## Abstract

Purpose: Different countries have different goals for social welfare policy. Consequently, it is reasonable to expect different outcomes after certain events. This article examines changes in the economic well-being of elderly women at widowhood in the United States and Germany. Design and Methods: Longitudinal data from the U.S. Panel Study of Income Dynamics and the German Socioeconomic Panel were used to prepare a sample of elderly widows. Economic well-being the year before the husband's death was compared with economic well-being the year after the husband's death. Results: Although the prevalence of poverty is different in the two countries, most widows in both countries experienced a decline in living standards, and many actually fell into poverty at widowhood. A fall in Social Security and pension income was the largest contributor to the fall in living standards. Implications: The retirement income system in both countries seems to be adequate for married couples but appears to fail for widows.

Laurence G. Branch, PhD

Poverty rates among unmarried elderly women in the United States are among the highest of any demographic group. In 1997, the poverty rate for unmarried elderly women was 19% compared with a poverty rate of less than 5% for married elderly couples (see Appendix, Note 1). Unmarried women make up almost 60% of the U.S. elderly poor while accounting for only a third of the total elderly population (National Economic Council 1998). One reason for the high poverty rate among elderly women in the United States is their reliance on Social Security as a sole or major source of income. Social Security was never meant to provide adequate income but rather to be a foundation to be built on with pensions and savings. On average, unmarried elderly women derive over half of their income from Social Security, and, for the poorest 40% of unmarried elderly women, Social Security accounts for over 75% of their total income (Social Security Administration 1998).

A commonly accepted fact is that widowhood increases the likelihood of being poor. Burkhauser, Holden, and Feaster 1988, using the Retirement History Study (RHS; see Appendix, Note 2), found that for nonpoor older married women who eventually become widows, widowhood and a fall in nonwage income are the two events most associated with a fall into poverty. Myers, Burkhauser, and Holden 1987 and Hurd and Wise 1989, also using the RHS, found that poverty rates of women who became widowed jumped dramatically after the deaths of their husbands. However, both studies also found a correlation between economic status before and during widowhood. Bound, Duncan, Laren, and Oleinick 1991, using the Panel Study of Income Dynamics and a much wider distribution of ages, concluded that "economic status prior to widowhood is the strongest predictor of status during widowhood" (p. 115). Obviously, the commonly accepted fact is only part of the story.

In this article, I examine two issues related to the death of an elderly spouse and economic well-being. The first issue deals with economic status before and after the death of a spouse: Are elderly women who are poor after becoming widows poor before? The answer to this question has important policy implications. Will policy proposals to reduce the Social Security spouse benefit and raise the survivor benefit just transfer income from one time of poverty to another time of poverty? The second issue is how the United States compares with a country that has a different philosophy toward social welfare programs. This study differs from other studies in two important respects: First, it examines living standards and income before and after the death of an elderly woman's husband. Second, it compares the United States with Germany in the 1980s and 1990s.

The results show that in both the United States and Germany, most elderly women experienced a fall in living standards at widowhood, and many fell into poverty. Also in both countries, poor widows were much more likely than other widows to be poor before the deaths of their husbands. Last, a fall in Social Security and pension income was the primary contributor to the fall in income at widowhood. This article is organized as follows: The next section briefly explains the retirement income systems in the two countries. The third section describes the data used for the analysis, and the results of the analysis are presented in the section that follows. Concluding remarks are offered in the final section.

## Retirement Income Systems in the United States and Germany

There are also some significant differences between the two countries' systems. In the United States, the retired worker's spouse will receive a benefit based on the retired worker's earnings equal to up to 50% of the retired worker's benefit. However, if the spouse has a substantial work history, he or she will receive the higher of the spouse benefit or a retired worker benefit based on his or her own earnings and age. The German Social Security system does not pay a spouse benefit, but women with some attachment to the labor market (at least 15 years of work with 10 years after age 40) will receive a benefit based on their earnings and age (see Appendix, Note 3). In the United States, after the death of the retired worker, the widow(er), if 60 years or older, receives a survivor's benefit equal to the retired worker's benefit (see Appendix, Note 4). In Germany, after the death of the retired worker, the survivor, if 45 years or older, receives 100% of the deceased worker's benefit for 3 months and then 60% thereafter.

Of the three pillars, Social Security is mandatory, whereas the other two pillars (pensions and savings) are voluntary. Consequently, many people rely on the first pillar as a sole source of income in old age because they were unable or unwilling to save and were not covered by a pension plan while working. Individuals in different parts of the income distribution rely on different combinations of income sources. Table 1 shows that those elderly persons in the bottom decile of the income distribution in both countries rely primarily on Social Security for income support—about 70% of income comes from Social Security. At the other end of the distribution, Social Security is a fairly important source of income, but it is not the only source. In both countries, earnings and pension each contribute more toward income support than Social Security for those in the top decile.

