| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Department of Policy Analysis and Management, Cornell University and German Institute for Economic Research, Ithaca, New York.
2 Statistics Canada, Ottawa, Canada.
3 Department of Economics, Otto-Friedrich-Universität Bamberg and German Institute for Economic Research, Bamberg, Germany.
Address correspondence to Richard Burkhauser, Cornell University, Department of Policy Analysis and Management, 119A MVR Hall, Ithaca, NY, 14853. E-mail: rvb1{at}cornell.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. With data from the Cross-National Equivalent File, we used event history analysis to track changes in the social security replacement rate and the more comprehensive total income replacement rate for women and to show how these changes vary across age and household income quintiles within and across countries.
Results. There were substantial differences across the countries in how income from specific sources changes, especially with respect to the mix of income from government and private sources, but the overall across-country pattern of total income replacement rates was remarkably similar both in size and in distribution across age and the woman's place in the income distribution prior to her husband's death.
Discussion. Studies that focus on a social security replacement rate will seriously understate the actual total income replacement rate of women following a husband's death. This will especially be the case in countries like the United States where private sources of income play a more important role in income replacement.
LABOR earnings are the primary source of income in most households, and researchers often focus on how social security programs replace lost earnings to gauge the potential income available to households after a worker's death. Because government-sponsored social security programs are larger (higher social security taxes and higher yearly benefits) in the 30 (mostly advanced industrial) countries that belong to the Organization for Economic Cooperation and Development (OECD) than in the United States, it is often found that social security benefits provide widows with a larger share of their husband's past earnings in these countries. Such findings are then used to argue that the decline in the economic well-being of women and their households is lower in these countries following the death of their spouse than in the United States. But such arguments, even when the analysis on which they are based is done appropriately, ignore how changes in other sources of income replace income lost following a husband's death.
This article is the first to use longitudinal data from four OECD countries (United States, Germany, Great Britain, and Canada) to trace the household income of women before and after the death of their spouse and to measure how it changes in its sources and amounts. As expected, we find large differences across countries in the degree to which government-sponsored social security programs replace lost earnings following the husband's death. Despite these differences, we find the average change in women's household size-adjusted income following their husbands' deaths to be remarkably similar across these countries.
A lack of comparable data usually restricts cross-national studies of how women fare after their husbands' deaths to two types. One uses a hypothetical average worker's earnings history to determine the husband's earnings before his death and the social security benefit his wife would receive based on that earnings history following his death to make this comparison (see Gruber & Wise, 1999
). The other uses cross-sectional data from countries to compare the economic well-being of married women and widows of a given age (see Yamada & Casey, 2002 and other studies using cross-sectional data from the Luxembourg Income Study www.lisproject.org/publications/wpapersentire.htm). More sophisticated cross-national studies use synthetic cohort analysis to measure the changes in a cohort's economic well-being as it ages and becomes more dominated by widows (see Williamson & Smeeding, 2005).
These studies are of limited value, in measuring the economic risks faced by a woman following the death of her husband, either because they fail to recognize variation in the importance of social security benefit programs or other government cash transfer programs in "income replacement" across countries or because they cannot directly trace changes in the economic well-being of women following the deaths of their husbands.
Literature tracing the actual changes in the economic well-being of women following the deaths of their husbands exists for the United States. The general conclusions are that the economic well-being of women falls following the deaths of their husbands and that widows of a given age cohort are more likely to experience declines in income and drops into poverty than are married women of that age cohort. Most recently, Haveman, Holden, Wilson, and Wolfe (2003) analyzed 10 years of administrative records data to show how much family income of older women who were married in 1982 and single in 1991 declined and the importance of social security benefits in offsetting that decline. Burkhauser, Butler, and Holden (1991) looked at this same transition using data from the Retirement History Survey for the 1970s. Burkhauser and Duncan (1989) used the Panel Study of Income Dynamics (PSID) data to compare economic well-being changes across several life events including widowhood over this same period (also see Holden & Smock, 1991; Zick & Holden, 2000; and Zick & Smith, 1991).
