【作者】 高宇明；

【导师】 齐中英；

【作者基本信息】 哈尔滨工业大学 ， 技术经济及管理， 2008， 博士

【摘要】 经济增长问题一直是经济领域内一个研究的热点问题,经济增长的影响因素中,资本和劳动投入因素是最重要的两个因素。在中国增长方式由“粗放式增长”向“集约式增长”转变的过程中,研究资本和劳动生产率以及两种要素生产率加权和组成的全要素生产率对经济增长的作用,分析中国经济增长的根源,具有更现实的意义。本文的研究是从资本和劳动的单要素生产率与经济增长关系的分析开始的。针对传统方法研究变量之间相互影响关系只能得到变量间样本区间内的平均影响关系,而不能得到变量间逐年变化的影响关系,本文采用了一种新的方法――卡尔曼滤波算法,通过建立状态空间模型研究了资本和劳动的单要素生产率和经济增长之间的时变参数关系。通过时变参数协整关系检验,证明了估计结果的可靠性,同时对时变参数估计结果进行了分析。利用带虚拟变量的最小二乘法和岭估计方法估计了中国传统总量生产函数,指出了传统总量生产函数的不足之处。针对传统总量生产函数的不足,建立了时变总量生产函数模型,应用两种时变参数方法:卡尔曼滤波算法和推广梯度递推算法估计了中国的时变总量生产函数,同时计算了两种总量函数体系下的全要素生产率增长率。对推广梯度递推算法估算时变总量生产函数的初值确定进行了详细讨论。对两种时变参数方法估算时变总量生产函数的残差进行了比较。分别用传统方法和两种时变参数方法在索洛增长速度方程框架下估计了全要素生产率变化率,指出传统方法估算全要素生产率增长的不足之处。比较了时变参数方法和传统方法估算全要素生产率中的假设和经济意义中的不同之处。对推广梯度递推算法估算索洛增长速度方程的初值确定进行了详细讨论。对两种时变参数方法估算索洛增长速度方程的残差进行了比较。对用六种方法估算得到的中国全要素生产率增长率1952～2005结果进行了比较。对于六种方法得到的全要素生产率增长率结果的趋势大体相似,但是传统方法得到的结果波动较大,时变参数方法波动较小这一情况给出了解释。从全要素生产率的六种结果中,选用时变参数方法中推广梯度递推算法得到的全要素生产率增长率估算结果作为中国宏观经济分析的基础数据,对于全要素生产率增长和经济增长关系进行了分析。采用参数和非参数两种方法对全要素生产率增长率进行量化分解,把全要素生产率分解为技术进步和技术效率两部分。计算技术效率和技术进步对全要素生产率增长的贡献,分析全要素生产率增长的主要原因。更多还原

【Abstract】 In the economic area, the economic growth issue has always been the hot issue. Capital and labor invest factor are the most important factors among the factor which affect economic growth. In China, the process of growth mode changes from“extensive growth”to“intensive growth”. It has more practical significance that study on capital and labor productivity, as well as the tatal factor productivity which is composed of capital and labor productivity weighted , and analysis the root of China economic growth..This paper starts from the relationship analysis between the single factor productivity as Capital and labor and economic growth. If we adopt traditional method to study on the mutual influence between the variations, we could only acquire the average influence relations in the variation sample, and we can not get the variation changes year to year. So the article adopt a new method, Kalman filtering algorithm.,to research the Time-varying Parameter relationship between capital and labor’s single Factor Productivity with the economic growth by establish the states space model. By the time-varying parameters cointegration test, the estimate result reliability is proven. simultaneously to the time-varying parameters estimated the result has carried on the analysis,and then the time-varying parameter estimation results are analyzed.The paper estimates the traditional aggregate production function by application of Least Square method with dmmy variable and Ridge Regression, and points out its deficiencies. According to these deficiencies, the paper established time-varying aggregate production function model,and esitimated time-varying aggregate production function in china by adopting two time-varying parameters methods: Kalman filtering algorithm and Multilayer Hierarchical Method. At the same time, the total factor productivity growth rate is estimated in the two kinds of system of aggregate production function.The discussion of initial value of Multilayer Hierarchical Method to estimate aggregate production function was carried out.The residual of two time-varying parameters methods estimating aggregate production function was compared.The paper estimated the growth rate of the total factor productivity by adopting the traditional method and the two time-varying parameters methods in the framework of Solow Growth Accounting Equation.,and pointed out the deficiencies of the traditional method.The diffrence of assumption and economic sense between time-varying parameters method and traditional method was perceived. The discussion of initial value of Multilayer Hierarchical Method to estimate Solow Growth Accounting Equation.was carried out. The residual of two time-varying parameters methods estimating Solow Growth Accounting Equation was compared.Compared the total factor productivity growth results in china from 1952 to 2005, the paper found that the result and general trend of the six methods are similar,and the result of traditional methods fluctuates over a large area but time-varying methods to be smaller,and explanations have been given. From the six results of TFP, estimation by the Multilayer Hierarchical Method is taken as fundamental data of the macroeconomic economic analysis in china.By these dates,the paper analyses the relationship between the input factor and economic growth,and the contribution of TFP to economic growth.The paper carried on the comparison of TFP by parametric and nonparametric methods,which indicated that TFP is composed of technical progress and technical efficiency.Then the prime cause of TFP growth is analyzed by calculation of the contribution made by the technology progresses and technical efficiency to TFP.更多还原