FINANSMARK-Finansmarkedet

People do not always have the necessary knowledge to make optimal choices for themselves, and may therefore rely on expert advice in order to make better choices. This is particularly salient in finance. Financial advisors constitute the connection between small investors with limited knowledge and complex financial markets, and they play an important role for millions of people who allocate their savings between different investment products. Many investors consider financial advisors as the most important information source, and financial advisors often serve as the true decision makers behind investments into actively managed mutual funds. It is thus challenging that financial advisors and their clients often have conflicting interests. What is good for the advisor may be bad for the client and vice versa. Indeed, research has demonstrated that financial advisors may be tempted to give advice that are based on self-interest rather than the interests of their clients. However, advisors (presumably) also give good advice, despite incentives to do otherwise. In this research project we use controlled experiments to investigate under which conditions advisors behave fairly, offering good advice, and under which conditions they act selfishly, offering advice that are unfavorable to their clients. We vary both the riskiness of the prospects that the advisors advice, and the responsibility that the advisors have for the choice their clients make. We also investigate to what extent the clients actually will follow the advisors' advice. In particular, we explore the role of robotization. An increasing number of financial service providers are using robo-advisors; online platforms that provide advice by complex computers. An important question is thus whether clients trust robot-based advice. Hence, we investigate experimentally how clients will respond to information that the advisor is an algorithm programmed to make particular advices. Our experimental results show that there are indeed personal costs associated with giving bad advice. We present results from a large-scale online experiment studying advisors' behavior under conflicting interests. We use a dictator game as a baseline and transform the game into a situation in which the dictator gives a binding advice and a free non-binding advice, respectively. We also vary the payoffs to include both certain and risky outcomes. Our results show that people are averse to giving bad advice. When subjects are given the role as advisors, they behave less selfishly, even when the economic and strategical considerations remain unchanged. Moreover, we find that the moral costs of giving a bad advice is larger when the advisors cannot dictate the clients' decision, but rather have to induce the clients to make bad choices for themselves. With respect to robotization, our intention was to examine if people trust algorithms more than their human counterparts. In collabaoration with two master students, we conducted two experiments. The first experiment suggest that people trust algorithms more than people. However, this does not seem to translate to the context of financial advisory (second experiment), where the participants relied equally on an advice given by a financial advisor and a robo-advisor. We have also organized workshops in June 2018 and 2019 where we discussed some of the results. Top european researchers within the field presented their work.

Financial advisors are tempted to give advice that are based on self-interest rather than the interests of their clients. However, advisors (presumably) also give good advise, despite incentives to do otherwise. In this research project we will investigate, by the use of controlled experiments, under which conditions advisors behave fairly, offering good advice, and under which conditions they act selfishly, offering advice that are harmful to their clients. We will also investigate under which condition clients actually follow the advisors' advice. In order to investigate this, we will run large-scale online experiments - as well as lab experiments - studying advisors' behavior under conflicting interests. We use dictator games as baselines where the advisor dictates an allocation between herself and the client. We then transform the game to situations where the dictator/advisor gives a real advice to the client on what choice the client should make. That is: The advisor can choose between advising prospect A or prospect B, where prospect A gives higher fee to the advisor than prospect B, while prospect B gives higher return to the client than prospect A. The transformation from a dictator game to an advisor game can then have two distinguishable effects. On the one hand, it might induce higher moral costs. Making an unfair advice may feel worse than dictating an unfair allocation, leading to more fair allocations in the advice treatments. On the other hand, the transformation can create a moral wiggle room. Since it is only an advice, the advisor may feel less obligated to induce a fair allocation. The project will give new insights into how one can induce financial advisors to give good advice, despite incentives to otherwise. This is highly relevant for policy makers, like the financial supervisory authority, who seek answers to the questions we pose.