James P. Scanlan, Attorney at Law

FEMINIZATION OF POVERTY

(Feb. 3, 2013)

Prefatory note:  This page discusses issues addressed since 1987 concerning the failure to recognize that, for reasons inherent in the shapes of income distributions, reductions in poverty, including the poverty of disadvantaged groups, invariably cause such groups to comprise a larger proportion of the poor (and of the non-poor).  Especially pertinent articles include "'Feminization of Poverty' is Misunderstood" (Plain Dealer 1987), "The Perils of Provocative Statistics" (Public Interest 1991), and "Comment on McLanahan, Sorensen, and Watson's 'Sex Differences in Poverty, 1950‑1980'" (Signs 1991), and to a lesser extent “Divining Difference” (Chance 1994) and “Race and Mortality” (Society 2000).  This page supersedes the Feminization of Poverty S sub-page of the Scanlan’s Rule page, which quotes several of those works at length.  It may be compared to the Discipline Disparities and Lending Disparities pages, which also involve situations where failure to understand fundamental statistical patterns may be deemed remarkable [i]  

 I will eventually create a sub-page to this page (or a separate page) pertaining to the so-called “suburbanization of poverty,” a pattern, much discussed since 2010, whereby persons in suburbs have lately come to comprise an increasing proportion of the poor.  As with the feminization of poverty, the suburbanization of poverty concept is an unfortunate one because it mixes changes in the proportion of the poor comprised by a subpopulation (a) resulting from changes in the proportion of the subpopulation comprises of the total population with those (b) resulting changes in differences between the susceptibility to poverty of the subject subpopulation and the other subpopulations.  But there exists a complication in contrasting the two concepts arising from the fact that appraisals of differences in poverty of female-headed families and other units that are often discussed as an element of the feminization of poverty invariably relied on relative differences between poverty rate and/or and comparisons of proportionate changes in poverty rates, while almost all work on the suburbanization of poverty have relied on absolute differences between rates and/or comparisons of percentage point changes in poverty rates.  For the present, I merely note that what in discussion of the feminization of poverty was perceived to be a dramatic worsening of the comparative situation of female-headed families and other units would, according to the methodology of those analyzing the suburbanization of poverty, have been perceived as a substantial improvement in the comparative situation of female-headed families.  For an example using data from the 1987 Plain Dealer article, between 1959 and 1974, as the overall poverty rate declined from 18% to 9%, the poverty rate for persons in female-headed families dropped from 40% to 28%.  Thus, the poverty rate for female-headed families declined 24% compared with a 50% decline for the population at large, and the relative differences in poverty almost doubled from 122% to 211%.  But, as such data would be analyzed by those  now discussing the suburbanization of poverty, the poverty-rate of female headed families declined by 12 percentage points compared with 9 percentage points for the population at large, and the absolute differences between rates was reduced by a quarter (from 12 to 9 percentage points).[ii]  Problematic aspects of reliance on absolute differences between rates and percentage point changes are discussed, among other places, at pages 18-23 of the Harvard University Measurement Letter.

 ***

 One of the more remarkable examples of the failure to recognize the pattern inherent in the shapes of normal distributions whereby when two groups differ in their susceptibility to an outcome, the more the outcome is restricted toward either end of the overall distribution the greater tends to be the relative difference in experiencing it and the smaller tends to be the relative difference in avoiding it[iii] involves the research and commentary regarding the so-called “feminization of poverty.” Because that concept (at least when the terms was first used in 1978) referred to an increase in the proportion of the poor comprised by female-headed families or households rather than an increase in relative differences between in the poverty rates of those units and more advantaged units, the more pertinent pattern is a corollary to that described in the first sentence:  to wit, the more the outcome is restricted toward either end of the overall distribution, the greater will tend to be the proportions the group more susceptible to the outcome comprises of those experiencing the outcome and of those avoiding the outcome.[iv] To put the matter more concretely, the less poverty there is, the greater will tend to be relative differences in poverty and the smaller will tend to be relative differences in avoiding poverty; correspondingly, the less poverty there is, the greater will tend to be the proportion disadvantaged groups comprise of the poor and of the non-poor.

The main reason that treatments of the feminization of poverty represent such a remarkable example of this failure of understanding is that the pattern whereby reductions in poverty, including the poverty of female-headed families and households, will tend to increase the proportion such units make up of the total poor – hence, (a) that reductions in poverty will tend to feminize it and increases in poverty will tend to defeminize it and (b) that the well-being of female-headed units will tend to increase as the feminization of poverty increases and decrease as the feminization of poverty decreases – is evident in the very data researchers examine in studying the feminization of poverty.  That is, such data are broken down by numbers and proportions of various demographic units, including such units as the female-headed family, living below various percentages of the poverty line (e.g., 200%, 125%, 100%, 75%).  Thus, the data as maintained for the public should make it pretty obvious that reducing poverty will tend to increase the proportion female-headed families comprise of the poor and increasing poverty will tend to reduce the proportion female-headed families comprise of the poor. 

