47 views

Uploaded by 0796105632

- Chapter5 Solutions
- chapter08 part 4
- Full Text
- Output yy
- 1259-3800-1-PB
- chapter08 part 5
- De Beer, y Otros, 2004
- final report
- Anova Reg Summary
- PP
- Arg02_Quantitative Tools and Excel
- Fooled
- 2014FA PS4 Solution
- Multiple_Regression_with_SPSS
- Answers Review Questions Econometrics
- SMChap003.docx
- Central Composite Design for the Optimization of r Absorbance
- molecules-17-03618
- Regression.ppt
- Applied Econ Slides 2010

You are on page 1of 5

It can be proved that a t-distribution is just a special case of the more general F-distribution. The square of a t-distribution with T-k degrees of freedom will be identical to an F-distribution with (1,T-k) degrees of freedom. ut remember that if we use a !" si#e of test, we will loo$ up a !" value for the F-distribution because the test is %-sided even though we onl& loo$ in one tail of the distribution. 'e loo$ up a %.!" value for the t-distribution since the test is %-tailed. ()amples at the !" level from tables T -k %* +* /* 1%* F critical value +.,! +.*. +.** ,.-% t critical value %.*%.*% %.** 1.-.

%. (a) 0* 1 , 2 % 'e could use an F- or a t- test for this one since it is a single h&pothesis involving onl& one coefficient. 'e would probabl& in practice use a t-test since it is computationall& simpler and we onl& have to estimate one regression. There is one restriction. (b) 0* 1 , 3 + 2 1 4ince this involves more than one coefficient, we should use an F-test. There is one restriction. (c) 0* 1 , 3 + 2 1 and ! 2 1 4ince we are testing more than one h&pothesis simultaneousl&, we would use an F-test. There are % restrictions. (d) 0* 1 % 2* and , 2 * and + 2 * and ! 2 * 5s for (c), we are testing multiple h&potheses so we cannot use a t-test. 'e have + restrictions. (e) 0* 1 %, 2 1 5lthough there is onl& one restriction, it is a multiplicative restriction. 'e therefore cannot use a t-test or an F-test to test it. In fact we cannot test it at all using the methodolog& that has been e)amined in this chapter. ,. T0( regression F-statistic would be given b& the test statistic associated with h&pothesis iv) above. 'e are alwa&s interested in testing this h&pothesis since it tests whether all of the coefficients in the regression (e)cept the constant) are jointl& insignificant. If the& are then we have a completel& useless regression, where none of the variables that we have said influence y actuall& do. 4o we would need to go bac$ to the drawing board6

1/5

The alternative h&pothesis is1 01 1 % * or , * or + * or ! * 7ote the form of the alternative h&pothesis1 8or9 indicates that onl& one of the components of the null h&pothesis would have to be rejected for us to reject the null h&pothesis as a whole. +. The restricted residual sum of squares will alwa&s be at least as big as the unrestricted residual sum of squares i.e. ::44 ;:44 To see this, thin$ about what we were doing when we determined what the regression parameters should be1 we chose the values that minimised the residual sum of squares. 'e said that <=4 would provide the 8best9 parameter values given the actual sample data. 7ow when we impose some restrictions on the model, so that the& cannot all be freel& determined, then the model should not fit as well as it did before. 0ence the residual sum of squares must be higher once we have imposed the restrictions> otherwise, the parameter values that <=4 chose originall& without the restrictions could not be the best. In the e)treme case (ver& unli$el& in practice), the two sets of residual sum of squares could be identical if the restrictions were alread& present in the data, so that imposing them on the model would &ield no penalt& in terms of loss of fit. !. The null h&pothesis is1 0* 1 , 3 + 2 1 and ! 2 1 The first step is to impose this on the regression model1 yt = 1 + %x%t + ,x,t + +x+t + !x!t + ut subject to , 3 + 2 1 and ! 2 1. 'e can rewrite the first part of the restriction as + 2 1 - , Then rewrite the regression with the restriction imposed yt = 1 + %x%t + ,x,t 3 (1-,)x+t + x!t + ut which can be re-written yt = 1 + %x%t + ,x,t + x+t - ,x+t + x!t + ut and rearranging (yt x+t x!t ) = 1 + %x%t + ,x,t - ,x+t + ut

2/5

(yt x+t x!t) = 1 + %x%t + ,(x,t x+t)+ ut 7ow create two new variables, call them Pt and Qt1 pt 2 (yt - x,t - x+t) qt 2 (x%t -x,t) 'e can then run the linear regression1 pt = 1 + %x%t + ,qt+ ut , which constitutes the restricted regression model. The test statistic is calculated as ((::44-;:44)?;:44)@( T-k)?m In this case, m2%, T2-/, k2! so the test statistic 2 !.A*+. Bompare this to an F-distribution with (%,-1) degrees of freedom, which is appro)imatel& ,.1*. 0ence we reject the null h&pothesis that the restrictions are valid. 'e cannot impose these restrictions on the data without a substantial increase in the residual sum of squares. /. ri = *.*.* + *..*1Si + *.,%1MBi + *.1/+PEi - *.*.+BETAi (*.*/+) (*.1+A) (*.1,/) (*.+%*) (*.1%*) 1.25 5.45 2. ! ". #" -".$""

