R&D Capital, R&D Spillovers, and Productivity Growth in World Agriculture

R&D Capital, R&D Spillovers, and Productivity Growth in World Agriculture Abstract Increasing the world’s food supply has depended heavily on increasing agricultural productivity, which in turn depends on investments in research and development (R&D). This article synthesizes findings from more than 40 studies on how R&D investments affect agricultural total factor productivity (TFP) in various parts of the world. The article breaks out the relative contributions to TFP growth of R&D by public institutions, private companies, and the CGIAR (a consortium of international agricultural research centers), including international technology spillovers. Major differences emerge between global regions in sources and efficiency of R&D capital. Developed countries appear to have benefitted more from private and international R&D spillovers than developing countries. Agricultural total factor productivity, R&D elasticities, R&D lags, technological obsolescence Agriculture may be unique in its reliance on productivity for growth. Whereas about one-third of total economic growth comes from increases in the total productivity of factor inputs (Jorgenson, Fukao, and Timmer 2016), in agriculture, total factor productivity (TFP) accounts for about three-fourths of growth at the global level and virtually all growth in industrialized countries (Fuglie 2015). This reliance on productivity reflects agriculture’s dependence on inherently limited resources like land and water, and it is these resource constraints that have given rise to concerns, since at least Malthus, that world population may soon overreach the capacity of what the world can sustainably afford to feed. The fact that agricultural productivity has been able to grow sufficiently to meet rising demand is no accident. Rather, it reflects to a large degree a deliberate choice to commit resources to agricultural research and development (R&D). In today’s advanced industrialized nations, the establishment of public agricultural research institutions in the late nineteenth century helped set in motion a process of technological and structural transformation of their agricultural systems (Ruttan 1982). That process is still going on today, and in fact, has been extended to most of the world. Nearly all countries now have national agricultural research institutions of one form or another. In addition, international agricultural research centers have been established (Alston, Dehmer, and Pardey 2006) and the role of the private sector in generating new agricultural technology has grown (Fuglie et al. 2011). Because positive externalities (spillovers) from R&D lead to undervaluation of innovation in the marketplace, governments have a central role in creating the knowledge capital required for economic growth.1 1 The government creates knowledge capital through direct investment in R&D, and by establishing intellectual property rights, creating excludability conditions for private inventors. Positive externalities from knowledge capital is central, for example, in New Growth Theory (Romer 1990). As articulated by Romer (1990), once new knowledge is created it is available everywhere to all, forever, except as constrained by insufficient human capital to make use of it, and by legal or other measures to protect the intellectual property of inventors. But because agriculture depends on environmental forces, and thus technology is sensitive to location, agricultural knowledge capital is likely to be much more constrained than the knowledge capital envisioned by Romer. Further, because environmental forces change over time (due to the co-evolution of pests and diseases, water and land resource degradation, and climate change), technological obsolescence in agriculture will eventually set in (Ruttan 2001). Olmstead and Rhode (2002) dubbed the need for continued research just to maintain agricultural productivity as the curse of the Red Queen.2 2 “Now, here, you see, it takes all the running you can do, to keep in the same place,” The Red Queen to Alice in Lewis Carroll’s Through the Looking Glass. Another factor, though not unique to but probably accentuated in agriculture, is the relatively slow uptake of technologies due to the highly dispersed, heterogeneous, and small-holder structure of producers. These characteristics of agriculture suggest the general shape of a time path for how investments in agricultural R&D are likely to affect farm productivity: a relatively long lag between R&D spending and when that spending results in significant improvements to aggregate farm productivity, and eventual depreciation of the productivity gains without renewal of R&D capital (Huffman and Evenson 2006; Alston et al. 2010). One implication of this view is that continuously raising agricultural productivity requires continuous growth in R&D spending. The objective of this paper is to provide a synthesis and assessment of how investments in agricultural R&D have affected productivity growth in world agriculture. First, I assemble historical data on public R&D spending from 150 countries, the private sector, and the CGIAR consortium of international agricultural research centers. Most of the estimates date from 1960, though for some formerly centrally-planned (transition) countries they start from the 1990s. I then construct estimates of accumulated R&D capital stock from each of these sources using a model conforming to the concepts outlined above (and formalized below). Through a review of 44 econometric studies on how R&D influences productivity growth in agriculture, I derive average R&D elasticities for the different sources of R&D and for different global regions. The R&D elasticity measures the percentage change in the TFP (the ratio between the gross output of crops and animal commodities and the combined input of labor, land, capital and materials employed in their production) given a 1% change in R&D capital. I use these elasticities to predict growth in TFP for each global region and compare these predictions against measured TFP growth from 1990 to 2011. The exercise sheds light on the role of public, private, CGIAR R&D investments, and, importantly, the role of R&D spillovers, on raising agricultural TFP around the world. Model of Agricultural R&D Capital and Productivity The Agricultural Production Function For assessing long-term growth in agriculture, it is useful to start with an aggregate production function: Qt=AtRtXtLabort,Landt,Capitalt,Materialst (1) where Q is the agricultural output of a country or region, A is technology or TFP (a function of variables R representing the creation and diffusion of knowledge and ideas), X consists of factor inputs, and t is time. From equation (1), growth in output over time can be decomposed into parts due to technological change and input accumulation: ∂Q∂t=∂A∂t+∂X∂t. (2) Empirical estimates of equations (1) and (2) applied to world agriculture find that during the twentieth century, the main source of growth shifted from input accumulation to productivity in most regions of the world (Hayami and Ruttan 1985; Federico 2005). For the United States and most other developed countries, this transition began in the mid-twentieth century. Since the 1950s, for example, growth rates for U.S. agricultural output and agricultural TFP have been almost synonymous, with aggregate input hardly changing, except in its composition (Wang et al. 2015). For developing countries, this transition began later. For the world as a whole, Fuglie (2015) estimated that since 1990, about three-fourths of the growth in world agricultural output was due to improvements in TFP, although for some low-income countries, particularly in Africa, factor accumulation continued to be the main source of growth. Understanding the future pace of agricultural growth will mostly involve forecasts of agricultural productivity. For example, the long-term world agricultural supply and demand projections of the United Nations Food and Agriculture Organization (FAO) suggest that between 2006 and 2050, global food demand will rise by around 60%, and at least 90% of this increase will come from raising agricultural yield and cropping intensity on existing farm land, rather than the expansion of farm land (Alexandratos and Bruinsma 2012). Given the overriding importance of productivity to agricultural growth, it is useful to have an explicit model of equations (1) and (2) that can give policy makers some leverage with which to influence future food supply. For this, I specify a Cobb-Douglas function where technology and productivity is driven by the stock of ideas, or knowledge capital, which arises from formal R&D investment from public and private sources: At=A∏i=1ISitδi (3) where Sit is the stock of R&D capital from one of i=1,2,…I sources of new technology for agriculture (e.g., from public research institutes, universities, private agribusiness, and international centers). The elasticities δ1,δ2,…,δI translate how a change in R&D capital from source i affects growth in TFP (i.e., a 1% increase in R&D capital Si increases TFP by δi percent). Empirically, the formulation of agricultural knowledge capital Si in equation (3) has been treated similarly to physical capital except for some special features. Like physical capital, knowledge capital (or R&D capital) is the accumulation of past annual investments in R&D and eventually depreciates. But R&D capital is likely to have a longer gestation period (time for research to lead to useable technologies and spread to farmers). It may also be longer-lived, given that “ideas” might not wear out as fast as machines or structures. Most importantly, knowledge capital is non-rival (its use in one place does not limit its use elsewhere). The non-rival nature of knowledge capital is what gives rise to potential spillovers. Numerically, knowledge capital with the above features can be measured using some version of the following: St=∑i=0TwiRt-i (4) where Rt-i is annual R&D spending i years ago, and wo,w1,…wT is a set of weights that are initially at or close to zero, rise to a peak, and then eventually fall back to zero after T years. The zero or low initial values of wi reflect the gestation period for research to result in new technology that can be used by farmers (e.g., the time to breed a new crop variety). The rising values of wi indicate the diffusion of new technology to farmers. The fact that the wis are assumed to eventually diminish reflects knowledge capital depreciation. Examples of R&D capital depreciation in agriculture abound; they are perhaps most in evidence by the emergence of new pests and diseases that threaten existing crop and animal yield. Knowledge deprecation also occurs when completely new forms of production technologies emerge (e.g., the development of tractors made obsolete many innovations in animal drafting). Changes to natural resources (soil degradation, groundwater withdrawals, and rising greenhouse gas concentrations) may also render ineffective many past innovations. Note several features of the model in equations (3) and (4). First, agricultural productivity depends on past investments in national R&D and on the inventive activities of others that are relevant to the conditions of a country. Second, R&D spending is not immediately translated into useable R&D capital. It takes time for R&D to produce technologies that are both adoptable and adopted by producers. Third, R&D capital depreciates over time. Fourth, the R&D elasticities and depreciation rate may vary by country, being conditioned by institutional and environmental factors. The relevance of inventive activity done elsewhere will be determined by the similarity of farming systems (in terms of ecologies, commodities produced, and scale of production).3 3 Note that New Growth Theory considers TFP growth to be proportional to some measure of constant scientific effort, or ∂lnAt/∂t=δSt, where St is research input and δ is a proportionality parameter (Romer 1990). In this paper, because of depreciation, TFP growth is proportional to the growth rate of research input, ∂lnAt/∂t=δ∂lnSt/∂t. Besides national public R&D, the model allows for external sources of technological change to affect a country’s agricultural productivity. Although agricultural technology is sensitive to local environmental conditions, direct spillovers of technology may be possible from other countries or regions that have similar environments or that create general-purpose technologies that can be put to use locally, though perhaps with some adaptation. Some technologies, like a new agricultural pesticide, may be toxic against a number of agricultural pests that inhabit different ecologies and infest different crops, but its usage may differ across them. Technological spillovers may also arise between sectors, such as from industry to agriculture. New human pharmaceutical discoveries may act against farm animal diseases, but may also require adaptive research for specific animal applications. Admittedly, the model in equations (3) and (4) takes a somewhat narrow view of what causes changes in TFP. By focusing exclusively on knowledge capital generated through formal investments in R&D, the model ignores productivity gains from specialization (made possible through greater openness to trade), economies of scale, and informal innovation (such as by farmers themselves). The model also does not explicitly account for factors that affect the “enabling environment” for technology diffusion (e.g., farmer’s education and health, agricultural extension and credit services, secure land tenure, and the rule of law), other than by acknowledging that these influence the R&D lag structure. Historically, trade policy has had an important influence on agricultural productivity. In late-nineteenth century Europe, Denmark and the Netherlands achieved more rapid agricultural growth than France and Germany because of their greater willingness to import cereals and specialize in animal production (Hayami and Ruttan 1985; Lains and Pinilla 2009). In late-twentieth century East Asia, partly for national food security objectives and partly to reduce rural-urban income disparity, Japan, South Korea, and to some extent China sought to protect local grain producers from international competition but at the expense of overall productivity and efficiency (Otsuka 2013). Policies and institutions have also been an important determinant of the rate of technology diffusion in agriculture. The dispersed structure of agriculture requires that before new technologies can actually affect productivity, they have to be adopted on thousands if not millions of small, family-run farms. How smoothly this occurs is likely to depend on several factors: the complexity, scale and cost of new technologies; the education and skill of farm managers and workers; their means of acquiring information; raising capital and insuring against unexpected losses; the ease of marketing farm surpluses; and how well farmers are remunerated for their efforts (see Feder, Just, and Zilberman 1985). Nonetheless, non-R&D factors are likely to more limited in their capacity to sustain agricultural TFP growth over the long run, due to diminishing returns. For example, productivity gains from trade liberalization or more rapid technology diffusion will be exhausted once a new equilibrium is reached. Informal farmer innovations and economies of scale may be primarily adaptations of knowledge capital arising from formal R&D. Evidence shows that the social returns to policies that strengthen the enabling environment, such as investing in farmer education and extension, are likely to be higher in an agricultural system undergoing rapid technological and structural change than in a technologically stagnant one (Schultz 1975; Foster and Rosenzweig 1996). While policy makers may view improvements in non-R&D factors as substitutes for investments in R&D, it may be considerably less costly to view these as complements. We may expect that formal R&D will be more strongly connected to TFP growth in countries where the enabling institutions are more developed. The Agricultural R&D Lag Structure To translate R&D investment into R&D capital and productivity growth I adopt a framework developed by Alston et al. (2010). These authors proposed an R&D capital lifespan of up to 50 years, and used a gamma distribution to capture the technology maturation and diffusion processes that occur in the early part of this period, and the obsolescence and dis-adoption that sets in toward the end. Huffman and Evenson (2006) proposed a similar concept using a trapezoid curve with a 35-year life span for R&D capital (which can be closely matched, with the right choice of parameters, by a gamma distribution) to represent the development, adoption, and obsolescence phases.4 4 Alston et al. (2010) and Huffman and Evenson (2006) specify the R&D lag structures as a set of weights that sum to one. In this way they apportion $1 of R&D spending to the time periods in which the impact of that R&D is expected to be felt. Here, the same relative weights are used but their maximum value is set to one, so that $1 of investment adds $1 to capital when the investment is fully operational (and less than $1 when it is partially operational), as is standard for measuring the accumulated stock of capital. Some gamma and trapezoid R&D lag distributions are shown in figure 1. In each distribution, R&D spending at time 0 slowly accrues to R&D capital, then peaks and depreciates until the end of its useful life. With the 35-year R&D lag structure, the full impact of R&D spending in year 0 is realized about a decade later, and with the 50-year lag structure, after about two decades. Other lag structures are certainly possible, and the gamma distribution is flexible enough to represent a range of possibilities, given the appropriate choice of parameters.5 5 Alston et al. (2010) tested 60 gamma distributions in modeling the impact of R&D on productivity growth in U.S. agriculture. The 50-year lag distribution shown in figure 2 reflects their preferred choice among these distributions. However, that goodness-of-fit tests often failed to distinguish between models and their preferred choice partly rests on historical information on the development and diffusion of major agricultural innovations. Huffman and Evenson (2006) rely on similar reasoning to choose a 35-year R&D lag structure, as did Fan (2000) in selecting a 27-year R&D lag structure for China. Figure 1 View largeDownload slide Alternative lag structures for R&D capital formation Note: Slightly departing from Alston et al. (2010), I measure the height of the gamma distribution in year k as bk=k-g-1ϕ1-ϕθk-gmaxk-g-1ϕ1-ϕθk-g where g is the R&D gestation lag , ϕ,θ are shape and scale parameters of the gamma distribution, and k = 1…L, where L is the maximum length of the R&D lag structure. Since the values of bkare very close to zero in early years, a value of g=0is used. The ϕ,θparameters for the 35-year and 50-year gamma distributions shown in the figure are 0.90,0.70 and 0.80,0.75, respectively. Alston et al. (2010) divide the numerator in the above equation by ∑k=1Lk-g-1ϕ1-ϕθk-g so that ∑bk = 1. I make a proportional shift to the weights so that the maximum value of bk = 1. This weighting scheme implies that $1 in R&D expenditure will add $1 to R&D stock once the technology from that R&D is fully utilized. Figure 1 View largeDownload slide Alternative lag structures for R&D capital formation Note: Slightly departing from Alston et al. (2010), I measure the height of the gamma distribution in year k as bk=k-g-1ϕ1-ϕθk-gmaxk-g-1ϕ1-ϕθk-g where g is the R&D gestation lag , ϕ,θ are shape and scale parameters of the gamma distribution, and k = 1…L, where L is the maximum length of the R&D lag structure. Since the values of bkare very close to zero in early years, a value of g=0is used. The ϕ,θparameters for the 35-year and 50-year gamma distributions shown in the figure are 0.90,0.70 and 0.80,0.75, respectively. Alston et al. (2010) divide the numerator in the above equation by ∑k=1Lk-g-1ϕ1-ϕθk-g so that ∑bk = 1. I make a proportional shift to the weights so that the maximum value of bk = 1. This weighting scheme implies that $1 in R&D expenditure will add $1 to R&D stock once the technology from that R&D is fully utilized. Different R&D lag structures might be appropriate for different kinds of research. More fundamental advances in agricultural technology are likely to take longer to mature and have a longer and wider impact on productivity compared with research that adapts existing technologies to new localities and uses. The initial effort to develop biotechnology traits for crops in the United States took about two decades, but it took considerably less time to transfer these traits to crop cultivars suitable for South America and elsewhere. The Green Revolution was typified by the global diffusion of semi-dwarf varieties of rice and wheat that had greater yield response to fertilizer. But these varieties often required hybridization with local cultivars that were adapted to pests, disease, day length, and soil types found in specific regions. The research to adapt or refine general-purpose technologies to local environments may involve shorter lags than research that expands the scientific frontier. Whether there are systematic differences in R&D lag structure for countries at different stages of development is an open question. On the one hand, we might expect a longer research gestation period in countries whose farmers already produce at a world technology frontier compared with countries where farmers are producing substantially below the frontier. Thus, developed countries may face longer R&D lag structures than many developing countries. On the other hand, R&D lag structures also incorporate the speed at which new technologies are taken up by a population of farmers. Technology diffusion may be relative slow, even for profitable technologies, in countries with poor enabling environments. This may argue in favor of a longer R&D lag structure in developing countries. Fan (2000), who examined the actual development and adoption time for a number of agricultural technologies in China, and Alene (2010), who estimated the parameters of an agricultural R&D lag structure that seemed to be the best fit for productivity patterns in Sub-Saharan Africa, seem to find evidence of shorter R&D lags compared with what has been estimated for the United States (Alston et al. 2010; Baldos et al. 2015) and the United Kingdom (Thirtle, Piesse, and Schimmelfpennig, 2008). However, none of this evidence is very robust, and R&D lag length is an issue deserving more attention. Agricultural R&D Spending and Capital Accumulation Agricultural R&D Spending This study uses data compiled from multiple sources in 150 countries to construct a continuous time series for public agricultural R&D spending since 1960. Public R&D includes research by government agencies and universities. For high-income countries, the data are from Heisey and Fuglie (forthcoming). For developing countries, fairly comprehensive annual series on agricultural R&D spending since 1981 are available from Agricultural Science and Technology Indicators. For years prior to 1981, R&D spending data are from Pardey and Roseboom (1989) and Pardey, Roseboom, and Anderson (1991). Further, R&D spending by (formerly) centrally-planned countries in Eastern Europe and the Soviet Union are drawn from Judd, Boyce, and Evenson (1991). In addition to these sources, a