ABSTRACT. Aim The purpose of this research is to explore gender differences in stock risk understanding. The 2019 Survey of Consumer Finances (SCF) includes three financial literacy questions, one of which focuses on stock risk. Researchers have consistently shown that women are less likely to invest in stock and score lower on financial literacy quizzes than men; however, women lack confidence and tend to respond to questions with “do not know” (e.g., Lusardi, 2011). Using SCF data, we investigate the gender difference in the likelihood of answering the stock risk question correctly while adjusting for other variables both with and without the “do not know” response included.

Methods The data for this study used come from the 2019 U.S. Survey of Consumer Finances (SCF), a large, nationally representative dataset that provides information on the assets, liabilities, and financial characteristics of households. The newly added financial literacy questions in the 2019 SCF focus on (1) stock risk, (2) interest rates, and (3) inflation. We investigate the characteristics associated with the first question on stock risk. This question is, “Do you think that the following statement is true or false: Buying a single company’s stock usually provides a safer return than a stock mutual fund?” We estimate a logistic regression model, with the dependent variable equal to 1 if the respondent answered the question correctly and 0 if not. Next, we estimate the model excluding the respondents who answered “do not know” to the stock risk question. We include gender as an independent variable in the logit model along with the following demographic, socioeconomic, and financial characteristics: net worth (log value), income (log value), age, education, race/ethnicity, marital status, expecting an inheritance, and use of a financial planner.

Results In the univariate analyses, men were significantly more likely than women to answer the stock risk question correctly when including the group who answered “do not know,” but the gender difference was eliminated after removing the “do not know” group from the sample. Logistic regression was used to explore the factors related to answering the stock risk financial literacy question correctly. When including the group who answered “do not know,” variables that were significant in explaining the likelihood of the question correctly were being female, net worth, being a college graduate, and being Black. After eliminating those who answered “do not know” to the stock risk question, being female was no longer significant, while the other variables remained significant in explaining whether this question was answered correctly. The current results show that many women are as financially knowledgeable as men on the risks associated with stocks and mutual funds. The results also show that categorizing "do not know" as an incorrect question may be problematic in research analyses, as researchers have shown that women are underconfident and thus may answer "do not know" even when they do know the correct answer.

References

Lusardi, A. (2011). https://www.nber.org/papers/w17107

ABSTRACT. We study if gender differences in financial literacy partly can be explained by differences in the response categories, more specifically, if gender differences are smaller when respondents are not given the option of answering “I don’t know” to questions about financial concepts. In addition, we investigate the role financial self-efficacy has in explaining gender differences in a) the tendency to prefer to answer “I don’t know” to questions about financial concepts and b) self-assessed financial literacy. The study is motivated by a recent Dutch survey experiment (Bucher-Koenen, Alessie, Lusardi, & van Rooij, 2021) that showed that the gender differences in financial literacy was reduced when the response option “do not know” was removed. Predictions based on a latent class model suggested that one-third of the financial literacy gender gap may be attributed to women’s lower confidence levels (Bucher-Koenen et al, 2021). In this study, we replicate this experiment using a higher number of literacy questions than in the original study. In addition, we include a direct measure of financial self-efficacy (Rothwell & Wu, 2019) to further explore the relationships between financial self-efficacy, measures of financial literacy, and self-assessed financial literacy.

Studies of the causes of gender differences in financial literacy are important since it is a potential explanation for observed gender differences in financial behaviour. The tendency for women to work less and invest their money more conservatively than men (e.g. Almeberg & Dreber, 2015), results in disadvantageous financial outcomes. Women typically hold lower amounts of wealth than men (e.g. Lusardi & Michell, 2008) and they are often found to be more financially fragile (Hasler and Lusardi, 2019). Hence, identifying the factors that produce the observed gender differences in financial behaviour is important when designing education programs in personal finance.

The tendency for women to be less financial literate than men has been found in nearly all countries in which financial literacy has been measured (Klapper & Lusardi, 2020; Hasler & Lusardi, 2017; OECD, 2015). One suggested explanation for this, is that women may be less confident than men. This notion was to a certain extent supported in Bucher-Koenen et al’s study (2021) where respondents were asked to rank their confidence in their answers to the financial literacy questions. The findings showed that women expressed significantly less confidence in their answers than the male respondents did.

In this study we include a direct measure of financial self-efficacy to further explore explanations for the gender differences in financial literacy. Self-efficacy refers to a person’s perception of own capability to accomplish a certain performance (Bandura, 1982). This perception may be accurate or faulty, but it will affect a person’s prospective behaviour, which in turn will strengthen or weaken the self-efficacy. Typically, self-efficacy determines if a person will avoid or pursue a certain action, or how persistent and how much effort he or she will put into a difficult task (Bandura, 1982). Self-efficacy is often seen as equivalent to “confidence”, but Bandura argued that confidence reflects a general strength of belief, while self-efficacy involves both a strength of belief and affirmation of the capability (Bandura, 1997). Financial self-efficacy refers to a person’s perception of financial capability and is expected to impact financial behaviour. Previous research has documented gender differences in financial self-efficacy, where women typically report lower levels of financial efficacy than men (Bannier & Schwarz, 2018; Rothwell & Wu, 2019).

Data will be collected from a representative sample of the Norwegian population above 18 years (n=2500) in January/February 2022: The participants are part of a large consumer panel administered by KANTAR. The questionnaire includes 25 financial literacy questions intended to measure understanding of concepts such as interest, inflation, risk, loans, and investment. The 25 questions are a subset of the scale developed by Nicolini (2019). These questions include the so-called Big Three financial literacy questions that often are used in surveys of financial literacy. “To measure financial self-efficacy a four-item FSE scale from Rothwell and Wu (2019) is combined with two items from a 6-item FSE scale (Lown, 2011). The six items are adapted and translated into Norwegian and included in the questionnaire. Survey participants will be randomly assigned into two groups. One group will be given the option of answering “I don’t know” to the literacy questions, while the other group will not get this option. The answers from the two groups will be compared to assess the effect of including a “I don’t know”-option. Further, we will analyse the relationships between the financial literacy scores, financial self-efficacy, and the respondents’ self-assessed financial literacy, with a special focus on gender differences.

References Almenberg, J., & Dreber, A., (2015). Gender, stock market participation and financial literacy. Economics Letters 137: 140-142. Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37(2), 122. Bandura, A. (1997). Self-Efficacy. The Exercise of Control. New York: WH Freeman and Company. Bannier, C. E., & Schwarz, M. (2018). Gender- and education-related effects of financial literacy and confidence on financial wealth. Journal of Economic Psychology, 67, 66-86. Bucher-Koenen, T. , Alessie, R., Lusardi, A., & van Rooij, M. (2021). Fearless Women: Financial literacy and stock market participation. Discussion Paper, no 21-015 03/2021. Hasler, A. & Lusardi, A. (2017) The gender gap in financial literacy: A Global Perspective. GFLEC Report. The George Washington University School of Business. Klapper, L., & Lusardi, A. 2020. Financial literacy and financial resilience: Evidence from around the world. Financial Management 49(3): 589-614. Lown, J. M. (2011). Development and validation of a financial self-efficacy scale. Journal of Financial Counseling and Planning, 22(2), 54. Lusardi, A., & Mitchell, O. S. (2008). Planning and financial literacy: How do women fare?. American Economic Review, 98(2), 413-17. Nicolini, G. (2019). Financial Literacy in Europe: Assessment Methodologies and Evidence from European Countries. Routledge. OECD/INFE. (2016). "OECD/INFE International Survey of Adult Financial Literacy Competencies". Retrieved from: https://www.oecd.org/finance/OECD-INFE-International-Survey-of-Adult-Financial-Literacy-Competencies.pdf Rothwell, D. W., & Wu, S. (2019). Exploring the Relationship between Financial Education and Financial Knowledge and Efficacy: Evidence from the Canadian Financial Capability Survey. Journal of Consumer Affairs, 53(4), 1725–1747.

ABSTRACT. Objectives and Motivation: Systematic differences between men and women have been consistently found in relation to their financial behavior, knowledge, and attitude. For instance, although more women than men in industrialized society attend college and university, a persistent gender gap in which men do better than women is usually observed in financial literacy (Bucher-Koenen et al., 2017; Fonseca et al., 2012; Lind et al., 2020; Lusardi and Mitchell, 2008). Another commonly observed difference is that men are more likely than women to be overconfident in the financial domain (Barber & Odean, 2001; Camerer & Lovallo, 1999), whereas women show higher levels of financial distress and anxiety (Fenton-O'Creevy, & Furnham, 2021; Lind et al., 2020). In addition, although women are increasingly part of the banking and finance sector (von Hippel et al., 2015), finance is still largely regarded as a masculine domain and stereotypical beliefs about gender and finance have been shown to exist already in young children (Driva et al., 2016). Such gender stereotypes in relation to money management can explain a large portion of gender differences in the financial domain (Tinghög et al., 2021), as they influence the way in which individuals evaluate themselves and their actual behavior. Understanding the exact nature and contents of such stereotypical beliefs is thus of paramount importance to develop policies aimed at empowering women, reducing gender inequalities and improving women’s financial behavior. Methods: A systematic review on gender differences about meanings attributed to money and about money management practices has been carried out. In addition, a qualitative study has been performed to better understand women’s relationship with money, finance, and banking sector. Five focus groups involving both women (as financial institutions clients) and financial advisors were conducted. Results. Results show that money has different gender connotations, and strong gender stereotypes are hold by women themselves. When associated with men, money stands for power and success and can be seen as an end in itself; when associated with women, money is mostly related to security, and it is seen as a means to an end. In addition, money is regarded by women as a way to reach autonomy and independence. Acquiring a large sum of money, however, has a different impact on male/female identity. Women tend to believe that being rich might enhance men’s identity, since having a lot of money is seen as a sign of high intelligence and entrepreneurial skills. Women’s wealth, instead, is less likely associated with personal skills or abilities, rather it is perceived as a matter of luck or family heritage. Other important differences between men and women are related to the financial education they received, the perceived self-efficacy in relation to money management and investment, and the relationship with bank institutions and financial advisors. In particular, women perceive financial matters and investments as too complex issues to be handled by themselves and are more likely to believe that such matters should be delegated to men or financial advisors. Despite the existence of such stereotypical beliefs – at least at implicit level – women show a great desire to gain a better understanding of financial matters and become more confident with such topics. Financial advisors also recognize that the relationship they develop with female clients is often very close to their ideal client relationship. Conclusion: Such preliminary results confirm the existence of numerous gender stereotypes in relation to money management and finance sector, which has traditionally been a male-dominated sector. Rising awareness and develop ad-hoc training for both financial institutions’ clients and financial advisors can be the first step towards breaking down the gender barriers and striving for parity.

ABSTRACT. This article describes a case study taking place at Sicredi, Brazil’s first credit union type financial institution, on financial education, economic psychology and behavioural science. Sicredi employs around 30 thousand people operating through over 2.200 branches spread over all states of the country. It offers different kinds of financial and non-financial products and services to its 5.5 million associates, in a cooperative system with 108 credit unions organized in a national system, based on a business model where funds remain locally, while strongly encouraging sustainable development.

Combining its cooperative nature and sustainability-oriented strategy to address Brazil’s social challenges, the institution has begun to provide financial education local actions to employees, clients/members, and the community around each of its branches at large. As a member of the UN Global Pact, these initiatives have been contributing to 2 of the UN 17 Sustainable Development Goals (end poverty and quality education), with two drives moving them: 1. the search for behaviourally informed evidence-based effective financial education, that has brought in economic psychology and behavioural science to their foundations, and 2. the importance of tailor-making it for greatly diverse audiences all over the country, while still keeping the same foundations, concepts, materials, and approach for all of them.

This process began in 2019, initially with workshops on mental functioning, decision-making, financial education with economic psychology and choice architecture, followed by the publication of a technical document addressing the rationale, foundations, goals, scope, team formation, method, benchmarks and general guidelines for a financial education program rooted on economic psychology and behavioural sciences, made available to the whole internal community after the program launch. This has been used as the basis to develop their program.

The method used to build the new program from conception and development to implementation, has been crowdsourcing, where employees from all over the country, holding different positions at the institution and with diverse educational, professional, and cultural background, have been brought together to contribute with insights and deep knowledge of the associates, their needs and the scenarios they are immerse in. Led by facilitators with expertise in innovative projects through collaboration, and divided into groups, they have discussed the designing of the initiatives aimed at adults, teens, children and microentrepreneurs, in all its details, addressing the following main topics: financial planning, saving, responsible use of credit, household financial organization, alternatives to raise income, avoiding scams and seeking reliable sources for help. They have thus offered their practical expertise to enhance content, language, techniques and tools at each stage of the project, followed by analysis and further research by an interdisciplinary team of financial educators, business analysts and designers, who would submit the consolidated version back to the first group, allowing it to advance in full collaboration towards increasingly finer versions of the program, under the supervision of an economic psychologist who works with financial education.

Economic psychology and behavioural science have been supporting the work in three pivotal modes: 1. Offering foundations for the design of the program itself, going from mental functioning, economic behaviours, motivations, decision-making and systematic errors, to more specific notions on economic socialization to guide the program for children at schools, including teachers and families as well, difficulties around self-control, the psychological meanings of money, and others, along with discussions about the role of financial education in society and how it cannot be effective if taken in isolation (preferably, it should come along with behaviourally informed regulation and consumer protection, plus the use of choice architecture strategies in addition to simple technical information); 2. A thorough systematic review of all the material to be used in each step of the program, from stimulating group discussions, to trainers’ training or addressing the final audience, always bearing in mind behavioural principles such as the true target being System 1/the hot self, the automatic way of thinking and behaving responsible for executing actions, therefore the importance of keeping it as simple as possible, dosing the amount of information provided, watching for involuntary biases of any nature, formatting for specific recipients, adjusting language and images, finding the best sequence for making sense and all the concerns about framing in general; 3. Choice architecture strategies to help close the gap between intention and action, stimulate actual behaviour change and maintenance of the new behaviour or habit, bringing in established studies, guidelines and proposals designed and tested by relevant actors throughout the world (the Behavioural Insights Team-BIT and Mindspace, from the UK, the World Bank, the Inter-American Development Bank-IDB, the Organisation for Economic Co-operation and Development-OECD, plus several instances of national strategies for financial education that have also embraced behavioural insights in their initiatives over the past few years).

The paper further details the process of developing the program and some highlights of the work dating back to 2019 up to early 2022 (the pandemic has made it virtual, but did not interrupt it), covering target audience, topics addressed and production of specific material throughout initial seminars, plus several workshops during the two phases of the program that have already been implemented. As a result, the Coops method has been conceived also using crowdsourcing, and is now being now disseminated at all Sicredi branches (Coops stands for Consciousness, Observation, Organization, Preparation, and Sustainability, also referring to cooperation). As a work in progress, assessment of the program is still under way, though already with some initial positive feedback from all those involved.

ABSTRACT. This study contributes to the understanding of the empirical link between risk and ambiguity aversion. It explores sources of heterogeneity in relationships between risk and ambiguity in the experimental economics and experimental psychology literature.

