#/20期檎滯當(dāng)劃舟冬竹《計(jì)量經(jīng)濟(jì)學(xué)》

課程實(shí)踐報(bào)告1系部：經(jīng)濟(jì)與管理系專業(yè)：國際經(jīng)濟(jì)與貿(mào)易任課教師：祖輝老師年級(jí)班級(jí)： 2013級(jí)2班組員：舒冠、淑琴、梁湘、冬雪稅收收入影響因素分析一基于Eviews模型的經(jīng)濟(jì)計(jì)量分析一、意義1960年以來，中國的經(jīng)濟(jì)基本處于高速增長之中。經(jīng)濟(jì)增長的高速發(fā)展，勢(shì)必會(huì)影響國家財(cái)政政策和國家福利水平。而稅收作為國家財(cái)政收入中最主要的部分對(duì)這些政策的實(shí)施也會(huì)有很大的影響。近些年來，國家的稅收也受到多種因素的影響。經(jīng)濟(jì)發(fā)展水平?jīng)Q定稅收收入水平，稅收同時(shí)也反作用于經(jīng)濟(jì)。要實(shí)現(xiàn)經(jīng)濟(jì)的持續(xù)增長，必須要求與經(jīng)濟(jì)緊密關(guān)聯(lián)的稅收符合其發(fā)展的要求，即政府籌集的稅收收入應(yīng)盡可能的滿足其實(shí)現(xiàn)職能的需求，同時(shí)又不至于損害經(jīng)濟(jì)的發(fā)展。影響未來的需求，我們需要研究影響中國稅收收入的主要原因，分析中央和地方稅收收入增長的數(shù)量規(guī)律，從結(jié)構(gòu)上對(duì)稅收收入的影響做一個(gè)很好的了解，對(duì)于預(yù)測(cè)中國稅收未來的增長趨勢(shì)具有重要的作用，對(duì)于我國的社會(huì)主義現(xiàn)代化建設(shè)具有重要意義。二、研究綜述影響稅收收入的因素有很多，如經(jīng)濟(jì)發(fā)展水平、稅收制度的設(shè)計(jì)、政府職能圍等。衛(wèi)剛認(rèn)為，影響稅收增長的因素是多元的，主要有經(jīng)濟(jì)增長、稅制結(jié)構(gòu)、稅收征管水平和價(jià)格因素；玉棟認(rèn)為，影響稅收收入增長的因素主要有經(jīng)濟(jì)增長、物價(jià)、稅收政策調(diào)整和稅收征管等幾個(gè)方面；安體富認(rèn)為，稅收收入主要受價(jià)格、經(jīng)濟(jì)結(jié)構(gòu)的變動(dòng)、經(jīng)濟(jì)效應(yīng)的變動(dòng)、稅收政策、財(cái)稅制度、稅收征管和稅款虛收的影響。以上學(xué)者都說明稅收收入的影響因素是多方面的，同時(shí)都認(rèn)為經(jīng)濟(jì)因素的重要性，但他們多集中于對(duì)稅收收入影響因素的全面分析，因而對(duì)經(jīng)濟(jì)性影響因素分析得不夠詳細(xì)。而郭慶旺認(rèn)為，稅收收入的經(jīng)濟(jì)增長彈性是1.536，經(jīng)濟(jì)增長對(duì)稅收增長的貢獻(xiàn)十分明顯。國家稅務(wù)總局科研所的研究結(jié)論為：在正常的經(jīng)濟(jì)運(yùn)行狀況下，經(jīng)濟(jì)稅源提供的收入應(yīng)占稅收總額的70%?80%或以上；而最為直接的、顯著的影響稅收增長的因素是經(jīng)濟(jì)增長和物價(jià)水平，而且物價(jià)水平對(duì)稅收收入增長的影響強(qiáng)于GDP的影響。胡才君認(rèn)為GDP與稅收收入負(fù)相關(guān)，進(jìn)出口總額、全國城鄉(xiāng)儲(chǔ)蓄存款年末余額、財(cái)政支出總量與稅收收入正相關(guān)。而歐陽若瀾認(rèn)為，GDP、財(cái)政支出和商品零售價(jià)格指數(shù)3個(gè)三個(gè)指標(biāo)與稅收收入存在正相關(guān)關(guān)系。鑫又認(rèn)為影響我國稅收增長的主要因素是財(cái)政支出和商品價(jià)格水平。以上文獻(xiàn)對(duì)影響稅收收入經(jīng)濟(jì)性因素的一個(gè)或幾個(gè)方面進(jìn)行了分析，而且由于數(shù)據(jù)選取不同和分析問題的角度不同，得出的結(jié)果在揭示各經(jīng)濟(jì)性影響因素對(duì)稅收收入的影響的程度方面還存在爭(zhēng)議：在影響稅收收入的經(jīng)濟(jì)性因素中各個(gè)因素對(duì)稅收收入增長的貢獻(xiàn)率是多少？占多少比重？因此，本文將重點(diǎn)研究影響稅收收入的經(jīng)濟(jì)性因素，對(duì)其進(jìn)行細(xì)分并進(jìn)行相應(yīng)的實(shí)證研究。根據(jù)前人的理論分析和數(shù)據(jù)的可得性等，本文從實(shí)證的角度，選取稅收收入、GDP、財(cái)政支出、商品零售價(jià)格指數(shù)和進(jìn)出口總額5個(gè)指標(biāo)做多元回歸模型，并采用計(jì)量分析工具對(duì)影響因素進(jìn)行分析。三、實(shí)證分析1、指標(biāo)的選?。焊鶕?jù)可操作性和科學(xué)性的原則,以與上文對(duì)稅收收入的定性分析,選取稅收收入為被解釋變量，在模型中用“Y”表示；選取國生產(chǎn)總值、財(cái)政支出、商品零售價(jià)格指數(shù)、進(jìn)出口總額為解釋變量，在模型中用分別用“X1”“X2”“X3”“X4”表示;原始數(shù)據(jù)主要來源于中國統(tǒng)計(jì)年鑒,數(shù)據(jù)采集的時(shí)間跨度為1960-2014年。2、模型的設(shè)定與數(shù)據(jù)的選?。海?）模型設(shè)定：研究影響中國稅收未來增長的主要原因，需要考慮以下幾個(gè)方面的容：A、對(duì)固定資產(chǎn)投資資金來源的衡量，用什么數(shù)據(jù)來表現(xiàn)呢？