基于平衡轮换样本调查的季节调整方法研究

传统季节调整方法对时间序列数据进行季节调整时，往往假定误差项为白噪声，不考虑其序列相关关系。为了进行更准确地季节调整分析，本文从连续性抽样调查的角度出发，研究基于平衡轮换样本调查的抽样误差对季节调整的影响，建立一般化的季节调整模型，利用卡尔曼滤波进行参数估计，并从预测误差、误差方差等角度评价模型精度。最后以中国城镇住户调查采用的12～0平衡轮换模式为例，对考虑抽样误差结构特征的季节调整模型进行实证分析，验证这套季节调整方法的有效性。

Research on Seasonal Adjustment Method Based on the Balanced Rotation Sample Survey

Traditional seasonal adjustment methods are focusing on building the model and estimation for time series economic sector, assuming that error is a white noise, ignoring the correlativity of the error’s structure. In order to develop seasonal adjustment accuracy, we study the influence of sampling error to seasonal changes based on the balanced rotation sample survey, processed from successive sampling survey. Then, this paper builds the general seasonal adjustment model, using Kalman filtering to estimate the parameters, and evaluates the model accuracy using forecast error and error variance. At last, taking 12～0 rotation scheme of China’s urban household survey as an example, we make the empirical analysis of the seasonal adjustment model including sampling error in order to verify the effectiveness of the method.