影響我國農(nóng)業(yè)總產(chǎn)值因素的實(shí)證分析資料

1、計量經(jīng)濟(jì)學(xué)課程論文影響我國農(nóng)業(yè)總產(chǎn)值 因素的實(shí)證分析小組成員：（保險學(xué)院 02 級）組長：侯 君 男 40205117組員：張 翠 青 40205107 石 小 航 40205074李 進(jìn) 40205129 陳 永 琴 40205076指導(dǎo)教師：任 棟日期： 2005年 4月 5 月 內(nèi)容摘要 ： 解決“三農(nóng)”問題是當(dāng)前我國完善社會主義市場經(jīng)濟(jì)制度的重頭戲，而實(shí)現(xiàn)農(nóng)業(yè)的繁榮興 旺又是其中的重中之重。改革開放以來，我國農(nóng)業(yè)發(fā)展取得的喜人的成績，但是制約因素也很 多，這使得入世后如何保護(hù)我國農(nóng)業(yè)的利益，保持農(nóng)業(yè)穩(wěn)定成為難題。深入了解農(nóng)業(yè)發(fā)展?fàn)顩r， 有足于認(rèn)清和解決問題。本文是根據(jù)我國農(nóng)業(yè)的現(xiàn)狀，想

2、從計量經(jīng)濟(jì)學(xué)的角度來驗(yàn)證一下是否 存在政府對農(nóng)業(yè)投入不足、農(nóng)業(yè)的現(xiàn)代化 程度（以農(nóng)業(yè)的機(jī)械化為衡量指標(biāo)） ，以及農(nóng)村中存在大量 的剩余勞動力。根據(jù)經(jīng)濟(jì)學(xué)原理，在模型中我們引入了五個變量：農(nóng)村居民家庭平均每戶生產(chǎn) 型固定投資，化肥施用量，農(nóng)業(yè)機(jī)械總動力，政府財政用于農(nóng)業(yè)的支出以及農(nóng)業(yè)從業(yè)人員。利 用EVIEW歎件對計量模型進(jìn)行了參數(shù)估計和檢驗(yàn)， 多重共線性的檢驗(yàn)，異方差的檢驗(yàn)和自相關(guān) 的檢驗(yàn)并加以修正。從我們所做的回歸結(jié)果看，我國農(nóng)村中確實(shí)存在政府對農(nóng)業(yè)投入不足、農(nóng)村中存在大量的 剩余勞動力，我國的農(nóng)業(yè)機(jī)械化程度是較低的，對我國的農(nóng)業(yè)增加值的貢獻(xiàn)十分低下等問題。 我們根據(jù)模型的回歸結(jié)果作了經(jīng)濟(jì)意

3、義的分析，并相應(yīng)提出一些政策建議。但是，鑒于水平有限，文中難免出現(xiàn)一些錯誤。另外還存在一些我們難以解決的問題，請 老師同學(xué)們多多包涵！ 關(guān)鍵詞 ：農(nóng)業(yè)總 產(chǎn)值 國家 財政對農(nóng)業(yè)的 基 礎(chǔ)性建 設(shè)投資 農(nóng)業(yè) 從業(yè)人 員人 數(shù) 農(nóng)村居民家庭平均每戶生產(chǎn)型固定投資 化肥施用量 農(nóng)業(yè)機(jī)械總動力一、 導(dǎo)論我國農(nóng)業(yè)的重要性 我國是農(nóng)業(yè)大國，農(nóng)業(yè)的發(fā)展程度直接制約著我國的第二、第三產(chǎn)業(yè)的發(fā)展，是工業(yè)品市 場；農(nóng)業(yè)的發(fā)展能為國民經(jīng)濟(jì)其他部門發(fā)展提供勞動力陣地。農(nóng)產(chǎn)品是輕工業(yè)的重要原料、重 要的出口商品。目前，我國 70%人口在農(nóng)村，農(nóng)業(yè)生產(chǎn)的發(fā)展直接關(guān)系廣大農(nóng)民生活的提高， 直接關(guān)系到國家經(jīng)濟(jì)建設(shè)目標(biāo)的實(shí)現(xiàn)。