Risk and ambiguity aversion have received considerable attention over decades, and their role in economic decisions is central. Although these preferences have been regarded for a long time as fundamentally distinct, there is still no consensus on their empirical link in the literature. This link is captured by the sign and value of their correlation. Indeed, there are as many correlations (positive or negative) as uncorrelation results observed among the authors who computed it. Moreover, since theoretical models remain silent on this topic, it is currently impossible to establish the actual empirical relationship between risk and ambiguity aversion. However, several issues arise from the exact nature of their correlation: it could, for example, explain the value effect observed in financial markets (Bossaerts et al., 2010), would allow understanding how learning generalizes in uncertain environments (Jia et al., 2020) and would be crucial for theoretical modeling and policy-making (Tymula et al., 2012).

In this paper, I attempt to explain the existence of the different signs of correlation between risk and ambiguity aversion observed in the literature. This amounts to investigating the conditions under which the frequency of risk-averse behavior coincides or not with the frequency of ambiguity-averse behavior.

To conduct this analysis, I use a methodology based on a statistical analysis of data directly collected in research papers. I begin by examining the plausibility of the hypotheses proposed by some authors. For example, Camerer and Weber (1992), point out measurement errors in ambiguity aversion elicitation procedures, particularly joint evaluation, which can bias the value given to risky and ambiguous options and lead to zero correlation. This phenomenon is a direct consequence of comparative ignorance highlighted by Fox and Tversky (1995). Koch and Schunk (2013) hypothesize that a positive correlation is more likely to emerge if subjects can suffer real losses and if these losses are sufficiently large. In addition to these explanations, I also examine the role of the experimental design employed, the elicitation format, the parametric nature of the preference measurement indices, the type of experimental population employed, and the sample size. To test the previous hypotheses, I collect 74 correlation data from 50 different papers in a database, to which I associate several experimental features. Thus, I analyze which experimental features influence the chances of observing a positive or zero correlation. Sources of negative correlations are simply commented. The statistical analysis is conducted by considering the full sample, on the one hand, then on the other hand by segmenting the sample into subsets, namely the choice-based ambiguity elicitation tasks subset, the gain domain subset, and finally, the real payoffs subset.

Estimating probit models on the full sample confirms the role of ”comparative ignorance” (Fox and Tversky, 1995) induced by the way ambiguous options are presented to individuals. Therefore, presenting the lotteries separately increases the chances of observing a positive correlation between risk and ambiguity aversion. These contrast effects persist on the other subsets alongside the impact of using a student population.

References

Bossaerts, P., Ghirardato, P., Guarnaschelli, S., & Zame, W. (2010). Prices and allocations in asset markets with heterogeneous attitudes towards ambiguity. Review of Financial Studies, 23(4), 1325–1359.

Camerer, C., & Weber, M. (1992). Recent developments in modeling preferences: Uncer- tainty and ambiguity. Journal of risk and uncertainty, 5(4), 325–370.

Fox, C. R., & Tversky, A. (1995). Ambiguity aversion and comparative ignorance. The quarterly journal of economics, 110(3), 585–603.

Jia, R., Furlong, E., Gao, S., Santos, L. R., & Levy, I. (2020). Learning about the ellsberg paradox reduces, but does not abolish, ambiguity aversion. PloS one, 15(3), e0228782.

Koch, C., & Schunk, D. (2013). Limiting liability?—risk and ambiguity attitudes under real losses. Schmalenbach Business Review, 65(1), 54–75.

Kocher, M. G., & Trautmann, S. T. (2013). Selection into auctions for risky and ambiguous prospects. Economic Inquiry, 51(1), 882–895.

Trautmann, S. T., & Van De Kuilen, G. (2015). Ambiguity attitudes. The Wiley Blackwell handbook of judgment and decision making, 1, 89–116.

Tymula, A., Glimcher, P. W., Levy, I., & Belmaker, L. A. R. (2012). Separating risk and ambiguity preferences across the life span: Novel findings and implications for policy. Unpublished manuscript.

ABSTRACT. Many economic decisions rely on a fast, intuitive system of numerical cognition. When trying to judge the sum of a sequence of numbers, this system produces non-random errors: on average, it underestimates. This underestimation bias is thought to be caused by a compressive scaling of numbers when they are encoded internally. We present two preregistered, incentive-compatible experiments that tested the economic relevance of the underestimation bias. We varied the economic frame of sequences to be summed and deployed both judgment and forced-choice elicitation methods. Experiment 1 (n = 104) showed significant underestimation in the judgment task, with an overall mean bias of approximately -6%. Experiment 2 (n = 501) recorded persistent underestimation bias in both judgment and forced-choice tasks. Clear learning effects imply the effect size likely marks a lower bound. These findings have implications for modelling the cognitive foundations of economic preferences. They also provide insight into how firms’ pricing structures can exacerbate biases, causing economic loss.

ABSTRACT. This paper develops an intertemporal choice model to understand the role of social security for intergenerational redistribution when decision makers are disposed to overconsume but have bequest motives. I specifically propose a unified model which covers the three behavioral models to describe over-consumption among the young agents (Quasi-hyperbolic discounting, Temptation and commitment, and Reference-dependence with loss aversion). This research analyzes the welfare consequences under Social Security for those who have joy of giving to their offspring, under the scheme that an economy moves from a less funded social security system to a more funded one. Using a unified social security model, in which different social security plans are specified via different degrees of fundedness, I examine the effects of social security on savings, consumption, and capital accumulation for an OLG production economy. I find that an increase in fundedness intensity unambiguously increases capital accumulation and savings, but decreases consumption in steady states. I also find that increases in tax rates decrease savings unambiguously for all plans considered, whereas the effect on consumption is not conclusive.

First, this paper analyzes consumption and saving behavior for those whose preferences are characterized by the tendency of overconsumption during young age. Over-consumption may be justified with preferences such as hyperbolic discounting, temptation and self-control, or reference-dependence with loss aversion. The case of naïve hyperbolic consumers has been widely analyzed with many applications including social security as demonstrated in Schwarz and Sheshinski (2007) and Imrohoroglu, Imrohoroglu, and Joins (2003). Moreover, as argued by Schwarz and Sheshinski (2007), this type of bounded rationality model can break the annuity aspect of equivalence between a funded pension system and an optimal pay as you go (PAYG) system established in Sheshinski and Weiss (1981).

The model of self-control suggested by Gul and Pesendorfer (2001) considers decision makers who would benefit from commitment. In their approach where temptation is modeled as the possible cause of a commitment preference, the decision makers use self-control to resist temptation if commitment is not an option. Using an example of a consumption-saving model, Gul and Pesendorfer (2004) extend the concept into a dynamic setting and denote the conflict between the temptation to consume the entire wealth of the agent and his long-run interest of moderating this, but without the time inconsistency problem common in the hyperbolic discounting models. The work by Kumru and Thanopoulos (2008) is an application of this self-control preference incorporated into a lifecycle framework with social security. They show that social security plays a role as a device to reduce the temptation cost in the consumer preference and thus can mitigate the welfare cost of the unfunded social security.

This paper, focusing on the third possibility of modeling over-consumption, analyzes the role of social security in determining intergenerational redistribution when decision-makers have reference-dependent preferences. By the term reference dependence, I posit that a decision-maker's utility depends not only on his actual consumption, but also on comparison to his beliefs about optimal consumption. A reference-dependent decision-maker derives utility from comparison to the reference beliefs: it may be a gain to the reference utility or a loss to it. A loss is considered more important than a gain of the same size. The recent development in reference-dependent preferences models suggests that reference points may be constructed within the model. I specifically use Koszegi and Rabin (2006, 2009)'s equilibrium approach to determine the reference points. Given any information regarding future income streams, the decision-maker forms beliefs on what should be the proper consumption in the standard world.

Second, this paper investigates the effect of Social Security on consumer welfare through intergenerational redistribution. Social security may change consumers' saving behavior and thus affect consumption and welfare via rebalancing resources over time. Both the direction and size of this effect depends on the type of the social security system in place. To thoroughly study the effect of each pension plan, I introduce a model of a comprehensive social security system. The social security system is represented by its degree of fundedness-the portion of payroll invested for future benefit in the form of social security income. This implies that given a payroll-tax rate, the funded portion is subject to investment for the retirement income of the currently young people, while the unfunded portion is to be transferred to the currently old people. Zero-fundedness specifies PAYG system, in which all taxed amounts are used to finance the pensions of the currently old. Full intensity, by contrast, indicates a fully-funded social security system. Between these two extremes, there are many partially-funded ones. With this model, it is possible to evaluate many types of combined social security plans.

ABSTRACT. Time preference is an important issue in behavioral economics and behavioral finance and has been studied extensively in many contexts including financial management, financial literacy, consumption, health, savings, and risk-taking. In economics and finance, time preference is a major component of intertemporal decision making that affects individuals’ decisions and behaviors.

Discounting of delayed risky assets has caught the attention of several studies in the past, including studies that tested the structure and the relationship between discounting delayed and uncertain rewards. However, no consistent conclusions on these issues have emerged. A comparison between the discounting of delayed risky and certain outcomes showed that in most cases delayed risky outcomes are discounted less, yet several studies contradicted these findings, and found that delayed risky outcomes are discounted more heavily than certain ones. However, the majority of this literature focused on the probabilities and payoffs that affect a delayed risky asset’s discounting rate and did not consider additional factors such as asset complexity or numerical cues (e.g., the magnitude of an underlying asset or benchmark).

The current study tests experimentally whether options' realization times affect individuals’ evaluation of derivatives with different magnitudes of underlying asset outcomes. We recruited 195 participants through Prolific online platform and assigned them randomly to four groups that differed by type of asset. Participants were asked to price a lottery (i.e., an option whose outcomes are derived from an underlying asset with numerical – high or low and non-numerical outcomes) to be realized in different points in time. In addition, they were asked to discount the immediate lottery certainty equivalent if it was offered to them in two different time points. The results show that option's complexity did not affect prices, and this similarity in prices was also found in the presence of a low numerical anchor. However, when a high numerical anchor was present, price differences emerged for both assets that are realized today and in the future. However, subjective discounting rates did not differ. This finding indicates that realization times do not affect the additional premium on the basic asset price that individuals demand. To the best of our knowledge, this is the first study that examines whether the underlying asset creates an anchor that affects an option’s pricing and test its persistence overtime.

ABSTRACT. Realising that the overconsumption of animal-based proteins is associated with several negative health and environmental outcomes, an increasing number of people hold intentions to reduce their consumption of meat. However, intentions to reduce meat consumption are often not realised.

A large body of literature aims to explain the reasons and the moderators of such intention-behaviour gaps. Most of this literature focuses on stable personality traits and cognitive variables. Less attention has been paid to the question why a given person’s propensity to follow through on their intentions varies over time, and how time-varying factors relating to contexts and emotions might impede or promote consistency between intentions and behaviours.

In this study, we first estimate the intention profiles around meat consumption of a representative sample of the UK population. We then follow up with those people who report intentions to reduce their meat consumption and identify the contextual correlates of meat consumption. Specifically, we elicit the physical context (where they were), the temporal context (meal type and day of the week), the social context (who they were with), and the mental context (their antecedent states and the self-reported decision factors that mattered most in the moment).

To measure the contexts of (meat) consumption, we employ the Day Reconstruction Method, which is designed to collect data on the experiences a person has on a given day, through a systematic reconstruction conducted on the following day (Kahneman et al., 2004). While the method was developed to measure happiness and other experiences, it is also an adequate tool to measure context variables linked to behaviours as they occur in everyday life.

We use a sample from the online participant recruitment service Prolific. To identify the prevalence of intentions to reduce meat consumption in the UK, we ask a representative sample of 1500 participants about their intentions to reduce meat consumption. We complement this sample with a non-representative booster sample of 3000 participants to increase the participants with intentions to reduce meat consumption. This approach allows us to analyse data from 1462 consumption episodes from 547 participants who intend to reduce their meat consumption and provided information about their consumption at yesterday's breakfast, lunch, and dinner.

The results of the nationally representative survey suggest that 22% of participants in the UK intend to reduce their meat consumption in the next four weeks. The results from multilevel regression models using the data from the day reconstruction method suggest that intention behaviour gaps in meat consumption are most likely to emerge: on Sundays; when eating out at a restaurant or café; at dinner time; when eating with others (especially friends); when they were hungry in the lead up; and when convenience, nutrition and environment are not key decision factors and craving is.

The findings thus indicate that some (but not all) physical, temporal, and social factors are predictive of whether people eat meat or not. Moreover, antecedent inner states can play an important role. By drawing an extensive picture of the contextual factors associated with meat consumption amongst those intending to reduce their meat intake, our study can inform future interventions in the form of boosts, changes of the choice-architecture, as well as self-nudges to help people follow their intentions.

ABSTRACT. How to effectively protect nature on commonly-owned lands – the vast majority of forests in developing countries – is among today’s key environmental questions. Offering financial compensation conditional on conservation outcomes is effective in inducing conservation on private lands, but it may give rise to strong free-riding incentives on common property lands. We implemented a Randomized Controlled Trial in arid Burkina Faso to test the relative effectiveness of two collective conditional payment schemes – a linear payment scheme, in which group payments increase linearly with tree survival rates, and a threshold payment scheme. While extant theory predicts the latter to (weakly) outperform the former, we develop new theory that shows that the reverse may also hold – but only if the relationship between effort and tree survival rates is very uncertain. Our field experiment shows that threshold group payments increase intermediate measures of cooperation, but also that – consistent with Burkina Faso’s harsh climatic conditions rendering tree survival quite stochastic – actual survival rates are higher with linear group payments. We present both survey evidence and lab-experimental results to explore the mechanisms giving rise to these results.

ABSTRACT. It is often assumed that where people are unwilling to change their habits or to support policies to tackle climate change, this largely reflects a failure to understand the issue that we all face. But is this true? If so, then improving ordinary people’s understanding is a vital part of climate mitigation. If not, then, other barriers to change need to be identified.

Many studies have indeed shown associations between knowledge and support, but experimental evidence supporting the idea that improving knowledge increases support is rare. Moreover, studies on motivated reasoning provide reason to doubt this causal link. We present a pre-registered online experiment that tests the link between comprehension and support for mitigation.

Method The study was run in October 2021 (prior to COP26). Participants were 1,000 adults recruited by a leading market research agency to be broadly representative of the adult population in Ireland. They were paid €5 for undertaking the study, which took 25 minutes to complete on average. The main task was a quiz about climate change. The 10-minute quiz was designed to engage participants and to measure their understanding, not of facts and figures, but of the scientific relationships behind climate change – of what causes what, and how. Half the sample was randomly assigned to see the answers to the quiz questions, while the other half was not. We then measured support for climate mitigation policy (in particular, a carbon tax) and intentions for pro-environmental behaviour in the future.

Results Opinions on whether carbon taxation can shift businesses and households towards sustainability were somewhat divided, although a slight majority reported they believe it to be effective (M = 4.7, SD = 1.6). Quiz score was not significantly related to belief in the carbon tax’s efficacy. Seeing the quiz answers, however, was significant at the 10% level (p = .068), with those who received the answers reporting that they more strongly believed the carbon tax can shift behaviour. To illustrate the size of this difference, 37% of participants who read the answers to the quiz gave a 6 or 7 when asked about whether the carbon tax would be effective, compared to 29.4% of those who did not see the answers.

