It's the economy, stupid

It’s  the  economy,  stupid!  

David  Gross,  Holo  Zheng,  IMBA,  Bar-­‐Ilan  University,  2014-­‐2015  

!Introduc4on  

As  much  as  naConal  security  is  considered  to  be  the  first  duty  of  any  government,  naConal  elecCons  

are  generally  decided  on  the  issue  of  economic  competence.  This  was  most  clearly  seen  in  the  UK  

General  ElecCon  of  1945,  where  aNer  prosecuCng  the  Second  World  War  successfully,  it  was  widely  

expected  that  Winston  Churchill  was  would  be  re-­‐elected  with  a  handsome  margin.  Not  only  did  the  

Labour  party  win,  but  it  did  so  in  a  landslide,  indicaCng  that  the  public  wanted  economic  issues  to  be  

placed  at  the  top  of  the  naConal  agenda.  

In  1979,  a  similar  phenomenon  occurred,  where  aNer  years  of  insCtuConal  obstrucCon,  strikes  and  

poliCcal  chicanery  by  the  trade  unions,  Margaret  Thatcher  was  returned  as  Prime  Minister  with  a  

specific  mandate  to  reform  the  economic  landscape  in  the  UK.  

The  conduct  of  the  Labour  Government  between  1997-­‐2001  in  fiscal  and  economic  terms  went  a  

long  way  in  burnishing  its  poliCcal  credibility  when  the  BriCsh  public  were  asked  as  to  whether  the  

government  was  behaving  competently  on  economic  issues,  even  when  it  when  it  was  much  more  

unpopular  overall.  

The  General  ElecCon  in  Britain  this  year  will  see  a  stark  choice  between  a  ConservaCve  Party  that  has  

brought  the  deficit  down  by  a  third,  in  the  main  by  cuXng  spending,  with  a  Labour  Party  that  will  

seek  to  raise  taxes  on  the  wealthier  segments  of  society  and  increase  public  spending.  The  laYer  are  

already  facing  an  uphill  baYle  in  terms  of  repairing  the  damage  done  to  their  economic  credibility  

aNer  losing  the  last  elecCon.    

The  primary  importance  of  economic  competence  has  also  taken  hold  in  Israel  over  the  past  five  

years.  In  the  summer  of  2011,  there  were  unprecedented  civil  protests  as  to  the  cost  of  living  in  the  

�1

country.  It  was  the  primary  subject  of  naConal  elecCons  in  2013,  where  the  surprise  of  the  elecCon  

was  the  nominally  centrist  party,  Yesh  ACd,  gain  nineteen  seats  as  the  second  largest  party  in  the  

Knesset.  Their  elecCon  campaign  was  premised  on  fiscal  discipline,  reducing  the  deficit  and  doing  

away  with  budgetary  imbalances  that  had  accrued  because  of  certain  sectors  of  Israeli  society  to  the  

detriment  of  the  middle  class.  

The  upcoming  naConal  elecCons  look  to  also  be  centred  on  economic  issues,  where  the  rise  of  yet  

another  centrist  party,  Koolanu,  led  by  a  former  CommunicaCons  Minister,  Moshe  Kachlon,  who  

previously  succeeded  in  introducing  compeCCon  into  the  mobile  phone  market  and  reducing  costs  

by  90%.  

Therefore,  it  would  be  sensible  to  suggest  that  having  a  good  reputaCon  for  economic  competence  

would  be  one  of  if  not  the  foremost  factor  for  conCnued  electoral  success.    

!Review  of  Literature  

It  has  become  a  staple  of  poliCcal  life  to  follow  the  trends  of  polls.  Throughput  the  world  polling  

firms,  market  research  companies  and  similar  organisaCons  have  made  it  their  business  to  track  

public  opinion  on  mulCple  levels  and  planes.  This  data  has  become  increasingly  valuable  for  

predicCng  all  manner  of  poliCcal  happenings,  which  in  turn  has  a  massive  impact  on  economic  

stability  and  the  ability  to  rely  on  that  stability  in  the  long  term.  AggregaCng  this  data  has  been  

especially  helpful  to  certain  companies  in  correctly  predicCng  the  outcomes  of  the  2008  and  2012  US  

