Case Study on Linear Regression Analysis

ISOM  201  Case  Study  on  Linear  Regression  Analysis   Due  Date:  November  5th,  2014  

 

CEO  Compensation  

Highly  publicized  salaries  of  corporate  chief  officers  (CEOs)  in  the  US  have  generated  sustained   interest  in  understanding  the  factors  related  to  CEO  compensation  in  general.  The   corresponding  Excel  spreadsheet  contains  data  on  the  annual  compensation  of  the  CEO’s  of  50   large  publicly  traded  corporations  in  the  US  in  the  previous  year,  as  well  as  other  information   that  we  would  hope  could  be  used  to  predict  CEO  compensation.    

In  column  6  in  the  spreadsheet,  a  dummy  variable  is  used.  A  “1”  is  used  to  indicate  that  the  CEO   has  an  MBA,  and  a  “0”  is  used  to  indicate  that  the  CEO  does  not  have  an  MBA.  

 

a) Construct  and  run  a  regression  model  to  predict  CEO  compensation  as  a  function  of  the   independent  variables.  

b) Evaluate  the  regression  output  of  your  regression  model.  Is  the  output  in  good  shape?   Are  there  any  modifications  you  would  like  to  do?  Is  there  any  evidence  of   multicollinearity  problem?  

c) Try  constructing  a  smaller  regression  that  uses  fewer  independent  variables.  You  might   need  to  experiment  with  several  different  regression  models  using  different   combinations  of  independent  variables.  Evaluate  your  best  regression  model  by  looking   at  the  regression  R-­‐square,  significance-­‐F  value,  p-­‐values,  and  confidence  intervals.  Are   you  satisfied  with  your  regression  model?  

d) Which  are  the  critical  factors  that  are  good  predictors  of  CEO  compensation?  In   particular,  does  having  an  MBA  have  an  effect  on  CEO  compensation?  Why  or  Why  not?  

e) Suppose  that  you  would  like  to  use  your  regression  model  to  predict  the  compensation   of  a  particular  CEO.  She  has  been  a  CEO  for  the  last  5  years,  her  company’s  stock  price   increased  15%  over  the  last  year,  her  company’s  sales  increased  30%  over  the  last  year,   and  she  has  an  MBA  degree  from  Suffolk  University.  What  would  your  regression  model   predict  for  her  CEO  compensation  package?