Scribd
Upload
35 views12 pages

Problem Set 3 Solutions

Uploaded by

giz
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
35 views12 pages

Problem Set 3 Solutions

Uploaded by

giz
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
METU Department of Economics 2020 Fall ECON 302 Instructor: Erdal Ozmen T.A.: Abdullah Gillett S.A.: Gillin Demirer PROBLEM SET 3: AUTOCORRELATION QUESTION 1: In Productivity Trends in the United States: Trends and Cycles, John Kendrick and Elliot Grossman construct index number series for real output, Y, real capital input, K, and real labor input, L, for the private business sector of the US for the period 1948-1978 (annual data). A researcher hypothesis that Y is related to K and L by a Cobb-Douglas function Y=AK*LheTy where T is a time trend (T=1, for 1948, 2 for 1949, etc) and v is a disturbance term. He fits the following equations (standard errors in parenthesis; SSR = residual sum of squares; d= Durbin- Watson statistics; C-O: Cochrane-Orcutt iterative technique; the subscript —1 indicates that the variable is lagged one period): () OLs lo8 ¥ = 1.21 + 0.32 log K + 0.30 log L + 0.026 T (2.13) (0.64) 0.22) (0.015) R? = 0.992, SSR = 0.0267, d = 0.64 Q) co log ¥ =-1.29 + 0.25 log K + 0.92 log L + 0.022 T (2.37) (0.68) 0.17)_—(0.016) R? = 0.997, SSR = 0.0091, d= 1.71 @) OLS lo Y = 1.84 + 1.23 log K + 0.70 log L + 0.010 T (0.88) (0.59) , (0.15) (0.007) + 0.69 log Y.1 — 1.17 log K-1 — 0.89 log Ls (0.09) (0.52) (0.15) R? = 0.999, SSR = 0.0034, d= 2.16 a, Show that regression (1) is correctly specified mathematically, given the hypothesized form of production function. , b. What assumptions are being made concerning the disturbance term in regression (1)? ¢. Give two reasons for believing that regression (1) is unsatisfactory. d. Explain the theory behind the Cochrane-Orcutt iterative technique, stating clearly the assumptions involving. e. Does Tegression (2) appear to be more satisfactory than the first? (Give reasons.) f. Explain why the researcher ran regression (3). Perform a common factor test, CamScanner ile tarandt > tT oR Lev 90 Teh ipovetes a) Coby-Daglas prod pres 4 a, take leg, oy pots sides, lagde= lagh + alogltet Bloglet Clog Te + Ve { F > constert tem ‘regression (UL), st is legarithence prod. ac end) carvectts tohach Ts Specipied - © Assunchors for olsturboices 4 a wt 2 VorlugheB ude cewter © Test for existonce 4 futocorre latte a j BeObl - Leg () is stetre model with fatercept so we om vse Dio tear statics: for 230 al E=3 md, 24214 dus {690 Smce 04 [Link] dp 24,214 _ptweni ae rey). Tp 3 kepok. tHe regression, both & snd |F stedtostres ove uacelrable | becavse € od F atutishies o@ cerised va (one HC od Arc. Mene2, We connat test He sigaiprons cy of cogppraects “t, a) gt : © CamScanner ile tarandt | | | | @ Cochrate- arat+ Heat’? Techara vy J)log Yee B+ a lglterplag ler They Tet VE with Mex p Meni t & FP eotimate lag} and 0 ~ a) given welt) : Ve sore Of the coepet cqeds are tetas cally vnsigaigteent Me ft dot tHe form 4 outocarrelatten right he Ke mey be Jue te omission is) Causes ca 2 eotinale He gt regression. £) since feseardar SUtPe neorrectly specdgted- that of relevert olyraniics Hace, » COMPAS Test Unrestricted Model eghet Igleg Tet Ay Igoe + Ye legke- +p leglert ue 2 + Deed from mode! Ls, bjggde Pat lagers! whore Ne= per tee Pestricted Hadel -» Model cas) lage = B+ dlegtet Beg Let Tlag Te #Y% leg key HP Blogger pt leg Tet? Ver plate: =ph +p lege -plag r= Wlkp)+ wlegke-palegker + ploye -ppleyle-1 t Thog ZiT leg Tepj ee ME = VER o% r Ve-! oad & rte fea Cem wr © CamScanner ile tarandi | =Ploy Me = 8C-p) + pleg Tet + alegky -galey kert plegLe - plagle-1 | *TlegTe - P TlegTe-1 + et =p Resttcteol ode) Coagprenedls of Ki jka oot 1) 2" nallicar Ta pareneter , we camot use F test Gut we can vse Compac test. Number of rest itio ty equal to at af explonatery voriables, Hee we have Say explenshery vertables ord restrictions ares VQ »Model log Vem Aor mlgterAalegle ‘ alg Ter AylagYaer +Aslag ke + elegheit ue Hot Mee -pad =-Ay Ae y= s/ar 2 oy Ay eo ph= Aue ae -Ay =- e/a X= LOL dg 20 =p To -2u Ag y= AAlage O S5L. 2.009! ) = 5,26 old . Couses ect Yo, So te resptetrer nat ¥ 2 @ Compac 7X3 3 He regression . > omtkd aly nomics m op ke | | | CamScanner ile tarandt QUESTION 2: A researcher has annual data on aggregate consumer expenditure on financial intermediaries, F , aggregate disposable income, X; , and rel price index for consumer expenditure on financial intermediaries, P; , for the United States for the period 1959-2003 and fits the following logarithmic regressions. 2:(01S) 2: (AR(1)) 3:(O1s) 4:(0Ls) In 7.40 + 1.56 InX,~0.26 InPy Ri=0.984 SSR=0.317 (0.77) (0.04) (0.21) d=065 BG =20. Ink 8.53 + 1.51 InX,+0,07 InP, R?=0,991 SSR=0.164 (1.04) (0.08) (0.21) 6=0.7 (0.11) 87 InFe = 1.68 + 0.04 InX_ +003 InP, + 0.7 InFus +.0.43 InXix—0.22 InP (1.02) (0.6) (0.2) (.41) (0.61) " (0.2) RP=0,993 SSR=0.138 d=168 8G=2.43 InFr =~ 1.8 +0.42 InX~0.1 InP +0.73 InFes R's 0.992 SSR=0.144 (1.01) (0.17) (0.15) (0.11) d=178 8G=083 a, Test the existance of serial correlation for the first regression. b. Show that regression 2 is the autocorrelation correction model for the first regression, ¢. Explain the relationship between the second and the third regressions, perform a common factor test. d. Given the all information available, choose the best model. CamScanner ile tarandt Qr) Tels @) LiFe 2 forBr IaXee Bala Pe + Ute Test por AC 5) Apply DIV tet chee city oleic model with ‘nbercerst tera 25 and ee dps 143 fer Tee duc tbe eG Oo oe eu fee ee > 4© regression W)Apply BG LM best; 4 AC de = At Jp Late tAglter 7 ter VE Hotoe ° 3No AC xy 28.84 48 Hrpto 2 FAC Greed 1 6 (mew) = 20-42 here ty AC peoblen b)dregression LY tte oper ple + palace ue where Uses ae } - eA } Wyant y apply o2d ard obtein PoE pouch Hat ii) multiply tHe previows puted op foch Megrossion “si plafinr= f por f pbken afplrtins Pres () subtract (2) £0" ) = polt-p) +P! + ay larer Ve A a save fboteae pale PbsaPet ef Ve WN dake-f daFe-1 with fe o-t ~ Dnt = P0* Ap axe ee) Dw tor gfatistres- due tsb2 Ld etdt 4 y-du Gro ke 14 seqondh ragrere (2: © CamScanner ile tarandt | ©) Regression 2 is tronpormed model ey assum prot while Regression 3 is vnreshfchecl medel Mrclediog all relevant slyonics, Reg & VR Model Ug Be Xe eile hy ME enPan # ee i . Model eg ap Rast, Ne oe bog, vpole pie pter ple per SP ga hee - a, gerenelet nonlinear in Pe og rests comFAc text = a exploratory coribes > Were 03 te UG, =-9 Pe byt > Gye - 4 [% Boece par 8% —> Ns = - 5 | Es oAbu . 48,3t Thy (4) ed Xe) U5) lag (ae Com AC = + gir rescron 7S 0.0488 ene) coeppiciet op T Corre ttn SSR op 9 . = ' ela i reson 1S (44) ey nF in re) Vv yecctten Imposed 18 1 Resock ee Acts on thea eynamics 1 © CamScanner ile tarandt eco ©) cree te bast matt Jive chou dat teres RO MF Choose pins Ore J gd oS regresston iR secora reg: vost regresster go we connot Wie compere 2" by Comyenc text, we shows a lie we che the resherctioy Fuso ved mate ol “6 Core 8 pearesston 03; LeoAll sbaoate -Mo Ke problen since vorubles are Sneludedd . 