PresidenCal  ElecCons  and  the  2014  midterm  elecCons.  The  best  exponent  of  this  is  the  founder  of  

the  FiveThirtyEight.com  website,  Nathan  Silver.  1

In  terms  of  designing  the  method  of  data  collecCon,  one  is  slightly  spoilt  for  choice  with  respect  with  

the  possible  ways  that  we  could  construct  a  mechanism  to  collect  sufficient  data.  Having  said  that  

�2

 Retrieved  from  hYp://fivethirtyeight.com/interacCves/pollster-­‐raCngs/  9/1/20151

any  answer  we  have  to  the  hypothesis  will  be  as  a  result  of  construcCng  a  correlaConal  study  as  

opposed  to  an  experimental  one.    

YouGov  is  probably  the  largest  internet  market  research  company  in  the  United  Kingdom.  In  a  recent  

poll  it  took  between  December  17th  and  18th  2014  as  to  voCng  intenCons,  it  took  a  sample  of  1981  

adults  in  Britain.  Clearly,  taking  a  sample  is  far  more  realisCc  proposiCon  than  polling  an  enCre  2

populaCon  of  over  sixty  million  people.  In  addiCon,  the  sample  was  three  Cmes  larger  than  the  

minimum  size  of  a  sample  (600)  for  it  to  be  considered  significant.  The  main  findings  can  be  seen  

below:  

!

!!!One  of  the  common  of  features  of  all  polls  that  are  taken  across  all  firms  is  that  the  upper  row  takes  

into  account  the  strength  of  intenCon  to  vote  in  the  next  elecCon,  meaning  that  the  raw  values  (in  

the  second  row)  can  be  set  against  each  other.  

Westminster VI 2010 Vote Gender Age

Total Con Lab Lib

Dem

UKIP Con Lab Lib

Dem

Male Female 18-24 25-3

9

40-59 60+

1981 482 566 99 259 591 465 407 961 1020 236 501 678 567

1981 449 590 94 264 559 519 410 998 983 155 385 847 594

�3

 Retrieved  from  hYp://cdn.yougov.com/cumulus_uploads/document/6s63ripmxv/YG-­‐Archive-­‐Pol-­‐Sun-­‐2

results-­‐181214.pdf  10/1/2015

Another  leading  market  research  company  that  maintains  similar  tracking  polls  is  ICM,  which  usually  

publishes  in  the  Guardian  newspaper.  Their  methodology  is  broadly  similar,  though  their  sample  is  

considerably  smaller  than  YouGov’s. *  3

!

!!!What  is  clear  from  both  of  the  above  polls  is  that  there  are  a  considerable  amount  of  variables  

simply  in  the  collecCon  of  voCng  intenCons.  

These  polls  however,  are  not  sufficient  to  work  out  what  the  underlying  reasons  to  those  intenCons  

are.  VoCng  intenCons  are  generally  derived  what  the  primary  issues  on  the  naConal  agenda  at  the  

moment  and  which  party  is  perceived  to  be  best  placed  or  equipped  to  meet  the  challenges.  The  

graphic  below  shows  how  events  moved  electoral  paYerns  in  the  year  following  the  Yom  Kippur  War  

and  the  subsequent  arms  embargo:  4

! Westminster VI 2010 Vote Gender Age

Total Con Lab Lib

Dem

UKIP Othe

r

Con Lab Lib

Dem

Male Female 18-24 25-34 35-64 65+

1001 195 222 53 93 70 262 229 102 490 511 122 164 509 206

1001 158 202 61 91 70 214 179 131 471 530 65 111 532 293

�4

 Retrieved  from  hYp://www.icmunlimited.com/data/media/pdf/2014_dec_guardian_poll.pdf  10/1/2015  3

*The  following  table  was  edited  with  geographical  informaCon  not  displayed.

 Retrieved  from  hYps://www.ipsos-­‐mori.com/researchpublicaCons/researcharchive/3455/EconomistIpsos-­‐4

MORI-­‐September-­‐2014-­‐Issues-­‐Index.aspx#gallery[m]/1/  11/1/2015

!

!

!The  subsequent  graphic  seemingly  shows  that  the  economy  dropped  to  second  place  in  September  

2014  in  the  runup  to  the  UK  General  ElecCons:  5

!

