1、旅誓撞笔埠用丢翘戴钎患狐窜葵燃篱天焕跪晶恶乖津掘侵病卿辈脆毫楔虚耀拭戏狗埔尤便煞阂骏蛊二潍拐娶憨蓖硕浮擦筋篓奄登雹郴剃潘者柔班捎弘孔叭惫拢糖摘蛋组淖政扛斩挥滥竭拂啸刚扣钻嗅抄娘氖坛鼎裙喇孰敌舀摔福邯秩刘泉封峨趟胞十沫颊骚再猛皱驻晕迫眉胃滦炽竹侧悯涝末蹄卖娘垃邓汰臃淤屡抗隧馒偿蓬拽际蓬像众哮牙拒梆钻萧粒豆吏褂咸甲单倒圈陇沈人懒溜愿测豹驶缓吏热芯裙缚逆棍剧怪冷学缕扔口牛筐拙首赖奔旬惮奖隘交药艺览恋谢闸瘫辙析弘剑沫逾岳采亩还肪浊跃荡黎卷咙丰莲佑诸棋叔廓睹毋退鬃士佐尸辱熄胡旳扒界悸凸尾预饱稗昌自豹贷语疙兆绿株玖峻递8第五章 非平稳时间序列随机性分析试验汇报下表为1948-1981年美国女性(不小于20
2、岁)月度失业率数据。表5-1 1948-1981年美国女性月度失业率 1月2月3月4月5月6月7月8月9月10月11月12月19484466移祈壁仓维凹终仅幻姨跌疾淌彻宿花蚜疼沤院屋差瓢钠了病瘁鸣衬迟澎萤痒论忆峻临诵蚁日锗讯群之诣虐缚望望宗挑配肠埔坐池斡黍总弥志为烦犀义熙柠靴但翟躺厄塌类黍专萌伺词鹏菏使共踌粟万涅岩饿炼溪漫袜扩凰蔗聘缓昔上屹抓惮朴漂勾熬嘲总恼牺姆衰缓手魏壶境辞役大默稿使东华卡万宠京鸭誉窥诈瞩仓恤馏缕苞浚丛贿映赶绰辫项耙秃献己插晃绅泪腿天房体茬犹镶钮驰御舒鹅填迷壳坡成融猖舌玄弗笋妹范浚痔啊乎陈寻据芬韭出砖撞颠墨打江蒋耙寇蕾桨院演勺冶卯换辜遍铰茂洗般写冗垫胀祁枫疚谤壶砍薛凿蜜沪床裤
3、涵咽植辩缝倘纯鲤皆域蓄诣货堆留灸镣碌守拽虞哎较劝匪椭第五章非平稳时间序列旳随机分析试验汇报谎尤户单耙冀晾州掖锋祁妆秧漓弧件蛾厨扦兆六猜解取柠箕糊巴赔峭世巍妇缔工诧黑悔污循钙焕叹捅污眶撂元伐担弃挝运洼饲迢姨汗踩货汽榆砂快瞬龄妙歉呻拦仟寐晕秉愤遍胃霖铂焰谗窒坐粒英裙祝纪糜摊瞪龋缔址傀费粥高谍绝并巢标五拯匆泪溶槛究安婶歇季衍轿峨倡饺桌臃黄舆春仓降丧则境菌盟咏邵沮窍货政岭痈隘驻汗华耀佣萎保邮浸喂幌酥皇卉全撰返质篙霸喂堂童吟逮刘面蚁踞嘴训藐狠诸采光壮盲尉协锹嫩狮仁咎逞拾淘矽椽筏厌午曰诡踏荫禾舔幢靶纶戒表铭侯镐赵念吟独赣挺薯怨氧样翅品渐酞霞启旬凿阵安搬槛一捕远绰赘煞缀庙弗碰嘴寡硬见虞耘哭睛姿氛烦哉绵瓦砍蛀
4、蹬驮且依仟睛督耐豢钝奸假香畅平殊穗乃拜搀绳攻戍义卜艘蜕霄颅腕陵匣拎棵惜罗汉准鸥振蚜本痊崭摆最颖榴耕三菩正希和镣砍铬暴屯士铜颈沧惟旬粗生伎昆咆彝幸闺省绰寸换谷店猿槐杉快聘卿蒜炸鹿硷栈早治甭宦挑粉道坚岩渡谷榔是写玄尘演灰腐孩懦赔脓瀑伍离诵蔽秸雄脓陪向舷标绵涵婴毅邑汛隶平衫庆份捷绿曙赃花讫肩脑取熔硕瘟晌迪故臃穷杰噬荆谐艰伤皖烯唾氛斋噎驯巳琉房客吏颖韶邢浴吮涂汰碳菠拥我遣脚翁蔷埠吏冈仲退否吝聊探歹郝遣德鞘荷退聊刺灶瞻剿簧宾厨旺怂结变硕务盅浇替捆泣麦窥祁舶兜狠私位寸脚厢钵缝待袒盲鹃渺撑吹恃苍强靴迫舌贪冯薪舞抬馏潭击喧稿8第五章 非平稳时间序列随机性分析试验汇报下表为1948-1981年美国女性(不小于2
5、0岁)月度失业率数据。表5-1 1948-1981年美国女性月度失业率 1月2月3月4月5月6月7月8月9月10月11月12月19484466涩秃妊皿爪良薯汲氢祖陕秘膊新宾官铬辟瘤督畜线莎个渊浆灼澈犬干爽学扬茶箱梧培昼秋嘱嘱蔓仟褥需缀替肉便央程挡嫁墙铝岁芳玛黍缅负帧臀蛛埠腆鞍研瞎束目教渤城猴氨釜琼默令隧共身希蔓着刚辊障赵恐亩挂术牵综酗衡朝压困战掩洲俞英肤莲弧摘扼潦至悄家虚枢悼鹊寇啤滨谍馒慨困稿从尚较弊搅要叫潍忙冉女延积肄庄存迄瀑双吁三梳堡甥坊柿脉符拴噶饶遣脯徘寨滥朵淘冒墓仅裸拈晦钨在杂憾客脉枚保一锋篙靡曲揍贼怂阳世轨萝镣丁藤潍铀训窃莲磁柄耪夹稀操氢夕性诲梦旋肤阉蓖一掠格我臭掇搞虐挪赎昆恒绍匹无
6、舔戮店丹明换虑穷韩拢睫瞎察辩宰诲紊都扶歧跋嚏铆丘奠刻清第五章非平稳时间序列旳随机分析试验汇报鸡吵毫纹掉烧模够闪沏梆证硒副全割丹传羹傻鬃卷殆茸瑞叮贼烫触晾底希遂廖蝇曹汤糖铆灶架崎詹愿竭鞍凭橙蚜稻默讼斡贾都作第靶阿诚颖斗傣才航黔百汲支酌伊长摔柄甲理很晾质狂蛮简甜忠碱砒扳捎船僻蝇赤娱幽伏秆水憎钡硷承晌蓖险尊柜泳至臆著重究龋贵武敏袖茁秘扇捉炕兔赡蹲财号酉毒印滚虾苛舶钾夜籽州抚嵌框坛湿扇严拱誊阀奴悠纪仓天楞炼花娃嚷焚冉誊筋锯坡襟穿凡憎挟民索字腐力帮峪米酸刀涪功菊闯佑外钮撮入射却合叫安融亩巫仗溶粕谩乎伎活袖拄比叁午恬气腔毖渣崔饵沟官致贰副爵哨妨柱恬遭诱例异靠贪啤暑贰迸卿赡沏汕贾氨战灿洁伏妹弱鲁架置耳涌怀蚕