But cross-national comparisons of widowhood, especially across the entire age distribution, are rarer. Burkhauser, Duncan, Hauser, and Berntsen (1991) were the first to use data from the PSID and the German Socio-Economic Panel (GSOEP) to compare the economic well-being of women following a marital split, but they did not have sufficient data to focus on economic consequences following the death of a spouse.
We took advantage of a newly expanded source of cross-national panel data, the Cross-National Equivalent File (CNEF), which contains comparable socioeconomic information on households in four countries (United States, Germany, Great Britain, and Canada), to trace the economic well-being of women following the deaths of their husbands.
Cross-national comparisons of this type can show the consequences of public policy because these countries have a mixture of private and public institutions to ameliorate the economic consequences of a husband's death. On the public side, their social insurance systems provide income to widows (survivor benefits). Such programs typically provide a minimum social safety net for nonworkers that may either be categorical or universal in design (see Aarts, Burkhauser, & de Jong, 1996, for a fuller taxonomy of social welfare systems in a comparative context and Burkhauser, Giles, Lillard, & Schwarze, 2004, for a discussion of how that taxonomy was used in this article).
While many studies of the economic consequences of widowhood focus on the ameliorative role of government programs, private institutions also play a part in replacing lost earnings following the death of a spouse. This is most importantly the case in the United States, but also in Great Britain and Canada and even to a lesser degree in Germany. Private employer compensation packages in these countries provide protection to many women following their husbands' deaths. In addition, some households in these countries are able to use income from their accumulated wealth, from the added market work of other household members, or from life insurance settlements to offset the earnings loss from a worker's death. Some women are able to alter their behavior to compensate for income lost when their husbands die. They remarry. They share living quarters with adult children or with unrelated adults to reduce average household costs such as heating or electricity (see Smith & Zick, 1986; Smith, Zick, & Duncan, 1991). Coresiding adult members of these households also contribute resources. Holden (1988) was the first to document the importance of including household composition changes in measures of the relative importance of social security benefit increases over time in improving the economic well-being of widows.
All these potential offsets to lost earnings are missed by studies that focus exclusively on social security survivor benefits. Conceptually then an analysis of the economic well-being of widows should consider all potential sources of income to assess how women fare after the deaths of their husbands. Such an analysis should consider both public and private sources of income, including earnings contributed by coresiding members of the widow's new household. This is the approach we take below.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
In each country sample, we pooled women by the age of her husband at death, regardless of his calendar year of death. We realigned our calendar year data into an event-history framework, where the event occurred in his year of death (t). His age was measured as his age at survey interview year minus 1. To avoid complications associated with husbands dying in different months of a given year, our analysis focused on economic well-being in the year before and the year after his death. Our data included income years 1970 through 1997 for the PSID, 1984 through 2000 for the GSOEP, 1991 through 2000 for the British Household Panel Study (BHPS), and 1993 through 2000 for the Canadian Survey of Labour and Income Dynamics (SLID). To increase our sample of PSID widows, we added 1970 to 1979 PSID data to our CNEF data, as well as the restricted access PSID file containing official date of death records for deceased PSID sample members (see Burkhauser, Lillard, & Valenti, 2001, for an example of this type of analysis where the event is a long-term exit from the labor force other than death).
CNEF includes standard demographic information, household income and its components, and individual information on employment and labor earnings, as well as a set of constructed variables (e.g., net-of-tax household income, estimates of annual taxes paid by respondents, etc.) that are not provided in the original surveys (for a fuller discussion of these data, see Burkhauser, Butrica, Daly, & Lillard, 2001).
Sources of Household Income
Each country in our sample had an array of government programs that offset lost labor earnings and provided some level of income protection for spouses and survivors. Burkhauser and colleagues (2004; available at www.human.cornell.edu/pam/cnef/social.pdf) described in detail each country's programs and how they categorized them. We categorized income from public sources as either social security or other government cash transfer programs. Defining "equivalent" categories across countries was necessarily somewhat arbitrary. Here we used the U.S. Old-Age, Survivors, and Disability Insurance program as our template and compared it to programs with similar goals in our other countries. To be included in our social security category, a benefit had to be based on a quid pro quo. That is, a person's retirement or disability benefit was related to that person's past contributions (taxes) into the system. In all four countries, lost labor earnings were protected by a set of programs primarily financed by a payroll tax whose benefits were related to past labor earnings. The relationship was not actuarially fair (i.e., the present discounted value of expected benefits did not equal taxes paid) but there was a significant quid pro quo relationship. Second, program benefits were not affected by current income. That is, they were not means tested.