number of other sources were consulted to fill gaps.6 6 The R&D data series and a full list of sources are available from the author upon request. To convert expenditures in national currencies to constant 2011 purchasing-power-parity dollars (PPP$), estimates were first adjusted to constant 2011 local currency units by the national implicit GDP price index, and then converted to PPP$ using the 2011 PPP$ exchange rate (both series are from World Development Indicators). For a complete picture of global agricultural R&D investments, one also needs the private sector, but here the data are less available. For the United States, Fuglie et al. (2011) provide estimates of farm-oriented R&D by U.S. companies from 1960 to 2007. At the global level, Pardey et al. (2016) report estimates of business-sector spending on food and agricultural research by decade from 1980 to 2010 for major global regions, but do not break out food-sector R&D from farm-oriented R&D. Fuglie (2016) gives annual estimates of global private-sector R&D on crops, livestock, and farm machinery world-wide from 1990 to 2010, but notes that it is difficult to assign private R&D to individual countries with much precision because of the multinational character of the largest firms conducing this R&D. For the purpose of this paper, I treat private agricultural R&D as contributing to a global agricultural knowledge stock. This R&D capital stock affects global regions differently through region-specific R&D elasticities. A higher elasticity value for developed countries, for example, implies that private R&D has a larger impact on productivity in these countries than elsewhere. Another important source of technology for agriculture are non-government, non-profit agricultural research centers. The most prominent of these is the CGIAR Consortium, a system of 15 centers focused primarily on developing-country agriculture that is funded by governments, private foundations, and intergovernmental organizations. The CGIAR was formed in 1971, but many of the centers that make up the system were established in the 1960s. Annual CGIAR R&D spending from 1971 is from Agricultural Science and Technology Indicators with pre-1971 figures from Alston, Dehmer, and Pardey (2006). Agricultural R&D Capital Stocks With long-term R&D expenditure series in constant PPP$, we can estimate and compare public agricultural R&D capital stocks assuming alternative R&D lag structures. Table 1 shows the evolution of public agricultural R&D spending between 1961 and 2011, and accumulated 2011 R&D capital for different regions of the world using the 25-year, 35-year and 50-year gamma lag structures from figure 1. In constant 2011 PPP$, global public agricultural R&D grew from about $7.5 billion in 1961 to $42.3 billion in 2011. Spending rose faster in developing countries than developed countries, with the developed-country share of the total falling from about 65% in 1961 to 47% in 2011. These estimates are broadly consistent with Pardey et al. (2016), who reported public agricultural R&D spending worldwide (not including transition countries) to be $38.13 billion in 2010 (in 2009 PPP$). Further, Pardey et al. (2016) estimated that global spending was 210% higher in 2010 than in 1980, while my estimates, excluding transition countries, give an almost identical 208% increase in real spending between these years. Table 1 Agricultural R&D Spending and R&D Capital by World Region (Constant 2011 PPP$, million) Region R&D expenditure R&D Stock in 2011 1 1961 1991 2011 Total R&D spending, 1962–2011 50-year gamma 35-year gamma 25-year gamma Public agricultural R&D Developing Regions Central America 123 550 866 24,038 9,879 10,336 7,861 South America 812 2,943 3,824 114,946 48,559 48,143 32,879 China 313 1,599 7,768 99,813 27,663 40,118 42,488 Southeast Asia 235 1,204 2,005 57,138 22,060 25,689 19,759 South Asia 278 1,461 3,798 71,658 23,249 31,913 28,482 West Asia 148 615 1,524 36,461 12,869 15,053 12,891 North Africa 119 395 728 20,775 7,663 9,206 7,304 Sub-Saharan Africa 522 1,439 1,910 59,837 24,394 23,407 17,214 Developed Regions Former Soviet Union 336 455 629 24,940 10,316 7,515 5,569 Eastern Europe 2 238 557 904 31,539 13,442 10,936 7,758 Western Europe 1,274 5,570 7,093 230,639 96,103 97,810 69,046 Canada-USA 1,711 4,980 5,501 217,023 88,761 89,571 64,103 Australia-NZ-S Africa 504 1,146 1,168 57,687 25,350 22,950 14,608 Japan-Korea-Taiwan 841 3,282 4,601 141,539 54,862 63,446 46,844 World public agricultural R&D 7,455 26,197 42,321 1,188,035 465,171 496,094 376,806 Developed country share 0.66 0.61 0.47 0.59 0.62 0.59 0.55 Private agricultural R&D 2,613 7,202 12,939 330,680 132,787 133,616 96,646 CGIAR and precursors 1 305 707 11,831 7,791 7,791 5,514 Total world - all sources 10,069 33,703 55,966 1,530,546 605,749 637,501 478,966 Region R&D expenditure R&D Stock in 2011 1 1961 1991 2011 Total R&D spending, 1962–2011 50-year gamma 35-year gamma 25-year gamma Public agricultural R&D Developing Regions Central America 123 550 866 24,038 9,879 10,336 7,861 South America 812 2,943 3,824 114,946 48,559 48,143 32,879 China 313 1,599 7,768 99,813 27,663 40,118 42,488 Southeast Asia 235 1,204 2,005 57,138 22,060 25,689 19,759 South Asia 278 1,461 3,798 71,658 23,249 31,913 28,482 West Asia 148 615 1,524 36,461 12,869 15,053 12,891 North Africa 119 395 728 20,775 7,663 9,206 7,304 Sub-Saharan Africa 522 1,439 1,910 59,837 24,394 23,407 17,214 Developed Regions Former Soviet Union 336 455 629 24,940 10,316 7,515 5,569 Eastern Europe 2 238 557 904 31,539 13,442 10,936 7,758 Western Europe 1,274 5,570 7,093 230,639 96,103 97,810 69,046 Canada-USA 1,711 4,980 5,501 217,023 88,761 89,571 64,103 Australia-NZ-S Africa 504 1,146 1,168 57,687 25,350 22,950 14,608 Japan-Korea-Taiwan 841 3,282 4,601 141,539 54,862 63,446 46,844 World public agricultural R&D 7,455 26,197 42,321 1,188,035 465,171 496,094 376,806 Developed country share 0.66 0.61 0.47 0.59 0.62 0.59 0.55 Private agricultural R&D 2,613 7,202 12,939 330,680 132,787 133,616 96,646 CGIAR and precursors 1 305 707 11,831 7,791 7,791 5,514 Total world - all sources 10,069 33,703 55,966 1,530,546 605,749 637,501 478,966 Note: Superscript 1 R&D capital stock is the aggregate amount of R&D expenditure contributing to productivity in 2011. Cumulative expenditure assumes no R&D lag or depreciation. The 50-year, 35-year, and 25-year gamma lag distributions assume a gestation period and eventual depreciation of R&D capita, according to the R&D capital-life profiles shown in figure 1. 2 Eastern Europe includes the transition economies of Poland, Hungary, Romania, Bulgaria, former Czechoslovakia, and former Yugoslavia. Sources: Data on R&D spending were compiled from multiple sources. Public R&D in high-income countries is from Heisey and Fuglie (forthcoming), except for the former Soviet Union and Eastern Europe, which are from Judd, Boyce, and Evenson (1991) and OECD; public R&D in developing countries since 1981 is from Agricultural Research and Technology Indicators, and prior to 1981 from Pardey and Roseboom (1989) and Pardey, Roseboom, Anderson, (1991). Private R&D since 1990 is from Fuglie (2016) and pre-1990 from Fuglie et al. (2011). CGIAR R&D since 1971 is from Agricultural Research and Technology Indicators and pre-1971 from Alston, Dehmer, and Pardey (2006). Some additional publications were consulted for specific countries and some extrapolations were made for missing data. Contact the author for a complete description of sources and data. Table 1 Agricultural R&D Spending and R&D Capital by World Region (Constant 2011 PPP$, million) Region R&D expenditure R&D Stock in 2011 1 1961 1991 2011 Total R&D spending, 1962–2011 50-year gamma 35-year gamma 25-year gamma Public agricultural R&D Developing Regions Central America 123 550 866 24,038 9,879 10,336 7,861 South America 812 2,943 3,824 114,946 48,559 48,143 32,879 China 313 1,599 7,768 99,813 27,663 40,118 42,488 Southeast Asia 235 1,204 2,005 57,138 22,060 25,689 19,759 South Asia 278 1,461 3,798 71,658 23,249 31,913 28,482 West Asia 148 615 1,524 36,461 12,869 15,053 12,891 North Africa 119 395 728 20,775 7,663 9,206 7,304 Sub-Saharan Africa 522 1,439 1,910 59,837 24,394 23,407 17,214 Developed Regions Former Soviet Union 336 455 629 24,940 10,316 7,515 5,569 Eastern Europe 2 238 557 904 31,539 13,442 10,936 7,758 Western Europe 1,274 5,570 7,093 230,639 96,103 97,810 69,046 Canada-USA 1,711 4,980 5,501 217,023 88,761 89,571 64,103 Australia-NZ-S Africa 504 1,146 1,168 57,687 25,350 22,950 14,608 Japan-Korea-Taiwan 841 3,282 4,601 141,539 54,862 63,446 46,844 World public agricultural R&D 7,455 26,197 42,321 1,188,035 465,171 496,094 376,806 Developed country share 0.66 0.61 0.47 0.59 0.62 0.59 0.55 Private agricultural R&D 2,613 7,202 12,939 330,680 132,787 133,616 96,646 CGIAR and precursors 1 305 707 11,831 7,791 7,791 5,514 Total world - all sources 10,069 33,703 55,966 1,530,546 605,749 637,501 478,966 Region R&D expenditure R&D Stock in 2011 1 1961 1991 2011 Total R&D spending, 1962–2011 50-year gamma 35-year gamma 25-year gamma Public agricultural R&D Developing Regions Central America 123 550 866 24,038 9,879 10,336 7,861 South America 812 2,943 3,824 114,946 48,559 48,143 32,879 China 313 1,599 7,768 99,813 27,663 40,118 42,488 Southeast Asia 235 1,204 2,005 57,138 22,060 25,689 19,759 South Asia 278 1,461 3,798 71,658 23,249 31,913 28,482 West Asia 148 615 1,524 36,461 12,869 15,053 12,891 North Africa 119 395 728 20,775 7,663 9,206 7,304 Sub-Saharan Africa 522 1,439 1,910 59,837 24,394 23,407 17,214 Developed Regions Former Soviet Union 336 455 629 24,940 10,316 7,515 5,569 Eastern Europe 2 238 557 904 31,539 13,442 10,936 7,758 Western Europe 1,274 5,570 7,093 230,639 96,103 97,810 69,046 Canada-USA 1,711 4,980 5,501 217,023 88,761 89,571 64,103 Australia-NZ-S Africa 504 1,146 1,168 57,687 25,350 22,950 14,608 Japan-Korea-Taiwan 841 3,282 4,601 141,539 54,862 63,446 46,844 World public agricultural R&D 7,455 26,197 42,321 1,188,035 465,171 496,094 376,806 Developed country share 0.66 0.61 0.47 0.59 0.62 0.59 0.55 Private agricultural R&D 2,613 7,202 12,939 330,680 132,787 133,616 96,646 CGIAR and precursors 1 305 707 11,831 7,791 7,791 5,514 Total world - all sources 10,069 33,703 55,966 1,530,546 605,749 637,501 478,966 Note: Superscript 1 R&D capital stock is the aggregate amount of R&D expenditure contributing to productivity in 2011. Cumulative expenditure assumes no R&D lag or depreciation. The 50-year, 35-year, and 25-year gamma lag distributions assume a gestation period and eventual depreciation of R&D capita, according to the R&D capital-life profiles shown in figure 1. 2 Eastern Europe includes the transition economies of Poland, Hungary, Romania, Bulgaria, former Czechoslovakia, and former Yugoslavia. Sources: Data on R&D spending were compiled from multiple sources. Public R&D in high-income countries is from Heisey and Fuglie (forthcoming), except for the former Soviet Union and Eastern Europe, which are from Judd, Boyce, and Evenson (1991) and OECD; public R&D in developing countries since 1981 is from Agricultural Research and Technology Indicators, and prior to 1981 from Pardey and Roseboom (1989) and Pardey, Roseboom, Anderson, (1991). Private R&D since 1990 is from Fuglie (2016) and pre-1990 from Fuglie et al. (2011). CGIAR R&D since 1971 is from Agricultural Research and Technology Indicators and pre-1971 from Alston, Dehmer, and Pardey (2006). Some additional publications were consulted for specific countries and some extrapolations were made for missing data. Contact the author for a complete description of sources and data. Over the 50 years from 1962 to 2011, total global spending for public agricultural R&D was $1,192 billion in constant 2011 PPP$. Applying the 50-year lag structure from figure 1 yields a 2011 R&D capital stock of $PPP 467 billion. In other words, about half of the R&D since 1962 was either obsolete or still in the development and diffusion pipeline. The 35-year lag structure for R&D capital yields a larger global R&D capital stock of $PPP 496 billion. Even though this R&D capital is shorter-lived than the estimate using a 50-year lag structure, it includes more of the spending that occurred since 1990 due to its shorter gestation period. Because much of the current R&D capital is based on R&D investment made two to four decades ago, the R&D capital shares for developed countries are larger than their R&D spending shares. In 2011, developed countries accounted for 55%, 59% or 62% of global public agricultural R&D capital (using the 25-,35-, and 50-year lag distributions), but only 47% of global R&D spending. Even if current expenditure shares were to remain constant over the coming decades, the R&D capital share of developing countries would rise (eventually matching their expenditure share) as their recent R&D spending matures and the technologies arising from it are adopted by farmers. The divergence between public R&D expenditure and capital shares is especially prominent for China. Chinese government investment in agricultural R&D grew by more than 10% per year between 2000 and 2011 in real terms. By 2008, China had overtaken the United States as the largest national investor in agricultural R&D, and by 2011 it accounted for about 18% of world (public) agricultural R&D spending. However, China’s global R&D capital share has yet to reflect this growth because much of this R&D spending has yet to translate into farm productivity. North American R&D capital share, on the other hand, is significantly larger than its expenditure share. This reflects the historical role of the United States as a world leader in agricultural science and technology. But as its public R&D capital ages and R&D spending stagnates, the United States is being overtaken by China and others. Including private-sector research in global R&D capital shares would likely shift the balance back somewhat to developed countries. However, as noted above, it is difficult to determine with much precision how to apportion private R&D to individual countries. Using the estimated global series for private agricultural R&D and a 35-year R&D lag structure, private agricultural R&D accounted for 23% of global expenditures and 21% of R&D capital in 2011. The CGIAR accounted for about 1.2% of both global R&D expenditures and capital stocks. Research and Development Capital and Agricultural Growth The evidence linking R&D investment to productivity growth in agriculture is compelling, whether assessed for specific commodities, at the sector level for a country, or through international comparisons. Studies comparing the long-term performance of national agricultural sectors have consistently found that countries that invested more in agricultural R&D achieved higher agricultural productivity growth (Evenson and Kislev 1975; Craig, Pardey, and Roseboom 1997; Thirtle, Lin, and Piesse 2003; Gutierrez and Gutierrez 2003; Evenson and Fuglie 2010). Moreover, the value of the productivity improvement has been large relative to the cost of the R&D. Hurley et al. (2014) provide a comprehensive, critical assessment of 372 studies on returns to agricultural R&D and find a median (social) internal rate of return of 39%. Evidence on R&D-to-TFP Elasticities in Agriculture Table 2 summarizes results from 44 studies that econometrically estimated agricultural R&D-to-TFP elasticities for a country or set of countries. Several criteria distinguish this set of studies from other studies on agricultural growth. First, they are all based on times series evidence that includes relatively recent years. All of them included data on TFP to at least 1980, with about half of the studies extending to the post-2000 era. Second, all studies use a measure of knowledge capital that is based on accumulated R&D investment (measured by R&D spending or scientist-years). This is an advance over older work that had to rely on proxies for R&D capital, such as Hayami and Ruttan (1985), who used the number of graduates from technical schools, and Evenson and Kislev (1975) who used counts of agricultural science publications. A third important feature of most (31 out of 44) of these studies is that they use panel data time series of cross-sections of countries or sub-regions (states, provinces, or districts) within countries. This substantially increases the degrees of freedom and explanatory power of the models. Fourth, nearly all use whole sector agricultural TFP as their explanatory variable, rather than productivity of just one or a few commodities. Exceptions are Evenson (2003), who quantified the impact of the CGIAR research on food crop productivity, Suphannachart and Warr (2012), who examined the productivity of Thailand’s crop and livestock sectors separately, and Jin et al. (2002), who focused on sources of TFP growth for China’s three principal crops (rice, wheat, and maize). Finally, the studies are geographically diverse: 26 address productivity in developing countries (including separate coverage of Asia, Latin America, and Africa); 14 focus on developed countries, one covers (previously) centrally-planned economies, and three have worldwide coverage that includes both developed and developing countries. Table 2 Estimates of Agricultural R&D Elasticities Study Geographic coverage1 Period Data R&D Elasticities Total -all sources National public Int'l spill-in CGIAR Private Craig et al. (1997) World 1965-1990 88-country panel 0.10 0.10 Wiebe et al. (2000) World 1961-1997 88-country panel 0.16 0.16 Gutierrez et al. (2003)2 World 1970-1992 47-country panel 0.88 0.25 0.63 Schimmel. et al. (1999) EU & USA 1973-1993 11-country panel 0.31 0.11 0.20 0.00 Jin et al. (2016) USA 1970-2002 48-state panel 0.25 0.25 Wang et al. (2012) USA 1960-2002 48-state panel 0.29 0.29 Alston et al. (2011) USA 1949-2002 47-state panel 0.29 0.29 Huffman et al. (2006) USA 1970-1999 48-state panel 0.47 0.35 0.11 Andersen et al. (2013) USA 1949-2002 National 0.37 0.37 Wang et al. (2013) USA 1970-2009 National 0.57 0.43 0.14 Baldos et al. (2015) USA 1949-2011 National 0.31 0.31 Sheng et al. (2011) Australia 1953-2007 National 0.23 (national & foreign R&D combined) Mullen et al. (1995) Australia 1953-1988 National 0.26 0.26 Hall et al. (2006) New Zealand 1927-2001 National 0.50 0.15 0.00 Thirtle et al. (2008) UK 1953-2005 National 0.61 0.23 0.37 Bouchet et al. (1983) France 1959-1984 National 0.75 0.36 0.28 0.11 Butault et al. (2015) France 1959-2012 National 0.16 (national & foreign R&D combined) Wong (1986) Socialist 1950-1980 8-country panel 0.07 0.07 Johnson et al. (2000) LDC 1960s, 70s, 80s 90-country panel 0.13 0.03 0.10 Fulginiti et al. (1993) LDC 1961-1984 18-country panel 0.07 0.07 Craig et al. (1997) LDC 1965-1990 67-country panel 0.09 0.09 Thirtle et al. (2003) LDC 1985,90,95 48-country panel 0.44 0.44 Fan et al. (1998) Asia 1972-1993 12-country panel 0.17 0.17 Thirtle et al. (2003) Asia 1985,90,96 11-country panel 0.34 0.34 Evenson (2003)3 Asia 1970-2000 10 food crops 0.14 Evenson et al. (1991) Philippines 1948-1984 9-region panel 0.31 0.31 Rada & Fuglie (2012) Indonesia 1985-2005 22-province panel 0.36 0.27 0.09 Suphan. et al. (2012) Thailand 1971-2006 National 0.20 0.17 0.04 Jin et al. (2002) China 1981-1995 16-province panel 0.37 0.33 0.04 Fan (2000) China 1975-1997 25-province panel 0.25 0.25 Fan et al. (2002) China 1970-1997 29-province panel 0.09 0.09 Pray et al. (1991) Bangladesh 1947-1981 National 0.12 0.12 0.004 Rahman et al. (2013) Bangladesh 1948-2008 National 0.13 0.13 Fan et al. (2000) India 1970-1993 17-state panel 0.30 0.30 Evenson et al. (1999) India 1956-1987 271-district panel 0.17 0.05 0.11 0.01 Rada et al. (2015) India 1980-2008 16-state panel 0.28 0.17 0.11 Thirtle et al. (2003) LAC 1985,1990 15-country panel 0.20 0.20 Evenson (2003)3 LAC 1970-2000 10 food crops 0.05 Fernandez- et al. (1997) Mexico 1960-1990 National 0.64 0.13 0.36 0.14 Rada & Buccola (2012) Brazil 1985,96,06 558-district panel 0.03 0.03 Bervejillo et al. (2012) Uruguay 1981-2000 National 0.68 0.57 0.12 Thirtle et al. (1995) Africa 1971-1986 22-country panel 0.02 0.02 Thirtle et al. (2003) Africa 1985,1990 22-country panel 0.36 0.36 Lusigi et al. (1997) Africa 1961-1991 47-country panel 0.05 0.02 0.03 Evenson (2003) WANA 1970-2000 10 food crops 0.07 Fan et al. (2006) Egypt 1980-2000 3-region panel 0.25 0.25 Frisvold et al. (1995) SSA 1973-1985 28-country panel 0.08 0.08 Block (2014) SSA 1981-2000 27-country panel 0.20 0.20 Alene (2010) SSA 1986-2004 15-country panel 0.20 0.20 Fuglie et al. (2013) SSA 1977-2005 32-country panel 0.08 0.04 0.04 Evenson (2003)3 SSA 1970-2000 10 food crops 0.03 Study Geographic coverage1 Period Data R&D Elasticities Total -all sources National public Int'l spill-in CGIAR Private Craig et al. (1997) World 1965-1990 88-country panel 0.10 0.10 Wiebe et al. (2000) World 1961-1997 88-country panel 0.16 0.16 Gutierrez et al. (2003)2 World 1970-1992 47-country panel 0.88 0.25 0.63 Schimmel. et al. (1999) EU & USA 1973-1993 11-country panel 0.31 0.11 0.20 0.00 Jin et al. (2016) USA 1970-2002 48-state panel 0.25 0.25 Wang et al. (2012) USA 1960-2002 48-state panel 0.29 0.29 Alston et al. (2011) USA 1949-2002 47-state panel 0.29 0.29 Huffman et al. (2006) USA 1970-1999 48-state panel 0.47 0.35 0.11 Andersen et al. (2013) USA 1949-2002 National 0.37 0.37 Wang et al. (2013) USA 1970-2009 National 0.57 0.43 0.14 Baldos et al. (2015) USA 1949-2011 National 0.31 0.31 Sheng et al. (2011) Australia 1953-2007 National 0.23 (national & foreign R&D combined) Mullen et al. (1995) Australia 1953-1988 National 0.26 0.26 Hall et al. (2006) New Zealand 1927-2001 National 0.50 0.15 0.00 Thirtle et al. (2008) UK 1953-2005 National 0.61 0.23 0.37 Bouchet et al. (1983) France 1959-1984 National 0.75 0.36 0.28 0.11 Butault et al. (2015) France 1959-2012 National 0.16 (national & foreign R&D combined) Wong (1986) Socialist 1950-1980 8-country panel 0.07 0.07 Johnson et al. (2000) LDC 1960s, 70s, 80s 90-country panel 0.13 0.03 0.10 Fulginiti et al. (1993) LDC 1961-1984 18-country panel 0.07 0.07 Craig et al. (1997) LDC 1965-1990 67-country panel 0.09 0.09 Thirtle et al. (2003) LDC 1985,90,95 48-country panel 0.44 0.44 Fan et al. (1998) Asia 1972-1993 12-country panel 0.17 0.17 Thirtle et al. (2003) Asia 1985,90,96 11-country panel 0.34 0.34 Evenson (2003)3 Asia 1970-2000 10 food crops 0.14 Evenson et al. (1991) Philippines 1948-1984 9-region panel 0.31 0.31 Rada & Fuglie (2012) Indonesia 1985-2005 22-province panel 0.36 0.27 0.09 Suphan. et al. (2012) Thailand 1971-2006 National 0.20 0.17 0.04 Jin et al. (2002) China 1981-1995 16-province panel 0.37 0.33 0.04 Fan (2000) China 1975-1997 25-province panel 0.25 0.25 Fan et al. (2002) China 1970-1997 29-province panel 0.09 0.09 Pray et al. (1991) Bangladesh 1947-1981 National 0.12 0.12 0.004 Rahman et al. (2013) Bangladesh 1948-2008 National 0.13 0.13 Fan et al. (2000) India 1970-1993 17-state panel 0.30 0.30 Evenson et al. (1999) India 1956-1987 271-district panel 0.17 0.05 0.11 0.01 Rada et al. (2015) India 1980-2008 16-state panel 0.28 0.17 0.11 Thirtle et al. (2003) LAC 1985,1990 15-country panel 0.20 0.20 Evenson (2003)3 LAC 1970-2000 10 food crops 0.05 Fernandez- et al. (1997) Mexico 1960-1990 National 0.64 0.13 0.36 0.14 Rada & Buccola (2012) Brazil 1985,96,06 558-district panel 0.03 0.03 Bervejillo et al. (2012) Uruguay 1981-2000 National 0.68 0.57 0.12 Thirtle et al. (1995) Africa 1971-1986 22-country panel 0.02 0.02 Thirtle et al. (2003) Africa 1985,1990 22-country panel 0.36 0.36 Lusigi et al. (1997) Africa 1961-1991 47-country panel 0.05 0.02 0.03 Evenson (2003) WANA 1970-2000 10 food crops 0.07 Fan et al. (2006) Egypt 1980-2000 3-region panel 0.25 0.25 Frisvold et al. (1995) SSA 1973-1985 28-country panel 0.08 0.08 Block (2014) SSA 1981-2000 27-country panel 0.20 0.20 Alene (2010) SSA 1986-2004 15-country panel 0.20 0.20 Fuglie et al. (2013) SSA 1977-2005 32-country panel 0.08 0.04 0.04 Evenson (2003)3 SSA 1970-2000 10 food crops 0.03 Note: Superscript 1 indicates that LAC=Latin America & Caribbean; SSA=Sub-Saharan Africa, WANA=West Asia & North Africa, LDC=developing countries, DC=developed countries. 