The unobserved variable in probit model is a function of a set of explanatory
variables (X i ’s). The relationship between the latent variable and the explanatory
variables is expressed as
Y
Ã
i ¼ X
0
i b þ ei; ei $ Nð0; r
2
Þ
Given that the model follows a normal distribution, the probabilities of having or
not having crop insurance is given as
P i ¼ P Y ¼ 1jX
ð
Þ¼PðY
Ã
i ! I i Þ ¼ PfX
0
i b þ e i [ 0g ¼ FðX
0
i bÞ
P Y ¼ 0jX
ð
Þ¼1 À P i ¼ 1 À FðX
0
i bÞ
The parameters in this model are estimated with maximum likelihood procedure.
If the coefficient of the explanatory variable, β is positive, it implies that the
probability of availing insurance increases with increase in the explanatory variable.
On the other hand, a negative coefficient would imply that the probability of having
insurance decreases with respect to increase in explanatory variable. The likelihood
function of the model is given below
LðbÞ ¼
Y N
i¼1
PðY i ¼ 1jx i ; bÞ
Y i PðY i ¼ 1jx i ; bÞ
1ÀY i
Since the present analysis is done using cross-sectional observations, there are
chances of heteroscedasticity problem. To solve it robust standard errors has been
used for hypothesis testing purpose instead of ordinary standard error.
3.3 Results
The results of the regression analysis are presented in the Table 3. The table shows
that both demand side factors and supply side factors play an important role in
determining crop insurance. It is observed that probability of availing insurance is
significantly high in inland north regions of Karnataka, which is a semi-arid region.
One expects supply of insurance to be less here. However, a positive sign implies
that demand side factors and not supply mainly determines crop insurance. Higher
presence of crop insurance in North Karnataka as well mean that presence of
production risk induces households to avail insurance.
The regression result shows that there are differences in probability of availing
insurance across caste and class. General caste households have a higher probability
of availing insurance. Similarly one also observes that land owned by the household
to be positively related with probability of availing crop insurance. Households
belonging to higher caste as well as higher class may be financially more aware.
Second, it may also happen that for having better social networking in financial
markets, they have better information.
248
M. Rajeev et al.
variables (X i ’s). The relationship between the latent variable and the explanatory
variables is expressed as
Y
Ã
i ¼ X
0
i b þ ei; ei $ Nð0; r
2
Þ
Given that the model follows a normal distribution, the probabilities of having or
not having crop insurance is given as
P i ¼ P Y ¼ 1jX
ð
Þ¼PðY
Ã
i ! I i Þ ¼ PfX
0
i b þ e i [ 0g ¼ FðX
0
i bÞ
P Y ¼ 0jX
ð
Þ¼1 À P i ¼ 1 À FðX
0
i bÞ
The parameters in this model are estimated with maximum likelihood procedure.
If the coefficient of the explanatory variable, β is positive, it implies that the
probability of availing insurance increases with increase in the explanatory variable.
On the other hand, a negative coefficient would imply that the probability of having
insurance decreases with respect to increase in explanatory variable. The likelihood
function of the model is given below
LðbÞ ¼
Y N
i¼1
PðY i ¼ 1jx i ; bÞ
Y i PðY i ¼ 1jx i ; bÞ
1ÀY i
Since the present analysis is done using cross-sectional observations, there are
chances of heteroscedasticity problem. To solve it robust standard errors has been
used for hypothesis testing purpose instead of ordinary standard error.
3.3 Results
The results of the regression analysis are presented in the Table 3. The table shows
that both demand side factors and supply side factors play an important role in
determining crop insurance. It is observed that probability of availing insurance is
significantly high in inland north regions of Karnataka, which is a semi-arid region.
One expects supply of insurance to be less here. However, a positive sign implies
that demand side factors and not supply mainly determines crop insurance. Higher
presence of crop insurance in North Karnataka as well mean that presence of
production risk induces households to avail insurance.
The regression result shows that there are differences in probability of availing
insurance across caste and class. General caste households have a higher probability
of availing insurance. Similarly one also observes that land owned by the household
to be positively related with probability of availing crop insurance. Households
belonging to higher caste as well as higher class may be financially more aware.
Second, it may also happen that for having better social networking in financial
markets, they have better information.
248
M. Rajeev et al.