Released: October 18, 2004
The Wealth of Hispanic Households: 1996 to 2002
IV. The Effects of the 2001 Recession on Household Wealth
One of the goals of this study is to determine the change in household wealth associated with the 2001 recession. It is estimated that the recession had a large impact on the net worth of Hispanic and Black households, eroding about one-quarter of their wealth within two years. The wealth of White households continued to grow through the recession and they expanded their advantage relative to other households. Because the Hispanic population is constantly changing with the influx of low-wealth immigrants, one challenge is to set aside the effect of population change and focus only on the impact of the business cycle. That requires some manipulation of the SIPP data.
The impact of the recession is measured by estimating the change in median wealth between the end of 1999 and the end of 2001. The duration of the recession, as declared by the National Bureau of Economic Research, was from March to November 2001. As mentioned in the introductory section, one problem that has to be resolved is that the estimates of wealth for 1999 may have an upward bias. The reason for this upward bias is that the 1996 SIPP panel is not replenished over its four-year life span to incorporate newly arriving immigrants. Since the net worth of new immigrants is typically lower than average, their arrival lowers the median wealth of the population. But this downward pull on wealth is missed by the SIPP sample and the extent of the error could be significant for population sub-groups where immigration is an important source of growth. Hispanics are the primary example of such a population sub-group. The Hispanic population is growing at roughly four percent per year and almost one-half of that growth can be attributed to immigration. There is, therefore, reason to believe that the estimates of Hispanic wealth in 1999 are overstated because the significant numbers of relatively poorer Hispanic immigrants which entered the country after 1996 are not reflected in the sample.12 The overstatement is automatically corrected with the selection of the 2001 SIPP panel because that the new sample is representative of the population at the time it is drawn, including recently arrived immigrants.
One way of eliminating the sample effect from the comparison of wealth across 1999 and 2001 is to adjust the composition of the 2001 SIPP panel, which represents the U.S. population in 2001, to better match the composition of the 1996 SIPP panel which represents the U.S. population in 1996. That is achieved by excluding immigrants who arrived in the U.S. in 1996 or later years from the 2001 SIPP panel. The wealth of this sub-sample from the 2001 SIPP can then be compared with the wealth of the 1996 SIPP panel, as measured in 1999, to answer the following question: How did the 2001 recession affect the wealth of households who were residents of the U.S. in 1996? In this framework, the wealth of a sample of households representing the U.S. population in 1996 is measured and compared across two points in time—1999 and 2001. The answer to this question may not be the same as comparing the wealth of the actual, but disparate, U.S. populations in 1999 and 2001, but it does isolate the effect of the business cycle by eliminating the sample composition effect. In other words, by this measurement rises or declines in the wealth of Latinos are the result of economic rather than demographic change.
The need to eliminate a group of immigrants from the 2001 SIPP triggers another restriction on the data. Immigrant status is known for most, but not all, members of a SIPP panel. Therefore, households whose immigrant status is unknown are also eliminated from the 2001 SIPP sample, and, for the sake of comparability, from the 1996 SIPP panel.13 As a result, the estimates of median wealth reported in this section differ from those reported in section 2 not just for 2001 and 2002 but also for the earlier years. However, the additional sample restriction has very small effects on the estimates of median wealth for 1996 to 1999.
The estimates of wealth adjusted to make the samples comparable across the two SIPP panels are reported in Table 5. For the 1996 to 1999 time period, the estimates in Table 3 differ slightly from those presented earlier in Table 1 due to the exclusion of households whose immigrant status is unknown. For example, the median wealth of the full sample of Hispanic households in 1999 is estimated to be $10,495 (Table 1). If the sample is restricted to households with a known immigrant status, the median wealth of Hispanic households in 1999 is estimated to be $10,660 (Table 5). Thus, restrictions on 1996 SIPP panel have minimal effect on the estimates of the net worth of households.
However, the restrictions on the composition of the 2001 SIPP panel have a notable impact on the estimates of wealth, especially for Hispanic households. The first restriction on this sample is to remove households whose immigrant status is unknown, and the second restriction is to eliminate immigrant households who entered the U.S. during or after 1996. For Hispanic households, this raises the estimated median wealth in 2001 from $6,213 in Table 1 to $7,791 in Table 5. The former is a better estimate of the true wealth of the population of Hispanic households in 2001, but the latter is more comparable to estimates derived from the 1996 SIPP panel. The difference between these two estimates is $1,578 and, although not shown in Table 5, all but $226 of this increase is due to the removal of immigrants entering the U.S. between 1996 and 2000. That underscores the importance of immigration in shaping the economic profile of the Hispanic population in the U.S. The wealth estimates for White and Black households in 2001 and 2002 are also impacted by the sample restrictions but in those cases the changes are almost entirely due to the removal of households with unknown immigrant status. Immigration itself is not an important phenomenon for those households.
The 2001 recession had a large effect on the net worth of Hispanic and Black households. Between 1999 and 2001 the median wealth of Hispanic households fell from $10,660 to $7,791, a decline of 26.9 percent (Table 5). A fair amount of this loss was apparently recovered in the following year as the median worth of Hispanic households in 2002 is estimated to be $9,845. However, this is still eight percent below the pre-recession level of wealth. Black households suffered a decline of 26.7 percent in their median wealth between 1999 and 2001, from $8,696 to $6,373. The recovery for Blacks was much weaker as their net worth increased by only 5.2 percent during 2002. In fact, their net worth of $6,706 in 2002 was 23 percent less than their net worth in 1999. On the other hand, the wealth of White households continued to increase through the recession, increasing by over two percent between 1999 and 2001 and then again by a similar percentage in 2002. Thus, the recession of 2001 and the economic slowdown that followed were harder on the net worth of minority groups.1314
- Another potential source of error with longitudinal data such as SIPP is “attrition bias.” That arises from the departure of households or persons from the sample over time. ↩
- This finding echoes that of earlier reports released by the Pew Hispanic Center that studied the impact of the latest recession on the employment and earnings of different racial/ethnic groups. See the PHC reports referenced in footnote 3 above. ↩
- Based on the analysis of data from the Survey of Consumer Finances, the Federal Reserve Board reported an increase of 10.4 percent in the median wealth of all U.S. households between 1998 and 2001 (see the reference in footnote 10). The gains for White and Black households were reported to be 16.9 percent and 13.1 percent respectively. Estimates from the SIPP, as shown in Table 5, also reveal double-digit percentage increases in the wealth of all households and White households between 1998 and 2001. The difference is that SIPP data show a decline in the wealth of Black households from 1998 to 2001. That may be a consequence of the fact that SIPP over samples low-income households and the opposite is true of the Survey of Consumer Finances. ↩