Comparative application of geographical detector and Tobit model in analysis of grain production efficiency in the western Guangdong region and its influencing factors

This study aims to clarify the grain production efficiency in western Guangdong and its key influencing factors and to optimize the efficiency of resource allocation, promoting the transformation and enhancement of grain production. Based on the grain production data of three cities in western Guangdong, from 1999 to 2018, a cross efficiency DEA model was employed to calculate the regional grain production efficiency and reveal almost two decades of temporal and spatial evolution characteristics of the grain production efficiency. The key influencing factors of the grain production efficiency were compared and analyzed by using the Tobit model and geographical detector model. Results showed that the average grain production efficiency in western Guangdong was 0.77; Maoming City exhibited the highest efficiency(0.88), followed by Yangjiang City(0.78)and Zhanjiang City(0.65). The most important input factor(contribution)of the Tobit and geographical detector models was the total power of agricultural machinery(-1.082 and 0.942, respectively). However, the Tobit model demonstrated that the seven input factors affecting the grain production efficiency were positively correlated only with the amount of chemical fertilizer applied(0.015); that the effective irrigated area, grain sown area, and pesticide use amount were not correlated with the grain production efficiency; and that the remaining factors were negatively correlated. All factors measured by the geographical detector had a significant influence on grain production efficiency. The contribution of input factors in Yangjiang was similar to that in western Guangdong Province, but the contribution of the amount of chemical fertilizer was 0.008 higher than that of the effective irrigation area. The contribution of chemical fertilizer and pesticide input in Zhanjiang and Maoming was second only to the total power of agricultural machinery and the contribution of effective irrigation area was the lowest. However, the contribution of the area in which grain was sown in Maoming was higher than that of the employed population and agricultural plastic film, whereas Zhanjiang showed the opposite trend. In general, the grain production efficiency of western Guangdong during the past 20 years has not been high, and has remained at a medium level. The data demonstrate that grain production efficiency decreases first and then increases significantly. There are no insignificant factors in the measurement results of the geographical detector and the influence of detection factors is more stable and less biased than in the Tobit model. The total power of agricultural machinery is the leading factor affecting the grain production efficiency of the three cities in western Guangdong, but the contribution of the factors in different cities is quite different, Zhanjiang and Maoming are more dependent on the chemical fertilizer and pesticide pollution.