The struggle of Cambodian farmers

Together with Andries Richter (Wageningen University) and Tum Nhim (UNDP Cambodia) we conducted an experiment in Cambodia to analyze what kind of institutional design would be appropriate to handle these social dilemma situations. Why Cambodia you ask? The problems of Cambodian farmers can be seen as exemplary for many regions in the Global South. Traditionally, Cambodian farmers rely on the rainy season to replenish lakes and rivers to provide sufficient water for water-intensive crops such as rice. Yet, climate change has disrupted the rainfall pattern: the rainy seasons became unreliable in terms of timing, duration, and amount. As a result, water became a commodity whose degree of scarcity varies somewhat unpredictably over time.

With less water available the risk of severe water scarcity increases for all farmers. The degree of individual water scarcity depends not only on rainfall patterns and the quantity available but also on the geographical proximity towards water sources such as rivers and lakes. As a result, one often finds that people nearer to the water sources try to ensure sufficient water for their own fields, thereby leaving others further away without water.

While water in Cambodia was traditionally managed through informal institutions such as social norms, these are not sufficient anymore. Informal institutions are deeply embedded in the social-ecological system they help to govern. These long-standing rules are challenged by the disruptive effect of climate change impacts. Given enough time, communities might be able to establish new informal rules but given the precarity of the livelihood of rice farmers, this might be too slow. Hence, to increase security in water access, some communities/villages have started to use more formal institutions called “Farmer Water User Communities (FWUCs)”. The arrangement is that the members pay an irrigation service fee and in turn this organization is responsible for keeping the irrigation infrastructure in good quality, ensuring provision of irrigation to their members, and enforcing the written rules. These FWUCs vary greatly in their effectiveness depending on the exact institutional design and the community. Thus, it is important to test what design works best for each community and not assume a one-size-fits-all solution.

Experimental Design

We designed experiments that mimic the basic setting that these Cambodian farmers are exposed to. There is a resource that is shared by a community that needs investments to ensure that the resource can be used (such as maintenance of irrigation infrastructure). The investments towards this public good will provide equal benefits to everyone in the community independent of how much each person contributed. To manage the investment, they can vote which rules apply (the institutional design). We have two settings. In the first one they choose between a voluntary contribution system (everybody contributes as they see fit) or a tax system (everyone is forced to contribute a minimum). The voluntary contribution system is free of charge whilst the tax system comes with a service fee. The second institutional choice is between a system that states a minimum contribution level and punishes people that undercut the amount, but the enforcement system is ineffective (only about half of the offenders get caught), or they can opt for a system that has an effective enforcement (every offender gets caught) but again, this institutional design can only be maintained with a fee.

Results

We find vast support for the costly institutions with about 62% choosing the institution that comes with a fee over the free one. Preferences between costly institutions differ but only 23% choose both institutions that were free of charge. The strongest support for functioning fee-based institutions comes from people that experienced water scarcity in the past. The people least likely to vote for a tax system are farmers who own land very close to the water source, indicating that they feel confident to access enough water independent of the institutional design.

Comparing the two fee-based institutions we find that the tax system provides a higher welfare than the effective enforcement system which is mainly driven by the higher contributions towards the public good in the tax system. The tax system provides the highest welfare of all four institutional settings even though the mandatory fee means that the voluntary contribution system as well as the ineffective enforcement can reach welfare levels that are not possible in the tax system while the costly punishment system produces the lowest welfare of all four systems. The high welfare in the tax system stems from the strong crowding-in effect (figure 1) which is probably driven by people’s desire to make an active positive contribution on top of the mandatory tax contribution.

Overall, we see that people are willing to pay for functioning formal institutions that are able to punish offenders reliably since this reduces individual uncertainty in water access. We found that Cambodian farmers value institutions that ensure equal benefits and equal contributions. When designing formal institutions, it is paramount to consider local norms and cultural aspects to gain acceptance. Cambodian society is very community oriented with strong preferences for equality and fairness. Thus, any formal institution that appeals to these values whilst ensuring a successful natural resource management will more likely be accepted.

References

The working paper can be found here: http://dx.doi.org/10.2139/ssrn.4252187

Hardin, G. (1968). The tragedy of the commons: the population problem has no technical solution; it requires a fundamental extension in morality. science, 162(3859), 1243-1248.

