Incentivising the Social Discounting Task: A Laboratory Experiment

• DOI: http://doi.org/10.1111/saje.12191
• Published date: 01 June 2018
• Date: 01 June 2018

INCENTIVISING THE SOCIAL DISCOUNTING TASK:

A LABORATORY EXPERIMENT

FREDERIK BOOYSEN*

‡

,ALISTAIR MUNRO

†

,SEVIAS GUVURIRO

‡

,TSHEPO MOLOI

‡

AND

CELESTE CAMPHER

‡

Abstract

Incentivising the social discounting task impacts the measurement of altruism in a student population.

Incentivised subjects are more altruistic at close social distances, especially subjec ts who are less altruis-

tic, thus providing evidence of reciprocal altruism. There is also some ev idence of hypothetical bias

among more altruistically inclined subjects. Making payments real also influences the subject’s choice

of recipient. Paid subjects select more geographically distant, but psychologically closer subjects,

because of prospects for increased anonymity and enforced reciprocity, respectively. Further research is

required to verify the robustness of these results, in the laboratory and especially in the field.

JEL Classification: C91, D64

Keywords: experiment, social discounting, incentives, South Africa

1. INTRODUCTION

Altruism is one of the most important social preferences driving pro-social human behav-

iour and is of particular interest to economists. In the social discounting task (SDT), a tool

used in Experimental and Economic Psychology to examine patterns of altruism, and com-

monly used in conventional laboratory experiments with students, payments typically are

hypothetical. Of the approximately 25 experiments on social discounting conducted to

date, few employ real payments. The exceptions are Locey et al. (2011), who introduce

payment as an experimental treatment, and the more recent work by Strombach et al.

(2013, 2015, 2016). Though the empirical evidence is mixed (Camerer and Hogarth,

1999; Ben-Ner et al., 2008; Vlaev, 2012), economists methodologically are largely scepti-

cal of the value of experiments in which no real payments are made (Hertwig and

Ortmann, 2001). Even within the SDT literature, Yi et al. (2012), recognise the limitation

of the widespread useof hypothetical incentives in social discounting experiments.

Incentivising the SDT has two potential methodological repercussions. Making pay-

ments real might affect the weight that subjects place on money received by recipients,

thus influencing their choices and the resultant estimate of altruism. This potential bias,

in terms of construct validity (i.e. the ability of the task to measure what it claims to

measure) and ecological validity (i.e. the ability of the task to represent a real-life setting),

may be particularly problematic where the focus is on studying how neurological

(Pornpattananangkul et al., 2017; Strang et al., 2017) and psychological processes (Olson

* Corresponding author: Population Health, Health Systems and Innovation (PHHSI),

Human Sciences Research Council (HSRC), 134 Pretorius Street, Pretoria 0002, South Africa.

E-mail: fbooysen@hsrc.ac.za

†

National Graduate Institute for Policy Studies (GRIPS)

‡

Department of Economics, University of the Free State.

V

C2018 Economic Society of South Africa. doi: 10.1111/saje.12191

153

South African Journal of Economics Vol. 86:2 June 2018

South African Journal

of Economics

et al., 2016; Strombach et al., 2016) and human behaviours (Bradstreet et al., 2012;

Sharp et al., 2012; Yi et al., 2012) are related to social discounting.

Having real payments may also change the set of people who are nominated as recipients,

an issue that has not been examined in the few studies that do have real payments. One reason

why the set of nominated recipients may change depending on payment rules is related to the

reason why real payments may change the level of payment. When payments are hypothetical,

the theoretical recipient never has a reason to think they have been the object of an experiment

and that someone made a decision about giving them money. However, when payments are

real, the recipient may receive money and in this case she or he may be able to identify the

donor even if this information is not formally given out. Anticipating this possibility, the sub-

ject may wish to be generous (to avoid social embarrassment for example, or because there is

some possibility of future reciprocity). At the same time, the subject faced with the possibility

of real payment may choose their target recipient in a different way when payments are real.

They may, for instance, choose someone who is more likely to reciprocate or someone distant

and anonymous with whom they would not feel social embarrassment at giving little.

According to Ol son et al.(2016:60),the“socialdiscountingparadigmisapowerfulmeans

of quantifying altruism in humans.” If behavioural and experimental economists are to adopt

the SDT as measure of human altruism and social discounting in their incentivised economics

experiments, it is necessary to have an understanding of how using real payments in the task

matters for the measurement of altruism, both in terms of the level of measured altruism and

its association with other preferences and related outcomes as well as the set of recipients and

the role their characteristics play in understanding giving. The paper is structured as follows:

Section 2 describes the hypotheses, while Sections 3 and 4 outline the materials and methods of

analysis, respectively. Section 5 contains the results and their discussion. Section 6 concludes.

2. HYPOTHESES

Based on the literature, two main hypotheses guide the first part of the analysis, i.e.theques-

tion as to the impact of incentives on social discounting. Firstly, it is hypothesised that sub-

jects completing the incentivised task may be less altruistic than subjects completing the

standard task. Generally, due to hypothetical bias (Vlaev, 2012), when faced with the pros-

pect of earning real money (as opposed to giving money to others), subjects are expected to

exhibit selfish behaviour and in this case to switch from A to B only in the latter rows of the

table, specifically for the more socially distant recipients (Hypothesis 1). Secondly, it could be

hypothesised that subjects, under the condition of real incentives, may be more altruistic

insofar as potential recipients, who receive actual money, can identify the donor (Locey and

Rachlin, 2015), who in turn, in an act of reciprocity, may potentially share that money with

the subject. In other words, receivers may potentially learn from whom they may receive a

payment, and senders correspondingly may give more. Osinski (2010), in line with this,

found that preferences for sharing increases as a function of prospects for reciprocity. Due to

the presence of reciprocal altruism (Simon, 1990; Ale et al., 2013), i.e.altruistic“behaviour

undertaken with an expectation of reciprocation” (Simon, 1990:1665), subjects completing

the incentivised task could therefore switch from A to B in the early rows, i.e.bemorealtruis-

tic, especially at closer social distances (Hypothesi s 2). In both instances, the null hypothesis is

that incentivising the task will not influence the estimate of the extent of altruism.

In terms of the second research question, whether or not making payment real might also

change the choice of the recipient. The null hypothesis here is that treatment has no effect

154 South African Journal of Economics Vol. 86:2 June 2018

V

C2018 Economic Society of South Africa.