virtual(reproduction(of(the((migration(flows(generated(by(aiesec( · 1" "...

18
1 Virtual reproduction of the migration flows generated by AIESEC P.G. Battaglia, F.Gorrieri, D.Scorpiniti Introduction AIESEC is an international nonprofit organization that provides services for university students and most of all offer internships in an international playground. The following project focuses on the analysis of the internship system and the migration flows, generated by individuals who want to find an internship abroad. These flows will be based on the combination of both endogenous characteristics of the agents and exogenous variables of the countries in which the AIESEC offices are based. Our world will be divided into four main blocks, virtually representing agent’s origins. The green square will be representing Europe, the brown one Africa, the blue one America and the yellow one Asia. On each block is settled an AIESEC office and the overall population is distributed around the world by a set of commands on the interface ruled by the user. Procedure Overview The first step of our program is representing the AIESEC advertising campaign, during which the organization will try to convince people to apply for the internships offered, the rate of success will be set by the user through a slider on the interface. An additional task of the offices is generating available positions. Indeed, each division will provide a number of available places for individuals coming from each block of origin, these values will have to be set by the user. The second step lies in the feedback influence. Indeed, a percentage of the workers at the end of the internship will provide its own feedback about the country in which it has done the experience and this will modify the preferences on that particular country of the new applying agents. This interaction of individuals who have already left and new potentially applying will condition the decision of each candidate about his or her destination. On the basis of this new modified endogenous variable, the candidates will go to the local AIESEC office and apply for an internship and the reiteration of this process will represent the migration flows.

Upload: others

Post on 10-Aug-2020

9 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

1    

Virtual  reproduction  of  the    migration  flows  generated  by  AIESEC  P.G.  Battaglia,  F.Gorrieri,  D.Scorpiniti  

 

Introduction  AIESEC  is  an  international  non-­‐profit  organization  that  provides  services  for  university  students  and  most  of  all   offer   internships   in   an   international   playground.   The   following   project   focuses   on   the   analysis   of   the  internship  system  and  the  migration  flows,  generated  by  individuals  who  want  to  find  an  internship  abroad.  These   flows   will   be   based   on   the   combination   of   both   endogenous   characteristics   of   the   agents   and  exogenous  variables  of  the  countries  in  which  the  AIESEC  offices  are  based.    

Our  world  will  be  divided  into  four  main  blocks,  virtually  representing  agent’s  origins.  The  green  square  will  be   representing   Europe,   the   brown   one  Africa,   the   blue   one  America   and   the   yellow   one  Asia.  On   each  block   is   settled   an   AIESEC   office   and   the   overall   population   is   distributed   around   the  world   by   a   set   of  commands  on  the  interface  ruled  by  the  user.  

 

Procedure  Overview  The   first   step   of   our   program   is   representing   the   AIESEC   advertising     campaign,   during   which   the  organization  will  try  to  convince  people  to  apply  for  the  internships  offered,  the  rate  of  success  will  be  set  by   the   user   through   a   slider   on   the   interface.   An   additional   task   of   the   offices   is   generating   available  positions.  Indeed,  each  division  will  provide  a  number  of  available  places  for  individuals  coming  from  each  block  of  origin,  these  values  will  have  to  be  set  by  the  user.    

The   second   step   lies   in   the   feedback   influence.   Indeed,   a   percentage   of   the   workers   at   the   end   of   the  internship  will  provide  its  own  feedback  about  the  country  in  which  it  has  done  the  experience  and  this  will  modify   the   preferences   on   that   particular   country   of   the   new   applying   agents.   This   interaction   of  individuals  who  have  already  left  and  new  potentially  applying  will  condition  the  decision  of  each  candidate  about  his  or  her  destination.  On  the  basis  of  this  new  modified  endogenous  variable,  the  candidates  will  go  to  the  local  AIESEC  office  and  apply  for  an  internship  and  the  reiteration  of  this  process  will  represent  the    migration  flows.  

Page 2: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

2    

Main  Procedures  After   having   divided   the  world   into   the   four   blocks,  we   generate   the   four   AIESEC   offices   and   the   global  population  and  then  we  distribute  it  all  around  our  world,  according  to  the  rates  defined  on  the  interface.  Below   is  provided  a  detailed  description  of   the  main  procedures  of   the   simulation,   in  particular  of   those  through  which   are  modified   endogenous   variable   of   the   agents,   in   order   to  make   it  more   clear   for   the  reader  to  understand  the  logic  of  the  simulation.  