2 Lucaeur) = 2-43 A —tad F teshare at reliable > lnPer ore indssiduall x 5 4s of InXet oe ne seer (au 7 3 crit 2|2.02| nspnigteatt 7 bay ay 2|e4h- old boas, 39 -| 22\- coon. me * modtun 14 resoter 4 Diet X3 Paar Uae a ket Ae -Wnfe=%e . : urbin oh Fes Test for pcp Dur CE us ct d- +) Ls ot ) a- yly (0.1) 1 held SE) Fools) Dench tay. ie es ne Laan) = 0-F3 Z Xe 384 PDorot Rey Ho? c ical mrpeant, Heree , er0ese Ml coegpicieats oe Gtatiotica lly apse , ie vest @) CamScanner ile tarandt QUESTION 3: A researcher has annual data on I, investment as a percentage of gross domestic product, and r, the real long-term rate of interest for a certain economy for the period 1973- 2002. He regresses I on r, (1) using ordinary least squares (OLS), (2) using an estimator appropriate for AR(1) residual autocorrelation, and (3) using OLS but adding I(-1) and r(-1) (1 and r lagged one time period) as explanatory variables. The results are shown in columns (1), (2) and (3) of the table. He then obtains annual data on g, the rate of growth of gross domestic product of the economy, for the same period, and repeats the regressions, adding g (and where appropriate, g(-1)) to the specifications as an explanatory variable. The results are shown in columns (4), (5) and (6) of the table. R and g are measured as percent per year. a) Q) GB) a) G) (6) : -0.87 0.83 -0.87 -1.81 -1.88 171 (0.98) | .05) | a.08) | 49) 40.50) __| (0.52) I) - - 0.37 7 - 0.22 (0.16) (0.18) 1D > - 0.64 z = -0.98 (1.08) (0.64) g - > - 1.61 1.61 1,92 (0.17) __| (0.18) (0.20) EC) > : > - - -0.02 (0.33) 6 . 037 > - -0.16 - (0.18) (0.20) constant [9.37 921 412 926 954 13.24 G6) 16.90) a4) La7) _| a6) (2.69) d 0.99 136 133 219 2.05 2.09 SSR 1205 103.9 103.5 274 268 23.5 Note: Standard errors are given in parentheses, p is the estimate of the autocorrelation parameter in the AR(1) specification. D is the Durbin-Watson d statistic. SSR is the sum of square residuals. plain why the researcher was not satisfied with regression (1). aluate regression (2). Explain why the coefficients of I(-1) and r(-1) are not reported, despite the fact that they are part of the regression specification, ¢. Evaluate regressions (3), (4), (5) and (6). 4, Summarize your conclusions concerning the evaluation of the different regressions. CamScanner ile tarandt ; 83 T=80 a} (Regt) Se= 9.81 BC oe Use Dw test shee Tt Okt —p TER Mor tie t ek ay gtette medel: Test gor for 280 end vcd 9 oe ette dus 1429 Strce ofd-00949, > There wy DAC 4 ragrossion i. 3s trons formed pode) obtain by pirsh regressiar b) Regression \2) airturbence Tom a Us =P deep tVe sith AR) process Y - Actually the exact ropreseatetion og ad et 2 regres then 73) Berga, Wey Wiggs Spe, (Reg 2) >See Soe Uircrve = vi ie a & , . - Woe porlines ity 1 poreeneters apty Arenopocrration j Z ond a Vortobls are renamed gad. e183 rm) ) 3) Ze= fot iter Pa Tent Baler t Mae 7 purlin h 2(4- 2) (Fa is) 2a. ez y teGb 2 Peg: bare (een u) dy Gd2er19 £y-du ae A and Ar are andatdvelly agen Tee Por Meet Mee + Une pc. 4 by t-test. (2) CamScanner ile tarandt (%g 3) Tee C0 rae + % Oe t We whee Use =P Use-1 + by Ths EGLS form Of yh regression. Since thee is no Wc ath reg.) Here iS no need to Fun oth regression Also; AC. ploblerr in ) ns atetivttcall trsppatpreert od d=207 ae ge (Lex ressien b) Ten bot bet batent leit Png AC =e Show it by oueelp are Sndividually insignificant. et bs Kit Ue by Durbn-h test ro Coopgrcierts Of Ja,hy, at gr a steve voriadle "9" apy 1S Omigaian 6) Couses 4 crybocore lotion a Cet ; all coeggiciedts rs 3? ragessien ey totic model, y" eolues AC) Locludig tase 6 reayesoion Lest model onong G2) CamScanner ile tarandt

You might also like