A laboratory experiment is used to test whether algorithm aversion occurs particularly in decision-making situations where serious consequences are at stake. It is shown that the willingness to use an algorithm that is recognizably more powerful than a human expert decreases when the decision is particularly important.

In an online experiment involving 402 participants from the United States, actual ex-ante stock price forecasts and actual ex-ante investment decisions are compared to the participants’ levels of confidence and overconfidence. The results show that confident and overconfident participants are more inclined to leaving themselves exposed by expressing strong opinions when forecasting stock prices. Furthermore, the results indicate that women and men differ in their levels of confidence and exhibit varying degrees of overconfidence. Finally, it can be concluded that confidence is a better predictor of subjects’ forecasting behavior than overconfidence.

A laboratory experiment is used to test whether algorithm aversion occurs particularly in decision-making situations where serious consequences are at stake. It is shown that the willingness to use an algorithm that is recognizably more powerful than a human expert decreases when the decision is particularly important.

Im Rahmen eines Online-Experiments, an dem 402 Probanden aus den USA teilnehmen, werden tatsächliche Ex-ante-Aktienkursprognosen und tatsächliche Ex-ante-Investitionsentscheidungen ins Verhältnis zu den Ausmaßen des Selbstbewusstseins (Confidence) und der Selbstüberschät zung (Overconfidence) der Probanden gesetzt. Es zeigt sich, dass selbstbewusste und sich selbst überschätzende Probanden eher geneigt sind, bei Aktienkursprognosen meinungsstark aufzu treten und sich mit den eigenen Prognosen zu exponieren. Es zeigt sich ferner, dass Frauen und Männer unterschiedlich selbstbewusst sind und in unterschiedlichem Maße Overconfidence an den Tag legen. Schließlich kann festgestellt werden, dass die Confidence besser geeignet ist, das Prognoseverhalten der Wirtschaftssubjekte zu erfassen als die Overconfidence.

Does a higher level of ignorance lead to higher subjective confidence in making stock price predictions? 150 subjects make stock price forecasts for three listed companies. A Likert scale is then used to measure how confident the subjects are that their predictions will actually come true. Additionally, the level of knowledge and experience relevant to the stock market was measured by knowledge questions. The results show that individuals with limited specialist knowledge and experience are particularly confident in their forecasts and vice versa. This finding is evident for men and is statistically highly significant, but it is not for women.

Im Rahmen eines Laborexperimentes wird geprüft, ob Algorithm Aversion insbesondere in Entscheidungssituationen auftritt, bei denen gravierende Konsequenzen drohen. Es zeigt sich, dass die Bereitschaft, einen im Ver-gleich zu einem Experten erkennbar leistungsfähigeren Algorithmus einzu-setzen, zurückgeht, wenn es bei der Entscheidung um besonders viel geht.

Die vorliegende Studie befasst sich mit dem Einfluss der Nachhaltigkeitsbemühungen von Immobiliengesellschaften auf deren ökonomischen Erfolg. Für den Zeitraum von 2017 bis 2023 werden 41 große europäische Immobiliengesellschaften daraufhin überprüft, ob die ESG-Ratings der London Stock Exchange Group (Refinitiv) die relative Jahresrendite wesentlich beeinflusst. Es zeigt sich, dass kein nennenswerter Zusammenhang zwischen den ESG-Ratings einerseits und den relativen Renditen andererseits besteht. Dies gilt sowohl, wenn man die ESG-Ratings in Beziehung mit den relativen Renditen jeweils desselben Jahres setzt, als auch, wenn man die ESG-Ratings mit den relativen Renditen des jeweils darauffolgenden Jahres in Beziehung setzt.

Welchen Einfluss hat die Anzahl der Handlungsalternativen auf das Ausmaß der Algorithmusaversion? Das ist die Forschungsfrage der vorliegenden Studie. Forschungsergebnisse im Bereich Choice Overload zeigen, dass eine Vielzahl von Alternativen häufig dazu führt, dass Wirtschaftssubjekte sich für eine leicht begründbare, zweckdienliche Alternative entscheiden. Choice Overload könnte somit die Neigung zur Algorithmusaversion dämpfen. Die Ergebnisse des vor-liegenden Laborexperiments bestätigen diese Vermutung jedoch nicht. Wäh-rend die Anzahl der Alternativen bei den männlichen Probanden keine Wir-kung entfaltet, zeigt sich bei den weiblichen Probanden sogar der entgegenge-setzte Effekt. Eine größere Zahl von Alternativen steigert bei Frauen die Nei-gung zur Algorithmusaversion signifikant.

