Risking Other People's Money: Experimental Evidence on the Role of Incentives and Personality Traits

Risking Other People's Money: Experimental Evidence on the Role of Incentives and Personality Traits. Ola Andersson et al. The Scandinavian Journal of Economics, March 18 2019. https://doi.org/10.1111/sjoe.12353

Abstract: Decision makers often face incentives to increase risk‐taking on behalf of others through bonus contracts and relative performance contracts. We conduct an experimental study of risk‐taking on behalf of others using a large heterogeneous sample and find that people respond to such incentives without much apparent concern for stakeholders. Responses are heterogeneous and mitigated by personality traits. The findings suggest that lack of concern for others’ risk exposure hardly requires “financial psychopaths” in order to flourish, but is diminished by social concerns.
---
I.Introduction

Risk  taking  on  behalf  of  others  is  common  in  many  economic  and  financial  decisions.  Examples  include  fund  managers  investing their  clients’  money  and  executives  acting  on  behalf of shareholders. To motivate decision makers, the authority to take decisions on behalf of others is often coupledwith powerfulincentives. A basic problem with this practice is that it  is  typically  hard  to  construct  compensation  schemes  that  perfectly  align  the  incentives  of  decision makers with the interests of stakeholders. Indeed, in the wake of the recent financial crisis,  actors  in  the  financial  sector  have  beenroutinely  accused  of  taking increasedrisk  on behalf of investors.  The  introduction  of  advanced  financial  products  has  expanded  opportunities  to  hedge  risks, creating further incentives for increased risk-taking. During a public hearing in the US Senate involving the CEO of a leading investment bank, it emerged from internal e-mails that the bank had taken bets against its own clients’ investmentsto hedge their profits. [***I do not agree with this mention here, it seems the authors support the view that these bets were wrong or immoral***.] Moreover, Andrew Haldane, director of the Bank of England, argues that the banking sector’s problems arerooted  in  the  fact  that  the private risks  of  financial  decision  makers  are  not  alignedwith social  risks,  and that the  latter  areof  a  much  greatermagnitude  (Haldane  2011).In  addition, Rajan  (2006)  suggests  that  new  developments  in  the  finance  industry—such  as  added  layers  of   financial   management   and   new   complex   financial   products—have   exacerbated   the   problem. The  argumentsmadein  the  previous  paragraph  suggestthat  increased  risk  takingis undesirablefrom  a  societal  point  of  view.  However,  theoretically  one  may  argue  that  increased risk  takingis  desirable.  It is  well  knownin  the  finance  literature  that  incentive  schemes may be used to increase risk takingbeyond what is motivated by the decision makers risk preferences (Shavell 1979). The argument made is usually that the owners of capital are well diversified and thereby interested in maximizing dividends payout (risk neutrality). The decision makers, on the other hand, are not well diversified and if risk aversethey  may  take  sub-optimal decisions if the reimbursement scheme does not compensate for the difference in risk exposure  and  risk  preferences.  Such  compensation  may  come  from  incentive  schemes  thatinduce a positive risk shift (e.g., by introducing competition or bonus schemes as in this paper).An alternative motivation is that owners of capital are risk averse, and aware of it, but would  like  their  decisions  to  reflect  dividend maximization. In  particular, from  a  normative  stance  they  agree  that  risk-neutral  decisions  are  optimal, but  when  facing  actual  decisions, they  cannot  refrain  frommaking  decisions  that  depart  from  this  principle.  It  may  then  be  preferred  to  delegate  to  a  decision  maker  whois,  for  example,less  emotionally  attached.  Inboth cases,  the  increased  risk  takingis  then  optimal  from  the  capital  owner’s  and  society’s  perspective and should be encouraged.  In this paper, we do not directly address whether increased risk taking on behalf of others is  welfare  enhancing  or  not,  wesimplycompare  the  level  of  risk  taken  for  others  under different  incentive  schemes. As  a  point  of  comparison,  we  estimate risk taking on  behalf  of  others  in a  situation  without distortive(orcorrective) incentives. Hence,  when  we  refer  to  increased risk taking, we mean risk taking above thelevel decision makers takeon behalf of others in such a neutral situation. Since we find that the level of risk taking on behalf of others without distorting incentives is indistinguishable to the level of risk that individuals take when making  decisions  on  their  ownbehalf,  it  is  natural  to  view  departures  from  this  level  as  detrimental  to  the  principal. However,  it  should  be stressedthat  in  line  with  the discussion above,  we  cannot rule  out the  existence  of  emotional  and  cognitive  constraints  that  impede decision  makers  to  act  in accordance  withtheir owninterest.  That  is,  a  higher  level  of  risk taking could be desirable although decision makers do not choose this for themselves.  From  previous  literature, we  know  that  competitive  incentives  increase  risk  taking  for  individuals  working  in  the  finance  industry  (Kirchler  et  al.  2018) and  students  (Dijk  et  al.  2016). Outstanding questions are whether such behaviour is present in the general population and whether it extends to situations where the decision has consequences for other people.  The  aim  of  this  paper  isto  study  such  incentive  schemes,  with  hedging  opportunities  or  misaligned  incentive  contract,  in  a  controlled  environment  using a  large  sample  of  people  fromall  walks  of  life.  In  particular,  we  let  decision  makers  takedecisions  on  behalf  of  two  other  individuals  under  bonus  and  competitive  incentives, which  may  distort  risk  takingas well as open up for hedging opportunities depending on the dividend correlation.  A  potential  counterbalancing  force  to  increasedrisk  takingmay  bethat  decision  makers  feel  responsible  to  broader  groups  or  have  altruistic  preferences,  i.e., they  intrinsically  care  about  the  outcome  they  generate  on  behalf  of  others(Andreoni  and  Miller  2002).  