2 The Gutierrez et al. elasticity for international spill-ins was judged to be an outlier. 3 Evenson estimated elasticities of CGIAR R&D on yield for ten food crops. The elasticity of CGIAR R&D on agricultural TFP is the average food crop R&D elasticity times the share of these food crops in the total gross agricultural output over 1970-2000 (FAOSTAT). Table 2 Estimates of Agricultural R&D Elasticities Study Geographic coverage1 Period Data R&D Elasticities Total -all sources National public Int'l spill-in CGIAR Private Craig et al. (1997) World 1965-1990 88-country panel 0.10 0.10 Wiebe et al. (2000) World 1961-1997 88-country panel 0.16 0.16 Gutierrez et al. (2003)2 World 1970-1992 47-country panel 0.88 0.25 0.63 Schimmel. et al. (1999) EU & USA 1973-1993 11-country panel 0.31 0.11 0.20 0.00 Jin et al. (2016) USA 1970-2002 48-state panel 0.25 0.25 Wang et al. (2012) USA 1960-2002 48-state panel 0.29 0.29 Alston et al. (2011) USA 1949-2002 47-state panel 0.29 0.29 Huffman et al. (2006) USA 1970-1999 48-state panel 0.47 0.35 0.11 Andersen et al. (2013) USA 1949-2002 National 0.37 0.37 Wang et al. (2013) USA 1970-2009 National 0.57 0.43 0.14 Baldos et al. (2015) USA 1949-2011 National 0.31 0.31 Sheng et al. (2011) Australia 1953-2007 National 0.23 (national & foreign R&D combined) Mullen et al. (1995) Australia 1953-1988 National 0.26 0.26 Hall et al. (2006) New Zealand 1927-2001 National 0.50 0.15 0.00 Thirtle et al. (2008) UK 1953-2005 National 0.61 0.23 0.37 Bouchet et al. (1983) France 1959-1984 National 0.75 0.36 0.28 0.11 Butault et al. (2015) France 1959-2012 National 0.16 (national & foreign R&D combined) Wong (1986) Socialist 1950-1980 8-country panel 0.07 0.07 Johnson et al. (2000) LDC 1960s, 70s, 80s 90-country panel 0.13 0.03 0.10 Fulginiti et al. (1993) LDC 1961-1984 18-country panel 0.07 0.07 Craig et al. (1997) LDC 1965-1990 67-country panel 0.09 0.09 Thirtle et al. (2003) LDC 1985,90,95 48-country panel 0.44 0.44 Fan et al. (1998) Asia 1972-1993 12-country panel 0.17 0.17 Thirtle et al. (2003) Asia 1985,90,96 11-country panel 0.34 0.34 Evenson (2003)3 Asia 1970-2000 10 food crops 0.14 Evenson et al. (1991) Philippines 1948-1984 9-region panel 0.31 0.31 Rada & Fuglie (2012) Indonesia 1985-2005 22-province panel 0.36 0.27 0.09 Suphan. et al. (2012) Thailand 1971-2006 National 0.20 0.17 0.04 Jin et al. (2002) China 1981-1995 16-province panel 0.37 0.33 0.04 Fan (2000) China 1975-1997 25-province panel 0.25 0.25 Fan et al. (2002) China 1970-1997 29-province panel 0.09 0.09 Pray et al. (1991) Bangladesh 1947-1981 National 0.12 0.12 0.004 Rahman et al. (2013) Bangladesh 1948-2008 National 0.13 0.13 Fan et al. (2000) India 1970-1993 17-state panel 0.30 0.30 Evenson et al. (1999) India 1956-1987 271-district panel 0.17 0.05 0.11 0.01 Rada et al. (2015) India 1980-2008 16-state panel 0.28 0.17 0.11 Thirtle et al. (2003) LAC 1985,1990 15-country panel 0.20 0.20 Evenson (2003)3 LAC 1970-2000 10 food crops 0.05 Fernandez- et al. (1997) Mexico 1960-1990 National 0.64 0.13 0.36 0.14 Rada & Buccola (2012) Brazil 1985,96,06 558-district panel 0.03 0.03 Bervejillo et al. (2012) Uruguay 1981-2000 National 0.68 0.57 0.12 Thirtle et al. (1995) Africa 1971-1986 22-country panel 0.02 0.02 Thirtle et al. (2003) Africa 1985,1990 22-country panel 0.36 0.36 Lusigi et al. (1997) Africa 1961-1991 47-country panel 0.05 0.02 0.03 Evenson (2003) WANA 1970-2000 10 food crops 0.07 Fan et al. (2006) Egypt 1980-2000 3-region panel 0.25 0.25 Frisvold et al. (1995) SSA 1973-1985 28-country panel 0.08 0.08 Block (2014) SSA 1981-2000 27-country panel 0.20 0.20 Alene (2010) SSA 1986-2004 15-country panel 0.20 0.20 Fuglie et al. (2013) SSA 1977-2005 32-country panel 0.08 0.04 0.04 Evenson (2003)3 SSA 1970-2000 10 food crops 0.03 Study Geographic coverage1 Period Data R&D Elasticities Total -all sources National public Int'l spill-in CGIAR Private Craig et al. (1997) World 1965-1990 88-country panel 0.10 0.10 Wiebe et al. (2000) World 1961-1997 88-country panel 0.16 0.16 Gutierrez et al. (2003)2 World 1970-1992 47-country panel 0.88 0.25 0.63 Schimmel. et al. (1999) EU & USA 1973-1993 11-country panel 0.31 0.11 0.20 0.00 Jin et al. (2016) USA 1970-2002 48-state panel 0.25 0.25 Wang et al. (2012) USA 1960-2002 48-state panel 0.29 0.29 Alston et al. (2011) USA 1949-2002 47-state panel 0.29 0.29 Huffman et al. (2006) USA 1970-1999 48-state panel 0.47 0.35 0.11 Andersen et al. (2013) USA 1949-2002 National 0.37 0.37 Wang et al. (2013) USA 1970-2009 National 0.57 0.43 0.14 Baldos et al. (2015) USA 1949-2011 National 0.31 0.31 Sheng et al. (2011) Australia 1953-2007 National 0.23 (national & foreign R&D combined) Mullen et al. (1995) Australia 1953-1988 National 0.26 0.26 Hall et al. (2006) New Zealand 1927-2001 National 0.50 0.15 0.00 Thirtle et al. (2008) UK 1953-2005 National 0.61 0.23 0.37 Bouchet et al. (1983) France 1959-1984 National 0.75 0.36 0.28 0.11 Butault et al. (2015) France 1959-2012 National 0.16 (national & foreign R&D combined) Wong (1986) Socialist 1950-1980 8-country panel 0.07 0.07 Johnson et al. (2000) LDC 1960s, 70s, 80s 90-country panel 0.13 0.03 0.10 Fulginiti et al. (1993) LDC 1961-1984 18-country panel 0.07 0.07 Craig et al. (1997) LDC 1965-1990 67-country panel 0.09 0.09 Thirtle et al. (2003) LDC 1985,90,95 48-country panel 0.44 0.44 Fan et al. (1998) Asia 1972-1993 12-country panel 0.17 0.17 Thirtle et al. (2003) Asia 1985,90,96 11-country panel 0.34 0.34 Evenson (2003)3 Asia 1970-2000 10 food crops 0.14 Evenson et al. (1991) Philippines 1948-1984 9-region panel 0.31 0.31 Rada & Fuglie (2012) Indonesia 1985-2005 22-province panel 0.36 0.27 0.09 Suphan. et al. (2012) Thailand 1971-2006 National 0.20 0.17 0.04 Jin et al. (2002) China 1981-1995 16-province panel 0.37 0.33 0.04 Fan (2000) China 1975-1997 25-province panel 0.25 0.25 Fan et al. (2002) China 1970-1997 29-province panel 0.09 0.09 Pray et al. (1991) Bangladesh 1947-1981 National 0.12 0.12 0.004 Rahman et al. (2013) Bangladesh 1948-2008 National 0.13 0.13 Fan et al. (2000) India 1970-1993 17-state panel 0.30 0.30 Evenson et al. (1999) India 1956-1987 271-district panel 0.17 0.05 0.11 0.01 Rada et al. (2015) India 1980-2008 16-state panel 0.28 0.17 0.11 Thirtle et al. (2003) LAC 1985,1990 15-country panel 0.20 0.20 Evenson (2003)3 LAC 1970-2000 10 food crops 0.05 Fernandez- et al. (1997) Mexico 1960-1990 National 0.64 0.13 0.36 0.14 Rada & Buccola (2012) Brazil 1985,96,06 558-district panel 0.03 0.03 Bervejillo et al. (2012) Uruguay 1981-2000 National 0.68 0.57 0.12 Thirtle et al. (1995) Africa 1971-1986 22-country panel 0.02 0.02 Thirtle et al. (2003) Africa 1985,1990 22-country panel 0.36 0.36 Lusigi et al. (1997) Africa 1961-1991 47-country panel 0.05 0.02 0.03 Evenson (2003) WANA 1970-2000 10 food crops 0.07 Fan et al. (2006) Egypt 1980-2000 3-region panel 0.25 0.25 Frisvold et al. (1995) SSA 1973-1985 28-country panel 0.08 0.08 Block (2014) SSA 1981-2000 27-country panel 0.20 0.20 Alene (2010) SSA 1986-2004 15-country panel 0.20 0.20 Fuglie et al. (2013) SSA 1977-2005 32-country panel 0.08 0.04 0.04 Evenson (2003)3 SSA 1970-2000 10 food crops 0.03 Note: Superscript 1 indicates that LAC=Latin America & Caribbean; SSA=Sub-Saharan Africa, WANA=West Asia & North Africa, LDC=developing countries, DC=developed countries. 2 The Gutierrez et al. elasticity for international spill-ins was judged to be an outlier. 3 Evenson estimated elasticities of CGIAR R&D on yield for ten food crops. The elasticity of CGIAR R&D on agricultural TFP is the average food crop R&D elasticity times the share of these food crops in the total gross agricultural output over 1970-2000 (FAOSTAT). To make the R&D elasticities reported in table 2 as comparable as possible, in some cases the econometric results reported by the studies were adjusted. One adjustment is to report a “total elasticity” for public agricultural R&D, which is simply the sum of elasticities for different parts of the public research system. For example, Alston et al. (2011) report R&D elasticities for the effects of state research on own-state productivity (0.15), R&D spill-ins from other states (0.07), and spill-ins from federal-level intramural research carried out by the USDA (0.07).7 7 These research elasticities are taken from the authors’ preferred model 1 (Alston et al. 2011, table 2) and do not include the effects of extension. The percentage change in national agricultural TFP from a 1% increase in the R&D capital of each state and the USDA would be the sum of these elasticities (0.29). A second adjustment is when the R&D elasticity reported by the study refers to only a sub-sector, such as research on food crops. In these cases, the elasticity is multiplied by the revenue share of the sub-sector so that it shows how food crop R&D affects the average TFP of the whole agricultural sector.8 8 Suppose the growth rate in agricultural TFP can be written as the revenue-weighted growth rate of its sub-sectors, that is, lnA=ϕlnAc+1-ϕlnAl, where Ac is the TFP indexes for food crops, Al is the TFP of the rest of the agricultural sector, and ϕ is the revenue share of food crops in total agricultural output. Evenson’s results relate R&D capital by the CGIAR to productivity growth in the food crop sector only. His results give an estimate of γ in the equation Ac=Scgiarγ. Substituting this into the previous equation and taking the derivate with respect to lnScgiar gives ϕγ, which is an estimate of how CGIAR R&D capital affects TFP growth of the whole agricultural sector. Finally, two studies—Bouchet, Orden, and Norton (1983) on France, and Fernandez-Cornejo and Shumway (1997) on Mexico—used U.S. agricultural TFP as an explanatory variable for spill-ins of technologies developed in the United States and adopted in these countries. The U.S. agricultural TFP is itself a function of R&D spending by the U.S. public and private sectors, and in order to relate French and Mexican agricultural productivity directly to these external R&D capital stocks, I adjust the econometric estimates reported by these studies as described in the following paragraphs. First, we note that what the raw econometric results give us is an elasticity for national public agricultural R&D and total spill-ins from the United States. Using France as an example, this can be written as follows: Afr=Sfr-publicδ1,frAusaβ (5) where Afr and Ausa are agricultural TFP indexes for France and the United States, respectively, Sfr-public is French public R&D capital, and the parameters δ1,fr and β are the elasticities associated with the right-hand-side variables. Second, taking the average values from the U.S. productivity studies in table 2, we have the following relationship for the effects of public and private R&D capital on U.S. agricultural TFP: Ausa=Susa-public0.322Sprivate0.127 (6) Substituting equation (6) into equation (5) gives Afr=Sfr-publicδ1,frSusa-public0.322Sprivate0.127β. (7) Thus, the effect that U.S. public (private) R&D capital has on French agricultural productivity is found by multiplying the estimated coefficient β by 0.322 (0.127). For example, in Bouchet, Orden, and Norton’s (1983) study of agricultural growth in France, the authors found that each 1% increase in U.S. agriculture TFP was associated with a 0.857% rise in French agricultural TFP, which they associated with French farmers adopting innovations imported from the United States. Using the above procedure implies a spillover elasticity of 0.857 * 0.322 = 0.28 for USA public R&D capital, and 0.857 * 0.127 = 0.11 for private R&D on French agricultural TFP. These adjusted elasticities are reported in table 2. Given the scant data available on private R&D, some studies have used patent counts as a proxy for this variable. Patents are an R&D intermediate output rather than an R&D input, and in principle a procedure like the one in equations (5)–(7) should be used to adjust the elasticities accordingly. However, lacking a quantified relationship between R&D spending and patent output, we are forced to assume that these measures are perfectly correlated. The elasticities for private R&D reported in table 2 for the United Kingdom and New Zealand are based on patent counts rather than R&D spending. One general finding from the studies listed in table 2 is that public R&D investment has been the dominant source of agricultural TFP growth around the world. All of the studies in table 2, whether focusing on a particular country or comparing growth across countries, found that national public R&D explained a significant share of the growth in agricultural TFP. While these studies generally treat spill-in or private R&D (if modeled at all) as independent sources of technology, it is very likely that national research institutions help to adapt and disseminate these technologies locally. Fuglie and Rada (2013), for example, found that CGIAR technologies spread more rapidly in African countries with more national agricultural R&D capital. Studies have also found complementarities between public and private agricultural research: one sector’s R&D appears to raise the returns to the research of the other sector, presumably because they specialize in complementary parts of the science-technology spectrum (Schimmelpfennig and Thirtle 1999; Fuglie and Toole 2014). A second result is that all of the total R&D elasticities in table 2 are less than 1. This implies that total R&D spending will tend to rise faster than productivity growth. This is consistent with the finding that as countries develop their agricultural sectors, they experience a rise in research intensity (the ratio of agricultural R&D to agricultural GDP; Pardey et al. 2016; Heisey and Fuglie, forthcoming). This is strongly at odds with the assumptions of New Growth Theory, where it is assumed that a constant level of R&D spending will generate a constant growth rate for TFP (see Romer 1990, and footnote 3). A third conclusion is that, despite environmental constraints, international technology transfer is an important source of agricultural TFP growth. With the exception of Gutierrez and Gutierrez (2003), the studies in table 2 find that international R&D spill-ins have occurred mainly among developed countries located in temperate climates. Gutierrez and Gutierrez (2003) constructed a variable of the foreign R&D available for a country by taking the weighted average of the domestic agricultural R&D of its trade partners, using total import shares as weights. It is unclear why total import share should correlate with agricultural technology transfer, other than the fact that both are likely to be higher among nearby countries. In contrast, using agricultural patent data, Johnson and Evenson (1999) find that most international technology transfer occurs between high-income, temperate countries, and Eberhardt and Teal (2013) find that agricultural TFP growth is strongly correlated across similar agro-climatic environments. The spill-in elasticity reported by Gutierrez and Gutierrez (2003) is an outlier compared with the other spill-in elasticities in table 2, but it suggests the need for more work on understanding the role of international technology transfer in agricultural productivity growth. A fourth conclusion from table 2 is that there appears to be systematic variation in the elasticities of R&D among global regions. The regional variation can be more readily seen in table 3, which gives the averages of the elasticities from table 2 for various developed and developing regions. These average elasticities are consistent with the following observations: Table 3 Average Agricultural R&D Elasticities by Region Region/Sub-region World agricultural output share Source of R&D Capital Total - all sources National public Int'l public spill-in CGIAR Private World 1.000 0.43 0.18 0.10 0.04 0.10 Developed 0.283 0.67 0.27 0.21 0.20  North America 0.126 0.63 0.30 0.20 0.13  Western Europe 0.116 0.72 0.23 0.24 0.24  Australia-NZ-S  Africa 0.023 0.64 0.18 0.12 0.35  Japan-Korea- Taiwan 0.018 (no region-specific elasticities available) Transition 0.088 0.07 0.07 Developing 0.628 0.38 0.18 0.07 0.07 0.07  Asia 0.399 0.30 0.21 0.08 0.01  Latin America 0.118 0.77 0.23 0.36 0.05 0.13  Africa & West  Asia 0.111 0.19 0.15 0.04  Sub-Saharan  Africa only 0.057 0.17 0.13 0.04 Region/Sub-region World agricultural output share Source of R&D Capital Total - all sources National public Int'l public spill-in CGIAR Private World 1.000 0.43 0.18 0.10 0.04 0.10 Developed 0.283 0.67 0.27 0.21 0.20  North America 0.126 0.63 0.30 0.20 0.13  Western Europe 0.116 0.72 0.23 0.24 0.24  Australia-NZ-S  Africa 0.023 0.64 0.18 0.12 0.35  Japan-Korea- Taiwan 0.018 (no region-specific elasticities available) Transition 0.088 0.07 0.07 Developing 0.628 0.38 0.18 0.07 0.07 0.07  Asia 0.399 0.30 0.21 0.08 0.01  Latin America 0.118 0.77 0.23 0.36 0.05 0.13  Africa & West  Asia 0.111 0.19 0.15 0.04  Sub-Saharan  Africa only 0.057 0.17 0.13 0.04 Note: Sub-region R&D elasticities are the simple average of the elasticities from the sub-region given in table 2. Region elasticities are a weighted average of elasticities from the sub-regions, where the weights are the average share of gross agricultural output over 1980–2010 (FAOSTAT). Table 3 Average Agricultural R&D Elasticities by Region Region/Sub-region World agricultural output share Source of R&D Capital Total - all sources National public Int'l public spill-in CGIAR Private World 1.000 0.43 0.18 0.10 0.04 0.10 Developed 0.283 0.67 0.27 0.21 0.20  North America 0.126 0.63 0.30 0.20 0.13  Western Europe 0.116 0.72 0.23 0.24 0.24  Australia-NZ-S  Africa 0.023 0.64 0.18 0.12 0.35  Japan-Korea- Taiwan 0.018 (no region-specific elasticities available) Transition 0.088 0.07 0.07 Developing 0.628 0.38 0.18 0.07 0.07 0.07  Asia 0.399 0.30 0.21 0.08 0.01  Latin America 0.118 0.77 0.23 0.36 0.05 0.13  Africa & West  Asia 0.111 0.19 0.15 0.04  Sub-Saharan  Africa only 0.057 0.17 0.13 0.04 Region/Sub-region World agricultural output share Source of R&D Capital Total - all sources National public Int'l public spill-in CGIAR Private World 1.000 0.43 0.18 0.10 0.04 0.10 Developed 0.283 0.67 0.27 0.21 0.20  North America 0.126 0.63 0.30 0.20 0.13  Western Europe 0.116 0.72 0.23 0.24 0.24  Australia-NZ-S  Africa 0.023 0.64 0.18 0.12 0.35  Japan-Korea- Taiwan 0.018 (no region-specific elasticities available) Transition 0.088 0.07 0.07 Developing 0.628 0.38 0.18 0.07 0.07 0.07  Asia 0.399 0.30 0.21 0.08 0.01  Latin America 0.118 0.77 0.23 0.36 0.05 0.13  Africa & West  Asia 0.111 0.19 0.15 0.04  Sub-Saharan  Africa only 0.057 0.17 0.13 0.04 Note: Sub-region R&D elasticities are the simple average of the elasticities from the sub-region given in table 2. Region elasticities are a weighted average of elasticities from the sub-regions, where the weights are the average share of gross agricultural output over 1980–2010 (FAOSTAT). The overall significance of R&D-led growth, given by the total elasticity of the effects of R&D from all sources, is higher for developed countries than developing countries. This is primarily due to stronger technological linkages to other countries and the private sector. R&D-led growth is least developed for Africa, due to relatively weak public R&D institutions, as well as the absence of strong linkages to private and international R&D, except for the CGIAR. In Latin America, along with national R&D, international R&D spill-ins and private R&D appear to have made significant contributions to agricultural productivity growth. For a recent review article on the role of the private sector in agricultural technology transfer and innovation in developing countries, see Pray and Fuglie (2015). Even though it is a relatively small component of the global agricultural R&D infrastructure and focuses heavily on staple food crops, the CGIAR has had a noticeable impact on aggregate agricultural TFP growth in Asia, Africa, and Latin America. R&D Capital and TFP Growth in World Agriculture, 1990–2011 One way to assess the dependence of agricultural productivity on R&D capital is to compare predicted and observed rates of agricultural TFP growth over some historical period. The model in equation (3) says that the rate of change in R&D capital stock multiplied by the R&D elasticity is the contribution of that R&D to agricultural TFP growth. Here, using the average R&D elasticities from table 3, I compare predicted with actual agricultural TFP growth rates for major global regions between 1990 and 2011. Applying the 35-year gamma lag structure from figure 1 to public, private, and CGIAR research expenditures from 1955–2011, I calculate the rate of change in R&D capital K^i,2011K^i,1990 for each technology source i between 1990 and 2011.9 9 The R&D expenditure series assembled for this paper start in 1960, expect for a few countries (the United States, United Kingdom, Netherlands, Japan, Australia, New Zealand, and Mexico), which go back earlier. For the rest, R&D spending is estimated back to 1955 assuming the 1960–1969 growth rate in R&D spending. For private R&D, Fuglie’s (2016) global estimates are extended back from 1990 using the private R&D growth rate by U.S. firms (Fuglie et al. 2011). Using the R&D elasticities ( δi) from table 2, the predicted rate of agricultural TFP growth over this period, lnA^2011A^1990, for a region is then given by lnA^2011A^1990=δ1lnK^1,2011K^1,1990+∑i=24δilnK^i,2011K^i,1990 (8) where the first right-hand-side term measures the productivity contribution of domestic R&D capital for all countries in the region, and the second term refers to technology spill-ins from public R&D in other countries, the private sector, and the CGIAR, respectively. For international R&D capital, I use the sum of public R&D capital in other developed countries, since studies in table 3 providing estimates of elasticities for international spill-ins refer primarily to spill-ins among the United States, Western Europe, and Oceania. Figure 2 shows the result of this exercise. The height of the bars indicate the predicted rate of agricultural TFP growth for a region between 1990 and 2011, with the different segments of the bar referring to the contributions from each of the four sources of R&D capital. The horizontal lines indicate actual TFP growth over the same period, according to the international agricultural productivity accounts maintained by the USDA Economic Research Service.10 10 The Economic Research Service estimates agricultural TFP indexes for each country and region using standard growth accounting: TFP growth is the difference between the rate of growth in gross agricultural output (from FAOSTAT) and the weighted-average growth rate in inputs (land, labor, farm machinery capital, livestock capital, fertilizer, and feed), where the weights are input-cost shares. See Fuglie (2015) for a complete description of methods and sources. Figure 2 View largeDownload slide Growth in R&D capital and productivity in world agriculture, 1990–2011 Note: The height of the stacked bars show the predicted growth of agricultural TFP between 1990 and 2011, with the bar segments showing the respective contributions of different sources of technology. The horizontal lines show