Take an agent with mean-variance utility over something that is uncertain: $$ U(x) = \mu_x^\theta - \sigma_x^\lambda $$

$A\in \{0,1\}$ happens if $U(x)>0$, and $x$ is a random variable

$$ A = \mathbf{1}\left(\mu_x^\theta - \sigma_x^\lambda >0\right) $$

Saha (1997) has shown that $\theta$ different from 1 corresponds to DARA or IARA preferences. If an agent isn't necessarily CARA, and $\mu = value - cost$, then the first term is $$ \mu_x^\theta = \left(E[value] - cost\right)^\theta $$ This assumes that cost isn't separable from value. In reality however, agents may not operate this way. The simplest example is a cash/credit constraint. You'd thus have $$ A = \mathbf{1}\left(\left(E[value] - cost\right)^\theta - \sigma_x^\lambda >0\right) s.t. cost < c $$

Alternatively, maybe cash expenditures are psychically painful. Like, even if I value something at \$5, the thought of not having that \$5 just bothers me. Maybe because of loss of option value, for example.

$$ A = \mathbf{1}\left(\left(E[value] - cost\right)^\theta - \sigma_x^\lambda - f(cost) >0\right) s.t. cost < c $$ where $f$ is weakly positive and monotonic.

My questions:

  1. Is there a specific term for this sort of separability in cost and value?
  2. Would it go away entirely if there were no credit constraint?
  3. And if I simply model $\left(value - cost\right)^\theta$ with a nonparametric $g(cost, value)$, would economists know what I am talking about if I justify it as represeting "non-classical behavior" or "deviations from the classical behavioral model?"

Your Answer

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

Browse other questions tagged or ask your own question.