基于动态贝叶斯网络的通货膨胀风险预测研究  

Study of Inflation Risk Prediction Based on Dynamic Bayesian Network

作　　者：贾德奎[1] 王双成[2] 卞世博[3] 

机构地区：[1]上海立信会计学院经贸学院,上海201620 [2]上海立信会计学院数学与信息学院,上海201620 [3]上海立信会计学院金融学院,上海201620

出　　处：《会计与经济研究》2014年第6期81-98,共18页Accounting and Economics Research

基　　金：国家自然科学基金项目(71301105)

摘　　要：利用动态朴素贝叶斯分类器对我国的通货膨胀风险进行了预测,并在此基础上计算了不同风险因素对通货膨胀风险等级的影响程度,结果表明,动态朴素贝叶斯分类器方法能够较好地预测和分析通货膨胀风险。利用我国实际经济数据进行模拟预测,结果显示在88.24%的可信度下,我国2012年第二季度的通货膨胀风险等级为2级,通货膨胀率预测值约为3.8%。通过分别使用包含和不包含企业家信心指数的动态朴素贝叶斯分类器进行预测,结果发现考虑了企业家信心指数的分类器大约能将预测准确率提高5%,这说明对通货膨胀风险等级预测有必要考虑经济主体的预期因素。通过计算6个指标对通货膨胀风险等级的影响程度,发现货币供应和产出缺口仍是解释我国通货膨胀风险的两个主要因素。In this paper,dynamic naive Bayesian classifier （DNBC ）is used to esti-mate the inflation risk in China and the effect of different risk factors on the inflation risk. DNBC is proven to work well for inflation rate forecasting and risk analysis.The result with real economic data in China estimated in DNBC shows that the level of inflation risk is 2 with 88.24%confidence level,that is,the predicted inflation rate is approximately 3.8%in second quarter of 2012 in China.Further estimate is conducted with entrepreneur confi-dence index taken into DNBC.The result shows that the estimation accuracy of DNBC is approximately 5% improved with the entrepreneur confidence index incorporated,which indicates that expectations of economic entities should be considered in inflation forecas-ting.At last,the paper calculates the effect of six different risk factors on the inflation risk,and finds that money supply and output gap are two key explanations for the inflation risk in China.

关 键 词：动态贝叶斯网络 通货膨胀风险 产出缺口 货币供应量 

分 类 号：F832[经济管理—金融学]