To  internship  The  internship  command  is  composed  by    the  contacting  procedure,  that   is  the  organization’s  advertising  campaign,  during  which  AIESEC  reaches  a  certain  percentage  of  the  overall  population  and  convinces  it  to  apply  for  an  internship  abroad.  The  individuals  applying  will  be  easily  recognised  as  they  will  become  white.  Next  we  focus  on  the  internship  selection.  Each  of  the  applying  candidates  will  move  to  his/her  local  AIESEC  office   and  will   randomly   choose   a   destination.  Here   is   an   example   for   the   European   candidates   and   this  procedure  is  repeated  for  each  of  the  four  origins:  

let  Eu  (  x  *  africarate  +  y  *  americarate    +  z  *  asiarate  )  

ask  europeans  [  if  (convinced  =  1  )          [  move-­‐to  office  0  set  color  green  ]  ]    

ask  europeans  with  [convinced  =  1]    [  set  destination  random  int  Eu  abroad_e]  

to  abroad_e      

   if  (destination  <=  x  *  africarate  )  [move-­‐to  office  1]  

   if  (destination  >=  (  x  *  africarate  +  y  *  Americarate  )  )  [move-­‐to  office  2]  

   if  (destination  >  x  *  africarate  and  destination  <  (  x  *  africarate  +  y  *  Americarate  )  )  [move-­‐to  office  3]            

end  

Practically   all   the   applying   candidates  will   randomly   select   an   integer   number   less   or   equal   to   “Eu”   and  according  to  the  belonging  interval,  the  person  will  move  to  a  specific  block.  “Eu”  is  the  sum  of  the  intervals  pertaining  to  each  destination,  multiplied  by  their  growth/decreased  rates  (initially  set  at  1),  which  is  given  by   the   feedbacks.   Initially   each   interval   will   be   of   the   same   size   but   in   a   second  moment   they     will   be  modified   by   the   feedback   rates,   that  will   influence   their   dimension   and   therefore   the   probability   of   the  future  interns  of  choosing  a  country  rather  than  another.  

To  select  This  is  probably  the  most  important  and  complicated  command  of  the  whole  code,  indeed  it  includes  many  procedures  and  it  is  the  main  developer  of  the  system.  The  task  of  this  command  is  to  allow  the  potential  interns,  who   have   not   found   an   available   job   position   in   their   first   destination   choice,   to   look   for   other  possibilities   around   the   world.     Moreover   it   includes   the   feedback   procedure,   which   represents   the  endogenous  variable  of  the  process.  

The  first  step  is  to  rearrange  the  first  wave  of  interns  incoming  in  each  block.  The  individuals  who  initially  have  decided  to  move  to  a  specific  destination,  go   there,  check   if   there  are  available   job  positions  and   if  they  find  an  internship  opportunity  stay  there.  At  the  end  of  this  step    we  count  how  many  free  positions  in  each   block     for   each   country   of   origin   are   left.   Below   is   related   the   command   for   the   Europeans   in   the  AfricanBlock,   note   that   the   same   is   repeated   for   each  of   the  nationalities  of   the   interns,   for   each  of   the  

Page 3: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

3    

blocks.  The  happiness  variable  refers  to  the  satisfaction  of  the  agents,  therefore,  if  the  candidate  finds  an  internship  at   this   step   the  happiness  variable  will   increase  by   two  units,   as   the   intern  has   found  a   job   in  his/her  first  destination  preference.  

   let  europeans_in_africa  count  europeans  with  [convinced  =  1  and  destination  <=  x  *  africarate  ]  

   if  (europeans_in_africa  >  eaf_places)  [  

   ask  n-­‐of  eaf_places  europeans  with  [convinced  =  1  and  destination  <=  x  *  africarate  ]  [  

   move-­‐to  one-­‐of  AfricanBlock  set  pastdestination  2  set  happinesseurope  happinesseurope  +  2    

   set  stage  1  set  free_eaf_places  0  ]  ]      

   if  (europeans_in_africa  <=  eaf_places  )  [      

   ask  n-­‐of  europeans_in_africa  europeans  with  [convinced  =  1  and  destination  <=  x  *  africarate    ]  [    

   move-­‐to  one-­‐of  AfricanBlock  set  pastdestination  2  set  stage  1  set  happinesseurope  happinesseurope  +  2          

   set  free_eaf_places  eaf_places  -­‐  europeans_in_africa  ]]  

In  a  second  moment  we  ask  to  those  interns  who  have  not  found  a  job  in  their  first  destination  choice  to  return   to   their   belonging   block   and   look   for   another   position   somewhere   else.   This   latter   procedure  represents  another  interesting  feature  of  the  code.  Indeed  what  we  do  is  to  compute  how  many  potential  interns   have   come   back,   not   finding   an   internship   in   their   first   destination   choice,   and   how  many   free  available  positions  are  left  all  around  the  world  for  the  individuals  coming  from  each  block.  Later  we  use  an  algorithm  to  embrace  all   the  situations  possible.  Here  are  reported  the  code   lines  for  the  Africans   in  two  cases:  there  are  places  only  in  one  country  and  there  are  places  in  two  countries.  As  always  the  procedure  will  be  repeated  for  each  of  the  four  nationalities,  moreover  we  report  only  the  cases   in  which  there  are  places  in  one  or  two  specific  blocks,  obviously  the  code  is  repeated  for  each  of  the  possible  combinations.  

