Parents’ Ambitions and Children’s Competitiveness Menusch Khadjavi Andreas Nicklisch WiSo-HH Working Paper Series Working Paper No. 0138 FeMbarurcahry 22001144 WiSo-HH Working Paper Series Working Paper No. 0138 MFeabrrcuha 2ry01240 14 Parents’ Ambitions and Children’s Competitiveness MenuschKhadjavi,UniversitätHamburg,Germany AndreasNicklisch,UniversitätHamburg,Germany ISSN 2196-8128 Font used: „TheSans UHH“ / LucasFonts Die Working Paper Series bieten Forscherinnen und Forschern, die an Projekten in Federfüh- rung oder mit der Beteiligung der Fakultät Wirtschafts- und Sozialwissenschaften der Uni- versität Hamburg tätig sind, die Möglichkeit zur digitalen Publikation ihrer Forschungser- gebnisse. Die Reihe erscheint in unregelmäßiger Reihenfolge. Jede Nummer erscheint in digitaler Version unter https://www.wiso.uni-hamburg.de/de/forschung/working-paper-series/ Kontakt: WiSo-Forschungslabor Von-Melle-Park 5 20146 Hamburg E-Mail: [email protected] Web: http://www.wiso.uni-hamburg.de/forschung/forschungslabor/home/ ∗ Parents’ Ambitions and Children’s Competitiveness 1 1,2,∗ Menusch Khadjavi, Andreas Nicklisch, 1 UniversityofHamburg,Germany 2 MaxPlanckInstituteforResearchonCollectiveGoods,Bonn,Germany ∗ Correspondingauthor;e-mail: [email protected] February 1, 2014 Abstract Individualwillingnesstocompeteisapersonalitytraitofhighimportance. Whilesub- stantial differences between individuals are documented in the literature, the sources of this heterogeneity are still not well understood. To contribute to this issue we conduct an incentivized field study with pre-school children. We assess the children’s willingness to compete and relate those inclinations to ambitions and preferences of their parents. The ambition levels of parents concerning their children’s success in the later professional life seemspredictivefortheirchildren’scompetitiveness. Inparticular,childrenofhighlyam- bitious parents tend to enter competition even if their chances to win are low. Moreover, thelinkbetweenparents’ambitionsandchildren’scompetitivenessisrelatedtothesocio- economicbackgroundoftheparents. Keywords: Children, Competition, Field Experiment, Parents, Socialization, Intergenera- tionalTransmission JEL-Classification: C91,C93,D01 ∗We gratefully acknowledge the many useful comments provided by participants of the Workshop on Self- Control, Self-Regulation and Education at Aarhus University, Denmark in December 2013. We are indebted to BernoBu¨chelforexcellentresearchassistance. 1 1 Introduction Akeycharacteristicofmostmodernsocietiesistheiremphasisoncompetition. Firmscompete forcustomers,employeescompeteforpromotionsandbonuses,andpoliticianscompeteforvot- ers. Indeed competition increases welfare and productivity in many settings, for instance labor contracts(LazearandRosen,1981),andcanresultinjoyofwinning(Dohmenetal.,2011). The ubiquitous emphasis of competition may however also be a burden for those individuals who experience losing frequently, resulting in stress, depression and connected health costs (e.g., Dohmen et al., 2011; Gilbert et al., 2009). Several studies show that competition for school placement and competition in classrooms negatively affects the quality of learning and other socialaspectsofstudents(e.g.,BelfieldandLevin,2002;LaddandFiske,2003). Consequently, itiscrucialtochooseone’scompetitionswisely. Little is known about the origins of immanent preferences for competition. Our study pro- vides evidence on factors that shape preferences for competition already in their early child- hood. Our data suggests that an important factor triggering children’s competitiveness are par- ents’ ambitions for their child’s later success in professional life. That is, parents’ ambitions drivesomeofthechildrentoentercompetitionregardlessoftheirchancestowin. A number of studies analyze whether gender causes differences in the competitiveness of children: Gneezy and Rustichini (2004) as well as Sutter and Ru¨tzler (2010) find greater com- petitivenessamongboysinarunningtask,particularlyforolderchildren. Theauthorsrelatethis finding to the tremendous degree of overconfidence concerning the rank of their performance of – especially older – boys. Other authors cannot confirm this result for other tasks (Khacha- tryan, 2012; Dreber et al., 2011; Samak, 2013), or present mixed evidence regarding decisions tocompeteandtheimprovementofperformance(Ca´rdenasetal.