我們選用中國稅收收入作為被解釋變量（y）分析影響中國稅收未來增長的主要原因。B、數(shù)據(jù)性質(zhì)的選擇?？紤]到截面數(shù)據(jù)受到制約，時(shí)間序列數(shù)據(jù)更加合理，所以本項(xiàng)目選擇了1960年到2014年的時(shí)間序列數(shù)據(jù)來建立模型。C、影響因素的分析。從宏觀經(jīng)濟(jì)看，經(jīng)濟(jì)整體增長是稅收增長的基本源泉，所以經(jīng)濟(jì)整體增長是影響中國稅收未來增長的主要原因的主要影響因素，所以選用國生產(chǎn)總值（GDP）作為經(jīng)濟(jì)整體增長水平的代表。除此之外，根據(jù)經(jīng)濟(jì)理論，還有眾多因素會(huì)影響中國稅收未來增長的主要原因：首先，公共財(cái)政的需求。稅收收入是財(cái)政收入的主體，社會(huì)經(jīng)濟(jì)的發(fā)展和社會(huì)保障的完善等都對(duì)公共財(cái)政提出了要求，因此對(duì)預(yù)算支出所表現(xiàn)的公共財(cái)政的需求（即財(cái)政支出）對(duì)當(dāng)年的稅收收入可能會(huì)產(chǎn)生影響，但是其數(shù)據(jù)獲得比較困難，因?yàn)楣藏?cái)政的需求與財(cái)政支出關(guān)系密切，所以選擇財(cái)政支出作為其代表。其次，物價(jià)水平。居民的收入水平、消費(fèi)水平與物價(jià)水平有一定的關(guān)系，我們選擇商品零售價(jià)格指數(shù)作為物價(jià)水平的代表。再次，進(jìn)出口總額。進(jìn)出口的收入水平與稅收收入存在一定的聯(lián)系，關(guān)系到關(guān)稅收入，所以我們選擇進(jìn)出口總額來作為解釋變量。因此，準(zhǔn)備將“國生產(chǎn)總值（XI）”、“財(cái)政支出（X2）”“商品零售價(jià)格指數(shù)（X3）”、“進(jìn)出口總額（X4）”作為解釋變量建立模型。D、模型形式的設(shè)計(jì)我們將方程形式設(shè)定為二次型Y=80+81X1+82X2+83X3+84X4+u然后將影響因素以某種方式引入模型。（2）數(shù)據(jù)：對(duì)稅收收入影響因素的模型分析單位：億元注：表中數(shù)據(jù)均來自《中國統(tǒng)計(jì)年鑒》

年份1ncn稅收收入（億元）Yqona匚國內(nèi)生產(chǎn)總值（億元）X1-\AULQ匚八財(cái)政支出（億元）X2a4。a0商品零售價(jià)格指數(shù)（％）X31no1n進(jìn)出口總額（億元）X4100an. . . . .1qc1 1izQna i990QO QICno lieon nnnc\1961^^^^^^^^^^^^^158.76^^^^^^^^^^^^^^^^^^^^^^"1220.90^^^^^^^^^^^^^^^^^^^^^^^^^^-356.09^^^^^^^^^^^^^^^^^^^^^^^^^^^^-116.20^^^^^^^^^^^^^^^^^^^^^^^^^-90.70 ■iq?9 16907 iicion 9Q4QQ inQon onon1962^^^^^^^^^^^^^^-162.07 1151.20^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^294.88^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^-103.80^^^^^^^^^^^^^^^^^^^^^^^^^^^^-80.90 i1QCQ 1aAQ1 19QAA0 QQ9HIZ CtA1A QfZ7八19631 164.31 1 1236.40 1 332.05 1 94.10 1 85.70 i1QC/I 1Q9AA 1/仄仄仄八 QQQ7Q QCQH Q7仄八1964 1 182.00 1 1455.50 1 393.79 1 96.30 1 97.50 i1Q6仄 90/IQ0 171790 AI^QQ7 Q7Qn iio/in19651 204.30 1 1717.20 1 459.97 1 97.30 1 118.40 i1Q仆仆 991Q? 1Q7Q1n 仄Q76仄 QQ7H 1971019661 221.96 1 1873.101537.65 1 99.70 1 127.10 i1QC7 1QC(20 17Q0QO JQQQJ QQQA 1199八19671 196.63 1 1780.30 1 439.84 1 99.30 1 112.20 iiqcq iqi仄6 i7Q0on Q仄7QJ innin ino仄八19681 191.56 1 1730.20 1 357.84 1 100.10 1 108.50 i1QCQ 9Q仄AA 1CM仄on 仄9仄QC QQQH 1f\7八八19691 235.44 1 1945.801525.86 1 98.90 1 107.00 i1Q7n 9QIon 9961Qn CiAQA1 QQQH IIOQH19701 281.20 1 2261.30 1 649.41 1 99.80 1 112.90 i1Q71 Q19仄6 9/Q仄Qn 79Q17 QQQH 1onQH1971 1 312.56 1 2435.30 1 723.17 1 99.30 1 120.90 iiQ79 qi7no O!zqnon 76仄QC QQon 1/16Qn19721 317.02 1 2530.20 1 765.86 1 99.80 1 146.90 i1Q7Q QJQQ(Z 97QQAA QAQ7Q 1AACA OOA仄八1973 1 348.95 1 2733.401808.78 1 100.60 1 220.50 i1Q7/I QCAAA 9QAQ7f\ 7QAQ仄 1AA|ZA 9Q99八19741 360.40 1 2803.70 1 790.25 1 100.50 1 292.20 i1Q7仄 /10977 QHiQin oonqq innon onnaa19751 402.77 1 3013.101820.88 1 100.20 1 290.40 iiQ7C /|H7qc oneiizn oneon innon 9101976 1 407.96 1 2961.50 1806.20 1 100.30 1 264.10 i1din /tcooi oqo11n 0/19co 1nonn 0r7。