set  africans_back  count  africans  with  [convinced  =  1  and  stage  !=  1]  

let  available_placesAF      (free_aas_places  +  free_aeu_places  +  free_ausa_places)  

ask  africans  with  [convinced  =  1  and  stage  !=  1]  [  

   if   (free_aeu_places   !=   0   and   free_aas_places   !=   0   and   free_ausa_places   =   0   )       [   set   destination   random   int   (   a   *  europerate  +  c  *  asiarate    )  retry-­‐abroadA1]  

if   (free_aeu_places   =   0   and   free_aas_places   =   0   and   free_ausa_places   !=   0   )     [move-­‐to   one-­‐of   AmericanBlock   set  pastdestination  3  set  stage  1  set  free_ausa_places  free_ausa_places  -­‐  1  set  happinessafrica  happinessafrica  +  1]  

if  (free_aeu_places  =  0  and  free_aas_places  =  0  and  free_ausa_places  =  0)  [ask  africans  with  [convinced  =  1  and  stage  !=  1][set  happinessafrica  happinessafrica  -­‐  1  die]]  

to  retry-­‐abroadA1  

if   (destination   <=   a   *   europerate   and   africans_back   <   free_aeu_places   )   [move-­‐to   one-­‐of   EuropeanBlock   set  pastdestination  1  set  stage  1  set  free_aeu_places  free_aeu_places  -­‐  1  set  happinessafrica  happinessafrica  +  1  ]  

 if   (destination   <=   a   *   europerate   and   africans_back   >=   free_aeu_places   )   [move-­‐to   one-­‐of   EuropeanBlock   set  pastdestination  1  set  stage  1  set  free_aeu_places  0  set  happinessafrica  happinessafrica  +  1]    

Page 4: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

4    

 if  (destination  >  a  *  europerate  and  africans_back  <  free_aas_places)  [move-­‐to  one-­‐of  AsianBlock  set  pastdestination  4  set  stage  1  set  free_aas_places  free_aas_places  -­‐  1  set  happinessafrica  happinessafrica  +  1  ]  

 if   (destination   >   a   *   europerate   and   africans_back   >=   free_aas_places   )   [move-­‐to   one-­‐of   AsianBlock   set  pastdestination  4  set  stage  1  set  free_aas_places  0  set  happinessafrica  happinessafrica  +  1]    

end  

A  chance   is  that  there  are  free  places  for  the  people  coming  from  a  specific  block   in  only  one  country,   in  that  case  a  number  of  interns  equal  to  the  free  available  positions  moves  to  this  new  destination.  Another  possibility  is  that  there  are  free  places  in  two  different  blocks,  in  that  case  the  interns  set  a  new  destination  choice   and   go   to   the   new  destination   chosen.   The   variable   happiness   is   increased   only   of   1   unit,   as   the  interns  do  find  an  job  abroad,  but  not  in  the  block  they  would  have  liked.  

Proceeding  in  the  select  command,  we  find  the  endogenous  part  of  the  code.  Here,  through  a  slider  on  the  interface,  the  user  selects  the  percentage  of  past  interns  who  comes  back  to  its  origin  block  and  provides  a  feedback  about  its  past  experience.    All  the  individuals  are  endowed  with  a  variable  “quality”,  and  for  those  who  come  back  to  provide  a  feedback,  we  randomly  modify  it  in  a  range  of  40  around  100.  Successively  we  sum  the  quality  of  all  the  feedback-­‐giving  people  and  we  compute  the  quality  rate  for  each  block.  This  rate  influences   the  probability   that   the  new  potential   interns   apply   for   a   block   rather   than   another.   Below   is  presented  the  example  of  the  quality  of  the  European  block.  

let  europeansfeedback    round  (  feedback-­‐rate  *  count  europeans)  

to  feedback        

ask  turtles  with  [  pastdestination  !=  0  and  stage  =  1  ][  

   ifelse  random-­‐float  1  >  0.5    [set  quality  quality  +  random  -­‐40]  

   [set  quality  quality  +  random  40]    ]  

ask  turtles  with  [  pastdestination  !=  0  ]  [  set  label  quality  set  label-­‐color  white  ]  

set   qualityinEurope   sum   [quality]   of   asians   with   [   pastdestination   =   1   ]   +   sum   [quality]   of   americans   with   [            pastdestination  =  1  ]  +    sum  [quality]  of  africans  with  [  pastdestination  =  1  ]  

let   pastpersoninEurope   count   asians  with   [   pastdestination   =   1   ]   +   count   americans  with   [   pastdestination   =   1   ]   +              count  africans  with  [  pastdestination  =  1  ]  

ifelse  pastpersoninEurope  !=  0  [    

   let  Europebase  pastpersoninEurope  *  100  

   set  Europerate  qualityinEurope  /  Europebase  ]  [  set  Europerate  1  ]  

 

 

The   last   command   of   the   program   is   called   “El   Farol”,   this   name   comes   from   the   study   of   the   El   Farol  minority   game,   that   has   inspired   this  work.   The   “new”   agents   collect   the   information   provided   by   “old”  agents,   and  according   to   that   improve   the  probability  of  moving   to  a   specific  block   rather   than  another.  Practically  the  agents  can  learn  from  the  experience  of  overcrowding  happened  in  the  past.  