## Data and Methods

The data sources for this study are the Panel Study of Income Dynamics (PSID) and the German Socioeconomic Panel (GSOEP). The PSID is a nationally representative longitudinal data set of the U.S. population that has been ongoing since 1968 (Hill 1992). The PSID interviewed a national sample of 4,800 households in 1968, and the number interviewed has grown to over 7,000 today. The replacement mechanism of the PSID for births is designed to yield a representative sample in each year.

The income measure used in the study was posttransfer, pretax income. There were several reasons for choosing this measure. First, it is the income measure used in the United States for official poverty statistics and is used by most researchers in poverty research. Second, tax information is not available in the PSID file after 1991; thus, almost half of the sample would be unusable. Third, taxes in each data set are calculated on the basis of household characteristics rather than actual taxes paid by the household. Fourth, if taxes were subtracted from income, not only would income be lower but so too would the poverty thresholds (see Appendix, Note 5). Household income is adjusted to reflect the composition of the household (household size and ages of household members), using the detailed official U.S. equivalence weight (see Appendix, Note 6). People are ranked on basis of the income of the household they live in adjusted for household size and composition (see Appendix, Note 7).

The concept of poverty is different in the two countries (Germany does not have an official poverty threshold), consequently a comparable measure of poverty needed to be developed. I use two different thresholds in this article, but both are based on median household income. The first is 35% of median equivalence-adjusted household income (denoted POVLIN35); the second is 45% of median equivalence- adjusted household income (denoted POVLIN45). Median income is determined for the entire PSID and GSOEP samples for each year. I chose a relative poverty threshold because an absolute measure (updated for inflation) would fall further behind living standards (measured by median income) during economic expansions. A relative measure has a fixed relationship to the standard of living, and changes in the standard of living of elderly widows are the main focus of this article. In the United States, it should be noted, the poverty rate, using a relative threshold, is generally higher than the official poverty rate (Danziger, Haveman, and Plotnick 1986). For the United States, POVLIN35 yielded about the same level of poverty as the official U.S. measure when averaged over several years but did not track year-to-year changes. This is because income and prices grow at different rates. POVLIN45 yielded a higher poverty rate on average and in all years. This measure yielded poverty rates that were 42% (United States) and 75% (Germany) greater than the 35% measure. I used POVLIN45 because many researchers and policymakers (including myself) believe that the U.S. official poverty threshold is too low. Burkhauser and colleagues 1995 provided a more detailed discussion of the differences between the United States and Germany regarding poverty.

Table 2 reports some descriptive information on the women in the sample. The age distribution in Period t − 1 was similar in the two samples: The mean in both was about 71 years, and the range was from 58 years to the late 80s. All women were 60 years or older in Year t + 1 and were therefore eligible to receive survivor benefits in both countries. Over 50% of the women in the German sample and 39% in the U.S. sample had less than a high school education. Another major difference between the two samples was that 22% of the women in the U.S. sample had education past high school, and only 1% of women in the German sample had education past high school. Part of this could be due to institutional differences in educational systems in the two countries (see Appendix, Note 9). Average household sizes in Period t − 1 and in Period t + 1 in the two countries were roughly the same.

The method used in this article was a tabular analysis of poverty and income dynamics. The poverty measures examined were the poverty rate and the average poverty gap. The poverty gap is the difference between the poverty threshold and household income if the difference is positive and is equal to zero otherwise. The income measures were income changes and mobility within the income distribution. When examining income mobility, mobility refers to movement within the distribution of the entire sample of households, not just the sample of elderly widows. Hypothesis tests were conducted for either Period t − 1 to t + 1 changes in each measure or for U.S.–German differences in these measures.

## Poverty Dynamics

Poverty dynamics in the two countries were sensitive to the poverty threshold used. Results for both the 35% (POVLIN35) and 45% (POVLIN45) thresholds are presented and discussed. Table 3 shows the poverty rate and poverty gap for Periods t − 1 and t + 1 in both countries. The hypothesis tested was whether the Period t − 1 and the Period t + 1 measures were equal. The first half of the table presents the results using POVLIN35, and the second half presents the results using POVLIN45.