匚八977468.273221.±0843.5302.00272.501070匚1coooa匚八on11QQno1nnqex。匚匚nn978 519.28 3650.20 22.09 00.70 355.00iQ7Q (zq7co anc77A i9017a 1nonn a;zacn19791 537.82 1 4067.70 1 1281.79 1 102.00 1 454.60 11non 「717H /I久久1cn 199Qqq 1ncnn 「7cnn19801 571.70 1 4551.60 1 1228.83 1 106.00 1 570.00 11noi aocon aono1n 11ooa1 1noan 口。匚on1C7O1(O<J.1QQO7nnnoizqqqnn1OOQQQ1nian771QA19821QQQI 700.0277(z久Q1 5333.00「Q71QB1 1229.981AAQ「O101.901niizn1 771.30 1ocn1n1983I 775.591 5975.30| 1409.521 101.501 860.10 11QQ/iQA7Q「700/2QAi70ino1noon1oninn19841 947.351 7226.301 1701.021 102.801 1201.00 11985 2040.79 9039.90 2004.25 108.80 2066.701qqc ohqh70 1OQOQQn 9onjq1 1ncnn 9ironaa1986 1 2090.73 1 10308.80 1 2204.91 1 106.00 1 2580.40 11Q07 01/in0a 19109on 00/2010 1A7on 0no/ion1987 2140.36^^^^^^^^^^^^^^^^^^^^^12102.20^^^^^^^^^^^^^^^^^^^^^^^^"2262.18^^^^^^^^^^^^^^^^^^^^^^^^^^^-107.30^^^^^^^^^^^^^^^^^^^^^^^"3084.20 11QQQ 9QQAA7 1|Z1A1 1A OJQ1O1 11Q(ZA QQ01QA1988^^^^^^^^^^^^"2390.4715101.10^^^^^^^^^^^^^^^^^^^^^^^^"2491.21118.50^^^^^^^^^^^^^^^^^^^^^^^"3821.80 i1QQQ 9797AA 17HQnQB OQQQ7Q 117QB /I1HQA1989^^^^^^^^^^^^"2727.40^^^^^^^^^^^^^^^^^^^^^17090.30^^^^^^^^^^^^^^^^^^^^^^^^"2823.78^^^^^^^^^^^^^^^^^^^^^^^^^^^-117.80^^^^^^^^^^^^^^^^^^^^^^^"4155.90 i1OOH oqoioa AQTlAQC\ qnoorn inoin H公八1n1990^^^^^^^^^^^^"2821.86^^^^^^^^^^^^^^^^^^^^^18774.30^^^^^^^^^^^^^^^^^^^^^^^^"3083.59^^^^^^^^^^^^^^^^^^^^^^^^^^^-102.10^^^^^^^^^^^^^^^^^^^^^^^-5560.10 i1QQ1 OQQA17 01QQC!ZA QQQ/2(20 1AOQA 7OO|ZQA1991^^^^^^^^^^^^"2990.1721895.50^^^^^^^^^^^^^^^^^^^^^^^^"3386.62^^^^^^^^^^^^^^^^^^^^^^^^^^^-102.90^^^^^^^^^^^^^^^^^^^^^^^"7225.80 i1QQO QOQttQ1 Q7A/2QQA Q7JOOA 1A(ZAA Q11(Z/2A19921 3296.91 1 27068.30 1 3742.20 1 105.40 1 9115.60 i19931QQ/i4225.3011℃QQ35524.30AQA(ZQ/2A4642.30|Z7QO(20113.201017A11271.00OAQQIQA1994I 5126.884 48459.601 5792.621 121.701 20381.90 11QQIZ/2AQQC\Aa-\1onon(2QQO7011/1QHQQJQQQA19951QQ/2| 6038.04/2QAQQA6 61129.8071(Z7QQA1 6823.727QQ7仄仄114.801n/21n1 23499.900/i100on19966 6909.841 71572.301 7937.551 106.101 24133.801QQ7QQQ/IAA7Qjaq(-nna00rz?1nnqaaZ2Q/27an19978 8234.047 79429.501 9233.561 100.801 26967.20 11998 9262.80 84883.70 10798.18 97.40 26849.701QQQ 1 久Q qai077c 10107an Q7nn oqqq/2qa19991 10682.58 1 90187.70 1 13187.67 1 97.00 1 29896.30 1onnn 10(zoi11 QQ77AQB 11QQAcn qqrn QQ97Qon20001 12581.51 1 99776.30 1 15886.50 1 98.50 1 39273.20 1200115301.38110270.4018902.5899.2042183.602002 17636.45 121220.00 22053.15 98.70 51378.20oaaq onni7qi 1QdA/AC 。