• No results found

The effect of pre-buyout inefficiency on post-buyout performance in private equity takeovers

N/A
N/A
Protected

Academic year: 2021

Share "The effect of pre-buyout inefficiency on post-buyout performance in private equity takeovers"

Copied!
40
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

on  Post-­‐Buyout  Performance  in  

Private  Equity  Takeovers  

          Abstract  

  This   thesis   uses   a   dataset   of   private   equity   buyouts   completed   in   the   United   Kingdom   from   2006   to   2010   to   research   the   effect   of   pre-­‐buyout   inefficiency   on   post-­‐buyout   performance   and   efficiency   gains.   Contrary   to   previous   research   this   thesis   focuses   on   efficiency   gains   realized   by   private   equity.   To   measure   efficiency   the   Altman   Z-­‐score   is   used,   a   novel   use   for   this   score   introduced   by   this   thesis.   This   thesis   finds   a   significant   positive   relation   between   pre-­‐buyout   inefficiency   and   the   change   in   ROA   and   the   change   in   efficiency   over   a   five-­‐year   period.   Firms   that   have   greater   pre-­‐buyout   inefficiency  improve  their  profitability  and  efficiency  score  more  over  the  sample   period.   The   same   is   true   for   specific   inefficiencies   such   as   credit-­‐days   and   collection-­‐days.   The   results   are   robust   to   winsorization   and   to   the   usage   of   individual  Altman  variables  rather  than  the  constrained  Altman  Z-­‐score.  Tests  of   the  robustness  of  the  contribution  of  PE  are  inconclusive.    

               

Name   Sebastiaan  Tito  

Student  Number   10002983  

Program   Economics  and  Business  

Track   Finance  and  Organization  

Supervisor   Timotej  Homar  

Date   28-­‐06-­‐2015  

(2)

Statement  of  Originality  

This   document   is   written   by   Student   Sebastiaan   Tito   who   declares   to   take   full   responsibility  for  the  contents  of  this  document.  

I  declare  that  the  text  and  the  work  presented  in  this  document  is  original  and   that  no  sources  other  than  those  mentioned  in  the  text  and  its  references  have   been  used  in  creating  it.  

The  Faculty  of  Economics  and  Business  is  responsible  solely  for  the  supervision   of  completion  of  the  work,  not  for  the  contents.  

(3)

2.  Literature  ...  6  

2.1  Private  Equity  and  its  history  ...  6  

2.2  Leveraged  Buyouts  ...  7  

2.3  Private  Equity  Research  ...  8  

3.  Research  Methodology  ...  10  

3.1  Efficiency  Scoring  ...  10  

3.2  Performance  Change  ...  11  

3.3  Efficiency  Change  ...  12  

3.4  Credit  and  Collection  Days  ...  13  

4.  Data  ...  14  

4.1  Data  on  Private  Equity  deals  ...  14  

4.2  Data  on  Portfolio  Companies  ...  15  

4.3  Sector  Data  ...  16  

4.4  Market  Data  ...  17  

5.  Results  ...  18  

5.1  Performance  Change  Analysis  ...  18  

5.2  Efficiency  Change  Analysis  ...  19  

5.3  Credit  and  Collection  Results  ...  22  

6.  Robustness  ...  23  

6.1  Individual  Altman  Variables  ...  23  

6.2  Private  Equity  Robustness  Test  ...  25  

6.3  Non-­‐Winsorized  Results  ...  26  

6.4  ROE  Robustness  Check  ...  29  

7.  Conclusion  ...  30  

References  ...  32  

(4)

1.  Introduction  

   In   the   1980s   the   Private   Equity   (PE)   sponsored   buyout   industry   grows   rapidly.  The  total  committed  capital  to  PE  funds  increases  from  $600  million  in   the   late   1970s   to   $100   billion   in   1994   (Fenn,   Liang,   &   Prowse,   1995).   The   increased  activity  in  this  market  attracts  the  attention  of  researchers.  

  Most   research   into   PE   focuses   on   the   level   of   the   PE   firm.   This   includes   research  into  the  risks  of  investment  in  PE,  research  into  the  performance  of  PE   adjusted   for   risks   and   outperformance   of   the   market,   and   research   into   the   wealth  effects  of  leveraged  buyouts  (LBOs).  However,  the  research  into  the  effect   of   PE   ownership   on   the   portfolio   company   is   limited   in   quantity   and   in   scope.   The   research   that   exists   is   mostly   from   the   1980s,   and   focuses   on   the   performance  of  a  portfolio  firm  post-­‐buyout.  The  results  from  these  papers  are   ambiguous.   Kaplan   (1989)   finds   that   PE   ownership   has   a   positive   effect   on   performance,  as  does  Jensen  (1989).  However,  Scellato  and  Ughetto  (2013)  find   that  PE  ownership  has  a  negative  effect,  while  Acharya,  Hahn,  and  Kehoe  (2009)   find  it  has  a  positive  effect  using  the  same  techniques.    

  This  thesis  intends  to  slightly  broaden  the  scope  of  the  existing  research   into   pre-­‐   and   post-­‐buyout   performance.   It   will   focus   on   the   efficiency   of   the   portfolio  company.  Central  to  this  research  is  the  PE  buyout  business  model  of   acquiring  a  business  and  making  operational  changes  to  trim  the  fat  and  create  a   more   efficient   company   (Mohan,   1990).   To   a   PE   investor,   pre-­‐existing   operational  and  organizational  inefficiencies  are  areas  to  potentially  improve  the   company.   This   thesis   will   research   if   companies   that   have   more   pre-­‐buyout   inefficiencies  manage  to  make  greater  improvements  during  their  time  as  a  PE   portfolio  company.  If  this  is  the  case  then  PE  does  indeed  create  more  firm  value   in  companies  that  are  in  worse  shape  pre-­‐buyout.  This  concerns  the  function  of   PE   for   individual   companies   as   well   as   the   benefit   of   PE   to   the   economy   as   a   whole.    

  The   research   question   of   this   thesis   is:   what   is   the   effect   of   pre-­‐buyout   inefficiencies  on  post-­‐buyout  performance?  Based  on  the  PE  business  model  the   pre-­‐buyout   inefficiencies   can   be   thought   of   as   potential   performance   gains.   Greater   inefficiencies   mean   more   potential   for   improvement   when   a   PE   firm   takes  over.  The  circumstances  of  a  buyout  with  the  increased  leverage  and  the  

(5)

realignment  of  the  incentives  can  be  expected  to  facilitate  this.  The  hypothesis  is   that  more  inefficiency  pre-­‐buyout  should  have  a  positive  effect  on  the  change  in   performance  and  efficiency  over  a  period  of  4  years.  This  period  is  at  the  end  of   the   planned   PE   holding   period   which   presumably   means   that   the   PE   firm   will   have  realized  most  of  the  efficiency  gains  they  thought  possible.  

  Empirical   research   will   be   done   to   test   this   hypothesis.   The   first   step   is   measuring   pre-­‐buyout   efficiency   and   performance;   the   second   step   is   measure   performance  and  efficiency  4  years  after  the  buyout.  A  multiple  regression  model   will  then  be  estimated  to  test  the  effect  of  the  pre-­‐buyout  efficiency  on  both  the   change   in   performance   and   the   change   in   efficiency   over   this   period.   The   measure   for   the   pre-­‐buyout   efficiency   is   expected   to   have   a   negative   sign,   indicating   a   positive   effect   of   lower   efficiency   and   a   negative   effect   of   greater   efficiency  pre-­‐buyout.  

