基于ARIMA与预期性指标的我国通货膨胀预测研究
发布时间：2018-07-15 13:51
【摘要】：中央银行货币政策的有效实施,企业和家庭对通货膨胀风险的理性规避,都需要对通货膨胀进行准确预测。由于通货膨胀的发生机制和影响因素非常复杂,通货膨胀的预测应该包含科学性和艺术性。先是采用科学性较强的ARIMA模型对我国通货膨胀进行预测。对2008年至2015年的月度和季度CPI数据,建立ARIMA模型,进行样本内预测,并与真实数据作比较。采用均方预测误差、平均绝对预测误差、平均正确预测方向三个预测精度指标,检验月度和季度通货膨胀预测值的预测精度。随后,采用艺术性较强的预期性指标法预测我国通货膨胀。选择四个典型的通货膨胀预期性指标,利用三个预测精度指标比较分析。结果发现,万得CPI预测指数和朗润预测的加权平均指数预测精度较高,能够分别有效预测月度和季度通货膨胀。再基于预期性指标法和ARIMA模型,构建科学性和艺术性都较强的组合预测模型,预测我国月度和季度通货膨胀。利用三个预测精度指标,比较预期性指标法、ARIMA模型和组合预测模型的预测精度。结果发现,组合预测模型的预测精度最高,预期性指标法的预测精度较高。政策建议是使用大数据技术进行通货膨胀数据挖掘,以提高通货膨胀预测中所需的数据质量;对于中央银行,因为货币政策效果主要在政策实施的半年后显现,需要构建出高精度的半年度和年度通货膨胀预期性指标,而且运用组合预测模型进行通货膨胀预测;对于中小企业和家庭,简单方便起见,可以运用预期性指标法预测通货膨胀。
[Abstract]:The effective implementation of central bank monetary policy and rational evasion of inflation risk by enterprises and households all need to predict inflation accurately. Because the mechanism and influencing factors of inflation are very complicated, the forecast of inflation should be scientific and artistic. First, Arima model is used to forecast inflation in China. Arima model was established for monthly and quarterly CPI data from 2008 to 2015. The mean square prediction error, the average absolute prediction error and the average correct forecast direction are used to test the forecast accuracy of the monthly and quarterly inflation forecast values. Then, the author predicts China's inflation by using the artistically strong expectation index method. This paper selects four typical inflation expectation indexes and makes a comparative analysis by using three prediction precision indicators. The results show that the weighted average index predicted by Wande CPI and Runrun is more accurate and can effectively forecast monthly and quarterly inflation respectively. Then, based on the expected index method and Arima model, a combination forecasting model with strong scientific and artistic character is constructed to forecast the monthly and quarterly inflation in China. The prediction accuracy of Arima model and combined prediction model is compared by using three prediction precision indexes. The results show that the combined prediction model has the highest prediction accuracy, and the predictive index method has higher prediction accuracy. Policy advice is to use big data technology to mine inflation data to improve the quality of data needed for inflation forecasting; for central banks, the effect of monetary policy is mainly apparent half a year after the implementation of the policy. There is a need to construct high-precision semi-annual and annual inflation expectations indicators and to use composite forecasting models to forecast inflation; for small and medium-sized enterprises and households, simplicity and convenience, We can forecast inflation by using the expected index method.
【学位授予单位】：青岛大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F822.5
本文编号：2124287
[Abstract]:The effective implementation of central bank monetary policy and rational evasion of inflation risk by enterprises and households all need to predict inflation accurately. Because the mechanism and influencing factors of inflation are very complicated, the forecast of inflation should be scientific and artistic. First, Arima model is used to forecast inflation in China. Arima model was established for monthly and quarterly CPI data from 2008 to 2015. The mean square prediction error, the average absolute prediction error and the average correct forecast direction are used to test the forecast accuracy of the monthly and quarterly inflation forecast values. Then, the author predicts China's inflation by using the artistically strong expectation index method. This paper selects four typical inflation expectation indexes and makes a comparative analysis by using three prediction precision indicators. The results show that the weighted average index predicted by Wande CPI and Runrun is more accurate and can effectively forecast monthly and quarterly inflation respectively. Then, based on the expected index method and Arima model, a combination forecasting model with strong scientific and artistic character is constructed to forecast the monthly and quarterly inflation in China. The prediction accuracy of Arima model and combined prediction model is compared by using three prediction precision indexes. The results show that the combined prediction model has the highest prediction accuracy, and the predictive index method has higher prediction accuracy. Policy advice is to use big data technology to mine inflation data to improve the quality of data needed for inflation forecasting; for central banks, the effect of monetary policy is mainly apparent half a year after the implementation of the policy. There is a need to construct high-precision semi-annual and annual inflation expectations indicators and to use composite forecasting models to forecast inflation; for small and medium-sized enterprises and households, simplicity and convenience, We can forecast inflation by using the expected index method.
【学位授予单位】：青岛大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F822.5
【参考文献】
相关期刊论文 前10条
1 应意楠;;美国的通货膨胀建模及预测:基于ARMA模型的实证研究[J];财经界(学术版);2016年17期
2 严钰童;崔自洪;;利率期限结构理论综述[J];全国商情;2016年24期
3 肖良;;基于季节性ARIMA模型的居民消费水平预测[J];统计与决策;2016年08期
4 薛晔;蔺琦珠;任耀;;我国通货膨胀风险的预测模型——基于决策树-BP神经网络[J];经济问题;2016年01期
5 郭永济;丁慧;范从来;;中国通货膨胀动态模型预测的实证研究[J];中国经济问题;2015年05期
6 杨颖梅;;基于ARIMA模型的北京居民消费价格指数预测[J];统计与决策;2015年04期
7 蒋传进;;组合预测中的预测精度与预测风险分析[J];统计与决策;2015年03期
8 徐国祥;代吉慧;;中国大宗商品现货价格指数的构建及预测能力研究[J];统计研究;2014年12期
9 贾凯威;韩家彬;杨洋;;基于SARIMA模型的短期通货膨胀预测[J];统计与决策;2014年22期
10 金雯雯;陈亮;毛德勇;叶茜茜;;利率期限结构内含的宏观经济信息——基于TVP-VAR模型的时变参数研究[J];经济评论;2014年05期
相关硕士学位论文 前2条
1 刘海萍;中国主要股指的分形分析与BP神经网络预测[D];大连理工大学;2013年
2 杨俏;基于调查数据的通货膨胀预测方法研究[D];中国人民大学;2008年
，本文编号：2124287
本文链接：https://www.wllwen.com/jingjilunwen/huobiyinxinglunwen/2124287.html