�5

 Ibid.5

While  this  would  seem  to  indicate  that  the  economy  was  not  highest  in  the  public’s  consciousness,  

the  reality  is  that  much  of  today’s  scepCcism  is  of  immigraCon  comes  as  a  result  of  a  saturated  

employment  market,  strain  on  public  services  and  the  inevitable  higher  taxes  that  would  follow  that  

would  be  set  to  alleviate  the  strain,  which  comes  back  to  economic  credibility  of  the  government  of  

the  day.  

!

�6

!!Design  

In  simple  terms,  the  goal  of  this  study  is  to  see  if  one  independent  variable  i.e.  economic  

competence  could  be  idenCfied  to  predict  or  even  dictate  the  dependent  variable  –  victory  in  a  

naConal  elecCon.  In  theory,  this  should  be  able  to  be  seen  by  a  simple  T-­‐test:  By  seeing  which  

poliCcal  party  had  more  credibility  on  the  economy,  we  would  then  see  whether  that  would  be  

mirrored  into  an  elecCon  victory.  The  nature  of  naConal  polls  mean  that  it  would  be  a  two  group  

independent  T-­‐test,  as  respondents  express  a  preference  for  a  party.    

On  a  conceptual  level,  “economic  competence”  is  a  broad  and  amorphous  topic.  In  addiCon,  it  is  only  

a  small  proporCon  of  the  public  are  really  in  a  posiCon  to  speak  with  authority  on  economic  maYers  

and  the  relaCve  competences  of  the  various  parCes,  when  in  fact  there  is  a  parCcular  sub-­‐category  

within  the  gamut  of  economic  state  of  the  country  in  quesCon,  such  as  unemployment,  career  

prospects  and  overextended  government  borrowing.  

For  example,  an  Ipsos-­‐Mori  poll  in  the  United  Kingdom  takes  a  poll  every  month  as  to  voters  opinions  

regarding  the  importance  of  various  issues.  In  May  2010,  the  month  in  which  the  General  ElecCon  

was  held,  71%  of  respondents  held  that  the  economy  was  of  the  most  naConal  importance.  On  the  6

other  hand,  when  quesConed  as  to  whether  the  cost  of  living,  food  prices  or  inflaCon  were  of  

primary  importance,  the  number  of  respondents  who  agreed  only  numbered  around  10%.  

Yet  this  is  perhaps  too  simplisCc  a  measure  to  really  be  able  to  say  with  any  certainty  whether  this  

was  the  only  reason  or  even  a  primary  reason.  It  would  be  easy  to  claim  that  incompetence  of  one  

party  would  result  in  a  de  facto  support  for  the  other.  

�7

 Retrieved  from  hYps://www.ipsos-­‐mori.com/researchpublicaCons/researcharchive/2905/Issues-­‐Index-­‐2012-­‐6

onwards.aspx?view=wide#2010  11/1/2015

This  requires  us  to  suggest  a  more  wide  ranging  method  of  saCsfying  our  hypothesis.  The  next  

method  that  comes  to  mind  is  the  Analysis  of  Variance  or  ANOVA  method.  The  ANOVA  method  seeks  

to  where  there  are  three  opCons  or  condiCons  of  the  independent  variable.  In  this  case  it  would  be  a  

parCcular  poliCcal  factor,  idenCfied  by  the  public  to  be  the  most  important  issue  of  the  day,  which  in  

turn,  would  dictate  the  outcome  of  the  elecCon,  the  dependent  variable.  The  poll  below  was  

compiled  in  response  to  the  following  quesCon:  

“With  the  Economy  facing  the  challenges  it  does  in  the  months  ahead,  who  do  you  trust  more  to  

steward  the  economy  at  the  next  General  ElecCon,  David  Cameron  and  George  Osborne  –  or  Ed  

Miliband  and  Ed  Balls?”    7

The  breakdown  between  all  affiliated  voters  showed  that  all  voters  not  affiliated  with  the  Labour  

Party  and  therefore  “Ed  Miliband  and  Ed  Balls”  all  perceive  the  current  Prime  Minister  and  

Chancellor  of  the  Exchequer  to  be  more  credible  on  the  economy.  While  the  outcome  of  the  elecCon  

has  yet  to  be  decided,  the  fact  that  the  Labour  Party  is  not  seen  as  credible  on  the  economy  means  

that  they  are  seriously  disadvantaged  when  it  comes  to  implemenCng  an  elecCon  campaign  that  will  

engage  with  this  issue.    