(2012)). Onepotentialreason forthismixedpicturemaybedifferentculturesinthecountrieswherethechildrengrewup. Psychologists, and very recently also economists, test another potential origin of prefer- ences for competition: they link different degrees of competitiveness to family factors (e.g. Hupp et al., 2010; McKee and Leader, 1955; Poulin-Dubois et al., 2002; Van Lange et al., 1997).1 For instance, McKee and Leader (1955) analyze behavior of 112 three-to-four-year 1Recent studies by Bauer et al. (2011), Dohmen et al. (2012) and Kosse and Pfeiffer (2012) have taken the sameavenuelinkingother-regardingpreferences,riskpreferencesandpatiencetotheirparents’attitudes.Allthree worksfindcorrelationsbetweenparents’andchildren’spreferencessuggestingthatthetransmissionofpreferences fromparentstotheirchildrenisakeydriverforthesepersonalitytraits. Likewise,Deckersetal.(2013)showthat parentalsocioeconomicstatussignificantlyinfluencespatienceandriskattitude:childrenfromfamilieswithhigher socioeconomicstatusaremorepatientandarelessriskseeking. 2 old children in child-care institutions in the Bay Area. They report that children in child-care institutions in low-class residential areas were more inclined towards competition compared to children from middle-class areas. Van Lange et al. (1997) measure competitiveness of 631 Dutchadultswiththesocialvalueorientationmethodandrelatetheirdatatobackgroundinfor- mation. Notably, they find that subjects with more siblings are more pro-socially oriented, but less competition-oriented compared to those with fewer siblings. Alma˚s et al. (2012) explore the relation between children’s willingness to compete and the socio-economic background of their families among Norwegian adolescents, 14 to 15 years old. One of their key findings is that children from families with low income and low education are less willing to compete (even when controlling for performance). Leibbrandt et al. (2013) provide evidence that part ofanindividual’scompetitivenessstemsfromnormsandcustomsinhersociety. Bartlingetal. (2012) analyzes whether child health can explain differences in their willingness to compete with others. Findings suggest that childrens willingness to compete is negatively related to health problems. More specifically, health has a strongly negative effect for children with low socio-economic background, whereas this connection cannot be found for children with high socio-economicbackground. In our study, we provide an investigation on another source of intergenerational socializa- tion: wetestwhetherparents’preferencesandambitionswithrespecttotheirchildren’ssuccess influence children’s competitiveness. In order to allow for maximal relevance of the family background on children’s decision, we analyze the behavior of very young children. Our in- vestigation comprises an experimental task with 84 pre-school children in Northern Germany. Like other recent studies, we borrow a running task from Gneezy and Rustichini (2004). The designallowsustomeasurechildren’swillingnesstocompeteinafamiliartaskwhichappears tobeeasilycomprehensibleforthreetosixyear-oldchildren,andwherecompetingis‘natural’. Prior to the experiment, we sent out consent forms including a short survey to the parents. Im- portantly,neithertheparentsnorthekindergartenteachersknewthetaskandresearchquestion. The survey included questions concerning topics which are related and unrelated to competi- tiveness. Wecollectedinformationonparents’valuesandambitionsconcerningtheirchildren. Ourresultsindicatehardlyanygenderdifferencesincompetitivenessatthisveryyoungage. Likewise, parents’ preferences do not appear to matter directly for children’s decision making. Rather,oneofthemostimportantfactorsinfluencingchildren’scompetitivenessisparents’am- bitions for their child’s later success in professional life. That is, children of parents stressing the importance of their offspring’s job success are significantly more likely to compete than 3 children of less ambitious parents: we find a (mean) marginal increase of about 14 percent in the likelihood to compete for a one point increase on the seven-point ambitions scale. Inter- estingly, when controlling for the individual likelihood of winning a competition, children of highly ambitious parents decide to enter competition even when the prospects of winning are