Social Preferences: An Introduction to Behavioural Economics and Experimental Research

By Michalis Drouvelis

This introduction to one of the key areas of behavioural economics – social preferences – explains in clear, nontechnical language how particular groups of experiments have been used by behavioural economists to shed light on the processes of economic decision making. These include bargaining games, trust games and public good games. The significance of determinants such as punishment, sanctioning, emotion, cooperation, reciprocity, leadership, framing and cross-cultural differences are demonstrated and explained, and students are provided with the understanding and resources needed to replicate the experiments themselves.

• Language: English
• Publisher: Agenda Publishing
• Release date: Sep 30, 2021
• ISBN: 9781788214193

Michalis Drouvelis

Michalis Drouvelis is Professor of Behavioural Economics at the University of Birmingham.

Book preview

SOCIAL PREFERENCES

An Introduction to Behavioural Economics and Experimental Research

MICHALIS DROUVELIS

To my family

© Michalis Drouvelis 2021

This book is copyright under the Berne Convention.

No reproduction without permission.

All rights reserved.

First published in 2021 by Agenda Publishing

Agenda Publishing Limited

The Core

Bath Lane

Newcastle Helix

Newcastle upon Tyne

NE4 5TF

www.agendapub.com

ISBN 978-1-78821-416-2 (hardcover)

ISBN 978-1-78821-417-9 (paperback)

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

Typeset by JS Typesetting Ltd, Porthcawl, Mid Glamorgan

Printed and bound in the UK by TJ Books

CONTENTS

Preface

1.Introduction

1.1 Homo economicus

1.2 Behavioural and experimental economics

1.3 Deception and monetary incentives

2.Bargaining games

2.1 Introduction

2.2 Dictator games

2.3 Competition from proposers’ and responders’ side

2.4 Psychological factors

2.5 Financial factors

3.Trust and gift exchange games

3.1 Introduction

3.2 Disentangling motives in the trust game

3.3 Behavioural determinants of trust

3.4 Gift exchange games

4.Public Good Games I

4.1 Introduction

4.2 Do people cooperate?

4.3 Why do people cooperate?

4.4 Conditional cooperation

5.Public Good Games II

5.1 Introduction

5.2 Can pre-play communication promote pro-social outcomes?

5.3 Income inequality and public good provision

5.4 Social identity and discrimination in public good experiments

6.Leadership

6.1 Introduction

6.2 Sequential vs simultaneous public good games

6.3 Leader appointment

6.4 Who leads more effectively?

7.Public good games with sanctioning I

7.1 Introduction

7.2 The role of emotions

7.3 Sanctioning mechanisms

7.4 Does the presence of monetary sanctions always promote cooperation?

8.Public good games with sanctioning II

8.1 Introduction

8.2 Voting on public good institutions with punishment and rewards

8.3 Voting on formal sanctions

8.4 Third-party punishment games

8.5 Factors determining the assignment of third-party sanctions

9.Cross-cultural experiments

9.1 Introduction

9.2 Fairness and bargaining behaviour

9.3 Trust games

9.4 Cooperative behaviour

9.5 Negative reciprocity

Appendix A Experimental instructions

Appendix B Practical information

Notes

References

Index

PREFACE

The present book offers an introduction in the experimental economics literature pertaining to aspects of individuals’ social preferences. In its nine chapters and two appendices, this book discusses some key experimental economics paradigms that behavioural scientists are frequently using in order to gain a better understanding of human behaviour. These paradigms consist of bargaining games, trust games, simultaneous public good games without and with punishment, and sequential public good games. The extant literature has produced a huge literature and has considerably advanced our knowledge. The aim of this book is not to present a comprehensive literature review of the existing social preferences experiments. Instead, we offer an accessible discussion of a selective part of the literature that can be used for a semester-long course in behavioural and experimental economics, with a focus on social preferences. The book is targeted at those who have little background in the field of experimental economics and the choice of the topics discussed reflects a bias towards my own research interests. However, it is hoped that the offered discussion presents a series of novel and influential experiments in the literature.

