How do consumers switch between close substitutes when price variation is small? The case of cigarette types

Abstract

Past empirical studies report ambiguous results regarding the magnitude and significance of substitution between different types of smoking tobacco. Since all types of tobacco contain nicotine this is quite surprising. Using a 20-year rotating panel data set of Norwegian households and a multinomial logit model, we find evidence that consumers switch between tobacco types: first, estimated price effects on choice probabilities have mostly expected signs, albeit their statistical significance vary across different metrics, second, household characteristics affect tobacco composition significantly. These findings suggest that consumers’ choices are ‘locked’ when the relative price variation is small, as has been the case in most of the data period, but that larger changes could induce large-scale switching between tobacco types. Our conjecture is that there is a latent potential for switching, which will become manifest if prices change sufficiently. Similar considerations are likely to have relevance for other close substitutes.