The t-ratios are given in the final row above, and are in italics. The& are calculated b& dividing the coefficient estimate b& its standard error. The relevant value from the t-tables is for a %-sided test with !" rejection overall. T-k 2 1-!> t%rit 2 1.-A. The null h&pothesis is rejected at the !" level if the absolute value of the test statistic is greater than the critical value. 'e would conclude based on this evidence that onl& firm si#e and mar$et to boo$ value have a significant effect on stoc$ returns. If a stoc$Cs beta increases from 1 to 1.%, then we would e)pect the return on the stoc$ to D5== b& (1.%-1)@*.*.+ 2 *.*1/. 2 1./." This is not the sign we would have e)pected on beta, since beta would be e)pected to be positivel& related to return, since investors would require higher returns as compensation for bearing higher mar$et ris$. A. 'e would thus consider deleting the price?earnings and beta variables from the regression since these are not significant in the regression - i.e. the& are not helping much to e)plain variations in y. 'e would not delete the constant term from the regression even though it is insignificant since there are good statistical reasons for its inclusion.

y t = 1 + 2 x 2t + ! x !t + y t 1 + u t y t = 1 + 2 x 2t + ! x !t + y t 1 + v t .

!/5

7ote that we have not changed an&thing substantial between these models in the sense that the second model is just a re-parameterisation (rearrangement) of the first, where we have subtracted yt-1 from both sides of the equation. (a) :emember that the residual sum of squares is the sum of each of the squared residuals. 4o lets consider what the residuals will be in each case. Dor the first model in the level of y

" " x " X " y "t = yt y " t = yt u 1 2 2t ! !t t 1

7ow for the second model, the dependent variable is now the change in y1

"t = y t y " t = y t "1 " 2 x 2t " ! x !t " y t 1 v

is the fitted value in each case (note that we do not need at this stage where y to assume the& are the same). :earranging this second model would give

If we compare this formulation with the one we calculated for the first model, we can see that the residuals are e)actl& the same for the two models, with " = " = " i (i 2 1, %, ,). 0ence if the residuals are the same, the " + 1 and i residual sum of squares must also be the same. In fact the two models are reall& identical, since one is just a rearrangement of the other. (b) 5s for &%, recall how we calculate &%1

R 2 =1 RSS for the first model and $ yi y # 2

RSS $y i y # 2 in the second case. Therefore since the total sum of

R 2 =1

squares (the denominator) has changed, then the value of &% must have also changed as a consequence of changing the dependent variable. (c) & the same logic, since the value of the adjusted &% is just an algebraic modification of &% itself, the value of the adjusted &% must also change. .. 5 researcher estimates the following two econometric models

y t = 1 + 2 x 2 t + ! x !t + u t

y t = 1 + 2 x 2 t + ! x !t + x t + v t

(a) The value of &% will almost alwa&s be higher for the second model since it has another variable added to the regression. The value of &% would onl& be

/5

identical for the two models in the ver&, ver& unli$el& event that the estimated coefficient on the x+t variable was e)actl& #ero. <therwise, the &% must be higher for the second model than the first. (b) The value of the adjusted &% could fall as we add another variable. The reason for this is that the adjusted version of &% has a correction for the loss of degrees of freedom associated with adding another regressor into a regression. This implies a penalt& term, so that the value of the adjusted &% will onl& rise if the increase in this penalt& is more than outweighed b& the rise in the value of &%. 11. &% ma& be defined in various wa&s, but the most common is

R2 = ESS TSS

4ince both ESS and TSS will have units of the square of the dependent variable, the units will cancel out and hence &% will be unit-free6

5/5

- Chapter5 SolutionsUploaded by0796105632
- chapter08 part 4Uploaded byapi-232613595
- Full TextUploaded byzgurocak
- Output yyUploaded byHira Damayanti
- 1259-3800-1-PBUploaded byAbdul Sattar Ansari
- chapter08 part 5Uploaded byapi-232613595
- De Beer, y Otros, 2004Uploaded byLeonardo Giraldo
- final reportUploaded byapi-465329396
- Anova Reg SummaryUploaded byMuhammad Imran
- PPUploaded bySaksham Agrawal
- Arg02_Quantitative Tools and ExcelUploaded byAndrea Franzini
- FooledUploaded byAnonymous zTqxnCKrl
- 2014FA PS4 SolutionUploaded bykelerong
- Multiple_Regression_with_SPSSUploaded byapi-19973711
- Answers Review Questions EconometricsUploaded byZX Lee
- SMChap003.docxUploaded byMohamedBanoon
- Central Composite Design for the Optimization of r AbsorbanceUploaded byAsif Fareed Qaisrani
- molecules-17-03618Uploaded byAhmad Al Naser
- Regression.pptUploaded byNivedita Nanda
- Applied Econ Slides 2010Uploaded byAlice Sapcaliu
- RA Sproj Airline Operating CostsUploaded byPrasad Marathe
- Analyzing GRT Data in StataUploaded byAnonymous EAineTiz
- mtk zia.docxUploaded bySamir Ridho
- Central Composite DesignUploaded byuse8716
- 2008 stat exam.pdfUploaded byElle Smart
- LAMPIRAN.docxUploaded byAjeng Larasati
- Answer 14Uploaded byGabriella Goodfield
- researchpaperii-100108075334-phpapp02Uploaded byMohamad Madkour
- full_eg (1).pdfUploaded bySana Javaid
- 1-s2Uploaded bymelissastgo