/1A/Qqiz qqqa 70/1QQcn2003 1 20017.31 1 136564.60 1 24649.95 1 99.90 1 70483.50 iOAA/I 0/11erZ2Q 1tin71/I/In OQJQ/2QQ 1AOQA QHQQ1H2004 1 24165.68 1 160714.40 1 28486.89 1 102.80 1 95539.10 i200528778.54Aor\A。匚185895.80q1口a匚公a.c\33930.284HzlOO70100.801clc\c\116921.801ahg74r\r\20061 34804.354匚公c1nr7I 217656.60OCOH1GAr\4 40422.73/tG7O1。匚1 101.001noOH1 140974.00200/OHHO1 45621.97C7zlOO97c2 268019.4091￡27C71口C\4 49781.35/?OC7QOQ.Q.1 103.801八匚rjn1 166863.701r7CCC1An2008OHHG1 54223.79匚C匚C1匚c3 316751.70。/1匚公ccon1 62592.667日cccno1 105.90noOH1 179921.471匚八aAOQ.Q.2009aninI 59521.5970ain7Q1 345629.20AHQQAQr\r\7 76299.93QQQ7/I1a98.801no1n1 150648.66an1700112010ani-11 73210.79QQ7QQQQ4 408903.00AQA-tOQ(ZA8 89874.161AQO/I77QI 103.101nannI 201722.15a0/2/in-1nr2011an-iq1 89738.391HH￡21AOO4 484123.50C794109匚八1 109247.79-iqerneqnr7104.901f\Q八八1 236401.95Ozl41￡2Hon2012aniq1 100614.2811neonicxI 534123.50(ZQQA1QQA1 125952.971/IHO101C\1 102.00iniar\1 244160.20o(ZQ1(20QA20131—20141 110530.701 119158.055 588018.806 636138.701 140212.101 151661.54101.40―1 101.001 258168.902 264334.493、實(shí)證分析：將方程形式設(shè)定為二次型Y=B0+B1X1+B2X2+B3X3+B4X4+u（1）、多重共線性EVIEWS的最小二乘估計(jì)結(jié)果為DependentVariable:YMethod:LeastSquaresDate:11/03/15Time:21:06Sample:19602014

Includedobservations:55VariableCoefficientStd.Errort-StatisticProb.C-3423.7921285.933-2.6624980.0104X1-0.0019670.007001-0.2810290.7798X20.6866520.02093432.801230.0000X332.4841112.435582.6121910.0118X40.0617930.0067479.1586550.0000R-squared0.999720Meandependentvar16116.57AdjustedR-squared0.999697S.D.dependentvar29942.14AkaikeinfoS.E.ofregression521.1390criterion15.43642Sumsquaredresid13579295Schwarzcriterion15.61890Hannan-QuinnLoglikelihood-419.5015criter.15.50699F-statistic44552.34Durbin-Watsonstat1.506966Prob(F-statistic)0.000000經(jīng)濟(jì)意義檢驗(yàn)：從回歸的結(jié)果可以看出，財(cái)政支出門2）、商品零售價(jià)格指數(shù)（X3）、進(jìn)出口總額（X4）符號(hào)均為正，符合經(jīng)濟(jì)意義。國生產(chǎn)總值（XI）符號(hào)為負(fù)，不符合經(jīng)濟(jì)意義。所以存在某種干擾。統(tǒng)計(jì)推斷檢驗(yàn)。該模型R/=0.999720,修正的R/=0.999697,可決系數(shù)很高，擬合優(yōu)度較好，F(xiàn)檢驗(yàn)值=44552.34,明顯顯著。但是當(dāng)a=0.05時(shí)，ta/2（n-k-1）=ta/2（55—4—1）=t0.025（50）=2.009,x1的系數(shù)t檢驗(yàn)不顯著。X1X2X3X4X11.0000000.994288-0.0741250.987209X20.9942881.000000-0.0806620.972683X3-0.074125-0.0806621.000000-0.068631X40.9872090.972683-0.0686311.000000由相關(guān)系數(shù)表可以看出，各解釋變量之間除了乂3之外的相關(guān)系數(shù)較高，證實(shí)確實(shí)存在嚴(yán)重的多重共線性。