Page 5: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

5    

The  command  is  structured  to  check  if  the  percentage  of  people  who  have  applied  for  a  specific  block  is  too  high.  As  an  example,  we  start  from  the  case  of  population  set  at  400  individuals,  an  applying  rate  at  level  1  and  with  a  given  set  of  available  positions,  which  has  as  final  result  100  interns  in  each  country.  Since  we  are  able   to  modify   the   set  of   available  positions   and   create  more   interesting   results,  we   can   study  what  happens   if   an  higher  percentage  of  people   is  willing   to  do   the   internship   in   a  particular   country.   The   “El  Farol”   command   checks   exactly   this   situation,   moreover   if   in   a   specific   block   there   has   been   an  overcrowding  phenomenon  in  the  last  period,  it  decreases  the  rate  of  choice  linked  to  that  destination.  

As   in   the   El   Farol  we  wanted   to   re-­‐create   the   issue  of   local   information,   focused  on   the   fact   that   in   the  reality  not  all  the  agents  have  the  same  quantity  of  information.  To  achieve  this  result  we  set  up  a  variable  and  we  activate  the  command  only  if  there  is  not  a  loss  of  information  greater  than  the  70%  of  the  total.  Here  is  reported  the  code  of  the  command:  

if   (   internspastdestination1   !=   0   and   internspastdestination2   !=   0   and   internspastdestination3   !=   0   and  internspastdestination4  !=  0  and  lostinformation  <  0.7  )    

[   if   (internspastdestination1  >  0.30  *  (   internspastdestination1  +   internspastdestination2  +   internspastdestination3  +  internspastdestination4  )  [set  europerate  europerate  -­‐  0.05]  

if   (internspastdestination2   >   0.30   *   (   internspastdestination1   +   internspastdestination2   +   internspastdestination3   +  internspastdestination4  )  )  

[set  africarate  africarate  -­‐  0.05]  

if   (internspastdestination3   >   0.30   *   (   internspastdestination1   +   internspastdestination2   +   internspastdestination3   +  internspastdestination4  )  )  

[set  americarate  americarate  -­‐  0.05]  

if   (internspastdestination4   >   0.30   *   (   internspastdestination1   +   internspastdestination2   +   internspastdestination3   +  internspastdestination4  )  )  

[set  asiarate  asiarate  -­‐  0.05]]  

;  3  country  with  places  

if   (   internspastdestination1   =   0   or   internspastdestination2   =   0   or   internspastdestination3   =   0   or  internspastdestination4  =  0  and  lostinformation  <  0.7  )  

[    if  (internspastdestination1  >  0.40  *  (  internspastdestination1  +  internspastdestination2  +  internspastdestination3  +  internspastdestination4  )  )  

[set  europerate  europerate  -­‐    0.05  ]  

if   (internspastdestination2   >   0.40   *   (   internspastdestination1   +   internspastdestination2   +   internspastdestination3   +  internspastdestination4  )  )  

[set  africarate  africarate  -­‐  0.05  ]  

if   (internspastdestination3   >   0.40   *   (   internspastdestination1   +   internspastdestination2   +   internspastdestination3   +  internspastdestination4  )  )  

 [set  americarate  americarate  -­‐    0.05  ]  

if   (internspastdestination4   >   0.40   *   (   internspastdestination1   +   internspastdestination2   +   internspastdestination3   +  internspastdestination4  )  )  

 [set  asiarate  asiarate  -­‐    0.05  ]]  

Page 6: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

6    

 

The  Graphs    

The  Graphs    

Past  destination:  with   this  graph  we  monitor  each  country  and  we  show   the  number  of  people   that  has  done   the   internship   in   there.   For   each   block   we   create   a   variable   called   “Internpastdestination”,   which  assumes   as   value   the   sum  of   agents,  who   has   been   doing   the   internship   in   that   specific   block.   Below   is  reported  an  example  from  the  code:  

set  Internspastdestination1  count  turtles  with  [  pastdestination  =  1]    

Quality  rate:  this  figure  exhibits  the  sum  of  all  the  values  of  the  variable  quality  of  those  people  that,  after  having  done  the  internship,  are  selected  to  give  a  feedback  to  the  new  generation.  This  sum  is  divided  by  the   number   feedback-­‐giving   individuals   multiplied   by   their   initial   quality   value     (100).   As   it   has   been  explained  before,   the  value  obtained  will   influence   the  probability  of  choosing  a  destination  block   rather  than  another  for  the  future  applying  agents.  

Happiness:    It  shows  the  happiness  of  the  agents,  in  terms  of  satisfaction  of  their  desires.  Each  agent  willing  to  leave  gets  2  points  if  he/she  finds  an  available  position  in  the  first  destination  choice.  Otherwise  if  he  is  able  to  find  an  available  position,  but  somewhere  else,  he/she  only  gets  1  point.  Finally  if  the  candidate  was  willing  to  leave,  but  he/she  has  not  found  an  available  job  anywhere,  than  the  happiness  value  decreases  of  one  unit.  The  sum  of   the  happiness  of   individuals   from  each  of   the   four  block,  divided  by   the  number  of  individuals  that  wanted  to  leave,  provides  us  the  graph  happiness  on  the  interface.  

