Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

Stated preference methods are increasingly used to measure household demand for a vaccine in developing countries. However, most studies interview only a single person within the household, thus assuming that the household follows a common preference model of intrahousehold resource allocation. As policy makers, in order to make efficient decisions, it is necessary to know whose opinion within the household counts, and the possible magnitude and direction of the error if we count the opinion of the wrong person.

Theoretical developments and empirical evidence have rejected the common preference model in favor of household bargaining. Although multiple interviews per household can identify the differences in household member preference and examine evidence for bargaining, aggregating preferences is difficult.

In this paper, I examine the differences in husband and wife preferences and evaluate the utility of a short private discussion between husbands and wives in aggregating household preferences. Husbands and wives from 422 households in Navi-Mumbai, India, were interviewed separately first and jointly thereafter in a stated preference framework to obtain their household willingness to pay for a malaria vaccine. This protocol is the first of its kind in a developing or an industrialized nation for any kind of stated preference survey. The fieldwork was conducted in the Spring of 2006.

When interviewed separately, husbands' and wives' stated demand for vaccines differed significantly at lower prices, where respondents had the freedom of budget space. The short private discussion enabled a majority of husbands and wives to reduce differences in their stated demand, with many couples choosing to purchase a vaccine either for all members of the household or for none. Respondents tended to be especially accommodative of their spouse's wishes at lower prices.

Both husband and wives changed their opinion during the joint interview indicating possibility of behavioral aggregation. Wives were about 40% more likely to change their opinion and acquiesce to their husband's wishes. Wives who had some source of income were about 30% less likely to change their opinion indicating that they had higher autonomy in decision-making. Significantly different coefficients for husband and wife non-labor income indicate some evidence of bargaining.

Tests of equivalence of coefficients for income between husbands and wives responses showed that the coefficient for non-labor income for husbands was significantly different from that for wives at 10% or less. This implies that marginal funds in hands of husbands could result in different consumption patterns as compared to wives. This is evidence in favor of rejection of the common preference model and in favor of non-pooling. The mean household willingness to pay (and 95% confidence interval) for a 95% effective malaria vaccine was 13.9 (13.3-14.5) for husbands and 11.9 (10.9-12.9) for wives when interviewed separately, and was 11.7 (10.9-12.4) and 11.2 (10.5-11.9) when interviewed jointly.

A montecarlo simulation of the conditions under which surveys are generally conducted?husbands or wives selected randomly from the household for interview? found that a traditional survey is likely to underestimate demand at lower prices and hence the associated welfare benefits.