The results estimated based on the spatial one way fixed effect are shown in Table
6. The parameter coefficients did not change as we move from one way fixed effect
panel data model to the one way fixed spatial effect model. The results show significant
decline in R2 in the spatial model. A spatial effect was present in quadratic specification
of phosphorus and quadratic and cubic specification of dissolved oxygen.
Conclusions
We estimated panel (regular and spatial) data models to determine whether the
quantifiable amount of social capital as used through social capital index can explain the
pollution differences across parishes. We used highly disaggregated water pollution data
collected at the watershed level. Results show significant role of social capital in
explaining nitrogen pollution but not phosphorus and dissolved oxygen. We did not find
an inverted U-shaped curve between pollutants and social capital. Rather, most of the
effects were U-shaped indicating higher nitrogen pollution is associated with both low
and high levels of social capital. The turning points for all pollutants were around 0.5
value of the social capital index. This indicates that the “middle amount” of social capital
is good for the environment. Spatial effects were found in phosphorus and dissolved
oxygen but parameter insignificance in these pollutants raises questions about the validity
of the models. Results indicated the need to further analyzed the data using a
nonparametric regression approach.
17