It is also important to underscore that the length of each chapter has been selected with the main principle being that the material presented should be no more than 20 pages, which seems to be a sensible length for students’ weekly workload. In parallel, it was aimed that for some important papers, the key aspects of the experimental design, hypotheses and results are presented. The primary objective was to make the content of the book friendly to students, keeping the language as simple as possible (unlike academic papers which may use more technical terms sometimes) and helping them absorb the key elements more easily. At the same time, some experiments which are discussed in less detail aim to give students a general idea of the corresponding literature. Providing all this level of detail (at least) for some of the papers comes at a cost of limiting the number of papers that can be discussed in a given chapter but on the other hand, this structure hopefully gives a clearer description of the key papers, compared to other behavioural economics textbooks which only focus on a brief discussion of the main findings of papers. Some additional references are provided in the footnotes but these are complimentary in nature and discussion of these papers are much less detailed. Each chapter concludes by offering the main highlights through five bullet points, summarizing key issues discussed in the corresponding chapters. This, of course, does not substitute the level of detail included in the main text but it is meant to help students easily remember the main concepts for each chapter. At the end of the book, there are two appendices with useful material. Appendix A contains a set of sample instructions corresponding to key games employed in experimental economics to measure aspects of individuals’ social preferences. This material is hoped to be useful for academics as well as students who design and plan to run an experiment either as part of their own research or an assessed component in a behavioural economics module. Appendix B includes some practical information for the conduct of economic experiments containing useful concepts and design options that are frequently discussed in experimental papers.

The main readership is expected to be undergraduate or postgraduate students who would like to receive an introductory treatment in the social preferences literature. Consequently, the writing style is not formal and tries to explain the relevant papers in a very simple and straightforward way. No previous knowledge in mathematics is needed as the explanation of the material is not based on technical terms and concepts. Beyond acquainting themselves with the realm of social preferences, students may also find the content of the book inspiring for their undergraduate and postgraduate dissertations. A number of key experimental economics frameworks are discussed and the simplicity of the discussion may be beneficial for students who think of possible research questions for the purpose of their dissertations or group presentations in experimental economics courses.

I hope that you find the content of the book fascinating and that this will be your first step towards obtaining a better understanding of experimental economics research. Economic courses are normally based on describing abstract theoretical models, omitting to discuss empirical evidence about how individuals actually behave. This book will provide you with such evidence and it will make crystal clear that how models assume people to behave and how they actually behave in reality are two completely different things in many occasions. The selfishness assumption positing that people only care about their own welfare and just ignore others is frequently refuted. To find out more about the bright side of human behaviour, read on!

As a final note, I would like to dedicate this book to my family whose constant and unconditional support helped me complete this project (as well as many other projects!).

Michalis Drouvelis

1

INTRODUCTION

1.1 Homo economicus

Economists use theories and mathematical models to describe certain economic phenomena observed in real-life. Such theories can be parsimonious and make specific assumptions about how individuals should and do actually behave. One very popular such theory is the standard economic theory assuming that individuals are rational and selfish, caring only about maximizing their own material self-interest. In economics, rational individuals are those who have a consistent objective that they want to achieve.¹ When individuals’ objective is to maximize their own material gains from undertaking a certain action, such decision-makers are often described as Homo economicus who are assumed to have strong computational abilities, allowing them to weight accurately the costs and benefits of a decision that monetarily benefits them the most. Standard economic theory has been the central framework of analysis for many decades. But why this theoretical model has attracted so much interest in economics?

The main reason is that standard economic theory is a very simple model of economic behaviour, abstracting from the complications of natural environments where different factors are operative at the same time. Building economic models is like drawing a map. In both cases, it is important to consider which details are relevant to be specified and which are less relevant: maps are drawn including those details that enable travellers to arrive at their destination and economic models are built including those details which are relevant in that they can make predictions that describe well observed behaviour.

