An application of a cardinality-constrained multiple benchmark tracking error model on a plant enterprise selection problem
研究了如何用带基数约束的多基准跟踪误差模型，从中国25年省级数据中选出高回报低风险的种植企业组合，帮助农民在多样化与风险间取得平衡。
© Oxford University Press and Foundation for the European Review of Agricultural Economics 2018; all rights reserved. Yield and return of plants grown in a region are generally closely related. Agricultural scientists are less likely to recommend a single-plant enterprise for a region because of risk and return concerns. From a risk/return perspective, a plant enterprise selection problem can be considered as a portfolio optimisation problem. We use a multiple benchmark tracking error (MBTE) model to select an optimal plant enterprise combination under two goals. A cardinality constraint (CC) is used to efficiently balance multiple objectives and limit over-diversification in a region. We use Chinese national and province level datasets from multiple plant enterprises over 25 years to identify the best plant enterprise combination with two objectives under consideration: return maximisation and risk minimisation. A simulated case using discrete programming is applied in order to analyse a farmer's choice of specific plant enterprise and the transaction cost during rotation. In the continuous problem, the MBTE model is found to be efficient in choosing plant enterprises with high returns and low risk. The inclusion of a CC in the MBTE model efficiently reduces the plant enterprise number and volatility while creating smaller tracking errors than the MBTE model alone in an out-of-sample test. In the discrete problem, a CC can be used to search for the optimal number of plant enterprises to obtain high returns and low risk. The study and methods used can be helpful in choosing an optimal enterprise combination with multiple objectives when there are over-diversification concerns.