When asked about what they would set the price per tonne of carbon to, almost half of the sample (46.5%) increased the tax from its current level and almost one-third (31.1%) decreased it. Ordinal logistic regression models showed again that receiving answers to the quiz had a significant effect on choices (p = .010). Those who saw the answers set the price of carbon to be higher on average than those who did not see the answers. Analysis of the distribution of responses shows that 35.4% of those who did not see the answers decreased the rate of carbon tax compared to 26.6% of those who received the answers, representing a change in behaviour of one-in-four participants.

After responding to the carbon taxation questions, participants were asked about their own likelihood of engaging with different environmentally friendly behaviours in the future. Participants were less willing to engage in higher-impact behaviours (e.g. taking fewer flights) than lower impact ones (e.g. buying local food), but there was an interaction between impact and having seen the quiz answers. Compared to those who didn’t see the answers, participants who did were marginally more willing to engage in moderate-impact behaviours (e.g. switching from a petrol to hybrid car; p = .089) and significantly more willing to engage in high-impact ones (e.g. switching to a plant-based diet; p = .042), although effects are again small at approximately 6% of a standard deviation. There was no difference in their willingness to engage in low impact behaviours.

All effects persisted when socio-demographic controls were added to the models.

Discussion Engagement with scientific information about climate change increased belief in the effectiveness of a carbon tax, even though the information was not directly related to carbon taxation. The effect of the quiz answers was not small: scientific information on the causes, effects and ways to tackle climate change resulted in 25% more people believing carbon taxation to be a highly effective way of motivating behavioural change. Similarly, seeing the answers led to a 25% decrease in the number of people who believed that the price per tonne of carbon should be reduced. Identifying which pieces of information had the strongest influence on carbon tax opinion will require further research but can plausibly be linked to the questions with poorest performance (which related to the relative environmental impact of various individual actions). Alternatively, the effects may be the result of simply engaging with the information, such as confirming a guess or learning the correct answer after giving an incorrect one.

The results suggest, however, that giving people good scientific information about climate change has a much stronger effect on support for mitigation policies than on individual behaviour. While our statistical models showed that information on climate change increased willingness to engage with moderate- and high-impact behaviours, the effects were small. It could be argued that even small shifts in intentions in the right direction are helpful but, given the already tenuous links between self-reported intentions and behaviour, such small changes in intentions are unlikely to translate into the kind of behaviour change required.

Providing good information on climate change to the public has the potential to generate support for policies that experts agree are likely to be effective. However, effects of information are limited. Further experimental tests that examine the interplay between psychological factors and proposed economic policy will be essential to identify effective ways to address concerns and motivate change.

ABSTRACT. The latest IPCC report urge that carbon dioxide and other greenhouse emissions must be reduced and the pace accelerated (IPCC, 2020). At the same time, emissions from the construction industry and buildings account for 40% of total energy consumption (IPCC, 5th building). One initiative aiming to improve energy efficiency in buildings was The Energy Performance in Buildings Directive, in effect in most EU countries in 2008, Sweden included, requiring owners of properties to obtain energy performance certification (EPC) and make it available to customers when a property is for sale or rent (EPBD, 2010).

One incentive to encourage investment in upgrading from one EPC level to another before selling a property is if it makes potential buyers more interested in the building, thus increasing the competition, and willing to pay a higher price (WTP), yielding a pay-off to the investment.

The purpose of this study is to examine the impact of different EPC levels on the interest in and WTP for a single-family house, and of how it is communicated, with a letter (level A, C or E) or visually with a chart, controlling for other factors known or expected to influence interest and house prices such as place (where the house is located), size, urgency to buy, and qualitative aspects such as characteristics of the area and of the house as such, environmental aspects of the house, and socio-economic factors, as well as competition at the time of the bidding (from other potential buyers and by other objects for sale) and marketing activities by the real estate agent, most of this not controlled for in earlier studies based on actual sales data.

An survey-based experiment was carried out on the housing market among visitors to the biggest site for houses for sale in Sweden (Hemnet) during September 2020, resulting in 750 respondents who stated a WTP for the house presented to them, with all the specified characteristics they wished for of the house and of the area controlled for. The six experimental stimuli mentioned were randomly assigned the respondents, plus a control group with no information about EPC level.

A structural equation model (SEM), including the experimental variables and all controlling variables as latent variables, was hypothesized, and then tested using SmartPLS. The results show a significant and large impact of WTP levels on the interest in a house, but a much smaller and only indirect (via interest) influence on WTP. Presenting the EPC level as a graph instead of a letter increases the impact, positively when the EPC level is above the expected EPC level (given by the control group, with no information about EPC level), and negatively when the EPC level is below the expected EPC level, as hypothesized.

When no EPC information was communicated, the interest in and WTP for the house was between the C and A levels, thus indicating that the expected EPC level of a house for sale is about the B level (also found when comparing the mean interest and WTP for the different experimental groups and the control group).

As to the influence of control variables, the more important environmental aspects of the house were considered by the buyers, the less interest in the house, but no direct effect on WTP. The more urgent buying a house was considered, the more interest in the house offered, but neither in this case any direct effect on WTP. As expected, place had the largest impact on WTP (the path coefficient being above 0.8), followed by size of the house, primarily then indirectly via place (thus, if wanting a big house, buyers seem to look for such a house in more expensive areas).

The main conclusion is that a higher EPC level does increase the interest in a house for sale, and thus should increase the competition which in turn could increase the final price for the house. This could thus encourage investments in raising the EPC level. However, EPC levels do not directly influence WTP. An implication concerning how to communicate the EPC level is that if it is above the B level, it should be communicated as a graph, and if below the B level, it should be communicated as a letter, if wanting to stimulate interest in and WTP for the house.

ABSTRACT. Subjective well-being (SWB) has been the topic of research in economics, psychology, sociology, and other disciplines. SWB is generally decomposed into the components life satisfaction (LS) and emotional well-being (EWB). The former refers to cognitive judgement of satisfaction with life obtained on a single scale (the Ladder) or a multi-item scale (e.g. the Satisfaction with Life Scale), the latter refers to recall of experiences of positive and negative affect during a time segment of everyday life. It is generally assumed that LS is related to relatively stable life circumstances including income, health, and marital status. In contrast, emotional well-being is shaped by daily hassles and “uplifts” that may not be strongly influenced by stable life circumstances. In empirical studies EWB is defined as the balance of the frequency (or duration) and intensity of experienced positive and negative affect. Factors that interfere with positive affect or amplify negative affect may therefore have negative impacts. Such factors identified in previous research include worries for economic hardship, poor sleep quality, and long work commutes. In two studies we focus on the role of time pressure commonly defined as the experience of being unable to meet time limits. Experimental laboratory studies demonstrate that time pressure has detrimental consequences for work performance; other studies show that time pressure due to long work hours impair performance of household tasks and caring duties, and that a majority of people feel “rushed” even in free time. In affluent consumer societies people face many competing opportunities that are attractive and affordable. It appears difficult then to not over-commit oneself. For instance, shopping is almost becoming a compulsion such that it presumably infringes on time for other activities. In Study 1 we conducted an online survey of 240 employees of a Swedish human relations corporation. Questions were asked about EWB, satisfaction with work, family life, and leisure, stress-related symptoms, and time pressure. Constructed indexes based on replies to the questions were used in regression analyses to show a negative relationship between EWB and time pressure, partially mediated by satisfaction with work, family life, and leisure. We conjecture that the negative relationship is due to time pressure acting as an impediment to attainment of important goals in different life domains. The results also showed that the index of stress-related symptoms is a parallel mediator. Failure to cope with time pressure is a possible explanation. Such failures may still be limited to people who are emotionally instable. Study 2 followed up on the results of Study 1 with the aim of investigating how individual differences in coping with time pressure influence EWB. Another Swedish sample of 1,102 small-business owners was surveyed online. Indexes were constructed based on replies to questions about EWB, satisfaction with work, family life, and leisure, and time pressure. In addition, the Big-Five personality trait of Neuroticism (emotional instability) was measured. The results of separate regression analyses showed that time pressure is directly and indirectly through satisfaction with work, family life, and leisure negatively related to EWB, thus replicating the results of Study 1 for a different and larger sample. In line with that Study 1 demonstrated that the index of stress-related symptoms was a parallel mediator of the negative relationship between EWB and time pressure, the results also showed that Neuroticism is negatively related to EWB in three ways, directly, indirectly through time pressure, and by moderating the direct negative relationship between time pressure and emotional wellbeing. In conclusion, we claim that emotional wellbeing has a negative relationship to time pressure because it impedes attainment of important goals in different life domains and causes stress responses in susceptible people. Our data are cross-sectional and based on self-reports. Additional studies are needed to verify our results in longitudinal correlational or experimental research designs, preferably including also objective measures.

ABSTRACT. It is widely recognized that GDP is not an adequate measure of national welfare. GDP growth raises average living standards but income inequalities and other consequences of economic growth can neutralise its benefits. Moreover, what individuals value go way beyond living standards. In this context, the attention devoted to subjective well-being (SWB) as a more complete and adequate measure to capture the progress of a society has been growing.

Despite its limited role in accounting for the things that matter for individuals, GDP explains more than half of the variability of life satisfaction across countries (Sacks et al 2012). The role of income for SWB and, in particular, of GDP in explaining country differences in SWB, has captured a great deal of attention. This attention grew out from Easterlin (1974), who found that as Western societies have become richer after WW2, their populations have not become happier (at least in the West). Many perspectives have been advanced to explain this paradox, namely on the basis of social comparison, adaptation of individuals to income increases, and the increase of financial aspirations. Another argument for the failure of growing income in generating increasing levels of SWB can be attributed to the growing income inequalities in some countries, which might neutralise income growth.

The economic policy implication of Easterlin's paradox would be a loss of prominence for the economic growth objective, which would be rather disappointing. However, Stevenson and colleagues (Stevenson & Wolters, 2008; Sacks et al., 2012), using data from several surveys and from different periods of time, concluded that, after all, there is a strong positive relationship between GDP and SWB.

International differences in levels of SWB have also been explained on the basis of culture (e.g. Arrindell et al., 1997; Diener & Diener, 1995). Hofstede (1991, p. 5) defines national culture as the "collective programming of the mind that distinguishes members of one group of human beings from another". This is a multidimensional construct, which distinguishes five different cultural dimensions, namely power distance, collectivism / individualism, femininity / masculinity, aversion to uncertainty and long-term orientation. It is recognised that "culture influences the ingredients of a good life and conceptualisations of happiness" (Scollon, et al., 2013, p. 1). However, studies focusing on the effects of culture on global measures of SWB have been scarce. Most contributions have focused on how culture and, in particular, how individualism, relates to one facet of well-being, job satisfaction. Hence, little attention has been given to studies on culture and SWB and, in particular, to cultural dimensions other than individualism (Steel et al., 2018). Notwithstanding, Steel et al. (2018) concluded that culture is important for individual and national well-being. Specifically, they found that SWB relates positively to high individualism, and to low masculinity, power distance and uncertainty aversion. They also concluded that culture (individualism and masculinity) interacts with GDP per capita in influencing SWB.

Given the saliency of income and its distribution (mirrored in the GINI index) and the paucity of analysis tapping the role of culture in explaining country differences in SWB across countries, our aim is to contribute to knowledge by exploring how these determinants of SWB combine to generate different recipes for eliciting subjective well-being. To do so, we rely on fuzzy set qualitative comparative analysis (fsQCA).

fsQCA is an analytic technique grounded in set theory that calibrates case membership into sets, conceptualize cases as set theoretical configurations, and views causality in terms of necessity and sufficiency relations between sets (Misangyi et al., 2017). This technique allows for equifinality and asymmetric causality, meaning that equally effective configurations of conditions may lead to the same outcome, and that asymmetric causality configurations that lead to an outcome are different from the configurations leading to the negation of it (Ragin, 2008). In different words, different combinations of GDP, GINI index and cultural dimensions may be capable to equally deliver high levels of SWB (equifinality). Moreover, the combinations that engender a low SWB are not necessarily the opposite of those that yield high levels of SWB (asymmetric causality).

The fuzzy set intermediate solution obtained for high levels of SWB comprises four equifinal configurations. The consistency values for all configurations as well as for the overall solution exceeds the threshold of 0.8 (Ragin, 2008). High GDP and high individualism are core conditions present in the four configurations. The importance of GDP meets the expectations resulting from the literature (e.g. Dynan & Sheiner 2018; Sacks et al. 2012). The requisite of individualism in the identified configurations also finds ample support in the literature, since individualistic societies encourage the pursuit of personal goals over and above the need to meet social obligations (e.g. Ahuvia 2001; Arrindell 2001).

The four equifinal configurations that ensure a high level of SWB are as follows: High GDG, high individualism, low Gini, low uncertainty avoidance, low power distance; High GDG, high individualism, low uncertainty avoidance, high masculinity, low power distance; High GDG, high individualism, low long-term orientation, low uncertainty avoidance, low power distance; High GDG, high individualism, low Gini, high long-term orientation, high masculinity, low power distance.

Interestingly, our findings emphasize that both high and low levels of long-term orientation and masculinity can contribute to high SWB, depending on how each of these dimensions combine with other cultural dimensions and the Gini index.

This study also analyses configurations that lead to low levels of SWB. The intermediate solution obtained for low SWB also comprises four equifinal configurations. These configurations are not mirror opposites of the configurations that predict high levels of SWB, which supports causal asymmetry. Low GDP is the single core condition present in all configurations. Long-term orientation, uncertainty avoidance, masculinity, individualism, and power distance, and the GINI index can each contribute to or mitigate low SWB, depending on how each combine with the remaining causal conditions.

Hence, in this study we innovate by addressing the research question: which combinations of GDP, GINI index, and cultural dimensions yield high SWB, and which combinations result in low SWB?

ABSTRACT. Several definitions and measures of financial well-being (FWB) have been proposed in the scientific literature. Most measures position FWB as a multidimensional construct, but do so based on inconsistent operational definitions in which the nature and number of dimensions vary greatly across measures. The present study seeks to fill this gap by refining and broadening our understanding of the multidimensional nature of FWB. To this end, we started by critically examining and comparing existing measures of FWB to identify the core dimensions essential to a comprehensive operationalization of FWB. Of those measures, one stood out in its ability to provide a comprehensive coverage of FWB: The Multidimensional Subjective Financial Well-Being Scale (MSFWBS).

Despite its interest, validity evidence for the MSFWBS currently only comes from a sample of emerging adults recruited in Italy and Portugal. As a result, the extent to which the psychometric properties of scores on this instrument would generalize to a wider age range and to different cultural or linguistic groups remains unknown. Moreover, evidence of factor validity of scores on the MSFWBS was obtained through confirmatory factor analyses (CFA), which revealed factor correlations that were high enough to call into question the discriminant validity of some factors.

These correlations, however, are not surprising given that the MSFWBS is specifically designed to assess conceptually related facets of FWB. In this situation, statistical research has revealed that measurement models tend to benefit from the reliance on an exploratory structural equation modeling (ESEM) representation of the factors which makes it possible to freely estimate all cross-loadings between items and non-target factors. More precisely, ESEM thus makes it possible to assess factors using all of the relevant information included in all items, which has been shown to result in a more accurate estimate of the factors, their correlations, and their associations with other constructs, while remaining unbiased when cross-loadings prove to be unnecessary.