4、農(nóng)產(chǎn)品在城鄉(xiāng)是人民的生活必需品，所以又直接關(guān)系到 城鄉(xiāng)人民生活的提高，物價穩(wěn)定，社會安定。我國農(nóng)業(yè)生產(chǎn)相對落后，已成為國民經(jīng)濟(jì)最薄弱 的環(huán)節(jié)，它已很難支撐國民經(jīng)濟(jì)其他部門的快速發(fā)展。因而，農(nóng)業(yè)生產(chǎn)的發(fā)展是我國人民生活 水平提高、現(xiàn)代化建設(shè)、社會穩(wěn)定的基礎(chǔ)，并最終決定著國民經(jīng)濟(jì)其他各部門的發(fā)展規(guī)模和速 度，是能否實(shí)現(xiàn)現(xiàn)代化戰(zhàn)略目標(biāo)的關(guān)鍵。二，模型的設(shè)定為了在更高層次上發(fā)展我國的經(jīng)濟(jì)，真正實(shí)現(xiàn)全民共同富裕的偉大目標(biāo)，保證糧食安全, 關(guān)注農(nóng)業(yè)總產(chǎn)值是必要的。而影響到農(nóng)業(yè)總產(chǎn)值的因素是多方面的。因此，我們提取了國家財 政對農(nóng)業(yè)的基礎(chǔ)性建設(shè)投資，農(nóng)業(yè)從業(yè)人員人數(shù)，農(nóng)村居民家庭平均每戶生產(chǎn)型固定投資，化肥

5、施用量，農(nóng)業(yè)機(jī)械總動力這五個對農(nóng)業(yè)總產(chǎn)值有較大影響的因素的時間序列數(shù)據(jù)來進(jìn)行分析， 希望通過建立一個合適的經(jīng)濟(jì)模型來從理論上找出影響農(nóng)業(yè)總產(chǎn)值的因素，從而提出增加農(nóng)業(yè) 總產(chǎn)值的方法。在此，我們將“農(nóng)業(yè)產(chǎn)總值”設(shè)為因變量，“農(nóng)村居民家庭平均每戶生產(chǎn)型固定投資”， “化肥施用量”，農(nóng)業(yè)機(jī)械總動力”，政府財政用于農(nóng)業(yè)的支出”，及“第一產(chǎn)業(yè)從業(yè)人員人數(shù)” 設(shè)為自變量，設(shè)定了以下經(jīng)濟(jì)學(xué)模型：Y =C+ 1 X2+ 2X3+ 3 X4 + 4 X5+ 5 X6+UY=農(nóng)業(yè)總產(chǎn)值（億元）X2 =農(nóng)村居民家庭平均每戶生產(chǎn)型固定投資（兀）X3=化肥施用量（萬噸）X4=農(nóng)業(yè)機(jī)械總動力（萬千瓦）X5=政府財政用于農(nóng)

6、業(yè)的支出（億元）X6=農(nóng)業(yè)從業(yè)人員（萬人）數(shù)據(jù)如下：obsYX2X3X4X5X619896534.7301126.0702357.10028067.00265.940032440.5019907662.0901258.0602590.30028707.70307.840033336.4019918157.0301401.0102805.10029388.60347.570034186.3019929084.7101643.9502930.20030308.40376.020034037.00199310995.531950.3103151.90031816.60440.450033258.20

7、199415750.472347.6303317.90033802.50532.980032690.30199520340.862774.2703593.70036118.10567.220032334.50199622353.703605.0703827.90038546.90700.430032260.40199723788.403896.5603980.70042015.60766.390032434.90199824541.903970.8104083.70045207.701154.76032626.40199924519.104045.4804124.30048996.101085

8、.76032911.80200024915.804676.9804146.40052573.601231.54032797.50200126179.604883.8004253.80055172.101456.73032451.00200227390.805221.3304339.40057929.901580.76031990.60200329691.805586.3404411.60060386.501754.45031259.60資料來源:，中國統(tǒng)計年鑒2004，中國統(tǒng)計年鑒1998三，參數(shù)估計模型為：丫 =C+ -1 X2 + -2X3+ -3 X4+4 X5+5 X6+UY=農(nóng)業(yè)總產(chǎn)

