Explaining Differences Between the Expected and Actual Duration Until Return Migration: Economic Changes

an den Be g, Ge a d J.; Weynand , Michèle A. A icle Explaining Di e ences Be ween he Expec ed and Ac ual Du a ion Un il Re u n Mig a ion: Economic Changes Schmolle s Jah buch – Jou nal o Applied Social Science S udies. Zei sch i ü Wi scha s- und Sozialwissenscha en P o ided in Coope a ion wi h: Duncke & Humblo , Be lin Sugges ed Ci a ion: an den Be g, Ge a d J.; Weynand , Michèle A. (2013) : Explaining Di e ences Be ween he Expec ed and Ac ual Du a ion Un il Re u n Mig a ion: Economic Changes, Schmolle s Jah buch – Jou nal o Applied Social Science S udies. Zei sch i ü Wi scha s- und Sozialwissenscha en, ISSN 1865-5742, Duncke & Humblo , Be lin, Vol. 133, Iss. 2, pp. 249-261, h ps://doi.o g/10.3790/schm.133.2.249 This Ve sion is a ailable a : h ps://hdl.handle.ne /10419/292409 S anda d-Nu zungsbedingungen: Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen Zwecken und zum P i a geb auch gespeiche und kopie we den. Sie dü en die Dokumen e nich ü ö en liche ode komme zielle Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich machen, e eiben ode ande wei ig nu zen. So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen (insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en, gel en abweichend on diesen Nu zungsbedingungen die in de do genann en Lizenz gewäh en Nu zungs ech e. Te ms o use: Documen s in EconS o may be sa ed and copied o you pe sonal and schola ly pu poses. You a e no o copy documen s o public o comme cial pu poses, o exhibi he documen s publicly, o make hem publicly a ailable on he in e ne , o o dis ibu e o o he wise use he documen s in public. I he documen s ha e been made a ailable unde an Open Con en Licence (especially C ea i e Commons Licences), you may exe cise u he usage igh s as speci ied in he indica ed licence. h ps://c ea i ecommons.o g/licenses/by/4.0/ Explaining Di e ences Be ween he Expec ed and Ac ual Du a ion Un il Re u n Mig a ion: Economic Changes By Ge a d J. an den Be g and Michèle A. Weynand * Abs ac This pape explo es he di e ence be ween in en ions and ealiza ions in e u n mig a- ion wi h he help o a du a ion model. Using he SOEP he esul s lend suppo o he ac ha people use simpli ying heu is ics when ying o o ecas he u u e; hei e u n in en ions indica e bunching in heaps o 5 yea s. Along hese lines we ind ha mig a ed indi iduals sys ema ically unde es ima e he leng h o hei s ay in he ecei ing coun y. We ind ha he di e ence dec eases he olde one ge s, bu is la ge he mo e disad an- aged one eels due o ones o igin as an example. The obus ness checks show ha he esul s do no hinge on a single de ini ion, o se o explaining a iables. The consis ency in he unde es ima ion may ha e impo an policy and modeling implica ions. JEL-Classi ica ion: C41, D84, F22 1. In oduc ion This pape explo es he ac ha mig a ed indi iduals unde es ima e he leng h o hei s ay in he ecei ing coun y. “Hedonic o ecas ing” e e s o he e o s ha indi iduals make in p edic ing changes in hei as es and eelings in he psychological li e a u e (see e.g; Read / an Leeuwen, 1998; Gilbe / Michael/Gill/Wilson, 2002; Nisbe / Kanouse 1968). The eade is p esen ed wi h e idence o a o ecas ing e o and con incing s a is ics p o ing ha i is no jus simple noise. Loewens ein / O’Donoghue/Rabin (2003) ha e de- ined he concep o p ojec ion bias as he sugges ion ha people unde s and Schmolle s Jah buch 133 (2013), 