This study was designed to conduct an in-depth examination of the psychometric properties of scores obtained on the MSFWBS in an age-diverse sample of French-speaking Canadian adults (n = 454). First, we examined the factor structure of these scores while contrasting CFA, bifactor-CFA, ESEM and bifactor-ESEM solutions. Second, the measurement invariance of these scores across subsamples defined on the basis of their age, sex, personal income and household income was examined. Third, the convergent validity of these scores was examined in relation to other validated measures of FWB. Finally, the criterion-related validity of these scores was examined in relation to validated measures of theoretically-related constructs (i.e., perceived stress, psychological distress and life satisfaction).

Overall, these results supported the superiority of the bifactor-ESEM solution, while suggesting that two of the items might have been associated with the wrong factors. As a result, we considered an alternative bifactor-ESEM factor structure where these two items were associated with a different factor. The parameter estimates from this solution revealed that the modifications resulted in superior psychometric qualities and therefore, this optimized version of the MSFWBS was retained.

Our results supported the measurement invariance of this solution across groups of participants defined on the basis of their age, sex, personal income and household income. Our results also supported the convergent (with other measures of FWB) and criterion-related validity of scores obtained on the MSFWBS. These results indicate that FWB explains a significant portion of the variance in perceived stress (R² = 35.20%), psychological distress (R² = 22.30%) and life satisfaction (R² = 49.20%). Furthermore, scores on the global FWB factor were significantly associated with all three criterion variables, predicting lower levels of perceived stress and psychological distress, and higher levels of satisfaction with life.

Our study emphasizes the importance of conceptual clarity as well as appropriate and empirically supported operationalization to foster a greater consensus among researchers on FWB, so that theory and research may move forward in a coherent manner and thus better serve the interests of practitioners and other potential users.

ABSTRACT. Older adults who experience severe health shocks must make important economic and financial decisions, such as when to retire and how much to save for future medical expenses. In this research, we follow common empirical practices in the literature to estimate the effect of health shocks on various decisions of older adults. To establish a causal effect, we rely on the assumption that a health shock is a sudden and unpredictable event that can affect an individual’s willingness to accept risky propositions (Decker & Schmitz, 2016). Using a nationally representative sample of adults aged 50 or older drawn from the 1998-2016 waves of the Health and Retirement Study (HRS), the present study investigates whether and to what extent exogenous health shocks influence a set of economic outcomes and whether heterogeneity in such relationships exists according to gender, retirement status, and overconfidence in cognitive functioning. In this work, we define health shocks, given their abrupt, unpredictable, and severe nature, as overnight hospitalizations (Dobkin et al., 2016; García-Gómez et al, 2013; Lindeboom et al., 2016; Parro & Pohl, 2021; Schurer, 2017). Further, our identification strategy uses the randomness of the timing of a health shock by creating a counterfactual control group consisting of older adults who experience the same shock two years after the treatment group (Fadlon & Nielsen, 2019, 2021; Jones & McVicar, 2020). With the restricted sample, we exploit a difference-in-differences approach by using individual-fixed effects models.

Similar to previous literature, our findings show that health shocks influence older adults’ projections of labor market activities by decreasing the perceived probability of working for pay in the future by 0.44 percentage points and increasing the perceived probability of having reduced Social Security benefits by 0.89 percentage points, the latter of which are no doubt driven by reduced lifetime earnings. Older adults further alter their perceived mortality risk after health shocks by reducing their subjective life expectancy by 0.79 percentage points. While increased use of healthcare relates to health shocks with an 18.18 percent increase in the number of nights in a nursing home and a 0.28 percentage point increase in nursing home residency, older adults also perceive increased future out-of-pocket medical costs by 3.14 percentage points. These increases in current medical costs and future medical care after health shocks reduce the perceived probability of leaving an inheritance to their heir(s) by 0.41 percentage points. It seems that older adults trade a demand for leaving a bequest for increased current and future health care consumption.

Health shocks also promote older adults’ precautionary behaviors to protect their wealth against similar future health shocks, in response to which they allocate more wealth to risky assets, especially stocks, by 0.32 percentage points, expecting to have higher returns in the long term. Further, they are also more likely to report difficulty with money management by 0.57 percentage points following a health shock and receive help from family or financial professionals by 0.43 percentage points to manage their finances. Although weakly associated (p<0.10), health shocks also lead to increased life insurance ownership by 0.44 percentage points. Because life insurance is used as an estate planning tool—to leave liquid assets to surviving family members to meet immediate financial obligations and compensate for income loss upon the death of the primary income earner—an increased perceived mortality risk might induce older adults with health shocks to purchase life insurance. On the other hand, health shocks reduce long-term care insurance ownership by 0.33 percentage points. The availability of informal care from family members and friends (Pauly, 1999) and of nursing home care through Medicaid may crowd out demand for long-term care insurance (Brown, Coe, & Finkelstein, 2007).

To study heterogeneous impacts of health shocks on sub-populations we parse our sample by gender, retirement status, and overconfidence. There were differential responses by gender to future labor market participation, with a decreased probability of working for pay for older men and the reverse for older women. Also, the impact of health shocks on the number of nursing home overnight stays is greater for women and retirees. Health shocks increase current and future health care utilization only among subgroups of older adults. They increase nursing home residency for men and retirees and the perceived probability of needing future medical expenses for men and those with an increased overconfidence level. We similarly found differences in precautionary behaviors after health shocks. Older men and those with a decreased overconfidence level are more likely to delegate money management to others after health shocks, but health shocks also tend to increase life insurance ownership for women and long-term care insurance ownership for non-retirees.

The present study contributes to the literature in three ways. First, it examines the impact of health shocks on several outcomes usually overlooked in other studies. Without being limited to labor market outcomes, medical expenses, debt, and health behaviors, it extends the previous findings by including additional outcomes such as asset allocation, financial health measured by using financial ratios adopted in the financial advice industry, and decisions about precautionary behaviors such as delegation of money management and insurance ownership. Second, it investigates heterogeneity in the impact of health shocks on outcomes according to factors that other studies have not addressed, such as retirement status and overconfidence in cognitive functioning. Although overconfidence is a prevalent behavior trait that explains financial and health decisions, as far as we know, it has never been studied in the context of people’s responses to health shocks. Third, following Fadlon and Nielsen (2019, 2021) and Jones and McVicar (2020) to ensure causality, the study uses a sample consisting of a treatment group that includes older adults with health shocks in the current period and a counterfactual control group composed of those with the same health shocks in the future. In this way, we balance the characteristics of those in the treatment and control groups and assume that the timing of health shocks for those affected is randomized.

ABSTRACT. Samuel Pufendorf’s natural law, expressed in De Jure Naturae et Gentium from 1672, integrated moral philosophy, jurisprudence, government, and political economy. His political economy embraces theories of human behaviour, private property and the four stages, value and money, foundation of states, and principles of taxation. His foundation is what we today will call psychological explanation of individual behaviour. The social life of man is the foundation for his work. Human behaviour is determined in a struggle between the power of the will, as an internal director of his actions, and the psychological inclinations or passions. The driving forces are self-love and sociability. Attentions is drawn to the passions, which deviates from right reason, such as craving for luxuries, ambitions of man, envy, jealousy, perversion, and rivalries of wit, to mention a few. His human behaviour is the foundation of his other economic theories. His theory of value is presented as a rudimentary supply and demand analysis of prices. He discusses how human passions influence the aptitude, or demand, and therefore affects the prices of goods or services. In the 18th century, Pufendorf was the most read and used philosopher in Europe. In Scotland he was the primarily source for Gershom Carmichael, Francis Hutcheson, and Adam Smith. When Smith wrote The Theory of Moral Sentiments and The Wealth of Nations, he had Pufendorf’s natural law work ready at hand. He therefore built very closely on Pufendorf when he proposed psychological explanations of human behavior and laid the foundation of classical economics.

ABSTRACT. Tipping is practiced in a variety of different contexts, including tipping couriers and delivery people. The current study focuses on factors that influence the tip given to a courier delivering food from a restaurant. Home deliveries provide value to restaurants by expanding their markets, and the relationship between consumer and courier is characterized by tipping, as a social norm. This research examines the different effects on tipping of food deliveries couriers. To the best of our knowledge, there is a lack of research on deliveries. Moreover, previous research on tipping rarely discusses tipping of couriers. This study strives to fill this gap. We also test whether including gifts in a delivery from the restaurant influences tipping behavior. The gifts are used as a manipulation mechanism to ascertain influences on customer behavior. Theories on tipping and gifts make different predictions on factors acting directly on customer behavior; thus, it is unclear how the use of gifts as manipulations will affect courier tipping.

To test our hypotheses, we analyzed data from home deliveries in Israel. Deliveries from restaurants are common in Israeli life, and deliveries, in general, have become more crucial during the COVID-19 pandemic. Therefore, it is important to expand the research regarding of deliveries. The data in the current study was collected prior to the COVID-19 pandemic.

This field experiment collected data concerning deliveries made by a restaurant. The data include manipulations in which a complimentary dish or gift was given to the customers, and a control group that received neither. Orders were randomly divided between the experimental and control groups, such that all deliveries during a specific shift were in the same group (manipulation or control group). It analyzes the effect of environmental stimuli, variables such as the type of payment and delivery time, and also the influence of the two manipulations in which gifts were added to the order. The manipulations were a complimentary dish or gift added to the original order, compliments of the house. The first factor that should influence tipping to restaurant couriers is the order size. In restaurant setting, the tip percentage is influenced by the bill amount. Previous research show that the percentage of the tip is decreasing when the bill amount is higher. Based on the previous research the first hypothesis is: H1: In lager order size, the courier will receive smaller tips.

The second factor considered in this research is delivery time and whether the order arrives promptly, which is related to the effort made by the courier. Research on tipping in restaurants has measured servers’ effort as a way of defining the service quantity (Bodvarsson & Gibson, 1994), which is defined in the literature in different ways (Azar, 2007a). Service quantity could be defined as the time the server dedicates to the table or the number of dishes ordered by the table. The tip amount should be positively correlated with service quantity since the patrons decide on tipping based on the server’s effort (Azar, 2007b). In the delivery situation, as examined in this research, the service quantity is the effort of the courier, which customers evaluate when they tip. The impact of delivery time on remuneration influences the effort exerted by the courier. According to the Prospect Theory, a reference point affects a decision by providing a baseline for making that decision (Kahneman, 1992; Kahneman & Tversky, 1979). Restaurants often commit to delivering within a specified time, thus creating a reference point. Food delivered earlier is considered a gain and the reverse is regarded as a loss. Hence, delivery prior to the specified time should have a positive influence on tip size. Restaurants make decisions based on business considerations rather than factors related to research design. The current study uses data from deliveries that were done with the regular practices of the restaurant (except for adding a free dish or gift); the couriers were not instructed to delay some deliveries, in order to avoid doing any harm to the reputation of the restaurant. Some deliveries were delayed beyond the reference time for various reasons, but these were not intentional delays. Consequently, the data include mostly deliveries made on-time (i.e., at the reference time) or earlier. Based on the previous research concerning factors that influence tip size, the second hypothesis is: H2: Late deliveries (according to the restaurant's definition of on-time delivery) will receive smaller tips than on-time deliveries.

Next, we consider how a gift included with the delivery influences customers’ tipping behavior (Marchand et al., 2017; Papadopoulou et al., 2019). Strohmetz et al. (2002) found that customers who received a small piece of chocolate along with the check tipped more than customers who did not receive a candy. Frank (2020) suggests that gifts’ impact on tipping is reliable and can be generalized beyond gifts of candy from waiters. Although Frank (2020) also found that gifts do not uniformly increase tips in all situations and their effect varies depending on personal traits, in service industries, gifts do cause customers to reciprocate with larger tips. Thus, the hypothesis is: H3: A complimentary dish or gift (non-food item) added to the order will increase the courier's tip percentage.

The payment of the bill (order) in cash or credit could influence the tip size to the courier. The payment in cash could encourage round payments, thus the tip amount would be effected more than the payment of the bill is in credit. Thus, the hypothesis is: H4: The bill payment type (cash or credit) will affect the courier's tip percentage.

As the world is changing due to COVID-19, deliveries have become an everyday commercial interaction between customers and businesses. The volume of deliveries has increased significantly; thus, it would be interesting to explore current reasons and motivations for tipping.

ABSTRACT. Sharing economy platforms have been exploding for some time now, with many praising its economic, environmental, and social benefits (e.g., Botsman & Rogers, 2011). Platforms that facilitate house, ride, knowledge, or tool sharing also allow complete strangers to meet face-to-face and form new, potentially long-lasting social ties, independent of socio-economic background or age differences (Fitzmaurice et al., 2016; Parigi et al., 2013; Schor, 2015). This seems to be especially beneficial, as sharing among strangers is a relatively novel phenomenon (Schor, 2014). However, an idea that seemingly emerged from a willingness to foster social and environmental change (Botsman & Rogers, 2011), is more and more often becoming a platform for generating profit (Martin, 2016; Slee, 2015). Exchange of good deeds and kindness seem to transform into exchange of services for money (Shor & Charles, 2017). Indeed, in a recent qualitative study of the users of a Norwegian sharing platform, payment was identified by some users as being in conflict with the idealistic image of the site and undermined their participation (Akin et al., 2021). So, could such economization of private property potentially harm, rather than support, social relations? According to Fiske’s (1991, 2004) social relational model, exchange of benefits for money is characteristic of market pricing relationships. Market relationships do not allow for intimacy and emotional connectedness (Jiang et al., 2014; Mead & Stuppy, 2014; Vohs, 2015). When people engage in such relationships, they care about how much they get out of their investment and whether the repayment is of comparable value (Fiske, 1992; Zaki et al., 2021). In stark contrast to that, in communal relationships people offer benefits not because they expect anything in return but because they intend to increase the recipient’s welfare or happiness. Interacting with each other in a communal mode, people find it natural to be helpful, friendly, generous, and even altruistic (Clark & Mills, 1993). What is more, people seem to be unwilling to mix these two types of relationships and once a relationship starts being perceived through the market-pricing lens it is difficult to turn back to communal norms (Gneezy & Rustichini, 2000). For this reason, in the current ongoing project we aim to investigate how the use of money in an exchange within a sharing economy interaction with a previously unknown neighbor affects the perception of that relationship, as well as people’s subsequent willingness to help and socialize. Experiment 1a (N = 158) confirmed that a monetary deal with the neighbor initiated through a tool sharing application was perceived as resembling a market-pricing rather than a communal-sharing relationship. On the other hand, a non-monetary deal was perceived as reflecting a relationship that was more alike to a communal-sharing than a market-pricing relationship. In Experiment 1b (preregistered; N = 189) we further showed that these results were not caused by the presence of money in general (paying for access to the app) but occurred only if money was involved in a deal with the neighbor. The next two experiments focused on the consequences of using money within a sharing economy interaction. Experiment 2 (preregistered; N = 197) showed participants declaring a lower willingness to help a neighbor in need if a previously described scenario presented a monetary deal compared to a non-monetary deal. This effect was mediated by the perception of the relationship with the neighbor as less communal. Finally, in Experiment 3 (preregistered; N = 285) we wanted to see whether these effects were solely attributable to it being the monetary exchange, or perhaps the same results would be observed if the relationship followed a favor for a favor approach (an exchange relationship, Clark & Mills, 2011). Additionally, we tested whether the effects extend beyond a relationship with a particular person and affect participants’ willingness to socialize during a neighborhood picnic organized by the neighbor they interacted with through the app. Here, we observed the indirect effect of the experimental manipulation on the dependent variable via perception of the relationship as following the market-pricing rules, with no total effect present. Participants who had their dog walked by a neighbour for a monetary fee (as compared to walking it for free) perceived this relationship as being less communal, which then negatively affected their willingness to participate in the picnic. This effect was not observed when the interaction was based on a quid-pro-quo deal (when compared to a non-monetary deal). To sum up, this preliminary set of studies suggest that the presence of money within the sharing economy interaction might negatively affect social ties. This result seems to be driven by a drop in the perception that the relationship is governed by the communal-sharing norms of behaviour. Perception that the relationship follows market-pricing rules, which are typical for strangers, does not seem to affect social ties in the studied context.