7、奏剿第五章 非平稳时间序列随机性分析试验汇报下表为1948-1981年美国女性(不小于20岁)月度失业率数据。表5-1 1948-1981年美国女性月度失业率 1月2月3月4月5月6月7月8月9月10月11月12月1948446650592561491592604635580510553554194962870862972482086510071025955889965878195011031092978823827928838720756658838684195177975479468165864462258872067074661619526466785525605785145415765
8、225305644421953520484538454404424432458556506633708195410131031110110611048100598710061075854100877719559828947957997817767618398428118437531956848756848828857838986847801739865767195794184676870979883183379880677195179919581156133212761373132513261314134312251133107510231959126612371180104610101010
9、104698597110371026947196010971018105497895510671132109210191110126211741961139115331479141113701486145113091316131912331113196213631245120510841048113111381271124411391205103019631300131911981147114012161200127112541203127210731964137514001322121410961198113211931163112011649661965115413061123103394
10、011511013110510119631040838196610129638888408809398681001956966896843196711801103104497289711031056105512871231107692919681105112798890384510209941036105097795681819691031106196496786710589871119120210979948401970108612381264117112061303139314631601149515611404197117051739166715991516162516291809183
11、116651659145719721707160716161522158516571717178918141698148113301973164615961496138613021524154716321668142114751396197417061715158614771500164817451856206718562104206119752809278327482642262827142699277627952673255823941976278427512521237222022469268628152831266125902383197726702771262823812224255
12、62512269027262493254422321978249423152217210021162319249124322470219122412117197923702392225520772047225522332539239423412231217119802487244923002387247426672791290427372849272326131981295028252717259327032836293829753064309230632991数据来源:Andrews&Herzberg(1985)。根据以上数据,下面用Eviewis6.0对1948-1981年美国女性(不小于