To the extent that the selfishness assumption can describe observed behaviour accurately, standard economic theory does a good job. It is therefore natural to gain a better understanding about whether and if so, under which conditions selfishness is a good or bad assumption? The general answer is that it depends on what our research question is. Sometimes selfishness is a relevant assumption for people’s behaviour; however, in many decision-making situations, it is not a good assumption and standard economic theory fails to explain economic behaviour. Take for example the case where we would like to understand why individuals engage in philanthropic activities or why they leave a tip in a restaurant or why they decide to volunteer in their free time? It easily becomes clear that, in all these cases, the construct of Homo economicus is going to provide us with a poor description of observed behaviour. The reason is that all these activities require that individuals sacrifice part of their own benefit to undertake an action that entails at least some monetary cost to them. So, a simple cost-benefit analysis would suggest behaviours that are against what we observe in real-life where people donate, leave tips and volunteer in their free time. In this book, we will discuss simple empirical paradigms frequently used in behavioural economics showing that individuals are not always driven by pure selfishness, but they exhibit social preferences in a wide range of economic activities (even if the future material benefits are zero). Understanding the behavioural content and forces describing non-selfish motives lies at the heart of behavioural economics, a rapidly growing field in economics.

1.2 Behavioural and experimental economics

Traditionally, economics as a field was regarded to be a non-experimental one. Looking at Milton Friedman’s (1953) essays on positive economics, it is noted that Unfortunately, we can seldom test particular predictions in the social sciences by experiments explicitly designed to eliminate what are judged to be the most important disturbing influences. Even almost three decades later, in the 1980s, Samuelson and Nordhaus’ famous book Principles of Economics mentioned that economists cannot perform controlled experiments like chemists or biologists because they can’t easily control other important factors. Just like astronomists or meteorologists, they usually have to solely use their observation (Samuelson & Nordhaus 1985: 8). While the role that experiments played at that point was peripheral, nowadays experiments are a well-established methodology in economics and their growing popularity is undoubted. An increasing number of researchers from different sub-fields within economics (e.g., in microeconomics, macroeconomics, public and development economics, among others) are using experiments as an appropriate method that can tackle their research questions. Among other indicators, the significance of the experimental research in economics is evidenced by the number of experimental papers being published in prominent general interest and top field journals in economics, but also by the number of several field journals whose titles contain or closely relate to the word experimental; for example, Experimental Economics, the Journal of Behavioural and Experimental Economics, the Journal of the Economic Science Association² and the Journal of Behavioural and Experimental Finance. The establishment of economics as an experimental science is nowadays a fact and the integration of experiments into the mainstream economic analysis has pushed the economics science towards new exciting directions that have improved our understanding of human economic behaviour.

While experimental and behavioural economics are two notions that sometimes maybe used interchangeably, they are actually separate ones but closely related to each other. Behavioural economics is a field within the economics science and experimental economics is the most commonly used methodology that is adopted to address questions which behavioural economists are concerned with. In particular, the field of behavioural economics uses interdisciplinary tools from other neighbouring sciences (such as psychology, politics, sociology and neuroscience) in order to enrich the arsenal of economics science and provide a more accurate explanation and description of real economic behaviour. Behavioural economists are interested in research questions such as: how do individuals make certain decisions in strategic and individual decision-making environments?³ Do people care about their own and others’ welfare? If so, how do they take into account the welfare of others and why do they do so even if caring about others yields no material benefit for them? How do people evaluate risky options and what are their attitudes towards risk? How do they reason when faced with options involving ambiguity or a trade-off regarding the receipt of goods or cash now vs at a later date? These are some of the questions that behavioural economists are trying to address and for which experiments is the most appropriate tool at their disposal. For the purposes of this book, our focus will be on strategic environments measuring aspects of individuals’ social preferences through the lens of experimental research.