249 –262 Duncke & Humblo , Be lin Schmolle s Jah buch 133 (2013) 2 *We hank he edi o Nicolas Zieba h and an anonymous Re e ee o e y help ul commen s and sugges ions. We g ea ly app ecia e Anne e Be gemann and Da id Ca d’s help o ge access o he da a. We a e g ea ul o Ma hew Rabin, S e ano DellaVigna, And ea Webe , Lena Janys, Johannes Schoch, Ve a Moli o and semina pa icipan s in he junio esea ch dialogue a he Uni e si y o Mannheim, SMYE 2012 and SOEP 2012, o many help ul commen s on an ea lie e sion o his pape . OPEN ACCESS | Licensed unde | h ps://c ea i ecommons.o g/abou /cclicenses/ DOI h ps://doi.o g/10.3790/schm.133.2.249 | Gene a ed on 2023-01-16 13:36:55 he quali a i e na u e o changes in hei as es, bu unde es ima e he magni- ude o hese changes.1 Looking a e u n mig a ion and he expec a ion o e u n ou p io is ha people unde es ima e hei a achmen o he coun y o mig a ion –when i s mo ing away om home one compa es e e y hing o home. Mos o he ime he cul u e in he coun y o mig a ion will be di e en , one will no know a lo o people and one may no e en ha e amily in he mig a ing coun y. All hese hings a e examples o wha one misses when i s mo ing o a coun y. Fu he mo e as ecen ly discussed in Ca d/Dus mann / P es on (2012) p eju- dices om na i es agains mig an s may hampe he adap a ion and he p ocess o eeling a home in Ge many. The e o e when asked abou whe he o no one wan s o e u n mos people say yes because hey miss he cul u e, he ood and so on.2 Once one has ully a i ed in he mig a ing coun y –Ge many o he cu - en analysis –one s a s o mee new people, one ge s o know people on he job –assuming ha you ha e a job –and one s a s o disco e hings abou Ge many ha one may no ha e known in ad ance. This p ocess o in eg a ing and eeling a home in Ge many is wha is unde s ood by ne a achmen in he ollowing. When i s coming o Ge many he ne a achmen is e y low, e en hough one decided o mig a e.3The decision why people mig a ed in he i s place unde lies he cu en analysis and he ocus lies on hose mig an s ha a e al eady in Ge many. The Ge man Socio-Economic Panel (SOEP) is used o he analysis as indi- iduals p o ide in o ma ion on hei e u n in en ions.4Using a du a ion model an exp ession o he p edic ed e u n ealiza ion (expec ed du a ion o he s ay in Ge many o he cu en analysis) is in e ed. This p edic ed e u n will be compa ed o he esponden s in en ions and will be eg essed on di e en se s 250 Ge a d J. an den Be g and Michèle A. Weynand Schmolle s Jah buch 133 (2013) 2 1Conlin/O’Donoghue/ Vogelsang (2007); Le y (2009); Acland /Le y (2010) a e pa- pe s ha a e able o pin down he exac alue o he p ojec ion bias pa ame e . 2Indi iduals who came o Ge many due o a wa o as e ugees on he o he hand may no wan o e u n o hei coun y e e . These indi iduals a e o no wo y o he cu en analysis, since hey should p edic ha hey wan o s ay in Ge many o e e . 3See Sjaas ad (1962) o a i s o mula ion o he decision o mo e. Fo a mo e ho - ough e iew on he e u n mig a ion decision, see Bo jas /B a sbe g (1996) who gene al- ize Bo jas (1991). O he ele an s udies a e; Dus mann (2003b); Dus mann (2003a); Dus mann /Weiss (2007). 