ABSTRACT. In this paper, we examine the persistent use of a misleading signaling device for quality fitness professionals (exercise physiologists, exercise scientist and personal trainers) in New Zealand, Aotearoa (NZ). This speaks to the use of ‘bad’ heuristics in decision-making. We look at the implications of this on society with respect to injuries and the growth of a more sedentary population in NZ. We start by examining REPs (New Zealand Register of Exercise Professionals) which provides a critically important signal to consumers of the quality exercise professionals and gyms in New Zealand. Evidence from the New Zealand fitness industry strongly suggests that the certification provided by REPs, New Zealand’s primary officially sanctioned body to signal the quality of trainers and of gyms, is very, very weakly correlated, at best, with the actual level of qualifications of New Zealand’s exercise professionals and the gyms that employ them (Altman 2020). In fact, in some instances, we have a negative correlation between REPs certification and the level of qualification of exercise professionals (Altman 2020). Therefore, REPs certification, cannot be taken as a measure of uniform quality in New Zealand’s fitness industry. A REPs certified trainer can be at either a very low or relatively higher level of qualification. The same can be said for gyms that are REPs certified. We develop a more objective measure of qualifications (an alternative to REPs) based by years of education and years of industry experience as a practical and analytical benchmark (Altman 2020).

In the real world of an abundant, imperfect and asymmetric information, signals play a vital role in affecting consumer choice, as a heuristic, with regards to exercise professionals and the gyms where they work (Akerlof 1970; Spence 1973, 2002; Stiglitz 1975). Using ratings by a respected agency is something that rational consumers have done for decades as it saves on the overall search costs (economic and non-economic) of locating a quality gym and a qualified exercise professional. In some cases, this would be a good example of the use of fast and frugal heuristics that Gigerenzer (2007) elaborates on. In his narrative, heuristics are (decision-making shortcuts) that smart or rational individuals employ to make best possible (satisficing) decisions in the most efficient way possible given the reality of bounded rationality. Herbert Simon makes the point, in his discussion of bounded rationality, that rational individuals are constrained by their decision-making environment. Therefore, as a result, individuals develop sensible decision-making procedures in the face of these constraints. However, heuristics that are based on misleading signals, such as REPs, can generate serious errors in decision-making and yield significant personal and social costs. And, this would then be an example of ‘bad’ or sub-optimal heuristics. Imperfect and asymmetric information, generates significant ‘noise’ in the decision-making space that hinders the ability of decision-makers to accurately identify the quality of signals. This plays an important role in the pre-eminent role of REPs in the decision-making process.

Given the importance of REPs on the demand side, we also find that it also affects the supply side of the market by affecting the employment opportunities of exercise professionals and, relatedly, which gyms are most likely to employ them. This would be especially the case if we assume gym owners tend to be most interested in maximizing their profits and therefore seeking the cheapest form of human capital. The demand side therefore should affect the supply of exercise professional-related labour supply or human capital and the quality (measured by the level of qualification) supplied overtime. Related to this, we find that the demand for exercise professionals has increased over time which, given the supply side (“Gyms and Fitness Centres,” 2019), must be filled by lesser qualified individuals. But given their lower qualifications, they have contributed to the higher injury rates that have been widely reported in the New Zealand exercise industry.

We therefore find that relatively poor signals for the quality of exercise professionals not only affects the type of demand for exercise professionals, they also affect the supply side yielding a less than optimal level of quality exercise professionals in NZ related human capital formation. These are nevertheless chosen because of the ‘noisy’ decision-making environment yielding ‘bad’ or sub-optimal heuristics. This damage can only be repaired by improving and, therefore, better regulating the decision-making environment, as Akerlof argued for in his classic market for lemons paper.

References

Altman, Hannah (2020). The behavioural economics of organizational inefficiency: The example of the New Zealand fitness industry. Master of Philosophy Thesis, Queensland University of Technology, Brisbane Australia.

Akerlof, George (1970). “The market for lemons: Quality Uncertainty and the market mechanisms.” Quarterly Journal of Economics, 84: 488-500.

Gigerenzer, G. (2007). Gut Feelings: The Intelligence of the Unconscious. New York: Viking.

Stiglitz, J. E. (1975). The Theory of "Screening," Education, and the Distribution of Income. American Economic Review, 65, 283-300.

Spence, M. A. (1973). Job Market Signalling. Quarterly Journal of Economics, 87, 355-374.

Statista (2016). Significant areas of growth of the fitness industry in New Zealand in 2014. Retrieved from https://www.statista.com/statistics/689966/new-zealand-growth-areas-offitness- industry/

IBISWorld (2019). Gyms and Fitness Centres-New Zealand Market Research Report. Retrieved from https://www.ibisworld.com.au/industry-trends/nz-market-research-reports/artsrecreation- services/gyms-fitness-centres.html

Word Count 894

ABSTRACT. Although the effects of climate change are visible nowadays in the form of heavier floods, increases in rainfall and longer heatwaves, one of the key characteristics of climate change risks is that these risks are psychologically distant. The individual perception of psychological distance regarding climate change is one of the determinants of the perceived value of risk change, and this directly affects individual preferences and adaptation to climate change. To inform both policymaking and research on behavioral adaptation to climate change, we aim to compute the context premium for extreme weather event fatality risks for the United Kingdom by examining individual trade-offs between extreme weather event fatality risks (floods and heatwaves) versus traffic accident fatality risks. Drawing on relevant psychological constructs, we include validated measures of Construal Level Theory and Regulatory Focus to measure perception, behaviors and attitudes towards both general and climate-change related risks. We find that psychological distance to climate change is one of the main drivers of the risk premium when systematic heterogeneity is incorporated to our analysis. Those psychologically close to climate change weigh extreme weather event fatality risks at almost two times the rate of traffic accident fatality risks, which translates to an average VSL value of £2,63 million per fatality per year. However, direct experience with traffic accidents dampens the effect of extreme weather events and keeps the context risk premium.

ABSTRACT. We contrast the literature on additivity neglect, based on asking participants for probabilities (psychology) with studies of ambiguity aversion deriving probabilities from participant’s preferences (economics). In five experiments, participants estimated five mutually exclusive and exhaustive outcomes of a real future event, followed by a choice between betting on either one of the uncertain outcomes just estimated (ambiguous lottery) or on a risky lottery. Participant’s probability estimates displayed additivity neglect adding to 160%-223% across experiments. If subjects choose according to their stated probabilities, they will strongly prefer ambiguous lotteries to risky lotteries, with probabilities adding to 100%. Actually, participants strongly prefer risk, choosing as if their probability assessment is almost an order of magnitude smaller than what they state. Participants also showed a tendency to avoid risky lotteries with 50% winning probability.

ABSTRACT. Introduction In the present paper, we investigate the idea that mental imagery can be considered as a cognitive factor that provides input to the decision-making process. We predicted that when people are confronted with a risky dilemma, they can generate mental images visualizing both their behavior and its potential consequences. These images, among other factors, are associated with both affective responses and cognitive evaluations of expected threats and benefits initiating the approach/withdrawal motivational tendency and leading to the final choice. Our theoretical approach is based on the notion that by producing vivid mental images, a decision maker can “pre-experience” how rewarding or threatening future outcomes of their choice will be. In other words, mental imagery “provides a window on the future by enabling people to envision possibilities and develop plans for bringing those possibilities about. In moving oneself from a current situation toward an envisioned future one, the anticipation and management of emotions and the initiation and maintenance of problem-solving activities are fundamental tasks” (Taylor et al., 1998, p. 429). Therefore, we propose that when people who face a risky dilemma generate vivid and positive mental images related to it, they are more willing to choose a risky course of action. On the contrary, when facing a risky situation leads to generating vivid but negative mental images, the willingness to accept risk is reduced. The main goal of the present project was to initially test these predictions.

Method We developed a novel cognitive task for investigating how mental imagery may inform decision making in risk-related situations—the Dynamic Mental Imagery Task (DMIT). This task is aimed at testing the iterative and dynamic nature of mental imagery in decision making by online registering spontaneously generated mental images until the decision maker selects either a risky or safe alternative. To illustrate, when completing DMIT, participants are presented with a decision problem of whether to engage in an uncertain activity such as participating in experimental medical treatment or investing in the stock market. They are then instructed to visually imagine themselves being actively involved in the activity and to indicate all mental images they generated by pressing a button each time a new mental image (either positive or negative) appeared in their mind. Participants are instructed that a mental image is a picture-like representation of the whole, like in comics, to make sure they report only visual mental images, and not other types of arguments (e.g., logical arguments like probabilities). When participants are ready to make a decision, they press another key either to accept risk or to withdraw. In the present study, we tested the psychometric properties of the DMIT and examined the main hypothesis that the valence of mental imagery is related to the willingness to take risk. We predicted that when people generate more negative (more positive) mental images when considering their engagement in a given risky activity, they would be less (more) likely to make a risky choice. We also tested another version of DMIT that does not involve pressing keys when a new mental image appears in participant’s mind. In this version of our method, participants are only requested to generate mental images without reporting them and to make a choice when ready. In both between-subject conditions (i.e., with and without reporting generated mental images by pressing appropriate keys), participants (N = 120) were presented with brief descriptions of 20 risky activities. Next, participants in the condition that involved reporting mental images were asked to indicate all negative and positive mental images produced in response to the activity by pressing an appropriate key. Finally, all participants indicated whether they intended to engage in a given activity or not. After completing DMIT, participants in both conditions were asked to assess their current emotional state. We also used an old-new task to test whether there were any differences between groups in correctly remembering the presented activities.

Results In the first part of the analysis, we compared the two experimental conditions. Three separate linear regression models demonstrated that both conditions did not differ between each other in: (1) the average positive affective state (b = 0.84, p = .473), (2) the average negative affective state (b = 0.51, p = .538) nor (3) correct remembering of activities (b = 0.01, p = .269). We fitted a generalized linear mixed model with participants and activities as random intercept effects to test whether there were differences between both conditions in the likelihood to take risk. We found that participants who were instructed to report their mental images (by pressing an appropriate key) were more likely to engage in risky activities than participants in the condition without such an instruction (b = 0.53, p < .001). It is possible that the requirement to report all images made the positive images more salient and, in consequence, had a positive effect on risk-taking propensity, but such an interpretation needs further examination. The second part of the analysis was carried out to test our main hypothesis which assumed the connection between the valence of mental imagery and declared risk taking. We fitted a generalized linear mixed model with participants and activities as random intercept effects to investigate whether the number of positive and negative mental images predicted the likelihood of engaging in risky activities. We found that, in line with our prediction, generating more positive mental images was related to a higher likelihood to take risk (b = 1.52, p < .001), while generating more negative images was linked to a lower likelihood to take risk (b = -0.34, p < .001; R2 = .69). To summarize, it seems that we successfully reached two aims. First, we showed that our task—DMIT—provides novel insights into studying decisions under risk and uncertainty and can be used to investigate the relationship between mental imagery and choices under risk. Second, we initially showed that the valence of mental images predicts the willingness to take risky activities with more positive mental imagery being predictive for higher risk propensity.

ABSTRACT. If it is commonly admitted that individuals widely differ in their risk attitudes, there is little understanding about the nature and the origin of this heterogeneity. In this paper, we provide evidence that all the heterogeneity in risk attitudes is driven by heterogeneity in probability weighting. Indeed, following recent advancements of Bayesian statistics in the field of decision under risk, we show that the Rank-dependent utility model's statistical performance remains unchanged when we suppose a unique utility function among subjects but significantly decreases when we suppose a universal probability weighting function. This result would indicate a homogeneous treatment of wealth but a heterogeneous treatment of probabilities from the subjects. More specifically, most of the individual heterogeneity in risk attitudes is captured in our estimates by the elevation of the probability weighting function, which is classically interpreted as a measure of optimism or pessimism. As an alternative to the existing risk-preferences elicitation procedures, we propose a "targeted heterogeneity approach" where only some elements in the models are estimated at an individual level. Consequently, we also underline that the estimations made entirely at a subject level may mistake pure noise for heterogeneity in the distribution of some of the models' parameters. Finally, on a methodological side, our paper also innovates by using both a parametric and a new non-parametric methods in its estimation of risk preferences.

ABSTRACT. ABSTRACT Communication is well known to increase cooperation rates in social dilemma situations, but the exact mechanisms behind this have not yet been entirely understood. This paper studies the impact of communication on public good provisioning in an artefactual field experiment conducted with 216 villagers from small, rural communities in northern Namibia. In line with previous experimental findings, we observe a strong increase in cooperation when face-to-face communication is allowed before decision-making. We additionally introduce a condition in which participants cannot discuss the dilemma but talk with their group members about an unrelated topic prior to learning about the public good game. It turns out that this condition already leads to significantly higher cooperation rates, albeit not as high as in the condition where discussions about the social dilemma are possible. After ruling out better comprehension of the game and increased expectations of one’s group members’ contributions as the main drivers for the communication effect, we can bring forward the relevance of interpersonal relationships and the personal norm of keeping promises. Finally, we discuss these findings in the light of existent theories and make suggestions for future empirical research.

