Inequality, Consumption, and Structural Change

Inequality is perhaps the predominant economic fact of the world today. Over the past four decades, wealth concentrations among the top fraction of society have skyrocketed, even as median wages have stagnated. In light of these trends, numerous studies have sought to determine the effects of inequality on different aspects of the economy, from the overall rate of growth to the amount of investment to the development of human capital. These studies have shed light on important consequences of rising income inequality, but on some level they skirt around its true impact. They fail to emphasize that the income distribution is a fundamental economic parameter, one that directly determines what products get made, which firms are profitable, and what jobs are created. From the perspective of firms, the distribution of income—independent of consumer preferences—is a key determinant of demand for their products.

In the basic economic model, each individual consumer is believed to purchase the combination of products that maximizes her utility subject to her budget constraint. Individual consumption, then, depends on both preferences and resources. At the aggregate level, overall demand for products depends both on the aggregate preferences of consumers and on the distribution of resources among them. Importantly, while economists and sociologists debate the extent to which preferences are exogenous, the income distribution is clearly endogenous to policy decisions and past economic choices.

The extent to which aggregate demand varies with the income distribution is an empirical question. It depends on the correlation of preferences across consumers and the extent to which people change their consumption as they grow poorer or wealthier. In this paper I estimate these parameters, in ways that allow me to simulate the effect of different income distributions on overall demand for specific product categories. Using data from the US Consumer Expenditure Survey I first document that consumption patterns vary dramatically with income and with family type. I use cluster identification techniques to identify types of consumer based on observed spending patterns and demographic characteristics (families with young children tend to buy different products than do empty nesters, for example). I then use these types to simulate what demand for various products would be like under alternative income distributions.