Introduc tion to E c onome tric s Cha pte r 5 E ze quie l Urie l - PowerPoint PPT Presentation

Introduc tion to E c onome tric s Cha pte r 5 E ze quie l Urie l Jimé ne z Unive rsity of Va le nc ia Va le nc ia , Se pte mbe r, 2013

5 Multiple re g re ssion a na lysis with qua lita tive informa tion 5.1 Introduc tion of qua lita tive informa tion in e c onome tric mode ls 5.2 A sing le dummy inde pe nde nt va ria ble 5.3 Multiple c a te g orie s for a n a ttribute 5.4 Se ve ra l a ttribute s 5.5 Inte ra c tions involving dummy va ria ble s 5.6 T e sting struc tura l c ha ng e s E xe rc ise s

5.1 Introduc tion of qua lita tive informa tion in e c onome tric mode ls 5 Multiple regression analysis with qualitative wage β 2 information β 2  1  1  1 +   educ 0 F IGURE 5.1. Sa me slope , diffe re nt inte rc e pt. [3]

5.2 A sing le dummy inde pe nde nt va ria ble E XAMPL E 5.1 Is the re wa g e disc rimina tion a g a inst wome n in Spa in? 5 Multiple regression analysis with qualitative (file wa g e 02sp) = b + d + b + wage female educ u ln( ) 1 1 2  = - + wage female educ ln( ) 1.731 0.307 0.0548 (0.026) (0.022) (0.0025) = = = RSS R n 2 393 0.243 2000 information   H : 0  0.3070 0 1    t 14.26   H : 0 0.0216 1 1 Pe rc e ntage diffe re nc e in ho urly wage between men and wo men    e 0.307 100 ( 1) 35.9% = [4]

5.2 A sing le dummy inde pe nde nt va ria ble E XAMPL E 5.2 Ana lysis of the re la tion be twe e n ma rke t c a pita liza tion a nd 5 Multiple regression analysis with qualitative book va lue : the role of ibe x35 (file bolma d11) = b + d + b + marketcap ibex bookvalue u ln( ) 35 ln( ) 1 1 2  = + + marketcap ibex bookvalue ln( ) 1.784 0.690 35 0.675ln( ) (0.243) (0.179) (0.037) = = = RSS R n 2 35.672 0.893 92 information   H : 0 0.690 t   0 1 3.85   H 0.179 : 0 1 1    e 0.690 100 ( 1) 99.4% Pe rc e ntage diffe re nc e =   H : 0 0.675 t   0 2 18   H 0.037 : 0 1 2 [5]

5.2 A sing le dummy inde pe nde nt va ria ble E XAMPL E 5.3 Do pe ople living in urba n a re a s spe nd more on fish tha n 5 Multiple regression analysis with qualitative pe ople living in rura l a re a s? (file de ma nd) = b + d + b + fish urban inc u ln( ) ln( ) 1 1 2  = - + + fish urban inc ln( ) 6.375 0.140 1.313ln( ) (0.511) (0.055) (0.070) = = = RSS R n 2 1.131 0.904 40 information   H : 0 0.140 0 1 t   2.55   H : 0 0.055 1 1 [6]

5.3 Multiple c a te g orie s for a n a ttribute Dummy var iable tr ap 5 Multiple regression analysis with qualitative E xample            wage small medium large educ u ln( ) 1 0 1 2 2   educ 1 1 0 0 1   educ 1 1 0 0   2 information   educ 1 0 1 0   3  X educ 1 0 1 0   4   educ 1 0 0 1 5   educ   1 0 0 1 6 Solutions:          wage medium large educ u ln( ) 1 1 2 2          wage small medium large educ u ln( ) 0 1 2 2 [7]

5.3 Multiple c a te g orie s for a n a ttribute E XAMPL E 5.4 Doe s firm size influe nc e wa g e de te rmina tion? (file wa g e 02sp) 5 Multiple regression analysis with qualitative = b + q + q + b + wage medium large educ u ln( ) 1 1 2 2  = + + + wage medium large educ ln( ) 1.566 0.281 0.162 0.0480 (0.027) (0.025) (0.024) (0.0025) = = = RSS R n 2 406 0.218 2000 information     H : 0 0 1 2 H H : is not true 1 0 = b + b + wage educ u ln( ) 1 2  = + wage educ ln( ) 1.657 0.0525 (0.026) (0.0026) = = = RSS R n 2 433 0.166 2000       RSS RSS q / 433 406 / 2  R UR   F 66.4   RSS n k / ( ) 406 / (2000 4) UR [8]