EXTENDED ABSTRACT One particular measure that has been observed to increase cooperation rates remarkably is allowing participants in social dilemma situations to talk to each other prior to making their decisions (Dawes 1980; Sally 1995; Balliet 2010; Ostrom 2010). Despite quite some research on the topic, exact mechanisms for this effect still remain mostly unclear (Lopez and Villamayor-Tomas 2017; Koessler et al. 2020). This study adds to solving the puzzle of identifying what determines decisions to cooperate and contributes to understanding how communication is able to raise cooperation in social dilemma situations [FOOTNOTE 1]. Disentangling single elements may help harnessing the potential of communication in solving the dilemma: Does the simple act of talking reduce social distance, generate mutual trust and affect social preferences? Does talking about the social dilemma problem increase comprehension and create norms of cooperation? Or does it require mutual commitment to cooperate from all members in order to build positive expectations about each other’s behavior? Based on previous theories and findings, we hypothesize that communication, in particular face-to-face communication, affects cooperation over various channels, some of which go beyond the mere content of the conversation. This study shortly reviews relevant literature and then presents results from an artefactual field experiment that compares three conditions: no communication as the control treatment (T1), an unrelated communication treatment (T2) and a dilemma-related communication treatment (T3). The crucial distinction between the two communication treatments lies in whether or not group members already know about the upcoming public good game when they talk to their group members. This design therefore allows us to distinguish what we call the dilemma-related elements in communication from the unrelated ones. Our experiment was conducted in northern Namibia with 216 villagers from small, rural communities which brings about two advantages: Firstly, participants do already know each other, so that a communication effect cannot only be attributed to simply identifying and getting acquainted with one’s group members. Secondly, it allows us to compare how communication interacts with different levels of previously existing social ties between participants, which has, to our knowledge, not yet been investigated before. Results show that talking about an unrelated topic is already conducive to cooperation. Discussing the actual dilemma, however, results in even higher rates of cooperation. We find that also interpersonal relationships of group members play a role, both through previously existent social ties as well as through spontaneous alterations in group members’ relations induced by communication. Interestingly, trust and expectations about one’s group members’ contributions remain unaffected by communication and increased comprehension of the dilemma after discussions cannot be identified as a driver for increased cooperation. Several possible reasons are then discussed that aim to explain these findings. While increased cooperation after talking about an unrelated topic can only be a consequence of altered intragroup relationships, we draw on the personal norm of keeping promises to explain the additional increment after dilemma-related discussions. This is partly supported by existent theories from adjacent disciplines.

[FOOTNOTE 1]: Communication, in particular in experimental settings, usually refers to unrestricted face-to-face discussions between a group of participants that face such a social dilemma (Bicchieri and Lev-On 2007). Procedural standards in economic experiments further ensure anonymity of individual decisions which allows all individuals to reveal their true preferences without having to worry about retaliation by other participants during or after the experiment. Participants consequently only get to know their own and the group outcome but are unable to discover the individual decisions of the other group members (unless all group members unambiguously defect or cooperate, in which case the other’s behavior can be deducted from the group outcome). Real money is offered in economic experiments to make preferences and decisions salient.

Funding and Acknowledgements Funded by the Southern African Science Service Center for Climate Change and Adaptive Land-Use Management (SASSCAL) through the German Federal Ministry for Education and Research [grant number 01LG1201B].

REFERENCES 1. Balliet, D. (2010). Communication and Cooperation in Social Dilemmas: A Meta-Analytic Review. Journal of Conflict Resolution, 54(1), 39–57. https://doi.org/10.1177/0022002709352443 2. Bicchieri, C., & Lev-On, A. (2007). Computer-mediated communication and cooperation in social dilemmas: An experimental analysis. Politics, Philosophy & Economics, 6(2), 139–168. https://doi.org/10.1177/1470594X07077267 3. Dawes, R M. (1980). Social Dilemmas. Annual Review of Psychology, 31(1), 169–193. https://doi.org/10.1146/annurev.ps.31.020180.001125 4. Koessler, A.-K., Ortiz Riomalo, J. F., Janke, M., & Engel, S. (2020). Structuring Communication Effectively for Environmental Cooperation (SSRN Scholarly Paper ID 3533910). Social Science Research Network. https://doi.org/10.2139/ssrn.3533910 5. Lopez, M. C., & Villamayor-Tomas, S. (2017). Understanding the black box of communication in a common-pool resource field experiment. Environmental Science & Policy, 68, 69–79. https://doi.org/10.1016/j.envsci.2016.12.002 6. Ostrom, E. (2010). Beyond Markets and States: Polycentric Governance of Complex Economic Systems. American Economic Review, 100(3), 641–672. https://doi.org/10.1257/aer.100.3.641 7. Sally, D. (1995). Conversation and Cooperation in Social Dilemmas: A Meta-Analysis of Experiments from 1958 to 1992. Rationality and Society, 7(1), 58–92. https://doi.org/10.1177/1043463195007001004

ABSTRACT. Artificial intelligence (AI) has become an essential part of our world. Intelligent machines can observe, analyze, learn from data, make decisions that mimic human behavior, and solve problems. This technology has transformed our lives as consumers and changed many aspects of production in a wide range of sectors, from finance to manufacturing, to retail. Given the rapid development of Artificial Intelligence the world is experiencing, the strategic interaction between humans and artificial agents is likely to become very relevant soon. In this paper, we study cooperation between humans and an AI in both one-shot and finitely Repeated Prisoner’s Dilemma (RPD) games in which each subject is partnered with an artificial intelligence that is trained to play as humans did in previous “all human” sessions. To better understand the motives behind humans’ decisions in this setting, in particular to rule out the lack of empathy with the machine, all AI’s are playing on behalf of real people present in the lab, who are randomly chosen among the participants to be only “passive” subjects, and who will receive the gains obtained by the AI’s. We design a rich setting controlling for beliefs, emotions and personal characteristics. In our experiment, we first form nine groups of participants, homogeneous in terms of gender composition and of the individuals’ level of altruism and cognitive ability previously measured in the laboratory. The first six groups of subjects play both one-shot Prisoner’s Dilemma and RPD games in “regular” sessions, with all human participants (the “Humans” treatment). The data from these sessions are then used as an input to train our artificial machine, which is the partner of each human participant in the following three sessions (the “AI” treatment). We then compare cooperative behavior in the two treatments. With the aim of understanding the reasons for the possibly different cooperation choices, we elicit the participants’ beliefs about their partner’s future behavior before each decision, and, after it, we ask whether this decision was mainly motivated by rationality or by some particular emotion. We observe that expectations of partner cooperation at the beginning of the one-shot and of the repeated tasks are not significantly different between the “Humans” and the “AI” treatments. Thus, ex ante, human players do not seem to fear the machine. Despite this, we observe that initial human cooperation in the RPD games is actually higher with other humans than with an AI. Cooperation continues to be higher in all periods of the RPD tasks: cooperation rates range between 50% and 80% in the baseline, while they range between 40% and 50% in the AI treatment. Moreover, decisions appear to be less emotion-driven in the AI treatment. Combining behavior and beliefs, we can identify an “outsmarting the partner” behavior, when the individual does not cooperate while expecting the partner to cooperate. We find that this behavior is more likely in the initial period of each game in the AI condition. We conclude that the strategic behavior of the human players trying to “outsmart” the machine at the beginning of each repeated game explains the persistent reduction of the cooperation levels in the AI treatment.

ABSTRACT. Much of the research on lying focuses on the lie-teller, whereas the lie-receiver remains understudied. To our knowledge, ours is the first study to investigate the way people respond to lying by providing feedback. We distinguish strategic and non-strategic responses. Previous studies cover different types of strategic ways in which people respond to in-cooperation in prisoner’s dilemma and public goods game, such as tit-for-tat (play the action that was previously played by the other), pavlov (win-stay, lose-shift), and grim trigger (punishment thereafter) (Dal Bó, P., & Fréchette,2019; Nowak & Sigmund, 1993; Sell,1997; Press and Dyson, 2012).

There a psychology literature that examines how people react to negative experiences such as betrayal or deception. For instance, betrayal evokes anger-related emotions (Leiser et al., 2013) In addition, betrayal trauma theory specifically discusses the negative impact of betrayal trauma perpetrated by someone won whom the victim depends for care and support. The victim may struggle to form intimate and secure attachment with others thereafter (Freyd, 1994; Gobin & Freyd, 2014).

Our study carries different features: First, we focus on passive receivers who do not have direct power in changing the final payoff outcome. Passive receivers’ responses will not influence senders’ financial results. However, we demonstrate in this study that even passive receivers would consider their responses as potentially influential and therefore, would play strategically in ways that could benefit themselves (the receivers) in the future. Second, we differentiate how people react to dishonest reports by a long-term partner instead of a one-time stranger, as a way to identify the strategic motivation behind their responses. Third, we study in the context of dishonest reports from the perspective of lie-receiver, which is a major problem in management where monitoring is imperfect.

Using a laboratory experiment with one-shpt and 18 repeated interactions with same parter, we provide an overarching view of how people respond to possible lies and articulate the motivations behind those responses. We differentiate non-strategic feedback responses and strategic responses. We find evidence that receivers are less likely to give negative feedback to a partner with whom they will interact in the future than to a stranger in a one-shot game. The difference is driven by strategic response to suspected misreports in expectation of future reciprocity, instead of pure emotional reactions. Strategic considerations temper emotions, so emotional responses are more forthcoming when they are costless.

ABSTRACT. The experimental research on trust has focused primarily on individual behavior. This focus has clear merits: it is simple, has immediate practical implications, and individual behavior lies at the base of more complex organizational behavior. However, we argue that in many environments the vehicles for trusting and trustworthy behavior are groups. Surprisingly, the experimental research on group-trust is scant and lacking. To that end, we developed a non-cooperative group trust game. The purpose of this study is comparing the effect of intra-group profit sharing rules (egalitarian vs. proportional) on the groups behavior.

We focused on trust games between two 3-member groups: A (trustor) and B (trustee). Each member of group A decided individually the amount to send to group B. The aggregated amount of the 3 members was tripled, and then divided equally between group B members. Then, each B member decided how much to send back to group A. In condition “egalitarian”, the amount sent back was equally divided between A’s members. In condition “proportional”, this amount was shared between A’s members in proportion to the initial group member contribution (amount sent). In condition “individual”, participants played a regular 2-person trust games. Each participant played two incentivized trust games that were separated by a filler. They received feedback regarding the outcome of both games only at the end of the session, and one game was randomly selected for payoff. Participants in the individual condition played two individual games. Participants in the group conditions played both the group-egalitarian game and the group-proportional game, and their order was counterbalanced across participants.

The results showed that mean amount sent by trustors in the group-egalitarian game ($4.27, se = .36) was lower than the amount sent in the group-proportional game ($5.54, se = .35). Interestingly, the amount sent in the individual game ($4.23, se =.54) was similar to the amount sent in the group egalitarian condition. A somewhat different picture emerged in player B’s behavior. The proportion of amount returned from the received amount was 0.32 (se = .033) in the group-egalitarian condition, 0.34 (se = .046) in the group proportional condition, and 0.31 (.037) in the individual condition, suggesting that groups were as trustworthy as individuals. Individual differences measures of trust and risk aversion did not seem to explain the group member’s behavior. The findings suggest that the profit sharing rules affect trustors behavior, but have less effect on the trustees. Implications to group cooperation and trusting behavior are discussed.

ABSTRACT. The present study implements the investment game with a novel tweak, in order to measure how behaviour is changed if participants are freed from norms regarding gender-typical behaviour. The subjects first respond to a survey, stating their gender. Next, they are randomly assigned a gender, i.e. a male can be assigned a female gender. Then, the subjects complete a standard “investment experiment,” to assess the degree of risk aversion (Charness & Gneezy 2012). The subjects complete this investment task as a financial investor on behalf of a client, assumed to observe their randomly assigned gender). The subjects who are assigned the role of a financial advisor of a different sex will no longer have to act in accordance with the gender-based social expectations about their risk attitudes. As such, the experiment reveals whether image concerns and fear of backlash from exhibiting gender-atypical behaviour play a role in gender differences in risk taking. The random assignment of gender is reminiscent of “gender swapping,” documented to be prevalent among videogame players. In gender-swapping, video game players choose an identity and avatar of a different gender (MacCallum-Stewart 2008), and thereafter, they exhibit a behaviour typical to this chosen gender. To our knowledge, no previous study has exploited “gender-swapping” in the examination of gender differences in latent characteristics, such as risk-taking. The data will be collected in January and February. The preliminary results will be ready by the time of the conference.

References Charness, G., & Gneezy, U. (2012). Strong evidence for gender differences in risk taking. Journal of Economic Behavior & Organization, 83(1), 50-58. MacCallum-Stewart, E. (2008). Real boys carry girly epics: Normalising gender bending in online games. Eludamos. Journal for Computer Game Culture, 2(1), 27-40.

ABSTRACT. Creativity represents one of the main drivers of development in any field, being it the foundation of any idea which made human civilization as it is. Creativity's central role is recognized both at a micro and at a macro scale, as it is seen as accountable for the well-being and self-determination of individuals on the former and for technological progress on the latter. The key link between what it may look like a pure psychological trait and economics lies on the fact that creativity is a key building block for innovation (Charness and Grieco, 2019). As for experimental economics, the interest in creativity is quite recent and most of the existing literature focuses on the relationship between the latter and incentives (Attanasi et al., 2021). We define GrI-creativity as the specific creative cognition process resulting in green innovation, i.e., directed toward the generation of green rather than non-green products. In this work, we developed an operational way to investigate the GrI-creativity process and its determinants through a lab-in-the-field experiment with primary school children aged from 7 to 11 years old. Under a tournament incentives scheme, subjects performed a common drawing task with constraints (the use of at least 3 triangles, 3 circles and 3 squares), but with different means: only a black marker, any color among twelve (including black), or three among the same twelve color set. Drawings were later evaluated relying on Amabile's Consensual Assessment Technique (Amabile, 1982): a jury of experts evaluated the creative outputs and the inter-scale reliability was measured through the Cronbach's alpha. Having verified that the alpha was large enough (practice in literature predicts 0.7 to be an acceptable threshold), we find weak evidence for boys to perform better when given three colors, while girls seem to be extremely sensitive to means made available. Specifically, females did significantly better when given the choice of one color, both with respect to the black marker and three colors conditions. Moreover, we combine our experimental method with insights from social psychology. We provide evidence of a highly significant positive effect on creativity of a high score at the Cognitive Reflection Test and, only for boys, the practice of sport in daily life. This latter result may also be the consequence of the fact that the incentive provided consisted in a coupon to be spent in a sports store, but it remains indeed interesting the fact that only one gender has shown such an effect. To give more robustness to our study, drawings were also catalogued depending on what they depicted. Two groups of categories were defined: the drawings' subject (landscape, humanoid or animal-like and not-defined) and if the participants fixated on the geometric figures that were provided as task constraints. We find that, other than increasing our regression's R-squared, this categorization suggests that judges evaluated as less creative drawings whose subject was not defined and those that were fixated on shapes. This work contributes to the existing literature about creativity differences between genders and finds a context where the interaction of sex and means made available determines changes in creative performances. Interestingly, a growing literature in neurosciences suggests that different cognitive strategies are put in place depending on gender in the creative domain, so that the main results of our study may well be the consequence of the latter. Caution is although recommended for this last statement, as participants' brain activity was not monitored during our task.

References Amabile, T.M., Social psychology of creativity: A consensual assessment technique. Journal of Personality and Social Psychology, vol. 43, 1982, pp. 997-1013. Attanasi, G., Chessa M., Gil Gallen, S., llerena, P., A survey on experimental elicitation of creativity in economics, Revue d'économie industrielle, vol. 174, no. 2, 2021, pp. 273-324. Charness, G., Grieco, D., Creativity and Incentives, Journal of the European Economic Association, vol. 17 no. 2, 2019, pp. 454-496.

