中國(guó)食品價(jià)格指數(shù)的影響因素分析

1、關(guān)鍵詞：食品價(jià)格指數(shù) 多因素分析 預(yù)測(cè)模型 模型檢測(cè)與修正二、模型設(shè)定在本文中，我們選取糧食價(jià)格指數(shù)、肉禽及制品價(jià)格指數(shù)、水產(chǎn)品價(jià)格指數(shù)、蔬菜價(jià)格指數(shù)作為解釋變量，選取食品價(jià)格指數(shù)作為被解釋變量，構(gòu)建多元線性回歸模型：y=0+1x1 +2x2 +3x3 +4x4 +i其中：y 食品價(jià)格指數(shù) x1 糧食價(jià)格指數(shù) x2 肉禽價(jià)格指數(shù) x3 水產(chǎn)品價(jià)格指數(shù) x4 蔬菜價(jià)格指數(shù)三、模型的估計(jì)與調(diào)整通過(guò)使用eviews計(jì)量經(jīng)濟(jì)學(xué)分析軟件，得到了一下回歸分析結(jié)果dependent variable: ymethod: least squaresdate: 05/03/14 time: 19:50sampl

2、e: 2014:01 2014:04included observations: 27variablecoefficientstd. errort-statisticprob. c7.2991204.8192871.5145640.1441x10.4531110.0604837.4915000.0000x20.2255630.02100210.740120.0000x30.1764920.0642352.7475760.0118x40.0593710.0123924.7909120.0001r-squared0.990031 mean dependent var108.2515adjusted

3、 r-squared0.988219 s.d. dependent var4.152074s.e. of regression0.450673 akaike info criterion1.409427sum squared resid4.468336 schwarz criterion1.649396log likelihood-14.02726 f-statistic546.2222durbin-watson stat0.901780 prob(f-statistic)0.0000001.多重共線性檢驗(yàn)。（1） 直觀的來(lái)看，x1、x3的相關(guān)系數(shù)達(dá)到了0.80，x2、x3的相關(guān)系數(shù)達(dá)到了0.

4、88。所以可以認(rèn)為存在較嚴(yán)重的多重共線性。（2） 修正多重共線性現(xiàn)剔除x3進(jìn)行回歸，結(jié)果如下：dependent variable: ymethod: least squaresdate: 05/03/14 time: 21:40sample: 2014:01 2014:04included observations: 27variablecoefficientstd. errort-statisticprob. c5.2102285.3941020.9659120.3441x10.5787620.04486712.899600.0000x20.2749320.01232422.308120.

5、0000x40.0758200.0122986.1650940.0000r-squared0.986610 mean dependent var108.2515adjusted r-squared0.984864 s.d. dependent var4.152074s.e. of regression0.510823 akaike info criterion1.630366sum squared resid6.001621 schwarz criterion1.822342log likelihood-18.00994 f-statistic564.9205durbin-watson sta

6、t0.921999 prob(f-statistic)0.000000由上圖可看出，剔除x3后，擬合優(yōu)度非常好，且顯著性明顯。再剔除x1進(jìn)行回歸，結(jié)果入下：dependent variable: ymethod: least squaresdate: 05/03/14 time: 21:43sample: 2014:01 2014:04included observations: 27variablecoefficientstd. errort-statisticprob. c32.394936.3853025.0733580.0000x20.1426790.0329004.3367320.0

7、002x30.5403430.0774786.9741530.0000x40.0144350.0199850.7222650.4774r-squared0.964601 mean dependent var108.2515adjusted r-squared0.959983 s.d. dependent var4.152074s.e. of regression0.830589 akaike info criterion2.602589sum squared resid15.86718 schwarz criterion2.794565log likelihood-31.13496 f-sta

8、tistic208.9094durbin-watson stat1.044482 prob(f-statistic)0.000000由上圖可以看出，剔除x1后，導(dǎo)致x4通不過(guò)t檢驗(yàn)。剔除x2進(jìn)行回歸，結(jié)果如下：dependent variable: ymethod: least squaresdate: 05/03/14 time: 21:41sample: 2014:01 2014:04included observations: 27variablecoefficientstd. errort-statisticprob. c16.3409511.595881.4092020.1722x1

