Discussion Paper No. 164 of 2014 on Alcohol Consumption and Healthcare Expenditure in Kenya

Abstract

This study investigates the factors influencing alcohol consumption and the effects of alcohol consumption on healthcare expenditure in Kenya using data from the 2005 Kenya Integrated Household Budget Survey (KIHBS). The analysis made use of the Cragg’s double hurdle model, also known as two-part model when applied to cross-sectional data. The model splits the households’ decision into two: the participation and the consumption decision. Participation decision is examined in the context of a logistic model, while the consumption decision given the decision to consume is analyzed with a log-linear regression. Descriptive results show that there are differences in alcohol consumption across different regions in Kenya, in terms of the number of households and the total expenditure on alcohol. Estimation results show that the participation and consumption decisions are significantly influenced by social, economic and demographic characteristics. In particular, presence of children, household head being female, and head being married reduces the likelihood of alcohol participation, while increase in age of the household head, tobacco consumption and increase in income increases the likelihood to drink alcohol. However, the increase in age has an inverted u-shape with a point where further increase in age reduces the probability of participation. On the other hand, female headed households, presence of children aged 0-14 years, and increase in household size by a member reduces alcohol expenditure by 30, 20 and 5 per cent, respectively. Households using tobacco spend 15 per cent more on alcohol while an increase in income by 1 per cent increases expenditure on alcohol by 0.73 per cent. Effects of alcohol consumption on healthcare expenditure was indeterminate, since it had different signs in the participation and the consumption decision. The study recommends that alcohol control programmes should target the groups that are more vulnerable, specifically the relatively young heads, mostly male who reside in urban areas and are smokers. There is also need to incorporate gender in alcohol control planning, and develop regional-specific alcohol control campaigns. Twin strategies for goods that are consumed together should be developed.