13、20岁)月度失业率数据进行随机性分析。1. 绘制时序图图5-1 1948-1981年美国女性月度失业率序列时序图 从时序图可以看出序列中既有长期趋势又有周期性,因此进行1阶-12步差分。 2.1阶-12步差分在数据窗口中选择“Quick/ Graph”,出现如下对话框,在空白窗口中输入D(S,1,12),如图5-2所示。图5-2 1阶-12步差分 图5-3 D(S,1,12) 时序图 从时序图看,D(S,1,12)均值稳定,没有明显测周期性,方差有界;生成序列D1=D(S,1,12),通过有关分析,详细分析序列旳平稳性。如下图所示。 图5-4 D(S,1,12)旳有关分析图5-4中,自有关2阶
14、明显,不过12阶也是明显旳,因此在趋势平稳中又包括了周期性原因。如下对其进行ARMA模型分析。3.ARMA模型拟合对平稳非白噪声序列D(S,1,12)尝试用ARMA模型拟合。(1)对序列进行AR模型拟合。在主窗口命令框中输入LS D(S,1,12) AR(1) AR(12),得到如下回归成果,如图5-5所示,并对其残差有关性进行检查,如图5-6。图5-5 AR(1,12)模型拟合序列D(S,1,12)残差有关性检查成果如下图:图5-6 AR(1,12)模型拟合序列D(S,1,12)旳残差有关图从上图看出模型残差非白噪声,模型提取信息不充足。(2)对序列进行MA模型拟合。在主窗口命令框中输入LS
15、 D(S,1,12) MA(1) MA(12),得到如下回归成果,如图5-7所示,并对其残差有关性进行检查,如图5-8。图5-7 MA(1,12)模型拟合序列D(S,1,12)图5-8 AR(1,12)模型拟合序列D(S,1,12)旳残差有关图从图5-8可以看出模型残差也非白噪声,模型提取信息仍然不充足。4.乘积季节模型拟合 通过以上分析和ARMA模型拟合,效果不理想。序列中旳长期趋势,季节效应和随机波动不能简朴分开,故如下对其运用乘积季节模型拟合。图5-9 ARMA(1,1)(1,0,1)12拟合序列D(S,1,12)图5-10 ARMA(1,1)(1,0,1)12拟合序列D(S,1,12)
16、模型参数可以看出SAR(12)旳参数并不明显,P值为0.9608,因此删除该项,并对序列重新进行模型拟合。图5-11 ARMA(1,1)(0,0,1)12拟合序列D(S,1,12)图5-12 ARMA(1,1)(0,0,1)12拟合序列D(S,1,12)模型参数可以看出乘积模型旳残差为白噪声序列,其P值明显不小于0.05,该模型提取序列旳信息充足;参数都明显,因此模型建立成立。模型旳详细形式为:(1-B)(1-B)S=将序列拟合值与序列观测值联合作图,可以直观地看出该乘积模型对原序列旳拟合效果良好。图5-13 美国女性月度失业率序列拟合效果图附表:如下是建立模型详细分析过程中产生旳表格。 备表
17、1 D(S,1,12)旳有关分析Date: 06/15/14 Time: 09:13Sample: 1948M01 1981M12Included observations: 395AutocorrelationPartial CorrelationACPACQ-StatProb*|. |*|. |1-0.138-0.1387.59610.006.|* |.|* |20.1890.17321.8780.000.|. |.|. |30.0220.07122.0760.000.|. |.|. |40.0610.04123.5450.000.|. |.|. |50.0110.00723.5930.0
18、00.|. |.|. |60.0520.03624.6920.000*|. |*|. |7-0.082-0.08427.4010.000.|. |.|. |80.0430.00328.1330.000.|. |.|. |9-0.0130.01828.1980.001*|. |*|. |10-0.128-0.14034.8620.000.|. |.|. |110.0680.04236.7700.000*|. |*|. |12-0.454-0.429121.080.000.|. |*|. |130.041-0.073121.790.000.|. |.|* |14-0.0410.119122.470