The period t − 1 poverty rate for these women in the United States was 12% and increased by 10 percentage points (almost doubling) to 22% in the year after widowhood. This increase in poverty is statistically significant at the 1% level. The average poverty gap tripled from $221 to$679 in the United States between Period t − 1 and Period t + 1. When examining just widows who were poor in Period t + 1, two points are worth emphasizing. First, the Period t − 1 average poverty gap of t + 1 poor widows was much larger than the average poverty gap of the whole sample (200% larger in the United States)—Period t + 1 poor widows started out in Period t − 1 (when married) much poorer than most widows. Second, the increase in the average poverty gap between Periods t − 1 and t + 1 was particularly startling—350% in the United States (this increase was statistically significant at the 1% level). In contrast, the poverty rate in Germany started out in Period t − 1 at just 3% and increased by 5.7 percentage points, or tripled, in Period t + 1. The increase in poverty gap in Germany at widowhood, however, was just as startling as the increase in the United States.

The second half of Table 3 presents the poverty results using POVLIN45 as the threshold. The overall Period t − 1 to Period t + 1 dynamics in both countries were pretty much the same, although the starting poverty rates and poverty gaps were much higher in both countries. The results in the halves of Table 3 suggest that in both the United States and Germany poor widows started out from a more disadvantaged position than nonpoor widows and become much poorer after the deaths of their husbands, although there was a smaller proportion of poor widows in Germany.

The next table, Table 4 , shows the distribution of income in relation to the poverty threshold. The columns in this table show the Period t − 1 and t + 1 distributions for the entire sample in both countries, plus the distribution in Period t − 1 of those widows who were poor in Period t + 1 (labeled t + 1 poor in t − 1). The first half of the table shows the results when using POVLIN35 as the poverty threshold. The Period t − 1 columns (a) reiterate the poverty rate results shown in Table 3 and (b) show that most women who became widows were not concentrated just above the poverty threshold. Consequently, it was not just small changes in income that pushed many of these widows into poverty. Over half of these widows in the United States lived in households with income over double the poverty threshold. In addition, 17% of Period t + 1 poor widows started out in Period t − 1 in households with income over twice the poverty threshold. The column for Period t + 1 shows that the distribution generally shifted down toward the poverty threshold, but still nearly half of the widows had income that was at least twice the poverty threshold. The overall pattern in Germany was not that much different than that in the United States.

As would be expected, with the POVLIN45 threshold poverty appeared to be much more persistent—a higher proportion of Period t + 1 poor widows were poor or near poor in Period t − 1 (see bottom half of Table 4 ). In Period t − 1, more U.S. women were concentrated near the poverty threshold, and more women were below the poverty threshold. Fewer than half of the women had income above 200% of the poverty threshold. The results were even more striking for the Period t + 1 poor widows: over 40% were poor in Period t − 1, and another 20% were near poor (i.e., 100% to 125% of the poverty threshold) in the United States. In Germany, many of the same qualitative results are evident. In addition, 28% of Period t + 1 poor widows were poor in Period t − 1, 24% were near poor, and an additional 27% were at 125%–150% of the poverty threshold.

Poverty transitions between Period t − 1 and Period t + 1 are shown in Table 5 . When POVLIN35 was used (top half), three quarters of the women in the U.S. sample were not poor in both Periods t − 1 and t + 1 compared with 90% of the women in the German sample (the U.S.–German difference was statistically significant at the 1% level). Slightly less than 10% in the United States and less than 3% in Germany were poor in both periods (again, the U.S.–German difference was statistically significant at the 1% level). Even though a very small minority in each country were always poor, it turns out that in the United States 40% of Period t + 1 poor widows were poor before widowhood in contrast to 29% of Period t + 1 poor German widows. The results were qualitatively similar when POVLIN45 was used (bottom half of Table 5 ).

## Income Dynamics

The results on poverty dynamics in the previous section show the proportion of elderly widows who moved above or below a particular threshold. However, these results do not provide much information on how living standards changed more generally. Table 6 shows how income changed for these women between Periods t − 1 and t + 1. The top two rows show the median equivalent real household income (in 1991 currency units) for these women in Periods t − 1 and t + 1. Median real income decreased slightly in both countries. On the basis of these median values, one might conclude that widowhood brings a small change in living standards. The results from the previous section show that this conclusion is certainly not valid for some widows (i.e., poor widows); however, is this conclusion valid for most widows? The next rows in Table 6 show that the median fall in equivalent real income after the husband's death was 16%, and about two thirds of the widows experienced a fall in equivalent real income in the United States. Almost 60% experienced a 10% or greater fall in income, and almost 25% of these women experienced a 10% or more increase in equivalent real income after the deaths of their husbands. Although a similar proportion of German widows experienced a fall in income after widowhood, the fall was not nearly as large as that of U.S. widows (the median decrease was 9%).

