影响黑龙江省经济发展因素的分析.docx

《影响黑龙江省经济发展因素的分析.docx》由会员分享，可在线阅读，更多相关《影响黑龙江省经济发展因素的分析.docx（11页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、影响黑龙江省经济进展因素的经济学分析09区域经济学 许文华 2022651008摘要：作为“东北老工业基地”振兴的主力战场，黑龙江省全省的结构调整和经济进展有 了长足的进步，但是在进展中也存在一些问题和困难，需要我们仔细重视、讨论并加以改进， 本文就影响黑龙江省经济进展的因素进行了计量经济学分析，并提出了一些的建议。关键词：GDP地区生产总值；固定资产投资；社会消费品零售总额怎样才能够使黑龙江省经济又快又好进展？在经济建设过程中应留意哪些方面，下面我 选取了固定资产投资、社会消费品零售总额、进出口总额等影响经济进展的几个因素进行计 量经济学分析。表1资料来源：黑龙江省统计年鉴（2022）年份地

2、区生产 总值（亿 元）y全社会固 定资产投 资（亿元）X1财政支出（亿元）x2城乡储蓄 存款余额（亿元）x3社会消费 品零售总 额（亿元）x4进出口总额（亿元）x5实际外资使用 额（亿美元）x61990715.20162.9092.70308.80341.0071.391.17771991822.30189.60110.10388.80352.20107.430.64621992959.70244.20102.50475.90403.00158.871.051619931198.40328.60124.90582.90459.50190.102.996919941604.90142.40790

3、.50209.064.905419951991.40487.50174.611091.10682.70199.297.499419962370.50568.60208.901418.90782.20203.637.872519972667.50669.90233.601689.70880.20204.1810.353719982774.401906.70949.70166.458.700919992866.30785.90359.702119.201016.20181.4011.130920003151.40408.752285.501094.00247.2111.035920013390.1

4、0972.90513.422578.401198.90280.1811.511420023637.201055.70565.862915.701320.00360.1012.365620034057.401190.70606.203342.401376.40441.2012.877220044750.601464.70758.533585.701557.30562.0114.454620055511.501731.90861.354078.591760.00783.9515.220220066201.442235.901064.804373.601997.701025.1717.4901200

5、77065.002864.201325.604478.182331.101315.4821.690820228310.003656.001717.705545.082838.571590.4326.5642其中：Y代表地区生产总值，XI代表全社会固定资产投资，X2代表财政支出，X3代 表城乡储蓄存款余额，X4代表社会消费品零售总额，X5代表进出口总额，X6代表实际外 资使用额。Date: 06/27/10 Time: 11:12Sample: 2 19Included observations: 18 after adjustmentsVariableCoefficientStd. Error

6、t-StatisticProb.C0.2860470.0942853.0338600.0079LNX4-0.903*LNX4(-1)0.7848300.1205806.5088100.0000R-squared0.725861Mean dependent var0.896812Adjusted R-squared0.708728S.D. dependent var0.072192S.E. of regression0.038962Akaike info criterion-3.548042Sum squared resid0.024288Schwarz criterion-3.449112Lo

7、g likelihood33.93238Hannan-Quinn criter.-3.534401F-statistic42.36360Durbin-Watson stat1.358678Prob(F-statistic)0.000007重复得到:得到d/VOW = L358 1.391 = d,j自相关已经消退。方程还原为:lny； = 0.295 + 0.7971nxlny = 2.81 + 0.7971nx45、总结分析由模型可以看出，经过经济学及计量经济学的一系列检验及修正，最终得出黑龙江省经 济进展影响因素的模型中，模型修正后的最终回归方程为：lny = 2.81 + 0.7971n

8、x4回归参数0.754403表示：在其他条件不变的状况下，社会消费品零售总额每增加1亿元,GDP相应增加0.754403亿元。这说明，促进黑龙江省经济进展的主要缘由是社会消费品零售, 该模型提示我们在进展经济的过程中应大力促进居民消费。一、建立模型并回归依据计算国民生产总值的支出法可知：GDP=消费支出+投资支出+政府支出+净出口。结 合本文选取的相关数据，模型的函数形式可为：Y=P（）+P1 X.+P2 X2+P3 X3+p4 X4+p5 X5+p6 X6+u；运用OLS估量方法对式1中的参数进行估量，得回归分析结果（表2）：Dependent Variable: YMethod: Leas

