Demand side of clientelism (2017 - 2019)
Date: 2017 - 2019
Research team: Harold Kincaid, Miquel Pellicer and Eva Wagner
Despite strong academic and policy interest in the phenomenon of political clientelism, little mainstream political science research exists on its demand side; i.e. research that focuses explicitly on the choices available to potential clients and the factors that affect such choices. This project seeks to undertake a systematic investigation of the demand side of clientelism in Tunisia, South Africa and Colombia. Importantly, we consider demand for different types of clientelism, such as vote buying and more “traditional” forms of clientelism. We build on several strands of literature as well as on our own previous work to provide a theoretical framework of demand for clientelism. From this framework, we develop hypotheses on the direct effect of certain factors such as risk aversion, beliefs in group political efficacy and perceptions of legitimacy of social inequalities. We also develop hypotheses on the role of contextual factors, such as community isolation, for clientelism and as moderator of the direct effects.
In order to test these hypotheses, we will conduct surveys in Tunisia, South Africa and Colombia. These surveys will introduce several state-of-the-art techniques for measurement and analysis of clientelism. First, we use recent techniques to obtain accurate measures of potentially problematic variables. Most importantly, we measure demand for clientelism using list experiments, a technique used to obtain truthful answers to sensitive questions. In addition to clientelism, we measure variables such as risk aversion and discount factors with simplified versions of state-of-the-art procedures in experimental economics. Second, we estimate the causal effect of perceptual factors by means of randomly assigned information treatments embedded in the surveys (i.e. survey experiments). We will design information treatments in order to affect perceptions such as efficacy or inequality legitimation. In order to develop our treatments, we will draw on an extensive literature in social psychology that has conducted such experiments mostly in the lab with students. Throughout the design phase of the project, focus groups will be conducted in the three countries to assess the relevance of our hypotheses and to help design meaningful treatments. Third, we analyse the role of contextual factors, such as community connectivity, by conducting our surveys in different locations within each country. We will use the lessons learned from our empirical analyses to refine our framework of clientelism choice and to propose context-sensitive policies to combat clientelism.
Decomposing trust into risk preferences, prosocial influences, and subjective beliefs 2 (2015 - 2018)
Date: 2015 - 2018
Research team: Harold Kincaid, Andre Hofmeyr, Brian Monroe, Rinelle Chetty (Masters) and Tarryn Beattie (Masters)
Trust is central to many, if not most, human interactions. It has been studied using a variety of methods but an economic experiment known as the trust game has become the behavioural measure of choice in social science investigations of trust. This measure, specifically, the amount sent in the game, is often used uncritically to compare levels of trust across people and cultures even though it may be confounded by a range of factors such as risk attitudes, prosocial influences, and subjective beliefs about return probabilities. We adopt an incentive-compatible experimental design and a full information maximum likelihood statistical framework to decompose the amount sent in a trust game into a measure free of these potential confounds and thereby determine whether the amount sent in this game is anything more than just the sum of its parts.
A successful pilot was run in 2015 and a more sophisticated follow-up study will be conducted in 2018.
A contingency management smoking cessation intervention (2017 - 2018)
Date: 2017 - 2018
Research team: Harold Kincaid, Andre Hofmeyr and Olivia Rusch (Masters)
Tobacco consumption is a pressing global issue, leading to approximately six million deaths each year. In South Africa, the smoking prevalence rate is stubbornly high, implying that a successful smoking cessation programme could have large social benefits, particularly if it targets young smokers. Contingency management (CM) interventions, which provide cash transfers conditional on biochemically-verified abstinence, have been effective in bringing about increased smoking cessation rates. However, the cost, complexity, and staff burden of CM interventions act as a barrier to their widespread adoption. For example, CM requires frequent monitoring of participants; tracking of participants’ substance use and reinforcement history; and, often significant, funds to pay for abstinence incentives.
As a response to this issue, we conduct a randomised controlled trial (RCT) to test whether a low-cost, low-intensity CM intervention is an effective form of treatment for tobacco addiction in a student population. The programme provides relatively modest abstinence-contingent cash rewards and had a low staff burden. The decision to target students was based on a number of factors: they are easy to recruit and track; their lifetime exposure to cigarettes and their smoking intensity tend to be low compared to older smokers so they may be more susceptible to an intervention (Mayhew et al. 2000); their relatively low incomes mean that the rewards on offer may be more salient to them than people in fulltime employment; and, if the programme is successful at getting them to quit smoking, the personal and societal benefits are large and continue to accrue over time.
In addition, our study goes beyond previous research that focusses on the efficacy of CM as a tool for treating tobacco addiction. Previous studies have taken an epidemiological approach to the impact of cessation programmes on smoking prevalence and intensity. In addition to taking measures of smoking outcomes, we use sophisticated techniques to elicit and estimate the risk preferences and time preferences of smokers involved in our cessation programme. The combination of these sources of information allow us to investigate whether a relationship exists between a smoker’s risk and time preferences and their participation and success in the cessation programme. This has clear policy relevance. For example, if people who are less risk averse and/or who discount the future at a higher rate are less likely to succeed in the programme, it may be necessary to adapt the CM design to target these ‘hard-to-treat’ smokers (e.g. front-loading the incentive schedule by offering larger rewards at the start of the cessation programme).