unfavorable. Consequently,relativelyslowchildrenwithveryambitiousparentsentercompeti- tions and, on average, earn fewer rewards than relatively slow children whose parents are less ambitious. Hence, parents’ ambitions drive some of the children to ‘overinvest’ into competi- tion. Onaverage,wefindparentswithlowerincomeandlowereducationtobemoreambitious thanfamilieswithhighrelativeincome. Thiseffectcouldstrengthenthealreadyexistingstatus andincomedisadvantagesofchildrenfromlowincomeandloweducationhouseholds. The remainder of the article is organized as follows: Section 2 presents the method we use to obtain data on children’s willingness to compete as well as parents’ characteristics and ambitions. Here we also formulate the hypotheses we want to test. In Section 3 we lay out the resultsofourstudywithregardtothedifferentpossiblesourcesofcompetitivenessandtestour hypotheses. Finally,Section4providesaconcludingdiscussionofourresults. 2 Method Intotal,ourstudyincludesdatafrom84childrenofagesbetweenalmostthreeandsixyears. 2 The settings of our study are two kindergartens in Northern Germany. The first kindergarten is located in Hamburg, in the suburb of Heimfeld. 52 children from Hamburg participated in our study,23boysand29girls. ThisdatawascollectedinMarch2012. Thesecondkindergartenis located in Lower Saxony, in Essenrode. Here we were able to collect the data for 32 children, 20boysand12girlsinMay2012. NoticethatbothkindergartensarelocatedinGermanmiddle classneighborhoods. Childreninbothkindergartenscomefromapproximatelythesamefamily structure (83% married parents in Heimfeld, 88% in Essenrode), the same number of siblings (mean is 0.78 in Heimfeld, 0.97 in Essenrode), parents’ age (median is 34 in Heimfeld, 37 in Essenrode) and parents’ education level (27% of the mothers have a university degree in Heimfeld,23%inEssenrode). Theprimaryaimofourstudyistoanalyzejointlytheeffectsofachild’scharacteristicsand her parents’ characteristics and ambitions for her decision concerning competition. First, we 2Overall,100childrenparticipatedintheexperiment.However,wehadtoexclude16observationsduetosome missingdataintheparents’survey. 4 describe the elicitation of the competition decision. Then we report the data on parents’ and children’scharacteristics. 2.1 Eliciting Children’s Willingness to Compete Weelicitchildren’swillingnesstocompetebyofferingcompetitiveandnon-competitivereward schemes in a running task. We borrow the idea of a running task from Gneezy and Rustichini (2004) who use it to study competition effects among 9-10 years old children in Israel.3 We employ the specific running task in our design to allow for easy comparison of our data with earlier findings. Moreover, running is one of the few tasks that presumably all children are familiarwithalreadyataveryyoungage. Itthereforeseemsnaturaltoconsidercompetitionin arunningtaskforkindergartenchildren. Therunningtaskwasconductedintherespectiveplaygroundsofthetwokindergartensinsix separatesessions.4 Thekindergartenteachersallocatedchildrenintothesesessionsaccordingto organizational criteria of the kindergartens. In each session, one group of children participated inourstudy,whileallotherswerelookedafterbyteachers. Onaverageagroupconsistedof16 children,whilegroupsizesvariedbetween12and23children. The children were asked to run as fast as possible from one pair of cones to another pair of cones which were placed at a distance of 30 meters. Each child ran twice and we recorded thetimewithastopwatch. Priortothefirstrecording,eachchildwasinformedthatshewould receive the reward if she completed the task faster than half of the other children. If she was notfasterthanhalfoftheotherchildren,thenshereceivednorewardforthisrecording. Notice thatwedidnotelicitthechild’ssubjectiveexpectationconcerningtherankorbeinginthefaster half of the group. Sutter and Ru¨tzler (2010) report an overwhelming degree of overconfidence among children of that age, so that statement seems to provide little information (they report that on average 91 percent of the children expect to be in the faster group). Nonetheless, we assumethatchildrencouldassessintuitivelytheirperformancewithinthegroupbasedonevery- day experience, since they were able to observe each other while running. Information on performanceandonearnedrewardswasnotdiscloseduntiltheendoftheexperiment.5 Afterthefirstrecordingeachchildwasinterviewedinprivate(ateacherwasalwayspresent 3RecentstudiesbyCa´rdenasetal.