This is a meaningful distinction because social security programs have historically been well financed in all four of our countries, while other government cash transfer programs have had a more varied level of support. Furthermore, other cash transfer programs usually focus on lower income households while social security benefits are more evenly received across the income distribution. Here we will show the relative importance of social security programs in mitigating the economic risks of the death of a husband at various ages across our four countries.
Accounting for Taxes
Household income was defined as the sum of all income held by individuals residing in a single dwelling, and it was measured as posttax, posttransfer money income. In the U.S. literature, pretax, posttransfer family money income, including cash government transfers, is the most common yardstick used to measure economic status. However, we were interested in making cross-national comparisons. Because taxes play a much larger role in Germany, Great Britain, and Canada than in the United States, we measured household income net of income and Social Security taxes in all four countries. These tax values are available on CNEF. In addition to measuring each household's disposable income, this measure embodied the two-pronged approach governments use to redistribute income (i.e., taxes and transfers) and thus accounted for both these traditional approaches to redistribution in measuring economic well-being.
To obtain a comprehensive income measure, we summed all sources of income for all household members during a calendar year and added the cash value of food stamps in the United States (see Burkhauser, Butricia, et al., 2001, for a detailed discussion). We adjusted for inflation using the CPI-UX series for the United States and the International Monetary Fund Consumer Price Index for the other countries. Income was converted to 1996 monetary units.
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
We also reported the sum of the household income losses (third from last row) and the sum of the household income gains (second from last row) as well as the net loss or gain (last row) in Table 1. Mean household income declined by $10,646 for households of U.S. women whose husbands died between the ages of 25 and 49 (unpublished appendix tables, available by request from the authors). In Table 1, we used mean values unadjusted for household size to focus on the changes in the relative importance of various sources of income following a husband's death. Burkhauser, Giles, Lillard, and Schwarze (2002) produced similar tables for male and female survivors.
Similar patterns of decreases emerged in Table 1 across countries and age of death. When the husband died at a younger age (aged 25–49 or 50–61) his labor earnings were the most important source of lost household income in all countries. Lost earnings continued to be most important in the United States and Great Britain for men who died aged 62–69, but pension income losses were more important in Canada, and social security losses were more important in Germany. Only when a husband's death occurred after age 70 were his lost earnings relatively unimportant in all countries, and even then, they accounted for 16% of lost income in the United States.
Patterns of gains varied more across countries and age of death. When a husband aged 25–49 died, gains in private sources were most important in offsetting lost income in all but the United States. Others' labor income was most important in Canada and Great Britain, and the widow's own labor income was most important in Germany. However, increases in others' labor income could have either been an increase in the labor earnings of those in the household prior to the husband's death or a change in household composition (e.g., a widow moves into the household of her child, a widow moves in with relatives, etc.) In the United States, increases in a public source—Social Security—were most important. When a husband aged 50–61 died, only in Canada was a private source—others' labor income—most important. In the United States and Germany reduced taxes were most important, while in Great Britain social security increases were the most important factors in offsetting lost income.
When a husband died at older ages, the total amount of income gained was much smaller and the sources were more varied. In Canada and Germany, gains from private sources dominated, while in the United States and Great Britain reduced taxes led. Canada stood out from the other countries in that increases in income from a private source—others' labor earnings—dominated at all ages. Closer examination of the data revealed, however, that it was changes in household composition rather than increases in the others' labor earnings behind this increase.