Experimental economics is a method that brings real people to the laboratory (or the field) where they make real decisions which have actual payoff consequences (i.e. people can earn (or lose) money). Czibor et al. broadly define an experiment as a study that generates primary, or original, data in a controlled environment (2019: 375). There are two key pieces of information included in this sentence which we discuss in turn: an experiment is (1) a controlled process and (2) a data generating process. First, an experiment is a controlled process in the sense that the researcher controls (i.e. holds constant) all aspects of the decision-making environment that may influence behaviour and only changes one of these aspects – the one that is of main interest depending on the research question studied – which varies at a time. Such a controlled process in a well-designed experiment achieves a very tight level of control (ruling out external behavioural factors), allowing the researcher to draw causal conclusions among the variables of interest, ceteris paribus (other things being equal). Thus, identifying causal effects lies at the heart of experimental economics research.

Next, it is important to distinguish the term of correlation with that of causation. Correlation refers to an association between two variables describing how they vary together. For example, a positive (negative) correlation between variables X and Y means that when variable X increases, variable Y increases (decreases). To establish causality, we need the presence of correlation between the two variables which should not be spurious (i.e. caused by a third variable) but also, we need to ensure that the cause (i.e. independent variable) has come before the actual effect (i.e. dependent variable). When designing an experiment, it is important that the researcher employs an appropriate methodology/design that will help them satisfy these conditions in order to capture the causal relationship between two variables (keeping other factors constant).

Second, apart from being a controlled process, an experiment is also a data generating process whereby the researcher can create their own data set (directly linked to the hypotheses tested in the experiment). In all sciences, and economics is no exception, theory offers the tools of analysis which outline how individuals should and do behave in certain economic phenomena. When building a theoretical model, a number of assumptions are in order (see also Croson & Gächter 2010). These relate to individuals’ preferences, their levels of sophistication (and beliefs about states of the world) and how predictions about economic behaviour can be made using game theoretic concepts. For example, in terms of social preferences (which this book is focusing on), standard economic theory would assume that people are rational, expect others to be rational and in terms of behaviour, it predicts that people should only care about maximizing their own material payoff (having high levels of sophistication). Yet, the extent to which theoretical predictions hold is a matter of empirical investigation.

Before the development of experimental methodology, economic theories could not be tested due to the lack of availability of appropriate data. Typically, economists were using observational data which were useful in their analysis but often important variables were missing as they could not be found in the field. This deprived empirical economists from the possibility to directly test for specific theory-based hypotheses. As a result, developing alternative theoretical models was not possible due to the limitations about data availability and theory testing. While standard economic theory made useful predictions for economic behaviour, the extent to which these predictions were empirically valid remained an open question. The revolutionary advancement of the experimental methodology changed this situation markedly, allowing researchers to charge the engine of science (following the terminology in Friedman & Sunder 1994). The alternation of theory with (experimental) evidence has substantially improved our ability to explain behaviour both theoretically and empirically, to a much greater extent than it used to be a few decades ago. In parallel, experiments enriched the empirical arsenal of economists who now have access to more sources of data: laboratory and field data.

The advantage of laboratory data is that (assuming that rigorous procedures have been followed in the design of the experiment) they have internal validity in that the researcher can draw reliable conclusions – i.e. the cause-and-effect relationship – from the implemented experiment. The issue of internal validity relates to how rigorous are the procedures that have been followed when the experiment was designed. As noted earlier, the process of a carefully designed experiment should enable the researcher to establish a causal relationship between two variables of interests. An important dimension for establishing such causality is the use of randomization. For example, suppose that the aim of an experiment is to test whether eating fruits affects performance in a maths task. In this simple thought experiment, the outcome variable is task performance. To assess the causal relationship of eating fruits on behaviour, the experimenter needs to consider two conditions: the control group where people perform the maths task (without eating fruits before doing so) and the experimental (treated) group consisting of individuals who eat fruit and then perform the maths task. To ensure internal validity, it is important that the assignment of experiment participants in the control and the treated group is randomized. The process of randomization allows the experimenter to control for variables that are not directly controlled in the experiment. For example, it might be that people behave differently depending on whether the experiment is conducted in the morning or in the afternoon. If one of the two treatments is always run in the morning and the other treatment group is always run in the afternoon, this introduces another variable that may interact with task performance, making the identification of the causal effect between eating fruits and task performance spurious. Randomizing which time to run the experiment rules out potential time effects. It is important to randomize as many as possible aspects of the experiment, for example, relating to who to invite when an experiment is scheduled, which experimental condition to run when, which computers participants will be allocated during the experiment, which role participants are allocated to subjects during the experiment.