9、值（億元）X3=化肥施用量（萬噸）X2=農(nóng)村居民家庭平均每戶生產(chǎn)型固定投資（元）X5二政府財政用于農(nóng)業(yè)的支出（億元）X6=農(nóng)業(yè)從業(yè)人員（萬人）X4=農(nóng)業(yè)機(jī)械總動力（萬千瓦）用Eviews估計結(jié)果為:Depe ndent Variable: 丫Method: Least SquaresDate: 04/30/05 Time: 12:51Sample: 1989 2003In eluded observati ons: 15VariableCoefficie ntStd. Errort-StatisticProb.C40016.9715190.732.6343030.0272X20.5509391

10、.9208270.2868240.7807X310.087662.2054494.5739720.0013X40.0665550.2303620.2889160.7792X5-2.6453473.625498-0.7296510.4842X6-1.8264690.535716-3.4093960.0078R-squared0.992862Mean depe ndent var18793.77Adjusted R-squared0.988896S.D.dependent var8203.735S.E. of regressi on864.4572Akaike info criteri on16.

11、65126Sum squared resid6725576.Schwarz criteri on16.93448Log likelihood-118.8844F-statistic250.3705Durbi n- Watson stat1.561839Prob(F-statistic)0.000000= =丫 =40016.97+0.55093 9X2+10.08766 X3+0.066555 X4+ (-2.645347)X5+ (-1.826469) X6T = (2.634303)(0.286824)(4.573972)(0.288916)(-0.729651)(-3.409396)R2

12、=0.992862四，檢驗(yàn)及修正1. 經(jīng)濟(jì)意義檢驗(yàn)從上表中可以看出，X5符號為負(fù)，應(yīng)剔出。而 X6雖然在理論上說不通，但卻符合中國現(xiàn) 實(shí)的國情，應(yīng)保留，其意義將在第四部分加以闡述。而其他因素不與經(jīng)濟(jì)原理向悖，說明具有 經(jīng)濟(jì)意義。2. 統(tǒng)計推斷檢驗(yàn)從回歸結(jié)果可以看出，模型的擬和優(yōu)度非常好（ R2 =0.992862 ），F(xiàn)統(tǒng)計量的值在給定顯 著性水平a =0.05的情況下也較顯著，但是 X2、X4的t統(tǒng)計值均不顯著(X、X4的t統(tǒng)計量 的值的絕對值均小于2)，說明X2、X4這兩個變量對Y的影響不顯著，或者變量之間存在多重 共線的影響使其t值不顯著。3. 計量經(jīng)濟(jì)學(xué)檢驗(yàn)(1) 多重共線性檢驗(yàn)檢驗(yàn)：

13、由F=250.3763 F.5 (5 , 15)=4.62 (顯著性水平a =0.05表明模型從整體上看農(nóng) 業(yè)的總產(chǎn)值與解釋變量間線形關(guān)系顯著。這里采用簡單相關(guān)系數(shù)矩陣法對其進(jìn)行檢驗(yàn)：X2X3X4X5X6X21.0000000.9745550.9779520.963494-0.707561X30.9745551.0000000.9259220.907169-0.636056X40.9779520.9259221.0000000.991186-0.666705X50.9634940.9071690.9911861.000000-0.668959X6-0.707561-0.636056-0.666

14、705-0.6689591.000000從結(jié)果可知 X , X3 , X4 , X5之間存在高度相關(guān)修正：采用逐步回歸法對其進(jìn)行補(bǔ)救。根據(jù)以上分析，由于X5不符合經(jīng)濟(jì)意義，首先剔出。由于 X3的t值最大，線形關(guān)系強(qiáng)，擬合程度最好，因此把X3作為基本變量。，將剩下的四個因素重新進(jìn)行參數(shù)估計：新模型估計結(jié)果：Depe ndent Variable: YMethod: Least SquaresDate: 04/30/05 Time: 12:57Sample: 1989 2003In cluded observati ons: 15VariableCoefficie ntStd. Errort-St

15、atisticProb.C40826.1614791.732.7600670.0201X20.7584191.8547160.4089140.6912X310.015652.1511104.6560380.0009X4-0.0691420.132722-0.5209550.6137X6-1.7603650.515507-3.4148200.0066R-squared0.992440Mean depe ndent var18793.77Adjusted R-squared0.989416S.D.dependent var8203.735S.E. of regressi on844.0038Aka