 

 

 

   

 

 

 

 

 

 

Page 7: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

7    

Experiments  

Test  1:  Abundance  and  scarcity  of    feedbacks  The  first  experiment  focuses  on  how  much  does  the  endogenous  procedure  of  giving  feedbacks  affect  the  outcomes  of  the  program.  We  set  the  population  equally  distributed  around  the  world,  the  population  size  (100)  and  the  number  of  available  places  will  be  240  and  fairly  spread  and  the  applying  rate  will  be  set  at  50%.  To  test  the  influence  of  the  feedback  procedure,  we  will  initially  set  the  feedback  rate  at    0,  then  50  %  and  finally  at  100%  and  looking  at  the  graphs  we  will  analyse  how  do  the  past  destinations  change  and  also  how  does  the  quality  rate  evolve  in  each  of  the  three  situations.  

The  first  case,  where  the  feedback  rate  is  set  equal  to  0,  that  means  that  no  past  intern  is  coming  home  to  provide  a  statement  about  the  past  experience  is  presented  below.  

 

The   trend  of   the   past   destinations   is   quite   chaotic  with   huge   and   frequent   variations.  Moreover  we   can  notice   that,   even   though   they   are   random,   because   of   the   El   Farol   procedure,   when   a   specific   country  becomes   to  much  popular  and  gets  problems  of  overcrowding,   successively   it  will   fall  down  because   the    quality  rate  referred  to   it  will  decrease.  The  graph  of  quality  rate   is  also  quite  static.   It  moves  around  the  value  1  as   the   code   is   structured   in   such  a  way   that   the  variable   is   influenced  mostly    by   those  who  are  coming  back  and  providing  a  feedback,  in  this  case  none.  The  variations  are  given  to  the  El  Farol  procedure,  that,  as  we  have  noticed  before,  reduces  the  quality  rate  for  those  blocks  that  have  been  the  most  chosen  destinations  in  the  last  period.  

The  second  test  is  done  setting  the  feedback  rate  at  0.5,  this  means  that  half  of  the  interns  who  have  been  abroad  will  come  back  and  share  their  opinion  with  the  new  applying  individuals.  

Page 8: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

8    

 

In   this   case   we   see   that   still   the   destinations   are   moving   frantically   and   the   variations   have   a   bigger  amplitude,   as   they  are   reflecting   the   trend  of   the  quality   rates.   This  means   that   the   feedback  procedure  influences  strongly  the  choice  of  the  destination  place.  The  quality  rate  trend  is  really  interesting.  We  can  see   that   it   is   lancing   randomly  around  1   in  a   range  of  0.45.  The   trend   is   completely   random  and   reflects  reality   as   it   is   strongly   accidental  whether   to   appreciate  more   or   less   an   experience   of  working   abroad,  depending  on  many  different  variables.  Also  in  this  case  the  El  Farol  procedure  is  applied,  reducing  for  the  interns  the  probability  to  apply  for  a  destination,  that  last  period  has  been  overpopulated.  

Finally  we  analyse  the  case  in  which  all  the  past  interns  are  coming  back  to  provide  their  feedback.  

 

In  this  final  case  the  variations  of  the  quality  rates  are  slightly   less  fluctuating  than  in  the  previous  event.  Interesting  is  also  the  evolution  of  the  rates,  that  look  calmer  and  less  chaotic  than  in  the  precedent  case.  The  variations   indeed  are  softer  and   less  agitated  than  before,  but   in  affecting  the  past  destinations  they  have  a  strong  influence.  This  influence  is  clearly  represented  by  the  increased  frequence  and  amplitude  of  the  deviations  in  the  past  destinations  graph.  

To   summarize   we   can   deduce   that   the   choice   of   the   interns   is   more   random,   when   the   probability   of  choosing  a  block  rather  than  another   is  the  same  for  each  destination.  The  more  the  feedbacks   influence  the   rates,   the  more   the   fluctuations   of   the   quality   rates   will   become   closer   to   the  mean   (1),   the  more  random  will  be  the  destination  choice.  

Test  2:  Abundance  and  scarcity  of  available  internship  positions  We  set  up  the  program  with  a  population  of  500  individuals  equally  distributed  around  the  four  blocks.  The  applying  rate  is  fixed  at  0.5,  therefore  there  will  be  250  people  applying  for  an  internship  abroad  and  the  feedback   rate  will  be  set  at  0.5.   Initially  we  set   the   total  amount  of   free  available   internship  positions  at  240,  fairly  spread  around  the  four  blocks,  practically  we  choose  20  as  input  for  available  positions  for  each  

Page 9: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

9    

of   the   four  blocks   for  people  coming   from  all  around   the  world.  Then  we  reduce   this  number   to  4,   for  a  total  amount  of  48  available  positions  and  then  we  set  it  a  40  for  a  total  amount  of  480  available  positions.  This   simulation   will   enable   us   to   see   what   happens   in   cases   of   extreme   abundance   or   scarcity   of  opportunities.  