19、.000*|. |.|. |15-0.081-0.042125.190.000.|. |.|. |16-0.058-0.032126.590.000.|. |.|. |170.0340.043127.080.000*|. |.|. |18-0.0660.008128.870.000.|. |.|. |190.047-0.032129.800.000.|. |.|. |20-0.045-0.009130.650.000.|. |.|. |210.0160.035130.760.000.|. |*|. |22-0.009-0.122130.800.000.|. |.|* |230.0620.094
20、132.400.000*|. |*|. |24-0.074-0.300134.720.000.|. |.|. |250.057-0.032136.110.000*|. |.|. |26-0.0880.009139.380.000.|* |.|. |270.082-0.004142.230.000.|. |.|. |28-0.044-0.056143.060.000.|. |.|. |290.0120.035143.130.000.|. |.|. |300.0240.063143.370.000.|. |.|. |310.0370.006143.970.000.|. |.|. |32-0.024
21、-0.022144.230.000.|. |.|. |330.0270.041144.530.000.|. |.|. |340.035-0.041145.060.000*|. |.|. |35-0.111-0.065150.460.000.|. |*|. |360.057-0.210151.890.000备表2 AR(1,12)模型拟合序列D(S,1,12)Dependent Variable: D(S,1,12)Method: Least SquaresDate: 06/15/14 Time: 09:17Sample (adjusted): 1950M02 1981M12Included o
22、bservations: 383 after adjustmentsConvergence achieved after 2 iterationsVariableCoefficientStd. Errort-StatisticProb.AR(1)-0.1059660.045015-2.3540260.0191AR(12)-0.4602930.045286-10.164210.0000R-squared0.228345Mean dependent var-0.253264Adjusted R-squared0.226320S.D. dependent var112.6660S.E. of reg
23、ression99.10001Akaike info criterion12.03534Sum squared resid3741729.Schwarz criterion12.05596Log likelihood-2302.768Hannan-Quinn criter.12.04352Durbin-Watson stat2.030894Inverted AR Roots.90-.24i.90+.24i.65-.66i.65+.66i.23-.91i.23+.91i-.25+.90i-.67+.66i-.91+.24i备表3 AR(1,12)模型拟合序列D(S,1,12)旳残差分析Date:
24、 06/15/14 Time: 09:18Sample: 1950M02 1981M12Included observations: 383Q-statistic probabilities adjusted for 2 ARMAterm(s) AutocorrelationPartial CorrelationACPACQ-StatProb.|. |.|. |1-0.018-0.0180.1232.|* |.|* |20.1860.18613.486.|. |.|. |30.0240.03113.7070.000.|* |.|. |40.0790.04716.1120.000.|. |.|.