the actual growth in agricultural TFP over the same year period. Sources: Predicted TFP growth are the author’s estimates. Actual TFP growth is from the Economic Research Service. Figure 2 View largeDownload slide Growth in R&D capital and productivity in world agriculture, 1990–2011 Note: The height of the stacked bars show the predicted growth of agricultural TFP between 1990 and 2011, with the bar segments showing the respective contributions of different sources of technology. The horizontal lines show the actual growth in agricultural TFP over the same year period. Sources: Predicted TFP growth are the author’s estimates. Actual TFP growth is from the Economic Research Service. The model predictions align fairly closely to measured TFP growth for all the developed regions, Latin America and the Caribbean (LAC) and South Asia. In other words, over the two decades since 1990, R&D capital growth accounts for nearly all the agricultural TFP growth in these regions. For SE Asia, China, and Africa-West Asia, however, predicted TFP growth accounts for only a portion of measured TFP improvement during these years. The gap between predicted and measured TFP growth suggests that either (a) the impact of R&D capital is not being adequately captured by these models, and/or (b) factors other than R&D are also major drivers of efficiency and productivity improvement in these regions. Recent studies of agricultural growth in these regions do report significant impacts of other drivers, especially institutional and policy reforms, on agricultural TFP. For China, Fan (1991) finds that reforms that strengthened producer incentives and liberalized markets contributed significantly to raising agricultural productivity. Policies that improved agricultural terms of trade or liberalized markets were found to increase agricultural TFP in Sub-Saharan Africa (Fuglie and Rada 2013), Indonesia (Rada and Fuglie 2012), and India (Rada and Schimmelpfennig 2015). Nonetheless, a major difference between developed and developing countries revealed by this review is that the former are more adept at capturing technology spillovers from other countries and the private sector, and this has helped them maintain agricultural productivity growth despite a slowdown in the growth of public agricultural R&D spending. At the same time, an apparent lack of international R&D spillovers emanating from developing countries might represent a lost opportunity to raise agricultural productivity globally. Summary and Conclusions The agricultural growth model proposed in this paper treats knowledge capital like physical capital—that is, as a long-lived productive asset. But unlike physical capital, knowledge capital has the potential to generate spillovers—that is, applications beyond the locality or application for which it was originally intended. But because the agricultural production function is conditioned by environmental factors, spillovers may not come freely everywhere and require adaptation. Moreover, because agriculture is subject to biological evolution and environmental and social change, many agricultural technologies become obsolete over time—the curse of the Red Queen—and need to be refurbished or replaced or productivity will fall. These features of agricultural knowledge capital—its location-specific nature and eventual depreciation, distinguish it from the treatment of knowledge capital in New Growth Theory. The public-goods nature of knowledge capital and the small-holder structure of farming has implied a major role of government in investing in agricultural R&D. As recently as 2011, public institutions accounted for about three-quarters of total global spending on agricultural research. Of the accumulated spending on public agricultural R&D over the 50 years from 1962 to 2011 of $PPP 1,192 billion (constant 2011 PPP$), slightly less than half of this could be considered as operational R&D capital in 2011 (the difference being due to technological obsolescent and technologies still in the development pipeline). This R&D capital stock is roughly one-tenth the estimated $5 trillion worth of physical capital (not including land) held by the world’s farmers (FAO 2012). But because of its increasing returns (due to spillovers), agricultural knowledge capital generates much higher social returns than physical capital and thus accounts for an outsized share of agricultural growth. From 44 empirical studies on determinants of agricultural productivity from around the world, public investment in agricultural R&D was found to be a consistent driver of agricultural TFP growth. From these studies, the elasticity (the percentage increase in TFP or output from a 1% increase in R&D capital) for national public R&D capital averaged 0.18 world-wide; it was higher for developed countries (0.27) and significantly lower for sub-Saharan Africa (0.13). However, these studies also show that despite the sensitivity of agriculture to local environmental conditions, R&D spillovers across national borders are also an important source of agricultural productivity growth. It is likely that national R&D capacity enhances the movement of these technologies through adaptive research, and countries with stronger national R&D systems appear to realize larger technology spill-ins from other sources. Including the effects of public R&D spill-ins as well as spill-ins from private R&D widens the innovation gap between developed and developing countries: The average total R&D elasticity (which includes the effects of spillovers, private R&D, and the CGIAR) was much higher in developed countries (0.67) compared with developing countries (0.38), and especially with sub-Saharan Africa (0.17). The growth model posited in this paper implies that R&D investment will need to increase in order for agricultural productivity to continue to grow. Moreover, total agricultural R&D spending will likely need to grow faster than the desired rate of agricultural output growth. This is partly due to limits on the transferability of agricultural technologies across agro-ecologies, but also because of the need for maintenance research to keep productivity from falling. Another finding from this review is that so far there is little evidence that agricultural R&D investment by developing countries have been a significant source of international technology spillovers. This suggests that most developing country research has been focused on local adaption rather than advancing the productivity frontier. As countries like China supplant the United States and other developed countries as the primary investors in public agricultural R&D, concerns emerge about where the next generation of frontier technologies might come from. It is likely that agricultural R&D systems in leading developing countries will need to grow in sophistication, and that they adopt policies to encourage sharing or trading new agricultural technology in order for them to become capable of generating technologies with large spillover potential. This would raise the total R&D elasticity of agricultural innovation systems and significantly reduce the cost of achieving the productivity growth necessary to feed a richer, more populous world. References Alene A. 2010 . Productivity Growth and the Effects of R&D in African Agriculture . Agricultural Economics 41 ( 3-4 ): 223 – 38 . Google Scholar Crossref Search ADS Alston J. , Andersen M. , James J. , Pardey P. . 2010 . Persistence Pays: US Agricultural Productivity Growth and the Benefits from Public R&D Spending . New York : Springer . Alston J. , Andersen M. , James J. , Pardey P. 2011 . The Economic Returns to U.S. Public Agricultural Research . American Journal of Agricultural Economics 93 5 : 1257 – 77 . Google Scholar Crossref Search ADS Alston J. , Dehmer S. , Pardey P. . 2006 . International Initiatives in Agricultural R&D: The Changing Fortunes of the CGIAR. In Agricultural R&D in the Developing World: Too Little, Too Late ?, ed. Pardey P. , Alston J. , Piggott R. , 313 – 60 . Washington DC : International Food Policy Research Institute (IFPRI ). Alexandratos N. , Bruinsma J. . 2012 . World Agriculture Towards 2030/2050: The 2012 Revision. ESA Working Paper No. 12-03 . Rome, Italy : Food and Agriculture Organization . Andersen M. , Song W. . 2013 . The Economic Impact of Public Agricultural Research and Development in the United States . Agricultural Economics 44 3 : 287 – 95 . Google Scholar Crossref Search ADS Agricultural Science and Technology Indicators . Online Database. Washington DC: International Food Policy Research Institute. Accessed October 15, 2016. Baldos U. , Viens F. , Hertel T. , Fuglie K. . 2015 . On the Linkages in U.S. Public R&D Spending, Knowledge Capital and Agricultural Productivity Growth: A Bayesian Approach. Conference Paper, Agricultural and Applied Economics Association, San Francisco, CA, July 26 – 28 . Bervejillo J. , Alston J. , Tumber K. . 2012 . The Benefits from Public Agricultural Research in Uruguay . Australian Journal of Agricultural and Resource Economics 56 4 : 475 – 97 . Google Scholar Crossref Search ADS Block S. 2014 . The Post-independence Decline and Rise of Crop Productivity in Sub-Saharan Africa: Measurement and Explanations . Oxford Economic Papers 66 2 : 373 – 96 . Google Scholar Crossref Search ADS Bouchet F. , Orden D. , Norton G. . 1983 . Sources of Growth in French Agriculture . American Journal of Agricultural Economics 71 1 : 280 – 93 . Butault J.-P. , LeMarie S. , Musolesi A. , Huart F. , Simioni M. , Schmitt B. . 2015 . Impact de la Recherche Agronomique sur la Produtivité Agricole Française: Une Approche par le Taux de Rentabilité Interne (TRI) de la Recherche Publique. Rapport d’étude à destination du Collège de Direction de l’INRA. Working paper, version 22 January 2015. Paris, France: French National Institute for Agricultural Research. Craig B. , Pardey P. , Roseboom J. . 1997 . International Productivity Patterns: Accounting for Input Quality, Infrastructure and Research . American Journal of Agricultural Economics 79 4 : 1064 – 76 . Google Scholar Crossref Search ADS Eberhardt M. , Teal F. . 2013 . No Mangoes in the Tundra: Spatial Heterogeneity in Agricultural Productivity Analysis . Oxford Bulletin of Economics and Statistics 75 6 : 914 – 39 . Google Scholar Crossref Search ADS Evenson R. 2003 . Production Impacts of Crop Genetic Improvement. In Crop Variety Improvement and Its Effect on Productivity: The Impact of International Agricultural Research , ed. Evenson R. , Gollin D. , 447 – 472 . Wallingford, UK : CAB International . Evenson R. , Fuglie K. . 2010 . Technology Capital: The Price of Admission to the Growth Club . Journal of Productivity Analysis 33 3 : 173 – 90 . Google Scholar Crossref Search ADS Evenson R. , Kislev Y. . 1975 . Agricultural Research and Productivity . New Haven, CT : Yale Univ. Press . Evenson R. , Pray C. , Rosegrant M. . 1999 . Agricultural Research and Productivity Growth in India . Research Report 109. Washington, DC : IFPRI . Evenson R. , Quizon J. . 1991 . Technology, Infrastructure, Output Supply, and Factor Demand in Philippine Agriculture. In Research and Productivity in Asian Agriculture , ed. Evenson R. , Pray C. , 195 – 205 . Ithaca, NY : Cornell University Press . Fan S. 1991 . Effects of Technological Change and Institutional Reform on Production Growth in Chinese Agriculture . American Journal of Agricultural Economics 73 2 : 266 – 75 . Google Scholar Crossref Search ADS Fan S. 2000 . Research Investment and the Economic Returns to Chinese Agricultural Research . Journal of Productivity Analysis 14 2 : 163 – 82 . Google Scholar Crossref Search ADS Fan S. , Al-Riffai P. , El-Said M. , Yun B. , Kamaly A. . 2006 . A Multi-level Analysis of Public Spending, Growth and Poverty Reduction in Egypt. DSGD Discussion Paper No. 41. Washington DC: International Food Policy Research Institute (IFPRI). Fan S. , Hazell P. , Thorat S. . 2000 . Government Spending, Growth and Poverty in Rural India . American Journal of Agricultural Economics 82 4 : 1038 – 51 . Google Scholar Crossref Search ADS Fan S. , Pardey P. . 1998 . Government Spending on Asian Agriculture: Trends and Production Consequence. In Agricultural Public Finance Policy in Asia . Tokyo : Asian Productivity Organization (APO ). Fan S. , Zhang L. , Zhang X. . 2002 . Growth, Inequality and Poverty in Rural China: The Role of Public investments . Research Report 125. Washington DC : International Food Policy Research Institute (IFPRI ). FAOSTAT. Online Database. Rome: Food and Agriculture Organization. Accessed September 15, 2016 . Food and Agriculture Organization . 2012 . State of Food and Agriculture 2012: Investing in Agriculture for a Better Future . Rome : Food and Agriculture Organization (FAO ). Feder G. , Just R. , Zilberman D. . 1985 . Adoption of Agricultural Innovations in Developing Countries: A Survey . Economic Development and Cultural Change 33 2 : 255 – 98 . Google Scholar Crossref Search ADS Federico G. 2005 . Feeding the World: An Economic History of Agriculture, 1800-2000 . Princeton, NJ : Princeton University Press . Fernandez-Cornejo J. , Shumway C. . 1997 . Research and Productivity in Mexican Agriculture . American Journal of Agricultural Economics 79 3 : 738 – 53 . Google Scholar Crossref Search ADS Foster A. , Rosenzweig M. . 1996 . Technical Change and Human-Capital Returns and Investments: Evidence from the Green Revolution . American Economic Review 86 4 : 931 – 53 . Frisvold G. , Ingram K. . 1995 . Sources of Agricultural Growth and Stagnation in Sub-Saharan Africa . Agricultural Economics 13 1 : 51 – 61 . Google Scholar Crossref Search ADS Fuglie K. 2015 . Accounting for Growth in Global Agriculture . Bio-based and Applied Economics 4 3 : 221 – 54 . Fuglie K. 2016 . The Growing Role of the Private Sector in Agricultural Research and Development World-wide . Global Food Security 10 : 29 – 38 . Google Scholar Crossref Search ADS Fuglie K. , Heisey P. , King J. , Pray C. , Day-Rubenstein K. , Schimmelpfennig D. , Wang S. , Karmarkar-Deshmukh R. . 2011 . Research Investments and Market Structure in the Food Processing, Agriculture Input and Biofuel Industries Worldwide . Washington DC : U.S. Department of Agriculture , Economic Research Report No. 130. Fuglie K. , Rada N. . 2013 . Resources, Policy and Agricultural Productivity in Sub-Saharan Africa . Washington DC : U.S. Dept. of Agriculture, Economic Research Service , Economic Research Report No. 145. Fuglie K. , Toole A. . 2014 . The Evolving Institutional Structure of Public and Private Agricultural Research . American Journal of Agricultural Economics 96 3 : 862 – 83 . Google Scholar Crossref Search ADS Fulginiti L. , Perrin R. . 1993 . Prices and Productivity in Agriculture . Review of Economics and Statistics 75 3 : 471 – 82 . Google Scholar Crossref Search ADS Gutierrez L. , Gutierrez M. . 2003 . International R&D Spill-overs and Productivity Growth in the Agricultural Sector: A Panel Cointegration Approach . European Review of Agricultural Economics 30 3 : 281 – 303 . Google Scholar Crossref Search ADS Hall J. , Scobie G. . 2006 . The Role of R&D in Productivity Growth: The Case of Agriculture in New Zealand, 1927 to 2001. Working Paper 06/01, New Zealand Treasury, Wellington. Hayami Y. , Ruttan V. . 1985 . Agricultural Development: An International Perspective . Baltimore, MD : The Johns Hopkins University Press . Heisey P. , Fuglie K. . Forthcoming. Agricultural Research Policy in High-Income Countries. Washington DC: U.S. Dept. of Agriculture, Economic Research Service. Economic Information Bulletin. Huffman W. , Evenson R. . 2006 . Science for Agriculture: A Long-Term Perspective . Ames, IA : Blackwell . Hurley T. , Rao X. , Pardey P. . 2014 . Re-examining the Reported Rates of Return to Food and Agricultural Research and Development . American Journal of Agricultural Economics 96 5 : 1492 – 504 . Google Scholar Crossref Search ADS Jin S. , Huang J. , Hu R. , Rozelle S. . 2002 . The Creation and Spread of Technology and Total Factor Productivity in China’s Agriculture . American Journal of Agricultural Economics 84 4 : 916 – 30 . Google Scholar Crossref Search ADS Jin Y. , Huffman W. . 2016 . Measuring Public Agricultural Research and Extension and Estimating their Impacts on Agricultural Productivity: New Insights from U.S. Evidence . Agricultural Economics 47 1 : 15 – 31 . Google Scholar Crossref Search ADS Johnson D. , Evenson R. . 1999 . R&D Spillovers to Agriculture: Measurement and Application . Contemporary Economic Policy 17 4 : 432 – 56 . Google Scholar Crossref Search ADS Johnson D. , Evenson R. 2000 . How Far away Is Africa? Technological Spill-overs to Agriculture and Productivity . American Journal of Agricultural Economics 82 5 : 743 – 49 . Google Scholar Crossref Search ADS Jorgenson D. , Fukao K. , Timmer M. , eds. 2016 . The World Economy: Growth or Stagnation ? Cambridge, UK : Cambridge University Press . Judd M. , Boyce J. , Evenson R. . 1991 . Investment in Agricultural Research and Extension Programs: A Quantitative Assessment. In Research and Productivity in Asian Agriculture , ed. Evenson R. , Pray C. 7 – 46 . Ithaca : Cornell University Press . Lains P. , Pinilla V. , eds. 2009 . Agriculture and Economic Development in Europe Since 1870 . London : Routledge . Lusigi A. , Thirtle C. . 1997 . Total Factor Productivity and the Effects of R&D in African Agriculture . Journal of International Development 9 4 : 529 – 38 . Google Scholar Crossref Search ADS Mullen J. , Cox T. . 1995 . The Returns from Research in Australian Broadacre Agriculture . Australian Journal of Agricultural Economics 39 2 : 105 – 28 . Google Scholar Crossref Search ADS Organisation for Economic Cooperation and Development. Online Database. Research and Development Statistics. Paris, France. Accessed October 12, 2016. Olmstead A. , Rhode P. . 2002 . The Red Queen and the Hard Reds: Productivity Growth in American Wheat, 1800-1940 . Journal of Economic History 62 4 : 929 – 66 . Google Scholar Crossref Search ADS Otsuka K. 2013 . Food Insecurity, Income Inequality, and the Changing Comparative Advantage in World Agriculture . Agricultural Economics 44 ( s1 ): 7 – 18 . Google Scholar Crossref Search ADS Pardey P. , Chan-Kang C. , Dehmer S. , Beddow J. . 2016 . Agricultural R&D is on the Move . Nature 537 (15 September): 301 – 3 . Google Scholar Crossref Search ADS PubMed Pardey P. , Roseboom J. . 1989 . ISNAR Agricultural Research Indicator Series: A Global Database on National Agricultural Research Systems . Cambridge : Cambridge University Press . Pardey P. , Roseboom J. , Anderson J. , eds. 1991 . Agricultural Research Policy: International Quantitative Perspectives . Cambridge, UK : Cambridge University Press . Pray C. , Ahmed Z. . 1991 . Research and Agricultural Productivity in Bangladesh. In Research and Productivity in Asian Agriculture , ed. Evenson R. , Pray C. , 114 – 32 . Ithaca : Cornell University Press . Pray C. , Fuglie K. . 2015 . Agricultural Research by the Private Sector . Annual Review of Resource Economics 7 : 399 – 424 . Google Scholar Crossref Search ADS Rada N. , Fuglie K. . 2012 . Shifting Sources of Agricultural Growth in Indonesia. In Productivity Growth in Agriculture: An International Perspective , ed. Fuglie K. , Wang S. , Ball E. , 199 – 214 . Wallingford, UK : CABI . Rada N. , Schimmelpfennig D. . 2015 . Propellers of Agricultural Productivity in India . Washington DC : U.S. Dept. of Agriculture, Economic Research Service , Economic Research Report No. 203. Rada N. , Buccola S. . 2012 . Agricultural Policy and Productivity: Evidence from Brazilian Censuses . Agricultural Economics 43 4 : 353 – 65 . Google Scholar Crossref Search ADS Rahman S. , Salim R. . 2013 . Six Decades of Total Factor Productivity Change and Sources of Growth in Bangladesh Agriculture (1948–2008) . Journal of Agricultural Economics 64 2 : 275 – 94 . Google Scholar Crossref Search ADS Romer P. 1990 . Endogenous Technological Change . Journal of Political Economy 98 : S71 – S102 . Google Scholar Crossref Search ADS Ruttan V. 1982 . Agricultural Research Policy . Minneapolis : University of Minnesota Press . Ruttan V. 2001 . Productivity Growth in World Agriculture: Sources and Constraints . Journal of Economic Perspectives 16 4 : 161 – 84 . Google Scholar Crossref Search ADS Schimmelpfennig D. , Thirtle C. . 1999 . The Internationalization of Agricultural Technology: Patents, R&D Spill-overs, and Their Effects on Productivity in the European Union and the United States . Contemporary Economic Policy 17 4 : 457 – 68 . Google Scholar Crossref Search ADS Schultz T. 1975 . The Value of the Ability to Deal with Disequilibria . Journal of Economic Literature 13 (3): 827 – 46 . Sheng Y. , Gray E. , Mullen J. , Davidson A. . 2011 . R&D and Extension: An Analysis of the Static and Dynamic Effects on Australian Broadacre Productivity. ABARES Research Report 11.7. Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. Suphannachart W. , Warr P. . 2012 . Total Factor Productivity in Thai Agriculture: Measurement and Determinants. In Productivity Growth in Agriculture: An International Perspective , ed. Fuglie K. , Wang S. , Ball E. , 215 – 36 . Wallingford, UK : CABI . Thirtle C. , Hadley D. , Townsend R. . 1995 . Policy-induced Innovation in Sub-Saharan African Agriculture: A Multilateral Malmquist Productivity Index Approach . Development Policy Review 12 : 323 – 48 . Google Scholar Crossref Search ADS Thirtle C. , Piesse J. , Schimmelpfennig D. . 2008 . Modeling the Length and Shape of the R&D Lag: An Application to UK Agricultural Productivity . Agricultural Economics 39 : 73 – 85 . Google Scholar Crossref Search ADS Thirtle C. , Lin L. , Piesse J. . 2003 . The Impact of Research-led Agricultural Productivity Growth on Poverty Reduction in Africa, Asia and Latin America . World Development 31 1 : 1959 – 75 . Google Scholar Crossref Search ADS U.S. Department of Agriculture, Economic Research Service. Online Database: International Agricultural Productivity. Washington DC. Accessed November 15, 2016. Wang S. , Heisey P. , Schimmelpfennig D. , Ball E. . 2015 . Agricultural Productivity Growth in the United States: Measurement, Trends, and Drivers . Washington DC : U.S. Department of Agriculture , Economic Research Report No. 189. Wang S. , Ball E. , Fulginiti L. , Plastina A. . 2012 . Accounting for the Impacts of Public Research, R&D Spillins, Extension, and Roads in U.S Regional Agricultural Productivity Growth, 1980–2004. In Productivity Growth in Agriculture: An International Perspective , ed. Fuglie K. , Wang S. , Ball E. , 13 – 32 . Wallingford, UK : CABI . Wang S. , Heisey P. , Huffman W. , Fuglie K. . 2013 . Public R&D, Private R&D, and U.S. Agricultural Productivity Growth: Dynamic and Long-run Relationships . American Journal of Agricultural Economics 95 5 : 1287 – 93 . Google Scholar Crossref Search ADS Wiebe K. , Soule M. , Narrod C. , Breneman V. . 2000 . Resource Quality and Agricultural Productivity: A Multi-country Comparison. Conference Paper, American Agricultural Economics Association, Tampa, FL. Wong Lung-Fai. 1986 . Agricultural Productivity in the Socialist Countries . Boulder, CO : Westview . World Development Indicators. Online Database. Washington DC: World Bank. Accessed October 15, 2016. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association 2017. This work is written by a US Government employee and is in the public domain in the US. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Economic Perspectives and Policy Oxford University Press

R&D Capital, R&D Spillovers, and Productivity Growth in World Agriculture

Loading next page...
