where Fi j is the area farm I allocates to grain j. Hence, vi is the weighted average of the
relative yields of all grains.
To estimate the effects of program participation on yields in 1997 we utilize the
following method: The actual relative yields observed in 1997 (vi,1997) can be explained by the
hypothetical yields we would have observed without agri-environmental program (hi,1997) and
the actual participation in agri-environmental programs. Program participation is modelled by
dummy variables (Di). For the four programs which require the farm as a whole to participate
dummies are set to 1 for participants and 0 for non-participants. For the six programs with
partial participation, the dummies are equal to the share of the acreage in the program to total
area under tillage (i.e., the dummy can range from 0 to 1). Hence, assuming a linear
relationship the impact of program participation can be estimated by
10
(3) V, 1997 = Ôo + Ô1 hi, 1997 + ∑(βkDk,1997 ) + ɛi
k=1
The coefficients оk depict the influence of program participation on relative yields. To
get an estimate of hi,1997, the yields we would have observed in 1997 without agri-
environmental program participation, we utilize the data of 1994, the year before agri-
environmental programs were introduced. In particular, we estimate relative yields in 1994
vi,1994 as dependent on the 1994-values of variables designed to model farm characteristics
that are thought to influence yields. Tilled area (AREA; as a proxy for economies of scale;
expected sign: + ), the ratio of tilled to total farm area (RATIO, a proxy for specialization;
expected sign: +), unit value per hectare (UV, this is a variable compiled for tax purposes; it
includes soil characteristics, climate etc. ; expected sign: +), and animal-units per hectare
(AUH, to somehow account for the unrecorded amount of manure, which is typically disposed