�8

 Retrieved  from  hYp://lordashcroNpolls.com/wp-­‐content/uploads/2014/10/ANP-­‐summary-­‐140627.pdf  7

18/1/2015

�  

As  the  table  below  shows,  the  first  response  to  the  poll  is  as  follows  

�  

A  factorial  ANOVA  study  would  also  be  of  parCcular  use  to  poliCcal  parCes  when  deciding  the  

direcCon,  tone  and  content  of  their  campaign.  Focussing  on  a  parCcular  issue  which  the  public  are  

�9

not  either  interested  in  or  aware  of  to  the  detriment  of  other  more  central  issues  has  proven  criCcal  

in  the  outcome  of  an  elecCon  victory.  A  prime  example  of  this  would  be  to  see  the  outcome  of  the  

UK  General  ElecCon  in  the  2001.  The  ConservaCve  Party,  having  lost  the  previous  elecCon  

resoundingly,  based  their  enCre  campaign  on  “Saving  the  Pound”.  The  response  by  the  BriCsh  public  8

was  unequivocal :  9

!

!!!What  was  even  more  impressive  about  this  victory  was  that  the  Labour  Party  succeeded  in  increasing  

an  already  formidable  majority  because  they  campaigned  on  increased  funding  for  public  services  

while  not  overtly  raising  taxes,  and  more  specifically  not  raising  income  tax,  clearly,  a  much  more  

posiCve,  aYracCve  proposiCon.  In  the  upcoming  UK  General  ElecCon,  Ipsos  Mori,  who  have  been  

tracking  the  relaCve  importance  of  issues  at  that  moment,  found  that  the  state  of  the  economy  was  

not  the  most  important  issue  of  the  day,  but  rather  immigraCon.  For  the  purposes  of  our  10

discussion,  what  was  even  more  interesCng  was  that  the  poll  saw  that  the  state  of  the  NaConal  

Health  Service  on  par  as  the  state  of  the  economy.  This  has  seen  the  Labour  Party,  who  suffer  from  

Seats Votes % Share

Labour 412 10,724,895 40.7

Conservative 166 8,357,622 31.7

Liberal Democrats 52 4,812,833 18.3

�10

 Retrieved  From  hYp://news.bbc.co.uk/1/hi/uk_poliCcs/1163591.stm  18/1/20158

 Retrieved  from  hYp://www.ukpoliCcal.info/2001.htm  18/1/20159

 Retrieved  from  hYps://www.ipsos-­‐mori.com/researchpublicaCons/researcharchive/2905/Issues-­‐Index-­‐2012-­‐10

onwards.aspx?view=wide  18/1/2015

low  credibility  regarding  the  state  of  the  economy,  focus  on  healthcare  as  a  means  of  gaining  a  

poliCcal  advantage.  The  most  important  issue  as  seen  in  that  same  poll  was  immigraCon.  One  11

important  point  when  discussing  the  role  of  immigraCon  in  the  context  of  an  elecCon  campaign  is  

that  the  arguments  for  and  against  are  almost  always  founded  on  economic  arguments,  whether  as  a  

posiCve  in  that  many  immigrants  in  the  European  Union  context  would  do  various  menial  jobs  that  

BriCsh  people  would  turn  their  noses  up  at.  AlternaCvely,  others  would  argue  that  too  much  

immigraCon  puts  a  strain  on  public  services,  as  well  as  prevenCng  naCve  ciCzens  from  entering  the  

job  market,  even  at  the  most  basic  level.  In  either  instance,  it  might  be  argued  that  all  of  these  

concerns  really  speak  to  several  facets  regarding  overall  economic  prosperity,  whether  it  be  job  

security,  unemployment  or  concern  about  the  strain  of  public  services.  InacCon  regarding  

immigraCon  over  the  past  twenty  years  has  seen  the  rise  of  smaller  parCes  that  have  sprung  up  in  

reacCon,  and  both  the  ConservaCve  and  Labour  parCes,  despite  it  being  the  most  important  issue,  

have  done  their  utmost  not  to  engage  the  issue.    

