Abstract
Building on Kihlstrom and Mirman (Journal of Economic Theory, 8(3), 361–388, 1974)’s formulation of risk aversion in the case of multidimensional utility functions, we study the effect of risk aversion on optimal behavior in a general consumer’s maximization problem under uncertainty. We completely characterize the relationship between changes in risk aversion and classical demand theory. We show that the effect of risk aversion on optimal behavior depends on the income and substitution effects. Moreover, the effect of risk aversion is determined not by the riskiness of the risky good, but rather the riskiness of the utility gamble associated with each decision.
Similar content being viewed by others
Notes
See Diamond and Stiglitz (1974) for an analogous approach using differentiability.
Note that the majority of the literature on risk aversion has been set in the context of the static portfolio problem, which is equivalent to the price of the risky good being random.
Diamond and Stiglitz (1974) points out the relationship between increasing risk aversion and preferences ordering over utility gambles.
In the three cases of randomness, an increase in risk aversion changes the weights attached to the marginal utilities corresponding to different outcomes of the uncertainty, while the income and substitution effects order the marginal utilities. For instance, with a normal good and uncertainty in income, the marginal utility is increasing in income, so that a lower income means a lower marginal utility and thus an increase in risk aversion adds more weight to the marginal utilities associated with lower incomes.
For \(I,I^{\prime } \in \{\underline {I},\overline {I} \}\), let \(y_{I}(x_{I^{\prime }}) \equiv (I - P_{x} x_{I^{\prime }})/P_{y}\).
This is a simplified version of the general proof of Kraus-Katz as used in the Appendix.
To generate the graph, we set P x = P y = 1 and I ∈ {2, 5}.
If \(x < \underline {I}/2\), then the outcome is strictly worse than choosing \(x = \underline {I}/2\), while, if \(x > \overline {I}/2\), then the outcome is strictly worse than choosing \(x=\overline {I}/2\). In this case, the optimal solution has both x and y positive, i.e., there is no corner solution in which either x = 0 or y = 0. However, on the interval \(\underline {I}/2 \leq x \leq \overline {I}/2\), there can be corner solutions, when \(x = \underline {I}/2\) or \(x = \overline {I}/2\), which correspond to the most risk-averse and the most risk-loving choices respectively.
Note that if the initial choice is \(x=\underline {I}/2\), then the consumer is making the most risk-averse choice. Therefore a more risk-averse transformation cannot reduce the level of x.
References
Arrow, K.J. (1965). Aspects of the theory of risk-bearing. Yrjo Jahnssonin Saatio.
Diamond, P.A., & Stiglitz, J.E. (1974). Increases in risk and in risk aversion. Journal of Economic Theory, 8(3), 337–360.
Katz, E. (1981). A note on a comparative statics theorem for choice under risk. Journal of Economic Theory, 25(2), 318–319.
Kihlstrom, R.E., & Mirman, L.J. (1974). Risk aversion with many commodities. Journal of Economic Theory, 8(3), 361–388.
Kraus, M. (1979). A comparative statics theorem for choice under risk. Journal of Economic Theory, 21(3), 510–517.
Pratt, J.W. (1964). Risk aversion in the small and in the large. Econometrica, 32(1–2), 122–136.
Ross, S.A. (1981). Some stronger measures of risk aversion in the small and the large with applications. Econometrica, 49(3), 621–638.
Acknowledgments
We are grateful to an Editor and an anonymous referee for their very helpful comments. We thank Elena Antoniadou for arousing our interest in the subject again and Toshihiko Mukoyama for helpful comments. A previous version of this paper entitled “Risk Aversion and Classical Demand Theory” was written with Hou Fei who has withdrawn his name from the project. We thank him for his contributions to the structure of the paper.
Author information
Authors and Affiliations
Corresponding author
Appendix: Proofs
Appendix: Proofs
We combine the proofs of Propositions 1, 2, and 3. We first state the following Lemma which combines Lemma 2 in Kraus (1979) and conditions stated in Katz (1981).
