 |
University of Pennsylvania Library
Expanded Academic ASAP |
Journal of Economic Issues, March 2002
v36 i1 p17(24)
Explaining the gender poverty gap in developed and transitional economies. (economic research and data) Steven Pressman. Full Text: COPYRIGHT 2002 Association for Evolutionary Economics
As economies throughout the world experience large and wrenching changes, poverty has increasingly become a problem in country after country. This is true regardless of whether these changes result from globalization, the economic transition from socialism to capitalism, increasing marketization and privatization, or some other major economic transformation (Aslanbeigui, Pressman, Summerfield 1994; Funk and Mueller 1993; Moghadam 1996).
A concomitant, disturbing aspect of rising poverty throughout the world is that poverty has increasingly become feminized--women are much more likely than men to be poor. This phenomenon was first noticed in the United States (Pearce 1978, 1989; Pressman 1988), but more recently the problem of the feminization of poverty has become an international concern as well (Casper, McLanahan, and Garfinkel 1994; Pressman 1998; Wright 1995).
This article employs the Luxembourg Income Study (LIS) to compare poverty rates for female-headed households (FHHs) with poverty rates for other households in a number of developed and transitional economies. It then seeks to explain why, in some countries, female-headed households are so much more likely to be poor compared with other families.
The next two sections, respectively, describe the LIS and discuss some of the problems encountered in measuring poverty. The paper then computes poverty rates in individual countries for female-headed households and for all other households using the LIS database. Given the problems associated with measuring poverty, I present several estimates of poverty for both types of household. Two sections then look at a couple of theoretical explanations for the gender poverty gap--human capital theory and a Keynesian approach that emphasizes the importance of fiscal policy as an antipoverty tool. The last section summarizes the main findings and draws some policy conclusions.
The Luxembourg In come Study
The Luxembourg Income Study began in April 1983 when the government of Luxembourg agreed to develop, and make available to social scientists, an international microdata set containing a large number of income and socio-demographic variables. Until that time, most cross-national studies of income distribution and poverty suffered because the national data that they used would define key terms differently. Most importantly, the notion of income itself was defined and measured differently in different countries.
One goal in creating the LIS database was to employ common definitions and concepts so that variables are measured according to uniform standards across countries. As a result, researchers can be confident that the coss-national income data that they are analyzing, and the socio-economic variables that they are examining, have been made as comparable as possible.
By 2001, the LIS contained information on twenty-five nations-Australia, Austria, Belgium, Canada, the Czech Republic, Denmark, Finland, France, Germany, Hungary, Ireland, Israel, Italy, Luxembourg, the Netherlands, Norway, Poland, Russia, the Slovak Republic, Spain, Sweden, Switzerland, Taiwan, the United Kingdom, and the United States. Negotiations are currently under way with Japan and several other countries to have their income data added to the US. Data for each country was originally derived from national household surveys similar to the US Current Population Reports, or (in a few cases) from tax returns filed with the national revenue service.
Currently four waves of data are available for individual countries. Wave I contains datasets for countries for one particular year in the late 1970s or early 1980s. Wave II contains datasets for some year in the mid 1980s. Wave III contains datasets for the late 1980s and early 1990s. Wave IV (currently in the process of being "Lissified" and put online) contains country datasets for the mid 1990s. Finally, historical data from the late 1960s and/or early and mid 1970s are available for a few countries.
LIS data are available for more than 100 income variables and nearly 100 socio-demographic variables. Wage and salary incomes are contained in the database for households as well as for different household members. In addition, the dataset includes information on in-kind earnings, property income, alimony and child support, pension income, employer social insurance contributions, and numerous government transfer payments and in-kind benefits such as child allowances, Food Stamps, and Social Security. There is also information on five different tax payments. Demographic variables are available for factors such as the education level of household members; the industries and occupations where adults in the family are employed; the ages of all family members; household size, ethnicity and race; and the marital status of the family or household head. (1)
This wealth of information permits researchers to do cross-national studies of poverty and income distribution and to address empirically questions about the causes of poverty and changing income distribution. It also allows great flexibility in how poverty is measured and how the middle class is defined.
Poverty Calculations Using the LIS
How to calculate poverty rates has been a matter of considerable controversy in the United States since the 1960s. The method currently employed was developed by Mollie Orshansky (1965, 1969) of the Social Security Administration in the early 1960s. Orshansky first calculated the cost of the minimum amount of food that different types of families would need during one year. Since Agriculture Department surveys found that families spent about one-third of their after-tax income on food, the cost of an economy food plan for families of different types and sizes was multiplied by 3 in order to arrive at poverty lines for each family type. Poverty lines for each type of family are increased annually with the increase in consumer prices. Poverty lines thus represent a real standard of living for families of a particular type and size that remains invariant over time. The poverty rate is calculated as the percentage of US families whose income, before taxes, falls below the poverty line (for their family size and t ype) in a given year.
