• Arbeitspapiere

    • Probability Updating When Drawing Signals Without Replacement (with H. Mohrschladt, S. Siedhoff, and S. Stitz)

      Abstract: Individual biases in probabilistic belief updating have been typically examined in experimental settings where signals are drawn with replacement. Motivated by the variety of real world applications, we investigate belief updating in without replacement settings. In such settings, drawing a specific signal has two opposed effects on the likelihood to obtain another signal of the same type in a further draw. First, and equivalent to drawing with replacement, the signal provides information on the state of the world resulting in a higher probability to draw the same signal type again. Second, the drawn signal is removed reducing the probability to draw the same signal type again. We find that subjects severely underinfer with respect to the first effect and mildly underestimate the second effect. (SSRN)

    • Experimental Exploration of Money Illusion in Long-term Financial Decision Making (with H. Cordes and N. Branger)

      Abstract: Inadequate consideration of inflation when making long-term financial decisions – so-called money illusion – can have severe consequences for future financial wellbeing. Thus, it is essential to understand how different methods of informing private investors about inflation affect their investment decisions. We propose a novel mechanism for experimental remuneration that can be used to systematically explore money illusion and its consequences. The mechanism imitates the disparity between nominal wealth and real purchasing power by a declining conversation rate. We demonstrate the flexibility and virtues of the new approach through two experimental studies. The objective of the first study is to generate a calibrated measure of money illusion that can be incorporated directly into economic modeling. Meanwhile, the second study is designed to generally explore how behavior in informational settings that are more or less susceptible to money illusion deviates from an unaffected preference benchmark. We find a substantial degree of money illusion in Study 1 and document nuanced behavioral patterns, including meta-cognitive components, in Study 2. (SSRN)

    • Getting more Wisdom out of the Crowd: The Case of Competence-Weighted Aggregates (with M. Goedde-Menke, E. Diecidue and A. Jacobs)

      Abstract: This paper shows that group discussions can serve as an instrument to improve individuals’ calibration, which in turn strongly increases the accuracy of competence-weighted, statistical aggregates. We conduct an experiment in which participants estimate quantities and report their self-perceived competence for various judgment problems. In addition, they engage in group discussions with other judges on unrelated judgment tasks. We find that prior to participating in the group discussions, judges’ self-perceived competence and their estimation accuracy are poorly aligned, which causes competence weighting to perform worse than prediction markets and simple averaging. However, the information exchange facilitated by the group discussions improved judges’ calibration, raising the accuracy of competence-weighted aggregates on subsequent judgment problems to prediction market levels and beyond. (SSRN)
    • Timing and Skewness of Information Revelation: Evidence on Information Structures and Compound Lotteries (with E. Diecidue S. Nolte and J. Schneider)

      Abstract: We design a comprehensive experimental setup to study (i) intrinsic preferences for gradual information revelation (ii) with different skewness (positive, negative, or symmetric) (iii) in two information environments. In a compound lottery environment, symmetric information revelation is preferred, while in an otherwise equivalent information structure environment, positively skewed information revelation is preferred. In both environments, early resolution is preferred over late resolution, with the three types of gradual resolution positioned in between. Our study integrates the three dimensions (timing, skewness, environment) of intrinsic preferences only studied separately to date, relates the findings to behavioral theories, and shows that careful consideration of the information environment is necessary when investigating preferences for gradual information revelation. (SSRN)
    • Biases in investment allocation decisions (with J. Sparks and C. Fox)


      Online platforms have made financial investing easier and more accessible than ever before, but little is known about how the choice architecture of these platforms— both in terms of how information is presented and how preferences are elicited— influences investment allocation decisions. Past research has shown that many people tend to “naïvely diversify” their investment funds, allocating 1/n of the total to each of n available instruments. The current work presents new experimental evidence that choice architecture can bias decisions among naively diversifying investors in three distinct ways. We first demonstrate partition-dependence, in which investor allocations vary systematically with the arbitrary grouping of investment options (by vendor and instrument). We next demonstrate unit-dependence, in which investor allocations vary systematically with the units in which the investment quantity is specified (numbers of shares purchased versus dollars invested). Finally, we demonstrate procedure-dependence, in which investment allocations vary systematically with the way in which preferences are elicited (choosing versus allocating). We observe these results among sophisticated participants, using both naturalistic investments and simple well-specified lotteries with incentive-compatible payoffs. We close with a discussion of implications for improving the design of asset allocation procedures.