修正多重共線性：運(yùn)用OLS方法逐一求Y對(duì)各個(gè)解釋變量的回歸。結(jié)合經(jīng)濟(jì)意義和統(tǒng)計(jì)檢驗(yàn)選出擬合效果最好的一元線性回歸方程。DependentVariable:YMethod:LeastSquaresDate:11/03/15Time:21:20Sample:19602014Includedobservations:55VariableCoefficientStd.Errort-StatisticProb.C-1579.503391.7915 -4.0314900.0002X10.1839990.002092 87.951270.0000R-squared0.993195Meandependentvar16116.57AdjustedR-squared0.993067S.D.dependentvar29942.14AkaikeinfoS.E.ofregression2493.187criterion18.51620Sumsquaredresid3.29E+08Schwarzcriterion18.58919Hannan-QuinnLoglikelihood-507.1954criter.18.54442F-statistic7735.426Durbin-Watsonstat0.076503Prob(F-statistic)0.000000DependentVariable:YMethod:LeastSquaresDate:11/03/15Time:21:21Sample:19602014Includedobservations:55VariableCoefficientStd.Errort-StatisticProb.C214.5864187.4265 1.1449100.2574X20.8023710.004470 179.48440.0000R-squared0.998357Meandependentvar16116.57AdjustedR-squared0.998326S.D.dependentvar29942.14AkaikeinfoS.E.ofregression1224.887criterion17.09477Sumsquaredresid79518466Schwarzcriterion17.16776Loglikelihood-468.1062Hannan-Quinn17.12300

criter.criter.32214.660.000000Durbin-Watsonstat32214.660.000000Durbin-Watsonstat0.849558Prob(F-statistic)DependentVariable:YMethod:LeastSquaresDate:11/03/15Time:21:21Sample:19602014Includedobservations:55VariableCoefficientStd.Errort-StatisticProb.C55391.2173901.59 0.7495270.4569X3-380.9575715.7482 -0.5322510.5968R-squared0.005317Meandependentvar16116.57AdjustedR-squared-0.013451S.D.dependentvar29942.14AkaikeinfoS.E.ofregression30142.85criterion23.50097Sumsquaredresid4.82E+10Schwarzcriterion23.57396Hannan-QuinnLoglikelihood-644.2766criter.23.52920F-statistic0.283291Durbin-Watsonstat0.027353Prob(F-statistic)0.596777DependentVariable:YMethod:LeastSquaresDate:11/03/15Time:21:21Sample:19602014Includedobservations:55VariableCoefficientStd.Errort-StatisticProb.C-915.0621932.0995 -0.9817210.3307X40.3796600.010499 36.160170.0000R-squared0.961045Meandependentvar16116.57AdjustedR-squared0.960310S.D.dependentvar29942.14AkaikeinfoS.E.ofregression5965.148criterion20.26094Sumsquaredresid1.89E+09Schwarzcriterion20.33394Hannan-QuinnLoglikelihood-555.1759criter.20.28917F-statistic1307.558Durbin-Watsonstat0.287228Prob(F-statistic)0.000000其中加入乂2的方程修正的RM最大，其方程為:Y=214.5864+0.802371X2(1.144910) (179.4844)修正的R^2=0.998326 SE=79518466 F=32214.66所以，以x2為基礎(chǔ)，順次加入其他的自變量逐步回歸：當(dāng)a=0.05時(shí)，ta/2(n-k-1)=ta/2(55-4-1)=t0.025(50)=2.009Y=-314.8053+0.583638X2+0.050579X1-2.11601419.932197.513019修正的R^2=0.999182SE=138129403F=32986.08應(yīng)為x1的引入改進(jìn)了修正的R^2和F值且其他回歸參數(shù)的t檢驗(yàn)在統(tǒng)計(jì)上仍然顯著，所以保留xl,方程為：Y=-314.8053+0.583638X2+0.050579X1繼續(xù)：Y=-3588.465+0.586619X2+0.049981X1+31.73905X3-1.72249720.273907.5168711.575298修正的R^2=0.999205SE=36360185F=22618.10式中X3不顯著，刪去。