Although Table 6 shows how income changed, it does not show which income changed. Elderly couples and widows receive income from several sources, all of which can change over time or after certain events such as widowhood. For Table 7 , each sample (U.S. and German) was split into three groups, based on the percentage of change in equivalent real household income (labeled %ΔInc in the table) between Periods t − 1 and t + 1. Average income from each source for each group was then calculated for Period t − 1 and Period t + 1.

For the U.S. widows who experienced a large decline in income (at least a 10% fall; see Appendix, Note 10), the largest absolute change in income was retirement income (Social Security plus other income; see Appendix, Notes 11 and 12), accounting for about 44% of the total decrease in income. After retirement income (Social Security and pension income), a fall in labor income was the second largest absolute change. Most of the Period t − 1 labor income for these women was earned by the husbands—about 20% (most in part-time jobs) of the U.S. sample worked in Period t − 1 (see Appendix, Note 13). The loss of the main breadwinner and a fall in Social Security and pension income accounted for the lower living standards of many widows. Together, the change in labor income, Social Security income, and other income accounted for 73% of the fall in income in the United States. In Germany, the fall in retirement income (i.e., Social Security) and labor income accounted for 94% of the fall in total income. Either the retirement income system (Social Security and pensions) is failing a substantial number of widows in each country, or many families are not adequately preparing for retirement in both countries.

The next group of widows was women experiencing either a small decrease or a small increase (±10%) in income. Generally, in both countries the change in each component of total income was fairly small, and only one of the t − 1/t + 1 differences (Social Security income in Germany) approached statistical significance (at the 10% level). The final group is the women who experienced a large (i.e., more than 10%) increase in real equivalent household income at widowhood. Again, there were few differences between the two countries. The single largest increase in income was in asset income (see Appendix, Note 14). There are two possible reasons for the large increase in this component of total household income. The first source could be the annuitization of a large life insurance distribution. The second source could be an increase in the imputed rental value (equivalent) of owner-occupied housing, as ownership of this asset is generally transferred to the widow (see Appendix, Note 15).

Next, I examine how widowhood affects a woman's position in the income distribution. Table 8 shows the distribution of movement within the income distribution (quintile movement) between Periods t − 1 and t + 1 (see Appendix, Note 16). In both countries, about 50% of widows lived in households in the poorest two income quintiles before they became widows (see Appendix, Table 1 A and Table 2A ). Furthermore, of the widows who started in the poorest two quintiles in Period t − 1, the vast majority (86% in the United States and 73% in Germany) remained in the poorest two quintiles in Period t + 1. Many widows were in the poorest 40% of the population before their husbands' deaths, and most remained there afterward. Very few women actually moved up in the income distribution (15% in the United States and 27% in Germany). The proportions for quintile mobility in Table 8 were based on the number of women who could potentially move the number of quintiles indicated. For example, the proportion who moved up two or more quintiles were those women who started out in the bottom three quintiles of the income distribution. In both countries, almost 45% of the widows did not change their quintile position in the income distribution. Roughly the same proportion of widows moved down one quintile in both the United States and Germany (35% and 30%, respectively). A larger proportion moved down two or more quintiles in the United States than in Germany, and the difference was statistically significant at the 1% level. Interestingly, 25% of the women who could move two or more quintiles did move two or more quintiles in Germany, compared with only 5% in the United States.

## Concluding Remarks

Widowhood, in both the United States and Germany, is a time of loss. Widows have lost not only their husbands but also, for many, their standard of living. Many American and German women become poor at widowhood, and many American women are poor before widowhood and then become even poorer after the loss of their husbands. In Germany, women are two to three times more likely to be in poverty after the deaths of their husbands than when they were married. In the United States, women who become widows are less than twice as likely to fall into poverty after the deaths of their husbands. In addition, many poor widows were poor before the deaths of their husbands. It must be emphasized, however, that the poverty rate for elderly widows in the United States was more than double the rate in Germany.

Most widows in the United States and Germany experienced a fall in real income after the deaths of their husbands. A fall in Social Security and pension income was the largest contributor to the fall in total income in both countries, and the loss of the husband's labor income was the second largest contributor. But not all widows were worse off financially after the deaths of their husbands; one in four widows actually experienced a large rise in income (i.e., by over 10%), mostly due to an increase in asset income.

The German retirement income system appears much more effective than the U.S. system in keeping retired married couples out of poverty because Social Security benefits are more generous (taxes are also higher, though). However, both the U.S. and German systems are much less successful in maintaining the living standards of women who become widows. These results suggest that proposals in the United States to reduce the spouse benefit and increase the survivor benefit would, for many, just transfer income from a time of poverty to another time of poverty (see Appendix, Note 17). A policy to increase survivor benefits only or to increase means-tested benefits could be effective in maintaining the living standards of elderly widows, but program costs would necessarily increase.