Although a carefully designed laboratory experiment has internal validity, the issue of external validity is a separate and an equally important one. External validity pertains to whether the conclusions drawn from an experiment can be generalisable to the real world. The increasing level of control in the laboratory raises the question of whether the behaviour observed in the laboratory is artificial and thus, whether conclusions drawn from laboratory experiments can be extrapolated outside of the lab. Falk and Heckman (2009) provide arguments in favour of conducting more laboratory experiments and explain that observing behaviour in the field is not more informative than observing laboratory behaviour. For example, the use of both lab and field methods may allow us to better understand mechanisms driving observed phenomena in the field but also help us to establish experimental validation of survey instruments. It is therefore important to combine different methodological tools – such as field, survey and experimental data – in order to advance our knowledge about human behaviour. The ultimate choice of a particular method depends on the exact research question to be studied, with both laboratory and field experiments being complements (not substitutes) to each other.

Understanding the extent to which behaviour in the lab and the field is correlated has received increasing attention among behavioural economists, with the existing evidence being mixed. While a substantial body of papers finds that lab and field behaviours strongly correlate, indicating that external validity of a diverse set of laboratory studies has been established, there are, however, exceptions (showing that there is no correlation of behaviour between the two environments). Evaluating the external validity data of laboratory measures is a fruitful area for further research (for a review on this topic, with an emphasis on experimental games measuring aspects of social preferences, see Galizzi & Navarro-Martinez 2018 and references therein).

The controlled environment of the laboratory allows researchers to test for the robustness and replicability of the reported findings. As noted above, the level of control induced in the laboratory abstracts from many elements that are present in real-life environments (similar to what a theory does). Such abstraction which oftentimes is done on purpose helps researchers to focus on the closely related aspects of the addressed research question. Whether laboratory findings in one context are robust in another context has become of utmost importance, with some of the key questions in the literature being: can evidence from one experiment be replicated in alternative contexts? Can previously reported results be reproduced when protocols from the original study are followed (using different samples)?

Providing convincing answers to such questions is crucial for the validity of experimental research in economics and further contributes to the progress of economics as a science.⁴ A popular study on replicability of well-published papers has been conducted by Camerer et al. (2016). In particular, the authors attempted to replicate 18 studies published in the American Economic Review and the Quarterly Journal of Economics between 2011 and 2014. All of these replications followed predefined analysis plans that were made publicly available beforehand and they all have a statistical power of at least 90 per cent to detect the original effect size at the 5 per cent significance level. Their evidence shows a replication rate of 61 per cent (i.e. 11 out of 18 studies were successfully replicated). While this is a relatively high replication rate (for example, similar exercises in the field of psychology shows a replication rate of 36 per cent), it also suggests that there is still room for improvement.

As a final note, which is also linked with our discussion in the next section, it is worth highlighting that the methodological differences between economics and psychology are likely to yield the observed differences in terms of reproducibility. Using Camerer et al.’s (2016: 1435) words, two reasons that have contributed to the relatively higher replication rate in economics compared to psychology are as follows:

First, experimental economists have strong norms about motivating subjects with substantial financial incentives and avoiding the use of deception. These norms make subjects more responsive and may reduce variability in how experiments are performed across different research teams, thereby improving replicability. Second, pioneering experimental economists were eager for others to adopt their methods; to this end, they persuaded journals to print instructions and even original data. These editorial practices created norms of transparency and have made replication and reanalysis relatively easy.

Next, we discuss in more detail the differences in standards and practices between economics and psychology experiments.

1.3 Deception and monetary incentives

The use of experiments is not a privilege that only belongs to economists but also to other scholars from different disciplines (for example, those doing research in the so-called hard sciences like in physics and chemistry) who have employed experimental methods long before this is done by economists. After