The Orshansky methodology for computing poverty rates has been criticized on a number of grounds. Harrell Rodgers (2000) argued that the minimum food requirements for a family were designed for short-term emergency situations only and would not be able to meet the nutritional needs of a family for an entire year. Since the food budgets used by Orshansky were 80 percent of what was necessary to provide a nutritional diet for the entire year, Rodgers argued that therefore the Orshansky poverty lines are 80 percent too low. John Schwarz and Thomas Volgy (1992) argued that food consumption has fallen from one-third to one-fifth of family spending, so current poverty lines should be based upon a food multiplier of 5 rather than 3. This would raise poverty lines by two-thirds and also make poverty-level incomes consistent with what public opinion surveys have found to be the amount of income people believe that a family requires to escape poverty. Taking a slightly different tack, Harold Watts (1986) argued that in the early 1960s the poor paid no income taxes and virtually no Social Security taxes. But in the 1970s and 1980s, poor families faced a considerable tax burden. Calculating poverty based upon pre-tax incomes ignores the fact that pre-tax incomes can buy less than a comparable or real pre-tax income from the 1960s. Although this point was undoubtedly a good one during the late 1980s, it may no longer be valid given sharp increases in the earned income tax credit during the 1990s.
The most frequent criticism of the Orshansky methodology, however, is a philosophical one rather than a technical one. Orshansky developed an absolute measure of poverty. Poverty is supposed to measure the minimum income necessary for a family to survive during the course of a year. But several authors (Dunlop 1965; Fuchs 1965; Rainwater 1974; Ruggles 1990) have argued that human beings are social animals, and so the standard of what is minimally necessary must vary from time to time and from place to place. For example, private baths, telephones, and television sets were not necessities in the 1920s or the 1930s, but they are necessities today. Likewise, childcare was not a necessity in the 1950s or 1960s. But as more and more families have two earners, or just one adult heading the household, childcare has become an important family expenditure. For this reason, many authors contend that poverty should be measured in relative terms, as some fraction of the average or median income at a particular time and i n a particular place. (2)
Additional problems arise when employing real, absolute poverty lines in cross-national studies. First, whenever we compare two countries with different national currencies we have to compare incomes that are measured in different units. Consequently, some way has to be found to convert one income into an equivalent income denominated in some other currency. Employing the actual exchange rates between two currencies at the time is a first, logical approach to this problem. But exchange rates vary considerably from day to day, from month to month, and from year to year; and they vary for speculative reasons that have nothing to do with changes in the relative value of the two currencies or the relative living standards in the two countries.
One attempt to get around this problem is to look at purchasing power parity (PPP). The basic idea behind this notion is rather straightforward. Some goods are sold virtually everywhere throughout the world; by comparing the cost of these goods from country to country we can obtain a good measure of the real value of two different currencies. If a McDonald's hamburger sells for $1 in the United States and 100 yen in Japan, then $1 and 100 yen should represent equivalent real incomes. According to the purchasing power parity theory, regardless of the exchange rate between the dollar and the yen, $1 = 100 yen should be used when comparing real incomes in the United States and Japan.
Unfortunately, serious problems with the notion of purchasing power parity make its use problematic when attempting to compare equivalent living standards in different nations. First, purchasing power parity assumes that domestic prices reflect only domestic costs. Domestic spending patterns thus become irrelevant. Yet in the real world, demand, as well as costs, is important in determining the prices of different goods.
Consider again the McDonald's hamburger. American diets include large quantities of meat, especially ground beef. Furthermore, few American families have an adult at home during the day to prepare the family dinner. As a result, the family is more likely to go out to eat, and fast food restaurants have become a popular choice for the family dinner. Contrast this now with Japan, where the family diet contains more fish and less beef and where the family dinner is likely to be served at home because someone stays home to prepare dinner. Given these cultural and socio-economic differences, demand for McDonald's hamburgers will be relatively greater in the United States than in Japan.
As a result, the price of a hamburger will be relatively greater in the United States than other goods, and the price of a hamburger in Japan will be relatively less than the other goods bought by a typical family. Using McDonald's hamburger prices (in part) to determine purchasing power parity will thus understate the relative income (and standard of living) of the Japanese family and overstate the relative income (and standard of living) of the American family.
A second problem concerns the notion of purchasing power parity itself. The standard empirical estimates of purchasing power parity were made in the late 1980s (OECD 1989) and early 1990s (Summers and Heston 1991). Studies that use purchasing power parity to compare real incomes across nations in other years typically adjust these figures for the inflation experienced within each country since the early 1990s. This procedure assumes that purchasing power parities remain the same over time. But there is no guarantee that this will be so. In theory, productivity growth differentials among countries should also affect living standards in different countries over time. Merely inflating PPP to reflect inflation differentials ignores this important cause of real income growth and changes in relative incomes across nations. Moreover, these studies assume that inflation is measured accurately in each nation, an assumption that does not hold for the United States (3) and likely does not hold elsewhere. Unless inflatio n is mismeasured everywhere to the same extent, estimates of PPP will get worse the further we move from the base year computations.