 
/lp/ou_press/r-d-capital-r-d-spillovers-and-productivity-growth-in-world-3dVAmQiDSu
Publisher
Oxford University Press
Copyright
Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association 2017. This work is written by a US Government employee and is in the public domain in the US.
ISSN
2040-5790
eISSN
2040-5804
D.O.I.
10.1093/aepp/ppx045
Publisher site
See Article on Publisher Site

Abstract

Abstract Increasing the world’s food supply has depended heavily on increasing agricultural productivity, which in turn depends on investments in research and development (R&D). This article synthesizes findings from more than 40 studies on how R&D investments affect agricultural total factor productivity (TFP) in various parts of the world. The article breaks out the relative contributions to TFP growth of R&D by public institutions, private companies, and the CGIAR (a consortium of international agricultural research centers), including international technology spillovers. Major differences emerge between global regions in sources and efficiency of R&D capital. Developed countries appear to have benefitted more from private and international R&D spillovers than developing countries. Agricultural total factor productivity, R&D elasticities, R&D lags, technological obsolescence Agriculture may be unique in its reliance on productivity for growth. Whereas about one-third of total economic growth comes from increases in the total productivity of factor inputs (Jorgenson, Fukao, and Timmer 2016), in agriculture, total factor productivity (TFP) accounts for about three-fourths of growth at the global level and virtually all growth in industrialized countries (Fuglie 2015). This reliance on productivity reflects agriculture’s dependence on inherently limited resources like land and water, and it is these resource constraints that have given rise to concerns, since at least Malthus, that world population may soon overreach the capacity of what the world can sustainably afford to feed. The fact that agricultural productivity has been able to grow sufficiently to meet rising demand is no accident. Rather, it reflects to a large degree a deliberate choice to commit resources to agricultural research and development (R&D). In today’s advanced industrialized nations, the establishment of public agricultural research institutions in the late nineteenth century helped set in motion a process of technological and structural transformation of their agricultural systems (Ruttan 1982). That process is still going on today, and in fact, has been extended to most of the world. Nearly all countries now have national agricultural research institutions of one form or another. In addition, international agricultural research centers have been established (Alston, Dehmer, and Pardey 2006) and the role of the private sector in generating new agricultural technology has grown (Fuglie et al. 2011). Because positive externalities (spillovers) from R&D lead to undervaluation of innovation in the marketplace, governments have a central role in creating the knowledge capital required for economic growth.1 1 The government creates knowledge capital through direct investment in R&D, and by establishing intellectual property rights, creating excludability conditions for private inventors. Positive externalities from knowledge capital is central, for example, in New Growth Theory (Romer 1990). As articulated by Romer (1990), once new knowledge is created it is available everywhere to all, forever, except as constrained by insufficient human capital to make use of it, and by legal or other measures to protect the intellectual property of inventors. But because agriculture depends on environmental forces, and thus technology is sensitive to location, agricultural knowledge capital is likely to be much more constrained than the knowledge capital envisioned by Romer. Further, because environmental forces change over time (due to the co-evolution of pests and diseases, water and land resource degradation, and climate change), technological obsolescence in agriculture will eventually set in (Ruttan 2001). Olmstead and Rhode (2002) dubbed the need for continued research just to maintain agricultural productivity as the curse of the Red Queen.2 2 “Now, here, you see, it takes all the running you can do, to keep in the same place,” The Red Queen to Alice in Lewis Carroll’s Through the Looking Glass. Another factor, though not unique to but probably accentuated in agriculture, is the relatively slow uptake of technologies due to the highly dispersed, heterogeneous, and small-holder structure of producers. These characteristics of agriculture suggest the general shape of a time path for how investments in agricultural R&D are likely to affect farm productivity: a relatively long lag between R&D spending and when that spending results in significant improvements to aggregate farm productivity, and eventual depreciation of the productivity gains without renewal of R&D capital (Huffman and Evenson 2006; Alston et al. 2010). One implication of this view is that continuously raising agricultural productivity requires continuous growth in R&D spending. The objective of this paper is to provide a synthesis and assessment of how investments in agricultural R&D have affected productivity growth in world agriculture. First, I assemble historical data on public R&D spending from 150 countries, the private sector, and the CGIAR consortium of international agricultural research centers. Most of the estimates date from 1960, though for some formerly centrally-planned (transition) countries they start from the 1990s. I then construct estimates of accumulated R&D capital stock from each of these sources using a model conforming to the concepts outlined above (and formalized below). Through a review of 44 econometric studies on how R&D influences productivity growth in agriculture, I derive average R&D elasticities for the different sources of R&D and for different global regions. The R&D elasticity measures the percentage change in the TFP (the ratio between the gross output of crops and animal commodities and the combined input of labor, land, capital and materials employed in their production) given a 1% change in R&D capital. I use these elasticities to predict growth in TFP for each global region and compare these predictions against measured TFP growth from 1990 to 2011. The exercise sheds light on the role of public, private, CGIAR R&D investments, and, importantly, the role of R&D spillovers, on raising agricultural TFP around the world. Model of Agricultural R&D Capital and Productivity The Agricultural Production Function For assessing long-term growth in agriculture, it is useful to start with an aggregate production function: Qt=AtRtXtLabort,Landt,Capitalt,Materialst (1) where Q is the agricultural output of a country or region, A is technology or TFP (a function of variables R representing the creation and diffusion of knowledge and ideas), X consists of factor inputs, and t is time. From equation (1), growth in output over time can be decomposed into parts due to technological change and input accumulation: ∂Q∂t=∂A∂t+∂X∂t. (2) Empirical estimates of equations (1) and (2) applied to world agriculture find that during the twentieth century, the main source of growth shifted from input accumulation to productivity in most regions of the world (Hayami and Ruttan 1985; Federico 2005). For the United States and most other developed countries, this transition began in the mid-twentieth century. Since the 1950s, for example, growth rates for U.S. agricultural output and agricultural TFP have been almost synonymous, with aggregate input hardly changing, except in its composition (Wang et al. 2015). For developing countries, this transition began later. For the world as a whole, Fuglie (2015) estimated that since 1990, about three-fourths of the growth in world agricultural output was due to improvements in TFP, although for some low-income countries, particularly in Africa, factor accumulation continued to be the main source of growth. Understanding the future pace of agricultural growth will mostly involve forecasts of agricultural productivity. For example, the long-term world agricultural supply and demand projections of the United Nations Food and Agriculture Organization (FAO) suggest that between 2006 and 2050, global food demand will rise by around 60%, and at least 90% of this increase will come from raising agricultural yield and cropping intensity on existing farm land, rather than the expansion of farm land (Alexandratos and Bruinsma 2012). Given the overriding importance of productivity to agricultural growth, it is useful to have an explicit model of equations (1) and (2) that can give policy makers some leverage with which to influence future food supply. For this, I specify a Cobb-Douglas function where technology and productivity is driven by the stock of ideas, or knowledge capital, which arises from formal R&D investment from public and private sources: At=A∏i=1ISitδi (3) where Sit is the stock of R&D capital from one of i=1,2,…I sources of new technology for agriculture (e.g., from public research institutes, universities, private agribusiness, and international centers). The elasticities δ1,δ2,…,δI translate how a change in R&D capital from source i affects growth in TFP (i.e., a 1% increase in R&D capital Si increases TFP by δi percent). Empirically, the formulation of agricultural knowledge capital Si in equation (3) has been treated similarly to physical capital except for some special features. Like physical capital, knowledge capital (or R&D capital) is the accumulation of past annual investments in R&D and eventually depreciates. But R&D capital is likely to have a longer gestation period (time for research to lead to useable technologies and spread to farmers). It may also be longer-lived, given that “ideas” might not wear out as fast as machines or structures. Most importantly, knowledge capital is non-rival (its use in one place does not limit its use elsewhere). The non-rival nature of knowledge capital is what gives rise to potential spillovers. Numerically, knowledge capital with the above features can be measured using some version of the following: St=∑i=0TwiRt-i (4) where Rt-i is annual R&D spending i years ago, and wo,w1,…wT is a set of weights that are initially at or close to zero, rise to a peak, and then eventually fall back to zero after T years. The zero or low initial values of wi reflect the gestation period for research to result in new technology that can be used by farmers (e.g., the time to breed a new crop variety). The rising values of wi indicate the diffusion of new technology to farmers. The fact that the wis are assumed to eventually diminish reflects knowledge capital depreciation. Examples of R&D capital depreciation in agriculture abound; they are perhaps most in evidence by the emergence of new pests and diseases that threaten existing crop and animal yield. Knowledge deprecation also occurs when completely new forms of production technologies emerge (e.g., the development of tractors made obsolete many innovations in animal drafting). Changes to natural resources (soil degradation, groundwater withdrawals, and rising greenhouse gas concentrations) may also render ineffective many past innovations. Note several features of the model in equations (3) and (4). First, agricultural productivity depends on past investments in national R&D and on the inventive activities of others that are relevant to the conditions of a country. Second, R&D spending is not immediately translated into useable R&D capital. It takes time for R&D to produce technologies that are both adoptable and adopted by producers. Third, R&D capital depreciates over time. Fourth, the R&D elasticities and depreciation rate may vary by country, being conditioned by institutional and environmental factors. The relevance of inventive activity done elsewhere will be determined by the similarity of farming systems (in terms of ecologies, commodities produced, and scale of production).3 3 Note that New Growth Theory considers TFP growth to be proportional to some measure of constant scientific effort, or ∂lnAt/∂t=δSt, where St is research input and δ is a proportionality parameter (Romer 1990). In this paper, because of depreciation, TFP growth is proportional to the growth rate of research input, ∂lnAt/∂t=δ∂lnSt/∂t. Besides national public R&D, the model allows for external sources of technological change to affect a country’s agricultural productivity. Although agricultural technology is sensitive to local environmental conditions, direct spillovers of technology may be possible from other countries or regions that have similar environments or that create general-purpose technologies that can be put to use locally, though perhaps with some adaptation. Some technologies, like a new agricultural pesticide, may be toxic against a number of agricultural pests that inhabit different ecologies and infest different crops, but its usage may differ across them. Technological spillovers may also arise between sectors, such as from industry to agriculture. New human pharmaceutical discoveries may act against farm animal diseases, but may also require adaptive research for specific animal applications. Admittedly, the model in equations (3) and (4) takes a somewhat narrow view of what causes changes in TFP. By focusing exclusively on knowledge capital generated through formal investments in R&D, the model ignores productivity gains from specialization (made possible through greater openness to trade), economies of scale, and informal innovation (such as by farmers themselves). The model also does not explicitly account for factors that affect the “enabling environment” for technology diffusion (e.g., farmer’s education and health, agricultural extension and credit services, secure land tenure, and the rule of law), other than by acknowledging that these influence the R&D lag structure. Historically, trade policy has had an important influence on agricultural productivity. In late-nineteenth century Europe, Denmark and the Netherlands achieved more rapid agricultural growth than France and Germany because of their greater willingness to import cereals and specialize in animal production (Hayami and Ruttan 1985; Lains and Pinilla 2009). In late-twentieth century East Asia, partly for national food security objectives and partly to reduce rural-urban income disparity, Japan, South Korea, and to some extent China sought to protect local grain producers from international competition but at the expense of overall productivity and efficiency (Otsuka 2013). Policies and institutions have also been an important determinant of the rate of technology diffusion in agriculture. The dispersed structure of agriculture requires that before new technologies can actually affect productivity, they have to be adopted on thousands if not millions of small, family-run farms. How smoothly this occurs is likely to depend on several factors: the complexity, scale and cost of new technologies; the education and skill of farm managers and workers; their means of acquiring information; raising capital and insuring against unexpected losses; the ease of marketing farm surpluses; and how well farmers are remunerated for their efforts (see Feder, Just, and Zilberman 1985). Nonetheless, non-R&D factors are likely to more limited in their capacity to sustain agricultural TFP growth over the long run, due to diminishing returns. For example, productivity gains from trade liberalization or more rapid technology diffusion will be exhausted once a new equilibrium is reached. Informal farmer innovations and economies of scale may be primarily adaptations of knowledge capital arising from formal R&D. Evidence shows that the social returns to policies that strengthen the enabling environment, such as investing in farmer education and extension, are likely to be higher in an agricultural system undergoing rapid technological and structural change than in a technologically stagnant one (Schultz 1975; Foster and Rosenzweig 1996). While policy makers may view improvements in non-R&D factors as substitutes for investments in R&D, it may be considerably less costly to view these as complements. We may expect that formal R&D will be more strongly connected to TFP growth in countries where the enabling institutions are more developed. The Agricultural R&D Lag Structure To translate R&D investment into R&D capital and productivity growth I adopt a framework developed by Alston et al. (2010). These authors proposed an R&D capital lifespan of up to 50 years, and used a gamma distribution to capture the technology maturation and diffusion processes that occur in the early part of this period, and the obsolescence and dis-adoption that sets in toward the end. Huffman and Evenson (2006) proposed a similar concept using a trapezoid curve with a 35-year life span for R&D capital (which can be closely matched, with the right choice of parameters, by a gamma distribution) to represent the development, adoption, and obsolescence phases.4 4 Alston et al. (2010) and Huffman and Evenson (2006) specify the R&D lag structures as a set of weights that sum to one. In this way they apportion $1 of R&D spending to the time periods in which the impact of that R&D is expected to be felt. Here, the same relative weights are used but their maximum value is set to one, so that $1 of investment adds $1 to capital when the investment is fully operational (and less than $1 when it is partially operational), as is standard for measuring the accumulated stock of capital. Some gamma and trapezoid R&D lag distributions are shown in figure 1. In each distribution, R&D spending at time 0 slowly accrues to R&D capital, then peaks and depreciates until the end of its useful life. With the 35-year R&D lag structure, the full impact of R&D spending in year 0 is realized about a decade later, and with the 50-year lag structure, after about two decades. Other lag structures are certainly possible, and the gamma distribution is flexible enough to represent a range of possibilities, given the appropriate choice of parameters.5 5 Alston et al. (2010) tested 60 gamma distributions in modeling the impact of R&D on productivity growth in U.S. agriculture. The 50-year lag distribution shown in figure 2 reflects their preferred choice among these distributions. However, that goodness-of-fit tests often failed to distinguish between models and their preferred choice partly rests on historical information on the development and diffusion of major agricultural innovations. Huffman and Evenson (2006) rely on similar reasoning to choose a 35-year R&D lag structure, as did Fan (2000) in selecting a 27-year R&D lag structure for China. Figure 1 View largeDownload slide Alternative lag structures for R&D capital formation Note: Slightly departing from Alston et al. (2010), I measure the height of the gamma distribution in year k as bk=k-g-1ϕ1-ϕθk-gmaxk-g-1ϕ1-ϕθk-g where g is the R&D gestation lag , ϕ,θ are shape and scale parameters of the gamma distribution, and k = 1…L, where L is the maximum length of the R&D lag structure. Since the values of bkare very close to zero in early years, a value of g=0is used. The ϕ,θparameters for the 35-year and 50-year gamma distributions shown in the figure are 0.90,0.70 and 0.80,0.75, respectively. Alston et al. (2010) divide the numerator in the above equation by ∑k=1Lk-g-1ϕ1-ϕθk-g so that ∑bk = 1. I make a proportional shift to the weights so that the maximum value of bk = 1. This weighting scheme implies that $1 in R&D expenditure will add $1 to R&D stock once the technology from that R&D is fully utilized. Figure 1 View largeDownload slide Alternative lag structures for R&D capital formation Note: Slightly departing from Alston et al. (2010), I measure the height of the gamma distribution in year k as bk=k-g-1ϕ1-ϕθk-gmaxk-g-1ϕ1-ϕθk-g where g is the R&D gestation lag , ϕ,θ are shape and scale parameters of the gamma distribution, and k = 1…L, where L is the maximum length of the R&D lag structure. Since the values of bkare very close to zero in early years, a value of g=0is used. The ϕ,θparameters for the 35-year and 50-year gamma distributions shown in the figure are 0.90,0.70 and 0.80,0.75, respectively. Alston et al. (2010) divide the numerator in the above equation by ∑k=1Lk-g-1ϕ1-ϕθk-g so that ∑bk = 1. I make a proportional shift to the weights so that the maximum value of bk = 1. This weighting scheme implies that $1 in R&D expenditure will add $1 to R&D stock once the technology from that R&D is fully utilized. Different R&D lag structures might be appropriate for different kinds of research. More fundamental advances in agricultural technology are likely to take longer to mature and have a longer and wider impact on productivity compared with research that adapts existing technologies to new localities and uses. The initial effort to develop biotechnology traits for crops in the United States took about two decades, but it took considerably less time to transfer these traits to crop cultivars suitable for South America and elsewhere. The Green Revolution was typified by the global diffusion of semi-dwarf varieties of rice and wheat that had greater yield response to fertilizer. But these varieties often required hybridization with local cultivars that were adapted to pests, disease, day length, and soil types found in specific regions. The research to adapt or refine general-purpose technologies to local environments may involve shorter lags than research that expands the scientific frontier. Whether there are systematic differences in R&D lag structure for countries at different stages of development is an open question. On the one hand, we might expect a longer research gestation period in countries whose farmers already produce at a world technology frontier compared with countries where farmers are producing substantially below the frontier. Thus, developed countries may face longer R&D lag structures than many developing countries. On the other hand, R&D lag structures also incorporate the speed at which new technologies are taken up by a population of farmers. Technology diffusion may be relative slow, even for profitable technologies, in countries with poor enabling environments. This may argue in favor of a longer R&D lag structure in developing countries. Fan (2000), who examined the actual development and adoption time for a number of agricultural technologies in China, and Alene (2010), who estimated the parameters of an agricultural R&D lag structure that seemed to be the best fit for productivity patterns in Sub-Saharan Africa, seem to find evidence of shorter R&D lags compared with what has been estimated for the United States (Alston et al. 2010; Baldos et al. 2015) and the United Kingdom (Thirtle, Piesse, and Schimmelfpennig, 2008). However, none of this evidence is very robust, and R&D lag length is an issue deserving more attention. Agricultural R&D Spending and Capital Accumulation Agricultural R&D Spending This study uses data compiled from multiple sources in 150 countries to construct a continuous time series for public agricultural R&D spending since 1960. Public R&D includes research by government agencies and universities. For high-income countries, the data are from Heisey and Fuglie (forthcoming). For developing countries, fairly comprehensive annual series on agricultural R&D spending since 1981 are available from Agricultural Science and Technology Indicators. For years prior to 1981, R&D spending data are from Pardey and Roseboom (1989) and Pardey, Roseboom, and Anderson (1991). Further, R&D spending by (formerly) centrally-planned countries in Eastern Europe and the Soviet Union are drawn from Judd, Boyce, and Evenson (1991). In addition to these sources, a number of other sources were consulted to fill gaps.6 6 The R&D data series and a full list of sources are available from the author upon request. To convert expenditures in national currencies to constant 2011 purchasing-power-parity dollars (PPP$), estimates were first adjusted to constant 2011 local currency units by the national implicit GDP price index, and then converted to PPP$ using the 2011 PPP$ exchange rate (both series are from World Development Indicators). For a complete picture of global agricultural R&D investments, one also needs the private sector, but here the data are less available. For the United States, Fuglie et al. (2011) provide estimates of farm-oriented R&D by U.S. companies from 1960 to 2007. At the global level, Pardey et al. (2016) report estimates of business-sector spending on food and agricultural research by decade from 1980 to 2010 for major global regions, but do not break out food-sector R&D from farm-oriented R&D. Fuglie (2016) gives annual estimates of global private-sector R&D on crops, livestock, and farm machinery world-wide from 1990 to 2010, but notes that it is difficult to assign private R&D to individual countries with much precision because of the multinational character of the largest firms conducing this R&D. For the purpose of this paper, I treat private agricultural R&D as contributing to a global agricultural knowledge stock. This R&D capital stock affects global regions differently through region-specific R&D elasticities. A higher elasticity value for developed countries, for example, implies that private R&D has a larger impact on productivity in these countries than elsewhere. Another important source of technology for agriculture are non-government, non-profit agricultural research centers. The most prominent of these is the CGIAR Consortium, a system of 15 centers focused primarily on developing-country agriculture that is funded by governments, private foundations, and intergovernmental organizations. The CGIAR was formed in 1971, but many of the centers that make up the system were established in the 1960s. Annual CGIAR R&D spending from 1971 is from Agricultural Science and Technology Indicators with pre-1971 figures from Alston, Dehmer, and Pardey (2006). Agricultural R&D Capital Stocks With long-term R&D expenditure series in constant PPP$, we can estimate and compare public agricultural R&D capital stocks assuming alternative R&D lag structures. Table 1 shows the evolution of public agricultural R&D spending between 1961 and 2011, and accumulated 2011 R&D capital for different regions of the world using the 25-year, 35-year and 50-year gamma lag structures from figure 1. In constant 2011 PPP$, global public agricultural R&D grew from about $7.5 billion in 1961 to $42.3 billion in 2011. Spending rose faster in developing countries than developed countries, with the developed-country share of the total falling from about 65% in 1961 to 47% in 2011. These estimates are broadly consistent with Pardey et al. (2016), who reported public agricultural R&D spending worldwide (not including transition countries) to be $38.13 billion in 2010 (in 2009 PPP$). Further, Pardey et al. (2016) estimated that global spending was 210% higher in 2010 than in 1980, while my estimates, excluding transition countries, give an almost identical 208% increase in real spending between these years. Table 1 Agricultural R&D Spending and R&D Capital by World Region (Constant 2011 PPP$, million) Region R&D expenditure R&D Stock in 2011 1 1961 1991 2011 Total R&D spending, 1962–2011 50-year gamma 35-year gamma 25-year gamma Public agricultural R&D Developing Regions Central America 123 550 866 24,038 9,879 10,336 7,861 South America 812 2,943 3,824 114,946 48,559 48,143 32,879 China 313 1,599 7,768 99,813 27,663 40,118 42,488 Southeast Asia 235 1,204 2,005 57,138 22,060 25,689 19,759 South Asia 278 1,461 3,798 71,658 23,249 31,913 28,482 West Asia 148 615 1,524 36,461 12,869 15,053 12,891 North Africa 119 395 728 20,775 7,663 9,206 7,304 Sub-Saharan Africa 522 1,439 1,910 59,837 24,394 23,407 17,214 Developed Regions Former Soviet Union 336 455 629 24,940 10,316 7,515 5,569 Eastern Europe 2 238 557 904 31,539 13,442 10,936 7,758 Western Europe 1,274 5,570 7,093 230,639 96,103 97,810 69,046 Canada-USA 1,711 4,980 5,501 217,023 88,761 89,571 64,103 Australia-NZ-S Africa 504 1,146 1,168 57,687 25,350 22,950 14,608 Japan-Korea-Taiwan 841 3,282 4,601 141,539 54,862 63,446 46,844 World public agricultural R&D 7,455 26,197 42,321 1,188,035 465,171 496,094 376,806 Developed country share 0.66 0.61 0.47 0.59 0.62 0.59 0.55 Private agricultural R&D 2,613 7,202 12,939 330,680 132,787 133,616 96,646 CGIAR and precursors 1 305 707 11,831 7,791 7,791 5,514 Total world - all sources 10,069 33,703 55,966 1,530,546 605,749 637,501 478,966 Region R&D expenditure R&D Stock in 2011 1 1961 1991 2011 Total R&D spending, 1962–2011 50-year gamma 35-year gamma 25-year gamma Public agricultural R&D Developing Regions Central America 123 550 866 24,038 9,879 10,336 7,861 South America 812 2,943 3,824 114,946 48,559 48,143 32,879 China 313 1,599 7,768 99,813 27,663 40,118 42,488 Southeast Asia 235 1,204 2,005 57,138 22,060 25,689 19,759 South Asia 278 1,461 3,798 71,658 23,249 31,913 28,482 West Asia 148 615 1,524 36,461 12,869 15,053 12,891 North Africa 119 395 728 20,775 7,663 9,206 7,304 Sub-Saharan Africa 522 1,439 1,910 59,837 24,394 23,407 17,214 Developed Regions Former Soviet Union 336 455 629 24,940 10,316 7,515 5,569 Eastern Europe 2 238 557 904 31,539 13,442 10,936 7,758 Western Europe 1,274 5,570 7,093 230,639 96,103 97,810 69,046 Canada-USA 1,711 4,980 5,501 217,023 88,761 89,571 64,103 Australia-NZ-S Africa 504 1,146 1,168 57,687 25,350 22,950 14,608 Japan-Korea-Taiwan 841 3,282 4,601 141,539 54,862 63,446 46,844 World public agricultural R&D 7,455 26,197 42,321 1,188,035 465,171 496,094 376,806 Developed country share 0.66 0.61 0.47 0.59 0.62 0.59 0.55 Private agricultural R&D 2,613 7,202 12,939 330,680 132,787 133,616 96,646 CGIAR and precursors 1 305 707 11,831 7,791 7,791 5,514 Total world - all sources 10,069 33,703 55,966 1,530,546 605,749 637,501 478,966 Note: Superscript 1 R&D capital stock is the aggregate amount of R&D expenditure contributing to productivity in 2011. Cumulative expenditure assumes no R&D lag or depreciation. The 50-year, 35-year, and 25-year gamma lag distributions assume a gestation period and eventual depreciation of R&D capita, according to the R&D capital-life profiles shown in figure 1. 2 Eastern Europe includes the transition economies of Poland, Hungary, Romania, Bulgaria, former Czechoslovakia, and former Yugoslavia. Sources: Data on R&D spending were compiled from multiple sources. Public R&D in high-income countries is from Heisey and Fuglie (forthcoming), except for the former Soviet Union and Eastern Europe, which are from Judd, Boyce, and Evenson (1991) and OECD; public R&D in developing countries since 1981 is from Agricultural Research and Technology Indicators, and prior to 1981 from Pardey and Roseboom (1989) and Pardey, Roseboom, Anderson, (1991). Private R&D since 1990 is from Fuglie (2016) and pre-1990 from Fuglie et al. (2011). CGIAR R&D since 1971 is from Agricultural Research and Technology Indicators and pre-1971 from Alston, Dehmer, and Pardey (2006). Some additional publications were consulted for specific countries and some extrapolations were made for missing data. Contact the author for a complete description of sources and data. Table 1 Agricultural R&D Spending and R&D Capital by World Region (Constant 2011 PPP$, million) Region R&D expenditure R&D Stock in 2011 1 1961 1991 2011 Total R&D spending, 1962–2011 50-year gamma 35-year gamma 25-year gamma Public agricultural R&D Developing Regions Central America 123 550 866 24,038 9,879 10,336 7,861 South America 812 2,943 3,824 114,946 48,559 48,143 32,879 China 313 1,599 7,768 99,813 27,663 40,118 42,488 Southeast Asia 235 1,204 2,005 57,138 22,060 25,689 19,759 South Asia 278 1,461 3,798 71,658 23,249 31,913 28,482 West Asia 148 615 1,524 36,461 12,869 15,053 12,891 North Africa 119 395 728 20,775 7,663 9,206 7,304 Sub-Saharan Africa 522 1,439 1,910 59,837 24,394 23,407 17,214 Developed Regions Former Soviet Union 336 455 629 24,940 10,316 7,515 5,569 Eastern Europe 2 238 557 904 31,539 13,442 10,936 7,758 Western Europe 1,274 5,570 7,093 230,639 96,103 97,810 69,046 Canada-USA 1,711 4,980 5,501 217,023 88,761 89,571 64,103 Australia-NZ-S Africa 504 1,146 1,168 57,687 25,350 22,950 14,608 Japan-Korea-Taiwan 841 3,282 4,601 141,539 54,862 63,446 46,844 World public agricultural R&D 7,455 26,197 42,321 1,188,035 465,171 496,094 376,806 Developed country share 0.66 0.61 0.47 0.59 0.62 0.59 0.55 Private agricultural R&D 2,613 7,202 12,939 330,680 132,787 133,616 96,646 CGIAR and precursors 1 305 707 11,831 7,791 7,791 5,514 Total world - all sources 10,069 33,703 55,966 1,530,546 605,749 637,501 478,966 Region R&D expenditure R&D Stock in 2011 1 1961 1991 2011 Total R&D spending, 1962–2011 50-year gamma 35-year gamma 25-year gamma Public agricultural R&D Developing Regions Central America 123 550 866 24,038 9,879 10,336 7,861 South America 812 2,943 3,824 114,946 48,559 48,143 32,879 China 313 1,599 7,768 99,813 27,663 40,118 42,488 Southeast Asia 235 1,204 2,005 57,138 22,060 25,689 19,759 South Asia 278 1,461 3,798 71,658 23,249 31,913 28,482 West Asia 148 615 1,524 36,461 12,869 15,053 12,891 North Africa 119 395 728 20,775 7,663 9,206 7,304 Sub-Saharan Africa 522 1,439 1,910 59,837 24,394 23,407 17,214 Developed Regions Former Soviet Union 336 455 629 24,940 10,316 7,515 5,569 Eastern Europe 2 238 557 904 31,539 13,442 10,936 7,758 Western Europe 1,274 5,570 7,093 230,639 96,103 97,810 69,046 Canada-USA 1,711 4,980 5,501 217,023 88,761 89,571 64,103 Australia-NZ-S Africa 504 1,146 1,168 57,687 25,350 22,950 14,608 Japan-Korea-Taiwan 841 3,282 4,601 141,539 54,862 63,446 46,844 World public agricultural R&D 7,455 26,197 42,321 1,188,035 465,171 496,094 376,806 Developed country share 0.66 0.61 0.47 0.59 0.62 0.59 0.55 Private agricultural R&D 2,613 7,202 12,939 330,680 132,787 133,616 96,646 CGIAR and precursors 1 305 707 11,831 7,791 7,791 5,514 Total world - all sources 10,069 33,703 55,966 1,530,546 605,749 637,501 478,966 Note: Superscript 1 R&D capital stock is the aggregate amount of R&D expenditure contributing to productivity in 2011. Cumulative expenditure assumes no R&D lag or depreciation. The 50-year, 35-year, and 25-year gamma lag distributions assume a gestation period and eventual depreciation of R&D capita, according to the R&D capital-life profiles shown in figure 1. 2 Eastern Europe includes the transition economies of Poland, Hungary, Romania, Bulgaria, former Czechoslovakia, and former Yugoslavia. Sources: Data on R&D spending were compiled from multiple sources. Public R&D in high-income countries is from Heisey and Fuglie (forthcoming), except for the former Soviet Union and Eastern Europe, which are from Judd, Boyce, and Evenson (1991) and OECD; public R&D in developing countries since 1981 is from Agricultural Research and Technology Indicators, and prior to 1981 from Pardey and Roseboom (1989) and Pardey, Roseboom, Anderson, (1991). Private R&D since 1990 is from Fuglie (2016) and pre-1990 from Fuglie et al. (2011). CGIAR R&D since 1971 is from Agricultural Research and Technology Indicators and pre-1971 from Alston, Dehmer, and Pardey (2006). Some additional publications were consulted for specific countries and some extrapolations were made for missing data. Contact the author for a complete description of sources and data. Over the 50 years from 1962 to 2011, total global spending for public agricultural R&D was $1,192 billion in constant 2011 PPP$. Applying the 50-year lag structure from figure 1 yields a 2011 R&D capital stock of $PPP 467 billion. In other words, about half of the R&D since 1962 was either obsolete or still in the development and diffusion pipeline. The 35-year lag structure for R&D capital yields a larger global R&D capital stock of $PPP 496 billion. Even though this R&D capital is shorter-lived than the estimate using a 50-year lag structure, it includes more of the spending that occurred since 1990 due to its shorter gestation period. Because much of the current R&D capital is based on R&D investment made two to four decades ago, the R&D capital shares for developed countries are larger than their R&D spending shares. In 2011, developed countries accounted for 55%, 59% or 62% of global public agricultural R&D capital (using the 25-,35-, and 50-year lag distributions), but only 47% of global R&D spending. Even if current expenditure shares were to remain constant over the coming decades, the R&D capital share of developing countries would rise (eventually matching their expenditure share) as their recent R&D spending matures and the technologies arising from it are adopted by farmers. The divergence between public R&D expenditure and capital shares is especially prominent for China. Chinese government investment in agricultural R&D grew by more than 10% per year between 2000 and 2011 in real terms. By 2008, China had overtaken the United States as the largest national investor in agricultural R&D, and by 2011 it accounted for about 18% of world (public) agricultural R&D spending. However, China’s global R&D capital share has yet to reflect this growth because much of this R&D spending has yet to translate into farm productivity. North American R&D capital share, on the other hand, is significantly larger than its expenditure share. This reflects the historical role of the United States as a world leader in agricultural science and technology. But as its public R&D capital ages and R&D spending stagnates, the United States is being overtaken by China and others. Including private-sector research in global R&D capital shares would likely shift the balance back somewhat to developed countries. However, as noted above, it is difficult to determine with much precision how to apportion private R&D to individual countries. Using the estimated global series for private agricultural R&D and a 35-year R&D lag structure, private agricultural R&D accounted for 23% of global expenditures and 21% of R&D capital in 2011. The CGIAR accounted for about 1.2% of both global R&D expenditures and capital stocks. Research and Development Capital and Agricultural Growth The evidence linking R&D investment to productivity growth in agriculture is compelling, whether assessed for specific commodities, at the sector level for a country, or through international comparisons. Studies comparing the long-term performance of national agricultural sectors have consistently found that countries that invested more in agricultural R&D achieved higher agricultural productivity growth (Evenson and Kislev 1975; Craig, Pardey, and Roseboom 1997; Thirtle, Lin, and Piesse 2003; Gutierrez and Gutierrez 2003; Evenson and Fuglie 2010). Moreover, the value of the productivity improvement has been large relative to the cost of the R&D. Hurley et al. (2014) provide a comprehensive, critical assessment of 372 studies on returns to agricultural R&D and find a median (social) internal rate of return of 39%. Evidence on R&D-to-TFP Elasticities in Agriculture Table 2 summarizes results from 44 studies that econometrically estimated agricultural R&D-to-TFP elasticities for a country or set of countries. Several criteria distinguish this set of studies from other studies on agricultural growth. First, they are all based on times series evidence that includes relatively recent years. All of them included data on TFP to at least 1980, with about half of the studies extending to the post-2000 era. Second, all studies use a measure of knowledge capital that is based on accumulated R&D investment (measured by R&D spending or scientist-years). This is an advance over older work that had to rely on proxies for R&D capital, such as Hayami and Ruttan (1985), who used the number of graduates from technical schools, and Evenson and Kislev (1975) who used counts of agricultural science publications. A third important feature of most (31 out of 44) of these studies is that they use panel data time series of cross-sections of countries or sub-regions (states, provinces, or districts) within countries. This substantially increases the degrees of freedom and explanatory power of the models. Fourth, nearly all use whole sector agricultural TFP as their explanatory variable, rather than productivity of just one or a few commodities. Exceptions are Evenson (2003), who quantified the impact of the CGIAR research on food crop productivity, Suphannachart and Warr (2012), who examined the productivity of Thailand’s crop and livestock sectors separately, and Jin et al. (2002), who focused on sources of TFP growth for China’s three principal crops (rice, wheat, and maize). Finally, the studies are geographically diverse: 26 address productivity in developing countries (including separate coverage of Asia, Latin America, and Africa); 14 focus on developed countries, one covers (previously) centrally-planned economies, and three have worldwide coverage that includes both developed and developing countries. Table 2 Estimates of Agricultural R&D Elasticities Study Geographic coverage1 Period Data R&D Elasticities Total -all sources National public Int'l spill-in CGIAR Private Craig et al. (1997) World 1965-1990 88-country panel 0.10 0.10 Wiebe et al. (2000) World 1961-1997 88-country panel 0.16 0.16 Gutierrez et al. (2003)2 World 1970-1992 47-country panel 0.88 0.25 0.63 Schimmel. et al. (1999) EU & USA 1973-1993 11-country panel 0.31 0.11 0.20 0.00 Jin et al. (2016) USA 1970-2002 48-state panel 0.25 0.25 Wang et al. (2012) USA 1960-2002 48-state panel 0.29 0.29 Alston et al. (2011) USA 1949-2002 47-state panel 0.29 0.29 Huffman et al. (2006) USA 1970-1999 48-state panel 0.47 0.35 0.11 Andersen et al. (2013) USA 1949-2002 National 0.37 0.37 Wang et al. (2013) USA 1970-2009 National 0.57 0.43 0.14 Baldos et al. (2015) USA 1949-2011 National 0.31 0.31 Sheng et al. (2011) Australia 1953-2007 National 0.23 (national & foreign R&D combined) Mullen et al. (1995) Australia 1953-1988 National 0.26 0.26 Hall et al. (2006) New Zealand 1927-2001 National 0.50 0.15 0.00 Thirtle et al. (2008) UK 1953-2005 National 0.61 0.23 0.37 Bouchet et al. (1983) France 1959-1984 National 0.75 0.36 0.28 0.11 Butault et al. (2015) France 1959-2012 National 0.16 (national & foreign R&D combined) Wong (1986) Socialist 1950-1980 8-country panel 0.07 0.07 Johnson et al. (2000) LDC 1960s, 70s, 80s 90-country panel 0.13 0.03 0.10 Fulginiti et al. (1993) LDC 1961-1984 18-country panel 0.07 0.07 Craig et al. (1997) LDC 1965-1990 67-country panel 0.09 0.09 Thirtle et al. (2003) LDC 1985,90,95 48-country panel 0.44 0.44 Fan et al. (1998) Asia 1972-1993 12-country panel 0.17 0.17 Thirtle et al. (2003) Asia 1985,90,96 11-country panel 0.34 0.34 Evenson (2003)3 Asia 1970-2000 10 food crops 0.14 Evenson et al. (1991) Philippines 1948-1984 9-region panel 0.31 0.31 Rada & Fuglie (2012) Indonesia 1985-2005 22-province panel 0.36 0.27 0.09 Suphan. et al. (2012) Thailand 1971-2006 National 0.20 0.17 0.04 Jin et al. (2002) China 1981-1995 16-province panel 0.37 0.33 0.04 Fan (2000) China 1975-1997 25-province panel 0.25 0.25 Fan et al. (2002) China 1970-1997 29-province panel 0.09 0.09 Pray et al. (1991) Bangladesh 1947-1981 National 0.12 0.12 0.004 Rahman et al. (2013) Bangladesh 1948-2008 National 0.13 0.13 Fan et al. (2000) India 1970-1993 17-state panel 0.30 0.30 Evenson et al. (1999) India 1956-1987 271-district panel 0.17 0.05 0.11 0.01 Rada et al. (2015) India 1980-2008 16-state panel 0.28 0.17 0.11 Thirtle et al. (2003) LAC 1985,1990 15-country panel 0.20 0.20 Evenson (2003)3 LAC 1970-2000 10 food crops 0.05 Fernandez- et al. (1997) Mexico 1960-1990 National 0.64 0.13 0.36 0.14 Rada & Buccola (2012) Brazil 1985,96,06 558-district panel 0.03 0.03 Bervejillo et al. (2012) Uruguay 1981-2000 National 0.68 0.57 0.12 Thirtle et al. (1995) Africa 1971-1986 22-country panel 0.02 0.02 Thirtle et al. (2003) Africa 1985,1990 22-country panel 0.36 0.36 Lusigi et al. (1997) Africa 1961-1991 47-country panel 0.05 0.02 0.03 Evenson (2003) WANA 1970-2000 10 food crops 0.07 Fan et al. (2006) Egypt 1980-2000 3-region panel 0.25 0.25 Frisvold et al. (1995) SSA 1973-1985 28-country panel 0.08 0.08 Block (2014) SSA 1981-2000 27-country panel 0.20 0.20 Alene (2010) SSA 1986-2004 15-country panel 0.20 0.20 Fuglie et al. (2013) SSA 1977-2005 32-country panel 0.08 0.04 0.04 Evenson (2003)3 SSA 1970-2000 10 food crops 0.03 Study Geographic coverage1 Period Data R&D Elasticities Total -all sources National public Int'l spill-in CGIAR Private Craig et al. (1997) World 1965-1990 88-country panel 0.10 0.10 Wiebe et al. (2000) World 1961-1997 88-country panel 0.16 0.16 Gutierrez et al. (2003)2 World 1970-1992 47-country panel 0.88 0.25 0.63 Schimmel. et al. (1999) EU & USA 1973-1993 11-country panel 0.31 0.11 0.20 0.00 Jin et al. (2016) USA 1970-2002 48-state panel 0.25 0.25 Wang et al. (2012) USA 1960-2002 48-state panel 0.29 0.29 Alston et al. (2011) USA 1949-2002 47-state panel 0.29 0.29 Huffman et al. (2006) USA 1970-1999 48-state panel 0.47 0.35 0.11 Andersen et al. (2013) USA 1949-2002 National 0.37 0.37 Wang et al. (2013) USA 1970-2009 National 0.57 0.43 0.14 Baldos et al. (2015) USA 1949-2011 National 0.31 0.31 Sheng et al. (2011) Australia 1953-2007 National 0.23 (national & foreign R&D combined) Mullen et al. (1995) Australia 1953-1988 National 0.26 0.26 Hall et al. (2006) New Zealand 1927-2001 National 0.50 0.15 0.00 Thirtle et al. (2008) UK 1953-2005 National 0.61 0.23 0.37 Bouchet et al. (1983) France 1959-1984 National 0.75 0.36 0.28 0.11 Butault et al. (2015) France 1959-2012 National 0.16 (national & foreign R&D combined) Wong (1986) Socialist 1950-1980 8-country panel 0.07 0.07 Johnson et al. (2000) LDC 1960s, 70s, 80s 90-country panel 0.13 0.03 0.10 Fulginiti et al. (1993) LDC 1961-1984 18-country panel 0.07 0.07 Craig et al. (1997) LDC 1965-1990 67-country panel 0.09 0.09 Thirtle et al. (2003) LDC 1985,90,95 48-country panel 0.44 0.44 Fan et al. (1998) Asia 1972-1993 12-country panel 0.17 0.17 Thirtle et al. (2003) Asia 1985,90,96 11-country panel 0.34 0.34 Evenson (2003)3 Asia 1970-2000 10 food crops 0.14 Evenson et al. (1991) Philippines 1948-1984 9-region panel 0.31 0.31 Rada & Fuglie (2012) Indonesia 1985-2005 22-province panel 0.36 0.27 0.09 Suphan. et al. (2012) Thailand 1971-2006 National 0.20 0.17 0.04 Jin et al. (2002) China 1981-1995 16-province panel 0.37 0.33 0.04 Fan (2000) China 1975-1997 25-province panel 0.25 0.25 Fan et al. (2002) China 1970-1997 29-province panel 0.09 0.09 Pray et al. (1991) Bangladesh 1947-1981 National 0.12 0.12 0.004 Rahman et al. (2013) Bangladesh 1948-2008 National 0.13 0.13 Fan et al. (2000) India 1970-1993 17-state panel 0.30 0.30 Evenson et al. (1999) India 1956-1987 271-district panel 0.17 0.05 0.11 0.01 Rada et al. (2015) India 1980-2008 16-state panel 0.28 0.17 0.11 Thirtle et al. (2003) LAC 1985,1990 15-country panel 0.20 0.20 Evenson (2003)3 LAC 1970-2000 10 food crops 0.05 Fernandez- et al. (1997) Mexico 1960-1990 National 0.64 0.13 0.36 0.14 Rada & Buccola (2012) Brazil 1985,96,06 558-district panel 0.03 0.03 Bervejillo et al. (2012) Uruguay 1981-2000 National 0.68 0.57 0.12 Thirtle et al. (1995) Africa 1971-1986 22-country panel 0.02 0.02 Thirtle et al. (2003) Africa 1985,1990 22-country panel 0.36 0.36 Lusigi et al. (1997) Africa 1961-1991 47-country panel 0.05 0.02 0.03 Evenson (2003) WANA 1970-2000 10 food crops 0.07 Fan et al. (2006) Egypt 1980-2000 3-region panel 0.25 0.25 Frisvold et al. (1995) SSA 1973-1985 28-country panel 0.08 0.08 Block (2014) SSA 1981-2000 27-country panel 0.20 0.20 Alene (2010) SSA 1986-2004 15-country panel 0.20 0.20 Fuglie et al. (2013) SSA 1977-2005 32-country panel 0.08 0.04 0.04 Evenson (2003)3 SSA 1970-2000 10 food crops 0.03 Note: Superscript 1 indicates that LAC=Latin America & Caribbean; SSA=Sub-Saharan Africa, WANA=West Asia & North Africa, LDC=developing countries, DC=developed countries. 