When to pay the performance incentive? Caveats of loss-aversion phenomenon
Rewards for performance are commonly used strategies to motivate employees to work harder. The emerging research in behavioural science is showing us innovative ways of designing these incentive systems. One such system based on the principle of loss aversion is gaining popularity. World Bank in its recent ‘Mind, Society and Behaviour’ report discusses the loss aversion phenomenon. Prof Raj Chetty also quotes it in a podcast on ‘Improving public policy through behavioural economics’.
The principle of loss aversion is based on this simple finding that people are sensitive to losses in income than gains in income. It is an income gain if the employees are paid at the end of the year based on their performance. Instead, if the money is given upfront at the beginning of the year and it is taken back at the end, on not meeting performance metrics, it is an income loss. An experiment conducted in Chicago schools by Prof List and others of the University of Chicago show that teachers who were subject to income loss design of incentives exerted more efforts and also resulted in higher student scores as compared to the teachers who were subjected to ‘income gain’ design. However, there are certain limitations to the incentive mechanisms designed on the principle of loss aversion depending on the context. This article, in particular, tries to point out five such differences in a context which can possibly change the nature of the results.
First, nature of work. Daniel Pink in his book ‘Drive: The surprising truth about what motivates us’ describes two categories of tasks; algorithmic and heuristic. Algorithmic tasks are those where you follow a set of defined instructions to solve the problem, like working as a grocery clerk. Heuristic tasks are those which require innovation, experimentation with multiple possibilities and devising a novel solution, like designing an ad campaign. Pink demonstrates through various examples and corroborating research that incentives work very well in former scenarios (algorithmic tasks) and might back fire in case of heuristic tasks. The argument being, the pressure to perform, which is likely to be highly especially if the incentive is designed as per loss aversion principle, narrows down the focus of people, limiting their world view and making them less risk averse. So, this type of design may not be desirable in highly creative environments.
Second, the time frame of incentive: Is the performance evaluated monthly, quarterly, annually or any other time frame? The perception of loss produces a huge internal drive to perform, not the positive drive but the drive associated with fear. If the performance is reviewed at short intervals of time, it may narrow down the focus of people and discourage cooperation between teams. News reports suggest that after the introduction of quarterly rating systems in Yahoo, accompanied with the action on employees receiving lower ratings, the cooperation among team members has reduced and people stopped taking up ambitious projects which will yield results only in long term. The fear of losing out something on nonperforming is common to both these designs. So, this type of design with performance evaluated at short intervals of time may not be desirable in environments where cooperation among teams is needed.
Third, incentive as fine. The incentive systems based on loss aversion as discussed in the beginning, are also a form of a penalty. The employees pay ‘fine’ for not meeting certain standards. This hypothesis assumes that this deterrence will make people work but this needn’t be the case always. Uric Gneezy and Aldo Rustichini demonstrated an exception to this in their famous paper ‘A fine is a price’. This experiment studies the effect of imposing fines on delay timings of parents who come late to day care centers to pick up their children. It is observed that after the imposition of fine, the delays have increased. The reason being, parents started perceiving the small penalty as a fine they are paying for coming late and hence felt entitled to come late.
There is a similar threat to the incentive systems based on loss aversion principle. If the incentive amount being given is above a basic salary (which is enough to satisfy their needs), and if the nonperformers are made to pay back money, in the end, it is a possibility that nonperformers can view this act of 'returning' money as paying the fine for their actions. They can then go on to justify their actions citing the penalty they are paying for the same. This isn’t a desirable situation since it takes out the last resort of ‘guilt’ from such employees.
Four, nature of performance metrics. The metrics to measure performance aren’t often as straight forward as test scores as in the case of example that we discussed in the beginning. It might not be possible to quantify the value created by people engaging in complex long-term projects, in the typical time frames of performance evaluation. For example, a mathematician trying to solve an age-old unproven theorem. In other cases, one might have to build a combination of metrics to evaluate the performance. In all such cases, if the standards designed are hard to achieve or if people perceive them as unfair, even the highly motivated people can lose steam in the long run, on having to pay the penalty even after working hard. This is harmful both to the employees and the organization.
Five, poor – good – excellent systems. An often missed out consideration is that the dynamics of incentive mechanisms are a function of the type of systems it is being incorporated in; poor, good and great systems. Hanushek et al explore an important question in their paper titled ‘Does school autonomy make sense everywhere?’ The authors analysed the test scores in international PISA tests of over one million students spanning across 42 countries. Their results suggest that ‘autonomy affects student achievement negatively in developing and low-performing countries, but positively in developed and high-performing countries’. There is an important lesson to be learnt from this about the dynamics of the relation between the policies and the type of systems.
If the system in consideration is extremely low performing, where there is near zero effort by the individuals, the incentives based on loss aversion principle may be effective because the marginal increase in productivity due to this is larger than the other effects owing to the lower base. If the system in consideration is a ‘good’ system and it is difficult to quantify the performance due to the nature of work, then this form of incentives may backfire due to its demotivating effects as discussed in point four above.
In conclusion, this isn’t to suggest or advocate against the incentives based on loss aversion principle. Its applicability depends heavily on context and one should appreciate its nuances while implementing it. One should also note that effects of incentives needn’t be the same in the short term and long term. It is essential to systematically track these effects and continuously keep iterating the incentive design and performance metrics before they backfire and cause repercussions.
Labels: Behavioural Economics