5.3 Multiple c a te g orie s for a n a ttribute E XAMPL E 5.5 In the c a se of L ydia E . Pinkha m, a re the time dummy va ria ble s introduc e d sig nific a nt individua lly or jointly? (file pinkha m) 5 Multiple regression analysis with qualitative = b + b + b + b + b + b + sales advexp sales d d d u 1 2 3 - t t t t t t t 1 2 3 1 4 5 6  = + + - + - sales advexp sales d d d t 254.6 0.5345 0.6073 133.35 1 216.84 2 202.50 3 - t t t t t 1 (96.3) (0.136) (0.0814) (89) (67) (67) = = R n 2 0.929 53 ì   q ï H information 0 ï i  0 i í 1,2,3 ï  q ¹ H 0 ï î i 1 - - 133.35 216.84 202.50        - t t t 1.50 3.22 3.02 q ˆ q ˆ q ˆ 89 67 67 1 2 3 ì      q q q ï H ï 0 1 2 3 í ï   H H is not true ï î 1 0   R 2 R 2 q ( ) / (0.9290 0.8770) / 3    UR R F 11.47     R n k 2 (1 ) / ( ) (1 0.9290) / (53 6) UR [9]

5.4 Se ve ra l a ttribute s E XAMPL E 5.6 T he influe nc e of g e nde r a nd le ng th of the workda y on wa g e de te rmina tion (file wa g e 06sp) 5 Multiple regression analysis with qualitative = b + d + f + b + wage female partime educ u ln( ) 1 1 1 2  = - - + wage female partime educ ln( ) 2.006 0.233 0.087 0.0531 (0.026) (0.021) (0.027) (0.0023) = = = RSS R n 2 365 0.235 2000 E XAMPL E 5.7 T rying to e xpla in the a bse nc e from work in the c ompa ny information Bue nosa ire s (file a bse nt) = b + d + f + b + b + b + absent bluecoll male age tenure wage u 1 1 1 2 3 4  = + + - - - absent bluecoll male age tenure wage 12.444 0.968 2.049 0.037 0.151 0.044 (1.640) (0.669) (0.712) (0.047) (0.065) (0.007) = = = RSS R n 2 161.95 0.760 48     H H : 0 : 0 0.968 t   0 1 1 1 1.45     H H 0.669 : 0 : 0 0 1 1 1 2.049 t       2.88 H H : 0 : 0 0.712 0 1 1 1 [10]

5.4 Se ve ra l a ttribute s E XAMPL E 5.8 Size of firm a nd g e nde r in de te rmining wa g e (file wa g e 02sp) 5 Multiple regression analysis with qualitative            wage female medium large educ u ln( ) 1 1 1 2 2       H : 0 0 1 1 2 H H : is not true 1 0 information  = - + + + wage female medium large educ ln( ) 1.639 0.327 0.308 0.168 0.0499 (0.026) (0.021) (0.023) (0.023) (0.0024) = = = RSS R 2 n 361 0.305 2000       RSS RSS q / 433 361 / 3  R UR   F 133   RSS n k / ( ) 361/ (2000 5) UR [11]

5.5 Inte ra c tions involving dummy va ria ble s E XAMPL E 5.9 Is the inte ra c tion be twe e n fe ma le s a nd pa rt- time work sig nific a nt? (file wa g e 06sp) 5 Multiple regression analysis with qualitative = b + d + f + j ´ + b + wage female partime female partime educ u ln( ) 1 1 1 1 2  = - - + ´ + wage female partime female partime educ ln( ) 2.007 0.259 0.198 0.167 0.0538 (0.026) (0.022) (0.047) (0.0024) (0.058) = = = RSS R n 2 363 0.238 2000 information   H : 0 0.167 t   0 1 2.89   H : 0 0.058 1 1 [12]

5.5 Inte ra c tions involving dummy va ria ble s E XAMPL E 5.10 Do sma ll firms disc rimina te a g a inst wome n more or le ss tha n la rg e r firms? (file wa g e 02sp) 5 Multiple regression analysis with qualitative = b + d + q + q wage female medium large ln( ) 1 1 1 2 + j ´ + j ´ + b + female medium female large educ u 1 2 2  = - + + wage female medium large ln( ) 1.624 0.262 0.361 0.179 (0.027) (0.034) (0.028) (0.027) information - female ´ medium - female ´ large + educ 0.159 0.043 0.0497 (0.050) ( 0.051) (0.0024) = = = RSS R 2 n 359 0.308 2000     H : 0 0 1 2 H H : is not true 1 0       RSS RSS q / 361 359 / 2  R UR   F 5.55   RSS n k / ( ) 359 / (2000 7) UR [13]

5.5 Inte ra c tions involving dummy va ria ble s 5 Multiple regression analysis with qualitative wage  2 information  2 +  1  1 educ 0 F IGURE 5.2. Diffe re nt slope , sa me inte rc e pt. [14]

5.5 Inte ra c tions involving dummy va ria ble E XAMPL E 5.11 Is the re turn to e duc a tion for ma le s g re a te r tha n for fe ma le s? 5 Multiple regression analysis with qualitative (file wa g e 02sp) = b + b + d ´ + wage educ female educ u 1 2 1  = + - ´ wage educ educ female ln( ) 1.640 0.0632 0.0274 (0.025) (0.0026) (0.0021) information = = = RSS R n 2 400 0.229 2000   H : 0 0.0274 0 1 t     12.81   H : 0 0.0021 1 1 [15]

5.6 T e sting struc tura l c ha ng e s 5 Multiple regression analysis with qualitative wage  2 information  2 +  2  1  1 +   educ 0 F IGURE 5.3. Diffe re nt slope , diffe re nt inte rc e pt. [16]