 

This   first   case   simulates   a   situation   in  which   the   agents   applying   for   an   internship   are   almost   the   same  number  as  the  available  position.  What  we  get  is  a  graph  in  which  the  agents  are  really  “happy”,  indeed  the  majority  of  them  has  been  able  to  perform  an  internship,  willing  to  do  that.  

The  second    eventuality  is  that  there  are  250  potential  interns  applying,  but  only  48  available  position.  This  is  a  situation  of  extreme  scarcity  of  job  offers.  Obviously  what  we  can  deduce  from  the  figures  is  that  the  happiness  of  the  agents  is  decreasing,  indeed  many  individuals  willing  to  move,  will  not  be  hired.  Predicted  is   also   the   outcome   of   the   past   destinations   graph.   As   there   are   a   few   free   places   and   many   interns  applying,  clearly  the  positions  will  all  be  takes  and  the  representing  line  will  adjust  at  the  maximum  level  of  available  positions  possible.  On  the  other  hand  is  quite  interesting  to  have  a  look  at  the  quality  rates.  The  lines  are  moving  frantically  and  widely.  The  reason  for  this  comes  from  the  fact  that  as  the  interns  providing  the  feedbacks  are  just  a  few,  the  rates  do  not  tend  to  a  smooth  function,  but  on  the  other  hand,  they  have  big  and  twitchy  variations.  

 

Page 10: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

10    

 

Finally   we   come   to   the   last   case,   that   of   abundance   of   available   positions   (   480   versus   240   applying  candidates).  What  we  get  is  a  chaotic  figure  for  the  past  destinations  and  a  smoother  graph  for  quality  rates  variations.  Moreover  we  can  notice  that  the  happiness  figure  has  a  positive  trend,  as  all  the  candidates  will  be  able  to  find  an  available  position,  and  furthermore  to  find  it  in  their  first  destination  choice.  The  slope  of  the  line  is  indeed  steeper  than  in  the  first  case,  as  the  interns  will  be  happier,  given  the  fact  that  they  will  probably  find  a  work  in  their  first  preference,  and  less  probably  be  pleased  by  a  second  choice.  

 

 

Test  3:  Rate  of  Applying    In  this  test  we  analyse  the  influence  of  the  applying  rate  on  the  results.  The  population  is  set  at  500  equally  distributed  in  each  block,  the  feedback  rate  is  set  a  0.5  and  20  available  places  are  given  to  each  block  to    the  candidates  coming  from  each  country,  for  a  total  amount  of  480  positions.  

We  focus  at  first  on  the  case  where  only  the  20%  of  the  population  is  willing  to  take  part  to  an  internship  abroad.  The  outcomes  are   really   similar   to   those  of   the  abundance  of  available  positions,  with  a  positive  

Page 11: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

11    

trend   for   the   happiness   figure,     a   frantically   moving   past   destinations   graph   and   a   quality   rate   graph  characterized  by  not  so  wide  variations.  

 

 

In  the  second  case  the  applying  rate  is  set  at  0.5,  that  means  that  half  of  the  people  in  the  world  wants  to  participate  to  an  AIESEC  project.  The  past  destinations  graph  is  stirring  around  a  constant  value,  that  is  the  maximum  number  of  available  positions.   Indeed   the  applying   interns  are  250  and   the  available  positions  240,   therefore   the   positions   are   all   taken,   apart   from   some   small   variations   in   the   trend   given   by   some  random  influences.  The  quality  rate   is  smoother  than  in  the  previous  case,  because  of  the  fact  that  there  are  more  people  providing  their  feedback.  Finally  the  happiness  figure  maintains  a  positive  trend,  although  with  a  smaller  slope,  given  by  the  reduced  possibility  of  the  interns  to  be  completely  satisfied.  

 

Page 12: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

12    

 

If   the   applying   rate   is   set   at   an   higher   level,   here   0.8,   the   past   destinations   figure   sets   statically   on   a  constant  value,  equal  to  the  number  of  available  destinations.  The  variations  present  in  the  previous  case  are  here  reduced  to  zero,  therefore  the  influence  of  the  random  variables  of  the  programs  is  reduced.  This  case  should  present  similar  characteristic  to  that  of  scarcity  of  available  positions,  but  this  does  not  happen.  Apart   from   the   first   graph   that   is   practically   the   same,   both   the   others   have   different   characteristics:   a  positive  trend  for  the  happiness,  even  with  a  slower  pace  and  smaller  variances  for  the  quality  rates.  

 

Finally  we  have  a  look  at  the  extreme  case  in  which  the  applying  rate  is  set  equal  to  1,  that  implies  that  all  the  people  in  the  world  want  to  take  part  to  an  AIESEC  program.  The  past  destinations  graph  is  statically  set  at   the   maximum   level   of   available   places.   The   quality   rates   are   moving   in   a   really   similar   way   to   the  previous   case  and  also   the  happiness  graph   is   following  a  positive   trend,  but  with  an  even   smaller   slope  than  in  the  previous  case.  

