Assessing the impact of RDP agri-environment measures on the use of nitrogen-based mineral fertilizers through spatial econometrics: The case study of Emilia-Romagna (Italy)

Agriculture is the main source of nitrogen loading (EEA, 2012) and is the sector with the largest residual emission reduction potential (Sutton et al., 2011). Moreover, surpluses of nitrogen are forecast to grow in the next decade (FAO, 2012). The objective of this study is to evaluate the determinants of the use of N-based mineral fertilizers in agriculture and the effectiveness of Agri-Environmental Schemes (AES) implemented in Emilia-Romagna, Italy, in preventing nitrate pollution. The indicator N mineral fertilizer application rate in agriculture was first estimated at the municipality level for the years 2000 and 2010. Thereafter, we performed a Moran's statistics and a LISA (Local Indicators of Spatial Association; Anselin, 1995) analysis to test the data for local spatial autocorrelations. Finally, in order to provide a quantitative evaluation of the impact of the agri-environmental measures on the application rate of N fertilizers, we constructed an aspatial model (Ordinary Least Square model) and two spatial models (spatial lag and error models). All of the models are able to explain more than 70% of the change in the N mineral fertilizer application rate between the years 2000 and 2010 (the dependent variable). The results indicate that the observed decrease in the application rate of N mineral fertilizers in the considered decade was positively influenced by both the uptake of specific AES and the location in Nitrate Vulnerable Zones (NVZ). Among the policy variables, the participation in AES is less important than the location in a NVZ for explaining the reduction in the N mineral fertilizer application rate in the municipalities of Emilia-Romagna. Other significant variables are farm size, population density and share of certified organic surface in the utilized agricultural area (UAA). The availability of finer scale data for the estimation of changes in nitrogen inputs would improve the robustness of the models.