## Notes

1. This 19% poverty rate is just slightly lower than the 20% poverty rate for children under 18 years.

2. The RHS sample consists of women married to men born between 1905 and 1911.

3. Borsch-Supan and Schnabel 1999 noted that most wives receive their own benefits.

4. This assumes the widow(er) is not eligible for a higher Social Security benefit on the basis of his or her earnings record.

5. The poverty threshold is based on median income. Incorporating taxes would change the income distribution since taxes are progressive. The calculated poverty threshold would be lower; consequently, the poverty rates in both countries would also be lower. Taxes are more progressive in Germany than in the United States. However, based on a limited analysis with the reduced data set the qualitative results hold if taxes are incorporated, but in many instances cell sizes are very small.

6. This equivalence weight is based on the official poverty threshold in the United States and accounts for economies of scale by adjusting for both the size and the composition of the family.

7. An implicit assumption is that income is freely shared among household members.

8. No German guest workers were included in the sample because they are not eligible to receive Social Security benefits.

9. From the income variable calculation appendix in the codebook (Butrica & Jurkat, n.d.), Social Security income in Germany included income from Social Security pensions plus income from employer or occupational old-age pensions. In the U.S. sample, Social Security income included only income from the Social Security Administration.

10. Other income in the United States included employer pension income as well as transfers from persons outside of the household. In the German sample, other income included only transfers from persons outside of the household.

11. Of the households where women worked in Period t − 1, their labor income accounted for 43% of total household labor income in the United States.

12. Asset income included income from assets plus the imputed rental value of owner-occupied housing. The imputed rental value of owner-occupied housing captures a cost that homeowners do not incur and was considered as income.

13. The imputed rental of owner-occupied housing most likely remains constant over Periods t − 1 to t + 1, but the equivalence factor becomes smaller after the husband's death. Consequently, the equivalent value of this income source will increase after the husband's death.

14. The transition matrices for both the United States and Germany are shown in Appendix Table 1 A and 2A. Income quintiles are defined for all households and are based on equivalent household income. Consequently, the rows and columns in Appendix Table 1 A and Table 2A do not necessarily have to sum to 20%.

15. It can also be argued that for families with high discount rates a transfer of income (inflation adjusted) to the future would make the family worse off.

Table 1.

Shares of Aggregate Income of the Elderly by Income Decile, United States and Germany

 United States, 1994 Germany, 1989 Income Decile Decile 1 Decile 10 Decile 1 Decile 10 Social Security 69.7 18.8 73.9 26.4 Pensions 3.9 20.0 3.7 31.3 Asset income 6.1 23.2 7.1 8.4 Earnings 2.6 37.9 6.5 33.6 Other income 17.9 0.1 8.9 0.3
 United States, 1994 Germany, 1989 Income Decile Decile 1 Decile 10 Decile 1 Decile 10 Social Security 69.7 18.8 73.9 26.4 Pensions 3.9 20.0 3.7 31.3 Asset income 6.1 23.2 7.1 8.4 Earnings 2.6 37.9 6.5 33.6 Other income 17.9 0.1 8.9 0.3

Source: Smeeding 1997.

Table 2.

Descriptive Statistics of Panel Study of Income Dynamics and German Socio-Economic Panel Samples

 Characteristic United States Germany Average Age 70.9 71.2 Age Range 58–89 58–86 Household size: Period t − 1 2.1 2.2 Household size: Period t + 1 1.2 1.2 Education (%) Less than high school 38.5 54.4 High school diploma 40.0 44.6 More than high school 21.5 1.0 Sample Size 259 177
 Characteristic United States Germany Average Age 70.9 71.2 Age Range 58–89 58–86 Household size: Period t − 1 2.1 2.2 Household size: Period t + 1 1.2 1.2 Education (%) Less than high school 38.5 54.4 High school diploma 40.0 44.6 More than high school 21.5 1.0 Sample Size 259 177

Note: Sample is weighted by household weights.

Table 3.