16、ike info criteri on16.57539Sum squared resid7123424.Schwarz criteri on16.81141Log likelihood-119.3154F-statistic328.1758Durbi n- Watson stat1.444093Prob(F-statistic)0.000000Y =40826.16+ X0.758419 X2+ 10.01565 X3+ (-0.069142 ) X4+(-1.760365) X6t= (2.760067)(0.408914)(4.656038)(0.520955)(-3.414820)R2=

17、0.992440可以看出個因素的T統(tǒng)計量都得到了不同程度的改善。在前一模型的基礎(chǔ)上剔出X6，擬合優(yōu)度變差，但對C的t值影響很大，統(tǒng)計檢驗(yàn)t=-0.799100，不顯著。而且X4的系數(shù)為負(fù)，與經(jīng)濟(jì)意義相悖。Depe ndent Variable: YMethod: Least SquaresDate: 04/30/05 Time: 12:59Sample: 1989 2003In cluded observati ons: 15VariableCoefficie ntStd. Errort-StatisticProb.C-6120.4427659.170-0.7991000.4411X24.67

18、97942.0438192.2897300.0428X36.0162192.5319572.3761140.0368X4-0.2860730.163526-1.7493980.1080R-squared0.983624Mean depe ndent var18793.77Adjusted R-squared0.979157S.D.dependent var8203.735S.E. of regressi on1184.370Akaike info criteri on17.21499Sum squared resid15430045Schwarz criteri on17.40380Log l

19、ikelihood-125.1124F-statistic220.2341Durb in -Watson stat1.460596Prob(F-statistic)0.000000剔出X2進(jìn)行回歸，X4不但經(jīng)濟(jì)意義違背而且T統(tǒng)計值較小，不能通過檢驗(yàn)Depe ndent Variable: YMethod: Least SquaresDate: 04/30/05 Time: 13:01Sample: 1989 2003In cluded observati ons: 15VariableCoefficie ntStd. Errort-StatisticProb.C42613.8213585.49

20、3.1367140.0095X310.819350.84049912.872530.0000X4-0.0196500.052361-0.3752870.7146X6-1.8908790.389190-4.8585010.0005R-squared0.992313Mean depe ndent var18793.77Adjusted R-squared0.990217S.D.dependent var8203.735S.E. of regressi on811.4261Akaike info criteri on16.45864Sum squared resid7242535.Schwarz c

21、riteri on16.64746Log likelihood-119.4398F-statistic473.3484Durb in -Watson stat1.382173Prob(F-statistic)0.000000剔出X4進(jìn)行回歸雖然擬合優(yōu)度略有改善，但 X2的T統(tǒng)計值為-0.166847，通不過檢驗(yàn)，應(yīng)剔出X2在做回歸。而其他因素的統(tǒng)計值都較好。Depe ndent Variable: YMethod: Least SquaresDate: 04/30/05 Time: 13:00Sample: 1989 2003In eluded observati ons: 15Variabl

22、eCoefficie ntStd. Errort-StatisticProb.C42252.9014046.333.0081100.0119X2-0.1227070.735446-0.1668470.8705X310.786301.5091097.1474590.0000X6-1.8889060.437375-4.3187370.0012R-squared0.992235Mean depe ndent var18793.77Adjusted R-squared0.990117S.D.dependent var8203.735S.E. of regressi on815.5728Akaike i

23、nfo criteri on16.46884Sum squared resid7316750.Schwarz criteri on16.65765Log likelihood-119.5163F-statistic468.5100Durbi n- Watson stat1.360316Prob(F-statistic)0.000000綜合考慮所得結(jié)果,選擇含有X2 X3X6這三個因素的模型。再做剔出X2的模型的參數(shù)估計:Depe ndent Variable: YMethod: Least SquaresDate: 04/30/05Time: 13:06Sample: 1989 2003In

24、cluded observati ons: 15VariableCoefficie ntStd. Errort-StatisticProb.C41516.5112783.363.2477000.0070X310.544100.39553826.657640.0000X6-1.8519090.361404-5.1242020.0003R-squared0.992215Mean depe ndent var18793.77Adjusted R-squared0.990917S.D.dependent var8203.735S.E. of regressi on781.8390Akaike info

