Essays in Behavioural and Experimental Economics: Self-selection and Incentives

Abstract

This thesis discusses the importance of self-selection to understand how individuals behave and respond to incentives. Individuals self-select themselves into their preferred contracts in a number of ways. Labour contracts are a clear example where individuals choose a job according to its characteristics, whilst also aligning to their personal preferences and skills. The reasons why a worker chooses a job can predict how likely she or he will respond to different incentives, such as financial or social incentives. The economics wage efficiency theory predicts that a higher wage can both attract more productive workers and enhance performance. Laboratory experiments have constituently proven this theory, showing that when employers pay workers a ‘fair’ wage, workers reciprocate by working harder, as in a gift exchange fashion. Other studies have argued that social incentives can be equally effective at achieving the same goal: employers that donate a portion of their profits to charity, often known as Corporate Social Responsibility (CSR), can be attractive to workers and motivate them to work harder even at wages that are lower than competitor employers. More recently the gift exchange theory was challenged by field experiments that found little evidence of reciprocity. In the first chapter on this thesis, I argue that an important gap has not been addressed by the literature, namely, how sorting mechanisms can affect reciprocal behaviour. In lab experiments, employers and workers are often randomly and bilaterally matched in every round of a game. Whilst outside the controlled laboratory environment, workers choose the employer they want to work for and then decide how hard they are willing to work. To address this gap, I designed a modified version of the gift exchange game – the most commonly used game to study experimental labour markets – where two employers and one worker are randomly grouped together. In this experiment employers can use their initial capital to make wage offers to workers and donate any percentage of their potential profits to charity (i.e. the level of CSR). In the control group workers are randomly matched to one of the two employers, as in other standard laboratory settings. In the treatment group, in every round workers can choose an employer from a set of two competitors, before determining their level of effort. The first key result I found is that workers always choose the employer offering the higher wage. Furthermore, workers choose the employers offering a higher level of CSR only when the wage offer is identical to that of a competitor. I also find that wages have a larger marginal effect than CSR at enhancing workers’ effort. The second contribution of this chapter is its reconciliation of the mixed evidence between field experiments and lab experiments. I find that the presence of competition among employers reduces the level of reciprocity from workers. I argue that workers already reciprocate higher wage offers by choosing an employer over a competitor, hence feeling less pressure to work harder once in the job. Based on findings from the first experiment, I returned to the lab to test the role of competition and self-selection in a modified version of the earlier gift exchange game, this time without the presence of CSR. The results of this second experiment are the topic of Chapter 2. I again find that the presence of competition reduces reciprocity, supporting findings from Chapter 1. Another contribution of this chapter is the study of how external wage offers affect workers’ behaviour. In the control condition of the experiment workers cannot select their preferred wage offer and are randomly matched to one of the two employers, but they can still see the offer of the unmatched employer. I find that this external wage offer influences workers’ behaviour as a reference point: after controlling for all other factors, when workers are randomly matched to the employer offering a higher wage they provide higher levels of effort. More striking and significant is the evidence of loss aversion: subjects were more responsive to subjective losses than gains – that is, being paired to an employer who offered a lower wage was more ‘painful’ to the worker and led to a stronger (negative) reaction, than being paired to the higher offer employer, which led to a weaker (positive) reaction. In other words, workers penalised more employers that offered a lower wage than rewarding employers that offered a higher wage. Previous studies showed that reference points can influence workers’ effort. These include a target income they set for themselves or the amount other similar workers earn. In this experiment we show that another important reference point is the wage offered by another employer. If the current employer offers a wage that is above that of an external employer, workers will reciprocate by working harder, beyond the reciprocal response that would have occurred without the presence of the outside offer. If the wage offer is lower than that of a competitor, the worker will punish the employer with significantly lower effort. In the last section of the second chapter I compare results from the two experiments and show that in a competitive environment employers must compete more aggressively by offering higher wages to attract and motivate workers. Shifting resources away from wage offers to increase the level of CSR can lead to lower levels of reciprocal behaviour from workers and, consequently, reduce employers’ earnings. Employers interested in engaging in Corporate Social Responsibility initiatives as a Human Resources Management strategy to attract and motivate workers should consider how different types of workers will have varied responses to social incentives compared to more traditional financial incentives. Moreover, employers should understand the characteristics of the job that attracted their workforce in the first place so as to design incentives that reflect their preferences. In the last chapter of this thesis I explore the role of self-selection in determining the effectiveness of defaults. Defaults proved to influence behaviour across a range of areas, from retirement savings to organ donations. Perhaps the main reason why defaults are effective is that individuals have a strong tendency to remain in their current situation rather changing to an alternative option. This is often referred to as ‘status quo bias’. Seeing as this bias strongly influences human behaviour, it is important to understand why and how individuals self-select into a situation or contract in the first place in order to design effective defaults. At the same time, a key challenge presented by defaults is that they can reduce an individuals’ sense of control, and be ineffective or counterproductive if they do not reflect the decision-maker’s preferences and past behaviour. I study the role of defaults in a previously unexplored setting where a preference for control might be stronger than in other contexts: charitable giving. I analyse results of a field experiment ran by an NGO hosting an online peer-to-peer microlending platform. Lenders who had their loans fully repaid, but did not take any action for more than a year, received an email inviting them to use their money by a certain date in any way they preferred – withdraw, lend, donate or leave idle. In two treatment groups, lenders were told that if they did not take any action by the given date, their money was automatically going to be donated to the organisation (‘default donation’) or re-lent to a group of borrowers on their behalf (‘default loan’). Results show that both defaults were more effective at increasing the proportion of individuals giving to charity and the average amount given, compared to a simple ask. However, the default loan treatment was significantly more effective than the default donation. This suggests that to influence behaviour without risk of backfiring, defaults should be designed to consider individuals’ past behaviour and self-selection. Government agencies, not-for-profit, and private sector organisations interested in implementing defaults should first aim to understand why and how individuals find themselves in a specific situation or contract in the first place. Defaults that reflect a decision-maker’s preferences have a higher chance of achieving their intended objectives. Preferences can often be elicited by observing past behaviour. Furthermore, it is important to understand that choices made in the past can influence behaviour as reference points. Testing and evaluating different types of defaults can help improve their effectiveness and avoid counterproductive consequences. The three chapters in this thesis can be considered as individual standalone papers. The reader can review each chapter separately without loss of context. Each chapter includes an introduction, a literature review and hypotheses, and a discussion of the results from the experiments. Recommendations for future research and policy are discussed in the conclusions of each chapter.