values. Furthermore, the aggregated studies do not take into account multiplier effects that result from stimulated growth in the nonfarm economy, or nonmarket benefits. As a result, even the most generous of the values reported may be considered as conservative (Raitzer, 2003).

8.2.4.6 Rates of return to agricultural R&D investments in sub-Saharan Africa
A compilation of the available case studies on ROR for African agricultural R&D investment support findings of four meta-analyses (Table 8-13). Of the 27 RORs to past investments in agricultural technology development and dissemination (TDT), 21 show RORs in excess of 12%. Detailed investigations into the lower RORs suggest that researchers had not yet found the right mix of activities to produce cost-effective solutions in challenging agroecologi-cal environments. Examining the future potential impact of innovations released or still in the development stage, 24 of 30 forward-looking RORs show expected returns in ex­cess of 12%. The second study reviewed the impact stud­ies conducted in Eastern and Southern Africa (ESA) during 1978-2005 (Anandajayasekeram et al., 2007). The RORs for those studies using the noneconometric methods ranged from 0 to 109. For those studies using the econometric methods ranged between 2 to 113%. Only 10 out of the 86 observations were below 12% under the worst case sce­nario. These compilations confirm that returns to research in SSA are similar to those found elsewhere, showing a high payoff for a wide range of programs.

8.2.5 Environmental impacts of AKST investments
The success of modern agriculture in recent decades has of­ten masked significant externalities that have positively and negatively affected natural resources. Externalities of agri­culture include the depletion of resources such as fossil fuel, water, soil and biodiversity; pollution of the environment by the products of fuel combustion, pesticides and fertilizers; and economic and social costs to communities. In the past, the objectives of AKST investments have been largely to in­crease quality, quantity and to improve food security. Thus

the environmental impacts of agricultural technologies were not usually considered in ROR and other decision-making tools. Their importance, however, is increasingly understood because of the positive link between ecologically sustainable development and poverty reduction (UNEP, 2004a). A good example at the national level is a study that estimates the total external environmental and health costs of "modern agriculture" in the United Kingdom at a total cost of £2343 million in 1996 (Pretty et al., 2000). This is equivalent to 89% of average net farm income and £208 per hectare of arable and permanent pasture. These estimates only include those externalities that give rise to financial costs, and so are likely to underestimate the total negative impacts to the environment.
         The quality and size of environmental impacts depend on many external forces. Different agroecological zones, market conditions, and financial and social incentives as well as specific technologies play significant roles in determining impacts. In order to quantify and value the environmental impact of an agricultural R&D investment, it is important to understand the source of the impact, the nature, and the relationship between the impact and those variables that can affect producers and consumers. Agriculture globally (including livestock and land use, but excluding transport of agricultural products) has an estimated contribution to GHG emissions of 32% (Stern, 2007).
          Although many economic valuation techniques have been developed and refined over the last twenty years, ob­taining monetary values of environmental impacts is difficult due to two basic reasons. First, the issue of time and scale complicates the data collection and valuation. Many of the environmental impacts are accumulative by nature, and thus time is a critical constraint in estimating values. Geographic scale is also critical, for example, captive shrimp production leads to a large-scale pollution of marine environments and destruction of mangroves (Clay, 2004). In general, environ­mental and ecological economists consider the scale either through an ecosystem-centric lens: the plot, the farm, the watershed, and region (Izac and Swift, 1994) or a human-centric lens: the individual farmer, the local community,