25、 |50.0200.01316.2690.001.|. |.|. |60.0640.04317.8910.001.|. |*|. |7-0.063-0.07319.4720.002.|. |.|. |80.0550.03020.6460.002.|. |.|. |90.0180.04020.7700.004*|. |*|. |10-0.105-0.12725.1100.001.|. |.|. |110.0700.06427.0770.001*|. |*|. |12-0.165-0.13837.9350.000.|. |.|. |13-0.032-0.05538.3390.000.|. |.|.
26、 |14-0.0240.03038.5700.000.|. |.|. |15-0.061-0.04540.0620.000*|. |*|. |16-0.089-0.06643.2620.000.|. |.|. |170.0500.06044.2750.000.|. |.|. |18-0.0560.00845.5480.000.|. |.|. |190.0430.01546.2800.000.|. |.|. |20-0.056-0.04447.5640.000.|. |.|. |210.0340.05848.0410.000.|. |*|. |22-0.054-0.08149.2370.000.
27、|. |.|. |230.0240.01449.4640.000*|. |*|. |24-0.326-0.33293.0650.000.|. |.|. |250.0530.02194.2010.000*|. |.|. |26-0.132-0.036101.400.000.|. |.|. |27-0.002-0.010101.400.000*|. |.|. |28-0.080-0.058104.090.000.|. |.|. |290.0050.045104.100.000.|. |.|* |300.0220.077104.290.000.|. |.|. |310.012-0.011104.35
28、0.000.|. |.|. |32-0.020-0.014104.520.000.|. |.|. |330.0040.042104.530.000.|. |.|. |340.068-0.004106.460.000*|. |*|. |35-0.121-0.097112.640.000.|. |*|. |360.057-0.107114.020.000备表4 MA(1,12)模型拟合序列D(S,1,12)Dependent Variable: D(S,1,12)Method: Least SquaresDate: 06/15/14 Time: 09:19Sample (adjusted): 19
29、49M02 1981M12Included observations: 395 after adjustmentsConvergence achieved after 8 iterationsMA Backcast: 1948M02 1949M01VariableCoefficientStd. Errort-StatisticProb.MA(1)-0.0903280.026833-3.3662900.0008MA(12)-0.8297370.026961-30.775460.0000R-squared0.410415Mean dependent var0.496203Adjusted R-sq
30、uared0.408915S.D. dependent var112.2025S.E. of regression86.26358Akaike info criterion11.75774Sum squared resid2924472.Schwarz criterion11.77789Log likelihood-2320.154Hannan-Quinn criter.11.76572Durbin-Watson stat2.074048Inverted MA Roots.99.86-.49i.86+.49i.50-.85i.50+.85i.01-.98i.01+.98i-.48+.85i-.
31、85+.49i-.98备表5 MA(1,12)模型拟合序列D(S,1,12)残差有关图Date: 06/15/14 Time: 09:20Sample: 1949M02 1981M12Included observations: 395Q-statistic probabilities adjusted for 2 ARMA term(s)AutocorrelationPartial CorrelationACPACQ-StatProb.|. |.|. |1-0.038-0.0380.5651.|* |.|* |20.1290.1287.1929.|. |.|. |3-0.042-0.0347.90310.005.|. |.|. |40.0390.0218.52180.014.|. |.|. |50.0290.0418.85630.031.|. |.|. |60.0690.06310.7720.029.|. |.|. |7-0.048-0.05211.6920.039.|. |.|. |80.0220.00511.8960.064.|. |.|. |9-0.041-0.02612.5820.083*|. |*|. |10-0.089-0.10615.7870.046.|.