  Chapter   2   will   go   more   in-­‐depth   into   the   history   of   PE   and   the   existing   literature.  The  goal  of  this  chapter  is  to  outline  the  different  views  on  PE  and  to   substantiate   the   research   decisions.   Chapter   3   will   outline   the   methodology.   It   states   how   the   data   will   be   used   to   generate   different   variables   and   what   the   actual  research  models  are.  Chapter  4  will  describe  and  test  the  dataset  used  for   this   research.   The   goal   of   this   chapter   is   to   promote   transparency   in   this   research   and   enable   the   reader   to   judge   the   validity   of   the   data.   Chapter   5   presents   the   results   from   this   research.   The   outcomes   of   the   different   models   and  their  meanings  will  be  explained.  Chapter  6  is  the  robustness  section.  This   section  tests  several  aspects  of  the  thesis  for  robustness.  The  goal  of  this  chapter   is  to  give  the  reader  insight  into  the  effect  of  the  research  setup  on  the  results.   Chapter  7  is  the  conclusion,  it  answers  the  research  question  and  discusses  the   limitations  of  this  research.  

(6)

2.  Literature  

2.1  Private  Equity  and  its  history  

  A  PE  fund  is  a  non-­‐transparent  investment  vehicle  that  invests  the  funds   of  its  investors  in  companies  (Fraser-­‐Sampson,  2007).  The  PE  fund  belongs  to  a   PE   firm   that   usually   manages   several   of   these   funds   and   the   companies   these   funds  hold  in  their  investment  portfolios  (Cumming,  2010).  The  PE  firm  intends   an  exit  after  5  years  and  aims  for  an  annualized  return  in  excess  of  20%  (Pearl  &   Rosenbaum,  2013).  In  practice  The  PE  firm  manages  these  companies  for  about   4-­‐7   years   (Fenn   et   al.,   1995).   In   this   period   the   PE   firm   can   make   operational   changes  as  well  as  investments  in  the  portfolio  company  to  boost  its  resale  value.     In  practice  PE  firms  are  usually  characterized  as  Venture  Capital  (VC)  or   as   buyout   funds   (Fraser-­‐Sampson,   2007).   Venture   Capital   funds   focus   on   early   investments   in   startup   companies   that   have   no   access   to   traditional   capital   markets.  In  return  for  a  share  of  the  control  they  provide  the  startup  with  access   to  capital  and  general  business  knowledge.  Their  goal  is  usually  to  sell  most  of   their   investment   in   an   initial   public   offering   (IPO)   (Damodaran,   2010).   Fraser-­‐ Sampson   (2007)   states   that   VC   is   a   powerful   tool   for   economic   growth.   He   further  states  that  by  the  end  of  2000,  VC  is  directly  responsible  for  the  creation   of  8  million  jobs,  roughly  one  job  for  every  $36  thousand  invested  (2007).  

  Buyout  funds  target  established  companies  rather  than  startups  (Fraser-­‐ Sampson,  2007).  The  deals  often  contain  a  large  portion  of  debt  financing,  and   buyout   firms   usually   take   a   more   active   approach   in   managing   the   portfolio   company   (2007).   Especially   in   Europe   they   are   seen   as   less   favorably   by   governments   than   VC   (2007).   This   is   because   their   operational   restructures   reduce  jobs  and  their  financial  restructures  reduce  tax  yield  (2007).  IPO  exits  are   less   likely   for   buyout   funds   than   for   VC,   especially   in   recent   years   so   called   secondary   buyouts   (selling   to   another   investment   firm)   are   becoming   more   prevalent  (Jong,  Roosenboom,  Verbeek,  &  Verwijmeren,  2007).  

  Private  Equity  firms  can  be  found  as  far  back  as  the  1940s,  although  these   firms   focus   mainly   on   Venture   Capital.   There   are   not   many   firms   involved   in   buyouts  at  this  time  (Liles,  1977).  It  is  not  until  the  development  of  the  limited  

(7)

partnership  in  the  1970s  that  both  buyout  and  VC  firms  become  more  popular   (Bygrave  &  Timmons,  1992).  However,  the  regulatory  changes  and  tax  benefits   of  the  late  1970s  and  the  early  1980s  are  what  really  start  the  boom  in  PE  (Pratt,   1982).   It   is   during   this   boom   that   leveraged   buyouts   become   very   popular.   In   1987   a   record   size   buyout   fund   by   Kohlberg,   Kravis,   and   Roberts   (KKR)   raises   $5.6  billion,  almost  twice  the  total  commitments  to  VC  in  that  year  (Fenn  et  al.,   1995).  In  1988  KKR  executes  one  of  the  largest  LBOs  in  history,  and  the  largest   ever  at  that  time,  when  it  buys  RJR-­‐Nabisco  for  $31.4  billion  (Copeland,  Koller,  &   Murrin,  2000).  The  increasing  size  of  the  industry  and  the  total  committed  funds   draw   more   attention   to   PE.   Especially   the   LBO   as   an   instrument   comes   under   scrutiny  (Axelson,  Strömberg,  &  Weisbach,  2008).    

2.2  Leveraged  Buyouts  

  The  LBO  is  the  deal  type  most  commonly  associated  with  a  PE  buyout,  and   refers   to   a   special   form   of   deal   financing.   Although   Fraser-­‐Sampson   (2007)   emphasizes  that  really  all  buyouts  are  ‘leveraged’,  because  they  all  involve  some   form  of  debt  financing,  there  is  a  marked  difference  between  a  regular  deal  and   an  LBO.  Central  to  an  LBO  is  the  use  of  a  target’s  own  assets  to  secure  acquisition   debt   (Fraser-­‐Sampson,   2007).   On   average   about   75%   of   the   acquisition   is   financed  using  debt  (Jong  et  al.,  2007).  

  There  are  different  views  on  the  benefits  and/or  drawbacks  of  LBO  deals.   After  the  M&A  wave  of  the  1960s  and  1970s  there  are  a  lot  of  large  inefficient   conglomerates  (Jong  et  al.,  2007).  In  this  landscape  LBOs  provide  a  key  economic   benefit,   because   they   enable   buyout   companies   to   purchase   and   restructure   these  conglomerates  (Jong  et  al.,  2007).  However,  in  the  period  1986-­‐1990,  41%   of   the   buyout   value   creation   is   the   result   of   the   increased   debt,   rather   than   operational   improvement   (2007).   These   leveraged   recapitalizations   are   of   significant   importance   in   the   operations   of   PE   firms   and   have   come   under   political  scrutiny  (Fraser-­‐Sampson,  2007).  

  However,   there   are   positive   effects   of   PE   ownership   and   the   associated   high  amount  of  debt.  Jensen  (1989)  argues  that  active  investors  are  a  return  to   form  for  companies  since  active  investors  are  able  to  maximize  firm  value.  This  is   consistent   with   the   empirical   results   of   research   by   Donaldson   (1984).  

