Apply Sascs and Economercs n Fnancal Research Obj. of Sudy & Hypoheses Tesng From framework objecves of sudy are needed o clarfy, hen, n research mehodology he hypoheses esng are saed, ncludng esng mehods. Secon 1 Secon Secon Secon 4 Secon 5 Daa Collecon Defnons of Varables - Concepualze vs Operaonalze Sample Selecon Crera Source of Daa Conssency of Daa Obj. L. Rev. Framework Hypoheses Mehodology - Daa H : Tesng. Resuls - Daa - Prelm. H : Tesng. Concluson. By Tare Janarakolca 1 By Tare Janarakolca Types of Varables Descrbng Daa or Sample Nomnal Level Ordnal Level Inerval Level Rao Level Measuremen Problem Ordnal Level vs Inerval Level Unvarae Sascal Analyss - Frequency Table, Graph, Char - Mean, Medan, Mode - Max, Mn, Range, Varance, SD., CV. - Skewness, Kuross Subsample Analyss By dvdng sample based on some ceran creron, subsample analyses can lead o a more clear undersandng of he sample group. By Tare Janarakolca By Tare Janarakolca 4
Apply Sascs & Economercs Mehods Objecves of Sudy & Hypoheses Tesng. Hypoheses Tesng - Unvarae and Bvarae Hypohess Tesng usng Basc Sascs - Mulvarae Hypohess Tesng usng Economercs Hypoheses Tesng Unvarae & Bvarae Hypoheses Tesng - Paramerc Tess - Nonparamerc Tess Mulvarae Hypoheses Tesng usng Economerc Technque - Indvdual Tes -es - Overall Tes F-es Resrced vs Unresrced Tes - Dummy Varables Tes - Specfc Tes By Tare Janarakolca 5 By Tare Janarakolca 6 Unvarae & Bvarae Hypoheses Tesng Paramerc Tess Unvarae Hypohess Tes One-sample -es Bvarae Hypohess Tes Two-sample Tes - Independen Sample -es - Dependen (Pared) -es One-way Analyss of Varance (ANOVA) Pearson s Correlaon Tes Independen Varable Measuremen Dependen Varables Measuremen - - Reurn Inerval or Rao Dvdend Pad Groups Before-Afer Groups Frm Sze > Groups Rsk Unvarae & Bvarae Hypoheses Tesng Nomnal Independen Nomnal Dependen Reurn Reurn Inerval or Rao Inerval or Rao Nomnal Reurn Inerval or Rao Inerval or Rao Reurn Inerval or Rao Sascal Tesng One-Sample -es Independen- Sample -es Dependen Pared -es One-way ANOVA Pearson s Correlaon By Tare Janarakolca 7 By Tare Janarakolca 8
Unvarae & Bvarae Hypoheses Tesng Nonnormal or Small Sample Nonparamerc Tess More appropraed for nonnormal dsrbuon daa or small sample case. Nomnal Daa Frequency - Conngency Table Analyss Ch-squared Tes Ordnal Daa Rank Dependen Samples - Sgn Tes & Wlcoxon Sgned Rank Tes Independen Samples - Wlcoxon Mann-Whney Rank-Sum Tes - Kruskal-Walls Tes By Tare Janarakolca 9 Independen Varable Unvarae & Bvarae Hypoheses Tesng Nonnormal or Small Sample Measuremen Dependen Varables - - Rang, Rankng Reurn Dvdend Pad Groups Before-Afer Groups Frm Sze > Groups Dvdend Pad, Frm Sze Rang, Rankng Nomnal Independen Nomnal Dependen Nomnal Nomnal Ordnal Lqudy Rao, Reurn Lqudy Rao, Reurn Lqudy Rao, Reurn Frm Sze, Indusry Rang, Rankng Measuremen Ordnal o Rao Inerval o Rao Inerval o Rao Inerval o Rao Nomnal Ordnal Sascal Tesng Sgn & Rank Tes Wlcoxon Tes Sgn & Rank Tes Kruskal-Walls Tes Ch-square es Spearman s Correlaon 1 Mulvarae Hypohess Tesng usng Economerc Technque Tradonal Lnear Regresson Model - Overall Tes F-es - Indvdual Tes -es - Tes for Equaly Resrcon - Resrced Regresson Tes F-es - Tes for Sably (Srucural Break) - Dummy Varable Technque.. Mcroeconomercs Models. Tme Seres Models By Tare Janarakolca 11 Hypohess Tesng Basc Tess e.g. Deermnans of frms performances = β1 + β + β + β4 4 Selec he Mos Appropraed Model Overall Tes (or F-es) H β β β4 Volaon of OLS Assumpon ncludes Mulcollneary, Auocorrelaon, : = = = Heeroscascy, Model Specfcaon, Endogeney problem Robusness of he Tess.. Tes Sgnfcan Impac of Each Varable Indvdual Tes H : β = By Tare Janarakolca 1
