The aggregate approach to the evaluation ofproduction oriented expenditure in the UnitedStates has consistently shown high rates ofreturn to agricultural research investments.These efforts, while showing the value of agricultural research, are limited in the information they can provide policy and budget decisionmakers.More recently, efforts such as the analysisof four major commodity groups by Bredahl andPeterson and cross sectional studies by Evenson,White, and Havlicek have begun refining the levelof analysis to individual regions and states andfor specific commodities (Evenson, Bredahl andPeterson, White and Havlicek).The objective of this study is to further disaggregate these commodity groupings of Bredahland Peterson into individual commodities and tobegin investigating the impact of interregionalresearch "spillovers." Case studies of corn,wheat, and sorghum are made, using individualstates as observation units over the time periodfor which research data on individual commoditiesare available. The empirical results presentedin this paper are the results of some firstattempts at estimation. Further work is beingdone to improve specification of variablesmeasuring weather, cash inputs, risk, and otherfactors influencing yield response of grain commodities.A special cross sectional-time seriesalgorithm is used in parameter estimation.Theoretical Framework, Model, and DataThe conceptual framework of this study isparticularly constrained by the availability ofonly 11 years of research expenditure data onindividual commodities (1967-1977). The shortnumber of years prohibits use of a 12-, to 13 yearpolynominal Almon distributed lag which hasbeen used in previous studies to investigate theresearch lag structure (White, et al., Quanceand Lu). The nonavailability of productioninput data for specific commodities for otherthan farm census years also limits utilizing thetraditional aggregate production functionapproach to the analysis of agriculture researchexpenditure.Faced with these data limitation problems,the framework of a supply response analysis isdeveloped as an alternative to investigating theimpact on research expenditures on productivityof individual grain commodities. The supplyresponse model which is derivable from the production function, expresses the quantity of acommodity offered for sale as a function of inputand output prices, technical parameters, and avariety of shifters, such as weather and technological change. Supply analysis studies havetypically represented the effects of technologicalchange as a linear trend variable. While the research evaluation studies have not focusedon the processes of technological adoption anddiffusion, their analyses have shown that publicinvestments in agricultural research activitieshave contributed to increases in productivity.On this basis, lagged research expenditures onindividual commodities will be used as the technological change shifter in the commodity response functions.