(8)

Donaldson   finds   that   top   managers   are   more   concerned   with   maximizing   the   corporate   purchasing   power   of   management,   rather   than   maximization   of   firm   value   (1984).   LBOs   facilitate   the   active   PE   investors   in   realigning   these   companies   to   make   them   perform   more   effectively   with   the   same   resources   (Jensen,  1989).  In  a  different  paper  Jensen  (1986)  discusses  the  agency  costs  of   free   cash   flow   in   an   organization.   His   theory   states   that   management   has   incentives  to  grow  the  business  beyond  its  optimal  size,  because  this  increases   their   remuneration   (1986).   This   is   consistent   with   the   empirical   findings   of   Murphy  (1985),  who  found  a  strong  positive  relation  between  firm  growth  and   increase   in   managerial   remuneration.   The   solution   Jensen   (1986)   proposes   is   more   debt,   because   the   increased   debt   service   payments   force   management   to   use   funds   more   efficiently   and   stop   them   from   undertaking   value   reducing   projects.  This  makes  management  more  aligned  with  the  broader  stakeholders.  

2.3  Private  Equity  Research  

  Research   into   the   performance   of   PE   firms   shows   that   unlike   mutual   funds,   PE   can   be   a   better   investment   than   a   diversified   portfolio.   Kaplan   and   Schoar  (2005)  do  empirical  research  into  the  performance  of  PE  firms.  They  find   that   on   average   the   risk-­‐adjusted   fund   returns   net   of   fees   equal   the   S&P500   index,   this   is   consistent   with   findings   by   Phalippou   and   Gottschalg   (2007)   and   Phalippou  (2009).  But  Kaplan  and  Schoar  (2005)  do  find  significant  evidence  of   heterogeneity   across   funds.   They   find   significant   evidence   of   persistent   outperformance  for  better  performing  funds,  indicating  that  well-­‐managed  funds   do   provide   an   alpha   for   investors   (Kaplan   &   Schoar,   2005).   However,   their   results   show   that   PE   funds   suffer   from   diseconomies   of   scale,   and   that   as   an   outperforming   firm   becomes   more   popular   and   its   committed   capital   to   funds   grows,   it   delivers   poorer   results   (2005).   The   diseconomies   of   scale   are   also   found   in   empirical   research   by   Lopez-­‐de-­‐Silanes,   Phalippou,   and   Gottschalg   (2013).   However,   contrary   to   these   findings,   a   recent   empirical   research   by   Harris,   Jenkinson,   and   Kaplan   (2014)   finds   that   buyout   funds   outperform   the   market  by  an  average  of  20%-­‐27%  when  adjusted  for  risk  and  net  of  fees.    

As   mentioned   in   the   introduction   the   research   into   the   effect   of   PE   ownership  on  portfolio  companies  is  ambiguous.  Kaplan  (1989)  does  empirical  

(9)

research   into   the   performance   of   portfolio   companies   of   buyouts   completed   between  1980  and  1986.  He  finds  that  these  companies  have  increased  operating   income,  decreased  capital  expenditures,   and   increases   in   net   cash   flow   3   years   after   the   buyout   when   he   controls   for   economy-­‐wide   and   industry   effects.   He   also   controls   for   divestitures   and   layoffs.   His   results   show   that   buyouts   experience   an   improvement   in   performance   that   is   due   to   operational   improvements  rather  than  layoffs  or  managerial  exploitation  of  shareholders.  

Scellato  and  Ughetto  (2013)  use  propensity  score  matched  peers  to  better   control   for   the   effect   of   PE   buyout   on   a   portfolio   company.   They   find   that   a   buyout   by   a   generalist   fund   negatively   impacts   operating   profitability,   while   a   buyout  by  a  turnaround  specialist  positively  impacts  operating  profitability.  This   is   contrary   to   empirical   findings   by   Acharya   et   al.   (2009).   They   use   the   same   propensity   score   matched   peers   but   find   that   PE   ownership   positively   affects   profitability.   Bacon,   Wright,   Meuleman,   and   Scholes   (2012)   do   empirical   research  into  the  effect  of  a  PE  buyout  on  human  resources.  They  find  that  a  PE   buyout  increases  average  skill  level  and  job  satisfaction.  

(10)

3.  Research  Methodology  

3.1  Efficiency  Scoring  

  This  thesis  intends  to  research  the  effect  of  portfolio  company  pre-­‐buyout   inefficiencies   on   post-­‐buyout   performance.   The   first   step   is   therefore   to   determine   the   level   of   pre-­‐buyout   efficiency.   An   extensive   due   diligence   is   required  for  an  accurate  evaluation  of  existing  inefficiencies  within  a  company.   However,  in  the  context  of  this  research  that  is  not  possible.  The  estimation  of   this  measure  is  therefore  limited  to  the  financial  and  operational  data  reported   by  the  portfolio  companies  and  the  ratios  computed  from  this  data.    

  It  is  important  that  this  score  is  a  broad  measure  of  both  the  operational   efficiency   as   well   as   the   efficiency   of   the   structure   of   a   firm.   This   requires   a   measure   that   takes   multiple   firm   characteristics   into   account.   The   measure   needs   to   score   liquidity,   profitability,   leverage,   solvency,   and   activity.   To   get   a   reliable   measure   with   adequate   weighing   of   these   different   categories   this   research  will  make  use  of  the  Altman  Z-­‐score.  This  score  is  designed  specifically   as  a  measure  for  these  characteristics  (Altman,  1968).  

  Because   this   research   focuses   on   companies   owned   by   a   PE   firm,   the   measure  of  inefficiency  will  be  the  Altman  Z-­‐score  for  private  companies.  This  so   called   Z’-­‐score   focuses   on   book   value   where   the   original   Z-­‐score   focuses   on   market  value  (Altman,  2000).  For  simplicity  there  will  be  no  focus  on  the  original   Altman  score.  All  further  mentions  of  the  ‘Z-­‐score’  will  refer  to  the  Z’-­‐score  for   private  companies.  

  The  Altman  Z-­‐score  has  the  following  form:  

𝑍 = 0.717 𝑋! + 0.847 𝑋! + 3.107 𝑋! + 0.420 𝑋! + 0.998 𝑋! .   Where:   𝑋! = 𝑊𝑜𝑟𝑘𝑖𝑛𝑔  𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝑇𝑜𝑡𝑎𝑙  𝐴𝑠𝑠𝑒𝑡𝑠,     𝑋! = 𝑅𝑒𝑡𝑎𝑖𝑛𝑒𝑑  𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑇𝑜𝑡𝑎𝑙  𝐴𝑠𝑠𝑒𝑡𝑠,     𝑋! = 𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠  𝐵𝑒𝑓𝑜𝑟𝑒  𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡  𝑎𝑛𝑑  𝑇𝑎𝑥𝑒𝑠 𝑇𝑜𝑡𝑎𝑙  𝐴𝑠𝑠𝑒𝑡𝑠,     𝑋! = 𝐵𝑜𝑜𝑘  𝑉𝑎𝑙𝑢𝑒  𝑜𝑓  𝐸𝑞𝑢𝑖𝑡𝑦 𝑇𝑜𝑡𝑎𝑙  𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠,     𝑋! = 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔  𝑅𝑒𝑣𝑒𝑛𝑢𝑒 𝑇𝑜𝑡𝑎𝑙  𝐴𝑠𝑠𝑒𝑡𝑠.      