(2012),Dreberetal.(2011)andSutterandRu¨tzler(2010)alsomakeuseof the task for research with 9-12 years old children in Colombia and Sweden, 7-10 years old children in Sweden, and3-8yearsoldchildreninAustria,respectively. 4Allsessionswereconductedincomparable,sunnyweatherconditions. 5Allchildrenfinallyreceivedatleastonerewardasa‘show-upfee’. 5 inthebackgroundtoensurecomfortforthechild). Weinformedeachchildthattherewouldbe a second recording and that this time the reward depends on her decision. Again, we told each child that we would record the time. Each child had to decide whether to run against another child(i.e.,tocompeteagainstanotherchild),ortoruninordertoimproveherowntime. Ifshe winsagainstanotherchild,shereceivesalargerewardplusasmallreward. Ifshelosesagainst another child, she receives no reward. Conversely, if she decides to improve her own time, she receivesalargerewardifsherunsfasterthanbeforeandasmallrewardifshedoesnotimprove her own time. The two options were illustrated by two large cards showing large and small wrapped rewards for the corresponding case (see Figure A.1 in Appendix A); children had to point to one of the two options and to declare explicitly their choices in order to make clear decisions. The interview with each child was always conducted by the same experimenter and fol- lowed,ascloselyaspossible,aprotocolofhowtophrasetheinstructionsandquestions. 6 The second recording of the running time was also conducted without another child running at the sametime(seeFigureA.2inAppendixA).7 Thematchingofcompetingchildrenwassetupby rankingfirstrecordingtimesandensuringcloseranks;noinformationconcerningthematching was provided to the children. Again, this procedure follows Gneezy and Rustichini (2004) and SutterandRu¨tzler(2010). Formally,achildineedstodecideonherbinarycompetitionchoicec ∈ {0,1}. Giventhat i childrendonotknowtheformoftherewardsuntiltheendofthestudy,itisfairtoassumethat 0 = vnone < vsmall < vlarge < vboth forallchildren.8 Thentheexpectedutilityofchildiis i i i i (cid:2) p vboth +0 ifc = 1 EU (c ) = i i i (1) i i q vlarge +(1−q )vsmall ifc = 0, i i i i i where p is child i(cid:3)s probability of winning in competition and q is her probability of running i i faster than her first record. The decision to compete (c = 1) maximizes expected utility if and i onlyif q vlarge +(1−q )vsmall p ≥ i i i i ∈ (0,1). (2) i vboth i 6SeeAppendixBfortheprotocol. 7Wedecidedtohavechildrenrunwithouttheircompetitorsasidetheminordertodisentanglethetwoeffects ofpressurefromhavinganotherchildnexttooneselfandbeinginacompetitionperse.Otherwise,wewouldhave hadthechildrenrunningnexttoeachotherforthefirstrecordingaswell,whichmayresultinissuesoffeelingin acompetitionwithoutbeinginone. 8Itispossiblethatvboth = vsmall +vlarge, whichhasthestandardinterpretationofriskneutrality. Smaller i i i vbothreflectsriskaversionsinceitisthediscretecounterparttoaconcaveutilityfunction. i 6 Note that this is always the case if p close enough to 1 and never satisfied if p close enough i i to 0 such that children who are very confident about winning should always enter competition, while children very sure not to win potential competition should not enter.9 For less extreme valuesofp theutilitymaximizingdecisiondependsonthetrade-offbetweenprobabilitiesand i valuation of the rewards. Thereby, the competitive choice becomes more attractive the larger the value of both rewards, vboth, is compared to the case of either reward, vlarge or vsmall and i i i thesmallertheprobabilityq isofrunningfasterinthesecondroundthaninthefirstround. i The probability of winning a competition, p , depends on the own ability to run fast in i comparisontotheabilitiesoftheothersintheowngroupwhoalsoenterthecompetition.10 Ex post, we can observe those who opted for competition and how they performed in the second round. Ex ante, the estimation of p depends on the beliefs about who enters the competition. i Still, it is reasonable to assume that the probability to win a competition, p , is increasing in i the performance in round one compared with the other children from the own group such that childrenwhoarefasterthanothershavebetterprospectsincompetition. Therefore,weconsider beingfasterthanothersinthefirstroundasanindicatorforahighp .11 i