9、0.1109050.1256320.8827720.3865x30.7667330.0812689.4346090.0000x4-0.0321650.021984-1.4630590.1570r-squared0.937763 mean dependent var108.2515adjusted r-squared0.929645 s.d. dependent var4.152074s.e. of regression1.101317 akaike info criterion3.166844sum squared resid27.89668 schwarz criterion3.358820

10、log likelihood-38.75239 f-statistic115.5183durbin-watson stat1.495176 prob(f-statistic)0.000000由上圖可知，剔除x2后，導(dǎo)致x1，x4都通不過(guò)t檢驗(yàn)，且可決系數(shù)大幅降低。剔除x4進(jìn)行回歸，結(jié)果入下：dependent variable: ymethod: least squaresdate: 05/03/14 time: 21:44sample: 2014:01 2014:04included observations: 27variablecoefficientstd. errort-statist

11、icprob. c21.060075.4100843.8927440.0007x10.3128540.0739924.2282150.0003x20.1563630.0213157.3358800.0000x30.3251700.0786274.1355880.0004r-squared0.979631 mean dependent var108.2515adjusted r-squared0.976974 s.d. dependent var4.152074s.e. of regression0.630052 akaike info criterion2.049924sum squared

12、resid9.130200 schwarz criterion2.241900log likelihood-23.67397 f-statistic368.7162durbin-watson stat2.010366 prob(f-statistic)0.000000 由上圖可看出，x4的存在不影響本文的分析結(jié)果，沒(méi)必要剔除。所以綜上所述，剔除x3，得到一下回歸分析結(jié)果：dependent variable: ymethod: least squaresdate: 05/31/12 time: 21:40sample: 2014:01 2014:04included observations:

13、 27variablecoefficientstd. errort-statisticprob. c5.2102285.3941020.9659120.3441x10.5787620.04486712.899600.0000x20.2749320.01232422.308120.0000x40.0758200.0122986.1650940.0000r-squared0.986610 mean dependent var108.2515adjusted r-squared0.984864 s.d. dependent var4.152074s.e. of regression0.510823

14、akaike info criterion1.630366sum squared resid6.001621 schwarz criterion1.822342log likelihood-18.00994 f-statistic564.9205durbin-watson stat0.921999 prob(f-statistic)0.000000得到的回歸方程為=5.210228+0.578762x1 +0.274932x2 +0.07582x4 (0.965912) (12.8996) (22.3081) (6.165094) r2=0.9866 adjusted r-squared =0

15、.9849 f=564.9205從回歸的結(jié)果可以得到r2=0.9866，修正的可決系數(shù)為0.9849，這說(shuō)明模型對(duì)樣本的擬合度非常好。 2.相關(guān)性檢驗(yàn)從估計(jì)的結(jié)果可以看出，模型擬合較好，可決系數(shù)r²=0.9866，修正的可決系數(shù)為0.9849，表明模型在整體上擬合比較好。3.顯著性檢驗(yàn)（1）對(duì)于ß1，t統(tǒng)計(jì)量為12.8996。給定=0.05，查t分布表，在自由度為n-4=23下，得臨界值t0.025(23)=2.069，因?yàn)閠>t0.025(23)，所以拒絕原假設(shè)h0: ß1=0，表明糧食價(jià)格指數(shù)對(duì)食品價(jià)格指數(shù)有顯著性影響；（2）對(duì)于ß2，t統(tǒng)計(jì)量