    • Generally Higher But Not Generally Better: How Subjective Expertise Influences Estimation  (In-)Accuracy (with L. Rettig)

      Abstract: In three studies, we analyse the effect of subjective expertise on estimation accuracy in an experimental setting. We confirm that estimates are overall biased towards the midpoint of the scale (50%). We argue that 50% serves  as an anchor reflecting estimation uncertainty and hypothesize that the impact of this nave prior lessens with increasing expertise. This prediction is confirmed on an aggregate level - individuals that stated higher expertise had more extreme estimates and, in our setting, lower absolute estimation errors. Surprisingly however, further analysis reveals that the magnitude of the true value is crucial for this relationship. Higher (lower) expertise did not generally lead to more (less) accurate, but to generally higher (lower) estimates. Thus, on average, experts only gave significantly better estimates when asked for estimates on high proportions (∼70%), while for the complementary low proportions (∼30%) non-experts performed even slightly better. We argue that our results bear many similarities to previous findings on the effect of expertise on probability estimations. We draw from theories on selective information seeking to connect our findings to previous literature and explain the underlying mechanisms. Our findings have practical implications for question design and the capture of expertise in estimation scenarios.
    • An Experimental Analysis of Annuity Aversion - The Role of Framing and Uncertainty (with S. Nolte)

      Abstract: To understand the factors that influence people in their retirement decisions is of great importance. Due to an ongoing shift in western societies from public to private pension systems it becomes more and more the task of individual retirees to insure against longevity risk. While economic theory suggests that annuities should account for large parts of a retiree’s income, only few retirees include private annuities in their retirement portfolio at all. Empirical data on what drives this annuity aversion is limited, so we suggest using laboratory experiments to test possible explanations for the low rates of annuitization. We propose an experimental setup and provide a possible framework for future experimental research in the context of annuities. In a first application, we use this framework to test the influence of framing and lifetime uncertainty on annuity aversion.  (SSRN)
    • Countering Money Illusion? How Personalization Affects the Consideration of Inflation in Consumer Financial Planning (with H. Cordes and C. Erner)

      Abstract: Financial institutions increasingly provide consumers with the opportunity to construct their personal inflation rate in the process of financial planning. As it can be expected that consumers perceive a personal inflation rate to be more relevant for their financial wellbeing than a general one, the provision of this service could counter their money illusion, the behavioral tendency to neglect the value-diminishing effects of inflation in financial planning. Contrary to this intuition, we find in a series of experiments that constructing a personal inflation rate reduces the consideration of inflation: The complexity of the personalization process reduces the subjective understanding of inflation and thereby the perceived competence in judging its effects properly. We conclude that the consumers’ perception of inflation is more nuanced than the naïve intuition predicts, and that current best practices of countering money illusion might have unintended consequences.
    • Do changes in reporting frequency really influence investors’ risk taking behavior? Myopic loss aversion revisited (with M. Weber and S. Zeisberger)

      Abstract: According to the behavioral concept of myopic loss aversion (MLA), investors are more willing to take risks if they are less frequently informed about their portfolio performance. This prediction of MLA has been confirmed in various experimental studies and the conclusion has been drawn that banks could in fact influence investors’ risk taking behavior by adjusting the frequency with which they give feedback. However, none of the existing studies has really provided an explicit test of this dynamic prediction. Instead it is simply assumed that the results from between-subject experiments translate to a within-subject scenario in which feedback frequency changes over time. To examine the scope of the phenomenon and to assess its practical relevance, we present the first experimental study of MLA that directly addresses the dynamic prediction and manipulates feedback frequency (and investment flexibility) within-subject. Our analysis reveals that the impact of such dynamic changes is not as straightforward as commonly assumed. Stickiness and a general introspection component superimpose the standard MLA effect and generate unexpected dynamic patterns of risk taking behavior.