Evaluating the Economic Well-Being of Individuals in Households of Different Size
To evaluate the actual change in economic well-being of a woman following her husband's death, it is important to account for changes in household size. Buhmann, Rainwater, Schmaus, and Smeeding (1988) proposed a formula that allows alternative returns to scale in household consumption:
|
|
show the sensitivity of income inequality and poverty measures to variations in the value of e but recognize that economic theory does not suggest a particular value. They point out, however, that in the international literature researchers commonly set e equal to 0.5. In Table 2, we have shown that the value chosen for e affects the inference one draws about economic well-being from measures of household income before and after the death of a spouse. We compared postgovernment household income in the year prior to the husband's death (t – 1) to income in the year following the husband's death (t + 1) using alternative values of e. As Table 2 shows, higher values of e reduced the postgovernment household income of each individual. More importantly for our purpose, mean household size-adjusted postgovernment income in t + 1 relative to mean household size-adjusted postgovernment income in t – 1 varied dramatically with the choice of e. When e = 0, the widow's household size-adjusted income fell in all countries following her husband's death. At the other extreme, when e = 1, the widow's household size-adjusted income rose for women in all of the age groups in each country. Differences in the household size-adjusted income ratios across values of e were in general greater than the differences across age groups within a country or within an age group across countries. While alternative choices of e will affect levels, they did not dramatically affect our patterns of relative benefits across countries or ages at death. In everything that follows, we used an e value of 0.5.
|
|
In all four countries, social security benefits provided substantial protection against income loss for the median woman following the death of her husband at older ages. The United States provided the highest social security replacement rate in the aged 70 and older group, but the differences across countries are small. For women whose husbands died between the ages of 62 and 69—an age range in which labor force participation of men in these countries varied quite a bit—the differences in replacement rates were far greater. In Canada, the replacement rate was 0.92, while the U.S. replacement rate was only 0.67.
In all countries the social security replacement rates were much smaller for the median widow at younger ages. The replacement rate was low for women whose husbands died at relatively young ages, largely because survivors did not automatically receive social security benefits. In the United States, for example, social security benefits are provided to women whose husbands die before age 62 only if there is a surviving child. Consequently, the median U.S. widow whose husband died between the ages of 50 and 61 receives no social security benefits.
In contrast, in the other countries, the median widow whose husband died between age 50 and 61 had higher replacement rates. In Germany, widows and widowers under the age of 45 receive 25% of their deceased spouse's covered workers pension (or estimated pension). Those aged 45 and above receive 60%. In Great Britain, widows qualify for social security benefits (National Insurance benefits) at any age as long as their husband worked. The Canadian social security program pays survivor benefits to widows and widowers in this age group immediately after the death of a covered worker. Benefits are based on the worker's accrued contributions to the Canada Pension Plan/Quebec Pension Plan. While median replacement rates in this age group in all three countries were much lower than at older ages, they were still substantially above the median value for the United States at this age.
This relative difference was less when a husband died between the ages of 25 and 49. The median replacement rate was 0.41 in the United States—these widows are much more likely to have dependent children entitling them to mothers' benefits. Thus, United States social security replacement rates substantially exceeded those of the other countries.
But as noted above, the replacement rate of total postgovernment household income provided a more complete understanding of how a woman's economic well-being changes after the death of her husband. As can be seen in Table 3, the substantial difference in replacement rates across age and in each country was dramatically narrowed. In all four countries the total income replacement rates were much higher than the social security replacement rates at younger ages and were almost always larger at older ages. The range of the total income replacement rates was much smaller across the four countries at all ages than was the range of the social security benefit replacement rates.
Differences in the Distribution of Replacement Rates Across Countries
Cross country similarities in income replacement rates found in Table 3 could mask substantial differences in these rates across the income distribution. Because median replacement rates did not vary much across the age distribution, we pooled all widows within a country to preserve sample size but otherwise used our total income replacement rate measure.