2 The Gutierrez et al. elasticity for international spill-ins was judged to be an outlier. 3 Evenson estimated elasticities of CGIAR R&D on yield for ten food crops. The elasticity of CGIAR R&D on agricultural TFP is the average food crop R&D elasticity times the share of these food crops in the total gross agricultural output over 1970-2000 (FAOSTAT). Table 2 Estimates of Agricultural R&D Elasticities Study Geographic coverage1 Period Data R&D Elasticities Total -all sources National public Int'l spill-in CGIAR Private Craig et al. (1997) World 1965-1990 88-country panel 0.10 0.10 Wiebe et al. (2000) World 1961-1997 88-country panel 0.16 0.16 Gutierrez et al. (2003)2 World 1970-1992 47-country panel 0.88 0.25 0.63 Schimmel. et al. (1999) EU & USA 1973-1993 11-country panel 0.31 0.11 0.20 0.00 Jin et al. (2016) USA 1970-2002 48-state panel 0.25 0.25 Wang et al. (2012) USA 1960-2002 48-state panel 0.29 0.29 Alston et al. (2011) USA 1949-2002 47-state panel 0.29 0.29 Huffman et al. (2006) USA 1970-1999 48-state panel 0.47 0.35 0.11 Andersen et al. (2013) USA 1949-2002 National 0.37 0.37 Wang et al. (2013) USA 1970-2009 National 0.57 0.43 0.14 Baldos et al. (2015) USA 1949-2011 National 0.31 0.31 Sheng et al. (2011) Australia 1953-2007 National 0.23 (national & foreign R&D combined) Mullen et al. (1995) Australia 1953-1988 National 0.26 0.26 Hall et al. (2006) New Zealand 1927-2001 National 0.50 0.15 0.00 Thirtle et al. (2008) UK 1953-2005 National 0.61 0.23 0.37 Bouchet et al. (1983) France 1959-1984 National 0.75 0.36 0.28 0.11 Butault et al. (2015) France 1959-2012 National 0.16 (national & foreign R&D combined) Wong (1986) Socialist 1950-1980 8-country panel 0.07 0.07 Johnson et al. (2000) LDC 1960s, 70s, 80s 90-country panel 0.13 0.03 0.10 Fulginiti et al. (1993) LDC 1961-1984 18-country panel 0.07 0.07 Craig et al. (1997) LDC 1965-1990 67-country panel 0.09 0.09 Thirtle et al. (2003) LDC 1985,90,95 48-country panel 0.44 0.44 Fan et al. (1998) Asia 1972-1993 12-country panel 0.17 0.17 Thirtle et al. (2003) Asia 1985,90,96 11-country panel 0.34 0.34 Evenson (2003)3 Asia 1970-2000 10 food crops 0.14 Evenson et al. (1991) Philippines 1948-1984 9-region panel 0.31 0.31 Rada & Fuglie (2012) Indonesia 1985-2005 22-province panel 0.36 0.27 0.09 Suphan. et al. (2012) Thailand 1971-2006 National 0.20 0.17 0.04 Jin et al. (2002) China 1981-1995 16-province panel 0.37 0.33 0.04 Fan (2000) China 1975-1997 25-province panel 0.25 0.25 Fan et al. (2002) China 1970-1997 29-province panel 0.09 0.09 Pray et al. (1991) Bangladesh 1947-1981 National 0.12 0.12 0.004 Rahman et al. (2013) Bangladesh 1948-2008 National 0.13 0.13 Fan et al. (2000) India 1970-1993 17-state panel 0.30 0.30 Evenson et al. (1999) India 1956-1987 271-district panel 0.17 0.05 0.11 0.01 Rada et al. (2015) India 1980-2008 16-state panel 0.28 0.17 0.11 Thirtle et al. (2003) LAC 1985,1990 15-country panel 0.20 0.20 Evenson (2003)3 LAC 1970-2000 10 food crops 0.05 Fernandez- et al. (1997) Mexico 1960-1990 National 0.64 0.13 0.36 0.14 Rada & Buccola (2012) Brazil 1985,96,06 558-district panel 0.03 0.03 Bervejillo et al. (2012) Uruguay 1981-2000 National 0.68 0.57 0.12 Thirtle et al. (1995) Africa 1971-1986 22-country panel 0.02 0.02 Thirtle et al. (2003) Africa 1985,1990 22-country panel 0.36 0.36 Lusigi et al. (1997) Africa 1961-1991 47-country panel 0.05 0.02 0.03 Evenson (2003) WANA 1970-2000 10 food crops 0.07 Fan et al. (2006) Egypt 1980-2000 3-region panel 0.25 0.25 Frisvold et al. (1995) SSA 1973-1985 28-country panel 0.08 0.08 Block (2014) SSA 1981-2000 27-country panel 0.20 0.20 Alene (2010) SSA 1986-2004 15-country panel 0.20 0.20 Fuglie et al. (2013) SSA 1977-2005 32-country panel 0.08 0.04 0.04 Evenson (2003)3 SSA 1970-2000 10 food crops 0.03 Study Geographic coverage1 Period Data R&D Elasticities Total -all sources National public Int'l spill-in CGIAR Private Craig et al. (1997) World 1965-1990 88-country panel 0.10 0.10 Wiebe et al. (2000) World 1961-1997 88-country panel 0.16 0.16 Gutierrez et al. (2003)2 World 1970-1992 47-country panel 0.88 0.25 0.63 Schimmel. et al. (1999) EU & USA 1973-1993 11-country panel 0.31 0.11 0.20 0.00 Jin et al. (2016) USA 1970-2002 48-state panel 0.25 0.25 Wang et al. (2012) USA 1960-2002 48-state panel 0.29 0.29 Alston et al. (2011) USA 1949-2002 47-state panel 0.29 0.29 Huffman et al. (2006) USA 1970-1999 48-state panel 0.47 0.35 0.11 Andersen et al. (2013) USA 1949-2002 National 0.37 0.37 Wang et al. (2013) USA 1970-2009 National 0.57 0.43 0.14 Baldos et al. (2015) USA 1949-2011 National 0.31 0.31 Sheng et al. (2011) Australia 1953-2007 National 0.23 (national & foreign R&D combined) Mullen et al. (1995) Australia 1953-1988 National 0.26 0.26 Hall et al. (2006) New Zealand 1927-2001 National 0.50 0.15 0.00 Thirtle et al. (2008) UK 1953-2005 National 0.61 0.23 0.37 Bouchet et al. (1983) France 1959-1984 National 0.75 0.36 0.28 0.11 Butault et al. (2015) France 1959-2012 National 0.16 (national & foreign R&D combined) Wong (1986) Socialist 1950-1980 8-country panel 0.07 0.07 Johnson et al. (2000) LDC 1960s, 70s, 80s 90-country panel 0.13 0.03 0.10 Fulginiti et al. (1993) LDC 1961-1984 18-country panel 0.07 0.07 Craig et al. (1997) LDC 1965-1990 67-country panel 0.09 0.09 Thirtle et al. (2003) LDC 1985,90,95 48-country panel 0.44 0.44 Fan et al. (1998) Asia 1972-1993 12-country panel 0.17 0.17 Thirtle et al. (2003) Asia 1985,90,96 11-country panel 0.34 0.34 Evenson (2003)3 Asia 1970-2000 10 food crops 0.14 Evenson et al. (1991) Philippines 1948-1984 9-region panel 0.31 0.31 Rada & Fuglie (2012) Indonesia 1985-2005 22-province panel 0.36 0.27 0.09 Suphan. et al. (2012) Thailand 1971-2006 National 0.20 0.17 0.04 Jin et al. (2002) China 1981-1995 16-province panel 0.37 0.33 0.04 Fan (2000) China 1975-1997 25-province panel 0.25 0.25 Fan et al. (2002) China 1970-1997 29-province panel 0.09 0.09 Pray et al. (1991) Bangladesh 1947-1981 National 0.12 0.12 0.004 Rahman et al. (2013) Bangladesh 1948-2008 National 0.13 0.13 Fan et al. (2000) India 1970-1993 17-state panel 0.30 0.30 Evenson et al. (1999) India 1956-1987 271-district panel 0.17 0.05 0.11 0.01 Rada et al. (2015) India 1980-2008 16-state panel 0.28 0.17 0.11 Thirtle et al. (2003) LAC 1985,1990 15-country panel 0.20 0.20 Evenson (2003)3 LAC 1970-2000 10 food crops 0.05 Fernandez- et al. (1997) Mexico 1960-1990 National 0.64 0.13 0.36 0.14 Rada & Buccola (2012) Brazil 1985,96,06 558-district panel 0.03 0.03 Bervejillo et al. (2012) Uruguay 1981-2000 National 0.68 0.57 0.12 Thirtle et al. (1995) Africa 1971-1986 22-country panel 0.02 0.02 Thirtle et al. (2003) Africa 1985,1990 22-country panel 0.36 0.36 Lusigi et al. (1997) Africa 1961-1991 47-country panel 0.05 0.02 0.03 Evenson (2003) WANA 1970-2000 10 food crops 0.07 Fan et al. (2006) Egypt 1980-2000 3-region panel 0.25 0.25 Frisvold et al. (1995) SSA 1973-1985 28-country panel 0.08 0.08 Block (2014) SSA 1981-2000 27-country panel 0.20 0.20 Alene (2010) SSA 1986-2004 15-country panel 0.20 0.20 Fuglie et al. (2013) SSA 1977-2005 32-country panel 0.08 0.04 0.04 Evenson (2003)3 SSA 1970-2000 10 food crops 0.03 Note: Superscript 1 indicates that LAC=Latin America & Caribbean; SSA=Sub-Saharan Africa, WANA=West Asia & North Africa, LDC=developing countries, DC=developed countries. 2 The Gutierrez et al. elasticity for international spill-ins was judged to be an outlier. 3 Evenson estimated elasticities of CGIAR R&D on yield for ten food crops. The elasticity of CGIAR R&D on agricultural TFP is the average food crop R&D elasticity times the share of these food crops in the total gross agricultural output over 1970-2000 (FAOSTAT). To make the R&D elasticities reported in table 2 as comparable as possible, in some cases the econometric results reported by the studies were adjusted. One adjustment is to report a “total elasticity” for public agricultural R&D, which is simply the sum of elasticities for different parts of the public research system. For example, Alston et al. (2011) report R&D elasticities for the effects of state research on own-state productivity (0.15), R&D spill-ins from other states (0.07), and spill-ins from federal-level intramural research carried out by the USDA (0.07).7 7 These research elasticities are taken from the authors’ preferred model 1 (Alston et al. 2011, table 2) and do not include the effects of extension. The percentage change in national agricultural TFP from a 1% increase in the R&D capital of each state and the USDA would be the sum of these elasticities (0.29). A second adjustment is when the R&D elasticity reported by the study refers to only a sub-sector, such as research on food crops. In these cases, the elasticity is multiplied by the revenue share of the sub-sector so that it shows how food crop R&D affects the average TFP of the whole agricultural sector.8 8 Suppose the growth rate in agricultural TFP can be written as the revenue-weighted growth rate of its sub-sectors, that is, lnA=ϕlnAc+1-ϕlnAl, where Ac is the TFP indexes for food crops, Al is the TFP of the rest of the agricultural sector, and ϕ is the revenue share of food crops in total agricultural output. Evenson’s results relate R&D capital by the CGIAR to productivity growth in the food crop sector only. His results give an estimate of γ in the equation Ac=Scgiarγ. Substituting this into the previous equation and taking the derivate with respect to lnScgiar gives ϕγ, which is an estimate of how CGIAR R&D capital affects TFP growth of the whole agricultural sector. Finally, two studies—Bouchet, Orden, and Norton (1983) on France, and Fernandez-Cornejo and Shumway (1997) on Mexico—used U.S. agricultural TFP as an explanatory variable for spill-ins of technologies developed in the United States and adopted in these countries. The U.S. agricultural TFP is itself a function of R&D spending by the U.S. public and private sectors, and in order to relate French and Mexican agricultural productivity directly to these external R&D capital stocks, I adjust the econometric estimates reported by these studies as described in the following paragraphs. First, we note that what the raw econometric results give us is an elasticity for national public agricultural R&D and total spill-ins from the United States. Using France as an example, this can be written as follows: Afr=Sfr-publicδ1,frAusaβ (5) where Afr and Ausa are agricultural TFP indexes for France and the United States, respectively, Sfr-public is French public R&D capital, and the parameters δ1,fr and β are the elasticities associated with the right-hand-side variables. Second, taking the average values from the U.S. productivity studies in table 2, we have the following relationship for the effects of public and private R&D capital on U.S. agricultural TFP: Ausa=Susa-public0.322Sprivate0.127 (6) Substituting equation (6) into equation (5) gives Afr=Sfr-publicδ1,frSusa-public0.322Sprivate0.127β. (7) Thus, the effect that U.S. public (private) R&D capital has on French agricultural productivity is found by multiplying the estimated coefficient β by 0.322 (0.127). For example, in Bouchet, Orden, and Norton’s (1983) study of agricultural growth in France, the authors found that each 1% increase in U.S. agriculture TFP was associated with a 0.857% rise in French agricultural TFP, which they associated with French farmers adopting innovations imported from the United States. Using the above procedure implies a spillover elasticity of 0.857 * 0.322 = 0.28 for USA public R&D capital, and 0.857 * 0.127 = 0.11 for private R&D on French agricultural TFP. These adjusted elasticities are reported in table 2. Given the scant data available on private R&D, some studies have used patent counts as a proxy for this variable. Patents are an R&D intermediate output rather than an R&D input, and in principle a procedure like the one in equations (5)–(7) should be used to adjust the elasticities accordingly. However, lacking a quantified relationship between R&D spending and patent output, we are forced to assume that these measures are perfectly correlated. The elasticities for private R&D reported in table 2 for the United Kingdom and New Zealand are based on patent counts rather than R&D spending. One general finding from the studies listed in table 2 is that public R&D investment has been the dominant source of agricultural TFP growth around the world. All of the studies in table 2, whether focusing on a particular country or comparing growth across countries, found that national public R&D explained a significant share of the growth in agricultural TFP. While these studies generally treat spill-in or private R&D (if modeled at all) as independent sources of technology, it is very likely that national research institutions help to adapt and disseminate these technologies locally. Fuglie and Rada (2013), for example, found that CGIAR technologies spread more rapidly in African countries with more national agricultural R&D capital. Studies have also found complementarities between public and private agricultural research: one sector’s R&D appears to raise the returns to the research of the other sector, presumably because they specialize in complementary parts of the science-technology spectrum (Schimmelpfennig and Thirtle 1999; Fuglie and Toole 2014). A second result is that all of the total R&D elasticities in table 2 are less than 1. This implies that total R&D spending will tend to rise faster than productivity growth. This is consistent with the finding that as countries develop their agricultural sectors, they experience a rise in research intensity (the ratio of agricultural R&D to agricultural GDP; Pardey et al. 2016; Heisey and Fuglie, forthcoming). This is strongly at odds with the assumptions of New Growth Theory, where it is assumed that a constant level of R&D spending will generate a constant growth rate for TFP (see Romer 1990, and footnote 3). A third conclusion is that, despite environmental constraints, international technology transfer is an important source of agricultural TFP growth. With the exception of Gutierrez and Gutierrez (2003), the studies in table 2 find that international R&D spill-ins have occurred mainly among developed countries located in temperate climates. Gutierrez and Gutierrez (2003) constructed a variable of the foreign R&D available for a country by taking the weighted average of the domestic agricultural R&D of its trade partners, using total import shares as weights. It is unclear why total import share should correlate with agricultural technology transfer, other than the fact that both are likely to be higher among nearby countries. In contrast, using agricultural patent data, Johnson and Evenson (1999) find that most international technology transfer occurs between high-income, temperate countries, and Eberhardt and Teal (2013) find that agricultural TFP growth is strongly correlated across similar agro-climatic environments. The spill-in elasticity reported by Gutierrez and Gutierrez (2003) is an outlier compared with the other spill-in elasticities in table 2, but it suggests the need for more work on understanding the role of international technology transfer in agricultural productivity growth. A fourth conclusion from table 2 is that there appears to be systematic variation in the elasticities of R&D among global regions. The regional variation can be more readily seen in table 3, which gives the averages of the elasticities from table 2 for various developed and developing regions. These average elasticities are consistent with the following observations: Table 3 Average Agricultural R&D Elasticities by Region Region/Sub-region World agricultural output share Source of R&D Capital Total - all sources National public Int'l public spill-in CGIAR Private World 1.000 0.43 0.18 0.10 0.04 0.10 Developed 0.283 0.67 0.27 0.21 0.20  North America 0.126 0.63 0.30 0.20 0.13  Western Europe 0.116 0.72 0.23 0.24 0.24  Australia-NZ-S  Africa 0.023 0.64 0.18 0.12 0.35  Japan-Korea- Taiwan 0.018 (no region-specific elasticities available) Transition 0.088 0.07 0.07 Developing 0.628 0.38 0.18 0.07 0.07 0.07  Asia 0.399 0.30 0.21 0.08 0.01  Latin America 0.118 0.77 0.23 0.36 0.05 0.13  Africa & West  Asia 0.111 0.19 0.15 0.04  Sub-Saharan  Africa only 0.057 0.17 0.13 0.04 Region/Sub-region World agricultural output share Source of R&D Capital Total - all sources National public Int'l public spill-in CGIAR Private World 1.000 0.43 0.18 0.10 0.04 0.10 Developed 0.283 0.67 0.27 0.21 0.20  North America 0.126 0.63 0.30 0.20 0.13  Western Europe 0.116 0.72 0.23 0.24 0.24  Australia-NZ-S  Africa 0.023 0.64 0.18 0.12 0.35  Japan-Korea- Taiwan 0.018 (no region-specific elasticities available) Transition 0.088 0.07 0.07 Developing 0.628 0.38 0.18 0.07 0.07 0.07  Asia 0.399 0.30 0.21 0.08 0.01  Latin America 0.118 0.77 0.23 0.36 0.05 0.13  Africa & West  Asia 0.111 0.19 0.15 0.04  Sub-Saharan  Africa only 0.057 0.17 0.13 0.04 Note: Sub-region R&D elasticities are the simple average of the elasticities from the sub-region given in table 2. Region elasticities are a weighted average of elasticities from the sub-regions, where the weights are the average share of gross agricultural output over 1980–2010 (FAOSTAT). Table 3 Average Agricultural R&D Elasticities by Region Region/Sub-region World agricultural output share Source of R&D Capital Total - all sources National public Int'l public spill-in CGIAR Private World 1.000 0.43 0.18 0.10 0.04 0.10 Developed 0.283 0.67 0.27 0.21 0.20  North America 0.126 0.63 0.30 0.20 0.13  Western Europe 0.116 0.72 0.23 0.24 0.24  Australia-NZ-S  Africa 0.023 0.64 0.18 0.12 0.35  Japan-Korea- Taiwan 0.018 (no region-specific elasticities available) Transition 0.088 0.07 0.07 Developing 0.628 0.38 0.18 0.07 0.07 0.07  Asia 0.399 0.30 0.21 0.08 0.01  Latin America 0.118 0.77 0.23 0.36 0.05 0.13  Africa & West  Asia 0.111 0.19 0.15 0.04  Sub-Saharan  Africa only 0.057 0.17 0.13 0.04 Region/Sub-region World agricultural output share Source of R&D Capital Total - all sources National public Int'l public spill-in CGIAR Private World 1.000 0.43 0.18 0.10 0.04 0.10 Developed 0.283 0.67 0.27 0.21 0.20  North America 0.126 0.63 0.30 0.20 0.13  Western Europe 0.116 0.72 0.23 0.24 0.24  Australia-NZ-S  Africa 0.023 0.64 0.18 0.12 0.35  Japan-Korea- Taiwan 0.018 (no region-specific elasticities available) Transition 0.088 0.07 0.07 Developing 0.628 0.38 0.18 0.07 0.07 0.07  Asia 0.399 0.30 0.21 0.08 0.01  Latin America 0.118 0.77 0.23 0.36 0.05 0.13  Africa & West  Asia 0.111 0.19 0.15 0.04  Sub-Saharan  Africa only 0.057 0.17 0.13 0.04 Note: Sub-region R&D elasticities are the simple average of the elasticities from the sub-region given in table 2. Region elasticities are a weighted average of elasticities from the sub-regions, where the weights are the average share of gross agricultural output over 1980–2010 (FAOSTAT). The overall significance of R&D-led growth, given by the total elasticity of the effects of R&D from all sources, is higher for developed countries than developing countries. This is primarily due to stronger technological linkages to other countries and the private sector. R&D-led growth is least developed for Africa, due to relatively weak public R&D institutions, as well as the absence of strong linkages to private and international R&D, except for the CGIAR. In Latin America, along with national R&D, international R&D spill-ins and private R&D appear to have made significant contributions to agricultural productivity growth. For a recent review article on the role of the private sector in agricultural technology transfer and innovation in developing countries, see Pray and Fuglie (2015). Even though it is a relatively small component of the global agricultural R&D infrastructure and focuses heavily on staple food crops, the CGIAR has had a noticeable impact on aggregate agricultural TFP growth in Asia, Africa, and Latin America. R&D Capital and TFP Growth in World Agriculture, 1990–2011 One way to assess the dependence of agricultural productivity on R&D capital is to compare predicted and observed rates of agricultural TFP growth over some historical period. The model in equation (3) says that the rate of change in R&D capital stock multiplied by the R&D elasticity is the contribution of that R&D to agricultural TFP growth. Here, using the average R&D elasticities from table 3, I compare predicted with actual agricultural TFP growth rates for major global regions between 1990 and 2011. Applying the 35-year gamma lag structure from figure 1 to public, private, and CGIAR research expenditures from 1955–2011, I calculate the rate of change in R&D capital K^i,2011K^i,1990 for each technology source i between 1990 and 2011.9 9 The R&D expenditure series assembled for this paper start in 1960, expect for a few countries (the United States, United Kingdom, Netherlands, Japan, Australia, New Zealand, and Mexico), which go back earlier. For the rest, R&D spending is estimated back to 1955 assuming the 1960–1969 growth rate in R&D spending. For private R&D, Fuglie’s (2016) global estimates are extended back from 1990 using the private R&D growth rate by U.S. firms (Fuglie et al. 2011). Using the R&D elasticities ( δi) from table 2, the predicted rate of agricultural TFP growth over this period, lnA^2011A^1990, for a region is then given by lnA^2011A^1990=δ1lnK^1,2011K^1,1990+∑i=24δilnK^i,2011K^i,1990 (8) where the first right-hand-side term measures the productivity contribution of domestic R&D capital for all countries in the region, and the second term refers to technology spill-ins from public R&D in other countries, the private sector, and the CGIAR, respectively. For international R&D capital, I use the sum of public R&D capital in other developed countries, since studies in table 3 providing estimates of elasticities for international spill-ins refer primarily to spill-ins among the United States, Western Europe, and Oceania. Figure 2 shows the result of this exercise. The height of the bars indicate the predicted rate of agricultural TFP growth for a region between 1990 and 2011, with the different segments of the bar referring to the contributions from each of the four sources of R&D capital. The horizontal lines indicate actual TFP growth over the same period, according to the international agricultural productivity accounts maintained by the USDA Economic Research Service.10 10 The Economic Research Service estimates agricultural TFP indexes for each country and region using standard growth accounting: TFP growth is the difference between the rate of growth in gross agricultural output (from FAOSTAT) and the weighted-average growth rate in inputs (land, labor, farm machinery capital, livestock capital, fertilizer, and feed), where the weights are input-cost shares. See Fuglie (2015) for a complete description of methods and sources. Figure 2 View largeDownload slide Growth in R&D capital and productivity in world agriculture, 1990–2011 Note: The height of the stacked bars show the predicted growth of agricultural TFP between 1990 and 2011, with the bar segments showing the respective