Back to Families and Children: The Common Wealth of KentuckyBy Miriam Fordham and Dan Jacovitch
Legislative Research Commission
From Exploring the Frontier of the Future: How Kentucky Will Live,
Learn and Work
pp. 31-42, published 1996
According to the most recent figures from the Bureau of the Census, 14.7 percent of Kentuckians were living in poverty in 1995. The key question addressed in this chapter is whether there have been significant changes in the level and distribution of poverty in Kentucky. The chapter starts with a discussion of the measures of poverty which are judged most reliable in identifying changing trends. Next is an overview of changes in poverty in Kentucky compared to the nation as a whole. Then a comparison between groups of Kentucky counties classified by population growth or decline highlights any differences over time. Finally, the chapter presents information regarding changes in the poverty status of specific segments of Kentucky’s population.
A necessary first step in evaluating trends in poverty in the state is to determine which measures of poverty present a reliable measure of change. As noted above, the Bureau of the Census estimated that Kentucky’s poverty rate in 1995 was 14.7 percent. Fifteen other states had poverty rate estimates greater than Kentucky’s. However, one can only say with confidence that Kentucky had a lower poverty rate than five states; South Carolina, Louisiana, Alabama, Mississippi, and New Mexico had rates ranging from 19.9 percent to 25.3 percent. Although the other 10 states had higher point estimates on their poverty rates than Kentucky in the March 1996 Current Population Survey (CPS), the relatively small sample (nearly 700 households in Kentucky) used to estimate annual poverty rates between census years means that state-level annual estimates carry margins of error in the range of 1 to 2 percentage points above and below the point estimate.
The same limitation of the survey sample holds true for measuring change over time. To alleviate some of the problems related to small sample size, the Bureau of the Census combines annual data into multiyear averages. In their latest release of state-level annual poverty rates, the Bureau provided the average for 1994-1995 and the average for 1993-1994. Kentucky registered a 2.9 percentage point decline between the two-year averages. The two-year moving average increased in 19 states, showed no change in one state, and decreased in 30 states. However, even with multiyear averages, when the margin of error is recognized, only 10 states had statistically significant changes (nine decreased and one increased) in the two-year moving average poverty rates. Kentucky’s 2.9 percentage point decline was statistically significant; only two states registered greater declines—Missouri and Louisiana.
The statistically significant decline of Kentucky’s poverty rate is welcome news; however, not much more can be said—the sample size does not allow one to accurately measure changes in the substate geographic distribution of poverty or detailed characteristics of those in poverty. Also, statistically significant changes in CPS measurement of state-level poverty rates are the exception rather than the rule. In the prior year’s CPS (March 1995), only three states had statistically significant changes between the two-year averages—Kentucky was not one of them. Kentucky’s standard error for annual poverty rates, as measured by the CPS, consistently runs around 2 percentage points each year. This means that if a survey measures a poverty rate of 18 percent, there is a 90 percent probability that the rate is somewhere between 16 percent and 20 percent. The problem in measuring poverty in nondecennial census years is clear: a real change in the poverty rate of 1 percentage point is a significant occurrence and yet it is well below the threshold of current sampling accuracy.
Given the limitations of state-level measures of poverty in noncensus years, clear understanding regarding changes in Kentucky’s poverty situation are only achievable through new uses of older decennial census information. This paper makes significant use of two pieces of information from the 1980 census and the 1990 census–county-level poverty data and Public Use Microdata Samples (PUMS), which will be described later.
Table 1 provides the 1990 census poverty rates and numbers (based on 1989 income) for children (related children under 18), elderly (persons of age 65 or older), and all others, as well as for female-headed families with no spouse present with and without children, and all other families. In addition to the 1989 poverty rates, this table provides the change from the 1980 census (based on 1979 income) in both quantity and percent, as well as comparable figures for the United States.
