Klevmarken, Lupton, and Stafford 341
as the young become middle-aged and middle-aged consume their wealth in retire- ment. On the other hand, variation can also be a result of phenomena exogenous to
individual choice. These range from various forms of discrimination that alter the ‘‘rules of game’’ for specific groups of people to inherent differences in ability.
Distinguishing between tastes and inherent impediments is a crucial step in monitor- ing economic inequality.
While in both countries there is a variety of interesting assessments of macroeco- nomic factors, in this paper, we instead focus on a reduced form version of quantile
mobility and examine the extent to which initial conditions affect quantile mobility without seriously ferreting out its fundamental causes. High wealth mobility in dol-
lars can contribute to a high wealth inequality, and a high mobility might make a high inequality more acceptable from an equity point of view. We thus turn to a
comparative analysis of mobility.
A. Background and Measures
How does mobility in the wealth distribution compare across the two countries? If the cross-sectional wealth distribution is widely dispersed is it still possible that the
rich and poor trade places frequently in either or both countries? What do we mean by trading places? One definition is movement in relative position in the wealth
distribution through time, such as changing location in the decile or quintiles of wealth through time. For comparison purposes we will rely primarily on wealth
quintiles for the two countries. This is to avoid excessive detail in the mobility tables and because the sample sizes in HUS are not large enough for additional quantile
disaggregation. The extent of wealth transitions across quantiles can be measured by Shorrocks’ index.
23
It measures the share of the off-diagonal elements in quantile transition tables, such as Tables 6 through 9 below, and it ranges from zero no
mobility to a value just above one no stability. It is not invariant to the choice of quantiles and it will of course depend on the time span used. Factors that influence
these transition tables were discussed in the introduction.
B. Measuring Mobility
For Sweden over the nine-year period, 198384–199293, the Shorrocks’ index for household decile wealth mobility has been estimated as 0.87 Bager-Sjo¨gren and
Klevmarken 1996, compared with the 0.804 for the United States over the ten-year period, 1984–94 Hurts, Luoh, and Stafford 1998. For the United States between
1984–89 and then 1989–94, the Shorrocks’ measure rose modestly from 0.733 to 0.754 Hurst, Luoh, and Stafford 1998. Looking at the period 1993–98 for Sweden,
the HUS data for quintiles show a value of 0.744 for the Shorrocks’ index and for the PSID, 1994–99 the data for quintiles show a value of 0.592. These estimates
are based on the transition matrices in Tables 6 and 7. While these measures suggest more wealth mobility in Sweden, it should be remembered that they are measures of
23. Here we use the Shorrocks’ measure, S ⫽ N ⫺ trPN ⫺ 1, where N is the number of groups quintiles and tr P is the trace of the NN transition matrix P. 0 ⬍ S ⬍ NN ⫺ 1. See Shorrocks
1978 for more details on the attractiveness of this index for capturing the properties of mobility.
342 The Journal of Human Resources
Table 6 Quintile Wealth Transition matrix for the United States 1994–99 quintiles in
1993 USD
Bottom Bracket 1
2 3
4 5
Quintile Value
2,949 29,730
80,752 206,195
1 0.583
0.273 0.099
0.031 0.015
2 1,043
0.267 0.435
0.223 0.058
0.016 3
22,011 0.087
0.208 0.419
0.232 0.055
4 66,118
0.048 0.079
0.193 0.481
0.2 5
165,345 0.014
0.022 0.051
0.2 0.713
Note: Shorrocks index of mobility equal to 0.592. Row numbers are 1994 quintiles while column numbers are 1999 quintiles. Bottom of quintile brackets reported adjacent to quintile numbers. Number of house-
holds is 4,383.
mobility across quintiles deciles in the previously cited research. If these quintiles themselves are wider apart and widening through time as shown for the United
States, one cannot straightforwardly conclude that there is more wealth mobility in Sweden. Remembering that the absolute spread of the U.S. wealth distribution has
been rising, it seems safe to conclude that wealth mobility is rising in the United States, a finding parallel to rising income mobility Gottschalk and Moffitt 1994.
Much of the quintile wealth mobility is in the mid-range quantiles. The top and bottom quantiles are characterized by substantial persistence, partly the simple con-
sequence of only single direction movement at the top and bottom quintiles. Of the families in the top U.S. wealth quintile in 1994, almost three-quarters 71.3 percent
are in the top quintile in 1999. Even in terms of deciles and extending the transition period to 10 years, between 1984 and 1994, for example, over half 53.3 percent
Table 7 Quintile Wealth Transition matrix for Sweden 1993–98 quintiles in 1993 USD
Bottom Bracket 1
2 3
4 5
Quintile Value
29,137 60,171
100,552 153,582
1 0.568
0.247 0.096
0.041 0.048
2 20,436
0.253 0.315
0.26 0.103
0.068 3
47,218 0.136
0.218 0.238
0.306 0.102
4 74,712
0.034 0.144
0.294 0.322
0.206 5
126,078 0.007
0.075 0.116
0.224 0.578
Note: Shorrocks index of mobility equal to 0.744. Row numbers are 1993 quintiles while column numbers are 1998 quintiles. Bottom of quintile brackets reported adjacent to quintile numbers. Number of house-
holds is 732.
Klevmarken, Lupton, and Stafford 343
of those initially in the top decile were still in the top decile.
24
For Sweden, of the families in the top quintile in 1993 almost three-fifths 57.8 percent remained in
the top quintile in 1998. At the other end of the spectrum, of those U.S. families in the bottom decile in 1984—which include numerous negative wealth families—
about half are still in the bottom decile in 1989 and about two-fifths are in the bottom decile a decade later in 1994. This is more important when we remember that the
1994 bottom decile for the United States was in much greater negative territory than the 1984 bottom decile See Figure 1.
For 1994 to 1999, Table 6 shows that 58 percent or almost three-fifths of U.S. families in the lowest quintile in 1994 were still in the lowest quintile in 1999. For
Sweden Table 7, of those in the bottom quintile in 1993, more than half 56.8 percent were in the bottom quintile in 1998. Of course, absolute numbers matter,
since the real value of assets of those in the bottom quintile in Sweden is well above the assets of those in the bottom quintile in the United States
There are families with persistently low and negative wealth in the United States, despite the overall drift toward greater wealth mobility. In the mid-quintiles, since
movements can occur both upward and downward from the initial position, there is less persistence. Using the overall summary of the Shorrocks’ index, we have a five-
year quintile-based value of 0.744 for Sweden 1993–98 and 0.592 for the United States 1994–99, indicating a higher degree of mobility in Sweden than in the United
States on the order of 25.7 percent.
C. Standardizing Mobility Differences