Agricultural Economics in the Context of Portuguese Rural Development - The Agricultural Economics of the 21st Century - page 130

Moran's I to evaluate the existence of spatial autocorrelation effects in the regres-
sion (null hypothesis means no spatial autocorrelation effects). The values for
Moran's I were obtained with GeoDa ( 2014 ). The presence of these statistical
infractions implies that the results are biased and consequently the conclusions
obtained are unadjusted. In some cases there is sign of heteroskedasticity that was
corrected with the OLS robust econometric method.
In the two tables the coefficient of the constant varies between 6 and 8 (always
with statistical significance), which signifies that there are other variables that
influence agricultural output in Portuguese municipalities in addition to those
considered (this may be an interesting topic for future studies).
On the other hand the coefficient for agricultural employment is, also, always
statistically significant and presents positive values around 1.
In relation to the other variables, only a number of farms with forestry and with
renewable energy production have negative effects on the agricultural output
( 0.305, respectively). Among the other activities related to agricul-

ture (Table 9.1 ), the stronger effect comes from the accommodation and restaurants
output and also from the wood and cork industry (respectively 0.121 and 0.108). In
multifunctional agricultural activities the most important effects come from the
number of farms with service provisions (0.104).
In any case the effects of the other activities related with farming and the
alternative production that can be developed within farms have a marginal effect
over agricultural performance, with values for the coefficients at around 0.1.
The values of the statistical tests (Breusch-Pagan/Cook-Weisberg test for
heteroskedasticity, the specification Ramsey RESET test using powers of the fitted
values, and Moran's I to evaluate the existence of spatial autocorrelation effects in
regression) show that there are no problems with the associated statistical infrac-
tion. There are, only, some problems with the heteroskedasticity that was resolved
by using the OLS robust.
These data and results show that more concentration upon economic activity is
needed, namely those related with agriculture, in the rural zones of the interior and
that more capacity to develop alternative activities is also needed, for farming,
inside and outside of farms that bring more income to the farmers. To have more
positive externalities in agricultural output from industry and services, it is deter-
minant whether these activities are located more in rural zones nearer to the farming
sector.

0.083 and

Conclusions
The revision of literature related to the issue raised in this study shows that
rural development has many problems, with difficult solutions, namely those
related with low population density, weak economic activity, and a lack of
infrastructures.

(continued)