Changes in Selected Measures of Economic Status

 Measure & Year United States Germany Poverty Threshold is 35% of Median Income Poverty rate t − 1 12.1% 3.0% t + 1 22.0%** 8.7%* Average poverty gap t − 1 $221 DM66 t + 1$679** DM255* Average poverty gap of poor in t + 1 t − 1 $684 DM688 t + 1$3,082** DM2,930** Poverty Threshold is 45% of Median Income Poverty rate t − 1 18.9% 9.7% t + 1 30.7%** 23.2%** Average poverty gap t − 1 $559 DM261 t + 1$1,272** DM828** Average poverty gap of poor in t + 1 t − 1 $1,406 DM866 t + 1$4,143** DM3,575**
 Measure & Year United States Germany Poverty Threshold is 35% of Median Income Poverty rate t − 1 12.1% 3.0% t + 1 22.0%** 8.7%* Average poverty gap t − 1 $221 DM66 t + 1$679** DM255* Average poverty gap of poor in t + 1 t − 1 $684 DM688 t + 1$3,082** DM2,930** Poverty Threshold is 45% of Median Income Poverty rate t − 1 18.9% 9.7% t + 1 30.7%** 23.2%** Average poverty gap t − 1 $559 DM261 t + 1$1,272** DM828** Average poverty gap of poor in t + 1 t − 1 $1,406 DM866 t + 1$4,143** DM3,575**

Notes: Sample weighted by household weights. Currency in constant 1991 currency units.

*

t − 1/t + 1 difference significant at 5% level, two-tailed test; **t − 1/t + 1 difference significant at 1% level, two-tailed test.

Table 4.

Distribution Around the Poverty Threshold in Periods t − 1 and t + 1

 United States Germany % of Poverty Line t − 1 t + 1 t + 1 poor in t − 1 t − 1 t + 1 t + 1 poor in t − 1 Poverty Threshold is 35% of Median Income 0–50 0.89 6.78 2.70 0.00 0.94 0.00 50–100 11.24 15.24 37.76 2.97 7.76 28.77 100–125 5.67 7.91 9.51 6.37 13.23 9.47 125–150 7.25 8.34 11.07 8.50 10.03 9.27 150–200 18.41 12.67 22.06 21.15 20.12 26.68 200+ 56.54 49.06 16.90 61.01 47.92 25.82 Poverty Threshold is 45% of Median Income 0–50 3.08 11.08 6.70 0.00 1.76 0.00 50–100 15.80 19.63 37.91 9.71 21.41 28.11 100–125 12.81 8.41 23.24 10.09 10.87 24.06 125–150 10.48 9.95 15.22 16.73 15.60 27.54 150–200 14.38 15.73 8.97 24.66 18.23 7.26 200+ 43.45 35.19 7.96 38.81 32.14 13.04
 United States Germany % of Poverty Line t − 1 t + 1 t + 1 poor in t − 1 t − 1 t + 1 t + 1 poor in t − 1 Poverty Threshold is 35% of Median Income 0–50 0.89 6.78 2.70 0.00 0.94 0.00 50–100 11.24 15.24 37.76 2.97 7.76 28.77 100–125 5.67 7.91 9.51 6.37 13.23 9.47 125–150 7.25 8.34 11.07 8.50 10.03 9.27 150–200 18.41 12.67 22.06 21.15 20.12 26.68 200+ 56.54 49.06 16.90 61.01 47.92 25.82 Poverty Threshold is 45% of Median Income 0–50 3.08 11.08 6.70 0.00 1.76 0.00 50–100 15.80 19.63 37.91 9.71 21.41 28.11 100–125 12.81 8.41 23.24 10.09 10.87 24.06 125–150 10.48 9.95 15.22 16.73 15.60 27.54 150–200 14.38 15.73 8.97 24.66 18.23 7.26 200+ 43.45 35.19 7.96 38.81 32.14 13.04

Note: Sample weighted by household weights.

Table 5.

Poverty Dynamics Between Periods t − 1 and t + 1

 Dynamic United States (%) Germany (%) Poverty Threshold is 35% of Median Income Never poor 74.75** 90.83 Poor to not poor 3.22* 0.47 Not poor to poor 13.11* 6.20 Always poor 8.91** 2.50 Poverty Threshold is 45% of Median Income Never poor 64.11* 73.63 Poor to not poor 5.17 3.20 Not poor to poor 17.01 16.66 Always poor 13.70* 6.51
 Dynamic United States (%) Germany (%) Poverty Threshold is 35% of Median Income Never poor 74.75** 90.83 Poor to not poor 3.22* 0.47 Not poor to poor 13.11* 6.20 Always poor 8.91** 2.50 Poverty Threshold is 45% of Median Income Never poor 64.11* 73.63 Poor to not poor 5.17 3.20 Not poor to poor 17.01 16.66 Always poor 13.70* 6.51

Note: Sample weighted by household weights.

*

U.S./German difference significant at 5% level, two-tailed test; **U.S./German difference significant at 1% level, two-tailed test.

Table 6.