Finally, even if purchasing power parity were an acceptable means of comparing disposable incomes across nations, there are still problems with using PPP to convert disposable incomes into equivalent living standards. Each country is different in terms of how it subsidizes goods like health care, housing, and education. Equivalent disposable incomes (adjusted using purchasing power parity) will therefore not measure equivalent levels of consumption. Put another way, purchasing power parity was meant to allow a comparison of average living standards. This is not the same as low-income or poverty-level living standards. Differences in public subsidies for the poor will make a big difference in living standards but will nor be reflected in different measures of income. As Timothy Smeeding, Lee Rainwater, and Gary Burtless (2000, 7) note, "in countries where in-kind benefits are larger than average, absolute poverty rates may be overstated because citizens actually face a lower effective price level than is refle cted by OECD's estimates of PPP."
Because of the arguments in favor of a relative notion of poverty and because of the many problems that arise when comparing real incomes and real living standards across nations, most LIS studies have employed a relative notion of poverty. A relative notion of poverty means that a household is poor if its income does not enable a standard of living that approximates what the average household is able to enjoy. LIS studies usually define poverty lines as 50 percent of median adjusted family or household income, after taxes, within a country for a specified year. Adjusted family income controls for the different sizes of different families and recognizes that $20,000 goes a lot further in a family of two than in a family of five. Most empirical studies using the LIS take the income needs of a second adult to be 70 percent of the income needs of a first adult and the income needs of children as 50 percent of the first adult. (4) These weights are similar to the implicit weights in the official US definition of poverty, as well as the explicit family equivalence scales used by the OECD.
Estimating the Gender Poverty Gap
Following the standard LIS methodology for computing poverty, table 1 presents poverty rates for countries currently in wave III of the LIS. Poverty rates are calculated for households headed by a single female and also for all other households. The last column of each table shows the difference between the poverty rate for female-headed households and the poverty rate for all other households.
For wave III, the difference between these two poverty rates (the gender poverty gap) ranges from about -2 percent (for Poland), meaning that poverty rates for female-headed households are 2 percentage points lower than other poverty rates for other families, to about +18 percent (for the United States), meaning that poverty rates for female-headed US households are 18 percentage points higher than poverty rates for other US households. For wave III datasets, the gender poverty gap averages 4.4 percent (unweighted).
A number of studies of the poverty gap (e.g., Casper, McLanahan, and Garfinkel 1994; Christopher et al. 1999) have looked at the ratio of poverty rates for female-headed households and other households rather than differences in these two rates. This approach may result from the habits of labor economists, who typically examine and study earnings ratios. Ratios are an acceptable means of comparison when looking at two different income levels and where the key issue is how much more men make or how much less women make. But looking at ratios of poverty rates is objectionable on two counts.
First, poverty rates are supposed to represent the probability that a family is poor. When comparing the poverty rate for female-headed households with the poverty rate for other households we usually want to know how much more likely it is that female-headed households will be poor. Differences in poverty rates give us this important information; ratios do not.
Second, with ratios of rates, small percentage point differences can lead to large ratio differences that can be misleading when we attempt to interpret the numbers or analyze the causes of the gender poverty gap. For example, if 1 percent of other households are calculated to be poor and 2 percent of female-headed households are poor (essentially the results for the Czech Republic), ratios focus on the fact that women are twice as likely to be poor as men. But given the reporting errors in survey data, plus the somewhat arbitrary nature of any equivalence scales and poverty lines, the difference between a poverty rate of 1 percent and a poverty rate of 2 percent is quite small and may not be robust or significant. Differences in poverty rates make this fact clear; poverty rate ratios do not. To the contrary, with ratios, a poverty rate for female-headed households of 20 percent and a poverty rate for other households of 10 percent (essentially the case of Canada) seem just as bad as the 2 percent and 1 perc ent case because it also yields a ratio of 2. But it should be clear that women in the Czech Republic are relatively better off than the women in Canada. To make this point it is necessary to focus on poverty rate differences rather than on ratios of poverty rates.
The gender poverty gaps reported in table 1 divide naturally into three different groups. First, there are countries with very small and insignificant gender poverty gaps. For Belgium (1992), the Czech Republic, Hungary, Italy, Luxembourg, the Slovak Republic, and Spain there is virtually no difference between poverty rates for female-headed households and for other households; and in two countries (Poland and Switzerland) poverty rates for female-headed households are slightly below poverty rates for other households. Second, eleven countries (Belgium (1988), Denmark, Finland, France, Germany, Israel, the Netherlands, Norway, Sweden, Taiwan, and the United Kingdom) have slightly higher FHH poverty rates. For these counties the gender poverty gap ranges from about 2 percentage points (Norway) to a little more than 6 percentage points (United Kingdom). Finally, four countries have extremely large gender poverty gaps. In Canada, the gender poverty gap is almost 10 percentage points; and in Australia, the gende r poverty gap exceeds 11 percentage points. Even worse performers are Russia, with a gender poverty gap of almost 15 percentage points, and the United States, where the gender poverty gap approaches 18 percentage points.