繼續(xù)：Y=-73.73216+0.683414X2+-0.001258X1+0.061678X4-0.74070730.98268-0.1703508.660592修正的RM=0.999662SE=15432613F=53312.68X4雖然顯著，但是它的引入影響了其他回歸參數(shù)的估計(jì)值的數(shù)值，使得隨機(jī)項(xiàng)和乂1的回歸參數(shù)通不過t檢驗(yàn)，所以刪去x4所以：Y=-314.8053+0.583638X2+0.050579X1-2.116014 19.93219 7.513019修正的R^2=0.999182SE=138129403F=32986.08最小二乘估計(jì)為：DependentVariable:YMethod:LeastSquaresDate:11/04/15Time:14:42Sample:19602014Includedobservations:55VariableCoefficientStd.Errort-StatisticProb.C-314.8053148.7728 -2.1160140.0392X10.0505790.006732 7.5130190.0000X20.5836380.029281 19.932190.0000R-squared0.999212Meandependentvar16116.57AdjustedR-squared0.999182S.D.dependentvar29942.14AkaikeinfoS.E.ofregression856.3047criterion16.39613Sumsquaredresid38129403Schwarzcriterion16.50562Hannan-QuinnLoglikelihood-447.8936criter.16.43847F-statistic32986.08Durbin-Watsonstat1.089983Prob(F-statistic)0.000000（2）、自相關(guān)（序列相關(guān)）令e=residDependentVariable:EMethod:LeastSquaresDate:11/04/15Time:14:49Sample(adjusted):19612014Includedobservations:54afteradjustments

C327.0372202.7418 1.6130730.1132X10.0301700.012040 2.5058240.0156X2-0.1111930.044899 -2.4765080.0168T1"2-0.9950640.417083 -2.3857720.0210E(-1)0.4376820.128433 3.4078570.0013R-squared0.293946Meandependentvar-1.778895AdjustedR-squared0.236309S.D.dependentvar848.0834AkaikeinfoS.E.ofregression741.1349criterion16.14226Sumsquaredresid26914768Schwarzcriterion16.32643Hannan-QuinnLoglikelihood-430.8411criter.16.21329F-statistic5.099957Durbin-Watsonstat1.521476Prob(F-statistic)0.001623LM檢驗(yàn)=54*RC2=15.873084>X/(1)=3.84。然后查看e(-1)的t檢驗(yàn)VariableCoefficientStd.Errort-StatisticProb.=3.407857>t(50)=2.009,所以，存在一階自相關(guān)。繼續(xù)：DependentVariable:EMethod:LeastSquaresDate:11/04/15Time:15:08Sample(adjusted):19622014Includedobservations:53afteradjustmentsVariableCoefficientStd.Errort-StatisticProb.C352.3572208.28231.6917290.0973X10.0257200.0123302.0859130.0424X2-0.0915240.046308-1.9764030.0540T1"2-0.9718900.422199-2.3019680.0258E(-1)0.5253890.1367553.8418180.0004E(-2)-0.2564210.146405-1.7514540.0864R-squared0.337121Meandependentvar-5.661248AdjustedR-squared0.266602S.D.dependentvarAkaikeinfo855.7146S.E.ofregression732.8223criterion16.13795Sumsquaredresid25240341Schwarzcriterion16.36101Loglikelihood-421.6558Hannan-Quinn16.22373criter.F-statisticProb(F-statistic)4.7805750.001298F-statisticProb(F-statistic)4.7805750.001298Durbin-Watsonstat1.530441LM檢驗(yàn)=54*RM=17.867413>X/(2)=5.99。