In Figure 1, we plotted the frequency distribution of total income replacement rates across six replacement rate categories. Again, we used e = 0.50 to control for changes in household size. The fraction of widows in each category was similar across countries. The modal widow in each country had a replacement rate of 0.75–0.99. The vast majority of women in each country had a replacement rate of 0.75 or more. But some women in all countries experienced larger declines in their household size-adjusted income. The United States had the highest share of widows in the two lowest replacement rate categories. About 13% of widows in the United States experienced a decline of more than one half. This fraction was almost twice the share of widows who experienced such declines in the other countries. By contrast, the share of widows with replacement rates of 0.50–0.75 was closer in the four countries.
|
|
That is, household size-adjusted income rose for women in the bottom tail of the distribution after their husbands' death in all four countries. This rise in resources may, in part, simply be regression to the mean but it was also the intended consequence of social policies in all four countries that are meant to ensure a minimum income level for widows.
Mean replacement rates tended to fall at higher income quintiles, with little differences in within-quintile replacement rates across the countries. Women in the highest income quintiles prior to their husbands' deaths experienced the greatest fall in their relative income. This decline at the upper tail of the distribution was, again, likely due, in part, to regression to the mean but it was also the intended consequence of social security and other government transfer programs that target benefits to the middle and lower ends of the income distribution.
Hence, while the overall replacement rates of women in the year following the death of their husbands varied more in the United States than in other countries, the bulk of the dramatic drops in replacement rates observed in the United States and, to a lesser degree, in other countries came from women in households in higher income quintiles.
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
In all four countries, mean household income for women, unadjusted for household size falls following the death of their husband. The main source of this decline is the same in all countries—his lost labor earnings at younger ages, his lost social security or pension income at older ages. Patterns in the sources of offsetting increases are less clear-cut. In general, increased income from private sources (labor earnings of the widow or other household members), dominates at younger ages, while income from public sources (social security or reduced taxes) dominates at older ages. Given this variation in offsetting income gains across the husband's age at death and country, it is important to focus on how the total household income of a woman changes after the death of her husband when making cross-national comparisons. Past studies that focus solely on differences in average social security replacement rates across countries to gauge the relative protection against household income drops for women following their husbands' deaths will not only understate that protection but will disproportionately do so for countries where social security programs play a smaller role in providing it.
Furthermore, we show that because the number of people living in a woman's household systematically falls following the death of her husband, comparisons of her economic well-being must control for this change. Replacement rates vary more across assumptions about household returns to scale than they do across countries, holding returns to scale constant.
Using an e equal to 0.50, we show that the median woman's social security replacement rate is uniformly high when her husband dies at ages 70 and older in all four countries, much more varied when he dies between ages 62 and 69, and much lower when he dies at younger ages in all four countries. But this variation across age and country is reduced substantially once a broader household size-adjusted income replacement rate measure is used.
Measures of replacement rates such as the mean or median can obscure substantial differences in the distribution of replacement outcomes across countries. We show that the distribution of replacements rates is greater in the United States than in other countries and that U.S. women are more likely to experience a greater than 50% decline in their household size-adjusted income following the deaths of their husbands than widows in the other countries. This suggests that greater reliance on non-government-based benefits in the United States results in greater uncertainty for U.S. women with respect to income replacement, including a higher risk of substantial drops in income, following their husbands' death. However, this greater risk appears to be borne primarily by women living in upper income households prior to their husbands' deaths in all countries, but especially in the United States. The mean replacement rates for women in the lower tail of the distribution who become widows in all four countries are between 1.30 and 1.54.
The main finding of our study is that across countries with widely different mixes of public and private support for widows, the change in the economic well-being of women following the deaths of their husbands is remarkable similar. Differences in outcomes across the husband's age at death appear to be greater than those across country of residence.
Our findings suggest that women in these four OECD countries on average experience similar changes in their economic well-being following the death of their spouse but achieve this end through different mechanisms. Women in low-income households are disproportionately protected in all four countries by government programs that if anything, on average, increase their household size-adjusted income following their husbands' deaths. Dramatic falls in household size adjusted income in all four countries are more likely to occur among women who lived in higher income households prior to the deaths of their husbands.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication July 9, 2004. Accepted for publication February 4, 2005.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. D. Fisher, D. S. Johnson, J. T. Marchand, T. M. Smeeding, and B. B. Torrey No Place Like Home: Older Adults and Their Housing J. Gerontol. B. Psychol. Sci. Soc. Sci., March 1, 2007; 62(2): S120 - S128. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