Studies using wave II of the LIS and examining female-headed households and poverty (Wright 1995; Pressman 1998) have found a similar pattern across different nations. Counties with a small gender poverty gap in one year tend to have a small gender poverty gap in the other year. Australia, Canada, and the United States do badly in both time periods (there is no Russian database for wave II) while Italy, Luxembourg, and Poland do well in both time periods. Countries falling in the middle ground in one time period also tend to fall in the middle ground in other time periods. There thus appears to be relatively little change from one wave or time period to the next when it comes to the rank ordering of different counties. Put another way, cross-national differences in poverty are much greater in one time period than intertemporal differences in poverty in one nation. This seems to indicate that the national tendencies, habits, and practices regarding women's employment and wages, as well as national policies des igned to assist FHHs, are more important in determining gender poverty gaps than are the economic or institutional changes that occur within countries over time. Policies and institutions within any country change slowly; but policy differences among nations are likely to be great, as they arise from different historical, cultural, and socio-economic traditions (Esping-Anderson 1990).
One interesting and related question is what has happened in transitional economies as a result of sharp reductions in the role of government in economic activity and giving greater sway to the market. Wave II datasets provide a benchmark for before the transition process; wave III datasets give a snapshot of the very beginning of the transformation process. These waves show only small gender poverty gaps. When waves IV and V datasets finally come online we will be able to see the impact of the full transition process. Other evidence of the impact of this transformation on women (Funk and Mueller 1993; Aslanbeigui, Pressman, and Summerfield 1994) must make one rather pessimistic about gender poverty gaps for these nations as the transition process moves forward.
A Sensitivity Analysis
Given the problems with survey data, as well as the problems with defining poverty that we discussed in the second section, one important question that needs to be addressed is how much hinges on the decisions that are made when measuring poverty. This section attempts to answer this question by means of a sensitivity analysis.
Table 2 uses wave III of the LIS and the standard equivalence scales for deriving adjusted family income. It differs only by using a slightly different definition of poverty. In table 2, households are taken to be poor if the family income falls below 40 percent of mean adjusted household income (rather than the usual 50 percent). Using this alternative poverty definition, the stylized facts presented in the previous section do not change very much. The United States still has the greatest problem of feminized poverty, although the poverty rate for FHHs and the gender poverty gap are both a bit lower due to the lower poverty line. Moreover, the same four countries (Australia, Canada, Russia, and the United States) still have the largest gender poverty gaps and the highest poverty rates for FHHs. Likewise, most of the countries with low gender poverty gaps using a 50 percent-of-median-income poverty line also have low or no gender poverty gaps when defining poverty as having less than 40 percent of adjusted mean family income. Poland has the lowest gender poverty gap in both instances. And the same set of countries (the Czech Republic, Hungary, Italy, Luxembourg, the Slovak Republic, Spain, and Switzerland) have negligible gender poverty gaps in both time periods. The only major change in our results is that a number of countries with moderate gender poverty gaps when we set a higher poverty line now have negligible poverty gaps. In the United Kingdom, for example, the gender poverty gap falls from 6.3 percent to 0.1 percent while in Israel the gender poverty gap falls from 4.8 percent to 0.9 percent. Overall, the correlation between the gender poverty gap using a poverty line set at 50 percent of median (adjusted) income and the gender poverty gap using a poverty line set at 40 percent of median (adjusted) income exceeds 30 percent.
This change also does not seem to have much impact on our story about women and poverty. The main change here is that poverty rates are higher when we assume that each child has the same income needs as the first adult in the family (rather than needs that are one-half of that). This pushes down adjusted household income and results in many more households with children that get categorized as poor.
Nonetheless, the trans-national story about women and poverty changes very little with our alternative measure of household income. Again, the United States has the largest gender poverty gap of all countries examined as well as the highest poverty rate for FHHs. Likewise, the same set of four countries (Australia, Canada, Russia, and the United States) still have the largest gender poverty gaps and the four highest poverty rates for FHHs. At the other end of the spectrum, Poland continues to have the lowest (negative) gender poverty gap, while the same set of countries generally tend to have the low gaps. The correlation between the gender poverty gap estimated in table and the gender poverty gap on this alternative definition of adjusted household income is 70 percent.
Possible Causes of the Gender Poverty Gap
Theoretical explanations for different gender poverty gaps among nations can generally be divided into three broad categories.
First, neoclassical economic theory attributes wage differentials primarily to productivity differences. Someone who is more valuable to a firm will be paid more than someone who contributes less to firm revenues. Human capital theory (Becker 1993; Mincer 1974; Schultz 1961) has taken this idea one step further and attempts to explain wage rates based upon the education and experience level of the individual. The insight of human capital theory is that more educated workers will be more productive and will thus receive higher pay. Likewise, more experienced workers will be more productive and should also be paid more money than less experienced workers.