25、 criteri on16.33803Sum squared resid7335267.Schwarz criteri on16.47964Log likelihood-119.5352F-statistic764.7022Durbi n- Watson stat1.352428Prob (F-statistic)0.000000= =可以看出擬合優(yōu)度很好 F統(tǒng)計量的值在給定顯著性水平 a =0.05的情況下也較顯著，C , X3 ,X6的T統(tǒng)計值也很顯著，表明對 丫的影響也很顯著新模型估計結(jié)果：丫 =41516.51+ 10.54410 X3 +(-1.851909) X6t= (3.247

26、7)(426.65764)(-5.124202)R2 =0.992215(2) 異方差檢驗(yàn)檢驗(yàn)：利用Goid_Quandt檢驗(yàn)法檢驗(yàn)?zāi)Ｐ褪欠翊嬖诋惙讲睢r間定義為19891993,然后對丫 CX3用OLS法求的下列結(jié)果:Y=-6225.673+5.317281 X3t= (-2.982843)(7.083533)22R =0.943584 e =634718.8Depe ndent Variable: YMethod: Least SquaresDate: 04/30/05 Time: 13:12Sample: 1989 1993In eluded observati ons: 5Varia

27、bleCoefficie ntStd. Errort-StatisticProb.C-6225.6732087.161-2.9828430.0585X35.3172810.7506547.0835330.0058R-squared0.943584Mean depe ndent var8486.818Adjusted R-squared0.924779S.D.dependent var1677.103S.E. of regressi on459.9706Akaike info criteri on15.38938Sum squared resid634718.8Schwarz criteri o

28、n15.23315Log likelihood-36.47344F-statistic50.17643Durbi n- Watson stat1.632631Prob(F-statistic)0.005786= =將時間定義為19992003,然后對丫 CX3用OLS法求的下列結(jié)果Y=-44209.20 +16.62678 X3t= (-5.018903) (8.034508)2R =0.9555917 ef =777592.5Depe ndent Variable: YMethod: Least SquaresDate: 04/30/05 Time: 13:13Sample: 1999 20

29、03In eluded observati ons: 5VariableCoeffieie ntStd. Errort-StatistieProb.C-44209.208808.537-5.0189030.0152X316.626782.0694218.0345080.0040R-squared0.955591Mean depe ndent var26539.42Adjusted R-squared0.940788S.D.dependent var2092.227S.E. of regressi on509.1144Akaike info eriteri on15.59240Sum squar

30、ed resid777592.5Schwarz eriteri on15.43617Log likelihood-36.98099F-statistie64.55332Durbi n- Watson stat1.966867Prob(F-statistie)0.004026= =22Z e2e =777952.5/634718.8=1.22566481409 小于 F0.05 ( 4, 4) =6.39 接受 H0 不存在異方差將時間定義為19891993,然后對丫 C X6用OLS法求的下列結(jié)果:Y=-20445.55 +0.864900 X6t= ( -0.473093)(0.669589

31、)R =0.130018 ei =9787897Depe ndent Variable: YMethod: Least SquaresDate: 04/30/05 Time: 13:19Sample: 1989 1993In eluded observati ons: 5VariableCoeffieie ntStd. Errort-StatistieProb.C-20445.5543216.71-0.4730930.6684X60.8649001.2916890.6695890.5510R-squared0.130018Mean depe ndent var8486.818Adjusted

32、R-squared-0.159975S.D.dependent var1677.103S.E. of regressi on1806.276Akaike info eriteri on18.12510Sum squared resid9787897.Sehwarz eriteri on17.96887Log likelihood-43.31274F-statistie0.448349Durbi n- Watson stat0.815079Prob(F-statistie)0.551047= =將時間定義為19992003,然后對丫 C X6用OLS法求的下列結(jié)果Y=126537.0 +(-3.

33、097615) X6t= (48.84461)(-38.60686)R2 = 0.9979917 e； =35171.96Depe ndent Variable: YMethod: Least SquaresDate: 04/30/05 Time: 13:18Sample: 1999 2003In eluded observati ons: 5VariableCoefficie ntStd. Errort-StatisticProb.C126537.02590.60248.844610.0000X6-3.0976150.080235-38.606860.0000R-squared0.99799

34、1Mean depe ndent var26539.42Adjusted R-squared0.997322S.D.dependent var2092.227S.E. of regressi on108.2774Akaike info criteri on12.49644Sum squared resid35171.96Schwarz criteri on12.34022Log likelihood-29.24111F-statistic1490.489Durbi n- Watson stat2.434963Prob(F-statistic)0.000038= =22Z e2 寧瓦 ei =0