Kraus-Katz Lemma
Let \(x_{\tilde {\theta }}^{\ast } =\arg \max _{x} \mathbb {E}_{\tilde {\theta }} \varphi ( Z(x,\tilde {\theta }))\) such that \(\partial ^{2} \varphi (Z(x,\theta ))/\partial x^{2} \vert _{x=x^{\ast }_{\tilde {\theta }}}<0\). If \(\partial Z(x,\theta )/\partial \theta \vert _{x=x^{\ast }_{\tilde {\theta }}}>0\) and \(\partial ^{2} Z(x,\theta ) / \partial x \partial \theta \vert _{x=x^{\ast }_{\tilde {\theta }}} < 0 (> 0)\), then \(x^{\ast }_{\tilde {\theta }}\) increases (decreases) along with an increase in risk aversion. Alternatively, if \(\partial Z(x,\theta )/\partial \theta \vert _{x=x^{\ast }_{\tilde {\theta }}}<0\) and \(\partial ^{2} Z(x,\theta ) / \partial x \partial \theta \vert _{x=x^{\ast }_{\tilde {\theta }}} > 0 (< 0)\), then \(x^{\ast }_{\tilde {\theta }}\) increases (decreases) along with an increase in risk aversion.
In the consumer’s problem, Z(x, θ) ≡ U(x, (I − P x x) / P y ), θ ∈ {I, P x , P y }. Suppose first that income is random as in Proposition 1, i.e., \(\tilde {\theta } \equiv \tilde {I}\). Then, \(\frac {\partial U( x,(I-P_{x}x)/P_{y})}{\partial I} > 0\), and \(\frac {\partial ^{2} U( x,(I-P_{x}x)/P_{y})}{\partial x\partial I} = I{}E_{I} \equiv U_{12} \cdot \frac {x}{P_{y}} - U_{22} \cdot \frac {P_{x}x}{P_{y}^{2}}\) is of the same sign as the income effect related to a change in income. By Kraus-Katz Lemma, \(x_{\tilde {I}}^{\ast }\) decreases along with an increase in risk aversion when the sure good is normal, i.e., \(I{}E_{I} \vert _{x=x^{\ast }_{\tilde {I}}} > 0\). Suppose next that the price of the sure good is random as in Proposition 2, i.e., \(\tilde {\theta } \equiv \tilde {P}_{x}\). Then, \(\frac {\partial U( x,(I-P_{x}x)/P_{y})}{\partial P_{x}} < 0\), and \(\frac {\partial ^{2} U( x,(I-P_{x}x)/P_{y})}{\partial x \partial P_{x}} = I{}E_{P_{x}} + S E_{P_{x}}\), where \(I{}E_{P_{x}} \equiv -\left [ U_{12} \cdot \frac {x}{P_{y}}-U_{22} \cdot \frac {P_{x}x}{P_{y}^{2}} \right ]\) and \(S E_{P_{x}} \equiv -\frac {U_{2}}{P_{y}} < 0\) are of the same sign as the income and substitution effects, respectively, related to a change in the price of the sure good. By Kraus-Katz Lemma, \(x_{\tilde {P}_{x}}^{\ast }\) decreases along with an increase in risk aversion when the sure good is normal, i.e., \(I{}E_{P_{x}} \vert _{x=x^{\ast }_{\tilde {P}_{x}}} < 0\). The effect of risk aversion for an inferior sure good depends on the relative strength of the income and substitution effects. Suppose finally that the price of the risky good is random as in Proposition 3, i.e., \(\tilde {\theta } \equiv \tilde {P}_{y}\). Then, \(\frac {\partial U( x,(I-P_{x}x)/P_{y})}{\partial P_{y}} < 0\), and \(\frac {\partial ^{2} U( x,(I-P_{x}x)/P_{y})}{\partial x \partial P_{y}} = I{}E_{P_{y}} + S E_{P_{y}}\) where \(I{}E_{P_{y}} \equiv -\left [ U_{12}\cdot \frac {x}{P_{y}}-U_{22}\cdot \frac {P_{x}}{P_{y}}\right ] \frac { I-P_{x}x}{P_{y}^{2}}\) and \(S E_{P_{y}} \equiv \frac {U_{2}P_{x}}{P_{y}^{2}}>0\) are of the same sign as the income and substitution effects, respectively, related to a change in the price of the risky good. By Kraus-Katz Lemma, \(x_{\tilde {P}_{y}}^{\ast }\) increases along with an increase in risk aversion when the good is inferior, i.e., \(I{}E_{P_{y}} \vert _{x=x^{\ast }_{\tilde {P}_{y}}} < 0\). The effect of risk aversion for a normal sure good depends on the relative strength of the income and substitution effects.
Rights and permissions
About this article
Cite this article
Mirman, L.J., Santugini, M. On risk aversion, classical demand theory, and KM preferences. J Risk Uncertain 48, 51–66 (2014). https://doi.org/10.1007/s11166-014-9182-3
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11166-014-9182-3