- Laurence Connors - ConnorsRSIUploaded byJoséRamón
- gretlguide.pdfUploaded by0796105632
- CFA Career GuideUploaded bylegolas55569
- Webtrader1.pdfUploaded by0796105632
- Garch Arima ModellingUploaded by0796105632
- Introduction_to_Financial_Markets.pdfUploaded by0796105632
- p 0141 July 2014Uploaded by0796105632
- Satrix ETF Form 2 Additional Investments (Lump Sums Only) Aug 2013 - Listing DocumentsUploaded by0796105632
- Value vs Growth Stock in Emerging Market-submittedUploaded by0796105632
- An Example in RugarchUploaded byGeorge MSenoir
- Business Plan GuidelinesUploaded bySibusiso V Mkhize
- Forbes USA - 24 January 2017Uploaded byCarlos Panao
- Laurence Connors - Bollinger Bands Trading Strategies That WorkUploaded bysamgans
- Laurence Connors - Bollinger Bands Trading Strategies That WorkUploaded bysamgans
- EuroStoxx50_Multiplier_ Presentation_Slides.pdfUploaded by0796105632
- Chapter7 SolutionsUploaded by0796105632
- New Reading 360Uploaded by0796105632
- Chapter12 SolutionsUploaded by0796105632
- alagidedeUploaded by0796105632
- Equinoxe - How to Launch a Hedge FundUploaded by0796105632
- Chapter10 SolutionsUploaded by0796105632
- Chapter8 SolutionsUploaded by0796105632
- Chapter4 SolutionsUploaded by0796105632
- Chapter11 SolutionsUploaded by0796105632
- Chapter9 Solutions (1)Uploaded byjhanggawan
- Chapter6 SolutionsUploaded by0796105632
- Chapter2 SolutionsUploaded byalantsang2
- Satrix ETF Form 1 New Business Application Aug 2013 - Listing Documents (1)Uploaded by0796105632

- Chapter 6Uploaded byKhairul Nidzam
- Stata Lecture3Uploaded byApam Benjamin
- Solutions for Chapter 8Uploaded byreloadedmemory
- Abridged CA Zambia Syllabus July 2017Uploaded byAugastusBwalyaLombe
- ACCA Relevance of accounting research 2010Uploaded byElzi Rahma
- Jurnal Perbandingan K-Means Clustering dan Spectral ClusteringUploaded byKristi Damayanti
- Download Assume That the Data Has a Normal DistributionUploaded bySolutionZIP
- Exploratory Data Analysis-Engineering Statistics Handbook NIST 2002Uploaded byAlbertoKrawczyk
- Measures of RelationshipUploaded byHarvin Bob Puno
- Program Integrity White PaperUploaded byGovtfraudlawyer
- edid6507 group assigment caseUploaded byapi-458205480
- Quiz Data MiningUploaded byVenkat Dot
- Multiple range MCQs on Introductory statisticsUploaded byKenny
- Stats Training Quiz 06-24-2013Uploaded bySunil Chadha
- statistics sample problemUploaded bylucas
- MANOVA-dgerUploaded byHoang Nguyen
- Regression Analysis and Forecasting ModelsUploaded bysooraj
- chapter2 Business Research MethodsUploaded byZubair1326
- MONOVAUploaded byAvishkar Jain
- measures of variationUploaded bySifat Mahmud
- 37 1497079367_10-06-2017.pdfUploaded byEditor IJRITCC
- Survival Analysis Using SPSSUploaded by786bismillah
- Inventory Management ProposalUploaded byIbrahim Ssenkubuge
- Project Final-simon Francis Koome Doc_2Uploaded byJohn Munene Karoki
- Mining Class ComparisonsUploaded bymurali_20c
- The Role of Total Management in Improving Product Acceptability of Consumers (a Case Study of Nigerian Bottling Company (Nbc) Owerri Plant)Uploaded byKanu Augustine Oluebubechi
- chin 2003Uploaded byBet9112003
- SPSS Introduction Course at PSB, UUMUploaded byRusnifaezah Musa
- How to Perform a One-Way ANOVA in SPSSUploaded bySwamy Rakesh
- Log Rank TestUploaded byDivyashri Jain