(11)

Originally  designed  by  Edward  I.  Altman  in  1968,  his  revisited  paper  from   2000   contains   this   version   designed   for   the   analysis   of   private   companies.   His   goal   was   to   design   a   score   that   mapped   the   default   probability   of   a   company   using   a   weighed   scoring   of   broad   financial   measures   from   several   key   firm   characteristics.  This  score  allows  the  comparing  of  different  firms  based  on  their   overall   health.   For   the   purposes   of   this   research   the   Altman   Z-­‐score   is   a   great   way   to   benchmark   efficiency   of   portfolio   companies   in   a   way   that   allows   for   comparison   across   structures   and   business   models.   The   inclusion   of   control   variables  for  industry  allows  for  an  even  broader  comparison.  A  higher  Altman   Z-­‐score  thus  reflects  greater  efficiency  and  a  lower  score  reflects  the  existence  of   more   inefficiencies,   and   thus   more   potential   unrealized   efficiency   and   performance   gains.   The   use   of   Altman   Z-­‐scores   as   a   measure   of   efficiency   is   a   novel   use   of   the   score   this   research   introduces   and   it   goes   a   long   way   to   overcome  the  obstacle  of  not  being  able  to  do  extensive  due  diligence.  

  In   this   thesis   the   Z-­‐score   is   central   to   the   identification   of   inefficiency.   However,   the   main   regressions   are   also   estimated   using   the   individual   Altman   variables.  This  is  to  test  the  robustness  of  using  the  Z-­‐score  without  the  weight-­‐ constraints  suggested  by  Altman.    For  these  regressions  please  refer  to  Section   6.1.  

3.2  Performance  Change  

  A   performance   measure   needs   to   be   specified   to   measure   the   effect   of   inefficiencies.   Based   on   existing   literature   on   the   subject   of   performance   measures,   this   research   will   use   Pre-­‐Tax   Return   on   Assets.   The   advantage   of   using  PTROA  is  that  is  it  not  influenced  by  possible  tax  benefits  from  accounting   changes  or  leverage.  For  simplicity  all  further  mentions  of  ROA  will  concern  Pre-­‐ Tax   Return   on   Assets.   The   advantage   of   using   ROA   as   a   measure   is   that   it   measures  the  return  to  the  broader  stakeholders,  rather  than  just  equity  holders.   It  is  also  less  influenced  by  changes  in  leverage  than  ROE,  which  is  an  important   advantage  since  leverage  varies  widely  across  PE  portfolio  companies  and  across   the   holding   period   (Acharya   et   al.,   2009).   Dess   and   Robinson   (1984)   also   describe   ROA   as   both   a   measure   of   economic   performance   of   a   firm   as   well   as   efficiency  of  a  firm  in  regard  to  the  profitable  use  of  its  total  asset  base.    

(12)

Because  of  the  existence  of  negative  ROAs,  the  change  in  performance  is   measured  as  the  simple  difference  in  ROA  stated  as:  

𝑅𝑂𝐴  𝐶ℎ𝑎𝑛𝑔𝑒 = 𝑅𝑂𝐴!!− 𝑅𝑂𝐴!"#  

  To   research   the   effect   of   the   pre-­‐buyout   efficiency   on   post-­‐buyout   performance   this   research   uses   multiple   regression   models   that   add   control   variables  to  control  for  outside  influences.  The  full  regression  model  is:  

𝑟𝑜𝑎_𝑐ℎ𝑎𝑛𝑔𝑒! = 𝛼 + 𝛽! 𝑧_𝑝𝑟𝑒! + 𝛽! 𝑟𝑜𝑎_𝑝𝑔𝑚𝑒𝑑_𝑐ℎ𝑎𝑛𝑔𝑒! + 𝛽! 𝑓𝑡𝑠𝑒_𝑟𝑒𝑡 +

𝛽! 𝑠𝑖𝑧𝑒! + 𝜀!    

Where:   𝑟𝑜𝑎_𝑐ℎ𝑎𝑛𝑔𝑒!   Change  in  ROA    

  𝑧_𝑝𝑟𝑒!   Pre-­‐Buyout  Altman  Z-­‐score  

  𝑟𝑜𝑎_𝑝𝑔𝑚𝑒𝑑_𝑐ℎ𝑎𝑛𝑔𝑒!   Change  in  Peer  Group  Median  ROA       𝑓𝑡𝑠𝑒_𝑟𝑒𝑡   FTSE100  return  between  pre  and  t4  

  𝑠𝑖𝑧𝑒!   Log(Total  Assets)  

 

3.3  Efficiency  Change  

  The   efficiency   is   scored   pre-­‐buyout   and   at   t4   using   the   Altman   Z-­‐score  

formula   stated   above.   This   research   will   also   focus   on   the   effect   of   this   pre-­‐ buyout   efficiency   on   change   in   efficiency   at   t4.   Because   of   the   existence   of  

negative   Z-­‐scores   this   efficiency   change   is   the   simple   difference   in   Altman   Z-­‐ score  stated  as:  

𝑍  𝐶ℎ𝑎𝑛𝑔𝑒 = 𝑍!!− 𝑍!"#  

  To   research   the   effect   of   the   pre-­‐buyout   efficiency   on   post-­‐buyout   efficiency  changes  this  research  uses  multiple  regression  models  that  add  control   variables  to  control  for  outside  influences.  The  full  regression  model  is:  

𝑧_𝑐ℎ𝑎𝑛𝑔𝑒! = 𝛼 + 𝛽!(𝑧_𝑝𝑟𝑒!) + 𝛽! 𝑟𝑜𝑎_𝑝𝑔𝑚𝑒𝑑_𝑐ℎ𝑎𝑛𝑔𝑒!

+ 𝛽! 𝑧_𝑝𝑔𝑚𝑒𝑑_𝑐ℎ𝑎𝑛𝑔𝑒! + 𝛽! 𝑓𝑡𝑠𝑒_𝑟𝑒𝑡 + 𝛽! 𝑠𝑖𝑧𝑒! + 𝜀!   Where:   𝑟𝑜𝑎_𝑐ℎ𝑎𝑛𝑔𝑒!   Change  in  ROA    

  𝑧_𝑝𝑟𝑒!   Pre-­‐Buyout  Altman  Z-­‐score  

  𝑟𝑜𝑎_𝑝𝑔𝑚𝑒𝑑_𝑐ℎ𝑎𝑛𝑔𝑒!   Change  in  Peer  Group  Median  ROA       𝑧_𝑝𝑔𝑚𝑒𝑑_𝑐ℎ𝑎𝑛𝑔𝑒!   Change  in  Peer  Group  Median  Z-­‐score       𝑓𝑡𝑠𝑒_𝑟𝑒𝑡   FTSE100  return  between  pre  and  t4  

(13)

3.4  Credit  and  Collection  Days  

  An  important  part  of  the  efficient  functioning  of  a  company  is  the  time  it   has  to  pay  back  creditors  and  the  time  it  takes  to  collect  on  its  bills.  This  section   will  research  the  changes  in  these  credit  and  collection  periods  specifically.       To   enable   comparison   across   industries   the   periods   are   benchmarked   against  the  median  periods  in  their  peer  group.  The  benchmark  is  calculated  for   the  year  pre-­‐buyout  and  for  t4.  The  dependent  variable  in  the  regressions  will  be  

the  change  of  this  benchmark.  These  measures  can  be  defined  as:  

