Behavior towards health risks: An empirical study using the “Mad Cow” crisis as an experiment

Abstract

The paper exploits the “Mad Cow” crisis as a natural experiment to gain knowledge on the behavioral effect of new health information. The analysis uses a detailed data set following a sample of households through the crisis. The paper disentangles the effect of non-separable preferences across time from the effect of previous exposure. It shows that new health information interacts in a non-monotonic way with disease susceptibility. Individuals at low or high risk of infection do not respond to new health information. The results show that individual behavior partly offsets the effect of new health information.

Appendix

The first order condition of model 1 is:

$$u_1-pu_2+\beta\pi_1 V+\beta\pi V_1=0$$

where u _i and V _i denote the partial derivative of the utility function and second period indirect utility function with respect to the ith argument. First differentiating this expression gives:

$$\Delta c_B [u_{11}-2pu_{12}+\beta V\pi_{11}+2\beta\pi_1 V_1+ \beta\pi V_{11}]+ \Delta y [u_{12}-pu_{22}]+\Delta p [-c u_{12}-u_2+pc u_{22}]\!+\!\Delta S \left[u_{13}-p u_{23}+\beta\pi_{11}V+2\beta\pi_1 V_1+\beta\pi V_{11}\right]+\Delta \kappa \beta\left[V \pi_{12}+\pi_2 V_1\right]=0$$

which can be written more compactly as:

$$\tilde{A}_B \Delta c_B+\tilde{A}_y \Delta y+\tilde{A}_p \Delta p+\tilde{A}_S \Delta S+ \tilde{A}_\kappa \Delta \kappa=0$$

so that:

$$\Delta c_B=A_p \Delta p+A_y \Delta y + A_S \Delta S+A_\kappa \Delta \kappa$$

Standard restrictions on the shape of the utility function imply that u _i  > 0, u _ii  < 0, V _i  > 0, V _ii  < 0. Moreover, the definition of the survival probability implies that $\partial \pi(S,\kappa)/\partial S=\pi_1\le 0$ and that $\partial \pi(S,\kappa)/\partial \kappa=\pi_2\le 0$, if individuals perceive that nvCJD is a threat to life.

If u ₁₂ ≥ 0 (beef and other meat products are complements) and the relationship between survival and beef consumption is concave (π ₁₁ ≤ 0, then $\tilde{A}_B\le 0$, $\tilde{A}_p\le 0$ and $\tilde{A}_y \ge 0$. The effect of health information on consumption of beef is equal to:

$$A_\kappa=\left(-\frac{\beta}{\tilde{A}_0}\right) (V\pi_{12}+\pi_2 V_1)$$