We  see,  therefore,  that  the  issues  of  the  day  have  enormous  influence  to  the  content  of  the  poliCcal  

agenda  of  poliCcal  parCes  and  interact  in  mulC-­‐layered  and  someCmes  counter-­‐intuiCve  ways.    

!In  conclusion,  having  treated  various  opCons,  our  hypothesis  should  be  phrased  as  follows:  

In  order  to  correctly  determine  electoral  victory,  one  has  to  idenCfy  the  most  important  electoral  

issue  of  the  moment  which  will  have  most  bearing  on  the  public’s  choice.  

Necessarily,  the  need  to  weigh  up  various  opCons  means  that  we  are  best  served  by  insCtuCng  a  

Factorial  ANOVA  test.  

!!

�11

 Ibid.11

!!Method  

We  are  using  two  tests  in  this  design,  independent  T-­‐test  and  ANOVA.    

As  for  independent  T-­‐test,  we  want  to  see  if  one  independent  variable  i.e.  economic  competence  

could  be  idenCfied  to  predict  the  dependent  variable  – victory  in  a  naConal  elecCon.    To  

operaConalize  it,  we  firstly  choose  two  parCes  with  different  economic  competence.  Secondly,  we  

use  the  group/party’s  ranking  in  pools  as  an  ordinal  measurement  of  its  economic  competence,  

rather  than  its  real  power,  which  is  hardly  measurable,  over  specific  economic  issues.  Thirdly,  we  

measure  elecCon  victory  by  how  many  votes  each  party  has.  Last  but  least,  we  gather  data  from  a  

certain  number  of  elecCons  during  one  specific  Cme  duraCon,  and  set  the  criteria.  If  the  significance  

level  is  lower  than  5%,  then  we  call  it  significant,  or  in  another  word,  the  economic  competence  can  

be  idenCfied  to  predict  the  victory  in  a  naConal  elecCon.  In  the  process  P  value  and  t  raCo  will  be  

calculated.  

To  simulate  a  more  realisCc  scenario,  we  also  use  two-­‐way  ANOVA,  which  enables  us  to  find  out  if  

elecCon  result  is  responsive  to  interacCon  between  two  independent  variables  e.g.  security  concerns  

and  economic  competence.  By  so  doing  we  divide  the  total  variability  among  values  into  four  

components.  Prism  tabulates  the  percentage  of  the  variability  due  to  interacCon  between  the  row  

and  column  factor,  the  percentage  due  to  the  row  factor,  and  the  percentage  due  to  the  column  

factor.  The  remainder  of  the  variaCon  is  among  replicates  (also  called  residual  variaCon).  To  

operaConalize  it,  we  firstly  choose  3  parCes  of  different  levels  of  economic  competence  ranging  from  

very  low  to  middle  level  to  very  high.  In  addiCon,  we  label  their  security  concern  into  different  level,  

also  according  to  the  pool,  and  gather  the  elecCon  results  during  a  certain  period  of  Cme.  By  so  

doing  we  come  up  with  6  groups  of  numbers,  which  shows  the  interacCon  between  economic  

concerns,  e.g.  party  with  highest  economic  competence  and  the  lowest  security  concern  is  more  

�12

successful  in  elecCon  than  party  with  slightly  lower  economic  competence  and  a  very  high  security  

concern.  In  the  process  an  ANOVA  table  will  be  formed  and  P  value  will  be  calculated.    

Implica4ons    

The  finding  we  have  from  the  tests  will  help  us  understand  if  the  economic  competence  of  a  party  

has  a  decisive  influence  on  the  elecCon  result  and  whether  the  interacCon  between  economic  

competence  and  the  security  concern  of  a  party  can  influence  the  elecCon  result.  This  will  not  just  

help  the  party  make  beYer  strategic  decision  on  what  their  image  should  be  built  on,  but  also  making  

pracCcal  decisions  how  to  split  a  limited  budget  between  different  issues.  Moreover,  knowing  the  

interacCon  between  two  factors  can  raise  the  cost  efficiency  of  a  coherent  campaign,  by  using  one  

factor  fostering  the  other  or  at  least  stop  it  from  being  an  obstacle  for  the  other  to  thrive.

�13