This theory can be applied to gender differences in earnings. If the education level of women who head households is much less than the education level of men who head married-couple families, we should expect the earnings and income of female-headed households to be much lower. Therefore, we should expect the gender poverty gap to be larger. Human capital theory traditionally proxies experience by looking at the age of the individual worker. Adopting this approach, we can look toward the age of household heads in order to explain the gender poverty gap. If female heads of house are younger than the men who head other households, then according to human capital theory the wages of these women should be lower than the wages of the men heading other families. Again, with lower relative wages, women should experience relatively greater poverty.
A second possible explanation for gender poverty gaps focuses on gender discrimination. Societal views about the worth of women and the work they do have led to a situation in which women receive lower pay than men, even when they do the same work and provide the same benefits to the firm. Another take on the discrimination angle is the claim that occupational sex segregation has put women into a set of jobs with low pay (Bergmann 1974; Sawhill 1976; Strober and Arnold 1987) or a set of industries (the service sector) that pay poorly (Northrop 1990). Obviously, the greater the discrimination against women in the marketplace, the lower the earnings of women relative to men and the higher the gender poverty gap will be.
Finally, government fiscal policies can affect the gender poverty gap in two main ways. Within a particular country, spending programs, or social transfer payments, can be geared more toward husband-wife households or more toward female-headed households. The more that social programs give to female-headed households relative to other households, the lower the gender poverty gap should be. Meager social insurance for female-headed families in the United States has been cited (Rodgers 2000; Zopf 1989) as a major cause of high poverty rates for female-headed households. This factor also may contribute to different national gender poverty gaps.
In addition to spending money, governments also collect taxes. Poverty calculations are usually made using after-tax, rather than before-tax, incomes. If government tax policy in one country favors married-couple households over single tax-paying units, female-headed households will do relatively worse after taxes than other households, and we should see a greater gender poverty gap.
Testing Alternative Theories of the Gender Poverty Gap
This section examines two of the three theories discussed above. We first explore how human capital considerations affect the gender poverty gap. Then we look at the impact of fiscal policy on the gender poverty gap. Given the usual time and space constraints, tests of the feminist approach, which looks to discrimination as the cause of the gender poverty gap, will be left for future research.
Table 4 examines one part of the human capital explanation for the gender poverty gap. It does so by raising the following empirical question--to what extent is the poverty of female-headed households due to the relative youth of the household head? To answer this question we take aggregate poverty rates as a weighted average of the poverty experienced by households whose heads fall into different age brackets. To derive the figures appearing in table 4, six age groups were distinguished--(1) under 30, (2) 30-39, (3) 40-49, (4) 50-59, (5) 60-69, and (6) over 69. For most countries, and especially for most developed countries, this yields six groups of relatively equal size for other households.
Poverty rates for each of these six age groups were calculated for both FHHs and for other households in each individual country. Table 4 recalculates poverty rates for FHHs as the weighted average of the (constant) poverty rates for each age group, assuming that female-headed households had the same age distribution as other households. The results of this computation are shown in column 3. Column 4 shows the change in poverty for FHHs in each country due to the actual age distribution of female household heads.
One reason age is unimportant is that in many countries FHHs are more likely to have older heads due to the greater life expectancy of women. And in virtually all countries older households are less likely to be poor due to the generous provision of retirement income to the elderly.
To take just one striking example, let us consider the Australian (1989) case. Younger FHHs (under 40) had around a 28 percent chance of being poor. In contrast, only around 15 percent of middle-aged FHHs (40-59) were poor and less than 10 percent of FHHs with an elderly head (60+) were poor. Since women live longer than men, there are proportionately more older FHHs than there are older other households in Australia. In 1989, about 21 percent of other households were 60 and over, but more than 36 percent of FHHs were 60 and over. The fact that FHHs are more likely to be older meant that the poverty of FHHs in Australia was lower by about 1.4 percentage points. If FHHs had had the same age distribution as other households, their poverty rate would have been 20.5 percent (rather than the actual 19.1 percent).
Table 5 looks at the other part of the human capital explanation for the gender poverty gap. It addresses the extent to which the poverty of FHHs is due to their lower levels of education. As noted above, we can regard poverty rates for FHHs as a weighted average of the poverty experienced by families with different characteristics. Here the relevant feature is educational levels rather than age.
The LIS does not have standard educational achievement classifications that are used in all country databases. But for each country, education categories are pretty much defined the same way for FHHs and for other households. In those few instances where categories were not identical, some minor recoding was needed. In these cases, only a very small percentage of households (less than one-half of one percent) had to be recoded, so recoding decisions should not affect the overall results. In a couple of cases (Israel and the United Kingdom) education data were available only by the age at which the individual last attended school; since this was not likely to be a very close proxy for educational attainment, these countries were excluded from table 5. For Russia, the recoding task was too large (since educational attainment categories differ substantially by gender) and recoding decisions would likely affect the final results because of the large number (thirty) of education categories in the Russian LIS datab ase. For this reason, Russia was excluded from the analysis of education and the gender poverty gap in table 5.
Column 3 of table 5 shows the poverty rates for FHHs under the assumption that they had the same educational distribution as other household heads. Column 4 of table 5 then shows the increase in poverty for FHHs that is due to the lower educational attainment of the household head.