Page 13: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

13    

 

Test  4:  Unequal  distribution  of  the  population  In  this  experiment  we  test  what  happens   if  we  distribute  the  population   in  the  world   in  an  unequal  way.  We   do   it   for   three   chance:   most   of   the   population   on   one   block,   few   people   on   a   block   and   all   the  population  spread  in  only  two  blocks.  The  total  amount  of  people  is  set  at  500,  the  feedback  rate  at  0.5  and  the  applying  rate  at  0.5.  There  are  20  available  internship  positions  in  each  block  for  interns  coming  from  each  of  the  four  blocks,  for  a  total  amount  of  480  jobs.    

First  we  test  what  happens  if  the  majority  of  the  population  is  living  in  one  of  the  four  blocks.  For  doing  this  we  modify  the  distribution  percentage  on  the  interface  and  we  set  the  Asian-­‐rate  at  0.6,  the  American-­‐rate  at  0.2  and  both   the  African-­‐rate  and   the  European-­‐rate  at  0.1.  What  we  get   is  a  graph   in  which   the  past  destinations  graph  presents  huge  variations,  but  it  is  clear  that  there  are  less  people  moving  to  America  and  mostly  to  Asia.  This  is  because  as  there  are  less  people  living  there  than  in  the  other  two  blocks,  there  will  be  less  people  coming  from  abroad  to  fill  the  available  working  positions  in  Asia.  The  second  figure  instead,  shows  the  trend  of  the  happiness  that  is  strongly  positive  for  all  of  the  unpopulated  blocks,  as  it  will  be  easy  for  them  to  satisfy  their  destination  and  working  desires.  On  the  other  hand  the  Asians  are  unhappier  and  the  reason  for  that  can  be  found  in  the  overcrowding  of  their  origin  country.  

Page 14: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

14    

   

A  second  chance  to  be  analysed  is  the  eventuality  of  having  a  block  under-­‐crowded.  This  is  simulated  fixing  the  distribution  rates  at  0.32  for  each  of  the  blocks  but  Europe,  that  is  set  is  at  0.04,  this  means  that  only  the  4%  of  the  overall  population  will  be  born  in  the  European  block.  The  outcome  is  that  obviously  Europe  is  the  most  popular  destination,  because  all  the  places  in  there  will  be  filled  by  interns  coming  from  abroad,  while  being  so  few  the  Europeans  will  not  be  able  to  compensate  the  offers  from  abroad.  In  the  same  way  has  to  be  read  the  happiness  graph.  The  Europeans  do  not  have  to  “fight”  to  find  an  available  place,  they  are  so  few  that  it  is  easy  for  them  to  find  a  proposal  that  fits  their  desires.  On  the  other  hand  for  the  other  countries  of  origin  there  is  more  competition.    

 

Finally  we  look  at  what  happens  in  the  unlikely  event  that  only  the  northern  side  of  the  world  is  populated.  This  is  done  through  setting  the  distribution  rate  of  the  American  and  European  blocks  at  0.5  and  0  for  the  other  two  blocks.  What  we  get   is  that  the  Asian  and  African  blocks  are  preferred  destination,  that  comes  from  the  fact  that,  as  there  are  no  Asians  or  African  to  go  abroad,  the  people  moving  to  America  will  only  be  the  20  Europeans  and  viceversa,  while  the  interns  moving  to  Asia  and  Africa  will  be  the  sum  of  the  two  groups  of  interns  coming  from  the  populated  blocs.  Moreover  the  representing  lines  for  Africa  and  Asia  will  be   overlapping   and   without   variations,   as   well   as   those   of   America   and   Europe.   This   situation   implies  overlapping   lines   also   in   the   happiness   graph.   Indeed   the   happiness   of   Europeans   and   Americans   is   the  same  and  higher  than  the  one  of  the  other  two  blocks.  

Page 15: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

15    

   

Test  5:  Unequal  distribution  of  the  internship  positions  A  similar  test  to  the  one  previously  done  is  that  of  setting  the  available  internship  positions  in  an  unequal  geoghraphical  way.   The  population   is   set   at   500,   the  applying   rate  and   the   feedback   rate  at   0.5   and   the  distribution  rates  for  the  populations  on  each  block  are  all  set  at  0.25.  We  initially  try  to  see  what  happens  if  only  country  wants  incoming  interns  from  abroad.  For  doing  this  we  set  the  available  places  in  Africa,  for  interns  coming  from  each  part  of  the  world  at  20,  for  a  total  amount  of  60  places,  and  we  set  0  in  all  the  other  countries.  What  we  obtain   is  a  graph   in  which  all   the   lines  referred  to  past  destinations  other  than  Africa  are  equal  to  0.  The  graph  for  Africa  is  steadily  set  at  the  number  of  available  positions,  both  because  this  number  of   job  offered   is  small   if   compared  to   the  candidates  and  also  because  all   the   interns  of   the  world  are  trying  to  go  there,  therefore  the  positions  are  all  taken.  The  quality  rate  of  each  block  is  also  not  moving,  as  there  is  no  interns  who  has  been  there  and  therefore  none  is  coming  back  to  provide  feedbacks.  Only  the  African  quality  rate   is  changing,  but  we  see  that  the  trend  is  that  of  the  reducing  the  variations,  therefore  we  expect  it  to  be  assimilated  to  a  the  constant  value  of  the  other  lines  in  the  long  period.  Finally  the  happiness  graph  shows  clearly  that  the  Africans  are  the  unhappiest  in  the  world.  Indeed  they  can  not  take  part  to  any  internship  abroad  and  therefore  they  are  strongly  unsatisfied.    