Table 1: Poverty in Kentucky and the United States, 1979-1989
The figures for children and elderly are interesting to compare. Note that between 1979 and 1989 the poverty rate for children in Kentucky increased 2.8 percent, to 24.5 percent, while for the elderly it declined 2.7 percent, to 20.6 percent. However, this does not mean that there were more children and fewer elderly in poverty. Quite the opposite was true: 292 fewer children and 358 more elderly lived in poverty (although for practical purposes, one could say these numbers were "flat" representing a change of less than 0.5 percent). The real change in both age groups occurred in those above poverty. There was a 14.9 percent decrease in the number of children above poverty and a 17.6 percent increase in the number of elderly above poverty. The same relative differences between those in poverty and those above poverty apply at the national level although the national figures reflect an overall growth in the population whereas Kentucky’s population experienced no appreciable growth between 1980 and 1990.
In particular, among all states, Kentucky had the third largest percent decline (11.9 percent) between 1980 and 1990 in the total number of children under 18. Only Iowa and West Virginia had greater declines, 13.1 percent and 20.7 percent respectively. The nation overall had a 0.3 percent decline in the number of children over the same period. The observed loss of children at a much greater rate for those above poverty could be due to any one, or a mix, of several factors—greater out-migration of those above poverty, lower birth rates for those above poverty paired with natural loss due to aging, or movement to below poverty status for those who were previously above poverty. Without data that allows the tracking of individuals over time, definitive statements are not possible.
The significantly higher growth rate of elderly above poverty is clearly due to an aging population paired with government financial aid for the elderly. The numbers may also reflect greater longevity for the elderly not living in poverty, as opposed to those who do live in poverty.
In Table 1, the "other persons" category, mostly working-age adults, makes up the majority of those in poverty. Although children have a higher poverty rate and experienced a greater percentage point increase in their poverty rate, working-age adults experienced an 18.2 percent increase in the number below poverty, while the number of poor children and elderly was basically unchanged between the two years.
The remaining information in Table 1 provides changes in poverty status for families with related children under 18. The two main categories presented here are female-headed families with no husband present and other families (married couple and male-headed with no wife present).(1) As one would expect, female-headed families with children but with no husband present had the highest poverty rate (51.8 percent); this type of family also had the highest percentage point increase (5.9 points). The percentage growth in the number below poverty was also the greatest (38.4 percent) for this type of family.
Again note that the number of families with children below poverty increased by 14,300 for female-headed families and 1,263 for other families, while the number of children in poverty dropped by 292. This indicates a significant change in the composition of poor families with children in Kentucky between 1979 and 1989. They must, on average, have fewer children, and a greater percentage are female-headed with no husband present. Relative to the United States however, Kentucky families in poverty have fewer female householders with no husband present. So, between 1979 and 1989, Kentucky experienced similarly high increases in numbers of female-headed families below poverty with no husband present (38.4 percent) as the United States (29.0 percent), but because Kentucky is generally a poorer state relative to the national average, other families compose a greater share of families in poverty both in real terms and relative to the nation.
The substantial decline (-8.9 percent) in the number of other families with children who had incomes above poverty is noteworthy. An offsetting increase of other families without children (not shown here) also occurred. These changes are likely due to a mix of factors, including less family creation, dissolution of families, postponement of childbearing, and, possibly, some movement from above poverty to below poverty classification due to changing economic circumstances. Again, without data which tracks individuals over time, it is not possible to determine which factors best account for these developments.
Table 2 is based on county-level poverty information for 1979 and 1989, extracted from the USA Counties 1994 CD-ROM. Table 2 presents similar information to Table 1, but rather than comparing Kentucky to the United States, the comparison is between totals for the 65 Kentucky counties that experienced a decline in population between the two decennial census years and the 55 counties that had population growth. The CD-ROM did not provide county-level information about families with children; therefore, that information is not provided in Table 2.
