Schedule - Parallel Session 7 - Indvidual Decision Making

IMC Room 246 - 09:00 - 10:30

A Dynamic Analysis of the House Money Effect

Hauke Jelschen; Ulrich Schmidt


We report results of an experimental approach to disentangle and investigate the influence of two widely known mechanisms on risk taking behavior in a two-stage dynamic setting involving real monetary gains and losses, namely the effect of windfalls gains (Arkes et al., 1994) and the house money effect (Thaler and Johnson, 1990). We define a windfall gain as a riskless and unanticipated gain with the effect’s intensity being associated with the history of the individuals’ endowment in the experiment, whereas house money is considered the difference between the current stake and the initial stake which has been acquired by taking risk.
We ran a two-stage lottery game which is a modified version of Gneezy and Potters’ (1997), to elicit individuals’ risk attitude and risk attitude changes after winning or losing in the first stage.
Employing four treatments, we aimed at effectively varying the extent to which participants regard their initial endowment as part of their own assets rather than a windfall gain. In the Standard Treatment, individuals were just endowed with 8€ (same amount in all treatments) at the beginning of the sessions. In both of two Time Treatments, we temporally separated the payment of the endowment from the actual gambling task by one week. The difference between Time 1 and Time 2 was that participants made an additional decision about their behavior in the first round in Time 2 at the time the endowment was paid, but could revise that decision one week later at no (monetary) cost. The fourth treatment involved participants completing questionnaires and being compensated with 8€ for their effort (Work treatment), the lotteries were played subsequently.
We observe an unambiguous and significant ranking of risk aversion between treatments by comparing first round decisions, with its degree being highest in the Work treatment, intermediate in the Time 1 treatment and lowest in the Standard treatment.
In contrast, we do not observe any clear effect of first round outcomes on risk attitude in the second round. The only significant change in risk aversion occurs in Time 2 for both, first-round winners and losers, but cannot reasonably be associated with an influence of the prior outcome and is best explained by the strong anchor of the first decision.
On the one hand our results reveal a strong effect of the endowment’s history on risk attitude, corroborating the existence of a windfall gain effect, but on the other hand do not exhibit any indication that winning or losing in the first round alters individual risk attitude in a predictable way, challenging the existence of a house money effect as defined in this context.
To the best of our knowledge, this approach is the first to vary the endowment’s history in a multi-stage gambling task in such a way. Our results suggest to carefully differentiate between a windfall gain effect and the house money effect. They should also intensify the awareness that paying experimental endowments as windfall money may decrease observable risk aversion significantly.

Hauke Jelschen

Student, Kiel Institute for the World Economy

Preference Reversal: a Common Behavior rather than an Individual Abnormal Conduct

Bodo Vogt; Shi Xing Han


This study aims to re-exam the preference reversal phenomenon individually of each subject rather than regard the subject group as a whole. The lottery pairs used in the elicitation questions include a P-bet and a $-bet however with the value to win in $-bet to be defined by the subject him/herself. The result indicates that although with large difference in frequency, every participant had a certain degree of preference inconsistency, which leads to the conclusion that preference reversal may be a common behavior rather than an individual abnormal conduct. Future research should aim to discover which personal characteristic has an impact on the extent of reversal.

The Behavioural Economics of Attention and the Implications for Choice

Stefano Testoni


I model choice using a two-stage framework. The first stage of information processing involves allocating attention over alternative channels, while the second stage of decision-making consists of maximising preferences over alternative choices conditionally on attention allocation. To model attention, I follow the dual-system paradigm in psychology, which posits that information processing is jointly regulated by the fast, automatic System 1, which does not consume attentional resources, and by the slow, deliberate System 2, which does consume resources. An agent allocates fluency and effort over System 1 and System 2 channels, respectively. The allocation of fluency is produced by a combination of exogenous contingencies and of the associative properties of System 1, and it takes place without the agent’s deliberation. The agent chooses the allocation of effort subject to scarcity constraints and processing costs, conceived both as opportunity costs and as psychological aversion to the exertion of effort. In equilibrium, effort is mobilised up until utility and cost equal at the margin. The choice of effort is though conditional on fluency, due to the influence that System 1 exercises over System 2. The allocation of attention then defines the preference ordering of alternative choices in the stage of decision-making, de facto selecting one set of preferences among an entire family of possible sets of preferences. The model of attention offers a psychologically grounded account of the processes of choice, and it can be extended to notions of welfare other than preference satisfaction.

Stefano Testoni

Research Student, London School of Economics and Political Science

Income in Jeopardy: How Losing Employment Affects the Willingness to Take Risks

Malte Preuss; Clemens Hetschko


The willingness to take risks strongly affects economically important outcomes such as entrepreneurial activity, migration and households’ allocation of financial assets. Some part of individual risk attitude is rooted in genetic dispositions, socialisation and personality development. Beyond that, life experiences such as poverty, child birth, being exposed to violence, the Great Depression or natural disasters shape people’s willingness to take risks. We analyse the risk-taking effect of another source of substantial individual risk concerning the vast majority of employees in market economies: losing one’s job. As approaching and experiencing job loss places, first and foremost, workers’ current and future income in jeopardy, this event facilitates a natural experiment for studying the impact of an extensive income shocks on risk attitude. To the best of our knowledge, Sahm (2012) provides the only existing study on the impact of job loss on risk attitudes. Besides various other insights pointing to time-invariant risk-taking, she does not find that elder workers in the US change risk attitude in the wake of being dismissed. In contrast to her, we focus on job losses due to the closure of a complete plant or firm and thus on a much more specific type of dismissal. To the extent that workers cannot completely insure the income risk associated with such a non-controllable job loss, its impact on risk attitude will be that of a background risk. As a result, we argue that increasing risk of job loss will cause workers to avoid other controllable risks more often as decreasing absolute risk aversion (DARA) characterises their utility function. To test this notion by estimating the causal effect of loss of work on risk attitude, we apply a difference-in-differences approach based on German Socio-economic Panel data (SOEP). We assign workers who experience job loss due to the closure of the complete plant or company to the treatment group and similar employees who do not lose their jobs to a control group. As a behaviourally valid measure of general risk attitude, we use the stated willingness to take risks. It turns out that exogenous job loss indeed decreases the willingness to take risks. The effect already begins to manifest itself before the job loss event ultimately occurs, as workers may perceive that employment is increasingly at risk. Pre-treatment hourly wage as proxy for the losses of earnings and nonwage benefits associated with job loss amplifies the negative impact of job loss on risk-taking. This confirms that the losses of current income and the fear of losing future income are driving forces behind the impact of job loss on risk-taking. In the aftermath of the event, the willingness to take risks gradually returns to its initial level as workers become reemployed. Additional empirical analyses point to the behavioural validity and economic significance of our findings.

Malte Preuss

PhD Student, Freie Universitaet Berlin