In order to assess the economic characteristics of the middle sectors of Latin American and Caribbean countries and compare these sectors over time and across countries, we need a precise definition. Briefly, we seek a measure with three characteristics. First, it must be based on data that are readily available for most countries in the region. Second, it should be a measure that allows us to compare countries at somewhat different levels of economic development, given that Latin America and the Caribbean countries span a considerable range of such levels; moreover, it would be useful to be able to compare Latin American countries with OECD countries, where development levels are higher on average. Third, our measure of the size of the middle sectors should be related in some consistent way to inequality in the economy: a larger middle sector should signal relatively lower inequality.

The key variable for identifying the middle sectors is income per head, which is taken from household surveys carried out in many Latin American countries. Income per head is computed on the basis of the household’s total income, adjusted for the number of household members.5 Income per head is converted to United States dollars and is further adjusted for differences in international prices – purchasing power parity – to allow comparison between one country and another. The household survey data sets furthermore contain information on the economic characteristics of middle-sector households that is useful for elaborating a statistical portrait of this group later in this chapter.

The rule for determining a middle-sector income level can be relative or absolute. Thus, many recent studies have defined middle-sector income levels in absolute terms: for example, the World Bank’s Martin Ravallion assigns households to the middle sector if their daily income per head is between USD 2 and USD 13 (in 2005 dollars on a purchasing-power parity basis).6 An influential study by Abhijit Banerjee and Esther Duflo of the Massachusetts Institute of Technology, meanwhile, defines the limit of the middle sectors at USD 2 and USD 10 per day (roughly USD 800 to USD 3 600 per year). The lower bound of the range in both studies – two dollars a day – is the standard international poverty line. Absolute definitions like these are transparent and easy to understand, but they make it difficult to compare the size of the middle sectors across countries with different levels of economic development. Thus, using either the Ravallion or Banerjee-Duflo definitions, there will be sizeable middle sectors in China and India, relatively smaller middle sectors in upper middle-income economies like those of many Latin American countries, and virtually all households in OECD economies will be in the income category above that of the middle sectors.

For these reasons, this Outlook’s definition of the middle sector will be anchored at the median level of income per head – which varies from one country to another. By definition, there are exactly as many households ranked below the median household as ranked above. Median household income therefore does not suffer the same potential distortions as the mean, which can be pushed upwards by a small number of very high-income households. The middle sectors can then be defined as the group within some specified distance of the median.7 Using a relative definition, of course, means that a Honduran with income close to the Honduran median household would be classified as belonging to the Honduran middle sector, but the same level of income would likely be too low to qualify for the Italian middle sector.

We consider the middle sectors to be those households with income per head between 50% and 150% of the median income. The 50% cut-off is frequently used by researchers as an internationally comparable poverty or low-income line in empirical studies of poverty and income distribution. A major OECD study on income inequality followed this practice and OECD statistics routinely use 50% of median income as a poverty line for OECD countries.8 This is reasonable given that the middle sectors are meant to comprise households not on the lowest rung of the ladder of income distribution. Given that the middle sectors are not meant to include the relatively well-off, a symmetrical upper bound of 150% of median income is straightforward.

Finally, a definition of the middle sectors anchored around median income in this way varies with income inequality in a way that other relative definitions do not. The Easterly study discussed at the start of this chapter, for example, defines the middle sectors as those households in the second, third and fourth income quintiles. Under the Easterly definition, the middle sector will invariably comprise 60% of the population. Our definition, in contrast, has the attractive property that the size of the middle sector varies from one country to another, and in particular varies with income inequality.

To summarise, we formulate a workable relative definition of the middle sectors:

  • The middle sector comprises those households with income between 50% and 150% of median household income. Those households whose income per head lies below the 50% threshold will be referred to as "disadvantaged"; those whose income lies above the 150% threshold will be referred to as "affluent."

This is the definition that will be used in this Outlook.9 For brevity we refer to it as the "50-150 definition".

Figure 1.1 illustrates the relative sizes of the middle sectors, the disadvantaged and the affluent in selected countries.

The figures are based on household-survey data using 2006 as the base year and use total household income adjusted for household size. The countries examined (with survey years in parentheses) are Argentina (2006), Bolivia (2005), Brazil (2006) Chile (2006), Colombia (2008) Costa Rica (2006), Ecuador (2006), Mexico (2006), Peru (2006) and Uruguay (2005). Between them these ten countries cover more than 80% of the population of Latin America and the Caribbean.10 Italy is included in the figure for purposes of comparison. The spectrum ranges from Uruguay (in which the size of the middle sectors is only 10 percentage points below Italy), through Mexico and Chile, with middle sectors around 50% of the population, to Bolivia and Colombia with middle sectors equal to just over a third of the population.

Figure 1.1. Size of the middle sectors in Latin America and Italy (as percentage of total households, 2006)

Figure 1.1. Size of the middle sectors in Latin America and Italy (as percentage of total households, 2006)