Algorithmusaversion beschreibt eine Verhaltensanomalie, nach der Menschen effizienteren, algorithmusbasierten Systemen misstrauen und stattdessen menschliches Urteilsvermögen bevorzugen. Wirtschaftssubjekte laufen damit Gefahr, nicht ihren maximal erreichbaren Nutzen zu realisieren. Diese Studie soll einen Beitrag zu der Frage leisten, wie Algorithmusaversion reduziert wer-den kann. Im Rahmen eines Laborexperiments wird dafür überprüft, ob die bereits intensiv erforschte, wirkungsvolle Verhaltensanomalie der Verlustaversion zur Reduktion von Algorithmusaversion beitragen kann. Tatsächlich zeigt sich, dass das Gegenteil der Fall zu sein scheint: Die Bereitschaft, einen im Vergleich zu einem menschlichen Experten erkennbar leistungsfähigeren Algorithmus einzusetzen, geht sogar zurück, wenn bei der Entscheidung ein Verlust droht. Dieser Befund stützt andere Forschungsergebnisse, wonach Algorithmusaversion bei schwerwiegenderen möglichen Konsequenzen verstärkt auftritt. Zur Verbreitung algorithmusbasierter Systeme scheint es daher angebracht zu sein, die mit ihrem Einsatz verbundenen Chancen auf Zugewinne zu betonen und sie nicht als Hilfsmittel zur Verlustvermeidung zu bewerben.

Algorithms already carry out many tasks more reliably than human experts. Nevertheless, some subjects have an aversion towards algorithms. In some decision-making situations an error can have serious consequences, in others not. In the context of a framing experiment, we examine the connection between the consequences of a decision-making situation and the frequency of algorithm aversion. This shows that the more serious the consequences of a decision are, the more frequently algorithm aversion occurs. Particularly in the case of very important decisions, algorithm aversion thus leads to a reduction of the probability of success. This can be described as the tragedy of algorithm aversion.

Although algorithms make more accurate forecasts than humans in many applications, decision-makers often refuse to resort to their use. In an economic experiment, we examine whether the extent of this phenomenon known as algorithm aversion can be reduced by granting decision-makers the possibility to exert an influence on the configuration of the algorithm (an influence on the algorithmic input). In addition, we replicate the study carried out by Dietvorst et al. (2018). This shows that algorithm aversion recedes significantly if the subjects can subsequently change the results of the algorithm—and even if this is only by a small percentage (an influence on the algorithmic output). The present study confirms that algorithm aversion is reduced significantly when there is such a possibility to influence the algorithmic output. However, exerting an influence on the algorithmic input seems to have only a limited ability to reduce algorithm aversion. A limited opportunity to modify the algorithmic output thus reduces algorithm aversion more effectively than having the ability to influence the algorithmic input.

Within the framework of a laboratory experiment, we examine to what extent algorithm aversion acts as an obstacle in the establishment of robo advisors. The subjects had to complete diversification tasks. They could either do this themselves or they could delegate them to a robo advisor. The robo advisor evaluated all the relevant data and always made the decision which led to the highest expected value for the subjects' payment. Although the high level of efficiency in the robo advisor was clear to see, the subjects only entrusted their decisions to the robo advisor in around 40% of cases. In this way, they reduced their success and their payment. Many subjects orientated themselves towards the 1/n-heuristic, which also contributed to their suboptimal decisions. As long as the subjects had to make decisions for others, they noticeably made a greater effort and were also more successful than when they made decisions for themselves. However, this did not have an effect on their acceptance of robo advisors. Even when they made decisions on behalf of others, the robo advisor was only consulted in around 40% of cases. This tendency towards algorithm aversion among subjects is an obstacle to the broader establishment of robo advisors.

Within the framework of an economic laboratory experiment, we investigate how Algorithm Aversion impedes the establishment of Robo Advisors. The participants have to cope with diversification tasks. They can do this themselves or they can entrust a Robo Advisor with this task. The Robo Advisor evaluates all relevant data and always makes the decision that leads to the highest expected value of compensation for the participant. Although the high performance of the Robo Advisor is obvious, the participants only trust the Robo Advisor in around 40% of all decisions. This reduces their success and their compensation. Many participants are guided by the 1/n heuristic, which contributes to their suboptimal decisions. Insofar as the participants have to decide for others, they noticeably put more effort into it and are also more successful than when they decide for themselves. However, this does not affect the acceptance of the Robo Advisor. Even in the case of proxy decisions, the Robo Advisor is only used in around 40% of cases. The propensity of economic agents towards algorithm aversion stands in the way of a broad establishment of Robo Advisors.