4To ge a ho ough o e iew o he da a, we e e you o Wagne /F ick /Schupp (2007). The Ques ion asked is: Do you wan o s ay in Ge many o e e ? This can be answe ed by yes o no. I he esponden s answe no, hey a e asked how long do you wan o emain in Ge many, which can be answe ed by a yea a mos and a ew mo e yea s. I he indi idual hen says a ew mo e yea s, hey a e asked how many yea s? And hey a e supposed o gi e an es ima ed amoun o yea s. E.g. i he indi idual esponds 5 i is supposed o mean she wan s o e u n in 5 yea s. OPEN ACCESS | Licensed unde | h ps://c ea i ecommons.o g/abou /cclicenses/ DOI h ps://doi.o g/10.3790/schm.133.2.249 | Gene a ed on 2023-01-16 13:36:55 o socio-economic a iables. These eg ession esul s highligh he d i ing ac- o s o he di e ence be ween e u n in en ions and e u n ealiza ions.5 A i s impo an inding is ha people’s in en ions exe bunching which al- eady poin s owa ds a simpli ying heu is ic. Taking a close look a he di e - ence be ween he in en ions and he ealiza ions we see ha he in en ions lie cons an ly below he ealiza ion. Indi iduals conside ably unde es ima e he du a ion o hei s ay. The a e age o ecas e o is he e o e mos ly nega i e bu dec eases he longe one s ayed in Ge many and he olde one ge s.6Using pooled OLS we a e also able o highligh a ew o he ac o s ha d i e he di e ence be ween in en ions and ealiza ions. Being olde han 60 yea s e- duces he di e ence conside ably, while i an indi idual eels disad an aged due o he o igin he o ecas e o inc eases. The clea unde s anding o he di e ence be ween in en ions and ealiza ions in e u n mig a ion is c ucial o in eg a ion policies. I mig an s consis en ly unde es ima e he du a ion o hei s ay, hey may no pu enough e o in o hei in eg a ion. Go e nmen in e en ion may help o imp o e he si ua ion o mig an s by emphasizing in eg a ion as ea ly as possible. I is impo an o unde s and hese di e ences o a oid con lic s o in eg a ion be ween cu en inhabi an s and mig an s. The se up o he pape is as ollows; sec ion 2 p esen s he du a ion model. Sec ion 3 p esen s he esul s, while sec ion 4 concludes. 2. Model Le T be he du a ion un il he e u n and le ð jxð Þ;x0Þbe he haza d a e, which can be in e p e ed as he e u n a e a o he e u n p obabili y, whe e p esen s ime since en y, xð Þa e ime a ying co a ia es, such as he cu en employmen s a us, he cu en amily income, and x0a e ime in a ian co a i- a es, such as he age a mig a ion, gende , educa ion, coun y o o igin. The amoun o money ha mig an s would ea n in hei home coun y and how he pu chasing powe s di e be ween he mig an s coun y o o igin and Ge many builds he amewo k o he analysis be ween expec a ions and eali- za ions. In o ma ion abou wha mig an s wages would be in hei home coun- y is no a ailable and GDP is used o in e how big he di e ences a e be- ween Ge many and he sending coun y. Since he ocus o he pape is o explain di e ences be ween e u n in en ions and e u n ealiza ions we need Expec ed and Ac ual Du a ion Un il Re u n Mig a ion 251 Schmolle s Jah buch 133 (2013) 2 5Please be awa e ha we a e no claiming a causali y o he esul s. We a e only in e - es ed in he d i ing ac o s o he o ecas e o . 6The di e ence be ween he in en ions and he p edic ed e u n and o ecas ing e o will be used in e changeably in he ollowing since hey e e o he same measu e. OPEN ACCESS | Licensed unde | h ps://c ea i ecommons.o g/abou /cclicenses/ DOI h ps://doi.o g/10.3790/schm.133.2.249 | Gene a ed on 2023-01-16 13:36:55 an exp ession o he e u n ealiza ion which will be in e ed h ough du a ion analysis. GDP is a good indica o o compa e coun ies and as men ioned in Bo jas/ B a sbe g (1996) o Dus mann/Weiss (2007) he decision o e u n may be a pa o he li e-cycle, o he sending coun y may ha e caugh up o he ecei - ing coun y in e ms o