Journal Issue

BACK TO BASICS: Trading Places: Measuring Income Mobility

International Monetary Fund. External Relations Dept.
Published Date:
March 2004
  • ShareShare
Show Summary Details

HOW MUCH income mobility is there in today’s economies? A high degree of mobility would imply, in the words of Columbia University’s Jagdish Bhagwati, that capitalism’s “inequalities then become tolerable, not because the rich deny themselves self-indulgence but because the poor fancy that these prizes may come to them someday too.”

Much of the evidence on income mobility is for the United States. In the State of Working America, researchers at the Economic Policy Institute conclude that while the U.S. evidence “does not reveal a great deal of income mobility, the data do show that mobility exists and that families move up and down as their relative fortunes change.”

The basis for their conclusion is shown in the chart. Individuals are grouped into quintiles based on their family income in 1969; then their position is observed again in 1994. This provides “before” and “after” snapshots that can reveal whether they have traded places with others in the income distribution or stayed put. Keep in mind that if the rankings in 1994 bore no resemblance to incomes in 1969, the number that stayed put in each quintile of the income distribution would be 20 percent, while the remainder would be dispersed equally among the other quintiles. This would correspond to what economists label “perfect mobility” because it indicates that the deck of cards of people’s incomes has been so thoroughly reshuffled that people are placed in income quintiles as if by pure chance.

On the move?

There’s substantial mobility in the middle of the income distribution, less so at the extremes.

Citation: 41, 1; 10.5089/9781451953411.022.A010

Source: Economic Policy Institute, 2003, “The State of Working America, 2002/03” (Washington, D.C.), based on unpublished tabulations by Peter Gottschalk, Boston College.

How far is actual mobility from this state of perfect mobility? Quite far at the extremes, less so in the middle. About 40 percent of those who were in the bottom quintile of the income distribution in 1969 were still in that quintile 25 years later. Likewise, nearly 40 percent of those who were in the top quintile maintained their relative position at the top 25 years later. These numbers are twice as large as those that would be indicated by a state of perfect mobility. In the intermediate quintiles, mobility is much greater. Only 24 percent of those in the middle quintile stayed put, and those who moved out were just as likely to move up in the income distribution as to move down. The other two intermediate quintiles also come close to what would be expected if perfect mobility held.

How much income inequality is erased because people move around in the income distribution over time? Boston College’s Peter Gottschalk calculates that U.S. inequality—as measured by the gap in labor earnings at the 90th and the 10th percentiles—is reduced by a third when mobility is taken into account. He points out, however, that the extent of mobility does not appear to have changed over time, whereas income inequality in the United States has increased in the last two decades. Hence, the extent to which mobility counteracts the effects of inequality may be waning. Moreover, the extent of mobility does not differ much across the major industrial countries, which comes as a surprise because labor market institutions and tax systems vary widely across countries. Gottschalk concludes that “U.S. mobility rates resemble those of countries as different as France, Italy, and Sweden.”

Read the fine print

Evidence on income mobility is difficult to assemble. It requires a longitudinal panel data set—that is, one that tracks the same people over time (and records their income). Ongoing surveys, such as the University of Michigan’s Panel Survey of Income Dynamics (on which the chart is based) and tax returns, are a way to track people’s income. However, the measure of income that is most readily available from these data sets is labor earnings; more comprehensive measures of earnings, or measures of wealth, are either unavailable or very difficult to compute.

Even after a data set is assembled, choices about the design of a study can greatly affect the conclusions drawn. One important choice is whether to base the calculations on an individual’s own income or on his (or her) family income. Studies that use the former tend to find much higher upward mobility than that reported above, but as University of Chicago economist Kevin Murphy explains, “This isn’t your classic income mobility. This is a guy who works in the college bookstore and has a real job by his early thirties.” In contrast, inclusion of people who are close to retirement may bias the results toward finding downward mobility. Thus, the sample selection has to ensure that the observed mobility is not simply due to individuals’ life cycle of income.

Another decision is how many years should elapse between the before and after snapshots? If the two snapshots are taken after only a short interval of time, there’s a danger of picking up mobility arising from transitory fluctuations in people’s income rather than more permanent shifts. Most researchers therefore prefer to use a decade or the 25-year interval used in the chart.

Are incomes of fathers visited on their sons?
(percent)Son’s earnings


Father’s earningsBottom quartile33282217
Top quartile20212535
Evidence from a study of 400,000 father-son pairs. See Miles Corak and Andrew Heisz, 1998, “The Intergenerational Earnings and Income Mobility of Canadian Men: Evidence from Longitudinal Income Tax Data,” Statistics Canada, Research Paper No. 113.
Evidence from a study of 400,000 father-son pairs. See Miles Corak and Andrew Heisz, 1998, “The Intergenerational Earnings and Income Mobility of Canadian Men: Evidence from Longitudinal Income Tax Data,” Statistics Canada, Research Paper No. 113.

Fathers and sons

Long accounting periods are also needed in studies of inter-generational mobility, which measure the impact that parents’ position in the income distribution has on the position of their offspring. To what extent are the offspring of rich parents able to stay at the top of the income distribution? To what extent is poverty passed on to the next generation?

The best answers come from income tax data. One study based on the income tax returns of 400,000 father-son pairs in Canada found patterns of mobility similar to those found in the U.S. data. This is shown in the table, which reports how a father’s position in the income distribution correlates with that of his son. The income distribution here is sliced into quartiles (rather than quintiles), so a state of perfect mobility would be indicated by the number 25 in each cell.

Once again, the evidence suggests departures from perfect mobility at the extremes of the income distribution and much more mobility in the middle. As shown in the upper-left-hand cell, about 33 percent of sons whose fathers had incomes in the bottom quartile also ended up with earnings in the bottom quartile. At the other end, 35 percent of those with fathers whose incomes were in the top quartile also had incomes in the top quartile. In contrast, the middle is characterized by close to perfect mobility—that is, the numbers do not deviate much from the 25 percent figure one would expect if incomes were assigned by chance.

Half full?

What’s the bottom line? The evidence on mobility is such that both defenders and critics of capitalism can stick to their guns. Critics can point to the behavior at the extremes of the income distribution to argue that many of those at the top are able to maintain their position, while many at the bottom find it difficult to claw their way up. Moreover, they would add, the mobility that does exist does little to erase the extreme concentration of income and wealth in the hands of a few. Defenders can point to the significant degree of mobility among the majority in the middle to suggest, as Bhagwati does, that people “feel that they can also make it: inequality is accepted because it excites not envy but aspiration and hope.”

Prakash Loungani is Assistant to the Director of the IMF’s External Relations Department.

Other Resources Citing This Publication