Distribution of Percentage Changes in Equivalent Real Household Income Between Periods t − 1 and t + 1

 Measure United States Germany Median equivalent real household income t − 1 $17,678 DM27,505 t + 1$15,073 DM23,467 Median % Change −16.22 −9.26 Distribution Less than −10% 58.59\|[dagger]\| 48.86 −10% to 0% 8.23* 16.04 0% to 10% 8.36 12.54 Greater than 10% 24.81 22.55
 Measure United States Germany Median equivalent real household income t − 1 $17,678 DM27,505 t + 1$15,073 DM23,467 Median % Change −16.22 −9.26 Distribution Less than −10% 58.59\|[dagger]\| 48.86 −10% to 0% 8.23* 16.04 0% to 10% 8.36 12.54 Greater than 10% 24.81 22.55

Notes: Sample weighted by household weights. Currency in constant 1991 currency units.

\|[dagger]\|

U.S./German difference significant at 10% level, two-tailed test; *U.S./German difference significant at 5% level, two-tailed test.

Table 7.

Income from Various Sources Periods t − 1 and t + 1 by Percentage Change in Total Household Income

 %ΔInc < −10% −10% ≤ %ΔInc < +10% +10% ≤ %ΔInc Type of Income t − 1 t + 1 t − 1 t + 1 t − 1 t + 1 United States (1991$) Labor 5,307** 1,245 4,320 2,984 1,635 3,190 Asset 6,827\|[dagger]\| 3,181 4,836 5,856 3,509** 8,197 Welfare 104 127 201 127 145 293 Social Security 8,074** 6,655 7,388 6,895 7,186 7,504 Other 7,861** 3,270 5,025 5,519 5,560 6,850 Total 28,173** 14,478 21,770 21,380 18,035* 26,035 Germany (1991 DM) Labor 3,012* 458 6,302 7,770 3,759 8,246 Asset 1,190 624 1,192 1,762 2,302* 8,217 Welfare 138 133 176 416 216\|[dagger]\| 1,208 Social Security 24,909** 18,635 22,433\|[dagger]\| 19,385 17,339 17,730 Other 2,993 2,041 2,913 3,600 4,206 5,856 Total 32,242** 21,891 33,015 32,933 27,822** 41,256  %ΔInc < −10% −10% ≤ %ΔInc < +10% +10% ≤ %ΔInc Type of Income t − 1 t + 1 t − 1 t + 1 t − 1 t + 1 United States (1991$) Labor 5,307** 1,245 4,320 2,984 1,635 3,190 Asset 6,827\|[dagger]\| 3,181 4,836 5,856 3,509** 8,197 Welfare 104 127 201 127 145 293 Social Security 8,074** 6,655 7,388 6,895 7,186 7,504 Other 7,861** 3,270 5,025 5,519 5,560 6,850 Total 28,173** 14,478 21,770 21,380 18,035* 26,035 Germany (1991 DM) Labor 3,012* 458 6,302 7,770 3,759 8,246 Asset 1,190 624 1,192 1,762 2,302* 8,217 Welfare 138 133 176 416 216\|[dagger]\| 1,208 Social Security 24,909** 18,635 22,433\|[dagger]\| 19,385 17,339 17,730 Other 2,993 2,041 2,913 3,600 4,206 5,856 Total 32,242** 21,891 33,015 32,933 27,822** 41,256

Notes: Three cases from the U.S. sample who experienced at least a 10% increase in income were omitted from these calculations due to an usually large increase in other income that substantially changed the means. %ΔInc = percentage of change in equivalent real household income.

\|[dagger]\|

t − 1/t + 1 difference significant at 10% level, two-tailed test; *t − 1/t + 1 difference significant at 5% level, two-tailed test; **t − 1/t + 1 difference significant at 1% level, two-tailed test.

Table 8.

Income Mobility Based on Relative Scale (Quintiles)

 Relative Scale United States Germany Up 2 or more quintiles 5.56** 25.84 Up 1 quintile 12.48 9.29 No change 45.79 44.95 Down 1 quintile 35.08 30.22 Down 2 or more quintiles 23.26** 7.49 1 15.59 1.89 1.69 0.57 0.77 2 12.40 13.93 3.21 0.44 0.26 3 3.86 7.87 6.30 2.86 0.26 4 0.29 1.66 5.03 5.59 2.94 5 0.51 1.00 4.16 2.51 4.41
 Relative Scale United States Germany Up 2 or more quintiles 5.56** 25.84 Up 1 quintile 12.48 9.29 No change 45.79 44.95 Down 1 quintile 35.08 30.22 Down 2 or more quintiles 23.26** 7.49 1 15.59 1.89 1.69 0.57 0.77 2 12.40 13.93 3.21 0.44 0.26 3 3.86 7.87 6.30 2.86 0.26 4 0.29 1.66 5.03 5.59 2.94 5 0.51 1.00 4.16 2.51 4.41

Note: Sample weighted by household weights.