9、t SquaresDate: 06/25/10 Time: 18:21Sample: 1990 2022Included observations: 19Substituted Coefficients:VariableCoefficientStd. Errort-StatisticProb.C-150.6556298.0553-0.5054620.6224X10.2851530.5212990.5470050.5944X2-2.5071660.695428-3.6052150.0036X30.4885850.2257772.1640130.0513X42.1695641.3655151.58

10、88250.1381X51.5800750.4036943.9140420.0021X615.6786533.047720.4744250.6437R-squared0.998931Mean dependent var3370.802Adjusted R-squared0.998396S.D. dependent var2164.602S.E. of regression86.68639Akaike info criterion12.03978Sum squared resid90174.37Schwarz criterion12.38773Log likelihood-107.3779Han

11、nan-Quinn criter.12.09867F-statistic1868.577Durbin-Watson stat2.010318Prob(F-statistic)0.000000二、多重共线性的检验及修正1、检验用简洁相关系数矩阵法进行检验，如下（表3）：X5X1X2X3X4X11X20.994451X30.941410.955151X40.984530.987690.984271X50.983320.977080.891280.9472611X60.952630.948860.971260.9824810.894583X6由上表可以看出，该模型存在多重共线性。2、用逐步回归法修正

12、多重共线性（1）用OLS方法逐一对Y各个解释变量回归。相关T值及R2值如下（表4）：X1X2X3X4X5X6t-Statistic20.2372820.9449724.2989360.6139911.9661521.21873R-squared0.9601450.9626940.9720220.9953940.8938750.963616结合经济意义选出拟合效果最好的变量，即X4社会消费品零售总额。可得出相应的一元线性回归方程：Y = -196.850368148 + 3.09630236447X4相关检验结果如下（表5）：Dependent Variable: YMethod: Least

13、SquaresDate: 06/25/10 Time: 20:43Sample: 1990 2022Included observations: 19VariableCoefficientStd. Errort-StatisticProb.C-196.850468.31501-2.8815100.0104X43.0963020.05108260.613990.0000R-squared0.995394Mean dependent var3370.802Adjusted R-squared0.995123S.D. dependent var2164.602S.E. of regression15

14、1.1608Akaike info criterion12.97387Sum squared resid388442.8Schwarz criterion13.07328Log likelihood-121.2517Hannan-Quinn criter.12.99069F-statistic3674.056Durbin-Watson stat0.805977Prob(F-statistic)0.000000（2）逐步回归。将其余解释变量逐一代入上式结合经济意义和统计检验，相关检验结果如下（表6）：Dependent Variable: Y Method: Least SquaresDate:

15、 06/25/10 Time: 20:35Sample: 1990 2022Included observations: 19VariableCoefficientStd. Errort-StatisticProb.C-277.1852119.5828-2.3179350.0340X43.3346620.29439411.327210.0000X1-0.1785650.217132-0.8223800.4229R-squaredAdjusted R-squaredR-squaredAdjusted R-squared3370.8022164.6020.995581 Mean dependent

16、 var0.995029 S.D. dependent var同理可对X2、X3、X5、X6进行回归(表7)X1X2X3X5X6t-Statistic-0.8238-1.7439861.3800490.0685740.452973R-squared0.9955810.996130.9958840.9953960.995153S.E. of regression Sum squared resid Log likelihood F-statistic152.6207 Akaike info criterion372689.5-120.85841802.38813.03773Prob(F-stat

17、istic)0.000000Schwarz criterionHannan-Quinn criter.Durbin-Watson stat13.1868513.062970.807045结合经济意义和统计检验，剔除其它全部解释变量，仅保存X4,所以，最终的回归议 程为：Y =-196.850368148 + 3.09630236447X4t-Statistic = 60.61399 R-squared=0.995394三、异方差检验及修正1、异方差检验(1)图示法检验首先绘制散点图，如下90,000-80,000-70,000-60,000-E250,000-40,000-30,000-20,