contributions of different sources of technology. The horizontal lines show the actual growth in agricultural TFP over the same year period. Sources: Predicted TFP growth are the author’s estimates. Actual TFP growth is from the Economic Research Service. Figure 2 View largeDownload slide Growth in R&D capital and productivity in world agriculture, 1990–2011 Note: The height of the stacked bars show the predicted growth of agricultural TFP between 1990 and 2011, with the bar segments showing the respective contributions of different sources of technology. The horizontal lines show the actual growth in agricultural TFP over the same year period. Sources: Predicted TFP growth are the author’s estimates. Actual TFP growth is from the Economic Research Service. The model predictions align fairly closely to measured TFP growth for all the developed regions, Latin America and the Caribbean (LAC) and South Asia. In other words, over the two decades since 1990, R&D capital growth accounts for nearly all the agricultural TFP growth in these regions. For SE Asia, China, and Africa-West Asia, however, predicted TFP growth accounts for only a portion of measured TFP improvement during these years. The gap between predicted and measured TFP growth suggests that either (a) the impact of R&D capital is not being adequately captured by these models, and/or (b) factors other than R&D are also major drivers of efficiency and productivity improvement in these regions. Recent studies of agricultural growth in these regions do report significant impacts of other drivers, especially institutional and policy reforms, on agricultural TFP. For China, Fan (1991) finds that reforms that strengthened producer incentives and liberalized markets contributed significantly to raising agricultural productivity. Policies that improved agricultural terms of trade or liberalized markets were found to increase agricultural TFP in Sub-Saharan Africa (Fuglie and Rada 2013), Indonesia (Rada and Fuglie 2012), and India (Rada and Schimmelpfennig 2015). Nonetheless, a major difference between developed and developing countries revealed by this review is that the former are more adept at capturing technology spillovers from other countries and the private sector, and this has helped them maintain agricultural productivity growth despite a slowdown in the growth of public agricultural R&D spending. At the same time, an apparent lack of international R&D spillovers emanating from developing countries might represent a lost opportunity to raise agricultural productivity globally. Summary and Conclusions The agricultural growth model proposed in this paper treats knowledge capital like physical capital—that is, as a long-lived productive asset. But unlike physical capital, knowledge capital has the potential to generate spillovers—that is, applications beyond the locality or application for which it was originally intended. But because the agricultural production function is conditioned by environmental factors, spillovers may not come freely everywhere and require adaptation. Moreover, because agriculture is subject to biological evolution and environmental and social change, many agricultural technologies become obsolete over time—the curse of the Red Queen—and need to be refurbished or replaced or productivity will fall. These features of agricultural knowledge capital—its location-specific nature and eventual depreciation, distinguish it from the treatment of knowledge capital in New Growth Theory. The public-goods nature of knowledge capital and the small-holder structure of farming has implied a major role of government in investing in agricultural R&D. As recently as 2011, public institutions accounted for about three-quarters of total global spending on agricultural research. Of the accumulated spending on public agricultural R&D over the 50 years from 1962 to 2011 of $PPP 1,192 billion (constant 2011 PPP$), slightly less than half of this could be considered as operational R&D capital in 2011 (the difference being due to technological obsolescent and technologies still in the development pipeline). This R&D capital stock is roughly one-tenth the estimated $5 trillion worth of physical capital (not including land) held by the world’s farmers (FAO 2012). But because of its increasing returns (due to spillovers), agricultural knowledge capital generates much higher social returns than physical capital and thus accounts for an outsized share of agricultural growth. From 44 empirical studies on determinants of agricultural productivity from around the world, public investment in agricultural R&D was found to be a consistent driver of agricultural TFP growth. From these studies, the elasticity (the percentage increase in TFP or output from a 1% increase in R&D capital) for national public R&D capital averaged 0.18 world-wide; it was higher for developed countries (0.27) and significantly lower for sub-Saharan Africa (0.13). However, these studies also show that despite the sensitivity of agriculture to local environmental conditions, R&D spillovers across national borders are also an important source of agricultural productivity growth. It is likely that national R&D capacity enhances the movement of these technologies through adaptive research, and countries with stronger national R&D systems appear to realize larger technology spill-ins from other sources. Including the effects of public R&D spill-ins as well as spill-ins from private R&D widens the innovation gap between developed and developing countries: The average total R&D elasticity (which includes the effects of spillovers, private R&D, and the CGIAR) was much higher in developed countries (0.67) compared with developing countries (0.38), and especially with sub-Saharan Africa (0.17). The growth model posited in this paper implies that R&D investment will need to increase in order for agricultural productivity to continue to grow. Moreover, total agricultural R&D spending will likely need to grow faster than the desired rate of agricultural output growth. This is partly due to limits on the transferability of agricultural technologies across agro-ecologies, but also because of the need for maintenance research to keep productivity from falling. Another finding from this review is that so far there is little evidence that agricultural R&D investment by developing countries have been a significant source of international technology spillovers. This suggests that most developing country research has been focused on local adaption rather than advancing the productivity frontier. As countries like China supplant the United States and other developed countries as the primary investors in public agricultural R&D, concerns emerge about where the next generation of frontier technologies might come from. It is likely that agricultural R&D systems in leading developing countries will need to grow in sophistication, and that they adopt policies to encourage sharing or trading new agricultural technology in order for them to become capable of generating technologies with large spillover potential. This would raise the total R&D elasticity of agricultural innovation systems and significantly reduce the cost of achieving the productivity growth necessary to feed a richer, more populous world. References Alene A. 2010 . Productivity Growth and the Effects of R&D in African Agriculture . Agricultural Economics 41 ( 3-4 ): 223 – 38 . Google Scholar Crossref Search ADS Alston J. , Andersen M. , James J. , Pardey P. . 2010 . Persistence Pays: US Agricultural Productivity Growth and the Benefits from Public R&D Spending . New York : Springer . Alston J. , Andersen M. , James J. , Pardey P. 2011 . The Economic Returns to U.S. Public Agricultural Research . American Journal of Agricultural Economics 93 5 : 1257 – 77 . Google Scholar Crossref Search ADS Alston J. , Dehmer S. , Pardey P. . 2006 . International Initiatives in Agricultural R&D: The Changing Fortunes of the CGIAR. In Agricultural R&D in the Developing World: Too Little, Too Late ?, ed. Pardey P. , Alston J. , Piggott R. , 313 – 60 . Washington DC : International Food Policy Research Institute (IFPRI ). Alexandratos N. , Bruinsma J. . 2012 . World Agriculture Towards 2030/2050: The 2012 Revision. ESA Working Paper No. 12-03 . Rome, Italy : Food and Agriculture Organization . Andersen M. , Song W. . 2013 . The Economic Impact of Public Agricultural Research and Development in the United States . Agricultural Economics 44 3 : 287 – 95 . Google Scholar Crossref Search ADS Agricultural Science and Technology Indicators . Online Database. Washington DC: International Food Policy Research Institute. Accessed October 15, 2016. Baldos U. , Viens F. , Hertel T. , Fuglie K. . 2015 . On the Linkages in U.S. Public R&D Spending, Knowledge Capital and Agricultural Productivity Growth: A Bayesian Approach. Conference Paper, Agricultural and Applied Economics Association, San Francisco, CA, July 26 – 28 . Bervejillo J. , Alston J. , Tumber K. . 2012 . The Benefits from Public Agricultural Research in Uruguay . Australian Journal of Agricultural and Resource Economics 56 4 : 475 – 97 . Google Scholar Crossref Search ADS Block S. 2014 . The Post-independence Decline and Rise of Crop Productivity in Sub-Saharan Africa: Measurement and Explanations . Oxford Economic Papers 66 2 : 373 – 96 . Google Scholar Crossref Search ADS Bouchet F. , Orden D. , Norton G. . 1983 . Sources of Growth in French Agriculture . American Journal of Agricultural Economics 71 1 : 280 – 93 . Butault J.-P. , LeMarie S. , Musolesi A. , Huart F. , Simioni M. , Schmitt B. . 2015 . Impact de la Recherche Agronomique sur la Produtivité Agricole Française: Une Approche par le Taux de Rentabilité Interne (TRI) de la Recherche Publique. Rapport d’étude à destination du Collège de Direction de l’INRA. Working paper, version 22 January 2015. Paris, France: French National Institute for Agricultural Research. Craig B. , Pardey P. , Roseboom J. . 1997 . International Productivity Patterns: Accounting for Input Quality, Infrastructure and Research . American Journal of Agricultural Economics 79 4 : 1064 – 76 . Google Scholar Crossref Search ADS Eberhardt M. , Teal F. . 2013 . No Mangoes in the Tundra: Spatial Heterogeneity in Agricultural Productivity Analysis . Oxford Bulletin of Economics and Statistics 75 6 : 914 – 39 . Google Scholar Crossref Search ADS Evenson R. 2003 . Production Impacts of Crop Genetic Improvement. In Crop Variety Improvement and Its Effect on Productivity: The Impact of International Agricultural Research , ed. Evenson R. , Gollin D. , 447 – 472 . Wallingford, UK : CAB International . Evenson R. , Fuglie K. . 2010 . Technology Capital: The Price of Admission to the Growth Club . Journal of Productivity Analysis 33 3 : 173 – 90 . Google Scholar Crossref Search ADS Evenson R. , Kislev Y. . 1975 . Agricultural Research and Productivity . New Haven, CT : Yale Univ. Press . Evenson R. , Pray C. , Rosegrant M. . 1999 . Agricultural Research and Productivity Growth in India . Research Report 109. Washington, DC : IFPRI . Evenson R. , Quizon J. . 1991 . Technology, Infrastructure, Output Supply, and Factor Demand in Philippine Agriculture. In Research and Productivity in Asian Agriculture , ed. Evenson R. , Pray C. , 195 – 205 . Ithaca, NY : Cornell University Press . Fan S. 1991 . Effects of Technological Change and Institutional Reform on Production Growth in Chinese Agriculture . American Journal of Agricultural Economics 73 2 : 266 – 75 . Google Scholar Crossref Search ADS Fan S. 2000 . Research Investment and the Economic Returns to Chinese Agricultural Research . Journal of Productivity Analysis 14 2 : 163 – 82 . Google Scholar Crossref Search ADS Fan S. , Al-Riffai P. , El-Said M. , Yun B. , Kamaly A. . 2006 . A Multi-level Analysis of Public Spending, Growth and Poverty Reduction in Egypt. DSGD Discussion Paper No. 41. Washington DC: International Food Policy Research Institute (IFPRI). Fan S. , Hazell P. , Thorat S. . 2000 . Government Spending, Growth and Poverty in Rural India . American Journal of Agricultural Economics 82 4 : 1038 – 51 . Google Scholar Crossref Search ADS Fan S. , Pardey P. . 1998 . Government Spending on Asian Agriculture: Trends and Production Consequence. In Agricultural Public Finance Policy in Asia . Tokyo : Asian Productivity Organization (APO ). Fan S. , Zhang L. , Zhang X. . 2002 . Growth, Inequality and Poverty in Rural China: The Role of Public investments . Research Report 125. Washington DC : International Food Policy Research Institute (IFPRI ). FAOSTAT. Online Database. Rome: Food and Agriculture Organization. Accessed September 15, 2016 . Food and Agriculture Organization . 2012 . State of Food and Agriculture 2012: Investing in Agriculture for a Better Future . Rome : Food and Agriculture Organization (FAO ). Feder G. , Just R. , Zilberman D. . 1985 . Adoption of Agricultural Innovations in Developing Countries: A Survey . Economic Development and Cultural Change 33 2 : 255 – 98 . Google Scholar Crossref Search ADS Federico G. 2005 . Feeding the World: An Economic History of Agriculture, 1800-2000 . Princeton, NJ : Princeton University Press . Fernandez-Cornejo J. , Shumway C. . 1997 . Research and Productivity in Mexican Agriculture . American Journal of Agricultural Economics 79 3 : 738 – 53 . Google Scholar Crossref Search ADS Foster A. , Rosenzweig M. . 1996 . Technical Change and Human-Capital Returns and Investments: Evidence from the Green Revolution . American Economic Review 86 4 : 931 – 53 . Frisvold G. , Ingram K. . 1995 . Sources of Agricultural Growth and Stagnation in Sub-Saharan Africa . Agricultural Economics 13 1 : 51 – 61 . Google Scholar Crossref Search ADS Fuglie K. 2015 . Accounting for Growth in Global Agriculture . Bio-based and Applied Economics 4 3 : 221 – 54 . Fuglie K. 2016 . The Growing Role of the Private Sector in Agricultural Research and Development World-wide . Global Food Security 10 : 29 – 38 . Google Scholar Crossref Search ADS Fuglie K. , Heisey P. , King J. , Pray C. , Day-Rubenstein K. , Schimmelpfennig D. , Wang S. , Karmarkar-Deshmukh R. . 2011 . Research Investments and Market Structure in the Food Processing, Agriculture Input and Biofuel Industries Worldwide . Washington DC : U.S. Department of Agriculture , Economic Research Report No. 130. Fuglie K. , Rada N. . 2013 . Resources, Policy and Agricultural Productivity in Sub-Saharan Africa . Washington DC : U.S. Dept. of Agriculture, Economic Research Service , Economic Research Report No. 145. Fuglie K. , Toole A. . 2014 . The Evolving Institutional Structure of Public and Private Agricultural Research . American Journal of Agricultural Economics 96 3 : 862 – 83 . Google Scholar Crossref Search ADS Fulginiti L. , Perrin R. . 1993 . Prices and Productivity in Agriculture . Review of Economics and Statistics 75 3 : 471 – 82 . Google Scholar Crossref Search ADS Gutierrez L. , Gutierrez M. . 2003 . International R&D Spill-overs and Productivity Growth in the Agricultural Sector: A Panel Cointegration Approach . European Review of Agricultural Economics 30 3 : 281 – 303 . Google Scholar Crossref Search ADS Hall J. , Scobie G. . 2006 . The Role of R&D in Productivity Growth: The Case of Agriculture in New Zealand, 1927 to 2001. Working Paper 06/01, New Zealand Treasury, Wellington. Hayami Y. , Ruttan V. . 1985 . Agricultural Development: An International Perspective . Baltimore, MD : The Johns Hopkins University Press . Heisey P. , Fuglie K. . Forthcoming. Agricultural Research Policy in High-Income Countries. Washington DC: U.S. Dept. of Agriculture, Economic Research Service. Economic Information Bulletin. Huffman W. , Evenson R. . 2006 . Science for Agriculture: A Long-Term Perspective . Ames, IA : Blackwell . Hurley T. , Rao X. , Pardey P. . 2014 . Re-examining the Reported Rates of Return to Food and Agricultural Research and Development . American Journal of Agricultural Economics 96 5 : 1492 – 504 . Google Scholar Crossref Search ADS Jin S. , Huang J. , Hu R. , Rozelle S. . 2002 . The Creation and Spread of Technology and Total Factor Productivity in China’s Agriculture . American Journal of Agricultural Economics 84 4 : 916 – 30 . Google Scholar Crossref Search ADS Jin Y. , Huffman W. . 2016 . Measuring Public Agricultural Research and Extension and Estimating their Impacts on Agricultural Productivity: New Insights from U.S. Evidence . Agricultural Economics 47 1 : 15 – 31 . Google Scholar Crossref Search ADS Johnson D. , Evenson R. . 1999 . R&D Spillovers to Agriculture: Measurement and Application . Contemporary Economic Policy 17 4 : 432 – 56 . Google Scholar Crossref Search ADS Johnson D. , Evenson R. 2000 . How Far away Is Africa? Technological Spill-overs to Agriculture and Productivity . American Journal of Agricultural Economics 82 5 : 743 – 49 . Google Scholar Crossref Search ADS Jorgenson D. , Fukao K. , Timmer M. , eds. 2016 . The World Economy: Growth or Stagnation ? Cambridge, UK : Cambridge University Press . Judd M. , Boyce J. , Evenson R. . 1991 . Investment in Agricultural Research and Extension Programs: A Quantitative Assessment. In Research and Productivity in Asian Agriculture , ed. Evenson R. , Pray C. 7 – 46 . Ithaca : Cornell University Press . Lains P. , Pinilla V. , eds. 2009 . Agriculture and Economic Development in Europe Since 1870 . London : Routledge . Lusigi A. , Thirtle C. . 1997 . Total Factor Productivity and the Effects of R&D in African Agriculture . Journal of International Development 9 4 : 529 – 38 . Google Scholar Crossref Search ADS Mullen J. , Cox T. . 1995 . The Returns from Research in Australian Broadacre Agriculture . Australian Journal of Agricultural Economics 39 2 : 105 – 28 . Google Scholar Crossref Search ADS Organisation for Economic Cooperation and Development. Online Database. Research and Development Statistics. Paris, France. Accessed October 12, 2016. Olmstead A. , Rhode P. . 2002 . The Red Queen and the Hard Reds: Productivity Growth in American Wheat, 1800-1940 . Journal of Economic History 62 4 : 929 – 66 . Google Scholar Crossref Search ADS Otsuka K. 2013 . Food Insecurity, Income Inequality, and the Changing Comparative Advantage in World Agriculture . Agricultural Economics 44 ( s1 ): 7 – 18 . Google Scholar Crossref Search ADS Pardey P. , Chan-Kang C. , Dehmer S. , Beddow J. . 2016 . Agricultural R&D is on the Move . Nature 537 (15 September): 301 – 3 . Google Scholar Crossref Search ADS PubMed Pardey P. , Roseboom J. . 1989 . ISNAR Agricultural Research Indicator Series: A Global Database on National Agricultural Research Systems . Cambridge : Cambridge University Press . Pardey P. , Roseboom J. , Anderson J. , eds. 1991 . Agricultural Research Policy: International Quantitative Perspectives . Cambridge, UK : Cambridge University Press . Pray C. , Ahmed Z. . 1991 . Research and Agricultural Productivity in Bangladesh. In Research and Productivity in Asian Agriculture , ed. Evenson R. , Pray C. , 114 – 32 . Ithaca : Cornell University Press . Pray C. , Fuglie K. . 2015 . Agricultural Research by the Private Sector . Annual Review of Resource Economics 7 : 399 – 424 . Google Scholar Crossref Search ADS Rada N. , Fuglie K. . 2012 . Shifting Sources of Agricultural Growth in Indonesia. In Productivity Growth in Agriculture: An International Perspective , ed. Fuglie K. , Wang S. , Ball E. , 199 – 214 . Wallingford, UK : CABI . Rada N. , Schimmelpfennig D. . 2015 . Propellers of Agricultural Productivity in India . Washington DC : U.S. Dept. of Agriculture, Economic Research Service , Economic Research Report No. 203. Rada N. , Buccola S. . 2012 . Agricultural Policy and Productivity: Evidence from Brazilian Censuses . Agricultural Economics 43 4 : 353 – 65 . Google Scholar Crossref Search ADS Rahman S. , Salim R. . 2013 . Six Decades of Total Factor Productivity Change and Sources of Growth in Bangladesh Agriculture (1948–2008) . Journal of Agricultural Economics 64 2 : 275 – 94 . Google Scholar Crossref Search ADS Romer P. 1990 . Endogenous Technological Change . Journal of Political Economy 98 : S71 – S102 . Google Scholar Crossref Search ADS Ruttan V. 1982 . Agricultural Research Policy . Minneapolis : University of Minnesota Press . Ruttan V. 2001 . Productivity Growth in World Agriculture: Sources and Constraints . Journal of Economic Perspectives 16 4 : 161 – 84 . Google Scholar Crossref Search ADS Schimmelpfennig D. , Thirtle C. . 1999 . The Internationalization of Agricultural Technology: Patents, R&D Spill-overs, and Their Effects on Productivity in the European Union and the United States . Contemporary Economic Policy 17 4 : 457 – 68 . Google Scholar Crossref Search ADS Schultz T. 1975 . The Value of the Ability to Deal with Disequilibria . Journal of Economic Literature 13 (3): 827 – 46 . Sheng Y. , Gray E. , Mullen J. , Davidson A. . 2011 . R&D and Extension: An Analysis of the Static and Dynamic Effects on Australian Broadacre Productivity. ABARES Research Report 11.7. Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. Suphannachart W. , Warr P. . 2012 . Total Factor Productivity in Thai Agriculture: Measurement and Determinants. In Productivity Growth in Agriculture: An International Perspective , ed. Fuglie K. , Wang S. , Ball E. , 215 – 36 . Wallingford, UK : CABI . Thirtle C. , Hadley D. , Townsend R. . 1995 . Policy-induced Innovation in Sub-Saharan African Agriculture: A Multilateral Malmquist Productivity Index Approach . Development Policy Review 12 : 323 – 48 . Google Scholar Crossref Search ADS Thirtle C. , Piesse J. , Schimmelpfennig D. . 2008 . Modeling the Length and Shape of the R&D Lag: An Application to UK Agricultural Productivity . Agricultural Economics 39 : 73 – 85 . Google Scholar Crossref Search ADS Thirtle C. , Lin L. , Piesse J. . 2003 . The Impact of Research-led Agricultural Productivity Growth on Poverty Reduction in Africa, Asia and Latin America . World Development 31 1 : 1959 – 75 . Google Scholar Crossref Search ADS U.S. Department of Agriculture, Economic Research Service. Online Database: International Agricultural Productivity. Washington DC. Accessed November 15, 2016. Wang S. , Heisey P. , Schimmelpfennig D. , Ball E. . 2015 . Agricultural Productivity Growth in the United States: Measurement, Trends, and Drivers . Washington DC : U.S. Department of Agriculture , Economic Research Report No. 189. Wang S. , Ball E. , Fulginiti L. , Plastina A. . 2012 . Accounting for the Impacts of Public Research, R&D Spillins, Extension, and Roads in U.S Regional Agricultural Productivity Growth, 1980–2004. In Productivity Growth in Agriculture: An International Perspective , ed. Fuglie K. , Wang S. , Ball E. , 13 – 32 . Wallingford, UK : CABI . Wang S. , Heisey P. , Huffman W. , Fuglie K. . 2013 . Public R&D, Private R&D, and U.S. Agricultural Productivity Growth: Dynamic and Long-run Relationships . American Journal of Agricultural Economics 95 5 : 1287 – 93 . Google Scholar Crossref Search ADS Wiebe K. , Soule M. , Narrod C. , Breneman V. . 2000 . Resource Quality and Agricultural Productivity: A Multi-country Comparison. Conference Paper, American Agricultural Economics Association, Tampa, FL. Wong Lung-Fai. 1986 . Agricultural Productivity in the Socialist Countries . Boulder, CO : Westview . World Development Indicators. Online Database. Washington DC: World Bank. Accessed October 15, 2016. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association 2017. This work is written by a US Government employee and is in the public domain in the US. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Applied Economic Perspectives and PolicyOxford University Press

Published: Sep 1, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off