Note: Table entries sum to 100%.

**

U.S./German difference significant at 1% level, two-tailed test.

Table 2A.

Transition Matrix (Quintiles) Weighted Cell Percentages, Germany

 Year t + 1 Year t − 1 1 2 3 4 5 1 11.70 1.91 0.35 0.00 4.54 2 12.43 11.38 1.93 0.00 7.03 3 1.91 7.73 3.08 2.47 7.09 4 1.05 0.00 2.81 4.74 1.45 5 0.31 0.38 0.00 1.66 14.05
 Year t + 1 Year t − 1 1 2 3 4 5 1 11.70 1.91 0.35 0.00 4.54 2 12.43 11.38 1.93 0.00 7.03 3 1.91 7.73 3.08 2.47 7.09 4 1.05 0.00 2.81 4.74 1.45 5 0.31 0.38 0.00 1.66 14.05

Note: Table entries sum to 100%.

I thank Maria S. Floro and Howard Iams for helpful comments on drafts of this article. The views expressed herein do not reflect the views of the Social Security Administration.

## References

Börsch-Supan A., Schnabel R.,
1999
. Social Security and retirement in Germany. Gruber J., Wise D. A., ed.
Social Security and retirement around the world

135
-180. University of Chicago Press, Chicago.
Bound J., Duncan G. J., Laren D. S., Oleinick L.,
1991
. Poverty dynamics in widowhood.
Journal of Gerontology

46:
S115
-S124.
Burkhauser R. V., Butrica B. A., Daly M. C.,
1995
.
The Syracuse University PSID-GSOEP equivalent data file: A product of cross-national research (Cross-National Studies in Aging Program Project Paper #25)
Syracuse University, Syracuse, NY.
Burkhauser R. V., Duncan G. J., Hauser R.,
1994
. Sharing prosperity across the age distribution: A comparison of the United States and Germany in the 1980s.
The Gerontologist

34:
150
-160.
Burkhauser R. V., Holden K. C., Feaster D.,
1988
. Incidence, timing, and events associated with poverty: A dynamic view of poverty in retirement.
Journal of Gerontology

43:
S46
-S52.
Butrica, B., & Jurkat, D. (n.d.). Codebook for PSID-GSOEP equivalent file 1980-1994. Syracuse, NY: Syracuse University.
Danziger S. H., Haveman R. H., Plotnick R. D.,
1986
. Antipoverty Policy: Effects on the Poor and the Nonpoor. Danziger S. H., Weinberg D. H., ed.
Fighting poverty: What works and what doesn't

50
-77. Harvard University Press, Cambridge, MA.
Hill M.,
1992
.
Panel study of income dynamics: User's guide
Sage, New York.
Hungerford T. L.,
1996
. The distribution and anti-poverty effectiveness of U.S. transfers, 1992.
Journal of Human Resources

31:
(1)
255
-273.
Hurd M. D., Wise D. A.,
1989
. The wealth and poverty of widows: Assets before and after the husband's death. Wise D., ed.
The economics of aging

117
-199. University of Chicago Press, Chicago.
Iams H. M., Sandell S. H.,
1998
. Cost-neutral policies to increase social security benefits for widows: A simulation for 1992.
Social Security Bulletin

61:
(1)
34
-43.
Myers D. A., Burkhauser R. V., Holden K. C.,
1987
. The transition from wife to widow: The importance of survivor benefits to widows.
Journal of Risk and Insurance

54:
752
-759.
National Economic Council
1998
.
Women and retirement security
National Economic Council, Washington, DC.
Sandell S. H., Iams H. M.,
1996
. Reducing women's poverty by shifting Social Security benefits from retired couples to widows.
Journal of Policy Analysis and Management

16:
(2)
279
-297.
Smeeding T. M.,
1997
.
Reshuffling responsibilities in old age: The United States in a comparative perspective. (LIS working paper No. 153)

Smolensky E., Danziger S., Gottschalk P.,
1988
. The declining significance of age in the United States: Trends in the well-being of children and the elderly since 1939. Palmer J., Smeeding T., Boyle Torrey B., ed.
The vulnerable

29
-54. The Urban Institute Press, Washington, DC.
1998
.
Income of the population 55 or older, 1996
Office of Research, Evaluation and Statistics, Washington, DC.
1997
.
Report of the 1994-1996 Advisory Council on Social Security

Wagner G. G., Burkhauser R. V., Behringer F.,
1993
. The English use file of the German socio-economic panel.
Journal of Human Resources

28:
(2)
429
-433.
World Bank
1994
.
Averting the old age crisis
Oxford University Press, Oxford, England.