The pattern is illustrated in Table 1 below, which is based on the 2011 Current Population Survey.  It shows for 2010 (1) total population and (2) population in female-headed families, and for each of the above-referenced three ratios of the poverty line, (3) total population falling below the point, (4) population in female-headed families falling below the point, (5) proportion persons in female-headed families comprise of those falling below the point, (6) proportion persons in female-headed families comprise of those falling above the point. 

Table 1:  Proportions Female-Headed Families Comprise of the Population Falling Below and Above Three Ratios of the Poverty-Line [ref zz2920]

It is the penultimate column ((5) FHF%ofBelow) that should make it evident that reducing poverty will tend to increase the proportion female-headed families comprise of the poor and that increasing poverty will tend to decrease the proportion female-headed families comprise of the poor.  Some additional (though hardly complex) calculations are required to show the patterns of changes in the proportion female-headed families comprise of the non-poor. 

Table 1 and Figure 1 of “Can We Actually Measure Health Disparities” (Chance 2006) provide similar information by race for twelve different ratios of the poverty line.  Table 1 from that article (which also appears as Table 1 of the Income Illustrations sub-page of the Scanlan’s Rule page) shows consistent patterns whereby the lower the ratio of income to the poverty line, the larger are both the proportion blacks comprise of blacks and whites falling below it and the proportion blacks comprise of blacks and whites falling above it.[v]  The table also shows patterns of measures of differences between rates at which blacks and white fall above and below each point.

Another remarkable aspect of treatments of the feminization of poverty involves the failure to recognize that the feminization of poverty did not continue to increase after it was first discovered.  In fact, the proportions female-headed families comprised of white and black poor changed inconsistently and only in minor ways subsequent to the middle 1970s.  The proportions that female-headed families comprised of black and white poor, drawn from the information in Section A of the Feminization of Poverty-S sub-page of the Scanlan’s Rule page (slightly augmented) are set out in Table 2.  Possibly I will at some point provide a table or figure with more complete data.

Table 2.  Proportions Persons in Female-Headed Families Comprised of White and Black Poor at Various Points in Time from 1959 to 2009 [ref zz2919].

One of the rhetorical observations about the feminization of poverty when it was first discovered that would be most quoted in the ensuing decades was the following 1980 statement of the National Advisory Council on Economic Opportunity:  "All other things being equal, if the proportion of the poor who are in female-headed families were to increase at the same rate as it did from 1967 to 1977, they would comprise 100 percent of the poverty population by about the year 2000!"  That statement was sufficiently provocative that it would continue to be quoted long after it was evident that female-headed families would never comprise 100% of the poor and after it was evident to anyone who bothered to look that the proportion female-headed families comprised of the poor had changed little since the statement was made.  A quote of the statement from a 2001 book may be found here.  

Rarely, however, has there been recognition that poverty was no longer being feminized.  A May 19, 2012 search for “’increasing feminization of poverty” yields 21,800 results.  Many of these of course involve discussion of some gender equity issue that has nothing to do with the feminization of poverty.  Some may involve the developing world where substantial reductions in poverty will tend to increase the proportion disadvantaged groups comprise of the poor just as happened in the United States between 1959 and the middle 1970s.  But even those who on occasion have noted that the feminization poverty has declined would fail to see any connection between such decline and increases in poverty.   

The quoted statement of the National Advisory Council on Economic Opportunity also warrants note because it reflects a notion that a society is better off when some part of the poor is comprised of persons who are not in economically disadvantaged circumstances.  In the unlikely event that the poor should even be comprised entirely of economically disadvantaged units, something that would occur only if society verged on the complete elimination of poverty, the circumstances of society and the circumstances of disadvantaged groups would both likely be better than they had ever been before.

The full story of the feminization of poverty is a good deal more complicated than described above, since the proportion female-headed families comprise of the poor is also a function of the proportion female-headed families comprise of the total population.  Such issues are discussed in references listed in the prefatory note.  One aspect of the matter not directly related to data issues that ought also to be deemed remarkable is the way that observers discussing the feminization of poverty often give attention to issues related to women’s employment opportunity without the least consideration of the way that the increasing employment of married women contributes to the disadvantage of the female-headed family compared with the married-couple family.  See “Comment on ‘McLanahan, Sorensen, and Watson's 'Sex Differences in Poverty, 1950‑1980’" (Signs 1991), "The Curious Case of Affirmative Action for Women" (Society 1992).  See also "What Helps the Single Working Parent?" (National Law Journal 1988).