35、.000352467193491 小于 F0.05 (4, 4) =6.39 接受 H 不存在異方差利用WHITE檢驗(yàn)法檢驗(yàn)?zāi)Ｐ褪欠翊嬖诋惙讲?。結(jié)果如下:White Heteroskedasticity Test:F-statistic0.34392Probability0.873777Obs*R-squared2.406316Probability0.790533Test Equati on:Depe ndent Variable: RESIDEMethod: Least SquaresDate: 04/30/05 Time: 17:28Sample: 1989 2003In eluded

36、observati ons: 15VariableCoefficie ntStd. Errort-StatisticProb.C-9.11E+081.56E+09-0.5824040.5746X346121.5269356.030.6649970.5227X3A2-1.1529751.207919-0.9545140.3648X3*X6-1.1820212.098834-0.5631800.5871X650954.9488666.680.5746800.5796X6A2-0.7165461.248717-0.5738250.5801R-squared0.160421Mean depe nden

37、t var489017.8Adjusted R-squared-0.306012S.D.dependent var1172075.S.E. of regressi on1339457.Akaike info criteri on31.34260Sum squared resid1.61E+13Schwarz criteri on31.62582Log likelihood-229.0695F-statistic0.343932Durbi n- Watson stat2.515047Prob(F-statistic)0.873777益5 (5)=9.48773 2.406316，所以接受H。，表

38、明模型中隨機(jī)誤差項不存在異方差。(3) 自相關(guān)檢驗(yàn)檢驗(yàn)：從模型設(shè)定來看，沒有違背 D-W檢驗(yàn)的假設(shè)條件，因此可以用 D-W檢驗(yàn)來檢驗(yàn)?zāi)Ｐ褪欠翊嬖谧韵嚓P(guān)。根據(jù)上表中估計的結(jié)果，由DW=1.352428，給定顯著性水平 a =0.05,查 Durbin-Watson表，n=15，k 2，得 d| =0.946du =1.543因?yàn)镈W統(tǒng)計量為dl 1.352428 du，根據(jù)判定區(qū)域知位于無決定區(qū)域，不確定是否存在 一階正自相關(guān)，需要進(jìn)行修正。修正：采用廣義差分法對模型進(jìn)行修正。由 DW=1.352428，根據(jù) p =1DW/2，計算出 p=.373786。用 GENR 分別對 X3，X6和 Y

39、作廣義差分。即：GENRDY= Y-0.4894*Y(-1)GENRDX3 = X3-0.4894 X3 (-1)GENRDX6 = X6-0.4894 X6 (-1)新修正為：DY=C+ 1D X3+ -2D X6+uDepe ndent Variable: DYMethod: Least SquaresDate: 05/01/05 Time: 12:36Sample(adjusted): 1990 1993In cluded observati ons: 4 after adjusti ng en dpo intsVariableCoefficie ntStd. Errort-Statis

40、ticProb.C19018.461293.82414.699420.0432DX34.8402810.17024328.431660.0224DX6-1.0403890.049918-20.842060.0305R-squared0.999733Mean depe ndent var6037.017Adjusted R-squared0.999199S.D.dependent var1105.154S.E. of regressi on31.28098Akaike info criteri on9.837603Sum squared resid978.5000Schwarz criteri

41、on9.377324Log likelihood-16.67521F-statistic1871.802Durbi n- Watson stat2.946105Prob(F-statistic)0.016342（4）確定模型DY = 19018.46 + 4.84021*DX3 + (-1.040389 )*DX6X3=化肥施用量（萬噸）X6=農(nóng)業(yè)從業(yè)人員（萬人）由于該模型的回歸結(jié)果、t值以及F統(tǒng)計值均顯著，且不存在計量經(jīng)濟(jì)學(xué)問題，因此最后定型為此。根據(jù)1989-2003年的數(shù)據(jù)建立的模型中可以看出每增加一萬噸的化肥使農(nóng)業(yè)增加值增長 了 4.84021億元，每減少一萬人的農(nóng)業(yè)從業(yè)人員數(shù)可以是農(nóng)