Dr. Esther Schuch 

Esther Schuch is a Senior Researcher at Behavia and an Affiliate at the Research Institute for Sustainability (RIFS) – Helmholtz Centre Potsdam. She holds a PhD in Environmental and Behavioural Economics and has conducted experimental and survey-based research in various countries. Esther has published policy briefs and academic articles addressing topics such as cognitive biases, decision-making, and social norms.

Thank you all for making this year so incredible.

As Artificial Intelligence (AI) applications grow, we face a variety of unprecedented ethical dilemmas and uncertainties over their adoption. One of the most pressing challenges for policymakers, companies, and consumers is the issue of liability.

Together with the European Commission, Kantar Public, and CEPS, we investigated the attitudes of Europeans towards AI in terms of trust, social acceptance, and willingness to take up AI applications. To generate insights about the potential impact of different liability regimes on consumer behavior, we tested three regulatory alternatives in a hypothetical scenario like the lawnmower thought experiment above and observed how they affect their willingness to pay for AI products.

Turns out that Europeans, while generally open to AI applications, are skeptical about adopting them for concerns such as privacy, transparency, and the potential damage these products can cause. In the experiment, consumers generally behaved rationally and demonstrated that they are aware of the serious ambiguities in determining who is at fault in accidents where AI applications are involved. As such, their willingness to buy AI products is in tight harmony with economic self-interest. If the burden of proof lies on the user when damage is caused to self, the willingness to pay decreases. In the scenario of damage being caused to other parties, the more liable the users are, the less they are willing to pay.

The findings of our study informed a proposal for an AI liability Directive by the European Commission which can be found here: https://ec.europa.eu/info/business-economy-euro/doing-business-eu/contract-rules/digital-contracts/liability-rules-artificial-intelligence_en

Link to the full study: https://op.europa.eu/en/publication-detail/-/publication/0f738be1-5911-11ed-92ed-01aa75ed71a1

In this post, I discuss how economists and psychologists view self-control, how they measure it for research, and I summarize the main findings on the impact of self-control and my related experimental study.

How self-control is viewed across research disciplines

Self-control is a widely debated term in psychology, generally described as a mental ability, but also a cognitive process, affected by genetics and childhood experiences. In economics, the more common concepts are the related notions of present bias and inconsistent time preferences. These refer to the preference of a smaller reward now over a larger reward later, while when being asked about future trade-offs, a preference to wait for a larger reward. There are however also economists who investigate how self-control predicts life outcomes. They understand self-control as the ability to override undesired behavioral tendencies.

A standard procedure to measure self-control is using the Brief Self-Control Scale (SCS) developed by Tangney et al. (2004). Survey respondents are asked to read 13 statements that describe behavioral mechanisms related to self-control, for example “I am good at resisting temptations,” and rate how much these statements generally apply to them. The responses are measured on a Likert scale from 1 to 5, where 1 indicates Completely disagree and 5 indicates Completely agree. Responses typically follow a normal distribution: many respondents choose 3, identifying with the statement to some extent, and fewer respondents agree or disagree completely with them.

An alternative to the SCS is using proxy variables relying on people’s real-life behavior that can be associated with low self-control, for example, being a smoker. Both the SCS and proxies are valid options to approximate the relationship between a person’s self-control ability and other variables.

Powerful predictor of all things good 

There seem to be only perks of higher self-control. Studies link the trait to many desirable life outcomes, such as higher educational outcomes, success in the labor market, better health, greater financial well-being, stronger relationships, interpersonal skills, and overall life satisfaction (see, for example, Cobb-Clark et al., 2019). Self-control is also positively correlated with well-being on a within-person level. This means that respondents who report higher self-control than they did in an earlier survey on average also report increases in their well-being. This raises the interesting question of how stable a person’s self-control is. The answer is likely that there is certainly correlation over time, but some variation within people remains.

Can we manipulate self-control?

In psychology, there is a controversial discussion on why these within-person fluctuations in self-control form and how researchers could manipulate this process. Since the early 2000’s, some psychologists suggest that self-control is a state and finite resource, in addition to being a personality trait. When self-control resources are strained within a person, the energy available to the self is low and the capacity to control the mind over habituated responses is impaired. This is thought to be the result of so-called ego depletion. As a consequence, the ability to exert further self-control decreases in the short term. To make this theory tangible, imagine if you resisted unhealthy food all day, but in the evening do not manage anymore to say “No” to some sweets.