16、為22.3081。給定=0.05，查t分布表，在自由度為n-4=23下，得臨界值t0.025(23)= 2.069，因?yàn)閠>t0.025(23)，所以拒絕原假設(shè)h0: ß2=0，表明肉禽價(jià)格指數(shù)對(duì)食品價(jià)格指數(shù)有顯著性影響。（3）對(duì)于ß4，t統(tǒng)計(jì)量為6.165094。給定=0.05，查t分布表，在自由度為n-4=23下，得臨界值t0.025(23)= 2.069，因?yàn)閠>t0.025(23)，所以拒絕原假設(shè)h0: ß4=0，表明蔬菜價(jià)格指數(shù)對(duì)食品價(jià)格指數(shù)有顯著性影響。 （4）對(duì)于f=564.9205>f(3,23)=3.03(顯著性水平為0.05)

17、，表明模型從整體上看食品價(jià)格指數(shù)與各解釋變量之間線性關(guān)系顯著。 4序列相關(guān)檢驗(yàn)（1） 由圖可知，存在一階自相關(guān)。（2） 修正：用科克倫-奧克特迭代方程法對(duì)模型進(jìn)行修正，得到如下結(jié)果：dependent variable: ymethod: least squaresdate: 05/03/14 time: 13:10sample(adjusted): 2014:01 2014:04included observations: 26 after adjusting endpointsconvergence achieved after 10 iterationsvariablecoefficie

18、ntstd. errort-statisticprob. c5.1691879.8464600.5249790.6051x10.5631620.0877866.4151880.0000x20.2821290.01771215.928550.0000x40.0845860.0103308.1882700.0000ar(1)0.5840820.1869773.1238250.0051r-squared0.989380 mean dependent var108.5496adjusted r-squared0.987357 s.d. dependent var3.928543s.e. of regr

19、ession0.441736 akaike info criterion1.374831sum squared resid4.097737 schwarz criterion1.616772log likelihood-12.87280 f-statistic489.0816durbin-watson stat1.380596 prob(f-statistic)0.000000inverted ar roots .58得到一階自相關(guān)系數(shù)估計(jì)為0.584082再次檢驗(yàn)是否存在自相關(guān)：由上圖可知，修正后不再存在自相關(guān)。綜上，本研究模型估計(jì)的最終結(jié)果為得到的回歸方程為=5.1691+0.5631x1

20、 +0.2821x2 +0.0846x4 (0.5250) (6.4152) (15.9286) (8.1883) r2=0.9894 adjusted r-squared =0.9874 f=489.0816經(jīng)濟(jì)意義檢驗(yàn)：從經(jīng)濟(jì)意義上來(lái)看,該模型說(shuō)明了在假定其他變量不變的情況下，糧食價(jià)格指數(shù)每上升1%，食品價(jià)格指數(shù)上漲0.5631%；肉禽及制品價(jià)格指數(shù)每上升1%，食品價(jià)格指數(shù)上漲0.2821%；蔬菜價(jià)格指數(shù)每上升1%，食品價(jià)格指數(shù)上漲0.0846%。由于各變量都通過(guò)了檢驗(yàn)，所以說(shuō)明各變量對(duì)被解釋變量都起到了很好的作用。5因果關(guān)系檢驗(yàn)對(duì)x1，y進(jìn)行因果關(guān)系檢驗(yàn)，結(jié)果如下：pairwise gr

21、anger causality testsdate: 05/03/14 time: 21:54sample: 2014:01 2014:04lags: 2 null hypothesis:obsf-statisticprobability y does not granger cause x125 1.09403 0.35409 x1 does not granger cause y 2.86637 0.08043由上圖可知，選定顯著性水平（如10%），0.35>0.1，則在該顯著性水平下，接受原假設(shè)，表示食品價(jià)格指數(shù)對(duì)糧食價(jià)格指數(shù)沒(méi)有影響；0.08<0.1，拒絕原假設(shè)，表示糧食價(jià)格指數(shù)對(duì)食品價(jià)格指數(shù)有顯著性的影響。對(duì)x2，y進(jìn)行因果關(guān)系檢驗(yàn)，結(jié)果如下：pairwise granger causality testsdate: 05/03/14 time: 20:32sample: 2014:01 2014:04lags: 1 null hypothesis:obsf-statisticprobability y does not granger cause x226 0.00457 0