42、業(yè)增加值增長1.040389，說明在我國的農(nóng)村中存在大量的剩余勞動力。模型還可表示為Yt =19018.46+4.84021 X3t-1.809202735 X3tJ +(-1.0140389 ) X6t +0.388882842 X6t j+0.373786 YtJGENR Zt =4.84021 X3t+(-1.0140389 ) Xet+( -1.809202735)X3tj+0.388882842 Xg則模型變?yōu)椋篩t =26245.95+1.712319 乙 +0.491976 YtDepe ndent Variable: YMethod: Least SquaresDate: 06

43、/06/05 Time: 12:49Sample(adjusted): 1990 2003In eluded observati ons: 14 after adjusti ng en dpo intsVariableCoefficie ntStd. Errort-StatisticProb.C26245.954065.2126.4562310.0000M1.7123190.2886635.9319010.0001Y(-1)0.4919760.0835535.8881660.0001R-squared0.992232Mean depe ndent var19669.41Adjusted R-s

44、quared0.990819S.D.dependent var7751.918S.E. of regressi on742.7666Akaike info criteri on16.24605Sum squared resid6068725.Schwarz criteri on16.38299Log likelihood-110.7224F-statistic702.4898Durbi n- Watson stat1.805031Prob(F-statistic)0.000000此模型存在滯后應(yīng)變量，因此 DW值失效，需要用德賓-H檢驗(yàn)h=(1-d/2)n 心匚 nV ar(：*) =0.09

45、74845*3.939100593=0.384001251取顯著性水平=0.05，查標(biāo)準(zhǔn)正態(tài)分布表得臨界值h/2 =1.96，由于|h|=0.384001251 h /2=1.96，則接受原假設(shè)=0，說明自回歸模型不存在一階自相關(guān)(5)單位根檢驗(yàn)ADF Test Statistic-3.3307951%Critical Value*-4.88705%Critical Value-3.828810% Critical Value-3.3588*MacKinnon critical values for rejection of hypothesis of a unit root.Augme nt

46、ed Dickey-Fuller Test Equati onDepe ndent Variable: D(Y)Method: Least SquaresDate: 06/07/05 Time: 20:03Sample(adjusted): 1991 2003Included observations: 13 after adjusting endpointsVariableCoefficie ntStd. Errort-StatisticProb.Y(-1)-0.4136580.124192-3.3307950.0088D(Y(-1)0.8828440.1961304.5013280.001

47、5C2072.293786.55032.6346610.0272TREND(1989)752.9303243.68733.0897390.0129R-squared0.728585Mean depe ndent var1694.592Adjusted R-squared0.638113S.D.dependent var1480.465S.E. of regressi on890.6042Akaike info criteri on16.66934Sum squared resid7138582.Schwarz criteri on16.84317Log likelihood-104.3507F

48、-statistic8.053172Durbi n- Watson stat1.926147Prob(F-statistic)0.006448即有:八丫=2072.293+752.9303匸0.413658丫t-1 +0.882844 Y-1其中：P=1，N=13單位根的T檢驗(yàn)結(jié)果為(H：行1 ):t =-0.413658/0.124192=-3.330795由表中給出的Mackinnon臨界值顯示，我們是不能拒絕 H）的，表明1989-2003年度的Y序列可 能是非平穩(wěn)序列。這點(diǎn)也可由 Y的時序圖得到驗(yàn)證。（見下圖）Y當(dāng)檢驗(yàn)結(jié)果不能拒絕零假設(shè)時，其結(jié)論尚待進(jìn)一步考證。由于水平有限對數(shù)據(jù)的平穩(wěn)性修正的 內(nèi)容省略。五、對模型的經(jīng)濟(jì)解釋及存在的問題1 .經(jīng)濟(jì)解釋從以上模型經(jīng)分析可得出：（1）從模型可以看出農(nóng)民對化肥的投入量，即模型中的化肥的使用量，是影響農(nóng)業(yè)產(chǎn)值增 長的最顯著因素。說明我國目前農(nóng)業(yè)生產(chǎn)中，農(nóng)民自己對農(nóng)業(yè)的投入所產(chǎn)生的效益最大（因?yàn)?化肥是農(nóng)民自己購買的，并且所占農(nóng)民支出份額甚大）。在最后確定的模型中，根據(jù) 1989-1993 年的數(shù)據(jù)建立的模型中可