Purpose This paper aims to assess the quality of interest rate forecasts for the money markets in Argentina, Brazil, Chile, Mexico and Venezuela for the period between 2001 and 2019. Future interest rate trends are of key significance for many business-related decisions. Thus, reliable interest rate forecasts are essential, for example, for banks that make profits by carrying out maturity transformations. Design/methodology/approach The data that we analyze were collected by Consensus Economics through a monthly survey with over 120 renowned economists and were published between 2001 and 2019 in the journal Latin American Consensus Forecasts. The authors use the Diebold-Mariano test, the sign accuracy test, the TOTA coefficient and the unbiasedness test to determine the precision and biasedness of the forecasts. Findings The research reveals that the forecasting work carried out in Brazil, Chile and Mexico is remarkably successful. The quality of forecasts from Argentina and Venezuela, on the other hand, is significantly poorer. Originality/value Over 50 studies have already been published with regard to the accuracy of interest rate forecasts, emphasizing the importance of the topic. However, interest rate forecasts for Latin American money markets have hardly been considered thus far. The paper closes this research gap. Overall, the analyzed database amounts to a total of 209 forecast time series with 28,451 individual interest rate forecasts. This study is thus far more comprehensive than all previous studies.

The neoclassical market model still has a decisive influence on important economic policy decisions today. A central role in this model is played by the formation of equilibrium prices, where aggregate supply functions and aggregate demand functions meet. We examine whether equilibrium prices are actually formed. For this purpose, we analyse 2,217 prices for homogeneous products that were collected by students between October 2020 and May 2022 in stationary and online retail. In 143 of 146 cases, no equilibrium price is found. The percentage price range is regularly over 100%. The presumed steering function of an equilibrium price does not materialise. The establishment of market mechanisms for the efficient solution of economic problems must therefore be questioned.

: Although algorithms make more precise forecasts than humans in many applications, decision-makers often refuse to resort to their use. In an economic experiment, we examine whether the extent of this phenomenon known as algorithm aversion can be reduced by granting decision-makers the possibility to exert an influence on the design of the algorithm (an influence on the algorithmic input). In addition, we replicate the study carried out by Dietvorst et al. (2018). This shows that algorithm aversion recedes significantly if the subjects can subsequently change the results of the algorithm – and even if this is only by a few percent (an influence on the algorithmic output). The present study confirms that algorithm aversion is reduced significantly when there is such a possibility to influence the algorithmic output. However, exerting an influence on the algorithmic input seems to have only a limited ability to reduce algorithm aversion. A limited opportunity to modify the algorithmic output thus reduces algorithm aversion more effectively than having the ability to influence the algorithmic input.

Obwohl Algorithmen in vielen Anwendungsgebieten präzisere Prognosen abgeben als Menschen, weigern sich Entscheidungsträger häufig, auf Algorithmen zurückzugreifen. In einem ökonomischen Experiment untersuchen wir, ob das Ausmaß dieses als „Algorithm Aversion“ bekannten Phänomens reduziert werden kann, indem Entscheidungsträgern eine Einflussmöglichkeit auf die Ausgestaltung des Algorithmus eingeräumt wird (Einflussmöglichkeit auf den algorithmischen Input). Zusätzlich replizieren wir die Studie von Dietvorst, Simmons & Massey (2018). Darin zeigt sich, dass die Algorithm Aversion deutlich zurückgeht, sofern die Subjekte am Ende die Ergebnisse des Algorithmus – und sei es nur um wenige Prozent – verändern können (Einflussmöglichkeit auf den algorithmischen Output). In der vorliegenden Studie bestätigt sich, dass die Algorithm Aversion bei einer Einflussmöglichkeit auf den algorithmischen Output signifikant zurückgeht. Eine Einflussmöglichkeit auf den algorithmischen Input scheint allerdings nur bedingt geeignet, die Algorithm Aversion zu reduzieren. Die begrenzte Möglichkeit zur Modifikation des algorithmischen Outputs reduziert die Algorithm Aversion effektiver als die Möglichkeit, Einfluss auf den algorithmischen Input zu nehmen.

In the context of an experiment, we examine the persistence of aversion towards algorithms in relation to learning processes. The subjects of the experiment are asked to make one share price forecast (rising or falling) in each of 40 rounds. A forecasting computer (algorithm) is available to them which has a success rate of 70%. Intuitive forecasts made by the subjects usually lead to a significantly poorer success rate. Feedback provided after each round of forecasts and a clear financial incentive lead to the subjects becoming better able to estimate their own forecasting abilities. At the same time, their aversion to algorithms also decreases significantly.