我国流通产业景气波动监测预警研究

发布时间：2018-04-25 14:37

本文选题：流通产业 + 景气预警　；参考：《浙江工商大学》2017年硕士论文

【摘要】：在国民经济中,流通产业占据着战略地位,它串联着经济体中的多个行业,一方面它为生产部门带来了需求信息、提供转移生产成果的去路,另一方面也为消费部门供应着商品与服务。随着经济模式持续变更,流通产业也发生着诸多变化,本文正是基于景气指数对其周期波动特征进行了探析,并构建出预警模型估计其发展方向与幅度。首先,本文确认了流通产业景气分析的基准指标——流通产业增加值增速,并从流通产业的产业整体发展情况、产业自支持度、宏观支持度三个方面构建了我国流通产业景气波动监测预警指标体系。其次,针对基准指标,本文通过EEMD时频分解法将其分解成了七个固有频率不同的分量,从而初步分辨出我国流通产业基准周期的结构及特征。结果显示,(1)从整个研究期限来看,短期因素的冲击对流通产业的影响较大,根据短期波动的特征,可以将流通产业划分为"低幅高频-高幅低频-低幅高频"三个阶段;(2)从临近年份来看,中长期因素的负面效应是导致近期流通产业存在滑落倾向的主因。再次,在经过数据预处理,取得指标体系中各指标的循环因子后,本文通过时差相关法、峰谷法与系统聚类法辨别出各指标与基准指标循环因子相比的先行、一致与滞后性,其中,系统聚类法对指标间距离的测量采用的切比雪夫距离,对类间距的测量采用的ward法;在此基础上,本文分别对这三类指标编制了景气合成指数,并用所有指标计算出综合景气指数。结合景气指数的走势与EEMD信号分解结果可知:(1)在研究期限内,我国流通产业经历了五个周期,其中第一周期长度最长,第二个周期震荡最剧烈,后三个周期正位于金融危机过后的调整期,波动相对平稳;(2)根据先行指数的走势,可以定性预计下一季度流通产业将加速下滑,并且这在很大程度上归因于流通产业固定资产投资增速的下滑以及货物运输行业的发展不济。最后,本文通过"μ-σ法"将流通产业综合景气指数划分为冷热不同的五个警情等级。然后将综合景气指数作为输出变量,经时间平移后的先行与一致指标作为输入变量,基于适用于连续型变量回归的BP-ANN、CART决策树、CHAID决策树与SVR建立了流通产业监测预警模型,同时针对在小样本问题中存在缺陷的前三类模型通过boosting算法提升其预测精度。经过比较,本文选择了训练和泛化能力均最为杰出的多项式核函数SVR模型对下一季度流通产业的景气程度进行了预测。结果表明,流通产业景气水平确实将加速下滑,估计值为98.8634,根据警情划分标准,它已处在跌破"正常"线的边缘。为此,本文提出了几项建议:合理化财富分配以促进消费;在工业化生产当中融入新技术以满足需求;创新商业模式、融合新旧市场,以挖掘潜在需求;投资交通基础设施,并促进物流仓储信息化发展。
[Abstract]:In the national economy, the circulation industry occupies a strategic position. It is connected with many industries in the economy. On the one hand, it brings demand information to the production sector and provides a way to transfer production results. On the other hand, it also supplies goods and services to the consumer sector. With the continuous change of economic model, many changes have taken place in the circulation industry. This paper analyzes the characteristics of its cycle fluctuation based on the boom index, and constructs an early warning model to estimate its development direction and amplitude. First of all, this paper confirms the basic index of circulation industry boom analysis, that is, the increase of value added of circulation industry, and from the overall development of circulation industry, the degree of industry self-support. Three aspects of macro-support to build the circulation industry in China's economic fluctuations monitoring and early warning index system. Secondly, aiming at the benchmark index, this paper decomposes it into seven components with different natural frequencies by EEMD time-frequency decomposition method, and thus preliminarily distinguishes the structure and characteristics of the benchmark cycle of China's circulation industry. The results show that the impact of short-term factors on the circulation industry is significant from the point of view of the whole study period. According to the characteristics of short-term fluctuations, The circulation industry can be divided into three stages: low amplitude high amplitude low frequency low amplitude high frequency and low amplitude high frequency. The negative effect of medium and long term factors is the main reason for the decline of circulation industry in the near future. Thirdly, after the data preprocessing, the cyclic factors of each index in the index system are obtained, and through the time difference correlation method, the peak and valley method and the system clustering method, we identify the first, consistent and lag between each index and the benchmark index cycle factor. Among them, the Chebyshev distance and the ward method are used to measure the distance between the indexes by the systematic clustering method. Using all the indexes to calculate the comprehensive climate index. Combined with the trend of the boom index and the decomposition of the EEMD signal, we can see that within the study period, China's circulation industry experienced five cycles, the longest of which was the first cycle, and the most violent was the second cycle. The latter three cycles are now in the adjustment period after the financial crisis. The volatility is relatively stable. According to the trend of the leading index, we can qualitatively predict that the circulation industry will accelerate its decline in the next quarter. And this is largely due to the decline in fixed asset investment growth in the circulation industry and the poor development of the cargo transport industry. Finally, the comprehensive prosperity index of circulation industry is divided into five different alert grades by "渭-蟽 method". Then the Synthetical climate index is taken as the output variable and the leading and consistent index after time translation is taken as the input variable. Based on the BP-ANNNCART decision tree which is suitable for the continuous variable regression, the chaid decision tree and SVR are used to establish the monitoring and warning model of circulation industry. At the same time, the boosting algorithm is used to improve the prediction accuracy of the first three kinds of models with defects in the small sample problem. Through comparison, this paper chooses the polynomial kernel function SVR model, which is the most outstanding training and generalization ability, to predict the prosperity of circulation industry in the next quarter. The results show that the prosperity level of circulation industry will really accelerate down, the estimated value is 98.8634. According to the criterion of police situation classification, it is on the verge of falling below the "normal" line. To that end, this paper makes several recommendations: rationalizing the distribution of wealth to promote consumption; incorporating new technologies into industrial production to meet demand; innovating business models and integrating new and old markets to tap potential needs; and investing in transport infrastructure. And promote logistics warehousing information development.
【学位授予单位】：浙江工商大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F724

【参考文献】