production variables for both commodities are available from the National Agricultural Statistics
Service (USDA), which maintains historical agricultural statistics for many commodities. The
cost of production is an average per acre cost of producing cotton and peanuts in the Southeast.
Chris McIntosh of University of Idaho provided the data for the amount of rainfall in the four-
state region. The expected prices and average costs of production were deflated in 1982 dollars
using the Consumer Price Index (CPI) published by the Bureau of Economic Analysis. The
public agricultural research expenditures are in constant 1984 prices, or real terms. The nominal
or current values were deflated by the Huffman and Evenson public agricultural research price
index (Huffman & Evenson, 1993). Based on the theory that research spillovers occur within
similar geo-climatic regions, all the data was averaged or aggregated to create a region-wide data
set for each commodity over the 33 years from 1963 through 1995.
The first step in calculating returns to research in cotton and peanuts for this study is to
directly estimate the supply function of each commodity. The exact functional form of supply
equations for both crops is as follows:
lnQt = β0 + βi(lnPt*) + βi(lnWt) + βi(lnτt) + ft(lnUt) +βi(lnAt), (13)
In both of the above equations, the state of technology (τt) is defined as a Pascal lag of research
expenditures. The research expenditure variable data is lagged seven years to account for the
time it takes for an investment in research to produce results3.
3 The research lag assumption of seven years is reasonable based on the findings of previous empirical studies such
as Evenson’s 1968 study that found a mean research lag for all agricultural research in the United States was
between 6 and 8 years.
11