GDP.7 As emphasized abo e, he decision o e u n elies on he economic model which builds he amewo k o he haza d a e. As an example, o an indi id- ual o ake he decision o mo e in 2005 i is needed ha he expec ed p esen alue o ea nings p oxied by GDP in he home coun y minus he mo ing cos s a e la ge han he expec ed p esen alue o ea nings p oxied by GDP in Ge - many (Sjaas ad, 1962). Mo e o mally, i one decides o mo e in 2005, X K ¼2005 1 ð1þ Þ E½UðXHð ÞÞ  E½UðXGð ÞÞðÞ>cþð1Þ needs o hold.8This can also be ew i en in e ms o he haza d a e in 2005, such ha : P< X K ¼2005 1 ð1þ Þ E½UðXHð ÞÞ  E½UðXGð ÞÞðÞc ! ð2Þ ¼PK ¼2005 1 ð1þ Þ E½UðXHð ÞÞ  E½UðXGð ÞÞðÞc  ! : This exp ession can also be ew i en o ge an exp ession o he haza d a e each yea . Since we a e ul ima ely in e es ed in he expec ed du a ion o a s ay, he du a ion amewo k allows us o w i e: yð0Þ¼EðTjx0;expec a ions o u u e pa h o xð ÞÞð3Þ ¼Z1 0 exp Z1 0 ðujxðuÞ;x0Þdu  dz 252 Ge a d J. an den Be g and Michèle A. Weynand Schmolle s Jah buch 133 (2013) 2 7The GDP le els o he di e en coun ies a e om Angus Maddison (h p://www. ggdc.ne /MADDISON/o iindex.h m) bu a e only a ailable un il 2008, which o ces he d op o he yea 2009 and lea es us wi h he pe iod 1984–2008. 8Xð Þ, a e co a ia es ha we con ol o . c ep esen s he cos o mo ing, Kis he maximum su i al ime o which we assume ha i equals he expec ed li e ime du a- ion app oxima ed by 100 –cu en age in he empi ical pa . is he in e es a e and is an e o e m. UðÞ ep esen s he u ili y unc ion. The subsc ip Gs ands o Ge many, and he subsc ip Hs ands o Home. OPEN ACCESS | Licensed unde | h ps://c ea i ecommons.o g/abou /cclicenses/ DOI h ps://doi.o g/10.3790/schm.133.2.249 | Gene a ed on 2023-01-16 13:36:55 in a con inuous ime amewo k. This equa ion can be ew i en o yð Þwhe e can ake any in ege alue in ½0;Twhich means ha we end up wi h possible yð Þ,yð 1Þ,…,yð0Þ. This exp ession allows he indi idual o adap he ex- pec a ions. In o he wo ds, yð0Þmay be di e en han yð1Þbecause people up- da e he u u e pa h o xð Þ. The model’s p edic ed expec a ions will be compa ed o he esponden s indica ed in en ions o see wha d i es he di e - ence and whe he people lea n; a e hei p edic ions e en ually con e ging o he “ u h”? In he empi ical pa , we a e going o es ima e a disc e e ime complemen a y log log model wi h a hi d o de polynomial in ime, which allows o in e he p edic ed e u n.9 3. Resul s The i s imp ession o a bias becomes isible when we compa e he ac ual and he in ended e u n. Looking a Table 1 abou 70 pe cen o non- e u ne s had he in en ion o e u n o hei home coun y in 1984. When e alua ing Table 1 keep in mind ha some people may ha e been w ongly coded as non e u ne s. They can s ill e u n bu i canno be obse ed due o igh censo ing. A u he hing o no e, is ha i is impossible o cap u e sho e m mig a ion las ing no longe han one yea . The SOEP su eys people annually he eby no allowing o accoun o people who mig a e and e u n wi hin a yea .10 Fo he es o he pape we conside only he mig an s who a e al eady in Ge many and p esen in he SOEP. Fu he mo e we only conside adul s who a e olde han 18 yea s in o de o only include hose indi iduals who ake he e u n decision hemsel es. As he use o he GDP Da a o ced he d op o he yea 2009, we a e le wi h 25 yea s o he analysis (1984 –2008).11 Be o e analyzing he di e ences be ween he in en ions and he p edic ed e u n, le us look a he indi iduals in en ions and wha may be d i ing ac o s Expec ed and