𝐶𝑟𝑒𝑑𝑖𝑡𝑑𝑎𝑦𝑠  𝐵𝑒𝑛𝑐ℎ𝑚𝑎𝑟𝑘 =   𝐶𝑜𝑚𝑝𝑎𝑛𝑦  𝐶𝑟𝑒𝑑𝑖𝑡𝑑𝑎𝑦𝑠  𝑝𝑟𝑒 𝑃𝑒𝑒𝑟  𝐺𝑟𝑜𝑢𝑝  𝑀𝑒𝑑𝑖𝑎𝑛  𝐶𝑟𝑒𝑑𝑖𝑡𝑑𝑎𝑦𝑠  𝑝𝑟𝑒   𝐶𝑜𝑙𝑙𝑒𝑐𝑡𝑑𝑎𝑦𝑠  𝐵𝑒𝑛𝑐ℎ𝑚𝑎𝑟𝑘 =   𝐶𝑜𝑚𝑝𝑎𝑛𝑦  𝐶𝑜𝑙𝑙𝑒𝑐𝑡𝑖𝑜𝑛  𝑑𝑎𝑦𝑠  𝑝𝑟𝑒 𝑃𝑒𝑒𝑟  𝐺𝑟𝑜𝑢𝑝  𝑀𝑒𝑑𝑖𝑎𝑛  𝐶𝑜𝑙𝑙𝑒𝑐𝑡𝑖𝑜𝑛  𝑑𝑎𝑦𝑠  𝑝𝑟𝑒   𝐶𝑟𝑒𝑑𝑖𝑡𝑑𝑎𝑦𝑠  𝐵𝑒𝑛𝑐ℎ𝑚𝑎𝑟𝑘  𝐶ℎ𝑎𝑛𝑔𝑒 = 𝐶𝑜𝑚𝑝𝑎𝑛𝑦  𝐶𝑟𝑒𝑑𝑖𝑡𝑑𝑎𝑦𝑠  𝑡4 𝑃𝑒𝑒𝑟  𝐺𝑟𝑜𝑢𝑝  𝑀𝑒𝑑𝑖𝑎𝑛  𝐶𝑟𝑒𝑑𝑖𝑡𝑑𝑎𝑦𝑠  𝑡4− 𝐶𝑜𝑚𝑝𝑎𝑛𝑦  𝐶𝑟𝑒𝑑𝑖𝑡𝑑𝑎𝑦𝑠  𝑝𝑟𝑒 𝑃𝑒𝑒𝑟  𝐺𝑟𝑜𝑢𝑝  𝑀𝑒𝑑𝑖𝑎𝑛  𝐶𝑟𝑒𝑑𝑖𝑡𝑑𝑎𝑦𝑠  𝑝𝑟𝑒   𝐶𝑜𝑙𝑙𝑒𝑐𝑡𝑑𝑎𝑦𝑠  𝐵𝑒𝑛𝑐ℎ𝑚𝑎𝑟𝑘  𝐶ℎ𝑎𝑛𝑔𝑒 = 𝐶𝑜𝑚𝑝𝑎𝑛𝑦  𝐶𝑜𝑙𝑙𝑒𝑐𝑡𝑖𝑜𝑛  𝑑𝑎𝑦𝑠  𝑡4 𝑃𝑒𝑒𝑟  𝐺𝑟𝑜𝑢𝑝  𝑀𝑒𝑑𝑖𝑎𝑛  𝐶𝑜𝑙𝑙𝑒𝑐𝑡𝑖𝑜𝑛  𝑑𝑎𝑦𝑠  𝑡4− 𝐶𝑜𝑚𝑝𝑎𝑛𝑦  𝐶𝑜𝑙𝑙𝑒𝑐𝑡𝑖𝑜𝑛  𝑑𝑎𝑦𝑠  𝑝𝑟𝑒 𝑃𝑒𝑒𝑟  𝐺𝑟𝑜𝑢𝑝  𝑀𝑒𝑑𝑖𝑎𝑛  𝐶𝑜𝑙𝑙𝑒𝑐𝑡𝑖𝑜𝑛  𝑑𝑎𝑦𝑠  𝑝𝑟𝑒  

  The  benchmarks  serve  as  measures  of  how  efficient  a  company  manages   its   credit   and   its   collections   compared   to   its   peer   group.   A   credit-­‐days   benchmark   >1   is   indicative   of   favorable   credit   terms   for   the   company.   A   collection-­‐days   benchmark   >1   indicates   a   company   takes   longer   than   the   peer   group  median  to  collect  from  debtors.  

  The  regressions  test  both  the  effect  of  the  pre-­‐buyout  benchmark  as  well   as  the  effect  of  the  pre-­‐buyout  Z-­‐score  on  the  change  in  the  benchmark.  Here,  the   pre-­‐buyout   Z-­‐score   is   again   used   to   measure   the   general   level   of   efficiency.   However,  the  pre-­‐buyout  benchmark  is  also  included  to  measure  the  pre-­‐buyout   efficiency  specific  to  the  credit  and  collection  periods.  The  full  regression  model   for  both  benchmark  changes  is:  

(14)

4.  Data  

4.1  Data  on  Private  Equity  deals  

  This   analysis   focuses   on   Private   Equity   buyouts   of   UK   companies,   executed   between   2006   and   2010.   The   focus   is   on   UK   companies   because   the   availability  of  financial  and  operational  data  on  private  companies  is  greater  for   UK  companies  than  for  US  companies.  The  UK  is  also  a  large  market  with  well-­‐ developed   PE,   making   it   unnecessary   to   include   other   countries   to   obtain   an   adequate   sample   size   (Engel   &   Stiebale,   2014).   Deal   data   is   collected   from   the   Zephyr  M&A  database,  with  Bureau  van  Dijk  identification  numbers  (BvDIDs)  for   the   target   companies.   To   accurately   examine   the   effect   of   PE   ownership   on   portfolio   company   management   and   operations,   the   research   is   limited   to   full   buyouts.   This   dataset   has   677   buyouts.   The   sample   has   been   reduced   to   133   buyouts,   because   this   research   requires   financial   and   operational   data   not   available  for  all  companies  in  the  full  dataset.  Financial  and  operational  data  for   the   portfolio   companies   is   collected   from   the   Orbis   database   using   the   companies’  BvDIDs.  For  an  overview  of  the  number  of  buyouts  per  year  please   refer  to  Table  1.  

Table  1  Deal  Distribution  by  Year  

Deals       Population   Sample   2006   150   29   2007   195   51   2008   141   31   2009   60   16   2010   131   6   Total   677   133    

  Several   firm   characteristics   from   the   sample   are   subjected   to   T-­‐testing   against  the  known  values  in  the  population  to  test  the  representativeness  of  the   sample.   The   means   of   the   sample   and   the   population:   total   assets   (t=0.7472),   number  of  employees  (t=0.3734),  and  operating  revenue  (t=0.8522)  are  tested.   The   null   hypotheses   are   not   rejected   at   an   alpha   of   10%,   5%,   or   1%.   There   is  

(15)

therefore   no   evidence   of   a   difference   in   the   means   introduced   by   using   this   sample.  For  detailed  results  of  these  tests  please  refer  to  Appendix  1.  

4.2  Data  on  Portfolio  Companies  

  Comprehensive   annual   data   is   gathered   on   each   of   the   133   portfolio   companies   through   the   Orbis   database.   This   data   includes   complete   balance   sheets,  key  ratios,  and  operational  data.  This  research  will  focus  on  the  company   pre-­‐buyout   (that   is,   one   year   prior   to   the   buyout   year)   and   4   years   after   the   buyout.  This  annual  data  is  reworked  for  each  company  to  fit  the  uniform  labels   of   ‘pre’   and   ‘t4’.   This   is   done   to   enable   comparisons   between   companies   with  

different  buyout  years.    

