Estimation Results
Equation (6) depicts the results of regression (4) (t-ratios in parentheses):
(6) vi,1994 = 0.638 + 0.0020 AREA1994 + 0.170RATIO1994 + 0.106 UV1994 + 0.060 AUH1994 +
(21.7) (5.1) (6.11) (8.8) (5.9)
+ 0.062 D1 1997 -0.169 D2 1997 - 0.165 D3 1997 + 0.023 D4 1997 - 0.027 D5 1997 -
(2.7) (7.8) (4.7) (1.5) (1.3)
- 0.074 D6,1997 - 0.053 D7,1997 - 0.0003D8,1997 - 0.132D9,1997 -? 0.129 D10,1997
(4.3) (1.1) (0.0) (2.1) (1.1)
R2 = 0.37
(the critical t-value for a regression with 1383 observations and 15 independent variables is tcrit=1.96)
For a panel regression, the value of R2 is quite satisfactory; moreover, it can be seen
that farms that later participated in the OEPUL-programs, even in 1994 exhibited quite
diverse grain yields (e.g., farms that in 1997 were to participate in the program #9 “Non-
application of fungicides” showed on average 17% lower than farms that were not to
participate).
The covariance matrix derived in regression (6) is used to perform the Monte Carlo
simulations described above. Table 2 presents the results of 2000 times running regression
(3). Mean coefficients values and t-ratios along with the lower and upper 5%-limits of their
respective distributions are presented and can be interpreted in the following way: For
example, participating in the OEPUL program “organic farming” reduces yields on average
(of our 2000 regressions) by 11%. In 95% of our 2000 regressions the negative impact on
yields from program participation is between 7% and 14.5%. The average t-value is -4.38 and
in 95% or our regressions it is between -2.65 and -6.04. Hence, organic farming a statistical
negative impact on yields. Beside organic farming significant negative impacts on yields are
only estimated for participation in the “extensive crop cultivation” program. Participation in
the program “Non-application of agro-chemicals, whole farm” has a negative impact on