A study covering twelve different commodities and us­ing a multicountry trade model, found that spillover effects from regions where research is conducted to over regions with similar agroecologies and rural infrastructures ranged from 64 to 82% of total international benefits (Davis et al., 1987). An analysis of 69 national and international wheat improvement research programs found that given the mag­nitude of potential spill-ins from the international research system, many wheat programs could significantly increase the efficiency of resource use by reducing the size of their wheat research programs and focusing on the screening of varieties developed elsewhere (Davis et al., 1987). The im­pact of research conducted within individual Latin Ameri­can and Caribbean countries covering edible beans, cassava, maize, potatoes, rice, sorghum, soybeans and wheat showed that when allowance was made for spillovers to other re­gions of the world, the resulting price impacts had impor­tant consequences for the distribution of benefits between producer and consumers and thus among countries within Latin America and the Caribbean (Alston et al., 2000b). At least for the United States, the locational range of spill-in effects for crop production is lower than for livestock pro­duction (Evenson, 1989). Crop genetic improvements in the United States had spillover effects into the rest of the world, with consumers in the rest of the world gaining but produc­ers outside the United States losing (Frisvold et al., 2003). Overall increases in net global welfare from United States crop improvements were distributed 60% to the US, 25% to other industrialized countries, and the remainder to de­veloping and transitional economies.
          Growth in public funding for international research has slowed over the last twenty years (see 9.1). Thus, un­derstanding the ROR of the CGIAR is very important, in­cluding the spill in and spill out impacts. Over the years, a number of studies have attempted to value the benefits to particular countries from research conducted at CG centers, in some cases comparing them against donor support pro­vided by the countries in question (Brennan, 1986, 1989; Burnett et al., 1990; Byerlee and Moya, 1993; Bofu et al., 1996; Fonseca et al., 1996; Pardey et al., 1996; Brennan and Bantilan, 1999; Johnson and Pachico, 2000; Brennan et al., 2002; Heisey et al., 2002). For the period 1973-1984, Australia gained US$747 million in terms of cost savings to wheat producers as a United States benefit from its adoption of wheat varieties from CIMMYT and rice varieties from IRRI (Brennan, 1986, 1989). Depending on the attribution rule used, the United States' economy gained at least US$3.4 billion and up to US$14.6 billion from 1970 to 1993 from the use of improved wheat varieties developed by CIMMYT and US$30 million and up to US$1 billion through the use of rice varieties developed by IRRI.8 These estimates did not account for the world price impact as a result of the rest of the world having adopted CIMMYT wheat varieties and thereby driving down the price of wheat.
         Assessments were made of Australia's benefits from research conducted by ICRISAT and ICARDA taking ex­plicit account of the world price impacts (Brennan and Ban­tilan, 1999; Brennan et al., 2002). Research on sorghum
8 For a discussion of the issues related to these estimate see Alston (2002) and Pardey et al. (2002).

(ICRISAT) resulted in a national benefit of US$3.6 mil­lion (producer loss of US$1.7 million and consumer gain of US$5.3 million) for Australia. Similarly, ICRISAT's re­search on chickpeas would have given a national benefit of US$1.2 million (producer loss of US$2.6 million and a consumer gain of US$3.8 million). The average estimated net gain to Australia as a result of the overall research effort at ICARDA in five crops (durum wheat, barley, chick pea, lentils and faba bean) is US$7.4 million per year (in 2001 dollars and exchange rates) over the period to 2002 (Bren­nan et al., 2002). This represents 1% of the gross value of Australia's production of the five crops. Most of those gains are achieved in the faba bean and lentil industries. Producers receive most of the welfare gains in Australia, amounting to US$6.5 million of the total. The main findings of the various studies are (Alston, 2002):
•   Intra national and international spillovers of public ag­ricultural AKST results are very important.
•   Spillovers can have profound implications for the dis­tribution of benefits from research between consumers and producers and thus among countries, depending on their trade status and capacity to adopt the technology.
•   It is not easy to measure these impacts, and the results can be sensitive to the specifics of the approach taken, but studies that ignore spillovers are likely to obtain se­riously distorted estimates of ROR.
•   Because spillovers are so important, research resources have been misallocated both within and among na­tions.

The estimation of these state, national or multinational im­pacts is data intensive, difficult, and adds to the measure­ment problems (Alston, 2002). However, there can be little doubt that agricultural AKST generates very large benefits and that a very large share of those benefits comes through spillovers. The omission or mismeasurement of spillover ef­fects may have contributed to a tendency to overestimate ROR to agricultural AKST in some instances. Clearly, the issue of international research spillovers is an important one for the allocation of resources for research both nationally and internationally. The spillover benefits to industrialized countries from international agricultural research have pos­itive funding implications. More work is needed in this area to develop better methods to measure spillovers and also to develop the necessary policy institutional arrangements to harness the full potential of spillover effects of AKST tech­nologies (Alston, 2002; Anandajayasekeram et al., 2007).
        Agricultural machinery and agricultural chemicals are obvious cases where industrial AKST is directed towards the improvement of agricultural inputs. Recent studies conclude that when new industrial products first come on the market, they are priced to only partially capture the real value of the improvement (most new models of equipment are better buys than the equipment that they replace) (Evenson, 2001). This produces a spill-in impact. Another type of spill-in that is recognized in few studies is the "recharge" spill-in from pre-invention science. Many of the studies summarized in the meta-analysis actually covered a wide range of research program activities including many pre-invention science ac­tivities. Some studies specifically identified pre-invention ex-