  The   reliability   of   this   research   is   highly   dependent   on   the   quality   of   the   gathered  data.  To  control  for  the  influence  of  severe  outliers  or  possible  faulty   reporting,   the   data   is   winsorized   at   the   1st   and   99th   percentile.   Section   6.3  

contains  the  regression  results  from  non-­‐winsorized  variables.  This  will  provide   this  research  with  transparency  and  allows  the  reader  to  make  inferences  on  the   effect   of   the   winsorization   on   the   outcomes   and   conclusions.   For   descriptive   statistics   on   the   most   important   portfolio   company   variables   please   refer   to   Table  2.  For  full  descriptive  statistics  please  refer  to  Appendix  2.  

Table  2  Descriptive  Statistics  Portfolio  Companies  

    N   Mean   Median   Std.  Dev.   Min   Max  

                            Altman  Pre   132   2.59   2.35   1.91   -­‐0.57   8.54   Altman  t4   131   2.49   2.53   1.75   -­‐1.59   7.12   Altman  Change1   131   -­‐0.09   -­‐0.12   1.53   -­‐5.41   5.65   ROA  Pre   133   10.13   8.15   19.02   -­‐72.65   58.55   ROA  t4   133   7.50   6.77   17.72   -­‐59.34   69.08   ROA  Change1   133   -­‐2.63   -­‐1.81   22.07   -­‐117.89   59.35   OR  Pre2   133   180.75   48.31   427.89   0.46   3077.41   OR  t42   133   188.21   71.78   378.17   1.96   2791.56   OR  Change  (%)   133   99.24%   22.00%   328.81%   -­‐74.36%   2734.94%  

Total  Assets  Pre2   133   279.15   32.34   765.36   0.55   4416.10  

Credit-­‐Days  Pre   130   32.75   28   39.25   0   298  

Credit-­‐Days  t4   131   31.13   27   27.45   0   176  

Collection-­‐Days  Pre   130   49.12   52   31.00   0   133  

Collection-­‐Days  t4   132   45.71   43   31.45   0   139  

1:  Simple  difference  t4  -­‐  pre  

         

2:  In  $  Millions  

(16)

4.3  Sector  Data  

  The  portfolio  companies  are  widely  spread  across  sectors.  The  sample  of   133   companies   has   76   unique   four-­‐digit   NACE   sector   identifications   and   one   company   missing   a   NACE   identifier.   The   highest   concentration   of   NACE   identifications  is  17  companies  with  NACE  7010,  the  indication  for  head  office  of   a   holding   company.   This   constitutes   12.78%   of   the   sample.   Other   significant   concentrations   are   7   companies   (5.26%)   in   ‘other   business   support’   (NACE   8299),  7  companies  (5.26%)  in  ‘other  financial  brokerage’  (NACE  6499),  and  5   companies   (3.76%)   in   ‘other   information   technology’   (NACE   6209).   For   information  on  sectors  with  a  lower  frequency  as  well  as  a  complete  breakdown   of  the  sample  by  sector  please  refer  to  Appendix  3.  

  This   wide   spread   of   sectors   across   the   sample   makes   an   internal   comparison   with   this   sample   size   impossible.   Therefore,   to   control   for   sector-­‐ wide  cyclical  performance  changes  as  well  as  booms  and  busts  there  is  a  need  for   specialized  peer  groups.  These  peer  groups  are  based  on  four-­‐digit  NACE  sector   and   company   size.   Using   the   Orbis   standard   peer   groups   that   consist   of   comparable  international  companies  operating  within  the  same  NACE  sector,  the   peer   group   is   narrowed   down   to   the   10   most   comparable   companies   by   total   assets  in  the  buyout  year.  For  some  companies  it  was  impossible  to  construct  a   representative  peer  group  and  some  peer  group  variables  were  unavailable.  This   means  that  including  peer  group  data  will  limit  the  sample  by  approximately  10-­‐ 13  observations,  depending  on  the  combination  of  variables  used.    

Comprehensive   annual   financial   and   operating   data   is   gathered   on   all   peer   group   companies   for   several   periods.   To   further   control   for   outliers,   the   median  value  of  the  financial  and  operating  variables  of  the  peer  group  is  used.   These   median   values   are   then   matched   with   their   respective   companies   and   together   form   the   ‘peer   group   median’   variables.   These   variables   are   again   reworked  into  the  format  of  ‘pre’  and  ‘t4’.  Peer  group  variables  are  winsorized  to  

the  same  fraction  as  the  portfolio  company  data.  For  descriptive  statistics  on  the   most  important  peer  group  variables  please  refer  to  Appendix  2.  

(17)

4.4  Market  Data  

  Based  on  the  availability  of  data,  the  coverage  period  of  this  research  is   between  2005  and  2014.  Analyzing  one  year  pre-­‐buyout  and  four  years  after  the   buyout  means  the  sample  period  for  this  thesis  is  deals  between  2005  and  2010.   To  control  for  the  obvious  effects  of  the  global  financial  crisis  as  well  as  several   mini-­‐booms   and   mini-­‐busts   surrounding   this   period   there   is   need   for   market   data.   To   proxy   for   these   events   the   FTSE100   index   will   be   used.   This   index   consists   of   the   100   largest   publicly   traded   companies   in   the   UK,   thus   it   should   provide  an  accurate  approximation  of  the  broad  UK  market.  Yearly  returns  are   calculated  on  a  simple  basis  as  the  percent  difference  in  the  index  compared  to   one-­‐year  prior.  These  returns  are  again  reworked  to  fit  the  buyout  timeline  for   each   company   depending   on   the   buyout   year.   The   variable   FTSE   return   is   therefore   the   simple   return   of   the   FTSE   between   ‘pre’   and   the   end   of   ‘t4’.   The  

FTSE  index  scores  are  winsorized  to  the  same  fraction  as  the  portfolio  company   data.    

(18)

5.  Results  

5.1  Performance  Change  Analysis  

  The  regression  analysis  starts  with  the  simple  regression  of  ROA  change   and   the   pre-­‐buyout   Altman   Z-­‐score.   Subsequent   regressions   are   estimated   by   adding  one  variable  at  a  time  to  control  for  outside  effects.  Table  3  contains  the   results  of  these  regressions.  Column  1  is  the  simple  regression  with  just  the  pre-­‐ buyout   Altman   Z-­‐score   and   a   constant.   The   Z-­‐score   has   a   negative   sign   and   is   significant   at   1%.   This   is   congruent   with   the   notion   that   more   inefficiency   (a   lower  Z-­‐score)  has  a  positive  effect  on  the  change  in  performance.  The  R-­‐squared   of  this  regression  is  0.1394.  