Theaboveexpectedutilityfunctionservesasasummaryoftheincentiveanddecisionstruc- ture;naturally,onemaywonderwhetherkindergartenchildrenareabletoprocessallthetrade- offs involved. Nevertheless, we employ the rewards in order to ensure an adequate incentive structure. While all children are informed about this incentive structure, we deliberately focus the attention of the children on the decision of whether to run against another child or to run to improve one’s own time. Thus, when confronted with the choice situation not only consid- erations of expected rewards, but also spontaneous inclinations to avoid or enter competition are likely to be present. In the latter case, we are not measuring risk preferences and beliefs about own and others’ performance, but rather a preference for or against competition per se. Ineithercase,itisimportanttounderstandwhichfactorsshapethedecision. Letusnowturnto thecharacteristicsandambitionsoftheparents. 9Similarly,highlyriskaversechildren(vboth closetovlarge)shouldnotentercompetition,whileriskseeking i i children(vbothsufficientlylarge)shouldenter. i 10Theassumptionvboth = vsmall+vlarge simplifiestheconditionsuchthatcompetitionmaximizesexpected i i i utilityifvsmall(p +q −1)≥vlarge(q −p ). Forinstance,oneshouldrejectcompetitionifp <min{q ,1}. i i i i i i i i 2 11Moreover,wewillshowlateronthatchildrenwhodonotbelongtothiscategoryof‘fast’children,maximize expectedutilitybynotenteringcompetition,iftheyareriskaverseorriskneutral. 7 2.2 Parents’ Preferences and Ambitions In addition to children’s decisions for or against entering into a competition, we collected data on parents’ preferences and ambitions which may allow us to identify parents’ influences on a child’s willingness to compete. In order to collect our data, we approached the parents by mail (sent via the kindergarten) to ask for their consent (with the permissions of the kindergartens’ directions and their administrative institutions). We combined the consent form with a survey part, so that we had the possibility to collect data on children’s and parents’ characteristics beyond the age and gender of the child. Note that neither parents nor kindergarten teachers wereawareofourresearchquestion. Toruleoutobviouscluesaboutourresearchquestion,the survey was a mixture of related and unrelated questions. Both kindergartens offered activities forthosechildrenwithoutparents’consentorthosewhooptedoutduringthesessions.12 Themainreasonforustostudykindergartenchildrenisthattheyareattheverybeginningof theirencounterofsocialnorms. Therefore,weexpectthatparents’influenceisfairlyundiluted compared to later stages in life. Amongst others, we collected data on parents’ age, number and ages of siblings, duration of breast feeding, marital status of the parents, the language that is spoken at home, and the education of the parents. Further, we asked parents to answer questionswithregardtotrust,competition,risk,importanceofathleticandprofessionalsuccess of their child, perceived relative household income and the mother’s share of household work onaseven-pointscale. 13 The survey items are guided by a list of hypotheses derived from multiple strands of the literature: Approaches suggesting the transmission of cultural traits argue that children adopt preferences similar to their parents’ (e.g., Bisin and Verdier, 2001, or see Bisin and Verdier, 2010 for a survey). In that light, competitive parents are expected to have competitive chil- dren.14In our data set we measure to which extent parents like competition on a seven-point scale. Thecorrespondinghypothesisisthathighlycompetitiveparentstendtohavecompetitive children. If this hypothesis is confirmed, then it is still an open question which mechanism causes the correlation. Parents not only share genes with their children but also serve as role modelsintheirsocializationprocess. Moreover,socializationisdependentonthe(social)envi- 12Despite the consent of her parents, a child was free to choose not to participate in the study at any point in time. Thisoptionwasnotusedapartfromafewexceptionsofveryyoungchildren. 13SeeAppendixCforanEnglishtranslationofthesurveysheet. 14Similarcorrelationsbetweenchildren’sandparents’preferencesarefoundrecentlyforriskpreferencesand trust(Dohmenetal.,2012),fortimepreferencessuchasimpatience(KosseandPfeiffer,2012),andforattitudes concerningfemalelaborforceparticipation(Fernandezetal.,2004). 8