Our 50-150 definition of the middle sector provides useful information about inequality in a country. A large middle sector, by this measure, means that a greater share of the total population is within reasonable distance of the median household income. A smaller middle class means that more households are at the extremes of the income distribution, most likely swelling the ranks of the disadvantaged. This section looks more closely at the income distributions in a selection of Latin American countries, in part inspired by the need for better information about prospects for social mobility.

If a substantial and economically healthy middle sector contributes to social welfare, social mobility becomes an important policy objective. Social mobility is often examined in terms of inter-generational mobility, comparing the socio-economic status of parents and children.17 Such mobility is the product of several components, ranging from inherited abilities and social context to environmental factors. The latter are shaped by the policies determining access to human-capital formation, such as public support for education at all its stages, as well as redistributive policies (such as tax and transfer schemes) that may influence access to higher education. These issues are covered in detail in the following chapters of this Outlook.

For all their detail, national household surveys tell us very little about social mobility. To examine the phenomenon properly we need panel data, generated by surveys that repeatedly gather information from the same set of households over many years. Such data would show disadvantaged households entering the middle sector and middle-sector households falling into the ranks of the disadvantaged, as well as providing information about how many middle-sector households retain that status over a given period.

Such panel data are available for Chile from

1996, 2001 and 2006 and studies of these show that there is considerable mobility both up and down – opportunity and risk are both evident.18 For example, 55% of households that were poor in 1996 were not poor in 2001; while 11% of households that were not poor in 1996 had fallen into poverty by 2001 (the poverty lines used in this analysis do not necessarily coincide with 50% of median household income, the threshold used in this Outlook). The data also reveal a relatively immobile group of poor households which seem to be excluded from opportunities for advancement.

Unfortunately such panel data are only rarely available. A promising alternative is retrospective data, derived from surveys which ask people about the socio-economic status of their parents. These provide information about inter-generational mobility at least.19

Simply comparing the size of the middle sector from one wave of a survey to the next is substantially less satisfactory, since it does not capture churning in the income distribution. This can be material and is certainly very important to the well-being of the individuals involved. If the middle sector grows from, say, 40% to 45% of the population between two household surveys, and at the same time the disadvantaged population drops by exactly 5 percentage points it is tempting – but false – to conclude that 5% of the population climbed out of disadvantage and into the middle sector. It may equally be the case that many middle-sector households fell into disadvantaged status and that many more disadvantaged households moved into the middle sector, or that there was substantial movement in both directions across the threshold separating the middle sector from the affluent. That said, such comparisons across time are readily calculated using available data and do enable some conclusions to be drawn.

Measures of mobility and resilience

disadvantaged are to the middle sector, and how "close" the middle sectors are to the lower threshold equal to 50% of median income. Precise measures of these notions of closeness are useful in two ways. They give a crude sense of the possibilities for social mobility and they illuminate the scale of intervention required by policy makers if they are to be effective.

We calculate two indicators of social mobility to test this: the "Disadvantaged Mobility-Potential Index" (DMP), and the "Middle Sector Resilience Index" (RES). The DMP measures the average distance of the income of disadvantaged people from the threshold of 50% of median income; it asks how "close" disadvantaged people are to entering the middle sector. DMP ranges in value between 0 and 1. A value near 1 implies a small average income shortfall from the threshold to the middle sector and so a greater potential for upward social mobility. Conversely, a value closer to 0 indicates that the average income shortfall among disadvantaged households is large.

RES, for its part, measures the mean distance above 50% of the median income of the incomes of those middle-sector households which earn less than 100% of the median income – what might be thought of as the "lower middle sector". RES is the mirror image of DMP in the sense that it provides a measure of the negative income shock that would be needed to push lower middle-sector households into disadvantaged status. Such shocks can take many forms, some of which are all too familiar to households in the developing world: things such as illness, accident, a death in the family, unemployment, or a natural disaster. RES again ranges from 0 to 1. A value close to 1 implies a lower risk of falling into disadvantaged status, or put differently, a greater resilience to staying in the middle sector. Detail on the definition and calculation of these indices is set out in Box 1.2.

Comparing several Latin American countries, Uruguay, with the largest middle sector in the region, exhibits the highest value of DMP (Figure 1.8). The Uruguayan disadvantaged, relative to the other countries depicted, are "closest" to crossing the threshold into the middle sector. It is perhaps surprising that Argentina, with its relatively large middle sector, has the lowest value of DMP. The implication is that the disadvantaged in Argentina, though less numerous than in other Latin American countries, are less able to move up into the middle sector. In this regard, the shape of Argentina’s income distribution most resembles Bolivia’s, though centred on a substantially higher median income.

