How do bonus structures affect risk and effort choices? Insight from a lab experiment

Qun Harris, Analise Mercieca, Emma Soane and Misa Tanaka.

The bonus regulations were introduced based on the consensus amongst financial regulators that compensation practices were a contributing factor to the 2008-9 financial crisis. But little is known about how they affect behaviour in practice. So we conducted a lab experiment to examine how different bonus structures affect individuals’ risk and effort choices. We find that restrictions on bonuses, such as a bonus cap, can incentivise people to take less risk. But their risk-mitigating effects weaken or disappear once bonus payment is made conditional on hitting a high performance target. We also find some evidence that bonus cap discourages effort to search for better projects.

The Bonus experiment

In the European Union (EU), the ‘bonus cap’ was introduced for staff who have material impact on the institution’s risk profile, restricting their variable pay to be no more than 100% of their fixed pay (or 200% with shareholders’ approval). A proportion of the bonus also needs to be deferred, and is subject to ‘malus’, meaning that the deferred bonus could be forfeited if certain conditions materialise before it vests. The UK regulators also introduced a clawback rule, which requires at least 40% of affected bankers’ variable pay to be deferred for a period of three to seven years, and enables their variable pay to be clawed back for a period of seven to ten years after it is awarded. But as bonus regulations do not fully control the pay structure, banks may choose to adjust pay contracts if they wanted their employees to take greater risks.

Our study examined how bonus restrictions affect incentives using the randomised controlled trial (RCT) method. In our lab experiment, all participants were first asked to select an asset to invest a hypothetical inheritance. Their choice in this task (Task 0) gave us information about their inherent risk preference – i.e. their willingness to take risks with their own money. No bonus was paid for this task. The participants were then randomly assigned to three different bonus groups, and were asked to engage in several asset selection tasks as an investment manager at a hypothetical bank. In these tasks, participants could earn a cash bonus that depended on the return on their chosen asset. 

In the first asset selection task (Task 1), a bonus was paid as long as the return on their investment was positive, depending on the bonus group the participants were assigned to:

  • Proportional bonus group: Participants were paid a bonus proportional to the asset return.
  • Bonus cap group: Participants were paid a bonus proportional to the asset return, but the maximum bonus in each task was capped.
  • Malus group: Participants were paid a bonus proportional to the asset return, conditional on the project not failing for the next time period.

In the subsequent asset selection task (Task 3), participants were told that the bonus will be paid according to the formula above, but only if the return on their investment meets a high performance target.  The participants were also given an effort task (Task 2), in which they could reveal an additional investment option for each correctly performed arithmetic calculation, before making the asset selection decision. These arithmetic tasks were onerous but basic, so we measured effort by the number of arithmetic questions attempted.

How did bonus influence risk and effort?

Chart 1 below shows the risk choice of different bonus groups: the green bars show the proportion of participants who chose low risk assets (Risklevel=1), the blue bars show the proportion that chose medium risk assets (Risklevel=2), and the red bars show the proportion that opted for high risk assets (Risklevel=3). 

Chart 1: Risk choices under different bonus schemes

Some interesting results emerge from our study. First, when individuals were rewarded proportionally for positive investment returns but not penalised for negative returns, they tended to take greater risks than they would with their own money: the red bar in Chart 1 becomes bigger as the proportional bonus group moves from Task 0 to Task 1.  This result appears to confirm the regulators’ belief that such bonus schemes can indeed encourage excessive risk-taking. 

Second, the bonus cap and malus groups were less likely to invest in high risk assets than the proportional bonus group: as shown in Chart 1, red bars are smaller and green bars are bigger for the bonus cap and malus groups, compared to the proportional bonus group in Task 1. This result might suggest that such restrictions on bonuses could curb excessive risk-taking. 

Third, however, when the bonus payment was made conditional on hitting a high performance target, all bonus groups became more likely to invest in high risk assets: red bars in Chart 1 become significantly larger for all bonus groups, as they move from Task 1 to Task 3. In fact, the differences in risk choice between the bonus groups were no longer statistically significant.

In the effort task, we found that the bonus cap group was more likely to shirk than other groups: nearly 20% of the bonus group did not attempt even a single calculation, compared to only 9% for the proportional bonus group and 7% for the malus group. So while we found no evidence that malus reduced incentive for effort, there was some evidence that a bonus cap might. This result is perhaps unsurprising: bonus cap limits the potential reward from effort, so it may be rational for individuals to ‘shirk’ when effort is costly. Malus, by contrast, does not limit the potential reward from effort. 

What do these results imply for remuneration regulations?

There are obvious limitations in drawing direct inferences about the effectiveness of the actual bonus regulations based on a lab experiment, where participants played investment games with small amounts of money at stake. That said, our experiment provides clear evidence that bonus structures affect individuals’ incentives, and consequently their risk and effort choices. 

Our study has two main policy implications. First, to curtail excessive risk-taking, it may not be sufficient for regulators to check banks’ compliance with the existing remuneration regulations, but it may also be necessary to examine risk-taking incentives embedded within the entire pay structure. Regulators need to be tuned to the possibility that features such as absolute and relative performance targets could be used to fuel bank executives’ risk-taking incentives, even in the presence of pay regulations.

Second, the effect of bonus cap on risk is far from clear. On one hand, we found some evidence that a bonus cap can mitigate risk-taking, though this might be undone if banks make the bonus conditional on a hitting a high performance target. On the other hand, our study also points to the possibility that a bonus cap could have the undesired side effect of reducing the project search effort. If a bonus cap indeed diminishes incentives for bankers to screen projects carefully, then it could potentially contribute to increasing, rather than reducing the riskiness of banks. 

Qun Harris works in the Strategy and Implementation Division, Analise Mercieca works in the Stress Testing Strategy Division, Emma Soane works at the London School of Economics and Misa Tanaka works in the Research Hub Division.

If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.