In sum, we investigate the efficacy of a low-cost, low-intensity CM smoking cessation intervention in a sample of university students and analyse the behavioural correlates of the decision to quit and the likelihood of remaining abstinent.
Understanding addiction: using economic experiments to understand the dynamics of tobacco smoking (2015 - 2017)
Date: 2015 - 2017
Research team: Harold Kincaid, Andre Hofmeyr, Don Ross and Glenn Harrison
Addiction is a pressing social problem internationally and especially in South Africa. It is important that the best tools of economics be applied to addictive behaviour to aid in the successful development of policy and interventions. This project seeks to advance that goal.
Addiction is an important problem for economic theory: why do most addicts expend resources to acquire their targets of addiction but then incur real costs to try and reduce or limit their consumption of these goods? Furthermore, why is the typical course of addiction characterised by repeated unsuccessful attempts to quit prior to final abstention? From the standpoint of standard consumer theory in economics these patterns of behaviour are difficult to rationalise. There is a theoretical literature in economics which models habit-forming behaviours, of which addiction is an extreme type, but there is a paucity of experimental economic studies eliciting and comparing the preferences – specifically, instantaneous risk preferences, time preferences, and intertemporal risk preferences – and beliefs that economic theory suggests may differ between addicts and non-addicts. The experimental research that has been conducted has been dominated by psychologists, and some economists have begun to follow their methodological lead. However, detailed reviews of the experimental literature on addiction highlight a number of methodological and statistical limitations in the ways these data have been collected and analysed.
We elicited the instantaneous risk preferences, time preferences, intertemporal risk preferences, and subjective beliefs of student and staff samples of smokers and non-smokers at the University of Cape Town using an incentive-compatible experimental design. We will use a cutting-edge maximum likelihood statistical framework, which is consistent with the data generating processes proposed by structural theories and accounts for subject errors in decision making, to draw robust inferences about the relationship between instantaneous risk preferences, time preferences, intertemporal risk preferences, subjective beliefs and addiction. The end product will be a better understanding of addiction to inform policy and interventions to treat addictive behaviours.
Impact of information on sure-thing principle (stp) violations (2017)
Research team: Glenn Harrison, Nicky Nicholls and Don Ross
This study followed up on a line of experiments that investigated the impact of information on violations of the Compound Independence Axiom (CIA). We aimed to correct methodological limitations in previous work reporting that Sure-Thing Principle (STP) violations do not decrease with information. Our design separated income information from information about event probability distributions and used a more robust, incentivized, process for eliciting subjective beliefs about the latter. We also use an established set of lottery choice questions to understand the risk attitudes that underlie choices and that condition stated subjective beliefs of respondents. This allowed us to jointly estimate subjective beliefs and structural details of respondents’ risk attitudes, including non-parametric tests of the CIA and ROCL axioms. Asking respondents to give detailed beliefs at multiple points where information is given allows us to better understand deviations from Bayesian decision making, and to see whether beliefs update with information in a Bayesian way. We included additional choices under uncertainty. This allowed us to estimate risk preferences such that the welfare implications of STP violations in terms of foregone Certainty Equivalent (CE) can be assessed.
Risk preference structures and evolution of rugby bets under changing information (2016)
Research team: Frauke von Bieberstein, Eberhard Feess, Glenn Harrison, Helge Mueller, Don Ross and Korey Rubinstein
We obtained years of data on bets placed by New Zealanders on international rugby tests. These data provided the basis for estimating a model relating pre-game odds to bet distributions in that population. We then recruited participants from amongst University of Auckland students to place bets on upcoming Tri-Nations Tournament fixtures at various distances from the events. Published odds were recorded at each betting point. We also elicited subjects’ risk preferences and subjective beliefs about probability distributions. Using the experimental betting data to parameterize the model of the historical data, we examine the relationship between dynamic receipt of information about odds, and beliefs and preferences concerning risk.
The behavioural identification and prevalence of pathological gamblers in Denmark (2015)
Research team: Glenn Harrison, Morten Lau, Don Ross and J. Todd Swarthout
We first designed and conducted a survey to estimate the prevalence of disordered gambling among Danish adults. We then randomly selected a subset of respondents, weighted so as to over-sample those with high disordered gambling risk scores, and conducted a series of experiments with them. These included experiments designed to elicit risk preferences, and experiments in which subjects gambled online on simulated slot machines we designed. All experiments were also conducted, for purposes of control and calibration, on a sample of Georgia State University undergraduates.
Relationships between product information, risk preferences and retail investment portfolio choices in cape town working adults (2015)
Research team: Glenn Harrison, Kanshukan Rajaratnum, Don Ross, Dave Strugnell, Brian Monroe, Mark Schneider and Gary White
We conducted experiments with staff at the University of Cape Town as subjects to investigate relationships between individuals’ risk preference structures, basic demographic profiles, and extent of success in choosing optimal retail investment portfolios (relative to their risk preferences). The purpose was to advise Allan Gray on which clients would derive benefit from an investor education protocol they had designed. There was prior agreement on which data would remain confidential, and which data could be used for scholarly publication.
School of Economics Building
University of Cape Town