 

Table  3  Regression  ROA  Change  and  Pre-­‐Buyout  Altman  Z-­‐score  

Dependent  variable:  Change  in  ROA  

    (1)   (2)   (3)   (4)   Constant   8.72***  (2.88)   10.44***  (3.42)   10.49***  (3.44)   28.9**  (2.26)   Prebuyout  Z   -­‐4.34***  (-­‐3.60)   -­‐3.9***  (-­‐3.96)   -­‐3.91***  (-­‐3.98)   -­‐4.4***  (-­‐4.03)   Ch  in  PGROA1     1.62***   1.61***   1.62***     (3.25)   (3.19)   (3.17)   FTSE  ret       -­‐1.58   -­‐2.28       (-­‐0.28)   (-­‐0.41)   Size2         -­‐1.59         (-­‐1.62)   R-­‐Squared   0.1394   0.2830   0.2832   0.2979   #  obs   132   122   122   122  

Note:  t-­‐stat  with  robust  std.  errors  in  parentheses,  significance  level  *  p<0.1,  **  p<0.05,  ***  p<0.01   1:    Value  at  t4  -­‐  value  pre  

2:  Log(Total  Assets)  

 

  Adding  control  variables  has  a  positive  effect  on  the  R-­‐squared.  Column  2   is   the   regression   with   the   added   sector   control.   This   control   is   the   change   in   median  peer  group  ROA.  The  Z-­‐score  has  a  negative  sign  and  is  still  significant  at  

(19)

1%.  The  sector  control  has  a  positive  sign  and  is  also  significant  at  1%.  This  is  in   line   with   logical   expectations,   if   the   median   peer   group   ROA   goes   up   during   a   period  this  could  point  to  a  boom  in  the  sector,  which  positively  affects  the  ROA   of   a   company   operating   within   that   sector.   The   R-­‐squared   of   this   regression   is   0.2830.  This  jump  is  in  line  with  the  logical  importance  of  controlling  for  sector   conditions  in  analyzing  the  performance  of  a  company.  As  mentioned  before,  the   observations  drop  from  132  to  122  because  of  some  missing  peer  group  values.     The  regression  in  column  3  controls  for  broad  market  conditions  with  the   inclusion   of   the   FTSE100   index   returns.   This   variable   is   not   significant   at   10%   and   the   inclusion   leads   to   a   marginally   higher   R-­‐squared   (0.2832).   It   is   not   directly  obvious  why  including  a  control  for  broad  market  conditions  would  be   insignificant,   especially   considering   the   market   volatility   in   the   sample   period.   However,   a   logical   explanation   could   be   that   the   change   in   median   peer   group   ROA   is   already   sufficiently   dependent   on   market   conditions   to   explain   this   influence.   For   the   purpose   of   completion   the   FTSE   return   is   left   in   because   it   marginally   improves   on   the   explanatory   power   of   the   model   and   leaves   the   significance  of  the  other  variables  unaffected.  

  Column   4   adds   a   pre-­‐buyout   size   control   to   the   regression.   The   peer   group  selection  already  intrinsically  controls  for  size  in  its  calculation.  However,   for  completion  this  inclusion  controls  for  any  remaining  size  and  scale  effects.  As   expected   the   size   control   is   not   significant.   The   Altman   Z-­‐score   and   the   peer   group  control  are  still  significant  at  1%  and  the  R-­‐squared  increases  to  0.2979.     Based   on   the   coefficients   of   regression   3   and   4,   and   the   standard   deviations   reported   in   the   descriptive   statistics,   the   effect   of   pre-­‐buyout   efficiency   on   the   change   in   ROA   is   significant.   A   firm   scoring   one   standard   deviation   lower   in   pre-­‐buyout   Altman   Z-­‐score   has   an   expected   change   in   ROA   7.43%  higher  than  mean-­‐efficiency  firms.  

5.2  Efficiency  Change  Analysis  

  The   regression   analysis   starts   with   the   simple   regression   of   Altman   Z-­‐ score   change   and   the   pre-­‐buyout   Altman   Z-­‐score.   Subsequent   regressions   are   estimated  by  adding  one  variable  at  a  time  to  control  for  outside  effects.  Table  4   contains  the  results  of  these  regressions.  Column  1  is  the  simple  regression  with  

(20)

just   the   pre-­‐buyout   Altman   Z-­‐score   and   a   constant.   The   Z-­‐score   has   a   negative   sign   and   is   significant   at   1%.   This   is   congruent   with   the   notion   that   more   pre-­‐ buyout   inefficiency   (a   lower   Z-­‐score   pre-­‐buyout)   has   a   positive   effect   on   the   change  in  efficiency  at  t4.  The  R-­‐squared  of  this  regression  is  0.2528.    

  The  second  regression  controls  for  the  sector  performance  by  adding  the   change   in   peer   group   median   ROA.   The   control   is   not   significant   and   the   R-­‐ squared  increases  slightly  to  0.2619.  The  sign  and  significance  of  the  pre-­‐buyout   Z-­‐score  is  unaffected.  The  control  is  added  not  because  of  a  specific  direct  link   between   sector   performance   and   firm   efficiency   but   to   control   for   any   distortions.   A   potential   sector   boom   could   lead   to   higher   operating   revenue   and/or   higher   EBIT,   causing   a   higher   Altman   Z-­‐score   without   any   changes   in   operational  efficiency.  

Columns  3,  4,  and  5  add  the  change  in  median  peer  group  Altman  Z-­‐score   in  the  place  of  the  change  in  peer  group  median  ROA.  This  controls  for  sector-­‐ wide   efficiency   changes   such   as   technological   developments   or   cyclical   sector   effects.  These  regressions  omit  the  change  in  peer  group  median  ROA  because  of   possible  multicollinearity,  since  the  ROA  is  a  part  of  the  Altman  Z-­‐score.  Just  as   the   sector   ROA   control,   the   peer   group   change   in   Altman   Z-­‐score   is   not   significant   and   slightly   increases   the   R-­‐squared   to   0.2622.   The   inclusion   of   the   peer  group  Z-­‐score  gives  a  slightly  higher  R-­‐squared  and  its  t-­‐stat  (0.83)  is  also   marginally  higher  than  that  of  the  peer  group  ROA  (0.52).  Therefore,  regressions   4  and  5  include  the  peer  group  Z-­‐score,  but  do  not  include  the  peer  group  ROA   control.  

(21)

Table  4  Regressions  Z-­‐Change  and  Pre-­‐Buyout  Altman  Z-­‐score  

Dependent  variable:  Change  in  Altman  Z-­‐Score  

    (1)   (2)   (3)   (4)   (5)   Constant   0.95***  (4.38)   1***  (4.28)   0.97***  (4.31)   0.98***  (4.26)   (0.61)  0.53   Prebuyout  Z   -­‐0.4***  (-­‐5.27)   -­‐0.4***  (-­‐5.08)   -­‐0.41***  (-­‐5.09)   -­‐0.41***  (-­‐5.06)   -­‐0.39***  (-­‐4.63)   Ch  in  PGROA1     0.01           (0.52)         Ch  in  PG  Z1       0.09   0.09   0.1       (0.83)   (0.80)   (0.91)   FTSE  ret         -­‐0.32   -­‐0.3         (-­‐0.50)   (-­‐0.48)   Size2           0.04           (0.57)   R-­‐Squared   0.2528   0.2619   0.2622   0.2649   0.2667   #  obs   131   121   121   121   121  

Note:  t-­‐stat  with  robust  std.  errors  in  parentheses,  significance  level  *  p<0.1,  **  p<0.05,  ***  p<0.01   1:    Value  at  t4  -­‐  value  pre  

2:  Log(Total  Assets)  

   

  Column   4   adds   a   control   variable   for   the   FTSE100   index   return.   This   because  the  FTSE  control  can  filter  out  broader  market-­‐effects  that  could  affect   the  change  in  Z-­‐score.  The  control  is  not  significant  and  the  R-­‐squared  increases   to  0.2649.  The  sign  and  significance  of  the  pre-­‐buyout  Z-­‐score  is  unaffected.     Lastly,  regression  5  adds  a  size  control.  This  variable  controls  for  possible   scale  effects  not  controlled  for  by  the  Z-­‐score.  The  size  control  is  not  significant   and   the   R-­‐squared   increases   to   0.2667.   Although   the   t-­‐stat   drops   a   little   (from     -­‐5.06   in   regression   4   to   -­‐4.63   in   regression   5),   the   sign   and   significance   of   the   pre-­‐buyout  Z-­‐score  is  unaffected.    