ABSTRACT. The purpose of this paper is to investigate whether having a compatriot course setter, influences performance of professional alpine skiing racers and also to examine whether there are any gender differences in this response. We use data on all 52,485 performances among men and 47,051 performances among women from 442 men’s and 445 women’s competitions in alpine skiing’s Slalom, Giant Slalom and Super Giant disciplines that took place in the World Cups, World Championships and the Olympic Games in the seasons from 2001-02 to 2017-2018. We use racer per season fixed effects model that allows us comparing the performance of racers when they compete on the course that was set by their compatriot to performance of the same racers when they compete on course that was not set by their compatriot. We find no effect of having a compatriot setter on the probability of failing to complete the first run. However, in the second, decisive run, women have a significantly lower probability of failing to complete their run, whereas men fail significantly more when the course was set by their compatriot. We explain this result by gender differences in choking under pressure when the stakes are larger. This is in line with Cohen-Zada et al. (2017) who showed that men lose their serve, which is considered as an advantage in tennis, significantly more in most decisive games in tennis match compared to women. The contribution of our paper is that it shows gender differences in choking in non-interactive settings where a racer competes against himself without any interference compared to tennis, where it is theoretically possible that a receiver enhances performance rather than a server chokes under pressure.

ABSTRACT. Creating and managing their public image is essential for leaders. From the role congruity perspective, female executives face a particular challenge; that they are disliked when following the societal image of agentic leaders (confident and assertive) but judged as ineffective when following female communal norms (benevolent and accommodating). However, there are reasons to believe that displaying communal qualities could be beneficial by enabling female leaders to avoid backlash, heightening the perception of competence (succeeded despite the gender obstacles), and establishing congruence to the new and contemporary androgynous leader image. This study empirically examines the spoken and written texts of 204 female and 318 male executives collected from various media outlets to reveal the images top female executives convey in public through communal and agentic language styles. Our findings show that female executives employ more communal language than male executives, but this effect is moderated by the industry sector such that executives in female-dominated industrial sectors tend to have a more agentic style. Female executives have a more agentic style than the general female language, while male executives exhibit more communal language than the general male language. We also find that female executives speak more about family and motherhood, irrespective of interview and non-interview occasions. This provides an additional contextual insight that family obligations generally being seen as obstacles for upward ascension for women in the labor market does not apply to the top executive levels. The overall findings are consistent with the notion that displaying communion benefits top executives.

ABSTRACT. In the UK, Norway and elsewhere, the Christmas period reliably yields substantial increases in consumer spending and borrowing. Following qualitative interviews using an adapted mental models approach, three scales were constructed, each with six to eight Likert scale items: the meaning of Christmas; propensity to spend, and propensity to borrow at Christmas. In two surveys in the run-up to Christmas, we investigated the role of the meaning of Christmas and other individual-level factors in predicting propensity to spend and borrow (UK, N = 190; Norway, N = 180). Factor analysis identified three components of the meaning of Christmas: financial concerns, indulgence and social meaning. Measures of propensity to spend and borrow were internally reliable and were successfully validated against post-Christmas reports of actual spending and borrowing. In both pre-Christmas surveys, hierarchical regression found that seeing Christmas as a time for indulgence, materialistic disposition, negative affect, and, in the Norwegian survey, internal locus of control predicted propensity to spend. Materialistic disposition and negative affect also predicted propensity to borrow in both surveys, as did money management skill and economic hardship. Finally, in the Norwegian survey, an obligation motive for gift giving and external locus of control predicted propensity to borrow. The implications for how best to support consumers’ financial well-being at times of high pressure to spend and borrow are discussed.

ABSTRACT. This paper closely examines consumers' inflation expectations against their nominal wage expectations. We use a large monthly Australian consumer survey from April 1997 to November 2021. Results show that inflation expectations are firmly anchored while wage expectations are not anchored and post a long term decline starting around 2011 suggesting persistent expected negative real wage growth. This is in line with actual weak growth in real wages in Australia and other developed economies and coincides with a break-down in historical Phillips curve relationships and a decreased pass-through from wages to inflation. Results here suggest that consumer pessimism, not originating from labor market developments, has been contributing to soft real wage growth. This pessimism toward wage increases has been disappearing during the COVID-19 pandemic but it is too early to tell whether this shift in pessimism is permanent. If it is permanent, Philips curve estimates that rely on parameters estimated over the last ten years may not be appropriate.

ABSTRACT. Impulse buying can be defined “as a consumer's tendency to buy spontaneously, unreflectively, immediately, and kinetically” (Rook & Fisher, 1995, p. 306). Modern societies offer consumers growing opportunities to engage in impulse buying behaviors (Vohs & Faber, 2007). Moreover, thanks to current information and communication technologies, such buying impulses can emerge twenty-four hours a day and can be met at any time, considering how easy it has become to purchase products (Vohs & Faber, 2007). While impulse buying is not always well regarded by society, companies see it as something positive, because it provides an additional source of income (Bossuyt, Vermeir, Slabbinck, De Bock, & Van Kenhove, 2017). Therefore, increasing the knowledge of the drivers of impulse buying is important for companies for better developing their strategies and meeting consumers’ urges. Furthermore, this knowledge can also be of use to inform consumers about their impulse buying behaviors. Unsurprisingly, extant research has examined the antecedents of impulse buying, having identified numerous personal factors, including consumer demographics, such as gender and age (Hultén & Vanyushyn, 2014; Wood, 1998), and psychological factors, such as self-esteem (Dhandra, 2020) and hedonic needs (Hausman, 2000; Iyer et al., 2020). Moreover, research has also uncovered a set of contextual variables such as the store environment (Sherman, Mathur, & Smith, 1997) and promotions (Hultén & Vanyushyn, 2014). This study contributes to this research stream by examining how personal values, namely resultant conservation and resultant self-enhancement, relate to impulse buying. Personal values are general goals that guide individuals’ assessments and choices (Schwartz, 1992), mirroring what is important for individuals in life (Huynh & Olsen, 2015). In particular, consumers pursue desired end states through their purchases (Babin, James, Camp, Jones, & Parker, 2019). The personal values of consumers have been related to diverse behaviors, such as the purchase of consumer durables (Corfman, Lehmann, & Narayanan, 1991), and the engagement in online brand communities (Marbach, Lages, Nunan, & Ekinci, 2019). Hence, past studies indicate that personal values play a pervasive role in guiding consumer behavior (Zasuwa, 2016), and this suggests that they might be related to impulse buying. Given the importance for companies, consumers, and the society at large of understanding the impulse buying phenomenon, this study is likely to yield relevant insights to multiple stakeholders by explaining it from a personal values’ perspective. In addition, considering the goal-setting theory and the hierarchical model for the influence of stable individual characteristics on the behavior of individuals (Mowen & Spears, 1999), this study considers the mediating role of hedonic shopping motivations. Specifically, personal values are regarded as motivational goals (Parks-Leduc, Feldman, & Bardi, 2015) that, being at a higher level, will exert their effects on impulse buying through lower order goals mirrored in hedonic shopping motivations. Hedonic consumption is related to “those facets of consumer behavior that relate to the multisensory, fantasy and emotive aspects of product usage experience” (Hirschman & Holbrook, 1982, p. 92). This type of buying motivation is a reflection of consumer demand for fun, fantasy and sensory stimulation (Babin, Darden, & Griffin, 1994). Research has identified six types of hedonic shopping motivations: adventure, value, role, idea, social and gratification shopping motivations (Arnold & Reynolds, 2003). The adventure shopping motivation concerns the extent to which individuals seek adventure, sheer excitement and a desire to experience a different universe of stimuli while shopping. As to value shopping motivation, it has consumers shop for sales, discounts, low prices and bargains. The role hedonic shopping motivation reflects the pleasure consumers draw from shopping for others. Regarding idea shopping, it is a motivation referring to the search for new trends, fashions, products and innovations. Social shopping concerns the desire for “shopping with friends and family, socializing while shopping, and bonding with others while shopping” (Arnold & Reynolds, 2003, p. 80). As for gratification shopping, it concerns consumers that shop to lower stress levels. These individuals find in shopping a way to relieve a bad mood, forget problems and obtain personal gratification (Arnold & Reynolds, 2003). Research suggests that hedonic purchases are related with emotional gratification, or with the minimization of negative feelings (Iyer et al., 2020). Not surprisingly, hedonic shopping is a key element of unplanned purchases (Parsad, Prashar, Vijay, & Kumar, 2021). However, few studies have addressed the multifaceted nature of hedonic shopping motivations. In this context, this study makes two major contributions to existing knowledge. First, this study innovates by looking into how personal values are related to impulse buying. Second, it also innovates by proposing that hedonic shopping motivations mediate the relationship between human values and impulse buying. To test the research hypotheses, we began by relying on a preliminary sample, collected through a snowball procedure relying on the distribution of a link to the questionnaire through social networks, such as Facebook, and by email. This yielded 510 usable responses. Following the encouraging results obtained in this sample, we collected a second sample through MTURK, with 332 respondents. The measures were obtained from previous studies and relied on a five-point Likert type scale. Confirmatory factor analysis support the measurement properties of the measures used. Given that the data is from a single source, we conducted a number of statistical assessments that indicated that common method variance should not be a relevant issue. The results of the estimations in the two samples are remarkably similar. Specifically, this study provides initial evidence that personal values are related to impulse buying, thus enriching the understanding of the latter phenomenon, which is important for companies and for society as a whole. Second, another implication for knowledge is that hedonic shopping motivations appear to fully mediate the relationship between personal values and impulse buying. Relatedly, the study indicates that personal values appear to shape hedonic shopping motivations, which have been related with an array of important outcomes, such as compulsive consumption (Arnold & Reynolds, 2003).

ABSTRACT. Introduction

The Covid-19 pandemic and rising energy prices in Great Britain have placed additional pressures on the finances of many households, raising concerns of a significant increase in the number of households struggling with their energy bills. Understanding and establishing what constitutes good practice for how energy suppliers communicate with their indebted customers is therefore increasingly important. To address this, GB’s energy regulator, Ofgem (the Office of Gas and Energy Markets), published their ‘Consumer Protection Report: Autumn 2021’ in October 2021, which set out criteria and recommendations that energy suppliers are expected to meet in their communications with indebted customers.

This presentation describes the methods and findings of a two-phased research project designed to inform these recommendations. It was undertaken in summer 2021 by Ofgem’s in-house Customer and Behavioural Insights team. The research aimed to understand how the emphasis, tone and content of energy debt communications can influence consumers’ intentions to engage in response to the communication, their comprehension of key information, and their attitudes towards receiving the communication. Its purpose was to provide empirical evidence to inform what constitutes ‘good’ practice for energy supplier debt communications.

The project combined qualitative interviews of indebted energy consumers with a nationally representative online behavioural experiment for over 1,500 GB domestic energy customers. This approach reflects the increasing application of behavioural science and controlled experimental methods in informing policy and regulatory decision-making in Great Britain and internationally.

Methods

30 qualitative interviews initially probed indebted consumers’ expectations, attitudes, and experiences of different energy debt communications. Three template debt communications, based on existing supplier communications and varying in content, tone, and emphasis, were shared with participants. These versions were iterated in response to participant feedback across three weekly ‘waves’ of interviews. A number of key attributes emerged – tone of language, the primary emphasis of the communication (e.g. whether focusing on the importance of immediate repayment or on available support/ solutions), and acknowledgment of circumstances outside of the customers control that could be a factor in their inability to pay (e.g. the Covid-19 pandemic).

An online behavioural experiment was then designed to quantify participants’ intentions to engage, their comprehension of key information and their attitudes towards receiving different hypothetical debt communications. Participants were asked to imagine they were falling behind on their energy bills and received a letter from their energy supplier. They were then shown one of four randomly assigned hypothetical debt communications.

A control version (V0) was styled on real existing supplier communications, and featured attributes that participants responded most negatively to in the qualitative research. Its primary emphasis was on the importance on immediate repayment and consequences of non-payment, it was written in a tone that was perceived to be harsh, and it did not acknowledge possible circumstances outside of the customer’s control.

Three alternative versions were sequentially adapted to incorporate attributes that were responded to positively in the qualitative research. V1, while still harsh in tone, emphasised available support/ solutions. V2 adopted a friendlier tone in addition to emphasising available support/ solutions, including acknowledging the potential for general circumstances to impact ability to pay. V3 was friendlier in tone still, and made specific reference to the possible impact of Covid-19.

Results

Self-reported intentions to engage were extremely high. 95% of respondents said they would contact either their supplier or independent support/ advice if they received the communication. Those who saw V0 were no more or less likely than those who saw any other version to say that they would contact someone.

Given a-priori concerns that self-reporting bias may contribute to a heightened positive response in the first task, a second task asked participants to read vignettes of hypothetical indebted consumers, and judge how that person would respond to the same communication. Four vignettes varied across two dimensions – whether the person was a first-time debtor or not, and whether or not Covid-19 had impacted their household finances. Overall, those who saw V1 or V2 were significantly more likely to think that the persons in the vignettes would contact their supplier after reading the letter, compared to those who saw V0.

Tests of participants’ objective comprehension of available support/ solutions showed that participants who saw V2 or V3 had significantly higher comprehension scores compared to those who saw V0. Further analysis suggests that this may have been driven at least in part by a reluctance to read the communication in full among those who saw V0 . Participants’ self-reported comprehension of the information did not vary by communication version. Importantly, this highlights the potential limitations of relying solely on self-reported understanding in the testing of communications, and the importance of gathering objective empirical evidence.

Later tasks also probed participants’ emotional response and a range of attitudes towards the supplier after receiving such a communication. Overall, participants who saw V2 or V3 reported significantly more positive attitudes towards the supplier. They also reported these communications to be less harsh and more friendly in tone, as intended, as well as less threatening.

Conclusions and implications

Collectively, the findings from this research project provide evidence in favour of communicating debt information in a compassionate tone, and that providing information about available support/ solutions should be the primary focus of initial debt communications from suppliers. Across all tasks, there was no evidence to support the use of communications that are harsh in tone and focus on immediate payment, as adopted by some suppliers. These findings provided empirical evidence to support the development of ‘good practice’ recommendations for suppliers communicating to indebted customers.

More broadly, the research project demonstrated the wider value of experimental pre-testing in principles-based regulation, as well as the importance of the triangulation of multiple research methods in informing regulatory decision-making. In addition, it demonstrated the value, opportunities and challenges associated with bringing the application of behavioural science expertise and experimental methods in-house within a national regulator.

ABSTRACT. Past instances of financial crises have emphasized the need to develop institutional devices to prevent bank runs. We set up an experiment to study the effectiveness of two of such devices: Stability funds that automatically limit depositors' possibility of withdrawing their assets, and individual insurance against the risk of default. We also vary the coverage that either mechanism permits. Depositors start off with all their monetary endowment deposited in a bank. They then have three rounds in which they can withdraw money, before and after the bank suffers a liquidity shock that can be either permanent or temporary. The bank defaults if the desired withdrawals exceed its available liquidity. We find that the only mechanism that is effective in reducing bank defaults compared to the baseline is the stability fund with high coverage. The insurance mechanism paradoxically generates more bank runs than the baseline. Large deposit withdrawals take place mainly, but not exclusively, in period one. Bank runs are less likely in groups with a higher share of female depositors.