Hypohess Tesng Specfc Tes on Ceran Condon e.g. Equaly of nfluences of neres rae and nflaon rae = β1 + β + β + β4 4 Tes for Equaly Resrcon H : β = β or ( β β ) = 4 4 e.g. Economy of Scale H : β + β = 1 Y = β e 1 β ln β = ln β1 + β ln + β ln Tes for Equaly Resrcon By Tare Janarakolca u 1 Hypohess Tesng Tes for Sably Chow Tes Whole PerodY = λ + λ 11 + λ + u for =1,,,n 1 +n Before Crss Y = + 1 1 + + u1 for =1,,,n 1 Afer Crss Y = β + β11 + β + u for = n 1 +1, n 1 +,, n 1 +n Hypohess H H : a α = β = λ and α1 = β1 = λ1 and α = β = λ : Oherwse F = ( S1 S S ) k ( S + S ) ( n + n k ) By Tare Janarakolca 14 1 Hypohess Tesng Tes for Sably Dummy Varables Technque Model wh Inercep and Slope Dummy Varable = β D + β11 1D 1 + β D where: D = before crss = 1 afer crss. Ths model can be nerpreed as: Before Crss: Afer Crss: = β + β11 + β = ( β ) + ( β1 1) 1 + ( β ) By Tare Janarakolca 15 Dummy Varable Alernave o Chow Tes Chow Tes Whole Perod Y = λ + λ11 + λ + u for =1,,,n 1 +n Before Crss Y = α + α11 + α + u1 for =1,,,n 1 Afer Crss Y = β + β 11 + β + u for = n 1 +1, n 1 +,, n 1 +n Dummy Varable Technque Whole Perod Before Crss Afer Crss = β + β11 + β = β + β11 + β = ( β ) + ( β1 1) 1 + ( β ) Dummy varable can be used as Chow Tes. Resrced F-es H: γ = γ1 = γ = By Tare Janarakolca 16
Dummy Varable Technque Dummy varable can also be used o es wheher specfc even has sgnfcan mpac. e.g. Wheher earnng announcemen has mpac on sock prce Wheher he proes has mpac on he sock marke Y = β D + β + β + β + u 1 1 where: D = for normal perod = 1 for even perod Indvdual Tes H : γ = Dummy Varable Technque Weekend Effec and Reverse Weekend Effec on Tha Sock Marke RQ: Wheher here exss evdences of weekend and reverse weekend effec and mpacs of frm sze on he weekend effec and he reverse weekend effec. Objecves: - To examne he evdence of weekend effec and reverse weekend effec n Thaland. - To examne he degree o whch he reverse weekend effec are relaed o frm sze. By Tare Janarakolca 17 By Tare Janarakolca 18 Weekend Effec Defnon Weekend Effec -- Dfferen Reurn on Monday Reverse weekend effec -- Dfferen Reurn on Frday 1 s Obj. Hypohess Tesng H : Excess Reurn Each Day = Where = 1 for Monday, Tuesday, Wednesday, 4 Thursday, and 5 Frday These hypoheses can be esed by usng Onesample -es for each day. If rejec H, means ha here exss excess reurn on each day, oherwse no excess reurn. Dummy varables regresson model: R = + β d + β d + β 4d 4 + β 5d 5 α + ε By Tare Janarakolca 19 If -es of β (=,,,5) s rejeced, means ha here exss excess reurn on each day. If no, here s no excess reurn on ha day. By Tare Janarakolca
nd Obj. Hypohess Tesng H : Dfferen frm sze has dfferen reurn μ 1 = μ = = μ 5 These hypoheses can be esed by usng One-way Analyss of Varance (ANOVA) for each day. If rejec H, means ha frm sze has sgnfcan effec on weekend effec. If no, here s no frm sze effec. By Tare Janarakolca 1