Table 2: Poverty in Kentucky Counties with Declining and Growing Populations, 1979-1989
The 55 growing counties as a group experienced lower poverty rates and substantially lower increases in those rates between 1979 and 1989 for the nonelderly (children and other persons). The 65 shrinking counties as a group experienced substantially greater percent declines in the two nonelderly groups above poverty compared to those below poverty—a 22.2 percent decline in the number of children above poverty versus a 2.8 percent decline in the number below poverty. The significant growth rate differentials between those above and those below poverty led to the greater increases in poverty rates. So, for the shrinking counties, a portion of the nonelderly above poverty are either moving away or falling into poverty.
For the elderly, the level of poverty rates or the change in poverty rates did not differ as much between the two groups of counties. Additionally, for those counties experiencing population loss, the elderly are the only component of the population with a net increase in numbers, with all of the increase coming in the above poverty category. These two factors indicate reduced mobility for the elderly while the nonelderly are more likely to move.
For those counties experiencing population growth, a drop of 7.4 percent in the number of children above poverty and a 3.9 percent increase in the number below poverty still occurred, with a net decline of about 27,000 children. This is remarkable in that the number of adults (elderly and nonelderly combined) increased significantly (141,000.) Together, these figures strongly indicate lower birth rates.
Similar analysis differentiating between high- and low-population density counties (not shown) yielded slightly different results. The counties with the higher population density in 1989 had lower poverty rates than, but about the same changes in those rates as, those with lower population density. The most noteworthy difference in the population density breakout was the differentiation for the elderly. Grouped by population growth, the poverty rate for the elderly differed slightly; however, when grouped by population density, the poverty rate for the elderly was much higher in the low-density areas. This indicates that, although the elderly may not be well represented among those who move into and out of areas of growth or decline, their well-being is somewhat related to the area in which they live, as measured by population density. Those in less dense or less developed areas are more likely to be poor, just like those in other demographic groups.
All of the previously presented information relied on standardized summary tabulations as released by the Bureau of the Census. Certain detailed cross-tabulations are not released because of the Bureau’s limited resources. To allow researchers to circumvent this limitation, the Bureau releases the actual census questionnaire responses of all persons living in a 5 percent random sample of households (addresses and names are withheld to protect confidentiality). This Public Use Microdata Sample (PUMS) allows researchers to formulate their own tabulations in order to answer questions which cannot be addressed with the standard set of summary tabulations. Table 3 provides tabulations from the 1990 PUMS along with changes from the 1980 PUMS. Note that because PUMS is a sample, all information in Table 3 is subject to some error. However, the large size of the sample (185,000 individuals) allows for analysis of fairly detailed segments of the population. The table is composed of three sections running horizontally across the page and representing, from left to right, three major groups of Kentuckians: adults in households with children, children alone, and the elderly.(2)
Table 3: Poverty Rates and Distribution by Detailed Population Characteristics, Kentucky, 1979-1989
When examining the poverty status of detailed segments of a group, it is important to consider both the frequency and magnitude of poverty for each segment relative to others. This can best be accomplished through simultaneous examination of two measures—poverty rate and share of poverty. The poverty rate represents the frequency of the occurrence of poverty within the population segment; it is a measure of how prevalent poverty is among the people within the segment. The share of poverty indicates the magnitude of poverty for a given segment relative to other segments; it is the share of all poor that is composed of that particular segment of the population. Table 3 provides the poverty rate and share of poverty for 1989, along with the changes from 1979, for detailed segments within each of the three major population groups under review.
An example of the different kinds of information provided by analyzing both the rate and share of poverty for various population groups can be seen in examining the relationship between race and poverty in Kentucky. Note in the left columns that the 1989 poverty rate for black adults in households with children (29 percent) was nearly double the rate for the white adults in the same category (16 percent). However, black adults accounted for only about 11 percent of all poor adults in households with children because blacks comprise a relatively small portion (7 percent) of the Kentucky population. The fact that black adults account for a larger share of those in the subgroup in poverty than they do of the entire population shows that they are overrepresented among the poor.