Again, the results of our analysis do not lend much support to the human capital explanation for the gender poverty gap. In eight cases out twenty (Belgium, Denmark, France, the Netherlands, Norway, the Slovak Republic, Sweden, and Switzerland), FHHs actually were less likely to be poor because of their relatively high education level. In three more cases (the Czech Republic, Finland, and Luxembourg), educational attainment made virtually no difference at all. In contrast, for only six countries (Germany, Hungary, Italy, Poland, Spain, and the United States) did educational deficiencies raise the poverty rate of FHH by more than 1 percentage point, and in only one of these (the United States) did it raise the poverty rate of FHH by more than 2 percentage points. The striking result of table 5 is that educational levels matter very little. On average (unweighted), lower education levels for women raised the poverty rate of FHH by one-half of a percentage point. Consequently, educational deficiencies by women c an explain only a little more than 10 percent of the gender poverty gap that we estimated in table 1.
While human capital theory does not help explain the gender poverty gap, Keynesian theory does considerably better. The Keynesian argument is that income distribution in general, and poverty rates in specific, depend on fiscal policy decisions made by the government. On the Keynesian view, the bigger the government safety net, and the broader and deeper (or more generous) the net, the lower the national poverty rate (see Pressman 1991). Because FHHs are more likely to be poor without any government assistance, the more generous the level of government transfer payments, the lower the gender poverty gap.
Tables 6,7, and 8 allow us to examine this theory empirically. Table 6 assumes no government benefits and that no taxes are imposed on earned incomes, It also assumes that there are no private transfers among households, such as child support or alimony payments. As a result, factor income (wages, interest, dividends, rent, and so on) is taken to be total household income. Using a poverty calculation analogous to our method in table 1--not receiving at least 50 percent of median (adjusted) household factor income--gives us enormously high poverty rates. This is especially so for FHHs, where poverty rates typically exceed 50 percent and reach as high as 70 percent. This, no doubt, stems from the fact that FHHs usually have only a single adult earner. When women head families with children, they may have childrearing responsibilities that limit the number of hours they can work each day and each week and, therefore, the sorts of jobs they could hold. Moreover, women typically earn less than men, and so they suf fer a further disadvantage. The result is that FHHs have low factor incomes and high poverty rates compared with other households.
The gender poverty gap in table 6 is also quite striking; it averages (unweighted) more than 30 percent when both fiscal policy and private transfers are excluded. This contrasts with an average poverty gap of 4.4 percent when taking into account the impact of government spending and taxes as well as private transfers (table 1). Also striking is the fact that when we look at just factor incomes, the US gender poverty gap lies a bit below the (unweighted) average gender poverty gap for all countries in table 6. Likewise, the poverty rate of FHHs in the United States is below the (unweighted) average for all LIS countries in wave III. What is true of the United States is also true of Canada and Russia, two of the other four countries with very high gender poverty gaps. Looking at only factor incomes, both have below average poverty rates for FHHs and below average gender poverty gaps. Canada, in fact, has the second lowest gender poverty gap and the third lowest poverty rate for FHHs when looking at just factor income. Australia, our last poorly performing country, has a below average poverty rate for FHHs but a gender poverty gap that is slightly above average.
Table 7 adds two important private transfers to factor income--child support and alimony payments. Poverty rates in each country are again computed based on whether adjusted household income falls below 50 percent of median adjusted household income, but here household income is taken to be the sum of factor income and private transfers. The main result of table 7 is that private transfers seem to make very little difference. Adding these payments to household income reduces poverty rates for FHHs a little and reduces the gender poverty gap a bit (each goes down by half a percentage point), but both these rates remain very high.
Table 8 looks at gross income before taxes. Here we include all government benefits in family income figures as well as all private transfers. Poverty rates here are calculated as the fraction of families whose gross income (adjusted for family size) falls below 50 percent of median (adjusted) gross income. As before, the poverty gap is the difference between the poverty rate for FHHs and the poverty rate for other households.
The first striking thing about table 8 is the sharp drop in poverty due to various government transfer payments. Government expenditures reduce the poverty rate of FHHs by about two-thirds and also reduce the poverty rate of other households by about two-thirds.
These declines, it is important to note, are not the result of just adding more types of income (and therefore more income) to each household. Poverty rates are computed based on a poverty line that is 50 percent of (adjusted) gross income; since gross income exceeds factor income for virtually every family, median income rises and the poverty line rises as well. In fact, if gross income rose proportionately to factor income for every household, there would be no change in poverty rates at all. So the sharp decline in poverty that we see in table 8 must be due to the equalizing effect of the added government expenditures.
The second thing to notice about the last column of table 8 is the sharp drop in the gender poverty gap. On average (unweighted), government expenditures reduce the gap by nearly 24 percentage points--from 30.7 percent to 7.2 percent--or by more than two-thirds. Moreover, there is a sharp drop in the gender poverty gap in virtually every country. Among the major exceptions here are the United States, Australia, Canada, and Russia, where fiscal expenditures do relatively little to lower the gender poverty gap. As a result, these countries have gender poverty gaps of between 15 to 20 percent when measured using (adjusted) family gross income.