然后查看e(-1)e(-2)的t檢驗(yàn)，經(jīng)查表可知e(-2)t值不滿足條件，所以，不存在二階自相關(guān)，即存在一階自相關(guān)。自相關(guān)的修正：使用迭代法修正得到：DependentVariable:YMethod:LeastSquaresDate:11/04/15Time:15:14Sample(adjusted):19612014Includedobservations:54afteradjustmentsConvergenceachievedafter11iterationsVariableCoefficientStd.Errort-StatisticProb.C2544.3662273.4391.1191710.2685X10.1889350.01520012.430090.0000X20.0931080.0544351.7104350.0935T1"2-5.7407271.700883-3.3751440.0015AR(1)0.9142670.06474614.120730.0000R-squared0.999710Meandependentvar16410.75AdjustedR-squared0.999687S.D.dependentvar30142.96AkaikeinfoS.E.ofregression533.5896criterion15.48515Sumsquaredresid13951177Schwarzcriterion15.66932Hannan-QuinnLoglikelihood-413.0991criter.15.55618F-statistic42271.44Durbin-Watsonstat1.146070Prob(F-statistic)0.000000InvertedARRoots.91DW=1.146070，查表得出上限為1.724，下限為DW=1.146070<1.414,所以存在正相關(guān)。再次修正自相關(guān)使用迭代法修正得到：DependentVariable:YMethod:LeastSquaresDate:11/04/15Time:15:19Sample(adjusted):19622014Includedobservations:53afteradjustmentsConvergenceachievedafter10iterationsVariableCoefficientStd.Errort-StatisticProb.C1521.278934.58871.6277510.1103X10.2108320.01191417.696330.0000X2-0.0058670.047559-0.1233700.9023T1"2-5.3034061.080340-4.9090150.0000AR(1)1.4427720.12832011.243560.0000AR(2)-0.5765840.129935-4.4374930.0001R-squared0.999786Meandependentvar16717.40AdjustedR-squared0.999763S.D.dependentvar30346.26AkaikeinfoS.E.ofregression466.9039criterion15.23639Sumsquaredresid10245966Schwarzcriterion15.45945Hannan-QuinnLoglikelihood-397.7645criter.15.32217F-statistic43923.39Durbin-Watsonstat1.883360Prob(F-statistic)0.000000InvertedARRoots.72+.24i.72-.24iDW=1.883360,查表得出上限為1.768,下限為1.374,因?yàn)?.768<DW=1.883360<2.232,所以不存在自相關(guān)。所以，本文模型估計(jì)的結(jié)果為：y=1521.278+0.210832x1+-0.005867x21.62775117.69633-0.123370R2=0.999786DW=1.883360F=43923.39(3)、異方差White檢驗(yàn)Y=B0+B1x1+82X2+uS八2=a0+a1x1+a2x2+a3(x1)八2+a4(x2)八2+a5(x1*x2)+eDependentVariable：EMethod：LeastSquaresDate：10/14/15Time：15:19Sample：19602014Includedobservations：55VariableCoefficientStd.Errort-StatisticProb.C127420.1222588.70.5724470.5696X1-70.8442739.79185-1.7803710.0812X2292.6836208.06051.4067230.1658(X1廠20.0023100.0008172.8294090.0067(X2)"20.0375460.0158812.3642820.0221X1*X2-0.0186260.007244-2.5712830.0132R-squared0.560929Meandependentvar693261.9AdjustedR-squared0.516126S.D.dependentvar1724521.AkaikeinfoS.E.ofregression1199595.criterion30.93554Sumsquaredresid7.05E+13Schwarzcriterion31.15452Hannan-QuinnLoglikelihood-844.7272criter.31.02022F-statistic12.51985Durbin-Watsonstat2.686561Prob(F-statistic)0.000000從表中可以看出，nR^2=55*0.560929=30.8511由white檢驗(yàn)知，在a=0.05下，查表得知x/(5)=11,07,因?yàn)閚R^2=30.8511>x^2(5)=11.07,所以拒絕原假設(shè)，說明模型中的隨機(jī)誤差存在異方差。