  Based   on   the   coefficient   of   regression   3,   and   the   standard   deviations   reported   in   the   descriptive   statistics,   the   effect   of   pre-­‐buyout   efficiency   on   the   change  in  efficiency  is  significant.  A  firm  scoring  one  standard  deviation  lower  in   pre-­‐buyout  Altman  Z-­‐score  has  an  expected  change  in  Altman  Z-­‐score  0.77  points   higher  than  mean-­‐efficiency  firms.  

(22)

5.3  Credit  and  Collection  Results  

  Table   5   contains   the   results   from   the   regressions   on   the   credit   and   collection   period   benchmarks.   For   both   dependent   variables   the   pre-­‐buyout   benchmark  is  significant  at  1%  with  a  negative  sign.  The  interpretation  of  this  is   that  if  a  company  scored  worse  on  its  pre-­‐buyout  benchmark,  it  is  expected  to   improve  more.  The  pre-­‐buyout  benchmarks  appear  to  be  capable  predictors  of   post-­‐buyout  efficiency  gains  achieved  by  the  company.  The  broad  Z-­‐score  is  not   significant  for  either  regression.  An  explanation  for  this  could  be  that  a  general   measure  such  as  the  Z-­‐score  cannot  account  for  very  specific  inefficiencies  such   as  those  associated  with  the  credit  and  collection  period.  

Table  5  Credit  and  Collection  Period  Regressions  

Dependent  variable:     Change  in  Creditdays   Benchmark  

    Change  in  Collectiondays   Benchmark           (1)   (2)       (1)   (2)   Constant   1.25***  (7.38)   1.44***  (4.83)     0.99***   1.17***     (8.06)   (4.19)   Benchmark  Pre   -­‐0.86***  (-­‐9.28)   -­‐0.87***  (-­‐9.19)     -­‐0.79***   -­‐0.79***     (-­‐11.38)   (-­‐11.29)   Pre-­‐Buyout  Z     -­‐0.06       -­‐0.06     (-­‐0.85)       (-­‐0.84)   R-­‐Squared   0.7063   0.7078       0.7512   0.7512   #  obs   114   114     114   114  

Note:  t-­‐stat  with  robust  std.  errors  in  parentheses,  significance  level  *  p<0.1,  **  p<0.05,  ***  p<0.01  

 

(23)

6.  Robustness  

6.1  Individual  Altman  Variables  

  To  measure  efficiency  this  thesis  uses  the  Altman  Z-­‐score.  As  is  explained   in   section   3.1,   the   Altman   Z-­‐score   is   a   weighed   score   of   measures   of   liquidity,   profitability,  leverage,  solvency,  and  activity.  It  has  the  following  form:  

𝑍 = 0.717 𝑋! + 0.847 𝑋! + 3.107 𝑋! + 0.420 𝑋! + 0.998 𝑋! .   Where:   𝑋! = 𝑊𝑜𝑟𝑘𝑖𝑛𝑔  𝐶𝑎𝑝𝑖𝑡𝑎𝑙 𝑇𝑜𝑡𝑎𝑙  𝐴𝑠𝑠𝑒𝑡𝑠,     𝑋! = 𝑅𝑒𝑡𝑎𝑖𝑛𝑒𝑑  𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑇𝑜𝑡𝑎𝑙  𝐴𝑠𝑠𝑒𝑡𝑠,     𝑋! = 𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠  𝐵𝑒𝑓𝑜𝑟𝑒  𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡  𝑎𝑛𝑑  𝑇𝑎𝑥𝑒𝑠 𝑇𝑜𝑡𝑎𝑙  𝐴𝑠𝑠𝑒𝑡𝑠,     𝑋! = 𝐵𝑜𝑜𝑘  𝑉𝑎𝑙𝑢𝑒  𝑜𝑓  𝐸𝑞𝑢𝑖𝑡𝑦 𝑇𝑜𝑡𝑎𝑙  𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠,     𝑋! = 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔  𝑅𝑒𝑣𝑒𝑛𝑢𝑒 𝑇𝑜𝑡𝑎𝑙  𝐴𝑠𝑠𝑒𝑡𝑠.  

  For   this   thesis   it   sufficed   to   calculate   the   Z-­‐score   with   the   coefficients   suggested   by   Altman   and   use   this   as   a   measure   of   overall   efficiency.   However,   understanding   the   effects   of   the   individual   variables   is   an   important   step   in   determining  the  direction  of  future  research.  Therefore,  this  section  will  perform   some  of  the  core  regressions  of  the  thesis  with  the  individual  Altman  variables,   rather  than  the  broad  Z-­‐score.  The  results  of  these  regressions  can  be  found  in   Table  6  and  Table  7.  For  the  ROA  regression,  the  change  in  peer  group  control  is   the   change   in   median   peer   group   ROA.   For   the   Altman   Z-­‐score   regression,   the   change  in  peer  group  control  is  the  change  in  median  peer  group  Altman  Z-­‐score.  

Table  6,  column  1  contains  the  regression  results  for  the  change  in  ROA   regression   with   the   individual   pre-­‐buyout   Altman   variables.   When   all   of   the   variables   are   included   only   the   EBIT   and   the   Equity   Book   Value   variables   are   significant  at  1%  and  5%  respectively.    The  peer  group  control  is  also  significant   at  1%.    

  Interestingly,   when   the   individual   Altman   variables   are   included,   the   R-­‐ squared   of   the   regression   (0.4438)   is   much   higher   than   in   the   corresponding   regression  using  the  Z-­‐score  (0.2830).  An  explanation  for  this  could  be  that  the   Z-­‐score   is   designed   to   score   a   company’s   default   risk,   and   that   the   constraints   imposed   on   the   joint   effect   of   the   variables   limit   the   explanatory   power   when   looking  at  profitability.  

Referenties

GERELATEERDE DOCUMENTEN

Dit raamplan beschrijft het door alle ULO’s ondersteunde kader waarbinnen voorstellen kunnen worden ingediend voor de opzet, uitvoering, evaluatie en consolidatie van

We try to analyze and understand various aspects of computer entertainment: besides “making new things”, we “analyze the things that we find in the world of computer

Het geweld van deze groepen is echter niet uitsluitend tegen zaydī’s gericht, maar treft ook de coalitie van president Hādī, zoals bij de inname in april 2015 van de

Actual curriculum development (second panel) is preceded by monitoring and analysis of trends in practice, science, and society (first panel).. Discourse about monitoring and

Since from literature it follows that timing of investments is an important performance driver I make an assessment of the different timing features of an individual investment:

This should be interpreted as follows: ranking positively influences the price paid in a club deal, while taking the (existing) relation between ranking and prices in the

Dit onderzoek dient getypeerd te worden als een wetenschappelijk exploratief onderzoek, daar het terrein van Controllersrollen in een buyout een relatief onbekend

Doctoraalscriptie - Matthijs van der Hoeven - Rijksuniversiteit Groningen 60 In sum, better public market conditions lead to high returns on exiting private equity