Diminishing marginal utility is a concept only in cardinal utility theory rather than ordinal utility theory. As diminishing marginal utility implies a concave shape of the utility function, does it necessarily mean that agents with diminishing marginal utility must be risk-averse (also has a function with a concave shape)?

Does cardinal utility theory assume the law of diminishing marginal utility in all of this theory? Could you have increasing marginal utility within cardinal utility theory?

  • $\begingroup$ I don't know if I fully understand the second part of your question. Are you asking if can have utility function with increasing marginal utility? I don't see why you couldn't theoretically have an agent with $e^{\beta \cdot W},\; \beta>0$ preferences. And quadratic utility preferences (which are common and standard in many applications) generally imply a region where marginal utility is increasing (though it may be at negative wealth levels. $\endgroup$ – BKay Jan 28 at 19:28
  • $\begingroup$ Thanks, so it is possible to have the increasing marginal utility in cardinal theory. $\endgroup$ – Aeeh Jan 28 at 19:41

Mas-Colell, Whinston, and Green Proposition 6.C.1 (p. 187) says that for an expected utility maximizer with a Bernoulli utility function $u(\cdot)$ on amounts of money, saying that "the decision maker is risk averse" is equivalent to "$u(\cdot)$ is concave".

However, these other assumptions matter. A concave utility function isn't enough. Diminishing marginal utility in amounts of money doesn't have to imply risk aversion. For example, with gamble $(F)$ taking two values $W_L,\: P(W_L) = \theta$ and $W_H>W_L,\:P(W_H) = 1 - \theta$, and a utility function $u{}$ with diminishing marginal utility in wealth, consider this person's welfare function over risky lotteries $(V)$: $$V(F, u(), \theta) = u(W_H)$$
This person is not risk averse. They love risk (for example, they prefer a risky lottery to the expected value of that lottery). But, in situations without risk, they have diminishing marginal utility of wealth.

| improve this answer | |
  • 1
    $\begingroup$ "these other assumptions" There are no assumptions in your answer that "these" could refer to. $\endgroup$ – Giskard Jan 28 at 16:31
  • 1
    $\begingroup$ The other assumptions are that the agent is 1) an expected utility maximizer and 2) has a Bernoulli utility function. $\endgroup$ – BKay Jan 28 at 17:09
  • $\begingroup$ Yes, of course. Sorry. $\endgroup$ – Giskard Jan 28 at 17:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.