The ego depletion theory has been the subject of countless controlled experiments around the world. So-called ego depletion methods rely on strenuous unrewarding exercises, demanding constant suppression of a habituated behavior. They cause mental fatigue and lead the participants to a state of depletion, assumingly reducing their ability to exert self-control in subsequent tasks. At the same time, control participants perform exercises which are not considered depleting. A classic ego depletion method is the ‘letter-crossing task’ where, first, a habit is instilled among participants who must cross out a certain letter from a given text for a short while. Then, only the treated participants are asked to follow a more complex rule: crossing out the letter if a vowel does not follow or precede the letter in a certain way. This experimental design aims to cause depletion among the treated by asking them to override their habit. This procedure is assumed to allow researchers to measure the causal impact of lowered self-control on decision-making observed in the experiment.

Experimental evidence on ego depletion and impulsive shopping

Relatedly, our experimental study investigates the impact of self-control on impulsive debt-taking for consumption. From a survey representative for households in Germany, we first showed that self-control correlated with a lower probability to get into repayment difficulties with consumer debt. Motivated by this observation, we aimed to causally investigate the relationship between low self-control and impulsive shopping on credit. We implemented a lab experiment and used the ‘letter-crossing task’ to produce exogenous variation in the self-control ability between the otherwise comparable treatment and control groups.

We found that the ego depletion treatment reduced the participant’s ability to concentrate significantly. When constructing an index from several manipulation checks, we showed that the treated were indeed, on average, significantly more depleted. Our main result was that the treated were about 7 percentage points more likely to shop on credit and spent, on average, 53 percent more money in our experimental setting compared to the control group. While these differences were in the expected direction and suggested that ego depletion increased the probability to impulsively borrow for consumption, they were not statistically significant on the common levels. However, we identified a sub-group of participants who reacted strongly to the treatment. Treated participants with low financial literacy were 21 percentage points more likely to shop on expensive credit compared to the treated with high financial literacy. This suggested that self-control problems could be compensated for with financial literacy.

So far, few experimental studies in behavioral economics focus explicitly on self-control problems. Ego depletion remains a method predominantly applied and discussed in psychology research. However, continued efforts to test and develop ways to influence self-control in experimental research can lay the basis to better understand the causal impact of the powerful ability.

Interestingly, recent research questions the ego depletion process and believes ego depletion only exists for those who believe in limited self-control resources: the theory may be a self-fulfilling prophesy. Further research in this direction could be helpful, for example to assess the impact of information treatments on ego depletion.

References

Cobb-Clark, D. A., Dahmann, S. C., Kamhöfer, D. A. & Schildberg-Hörisch, H. (2022). The predictive power of self-control for life outcomes.  Journal of Economic Behavior & Organization 197: 725-744.

Grohmann, A. & Hamdan, J. S. (2021). The effect of self-control and financial literacy on impulsive borrowing: Experimental evidence. DIW Discussion Paper No. 1950.

Tangney, J. P., Baumeister, R. F. & Boone, A. L. (2004). High self-control predicts good adjustment, less pathology, better grades, and interpersonal success. Journal of Personality, 72(2): 271-324.

Dr. Jana Hamdan 

Jana Hamdan is a Senior Researcher at Behavia and a Post-doctoral researcher at the German Institute for Economic Research (DIW Berlin). She holds PhD in Economics and has conducted experimental and survey-based research in various countries. Jana has published policy and academic articles, for example in the peer-reviewed Journal of Development Studies.

Have you ever day-dreamed about winning the lottery? You can almost see yourself being happy until the end of your days, spending great holidays in paradisiac beaches while at the same time investing your money wisely to prosper. And what about if you got dismissed from your job for professional misconduct? You would not have time to feel sad or concerned – you’d apply to two or three amazing positions and would certainly get two (or even three!) offers.

It is common to think about the future and to make predictions about how certain events will affect us. Some of these predictions help us to make decisions, such as whether we should buy a lottery ticket today or try to meet our job deadlines more often! The only problem is that every time we forecast feelings associated with great or terrible events, we tend to over- or underestimate them and their consequences.