Childhood poverty continues to be an area of focus for public policy proposals. The increasing percentage of children in poverty raises concerns about future economic opportunities for Kentucky and the future fiscal impact on the state. Keeping all other factors constant, an increase in poverty in the state tends to decrease the amount of revenues received by state and local governments while increasing their expenditures. However, children are not the only focus of policy concern. Children are not poor by their own design, but rather because the adults responsible for their care cannot, or do not, provide a sufficient income to raise themselves and their children out of poverty. Therefore, in order to come to a fuller understanding of the causes of poverty among children, it is necessary to examine the characteristics of the adults responsible for their care.
One change among adults with children in poverty was in the distribution of poverty by age segment. Between 1979 and 1989, the two segments of older working-age adults with children (36 to 65) experienced little change in poverty rate while the two segments of younger working-age adults with children (19 to 35) experienced significant increases. The share of poor adults with children comprised of those in the 26 to 45 age segment increased while the shares comprised of those both older (46+) and younger (19-25) decreased. The aging of the Baby Boom generation likely accounts for this increased share, as it accounts for such a large share of the population in general.
Tabulations regarding household size, number of persons in the household, and number of associated children indicate a significant decline in the share of poverty represented by large families with many children. These larger families have the highest poverty rates (which is not surprising given that poverty status is determined partly by household size) as well as the greatest increases in those rates, but their share of poverty is declining. This drop in share is likely due to a general reduction in family size and an increased incidence of poor, single-parent families. These are likely to be smaller because of the absence of additional adults and because the household heads are likely to be younger and, therefore, have fewer children.
Families with two parents present may be better able to survive economically during a downturn in the economy, or anytime for that matter, because they have the potential for two wage earners in the family. Also, average earnings of female-headed families with one wage earner is only 64 percent of those of married couple families with one wage earner and 31 percent of those of married couple families with two wage earners (and 72 percent of those of male-headed families with one wage earner, although very few of these exist).(3) Not surprisingly then, adults in female-headed households with children had the highest poverty rate (42.2 percent) among all family types in 1989. This was up 3 percentage points from 1979. Adults with children in married couple families still accounted for the largest share (63.1 percent) of all poor families, but that share was nearly 7 percentage points lower than in 1979, while the poverty rate for the group was essentially unchanged. This means that married couple families made up a smaller share of all families. Conversely, female-headed households with children increased as a share of all families, but particularly, as a share of poor families.
The tabulation of marital status further outlines the changing characteristics of households in poverty. Those adults in households who have children and who are divorced, separated, or never married had the highest poverty rates, experienced the greatest increases in those rates and showed the greatest increase in the share of those below poverty. These observations highlight, once again, the trend toward an increasing number of single-headed households (mostly female-headed) and point to their heightened vulnerability.
As the tabulation for educational attainment indicates, the two population segments with a high school education or less continue to make up a preponderance of those with incomes below the poverty level (86.6 percent). Poverty rates are much lower for those who have at least some college attendance. Adults with the least education (eight years or less) had by far the highest poverty rate (42.1 percent) and the greatest increase in that rate (10 percentage points). However, the share of the poor who had less than an eighth grade education declined significantly. As was the case with the population in general, those below poverty were increasingly more educated in 1989 than in 1979. Note the 9.8 and 4.8 percentage point increases in the share of poverty for groups with 9 to 12 years of school and one to four years of college, respectively. However, this increased level of education is not just a reflection of changes in the general population, for those with 9 to 12 years of school experienced a significant increase in poverty rate (5.9 percentage points), while those with college experienced little change in poverty rate. Today’s labor market demands highly skilled workers, and those workers armed with only a high school education have fewer prospects. For workers with less than a high school education, the outlook is even more bleak.
Measures of the degree to which a person is a part of the workforce are provided in the tabulations of "Number of Weeks Worked During the Year" and "Usual Hours Worked per Week." As can be seen, of the adults in households with poor children, more than 50 percent did not work at all in the preceding year, and 73 percent worked 26 weeks or less. The poverty rate for the group who did not work at all during 1989 was 39.5 percent, 10 percentage points higher than in 1979. There was very little change in the distribution of number of weeks worked among poor adults with children. So, for this group the extent of work through the year did not change much between the decennial census years, but those working the fewest number of weeks had the highest poverty rates and the greatest increase in those rates.
