Variation and determinants of rice yields among individual paddy fields: Case study of a large-scale farm in the Kanto region of Japan

This study aims to identify the variation and determinants of rice yield measured by IT combine, among different rice yields within large-scale farms. In addition to the yield of the paddy with 15% of moisture, unsorted and sorted brown rice, variations of concerning ratios are studied as well. The sample includes 351 paddy fields from a farm corporation scaled over 113 ha, locating in the Kanto region of Japan. The candidate determinants include the field area and condition, nitrogen amount, time of transplanting or seeding, stage-specific growth indictors for the contain of chlorophyll, number of panicles, plant height, and leaf plate value. In addition, soil properties, average temperature and solar radiation are incorporated. Meanwhile, varieties, cultivation methods and soil types are adopted as dummy variables. The empirical analysis is conducted using a multivariate linear regression, with logarithmic transformations of the continuous variables. The estimation result indicates that panicle numbers in full-heading stage and transplanting or sowing time are the most important continuous determinants, following by nitrogen amount, humus content, and so forth. Within the significant discrete determinants, Akidawara and Milky queen are measured as productive and unproductive variety, respectively; while the well-drained direct sowing method is identified as negatively affecting the yield.