Ac ual Du a ion Un il Re u n Mig a ion 253 Schmolle s Jah buch 133 (2013) 2 9The esul s o he cloglog model a e no epo ed he e, bu a e a ailable om he au ho s upon eques . We con ol o a hi d o de polynomial in ime, gende , age a mig a ion, di e ence in GDP be ween Ge many and he home coun y, ma i al s a us, whe he o no one has amily a home, a spouse a home, a ended school in Ge many, GDP g ow h in Ge many and a home, income, whe he o no one is abo e 60 and coun y o o igin. The dependen a iable is 1 i he indi idual has le Ge many and 0 i we do no obse e he esponden lea e. Whe he o no an indi idual has le is in e ed om he add ess log o he SOEP. 10 These indi iduals do no play an impo an ole o he analysis o he unde es ima- ion o he ip du a ion. 11 Table 1 only includes s a is ics o hose indi iduals who answe ed he in en ion o e u n ques ion in 1984. In able 2 he numbe o indi iduals is highe and i depends upon he speci ica ion exac ly how many. Fo he cloglog models we had 3152 indi id- uals whe e 574 du a ions un il e-mig a ion a e no igh censo ed. OPEN ACCESS | Licensed unde | h ps://c ea i ecommons.o g/abou /cclicenses/ DOI h ps://doi.o g/10.3790/schm.133.2.249 | Gene a ed on 2023-01-16 13:36:55 Table 1 In en ions and Realiza ion 1984 –2009 Re u n be ween 84 and 09 In ended Re u n (84) No Yes To al No 682 82 764 Column Pe cen age 30.00 16.05 27.44 Row Pe cen age 89.27 10.73 Yes 1591 429 2020 Column Pe cen age 70.00 83.95 72.56 Row Pe cen age 78.76 21.24 To al 2273 511 2784 This Table only p esen s s a is ics o people p esen in 1984. Sou ce: SOEP, own calcula ions. o changes in hese in en ions. Figu e 1 plo s he In ended Du a ion o S ay, in igu e 1a we impu ed he in ended du a ion o hose who wan ed o s ay o - e e as 100  hei cu en age, while in igu e 1b we only ake a look a hose ha ac ually ell us how long hey plan on s aying. In bo h panels we can see ha he indi iduals show bunching beha io a ound 5, 10, 15, 20 yea s. This bunching may al eady poin owa ds a simpli ying heu is ic a wo k when indi- iduals o m hei in en ions.12 Figu e 2 plo s he di e ence be ween he in en- ions and he p edic ed e u n. Again igu e 2a plo s he di e ence o he whole sample whe e o hose who in ended o s ay o e e we impu ed hei maximal su i al ime as 100  hei cu en age. Figu e 2b plo s he di e - Figu e 1: In ended Du a ion o S ay 254 Ge a d J. an den Be g and Michèle A. Weynand Schmolle s Jah buch 133 (2013) 2 0 2 4 6 8 10 020 40 60 80 100 0 5 10 15 20 25 020 40 60 80 10 0 Expec ed Du a ion o S ay in Ge many Expec ed Du a ion o S ay in Ge many Pe cen Pe cen (a) In ended Du a ion o S ay o hose who in end o s ay o e e = 100–cu en age (b) In ended Du a ion o S ay wi hou hose who in end o s ay o e e 12 We also an eg essions on wha d i es he changes in in en ions, hese esul s a e a ailable om he au ho s upon eques . OPEN ACCESS | Licensed unde | h ps://c ea i ecommons.o g/abou /cclicenses/ DOI h ps://doi.o g/10.3790/schm.133.2.249 | Gene a ed on 2023-01-16 13:36:55 ence o he educed sample whe e we lea e hose ou who in end o s ay o - e e . Figu e 2a gi es us hope ha he e seem o be many people p edic ing he du a ion o hei s ay co ec ly, bu when we ake hose ou who in end o s ay o e e (which a e hose ha we imposed on how long hey emain in Ge many – igu e 2b) p ac ically no emaining indi idual has a co ec p edic ion. Figu e 2b shows ha indi iduals o e es ima e he e u n o hei home coun y, equi alen ly s a ed, unde es ima e hei ime spen in Ge