Page 16: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

16    

 

In  the  following  example  we  analyse  the  case  in  which  one  country  does  not  want  any  incoming  intern  from  abroad.  We  do   this   setting   30   available  positions  on  each  block   for   candidates   coming   from  each  of   the  other  blocks,  but  in  Africa,  where  there  are  no  available  job  positions.  The  outcome  is  a  situation  in  which  the  past  destination  figure  is  quite  chaotic  for  each  of  the  blocks,  but  for  Africa,  were  the  graph  is  constant  at  value  0.  That  is  because  none  can  go  there.  The  quality  rate  graph  shows  small  variations  for  each  of  the  country   in  which   is  possible  to  find  an   internship  and  a  constant  value  for  Africa.  The  happiness   is  clearly  higher  for  Africans  than  for  others  to  be  satisfied,  as  the  others  miss  the  chance  to  move  to  Africa.    

 

Page 17: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

17    

Test  6:  Racism    

This  final  test  focuses  on  the  contingent  presence  of  racism,  therefore  we  simulate  what  can  happen  if  no  block   wants   interns   coming   from   a   specific   block,   for   example   Europe.   We   set   up   a   population   of   500  individuals  equally  distributed  in  the  4  blocks.  The  applying  rate   is  set  at  0.5  as  well  as  the  feedback  rate.  Each  of  the  input  boxes  is  set  at  20,  but  the  ones  referred  to  available  positions  for  Europeans,  that  are  set  equal  to  0,  the  total  amount  of  available  places  will  be  180,  of  which  0  destined  to  Europeans.  

   

What  we  can  infer  from  these  charts   is  that  obviously  the  value  around  which  the  Europe  line  sets   in  the  past  destination  graph  is  bigger  than  for  the  other  blocks.  Indeed  Europe  is  accepting  incoming  interns  from  all   over   the   world,   while   the   other   three   blocks   are   refusing   European   interns   and   offering   the   same  number  of  positions  to  incoming  workers  of  other  origins.  The  happiness  of  Europeans  obviously  follows  a  negative   trend,   as   they   have   no   chance   to   leave   for   an   internship   abroad   and   therefore   to   satisfy   their  desires.    If  we  leave  the  same  conditions,  but  we  change  the  number  of  incoming  interns  from  nationalities  different  from  Europe,  therefore  we  leave  the  inputs  to  go  in  Europe  al  set  at  20,  while  the  inputs  of  other  blocks  equal   to  0   for  Europeans  and  to  30  for  the  other  nationalities,   for  a  total  amount  of  240  available  positions.  This  option  makes  the  model  more  realistic,  as   ideally   if  a  block  needs  60  interns,  but  does  not  want  Europeans,  will  hire  more  people  from  other  nationalities.  What  we  get  is:  

Page 18: Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC( · 1" " Virtual(reproduction(of(the((migration(flows(generated(by(AIESEC(P.G."Battaglia,"F.Gorrieri,"D.Scorpiniti"

18    

 

These   are   very   chaotic   graphs,   in   which   the   past   destinations   are   doing   huge   deviations,   but   with   the  feeling  that  the  Europe  line  is  still  overlooking  the  others.  The  reason  might  be  the  wider  acceptance  of  the  block  that  makes  it  easier  in  a  probabilistic  way  to  choose  to  go  there  or  the  influence  of  the  feedback  rate  on  the  decision  of  the  future  interns.  The  quality  rate  graph  is  more  regular  and  with  smaller  variations  that  in  the  previous  case.  To  conclude  the  happiness  graph  keeps  the  same  feature,  with  a  negative  trend  for  the  happiness  of  Europeans  and  a  positive  one  for  the  happiness  of  the  other  interns.  Moreover  the  slope  of  the  graph  for  the  other  three  remaining  blocks  is  steeper,  as  being  the  number  of  available  internships  increased,  it  is  easier  for  them  to  find  a  internship  abroad  and  therefore  their  happiness  increases.  

References    

1. Emergence of Cooperation and Organization in an Evolutionary Game. CHALLET, D., ZHANG, Y.C., 1997

2. Gossip, Sexual Recombination and the El Farol Bar: modelling the emergence of heterogeneity, Journal of Artificial Societies and Social Simulation. EDMONDS, B., 1999

3. Seeing the light at El Farol. Complexity Vol. 1 CASTI, J.L., 1996.

4. Shedding light on El Farol. Physica A: Statistical Mechanics and its Applications , Volume 332, Pages 469–482 CHALLET, D., MARSILI, M., OTTINO, G., 2004.