Box 1.2 Mobility-potential indicators

The Disadvantaged Mobility-Potential Index (DMP) is calculated as follows. For a given country, first calculate the difference between a disadvantaged household’s income and 50% of the median income for that country. This is the shortfall between actual income and the minimum needed to be in the middle sector, on our 50-150 definition. Second, sum these income shortfalls over all disadvantaged households. Third, divide this aggregate shortfall by the total income that all disadvantaged households would earn if they each earned exactly 50% of the median income. Expressed algebraically the formula is: 

where: M1 = number of people in the disadvantaged group (income less than 50% of the median); ym = median income; yi = income of the ith household; wi= weights.
DMP is a variant of standard poverty-gap indices, which seek not only to measure the incidence of poverty, but also its depth. The DMP index can be interpreted as the average distance between disadvantaged households and the middle sector.20
The Middle Sector Resilience Index (RES) measures the mean distance between the incomes of those middle-sector households earning less than the median income and 50% of the median income. The following formula is used:

where: M2 = number of people in the lower middle-sector group (income between 50% and 100% of the median); ym = median income; yi = income of the ith household; wi= weights.
It is straightforward to construct in the same fashion an index of the ease with which middle-sector households with incomes above the median income – the upper middle sector – can move into the ranks of the affluent. A Middle Sector Mobility-Potential Index (MSMP) can be calculated according to the formula:

where: M3 = number of people in the upper middle-sector group (income between 100% and 150% of the median income); ym = median income; yi = income of the ith household; wi= weights.
The closer the value of MSMP to 1, the smaller the average income shortfall from the lower threshold of the affluent category, and the higher the potential for the upper middle-sector to move up into the ranks of the affluent.
Finally, the Middle Sector Cohesiveness Index (COH) is defined as the mean distance of the middle sector from the median income as a proportion of the median income. The mean is taken over the whole middle-sector population, according to the following formula:

where: M4 = number of people in the middle sector (income between 50% and 150% of median); ym = median income; yi = income of the ith household; wi= weights.
COH is a rough measure of the spread of middle-sector incomes. A value close to 1 implies incomes are clustered near the median income and, therefore, greater cohesiveness of the middle sector.
See Castellani and Parent (2010) for more details on all of these measures and an overview of the evolution of inter-category mobility over time.

Figure 1.8. Indicators of social-mobility potential in Latin America (household level, 2006)

Figure 1.8. Indicators of social-mobility potential in Latin America (household level, 2006)

Uruguay’s middle sector is relatively resilient to the risk of falling into disadvantaged status, with a value of RES near 0.5 (Figure 1.8, top right-hand panel). What is perhaps more surprising is that Chile’s lower middle sector is the least resilient among the countries surveyed. This may reflect Chile’s remarkable success in reducing poverty over the last two decades: as a result, there are disproportionately many lower middle-sector households just over the 50% of median income threshold, and therefore close on our measure to falling back into disadvantaged status.

Argentina, Chile, Costa Rica and Mexico, 1996-2006

This section looks at how the size of the middle sector and the indices of potential mobility have developed over time in four countries. These countries have been chosen both because they have available the necessary longitudinal household-survey data, and because of the variety of stories that their experiences tell (Figure 1.9).

Figure 1.9. Changes over time in the middle sectors: size and mobility potential in Argentina

Figure 1.9. Changes over time in the middle sectors: size and mobility potential in Argentina

Figure 1.9. Mexico

Figure 1.9. Mexico

The data show a substantial retrenchment for the Argentinean middle sector. Between 1996 and 2006, the middle sector there shrank by almost 20%. At the same time, the disadvantaged population grew while the affluent stratum remained unaffected. Unstable economic performance over the decade – most notably the economic crisis of 2001 – hit lower-income groups disproportionally and dragged down the indices of potential social mobility. Since 2003, conditions have been improving for the disadvantaged. The middle sector on the other hand still looks immobile based on its index levels, historically or when compared to other countries.

The experience of Chile contrasts sharply. The middle sector there is stable in size over the period. This stability extends also to the indices of potential social mobility which change little over the years for which survey data are available.

Costa Rica exhibits progress on reduction of the size of the disadvantaged population and growth of the middle sector until 2007. Since then, however, the disadvantaged proportion has surged and indices of potential social mobility fallen. Both are linked to poorer economic performance with higher inflation and lower growth. The resilience of the lower middle sector has partially recovered in recent years suggesting less vulnerability to falling into disadvantaged status.

Indicators in Mexico picked up following the crisis at the end of the 1990s. Nevertheless, unsatisfactory economic performance since has pushed some people from the middle sector back into disadvantaged status. The middle sector has shrunk and disadvantaged households are displaying lower potential mobility.