many. When looking a he di e ence he in ended e u n is cons an ly below he ac ual p edic ed e u n which makes he di e ence nega i e. This is an impo an inding and may poin owa d o e con idence; a opic e y nicely in oduced in Kahneman (2011). In he cu en wo k o e con idence would ha e o go along wi h ne a achmen in he sense ha indi iduals a e o e con iden abou he ac ha hey will be ue o hei amily ( o hei “ oo s”) and wan o e u n home, and hus unde es ima e hei a achmen o Ge many. Figu e 2: Di e ence Be ween In en ions and P edic ed Realiza ions Figu e 3 and 4 hen plo he a e age o ecas e o (equi alen o he a e age di e ence) o e di e en ime speci ica ions. Figu e 3a plo s he a e age o e- cas e o o e ime spen in Ge many, and wha was no ob ious be o e now seems o become ele an . The longe people ac ually a e in Ge many, he mo e accu a e hey ge on a e age. The la ges e o ha hey make is when hey ha e spen 20 yea s in Ge many while hei e o is p ac ically ze o once hey ha e spen 60 yea s in Ge many. This could go along wi h he ac ha ha ing spen 20 yea s in a coun y you may s ill belie e ha you e en ually e u n, bu he olde you ge , he be e you a e a es ima ing you ac ual chance o lea ing and so you seem o be mo e accu a e wi h you o ecas . Figu e 3b helps us explain a wha age you seem o ge be e a p edic ing you u ili y o you u u e choice a iables. Towa d his end he e is a clea di ec ion; he olde you ge he be e you ge a p edic ing you emaining Expec ed and Ac ual Du a ion Un il Re u n Mig a ion 255 Schmolle s Jah buch 133 (2013) 2 0 5 10 15 20 -1 0 0 -50 050 0 1 2 3 4 -1 0 0 -50 050 Di e ence be ween in en ions and p edic ed ealiza ions Di e ence be ween in en ions and p edic ed ealiza ions Pe cen Pe cen (a) Those who in end o s ay o e e maximal su i al ime = 100–cu en age (b) Wi hou hose who in end o s ay o e e OPEN ACCESS | Licensed unde | h ps://c ea i ecommons.o g/abou /cclicenses/ DOI h ps://doi.o g/10.3790/schm.133.2.249 | Gene a ed on 2023-01-16 13:36:55 du a ion. This inding is no su p ising as he olde one is he sho e he e- maining ho izon ge s, and he e o e one may also be be e a p edic ing he du a ion o he s ay. Figu e 4 hen plo s he a e age o ecas e o o e ime. We included his esul in o de o show ha he e do no seem o be ele an mac o shocks ha may d i e ou esul s. Figu e 3: A e age Fo ecas E o Figu e 4: A e age Fo ecas E o Table 2 hen inally akes a close look a he di e ences be ween he in en- ions and he p edic ed e u n. In able 2 columns (3), (5) and (6) include indi- idual ixed e ec s whe e he s anda d e o s a e also clus e ed a he indi idual le el. The OLS esul s in columns (1), (2) and (4) a e shown jus o com- pa ison.13 The coe icien on ‘a ended school in Ge many’changes sign and 256 Ge a d J. an den Be g and Michèle A. Weynand Schmolle s Jah buch 133 (2013) 2 -20 -10 010 20 020 40 60 -30 -20 -10 010 20 40 60 80 100 age a e age o ecas e o Time in Ge many a e age o ecas e o Whole Sample WO in end o s ay o e e Whole Sample WO in end o s ay o e e (a) O e ime spen in Ge many (b) O e age a ime o o ecas -30 -20 -10 0 1985 1990 1995 2000 2005 2010 Whole Sample WO in end o s ay o e e a e age o ecas e o e hebj 13 The iden i ica ion wi h he use o indi idual ixed e ec s is d i en by a ia ions ac oss ime by each indi idual. Since many o he a iables included in he eg ession OPEN ACCESS | Licensed unde | h ps://c ea i ecommons.o g/abou /cclicenses/ DOI h ps://doi.o g/10.3790/schm.133.2.249 | Gene a ed on 2023-01-16 13:36:55