Scale gains in household consumption and their modeling implications in poverty and distribution analyses

In order to perform income inequality, poverty and welfare analyses, splitting households in different groups is a mandatory approach. Depending on the research purposes, groups can be classified according to income, race, region, and so on (Bourguignon and Pereira da Silva, 2008; Combet et al, 2010; Mathur and Morris, 2014). Ranking households according to their level of income is a standard approach, but the varying profiles that can be seen along the spectrum of the population must be taken into account. For example, the size of the household and the average age of its members generally change in accordance to income. In this sense, modeling choices in economic analyses may lead to very different results, and hence to distinct policy implications. This kind of analysis often considers per capita expenditures or per capita income as indicators. Nonetheless, such approach fails at taking into account potential economies of scale that may arise from cohabitation. Households' needs grow with each additional member, but generally not in a proportional way: people living in the same household may share, to a certain extent, appliances (washing machine, television set, refrigerator, etc.), furniture, as well as room lighting and heating or cooling. Equivalence scales adjust the incomes of households in a way that recognizes differences in the needs of individuals and the economies that flow from sharing resources. The variables that are usually taken into account are the size of the household and the age of its members (OECD, 2013). In this paper, we will compare the profiles of household expenditure by living standard deciles (that is, using equivalence scales ) and per capita income deciles for Brazil and France. The datasets use microdata from national household budget surveys, namely the 2008-2009 POF for Brazil and the 2006 BdF for France. These profiles will be thoroughly analyzed for each of these two countries and then compared. We will also assess the indices of concentration of income and of the various categories of consumption using the two approaches. We will discuss the potential modeling and policy implications of using the two approaches in poverty and income distribution analyses, especially while performing carbon-pricing simulations. Carbon taxes are generally related to energy consumption profiles, which vary according to the household income and possibility of scale gains. Thus, modeling choices may bring about different levels of progressivity, that is, the extent to which the burden of the tax falls in the richer and poorer strata of the population.