As stated earlier, the defining characteristic of children in poverty is that they live with adults who cannot provide a sufficient income to raise them out of poverty.(4) For the most part, the changes in rates and distribution of poverty noted above for adults apply similarly to children, where applicable. However, the poverty rates for children are higher than the corresponding rates for adults across the board.
Just as the youngest adults had higher poverty rates, so do the youngest children (these are likely the children of those youngest adults). Poor children aged five or under represented more than one third of all children in poverty, and this youngest age group experienced the greatest percentage point increases in poverty rate and share of poverty. Children in female-headed households comprised 40.1 percent of those below poverty. These children had a poverty rate of 53.8 percent, much higher than the 16.2 percent for children living in married couple households.
Additional insight can be gained from examining the distribution of those below poverty in terms of their depth of poverty. After all, a family with income at 50 percent of the federally defined poverty level has only about half as much income as a family that is slightly below the poverty level. For children, the distribution of those below poverty is disturbing, in that over half were in households with incomes at or below 50 percent of the poverty level. There was an increase of 4.5 percentage points in the share these poorest of children comprised of all poor children. Also troubling is that the segment of poor children in households with 76 to 100 percent of poverty level was the segment which experienced the corresponding decline in share of children below poverty. Thus, although there was a slight decline in the total number of poor children (as shown in Table 1), these slightly fewer children are deeper in poverty.
Between 1979 and 1989, the elderly in Kentucky did not experience the same changes in poverty as the other groups. The poverty rates of all but the oldest age segment declined over the period. The rate for the oldest segment was essentially unchanged. Even with these observed declines, the elderly still have higher poverty rates than working-age adults. Also note that, for the elderly, the oldest have the highest poverty rates but represent the smallest share of the elderly poor. By way of comparison, among children the youngest have the highest poverty rates and represent the greatest share of poor children. The former may often be a case of persons exhausting financial resources and, thereby, falling below the poverty level, while the latter may be a case of children starting out in families with insufficient financial resources.
In 1989 more than 70 percent of the elderly poor were female, a share which increased by 5.5 percentage points over the decade. The increase is likely attributable to the greater life expectancy and somewhat limited attachment to historical income (such as through pensions or social security payments based on lifetime earnings) for elderly women. The poverty rate for both elderly females and younger adult females was greater than that for the males in each of the respective age groups. However, the poverty rate for elderly females (27.2 percent) was significantly higher than the poverty rate for younger adult females (19.4 percent).
Another major change for the elderly poor was the substantial decline in the poverty rate for the elderly in nonfamily households (down 44.1 percent) even as this living arrangement accounted for a growing share of both poor and nonpoor elderly households. It would appear that one way in which the elderly have attempted to improve their economic security is by joining with nonfamily members to share scarce resources.
The final tabulation provides the distribution of elderly below poverty in terms of their household income as a percent of the federal poverty level. The figures provide a further contrast between the nature of poverty among the elderly compared with poverty among children. Note that the distribution for the elderly poor is inverted compared to that for poor children. Over 80 percent of the elderly poor were in households with incomes greater than one half of the poverty level, mostly unchanged from 1979. The comparable figure for the share of poor children was less than 50 percent with a decline of over 4 percentage points from 1979. Thus, the elderly poor do not experience the concentration of deep poverty which afflicts Kentucky’s poor children.
Definitive state-level poverty measurement is only available in snapshots taken once every 10 years with the decennial census. While care must be taken to not overrepresent changes between two points in time as indicative of trends, insights can be gained from observing the changed poverty situation of Kentucky vs. the United States, of Kentucky’s growing vs. declining areas, of Kentucky’s elderly compared to young adults with children, and of the changing composition of poor families in Kentucky.