This is a result of our wishes and desires! Wishing for the excitement of positive events and ignoring the burden of negative ones is a strategy used to maximize our well-being and to avoid preparing ourselves for painful moments.  In 2018, I was part of a research team that investigated this assumption by gathering information from hundreds of people on how they estimated the impact of desirable and undesirable events, imagining as if these were to happen to themselves or to others. We predicted that excitement of positive events would be overestimated when compared to grief of negative events. Since we see ourselves as better-than-average, we also predicted that this asymmetry would not be true when people made predictions about how others would feel.

In one of the experiments we conducted, some participants were randomly exposed to a series of hypothetical positive (e.g., having their work recognized with an international award; graduating in the top three of the class; winning the lottery) and negative events (e.g., being betrayed by a significant other; having cancer; getting fired) as if they were happening to themselves. Other participants were exposed to the same hypothetical events as if they were to happen to someone else. After each positive/negative event everyone was asked to estimate for how long they (vs. someone else) would feel happy/sad and whether each event was desirable.

Individuals estimated their own happiness associated to positive events to extend for a longer period than their sadness associated to negative events. What about others? Others lived for a similar period both their happy and sad feelings. But it is not new that what is positive is more desirable. We also confirmed that association between positivity and desirability. To disentangle positivity from desirability, in one of the other experiments, some participants were randomly exposed to scenarios where both desirable and undesirable events were inducive of either negative or positive feelings as if they were happening to the self. Other participants were randomly exposed to the same event as if they were happening to someone else. After each event, everyone was asked to estimate for how long they would feel e.g., grateful/angry. Regardless of their valence, feelings associated with desirable events are estimated to last longer than those associated with undesirable ones. Once again, this asymmetry between desirable and undesirable events was not true for other people.

Motivated thinking, or our tendency to accept more easily what we want to believe in, is a powerful tool that influences decision-making by maximizing the experience of desirable affective states. However, we believe that desirable asymmetric forecasts for the self may have harsh consequences on different grounds. Particularly, in risk-taking behavior: If we believe that our own desirable feelings are supposed to last longer than our undesirable ones, we might underestimate the consequences of our undesirable behaviors, becoming more vulnerable to irrational financial decisions or unhealthy behaviors.

Reference: Mata, A., Simão, C., Farias, A. R., & Steimer, A. (2018). Forecasting the Duration of Emotions: A Motivational Account and Self-Other Differences. Emotion.

Dr. Cláudia Simão 

Cláudia Simão is a Senior Researcher at Behavia and an Adjunct Professor at Católica-Lisbon School of Business and Economics. She has a PhD in Experimental Social Psychology, and over the last 10+ years she has designed 500+ research studies/experiments on human behavior. Cláudia published 35+ scientific articles in top journals addressing topics such as prosocial behavior, decision-making or positive emotions.

A recent analysis by Hall and Madsen (2022) tells a different story. After measuring the effect of a traffic safety intervention in Texas that uses billboards to display fatality messages, the authors found that displaying these messages increases the number of crashes by 4.5% in a 10 km downstream, which is comparable to raising the speed limit by 5-8 km/h. In fact, the larger the number of fatalities reported in the messages, the more harmful they are. This suggests that fatality messages are too salient in this context, leading to cognitive overload at a time when we need less distractions. As for the long-term effect, the results showed no evidence that fatality messages affect driving behavior outside the intervention period.

This paper is another reminder that behavioral interventions are tools and not truths about behavioral change. Truths about behavior is always found in the context, by which the intervention design is largely inspired.

Full paper: https://doi.org/10.1126/science.abm3427 

To decrease this participation gap, an award-winning paper by De La Rosa et al. tested the influence of 3 messages with different frames on people’s desire to claim government benefits: (1) psychological ownership, (2) social norms, and (3) urgency. The message that accentuated psychological ownership increased interest in government benefits by 20-128% across four experiments and performed consistently better than the urgency and social norm frames, suggesting the existence of serious aversions and psychological barriers to claiming government assistance.

The study shows how subtle variations in language can go a long way in increasing people’s fundamental desire to claim the benefits that are rightfully theirs.

Full paper: https://www.pnas.org/doi/full/10.1073/pnas.2106357118