Moving from the last column of table 8 back to table 1 enables us to see the impact of taxes on poverty and the gender poverty gap. On average (unweighted), the tax system reduces the gender poverty rate for FHHs by 4.4 percentage points and the poverty rate for other households by 1.5 percentage points. Thus the poverty gap falls by 2.8 percentage points due to taxes.
Table 9 pulls together the results of our analysis in this section. It starts where most families start, with factor incomes, the money earned from market activities. Had this been the only source of income for families, the gender poverty gap would have been nearly 30 percent in most countries. Adding private transfers (child support payments and alimony) slightly lowers the gender poverty gap in virtually all nations and on average it slightly lowers the gender poverty gap. Most of the action in lowering the gender poverty gap, however, occurs as a result of fiscal tax and transfer policies, especially the latter. Countries that provide large social transfers generally experience the largest reductions in the gender poverty gap (see table 10). And countries without a fiscal policy that aids or favors low-income FHHs generally have high gender poverty gaps and experience little reduction from the high gender poverty gaps that result when looking at only factor incomes (Pressman 1998, table 3).
Summary and Conclusions
This paper has examined the gender poverty gap in a wide set of countries using wave III of the Luxembourg Income Study. It finds that the gender poverty gap was relatively large in some countries during the late 1980s and early 1990s, was moderate in other countries, and was very low or negative in yet other countries. These results were fairly robust with different attempts to measure poverty.
Next, the paper sought the causes of different gender poverty gaps across countries. It found the human capital explanation wanting. Neither age nor education can explain much of the gender poverty gap. A more Keynesian explanation for the gender poverty gap proved more fruitful. Fiscal policy is able to explain a large proportion of the gap. Excluding government, the poverty rate of FHHs and the gender poverty gap are both very large in all countries. Some nations use fiscal policy aggressively to assist low-income households; other nations spend less money to assist low-income households. Nations that do more have much lower poverty rates for FHHs and much lower gender poverty gaps. In contrast, nations like Australia, Canada, Russia, and the United States fail to employ fiscal policy aggressively in an attempt to assist poor families; as a result they wind up with large poverty rates. These counties also do not focus their fiscal assistance on FHHs, and so these nations have high poverty rates for FHHs and large gender poverty gaps. The results of this paper thus support other studies which have found that the type of welfare state and the character of social policies and spending programs affect poverty rates for single mothers (Duncan and Edwards 1997; Lewis 1997).
This analysis also leads to two policy conclusions. First, attempts to improve the relative economic condition of poor FHHs by developing the skills and improving the education level of women are not likely to be effective. The reason for this is that we found human capital factors seem to have very little effect on the gender poverty gap. Human capital policies are thus likely to result in large costs but have small benefits in terms of reducing the poverty of FHHs. Also, there are likely to be pragmatic difficulties with such an approach. From a political perspective, it will be hard to justify human capital spending that would disproportionately benefit women over men. Second, fiscal policy must focus more on the problems facing FHHs and spending must be directed more to low income FHHs. If countries are to effectively deal with problems of feminized poverty, then fiscal policy must be used to assist FHHs.
Notes
(1.) For more information about the Luxembourg Income Study, and for information on how to access the LIS databases, see Smeeding et al. 1985.
(2.) In response to this it is sometimes argued chat migration from low-income countries to high-income countries shows that absolute incomes are more important than relative incomes. The case is basically that people move from low-income countries to improve their absolute standard of living although in their new environment they are at the bottom of the income distribution. Moreover, there is no reverse migration of people moving from high-income countries in order to improve their relative status. There are, however, a number of problems with this argument. While it is true that some people migrate to increase their absolute incomes (while lowering their relative income), many people do not do this. If migration is supposed to be evidence that people care about absolute incomes, then by the same token each failure to migrate should be taken as evidence that people care more about relative incomes-they have decided to remain big fish in a small pond rather than migrating and increasing their absolute incom es. In addition, there is a good deal of empirical evidence that people in developed countries do care a great deal about relative incomes. Robert Frank does an excellent job of summarizing this literature in Luxury Fever (Frank 1999).
(3.) The Boskin Commission Report of December 1996 makes the case that inflation in the United States is considerably overstated. The report, with critical commentary, appears in Baker 1998.
References
Aslanbeigui, Nahid, Steven Pressman, and Gale Summerfield. Women in The Age of Economic Transformation. New York & London: Routledge, 1994.
Baker, Dean, ed. Getting Prices Right: The Debate Over the Consumer Price Index. Armonk, N.Y.: M. E. Sharpe, 1998.
Becker, Gary. Human Capital, 3rd ed. Chicago: University of Chicago Press, 1993.
Bergmann, Barbata. Occupational Segregation, Wages, and Profits when Employers Discriminate by Race/Sex." Eastern Economic Journal 1(1974): 103-110.