18、000-10,000-0-0500 1,000 1,500 2,000 2,500 3,000X4由图可以看出，残差平方E2对解释变量X4的散点图主要分布在图形中的下三角局部, 大致可以看到，残差平方E2随X4的变动呈增大的趋势，因此可能存在异方差，待进一步 检验。(2) Goldfeld-Quanadt 检验由于该模型是小样本，不适用White检。用Goldfeld-quandt检验。以X4得升序排列，去 掉中间1/4的数据，19个数据大约去掉5个数据。剩下的分为前后两局部，每局部各有(n-5)/2=7个数据。分别对这两局部进行回归，纪录其残差平方和。如下(表8)：Dependent Var

19、iable: Y Method: Least SquaresDate: 06/25/10 Time: 22:48Sample: 1 7Included observations: 7VariableCoefficientStd. Errort-StatisticProb.C-503.584854.06913-9.3137210.0002X43.6915970.10125136.459780.0000R-squared0.996253Mean dependent var1380.343Adjusted R-squared0.995503S.D. dependent var628.2381S.E.

20、 of regression42.12798Akaike info criterion10.55426Sum squared resid8873.832Schwarz criterion10.53880Log likelihood-34.93990Hannan-Quinn criter.10.36325F-statistic1329.316Durbin-Watson stat2.443749Prob(F-statistic)0.000000(表9)Dependent Variable: YMethod: Least SquaresDate: 06/25/10 Time: 22:50Sample

21、: 13 19Included observations: 7VariableCoefficientStd. Errort-StatisticProb.C-25.68395329.3928-0.0779740.9409X43.0128760.16884217.844370.0000R-squared0.984540Mean dependent var5647.591Adjusted R-squared0.981448S.D. dependent var1673.206S.E. of regression227.8980Akaike info criterion13.93063Sum squar

22、ed resid259687.5Schwarz criterion13.91518Log likelihood-46.75720Hannan-Quinn criter.13.73962F-statistic318.4216Durbin-Watson stat0.802204Prob(F-statistic)0.000010基于以上两图，样本范围在1990-1996的残差平方和为El=8873.832o样本范围在 2002-2022的残差平方和为E2=259687.5。依据Goldfeld-quandt检验，构造F统计量 F*=E2/E1 =29.2644在给定的显著性水平&=0.05下查F表的

23、临界值F(5,5)=5.05。由于F*F,表 明误差项存在异方差。2、异方差修正（加权最小二乘法）设权数w=l/屋2,回归得:（表10）Dependent Variable: YMethod: Least SquaresDate: 06/25/10 Time: 23:14Sample: 1 19Included observations: 19Weighting series: WProb（F-statistic）0.000000VariableCoefficientStd. Errort-StatisticProb.C-397.661840.12022-9.9117750.0000X43.3

24、954790.09261736.661680.0000Weighted StatisticsR-squared0.987510Mean dependent var1525.075Adjusted R-squared0.986775S.D. dependent var810.4947S.E. of regression89.79404Akaike info criterion11.93221Sum squared resid137070.5Schwarz criterion12.03163Log likelihood-111.3560Hannan-Quinn criter.11.94904F-s

25、tatistic1344.079Durbin-Watson stat1.028456Unweighted StatisticsR-squared0.981436Mean dependent var3370.802Adjusted R-squared0.980344S.D. dependent var2164.602S.E. of regression303.4802Sum squared resid1565704.Durbin-Watson stat0.304833据上图，估量结果如下Y = -397.661834278 + 3.3954789304X4四、自相关检验1、DW检验对yx4去对数

26、，令Iny = logy,lnx4 = logx4,作一元回归，结构如下：（表11）Dependent Variable: LNYMethod: Least SquaresDate: 06/27/10 Time: 10:37Sample: 1 19Included observations: 19VariableCoefficient Std. Error t-Statistic Prob.R-squared0.990788Mean dependent var7.903101Adjusted R-squared0.990246S.D. dependent var0.720028S.E. of

