incentives, utilityWhen giving incentives to an agent, whats the difference (in economic terms) between offering the agent an award versus a punishment? What I’m thinking about is utility.
Say that the agents current utility is 10, the reward is +1 and the punishment is -1. What will be the better incentive? Has there been any studies on this?
I’ve learned that most agents are risk averse - and, i would conclude that they are thereby not willing to risk losing current utlity stock. But is this true? What arguments or studies are there?
In terms of rational economic models, it should not matter. Say, for example, you give an employee a “base salary” of $100, and a “reward” of $50 for meeting some particular objective, it should be equivalent to offering a base of $150, and a “punishment” of $50 for failing to meet that objective. This is regardless of the probabilities over the two events, so pure “risk aversion” would not change the analysis.
On the other hand, if you read some of the research of Daniel Kahneman, he finds that people behave in a manor that is strongly “loss averse”. That is, essentially, if I understand your phrasing, people will be better incentivized by a punishment than a reward, even if economically, the two options should be functionally equivalent.
I’m going to answer the question in the terms it was asked, comparing 10 plus a possible reward of +1 and 10 with a possible punishment of -1.
Risk-aversion is not very relevant here I think. Risk, in economics, refers equally to the upside and the downside. In this sense (possibly different from the common usage of the term) a possible -1 is no more risky than a possible +1. It is not a case of an agent choosing a risky option that might lead to +1 but might also lead to -1, and preferring a certain 10. The economic concept that is relevant, especially if 1 unit is large, is diminishing marginal utility. However, before applying this here one other point needs clarification.
No employer or other authority can directly award, or take away, utility. What they can perhaps do is award or withhold money or other benefits. This will almost certainly affect agents’ utility, but the precise effect depends on each individual (utility is often said to be subjective). So the practical question for an employer or authority is not about utility but about a monetary or other reward or punishment. And this is where diminishing marginal utility applies: if the monetary alternatives are say, 10,000 plus a possible reward of 1,000, and 10,000 minus a possible punishment of 1,000, most people’s utility will be more affected by the difference between 9,000 and 10,000 than by that between 10,000 and 11,000. There are many online sources on diminishing marginal utility: eg http://economicsconcepts.com/law_diminishing_marginal_utility.htm
Also relevant here are agents’ expectations. Incentives are often sequential or periodic rather than one-off. If an agent expects income of 10,000 and has planned his/her outgoings accordingly, they will have a strong reason to try to avoid -1,000. Equally, however, if in the past they have regularly performed well enough to achieve +1,000, they may have adjusted their outgoings so that they have a strong incentive to keep on achieving +1,000.
Finally, from personal experience the presentation of incentive systems is extremely important. Cultural and linguistic factors are likely to affect how systems are perceived. To me at least, “punishment” suggests a response to wrongdoing, and if an authority attempts to punish me my reaction is likely to be to show that I have not done wrong. If the issue is only under-performance, then some other terminology might be more appropriate, eg penalty. Employment law in the UK at least would not allow an employer to make a deduction against a contracted salary on grounds of under-performance: hence employers who wish to offer monetary incentives have little choice but to adopt the reward option (with any punishments in other forms, eg formal warnings and ultimately dismissal).
All content is licensed under CC BY-SA 3.0.