Casper, Lynne, Sara McLanahan, and Irwin Garfinkel. "The Gender-Poverty Gap: What Can We Learn from Other Countries?" American Sociological Review 59 (August 1994): 594-605.
Christopher, Karen, Paula England, Katherin Ross, Timothy Smeeding, and Sara McLanahan. "The Sex Gap in Modern Nations: Single Motherhood, the Market, and the State." LIS Working Paper, 1999.
Duncan, Simon, and Rosalind Edwards. Single Mothers in an International Context. London: UCL Press, 1997.
Dunlop, John T. "Poverty: Definition and Measurement." In The Concept of Poverty. Washington, DC: Chamber of Commerce of the United States of America, 1965.
Esping-Andersen, Costa. The Three Worlds of Welfare Capitalism. Princeton: Princeton University Press, 1990.
Frank, Robert. Luxury Fever. New York: Free Press, 1999.
Fuchs, Victor R. "Toward a Theory of Poverty." In The Concept of Poverty. Washington, DC: Chamber of Commerce of the United States of America, 1965.
_____. Women's Quest for Economic Equality. Cambridge: Harvard University Press, 1988.
Funk, Nannete, and Magda Mueller, eds. Gender Polities and Post-Communism: Reflections from Eastern Europe and the Former Soviet Union. New York: Routledge, 1993.
Lewis, Jane, ed. Lone Mothers in European Welfare Regimes. London: Jessica Kingsley, 1997.
Mincer, Jacob. Schooling, Experience, and Earnings. New York: National Bureau of Economic Research, 1974.
Moghadam, Valentine M., ed. Patriarchy and Economic Development: Women's Positions at the End of the Twentieth Century. New York: Clarendon Press, 1996.
Northrop, Emily. "The Feminization of Poverty: The Demographic Factor and the Composition of Economic Growth." Journal of Economic Issues 24 (March 1990): 145-160.
OECD National Accounts: Main Aggregates. Paris: OECD, 1989.
Orshansky, Mollie. "Consumption, Work, and Poverty." In Poverty as a Public Issue, edited by Ben B. Seligman. New York: Free Press, 1965.
Orshansky, Mollie. "How Poverty Is Measured." Monthly Labor Review 92 (February 1969): 37-41.
Pearce, Diana M. "The Feminization of Poverty: Women, Work, and Welfare." Urban and Social Change Review 11(1978): 28-36.
_____. The Feminization of Poverty: A Second Look. Washington, DC: Institute for Women's Policy Research, 1989. Pearce, Diana M., and Hariette McAdoo. Women and Children: Alone in Poverty. Washington, DC: National Advisory Council on Economic Opportunity, 1981.
Pressman, Steven. "The Feminization of Poverty: Causes and Remedies." Challenge 31, no. 2 (March-April 1988): 57-61.
-----. "Keynes and Antipoverty Policy." Review of Social Economy 49, no. 3 (fall 1991): 365-82.
-----. "The Gender Poverty Gap in Developed Countries: Causes and Cures." Social Science Journal 35, no. 2 (1998): 275-286.
Rainwater, Lee. What Money Buys: Inequality and the Social Meaning of Income. New York: Basic Books, 1974.
Rodgers, Harrell R., Jr. American Poverty in a New Era of Reform. Armonk, N.Y.: M. E. Sharpe, 2000.
Ruggles, Patricia. Drawing the Line: Alternative Poverty Measures and Their Implications for Public Policy. Washington, DC: Urban Institute Press, 1990.
Sawhill, Isabel. "Discrimination and Poverty Among Women-Headed Families." In Women and the Workplace: The Implications of Occupational Segregation, edited by M. Blaxall and B. Reagan. Chicago: University of Chicago Press, 1976.
Schultz, Theodore W. "Investment in Human Capital." America Economic Review 51, no. 1 (March 1961): 1-17.
Schwarz, John E., and Thomas J. Volgy. The Forgotten American. New York: Norton, 1992.
Smeeding, Timothy, Gunther Schmaus, and Serge Allegraza. "An Introduction to LIS." LIS Working Paper #1, 1985.
Smeeding, Timothy M., Lee Rainwater, and Gary Burtless. "United States Poverty in a Cross-National Context." LIS Working Paper #244, 2000.
Strober, Myra H., and Carolyn Arnold. "The Dynamics of Occupational Segregation among Bank Tellers." In Gender in the Workplace, edited by C. Brown and J. A. Pechman. Washington, DC: Brookings Institution, 1987.
Summers, Robert, and Alan Heston. "The Penn World Table (Mark 5): An Expanded Set of International Comparisons, 1950-1988." Quarterly Journal of Economics 106 (May 1991): 327-368.
Wright, Robert E. "Women and Poverty in Industrialized Countries." Journal of Income Distribution 5, no. 1 (1995): 3 1-46.
Zopf, Paul E. American Women in Poverty. New York: Greenwood Press, 1989.
| |||
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)
Copyright and Terms of
Use