修正異方差：使用w=1/abs(resid)作為權(quán)數(shù)，得出:DependentVariable:YMethod:LeastSquaresDate:11/04/15Time:15:28Sample:19602014Includedobservations:55Weightingseries:1/ABS(RESID)VariableCoefficientStd.Errort-StatisticProb.C-312.89706.083272 -51.435630.0000X10.0513790.001756 29.261060.0000X20.5795030.008802 65.840110.0000WeightedStatisticsR-squared0.999754Meandependentvar3107.385AdjustedR-squared0.999744S.D.dependentvar5593.490AkaikeinfoS.E.ofregression97.82912criterion12.05732Sumsquaredresid497667.9Schwarzcriterion12.16681Hannan-QuinnLoglikelihood-328.5764criter.12.09966F-statistic105637.8Durbin-Watsonstat0.656664Prob(F-statistic)0.000000UnweightedStatisticsR-squared0.999211Meandependentvar16116.57AdjustedR-squared0.999181S.D.dependentvar29942.14S.E.ofregression856.8133Sumsquaredresid38174711Durbin-Watsonstat1.079339估計(jì)的結(jié)果為：y =-312.8970+0.051379x1+0.579503x2(-51.43563)(29.26106)(65.84011)R/=0.999754DW=0.656664F=105637.8再次做懷特檢驗(yàn)：

DependentVariable:EMethod:LeastSquaresDate:11/04/15Time:15:34Sample:19602014Includedobservations:55Weightingseries:1/ABS(RESID)C119.8746 31.15437 3.8477630.0003X10.021442 0.003928 5.4582490.0000X2-0.178276 0.039257 -4.5412850.0000X1"2-5.07E-07 1.87E-07 -2.7096590.0093X2"2-1.41E-05 3.74E-06 -3.7746330.0004X1*X25.63E-06 1.65E-06 3.4124310.0013WeightedStatisticsR-squared0.394156 Meandependentvar26.71224AdjustedR-squared0.332335 S.D.dependentvar91.44191AkaikeinfoS.E.ofregression77.76088criterion11.64782Sumsquaredresid296291.0 Schwarzcriterion11.86680Hannan-QuinnLoglikelihood-314.3151criter.11.73251F-statistic6.375776 Durbin-Watsonstat0.702770Prob(F-statistic)0.000124UnweightedStatisticsR-squared0.379217 Meandependentvar-1.41E-12AdjustedR-squared0.315872 S.D.dependentvar840.2976S.E.ofregression695.0273 Sumsquaredresid23670088Durbin-Watsonstat0.964671VariableCoefficientStd.Errort-StatisticProb.從表中可以看出，nR八2=55*0.394156=21.67858,由white檢驗(yàn)知，在a=0.05下，查表得知X/(5)=11.0705,因?yàn)閚R八2=21.67858>X八2(5)二11.0705，所以拒絕原假設(shè)，說明模型中的隨機(jī)誤差存在異方差。繼續(xù)修正：使用w=1/(resid)M作為權(quán)數(shù)，得出：

DependentVariable:YMethod:LeastSquaresDate:11/04/15Time:15:39Sample:19602014Includedobservations:55Weightingseries：1/(RESID廠2VariableCoefficientStd.Errort-StatisticProb.C-291.44350.512662 -568.49020.0000X10.0569050.000157 361.72320.0000X20.5375000.000901 596.47750.0000WeightedStatisticsR-squared1.0000