基于GRNN算法的潜在产出与产出缺口估算模型

Potential output and output gap model based on GRNN algorithm

基于神经网络学习性、非参数估计的优势,建立估算经济潜在产出和产出缺口的广义回归神经网络(GRNN)模型,对中国产出缺口和潜在产出增速的1978—2015年数据进行估算、2016—2020年数据进行多步预测,实证分析较好地克服了生产函数法中由于事先假设确定的函数形式及选择不同函数结果不同的问题,并且所估算的潜在产出增速与经济增速具有较高的契合度。模型分析得出:中国经济增长速度大致维持在潜在产出增长率水平附近,便可实现通货膨胀率维持在合理水平,且较好地实现总体供需平衡的目的。

Based on the advantages of learning and non-parametric estimating of neural networks, we built a general regression neural network （GRNN） model to estimate the output gap and the potential output growth of China according to the data from 1978 to 2015, and predict the data from 2016 to 2020. Meanwhile, the relationship between the mentioned data and the economic growth was empirically analyzed. The model can overcome the problems of the function form determined by the presupposition in the production function method and different results coming from different functions, and the estimated output growth has a high degree of fit with the economic growth. Based on the model analysis, it is concluded that if China＇s economic growth rate is maintained at the level of potential output growth rate, the inflation rate will be maintained at a reasonable level, and the balance between supply and demand will be well achieved.