Coping with Heterogeneous Opinions – CHOp

Diverse opinions and interests show up in all social issues. Individuals disagree on almost any issue one can think of. Even when scientific evidence is available, diversity of opinions often exists concerning its interpretation. Heterogeneity is a basic ingredient of any society, often a source of richness, sometimes a puzzle and always a major source of economic exchange.

In Economics, one distinguishes between tastes and beliefs. Tastes are represented by a utility function while beliefs are represented by a probability distribution over the set of future contingencies. Heterogeneous tastes are taken as given. Heterogeneous beliefs are often viewed as stemming from different information, leading to the conclusion that individuals with the same information should hold the same beliefs.

We will enlarge this simplistic view: tastes are not always well-formed, beliefs are not always probabilistic or formed according to Bayesian updating of some initial prior, and individuals are not omniscient about the future. Therefore equating an opinion, by which we mean a subjective evaluation of a situation, with a probability distribution over a well defined set of events is overly restrictive. This enlarged perspective on the way we form opinions leads to study their heterogeneity and raises two broad questions. The first concerns the sources and consequences of heterogeneous opinions. The second is how to build an aggregate opinion from individual ones. These positive and normative questions call for analytical and experimental investigations.

From a descriptive viewpoint, instead of viewing disagreement as a pathology or a mere informational issue, we will explain how it comes about and how the social context matters in reducing or widening it. We will study the consequences of heterogeneity and show how one can use the structure that arises when it stems from underlying ambiguity. Once heterogeneity is recognized, a natural question emerges: do investors with the «wrong« opinions get wiped out of the market? A question we will study adopting our enlarged perspective. Heterogeneous opinions also affect usual economic arrangements and we intend to provide results that go beyond the view that anything goes when such heterogeneity is allowed. Finally, we will study how collective beliefs (public opinions) affects individuals' beliefs.

From a normative viewpoint, in many situations, opinions of various actors on a range of relevant but complex issues are required, and will ultimately need to be aggregated. Such opinions often involve a double source of diversity: different actors may be more or less competent on a given topic and the same actor may have different levels of expertise on different issues.

Our objective is to build aggregation rules that can be implemented in actual settings. Several issues have to be addressed. The degree of competence of experts should be relevant in the aggregation procedure. The source of experts' disagreement should be identified and taken into account: the aggregation rule should not be the same in a situation where agents are Bayesian agents each with a piece of information and in a situation in which they have the same information but process it differently. We will experimentally test various rules and investigate the influence of the aggregation rule on the relationship between individuals' and social rationality.

The impact we seek first and foremost is a scientific impact. We will produce papers that will have an impact in the profession and hopefully drive other scientists to introduce opinion heterogeneity to a larger extent and in a more encompassing format than what is the case in Economics in particular. Thus, the project's first aim is to advance scientific knowledge, along various directions. Besides publications, presentations in conferences, specialized workshops and seminars will also be a way to spread the research done among the international scientific community.

We do think that the results we expect could have an important impact beyond the academic circles. It is a triviality to observe that Western societies in particular are increasingly divided on political and societal issues. Part of the division rests on very different ways of apprehending the future and not confronting others' views. We expect our results to provide ways to tackle the issue of how to take public actions in the best way when opinions differ. We will not tackle the sources of these differences, but rather work on how this affects welfare criteria on which to base public policy. We will also provide understanding of how these differences affect basic economic arrangements and mechanisms. The normative side of our project is thus for instance highly relevant for the work of the IPCC on how to convey to the policy makers the nature of the consensus on various climate change related issues.

After 18 months, the CHOp project has generated 2 publications as well as 24 working papers, and given rise to 23 presentations at conferences and seminars.

The project has also given rise to intense exchanges and discussions, both within and between different communities. Besides internal workshops gathering the group members, CHOp organized, in collaboration with the ANR ColAForm project, the DBAED conference (Deliberation, Belief Aggregation, and Epistemic Democracy), which took place in June 2018 in Paris. This an inter-disciplinary conference gathering researchers from economics, philosophy, psychology, political science, sociology, mathematics, computer science, and physics. During three days classical presentations alternated with extended discussions and round tables in order to stimulate exchanges across disciplines. This conference was a great success and it was decided to organize a second edition in June 2019, in the same format and again in collaboration between CHOp and ColAForm.

CHOp also organizes in June 2019 the RUD conference (Risk, Uncertainty, and Decision). This conference gathers each year the best internal specialists in decision sciences, and is this year coordinated with the D-TEA conference (Decisions – Theory, Experiments, and Applications), both conferences taking place at the same venue over a full week and with a high number of common participants. These two conferences are concerned in particular with models of agents' beliefs, and especially the notion of ambiguity that is a central topic in the CHOp project.

This project tackles the issue of the diversity of opinions in a society. It is grounded in Economics but is at the intersection with several other fields: Psychology, Statistics, Decision Theory, Analytical Philosophy, Social Choice, and Political Science. It goes beyond equating individuals' opinions with their probabilistic beliefs, a standard practice in Economics, and aims at encompassing notions like unawareness, ambiguity, and competence to analyze opinion heterogeneity more broadly. The general questions we will explore are the following ones:
- How does diversity of opinions come about and how can it survive?
- What is the impact of this diversity on economic arrangements?
- How do individuals react when confronted with other's opinions?
- How can one aggregate the various opinions to make the best decisions?

These questions deserve to be treated both normatively and positively, calling for various analytical approaches (e.g. axiomatic treatment, modelling of markets, etc.) and experimental approaches (from psychophysics and social psychology). From a positive side, we will provide models explaining how individuals exposed to the same information can disagree. This requires to think outside of the usual Bayesian, common prior framework. Getting outside of this framework is also required to provide an understanding of how disagreements can survive in the long run, as we aim to show, contrary to a long-standing view in economics that "irrational" opinions get wiped out of the market. Experimentally, we will focus on how individuals integrate heterogeneous opinions when making decisions and, in particular, how social pressure affects individuals' opinions. We will finally analyze how opinion heterogeneity affects economic arrangements such as bargaining, risk sharing and financial markets, incentive provision, etc.

From a normative side, we will concentrate on the many ways different opinions can be aggregated to form a social or group opinion. A particular attention will be given to procedures that allow one to extract a measure of competence or expertise from opinions expressed by members of a society or group. These degrees of competence will then be used to weight the various opinions in order to make the best informed decision for the group. We will provide an experimental assessment of these rules, compared with other rules studied in the literature. We will also go beyond taking opinions as exogenous and will see how correlations (and, possibly, failure to recognize these) among the sources of information accessible to individuals changes the way aggregating procedures like voting work. This study will fit into a more general research question which is how to aggregate "ill-defined" or "biased" opinions.

The project's aim is mainly to advance scientific knowledge on these issues. It could lead to define practical ways of aggregating opinions or rankings that could be of interest for the private and public sectors.

The team of researchers assembled for this project have expertise in the various fields the project is touching. They have already worked, produced scientific papers published in the best international outlets and organized events together.