Indeed,  if  such  a  concern  is  sufficiently  strong, it  may  operate  as  a  natural  moderator  of  extrinsic  incentives  to  take  on  more  risk.  Determining  the  strength  of  these  forces  is  an  empirical  question, made especially difficultbecause it is likely that behavioral responses to misaligned incentives  differ  between  individuals.  Understanding  this  heterogeneity  is  important  because  sometimes  we  can  choose  upon whom  to  bestow  the  responsibility  of  making  decisions  on  behalf of others, and we can select people according to their characteristics. To study this, we employ several measures of personality traits, both survey-based and behavioural measures. Our  focus  here  is  on  risk-taking  behaviour when  there  are  monetary  conflicts  of  interest  between  the  decision  maker  and  investors  (henceforth  called  receivers).  In  our  experiment,  decision  makers  take  risky decisions  on  behalf  of  two  receivers,  whose  payoffs  may  be  negatively or positively correlated. When the payoffs of the receivers are perfectly negatively 
correlated,  the  decision  makers  can  exploit  the  correlation  to  increase  their  ownpayoff without  increasing  their  ownrisk  exposure.  On  the  contrary,  when  payoffs  are  perfectly  positively correlated, such risk-free gains are not possible. We allow decision makers to take decisions under both regimes.  For  decision  makers, we  incorporate  two  types  of  incentive  structures  common  in  the  financial sector. First, we consider a bonus-like incentive scheme where the decision maker’s compensation   is   proportional   to   the   total   payoffs   of   the   two   receivers.   Within   our   experimental setup, we show theoretically that such bonus schemescreate material incentives for increased risk-taking if the receivers’ returns are negatively correlated. Second, we study winner-take-all competition between decision makers who are matchedin pairs. The decision maker  who  generates  the  higher  total  payoff  on  behalf  of  her  receivers  earns  aperformance fee  as  a  percentage  of  the  total  payoff  to  the  receivers,  while  the  otherdecision  makerearns nothing. Competitive incentives are commonplace in financial marketsand create option-like convex compensationschemes(Chevalier and Ellison 1997).We show theoretically that such compensation schemes create material incentives for increased risk taking, independent of the correlation structure of the receivers’ returns. The intuitionis that increasing the risk exposure increases  the  chance  of  outperforming  peers,  and  this  mechanism  trumps  any  concerns  for  individual risk-taking by the decision maker. We believe the research reported here is the first to experimentally investigate the effects of such incentives on risk-taking on behalf of others on a large scale using a random sample of the general population.  Our  experimental  study  yields  two  main  findings.  First,  ordinary  people  respond  to  powerful incentives to take risks. In particular, in line with our hypotheses, we find that bonus schemes  trigger  increased  risk-taking  on  behalf  of  others  only  when  receivers’  returns  are  negatively correlated. Hence, a bonus scheme with well-aligned risk profiles between decision makers  and  receivers  does  not  distort  risk-taking  in  our  setting.  Competition,  on  the  other  
hand,  triggers  increased  risk-taking  irrespective  of  the  correlation  structure  of  receivers’  returns.  For  the  receivers,  competition  between  the  decision  makers  thereby  always  leads  to  higher risk exposure.  Overall,  we  find  that  individual  incentives  dominate  oversocial  concerns  in  the  settings  studied  here.  However,  we  also  findconsiderable  heterogeneity  in  how  people  respond  to  such  financial  incentives.  We  have  access  to  a  large  and  heterogeneous  sample  along  with  a  wealth  of  measures  from  earlier  surveys  and  experiments.  This  unique  data  enables  us  to  identify  and  investigate  who  chooses  to  expose  others  to  risk.  We  find  that  measures  of  personality related topro-social orientation are associatedwithrisk-taking on behalf of others. Indeed,  individuals  with  more  pro-social orientations expose  receivers  to  significantly  lessrisk.  It  has  been  popular  to  decry  decision  makers  in  the  financial  industry  as  “financial  psychopaths” (see, e.g., DeCovny, 2012).We are not in a position to judge whether this is an accurate  description,  but  our  observations,  based  on  a  fairly  representative  sample  of  the  general  populationcoupled  with  individual  personality  measures,  allow  us  to  conclude  that  lack  of  concern  for  others’  risk  exposure  hardly  depends  on  “financial  psychopaths”  to  flourish.  Ordinary  people  tend  to  do  it  when  the  incentives  of  decision  makers  and  receivers  are not aligned. The general lesson is that policymakersshould become more circumspect in designing incentives and institutionsbecause they impact the risks that are takenon behalf of others. Scientific evidence on the characteristics of individuals working in the financial sector is scant.  Concerning  risk  preferences,  Haigh  and  List  (2005)  find  that  professional  traders  exhibit behaviour consistent with myopic loss aversion to a greater extent than students. In a small sample (n= 21) of traders, Durand et al. (2008) find that average Big 5 scores among traders are not significantly different from the population averages. Along similar lines, using 
a  small  sample  of  day  traders,Loet  al.(2005)  were  unable  to  relate  trader  performance  to  personality   traits.   Oberlechner   (2004)   investigates   which   personal   characteristics   are   perceived  as  important  for  being  successful  as  a  foreign  exchange  trader.  However,  the  characteristics emphasized are not directly comparable with the Big 5 inventorythat we use to measure  personality  traits.  The  closest  match  to  agreeableness  and  extraversion  (which  we  find  to  be  important  in  Table  3)  is  probably  social  skills.  Interestingly,  social  skills  were  considered  the  least  important  of  the  23  delineated  skills.Sjöberg  and  Engelberg  (2009)  compare financial economics students with a sample from the Swedish population. They find that  compared  to  the  overall  population  financial  economics  students  are  less  altruistic  (as  measured by interest in peace and the environment) and less risk averse.