27、regression0.071112Akaike info criterion-2.349829Sum squared resid0.085967Schwarz criterion-2.250415Log likelihood24.32338Hannan-Quinn criter.-2.333004F-statistic1828.349Durbin-Watson stat0.485916Prob(F-statistic)0.000000cLNX4cLNX40.1443771.1294970.1821810.0264150.79249042.759720.43900.0000对y,x4去对数，令

28、Iny = logy,lnx4 = logx4,作一元回归，结构如下：(表12)Dependent Variable: LNYMethod: Least SquaresDate: 06/27/10 Time: 10:37Sample: 1 19Included observations: 19VariableCoefficientStd. Errort-StatisticProb.C0.1443770.1821810.7924900.4390LNX41.1294970.02641542.759720.0000R-squared0.990788Mean dependent var7.903101

29、Adjusted R-squared0.990246S.D. dependent var0.720028S.E. of regression0.071112Akaike info criterion-2.349829Sum squared resid0.085967Schwarz criterion-2.250415Log likelihood24.32338Hannan-Quinn criter.-2.333004F-statistic1828.349Durbin-Watson stat0.485916Prob(F-statistic)0.000000残差图为:由上面结果DW = 0.486

30、 1.18 = d/和图知道：必 存在正自相关。2、自相关修正：科克伦一奥特克迭代法一元线性回归模型lnK = 1+21n%4+4,假定,为一阶自回归形式：N = PN首先对进行最小二乘法估量：得出残差：匕，再做如形式，得到的结果如下：（表13）Dependent Variable: E11Method: Least SquaresDate: 06/27/10 Time: 10:50Sample: 1 19Included observations: 19VariableCoefficientStd. Errort-StatisticProb.E11（-1）0.6246260.1550654.

31、0281550.0009R-squared0.476251Mean dependent var0.008828Adjusted R-squared0.476251S.D. dependent var0.059067S.E. of regression0.042747Akaike info criterion-3.413075Sum squared resid0.031064Schwarz criterion-3.363630Log likelihood31.71768Hannan-Quinn criter.-3.406255Durbin-Watson stat1.108973得到夕= 0.62

32、4,再作广义差分方程得到：In Yt-Pln Kt = a - 夕）+02（ln- -In %：_1）+M - A-i 将夕=0.624代入：做出估量：（表14）Dependent Variable: LNY-0.624*LNY（-1）Method: Least SquaresDate: 06/27/10 Time: 10:55Sample: 1 19Included observations: 19VariableCoefficientStd. Errort-StatisticProb.C0.3791540.1435542.6411980.0178LNX4-0.624*LNX41.03868

33、30.05492418.911280.0000R-squaredAdjusted R-squaredS.E. of regression3.0843850.239347-3.0034740.957178 Mean dependent var0.954501 S.D. dependent var0.051054 Akaike info criterionSum squared resid0.041704Schwarz criterion-2.908504Log likelihood29.06691Hannan-Quinn criter.-2.993793F-statistic357.6365Du

34、rbin-Watson stat0.681163Prob(F-statistic)0.0000001.0386.OW = 0.68L18 = d/，再依据图形知道：自相关没有消退：作其次轮迭代由上面结果知道：In y； =379 +LO3861nX；计算出= LOO&p?=对e/ = lnyATnx的残差作回归，结果如下：(表Dependent Variable: E1Method: Least SquaresDate: 06/27/10 Time: 11:08Sample: 1 19Included observations: 18 after adjustmentsVariableCoef

35、ficientStd. Errort-StatisticProb.E1(-1)0.9029450.04035622.374360.0000R-squared0.583758Mean dependent var-0.225167Adjusted R-squared0.583758S.D. dependent var0.067813S.E. of regression0.043751Akaike info criterion-3.366660Sum squared resid0.032540Schwarz criterion-3.317195Log likelihood31.29994Hannan-Quinn criter.-3.359839Durbin-Watson stat1.270530夕= 0.903,重复上面步骤、得到如下结果：（表16）Dependent Variable: LNY-0.903*LNY(-1)Method: Least Squares