ABSTRACT. Statistically inaccurate beliefs are the primary drivers of gender discrimination and even updating – in light of new information – might be biased by stereotype-based expectations. We use a lab experiment, in which participants can decide to take over an intelligence task, to investigate both beliefs about others’ performance and actual decisions. Such take over decisions are widespread in organizational contexts where stereotypes hindering the most qualified person to take over can lead to inefficient task allocations. Additionally, we test whether implicit discrimination drives a wedge between beliefs and actions. Consider the following example: An employer is asked to evaluate one of his female employees. Driven by the injunctive norm that women perform just as good as men, he complies with the norm and gives a positive evaluation of her performance. The next time she is working on a task the employer has now the opportunity to take over from her. Even though the employer observes the exact same performance as before, this time he decides to take over. The upcoming days the same story occurs with a male employee and the exact same performance, except, the employer doesn’t take over. That is, although reported preferences claimed no gender discrimination, the employer discriminated through his actions. Whether people have stereotypical beliefs and whether discrimination is more prevalent through actions than through reported preferences are the focus of this study.

ABSTRACT. Even though we are aware of the serious consequences of unhealthy eating habits, most of us find it challenging to stick to our dietary goals. Public policy interventions such as setting vegetarian food as the default in canteens or taxing sugary drinks are designed to promote healthier choices. However, whether a certain policy succeeds in changing food choices of people who struggle to meet their dietary goals depends heavily on whether they endorse the policy. This project sheds light on people's attitudes towards public policy making by investigating whether and when dietary goal failures accept public policy interventions promoting a healthier diet. We hypothesize that the extent to which people fail to achieve their dietary goals is related to their willingness to accept health policies. We further believe this link between dietary goal failure and policy acceptance to depend on (1) whether people rather blame themselves or others (such as the government) for their dietary goal failure and (2) the extent to which people perceive the policy as helpful in achieving its goal of promoting a healthy diet. To test our hypotheses, we analyzed experience sampling data (n = 409 with 6,447 observations) on food choices and policy attitudes collected and provided by Bauer and colleagues (in prep). Via a mobile app participants regularly reported on all their significant food consumptions and evaluated the extent to which their meals aligned with their dietary goals. The researchers asked participants to provide at least seven entries and reports ranged over a period between six to eleven days. In addition, in a pre- and post-questionnaire participants estimated ten popular public policies regarding their acceptance and effectiveness and answered questions measuring blame attribution, self-control and demographics, among others. Conducting linear regression analyses, we found that dietary goal failures accept public policies less compared to those succeeding with their dietary goals. Blame attribution and perceived effectiveness across all policies did not impact this relationship. For subsidies on healthy food, lower perceived policy effectiveness appeared to strengthen the negative link between dietary goal success and policy acceptance. Our findings suggest that those who have difficulties in achieving their dietary goals do not necessarily have self-serving political attitudes but, on the contrary, are more inclined to reject supporting policies. Understanding policy attitudes thus appears to be more complex than current policy literature suggests. This project further contributes to clarifying significant drivers and barriers of policy acceptance among dietary goal failures. Implications are drawn for public policy making.

ABSTRACT. The motivation for the paper is the important role attributed to Tacit Knowledge (Polanyi, 1958) as critical for lack of replicability in experimental research in natural science (Collins, 1974, 1975), and Tacit Knowledge has more recently also been explicitly invoked for the same purpose and the same reason in social psychology experiments (Schmidt, 2009): it may be necessary to learn how to conduct an experiment in presence of an experienced successful experimenter through (tacit) transmission of knowledge in order to obtain the same results. With human beings as experimental subjects, we argue that Tacit Knowledge has a potentially much stronger role to play for lack of replicability in social science experiments than in hard sciences through not only the experimenter-experimenter relationships, but also the experimenter-subject (e.g., the experimenter demand effect) and subject-subject relationships (Festré & Østbye, 2021). There is a relevant literature on the experimenter-subject relationship without reference to Tacit Knowledge (see, e.g. Bardsley, 2005; Zizzo, 2010). There is also a literature more specifically related to framing, emphasising the implicit information conveyed by language and communication (see, e.g. Bruine de Bruin, 2011; Keren, 2011b) and calling for a conversational approach following Grice (1975) in order to examine the interaction between speakers (experimenters) and listeners (subjects). This is also a theme for Polanyi, discussing Tacit Knowledge in the context of sense-giving and sense-reading. On basis of this, we discuss the potential for teasing out the extent of Tacit Knowledge through systematically varying the content of explicit information in the experimental instructions. It is well known in the psychological and the economic experimental literature (Keren, 2011a; Tversky & Kahneman, 1981) that choices are highly sensitive to frames, primes or levels (normative vs. descriptive) of information (Sher & McKenzie, 2011). With minimal context, it is plausible that variance tends to increase and that subjects make their own frames to make sense of the task (Engel & Rand, 2014; Eriksson & Strimling, 2014). Many experiments in economics and psychology are based on game theoretic models. It would therefore be tempting to look at games straight away, but maybe more wise to consider simpler situations first. We therefore suggest a stepwise approach with increased complexity as a second dimension of the research strategy in addition to varying the content of explicit information in the instructions already mentioned. Starting with, say, a raw decision task and what this implies in terms of controls if we are to avoid implicit framing (or Tacit Knowledge) to be a confound, it may be enough to elicit beliefs of subjects about the intentions of the experimenters or to place the subjects in the shoes of the experimenters as suggested by Keren (2011b). For a more complex situation, say, a task with social concern (Negotiating Game), additional elicitation of beliefs about others might be necessary. For a full-fledged strategic game, elicitation of beliefs about what kind of game is being played may come on top of everything else.

References Bardsley, N. (2005). Experimental economics and the artificiality of alteration. Journal of Economic Methodology, 12 (2), 239-251. Bruine de Bruin, W. (2011). Framing effects in surveys: How respondents make sense of the questions we ask. In G. Keren (Ed.), Perspectives on framing. Psychology Press. Collins, H. M. (1974). The tea set: Tacit knowledge and scientific networks. Science Studies, 4 (2), 165–185. Collins, H. M. (1975). The seven sexes: A study in the sociology of a phenomenon, or the replication of experiments in physics. Sociology, 9 (2), 205-224. Engel, C., & Rand, D. G. (2014). What does "clean" really mean? The implicit framing of decontextualized experiments. Economics Letters, 122 (3), 386-389. Eriksson, K., & Strimling, P. (2014). Spontaneous associations and label framing have similar effects in the public goods game. Judgment and Decision Making, 9 (5), 360-372. Festré, A., & Østbye, S. (2021). Faith in science: What can we learn from Michael Polanyi (GREDEG Working Papers No. 2021-40). Groupe de REcherche en Droit, Economie, Gestion (GREDEG CNRS), Université Côte d’Azur, France. Grice, P. H. (1975). Logic and conversation. In P. Cole & J. Morgan (Eds.), Syntax and semantics, vol.3 (p. 41-58). Academic Press. Keren, G. (2011a). On the definition and possible underpinnings of framing effects: A brief review and a critical evaluation. In G. Keren (Ed.), Perspectives on framing. Psychology Press. Keren, G. (2011b). Perspectives on framing. Psychology Press. Polanyi, M. (1958). Personal knowledge: Towards a post-critical philosophy. Chicago: Chicago University Press. Sher, S., & McKenzie, C. R. (2011). Levels of information: A framing hierarchy. In G. Keren (Ed.), Perspectives on framing. Psychology Press. Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211 (4481), 453-458. Zizzo, D. J. (2010). Experimenter demand effects in economic experiments. Experimental Economics, 13 (1), 75–98.

ABSTRACT. Objectives and motivation: Data obtained using bipolar psychometric scales fulfill the definition of compositional data. The order of magnitude of agreement towards an item assertion implies complementary information about the order of magnitude of disagreement towards the item assertion. Thus, the information obtained using bipolar scales are bivariate. The analysis of compositional data is not straight forward because of the underlying Aitchison metric. Ignoring the complementary information induces serious bias to statistical analysis. The question arises, whether or not, considering the compositional data structure significantly increases the statistical power of standard statistical procedures.

Methods and results: We proposed to apply the isometric log-ratio (ilr) transformation because it transforms compositional data towards the interval scale where standard statistical procedures can be applied. Different data generating processes (e.g., Normal distribution, Laplace distribution and Cauchy distribution) are used to simulate random data. The well-known two samples Welch t-test (independent data) and the two samples t-test (paired data) are applied to both, untransformed and ilr-transformed simulated bipolar data. Comparing the statistical powers we found that the ilr approach should be preferred. Depending on the order of magnitude of different parameters (e.g., Cohen’s d, dispersion) a mean increase of the statistical power of approximately 5% can be achieved using the ilr approach. That is, less effects are overseen and sample size can be reduced while retaining statistical power.

ABSTRACT. Taxes are inevitable but people pay them with a different mindset. The concept of five tax-related motivational postures (Commitment, Capitulation, Resistance, Disengagement and Game Playing) developed by Braithwaite (2002) showed that through the adopted posture people communicate the quality of a person’s relationship with tax administration. Therefore, motivational postures are external manifestations of a person’s degree of approval and respect for authority, and determining the manner in which a person presents themselves in specific social contexts. One of the assumptions adopted in the concept of Braithwaite (2002; 2011) relates to the belief that the source of tax motivational postures is the different degree of threat to personal freedom perceived by an individual. Three areas of freedom that may be threatened by the actions of tax authorities refer to three dimensions of the self: moral self, democratic collective self and status seeking self. The moral self supports the attitudes of Commitment and Capitulation, in which the individual treats the diligent payment of taxes as an expression of civic commitment, dictated by morally right actions for the common good. The study attempted to empirically verify the above assumptions on a sample of Polish taxpayers. In particular, the relationship between the following variables was analyzed: 1) self-importance of moral identity - in which an individual thinks of himself as having certain moral features (e.g. honest, fair), but at the same time relates the above moral characteristics to the social space (e.g. taking pro-social activities, charity); 2) civic engagement, which is apolitical, shaped by the norms and values of the community and expressed in the motivation to take action for the benefit of the community; 3) need for control, which in the case of taxes is related to the third area of freedom, and refers to the inability to realize individual aspirations and plans due to the need to comply with tax law regulations which means a reduction in individual resources. The study was conducted using the questionnaire method on a sample of Polish taxpayers. The sample consisted of N = 388 (162 women), M (age) = 46.09. Results of the analysis will be presented on the poster.

ABSTRACT. Introduction Most goods and services are traded in markets. However, consumers also enjoy goods and services for which no market exists, particularly environmental goods and services. In the absence of observable market behavior, the values of non-market environmental goods and services such as clean air, recreational opportunities, and regulating services (e.g., water purification and carbon sequestration) must be elicited directly from consumer preferences. To this end, stated preference (SP) research use surveys to elicit willingness to pay (WTP). Given the high stakes and public spending of tax payers’ money to restore or preserve environmental goods, a deeper understanding of the WTP elicitation process has been of intense interest to SP-researchers ever since the 1989 Exxon-Valdez oil spill in Alaska, after which the contingent valuation method (CVM) was used to litigate a $4.9 billion damage claim for lost environmental values [1]. Specifically, issues of sensitivity in elicited WTP to the scope of environmental goods being valued has been a point of controversy in the SP literature for decades [2]. Proponents of CVM point to numerous studies that yield increasing WTP as environmental benefits increase [3]. Opponents of CVM, of which many are psychologists, likewise point to studies demonstrating its inability to yield such sensitivity in WTP to scope [4–7].

Psychological insights have made their way into many areas of economics, although to a smaller degree in environmental and resource economics [8]. The current research will assume a psychological viewpoint by investigating the anchoring effect of a typical CVM payment instrument on scope sensitivity. The payment instrument may take the form of a horizontal scale with multiple pre-defined amounts, among which participants are to select their maximum WTP. In effect the amounts form intervals. The main hypothesis to be tested is that, while holding scale range constant, increasing the interval length of the first half of the scale will increase sensitivity to scope in terms of WTP.

Methods In a CVM survey for prevention of marine oil spills in Norway (N = 2500) we will use a split-sample (between-subjects) design to randomly impose a customized payment scale. Participants (n = 1250) in the experimental condition will be exposed to twice the interval length on the first half of the scale compared to the interval length encountered by participants in the control condition (n = 1250).

Welch's t-test will be used to test for differences in WTP scope sensitivity across four scopes of environmental changes between conditions. As the appearance of the payment scale differs across response platforms, we will further test whether response platform (smartphone vs. tablet vs. PC) moderates the presumed effect of condition on scope sensitivity through a 2 (condition: control vs. experimental) × 3 (platform: smartphone vs. tablet vs. PC) between-subjects ANOVA. In addition, we will test a hypothesis of anchoring effects on mean WTP, namely that mean WTP is larger among participants in the experimental (vs. control) condition, again by means of Welch’s t-test and two-way ANOVA.

For discussion purposes regarding scope sensitivity, we will estimate an OLS-regression model of scope sensitivity in the pooled sample (N = 2500) as an exploratory analysis. The model will include two valid and reliable psychometric scales for environmental concern and social desirability bias, respectively [9,10], as well as a self-reported measure on preference uncertainty. Interactions between the experimental condition and these three variables will be included, as well as controls for income, gender, and response platform. Our analyses will be pre-registered prior to data collection during the spring of 2022.

Expected Results The pre-registered hypothesis is that, while holding scale range constant, increasing the interval length of the first half of the scale for participants in the experimental condition will increase their sensitivity to scope in terms of WTP. Numerous examples exist of how mean WTP is anchored on information provided in CVM surveys in health economics [11–15], while examples from environmental economics are scarce, mixed, and typically with small samples [16,17]. We expect anchoring effects to apply in the environmental context as well. We also expect such effects to apply for consecutive (within-subjects) elicitations of WTP over multiple changes of environmental states (i.e., scope sensitivity).

Discussion and Contribution CVM data (WTP) is typically used for management of public goods. For such management to be optimal, and hence sustainable, elicited WTP must reflects people’s true preferences. As there is tension between the fields of psychology and economics regarding the view on preference construction [18–20], what constitutes true preferences remains unsettled. As WTP is normally elicited per household (or person) in CVM surveys, a mean value must be aggregated for a total population affected by the environmental changes to estimate the effect on total welfare. Even small changes in the estimate of mean WTP (e.g., one standard error) might result in huge changes in welfare estimates, depending on the size of the affected population. It is therefore of vital importance to public goods managers to enhance the integrity of the data informing decisions, by gaining a deeper understanding of the WTP elicitation process.

While there are numerous indications of anchoring effects on mean WTP in CVM, no study has yet investigated anchoring effects on differences in WTP over consecutive environmental changes (i.e., WTP scope sensitivity), as far as can be ascertained. This is important because CVM studies designed for direct use in management of specific public goods need to span a certain scope of environmental changes to produce WTP data for management schemes of different types and sizes in order to be of practical value to decision makers. The central (expected) contribution of this research will be knowledge on the extent to which researchers’ choice of pre-defined amounts on a CVM payment scale affects the sensitivity of WTP to changes in environmental states. This research will extend the theoretical understanding of preference construction and should affect survey design for all future CVM studies intended for direct use in public goods management.

References are available upon request.