement of concentration has been quite fast, and occurred in the late 1990s and early 2000s. Before a massive wave of mergers and acquisitions among the agricultural biotech firms, the market of seeds and crops was rather atomized. Numerous small firms and start-ups owed a significant market share: in 1995, 37% of the total value of seeds throughout the world were sold by the 10 largest seed producer firms.7 In 2009, this figure had increased to 73%. Monsanto and DuPont (Pioneer) accounted for 27% and 17% respectively, and more generally United States firms’ market share was higher than 50% (Fuglie et al., 2011). This concentration trend is more obvious in some sectors: on the vegetable seeds market only, the top 8 companies accounted for 94% of global sales in 2007 (Heisey and Fuglie, 2011). Finally, for some specific seeds and crops, figures are even more striking: in 2007, 85% of global GM crops cultivated surfaces held a Monsanto trait, and 98% of these surfaces held traits developed by 5 firms (Monsanto, DuPont, Syngenta, Bayer and Dow).8 Figures on intellectual property protection granted over plant varieties, as provided by Pardey.
It is quite remarkable that the major seeds and crops companies are also the major investors in R&D on varieties (one only notable exception is BASF, whose R&D effort on varieties is comparable to Bayer’s, despite the fact their direct sales of seeds and crops are negligible). In the early 1990s, most of R&D effort on seeds and crops (GM and non-GM) was undertaken by large, traditional seed companies focused almost exclusively on plant breeding. In 1994, they represented 66% of the total of 2006US$ 1,462 millions spent by private actors in R&D on varieties, while the aggregate of firms producing both seeds and chemicals, the “Big 6”, accounted for 22% and small and medium biotechnology firms for 11%.”9 In the early 2010s, however, both the absolute and relative R&D effort of traditional seed companies and small and medium biotechnology firms had decreased. In 2010, they accounted for 21% and less than 3%, respectively, of the total 2006US$ 1,462 millions spent in crops R&D by private actors. On the contrary, the “Big 6” firms had taken the major part of effort, as their share had increased to 76% (Heisey and Fuglie, 2011). These figures are quite consistent with the historical trend towards concentration of the market in the hands of large seeds and chemical firms. First, the seed market has shifted from traditional varieties towards more and more GM crops. Second, most biotech firms have either been chemical producers before 1980 (e.g. Monsanto or Hoechst) and/or largely merged with or been bought by chemical firms (e.g. Pioneer bought by DuPont or Mycogen bought by Dow). Heisey and Fuglie (2011) finally note that the major seeds firms have cross-licensing agreements with each other (especially those inherited from the agreements signed in the 1990s, for example between Pioneer and Monsanto on Roundup Ready and Bt traits). It is also notable that the “Big 6” firms hold together 71% of market shares on agro-chemicals sales (Shand, 2012). Heisey and Fuglie (2011) estimate the share of R&D dedicated to biotechnologies (opposed to the share of R&D dedicated to conventional breeding) to have been around 50% in 2003.
Table of contents :
2 Material and methods
I Agricultural innovation, outcome of the R&D process
1 A century of transformations of agricultural production
1 From crops picking to pharming: a brief history of varieties R&D in the 20th century
1.1 The publicly-funded development of selected and cross-bred traditional varieties, and the emergence of the private sector .
1.2 Biotechnologies in agriculture
2 Highly productive varieties developed and commercialized by a concentrated private sector
2.1 R&D on plants varieties has allowed the development and adoption of highly productive varieties
2.2 Concentration in the private sector of R&D on plant varieties
3 Different channels for the impact of plant varieties R&D on the environment
3.1 The role of crops R&D in sustainable intensification and land sparing
3.2 Crops R&D and greenhouse gas emissions
3.3 Crops R&D and chemicals use
3.4 Agricultural R&D and land use
3.5 Coexistence of newly developed crops and other organisms .
3.6 Innovation and resistance
2 May innovation on varieties share agricultural land with nature, or spare land for it?
1 Literature review and motivation
2 The model
2.1 Herbicide tolerant varieties
2.2 Drought tolerant varieties
2.3 Insect resistant varieties
2.4 Comparison with empirical results
3.1 Market effects of agricultural innovation on land sparing
3.2 Crops at the crossroads of agricultural research and conservation policies
II Institutions for R&D on varieties
3 Evolution of plant intellectual property protection regimes
1 Introduction to intellectual property protection
1.1 From idea to IP
1.2 Requirements and rights associated to intellectual property protection
1.3 Timing of IP rights – First-to-file and first-to-invent – Prior user defense
2 Intellectual property protection on agricultural varieties
2.1 Different intellectual property protection regimes
2.2 Exemples of agricultural IPR implementation
3 The influence of the intellectual property protection regime on the agricultural R&D market
4 Economic assessment of R&D on varieties
1 Theoretical frameworks of analysis of the social value of R&D on varieties
1.1 Valuing competitively supplied innovation – Griliches (1958) .
1.2 Valuing oligopolistically supplied innovation – Moschini and Lapan (1997)
2 Quantitative valuation of R&D on varieties
2.1 Empirical evaluations of publicly or competitively supplied funded innovation
2.2 Empirical evaluation of monopolistically supplied innovation .
3 (Why) is investment in agricultural research on varieties insufficient ?
4 Social value of agricultural innovation in presence of environmental externalities
4.1 Direction and magnitude of the environmental externalities of agricultural R&D
4.2 Valuation of the environmental externalities of agricultural R&D
III R&D on varieties: a competitive process
5 Models of patent races and theoretical analysis of IPP design
1 Stochastic static model of innovation race
1.1 Loury (1979)
1.2 Lee and Wilde (1980)
2 Theoretical analysis of IPP design
2.1 The basic characteristics of IP: duration and breadth
2.2 IP regime and subsequent innovations
2.3 Competition for R&D and IPP design
3 Modeling R&D on agricultural varieties and its specific features .
3.1 Modeling appropriability and markets in the agricultural innovation process
3.2 Pest adaptation and innovation
3.3 Comparing different regimes of intellectual property protection in agriculture
3.4 Optimal R&D effort and IP protection on plant varieties in presence of environmental externalities
6 Model of patent race in presence of environmental effects
1 Literature review and motivation
2 The model
2.1 Duopoly R&D effort
2.2 Social planner’s R&D effort
2.3 Monopoly’s R&D effort
3.1 Environmental effects in the innovation analysis framework
3.2 Policy implications for intellectual protection and public support to R&D
1 Main results