Throughout the analysis, several themes recur. Clearly, the young in Kentucky were worse off in 1989 than they were in 1979, and the elderly were better off, yet still experienced high rates of poverty. The general increase in poverty rates in Kentucky between 1979 and 1989 appears to be driven both by higher poverty rates among younger working-age adults with children and by greater numbers of female-headed households with children (not mutually exclusive groups). These groups are comprised of individuals most likely to have small children, which explains why over one fourth of the youngest children were below poverty and why their poverty rates increased so much.
The increases in poverty rates among the younger working-age population between 1979 and 1989 was indicative of an economy that did not provide economic opportunities matched to the skills and location of its younger citizens. Persons who are above poverty are more likely to be able to avail themselves of education and leave impoverished areas, hence the significant decline in numbers of persons above poverty, and the widely observed out-migration from counties with the highest poverty rates. However, even counties that had the higher growths in population experienced a loss of children above poverty. This could be due to a lifestyle decision observed among those with higher incomes to postpone having children until later in life, which may result in a disproportionate lowering of the birthrate among those above poverty.
The elderly on the other hand, while having higher poverty rates than the overall population, experienced significant declines in those rates between 1979 and 1989 across all breakdowns for analysis—indicative of the benefits afforded by federal support programs. Elderly women had a significantly higher poverty rate than elderly men and comprised about 70 percent of the elderly poor. This is due both to less attachment to historical income for elderly women and to their longer life expectancies. The experience of the oldest segments of the elderly (higher poverty rates and an increasing share of the elderly in poverty) is likely due to an exhaustion of fixed or limited incomes, particularly relating to medical expenses, for an ever increasing number of people living past the life expectancies that existed when they or their spouses entered retirement.
The driving forces behind Kentucky’s poverty might be broken into several categories, each with its own possibilities for policy intervention. Poverty associated with the life cycle category might be defined as that related to older people exhausting their assets and younger people trying to get started in life. Poverty associated with societal composition factors might be defined as that related to changes in family size, structure, and demographic characteristics. Examples are the increase in the number of female-headed families, possible differences in fertility rates between poor and nonpoor families, and an increase in the general educational level which means that those with some high school or a high school diploma account for an increasing share of the poor. Finally, poverty associated with regional economic factors is related to a general lack of economic opportunities in the surrounding area and the propensity of those with the ability to seek opportunities elsewhere to leave.
While the poor of Kentucky fall into all of these categories, the last category appears to play a prominent role in distinguishing Kentucky’s experience over the decade from that of the nation as a whole. For the future, it is safe to say that areas which remain economically depressed are likely to continue to experience out-migration that will be comprised mostly of those able to find work elsewhere. Those unable to move (the elderly), and those unable to find jobs with sufficient incomes (the unskilled, female heads of households) and their children, are likely to find themselves living in areas where poverty is increasingly common. Given that Kentucky has a declining number of children, who are increasingly living in the deepest poverty, and given the well documented research which indicates that poor children are less likely to grow into economically productive adults (particularly in an economy which demands increasingly complex skills), the greatest change indicated by the analysis is not between the nature of Kentucky’s past and its present, but rather that its future seems so much more at risk.
To view a list of all chapters in this book, click here. To read the chapters in sequential order, please follow the arrows below.
Back to Families and Children: The Common Wealth of Kentucky
Ahead to Assessing the Future of Housing in Kentucky
For
the purposes of this section, "children" are defined as those under age 18 who are related to the householder. Children related to the householder can include any child related by blood or marriage to the householder. Return to text.Poverty
status as calculated here is identical for all individuals in the same household and is based on household income. This differs from the traditional census measure reported thus far where the calculation of an individual’s poverty rate depends on whether the person is related to the head of the household (householder). Individuals related to the householder who are under 18 or are the spouse of the householder all share the same poverty rate based on the family’s income; people 15 years or older who are not related to the householder, or are the householder of a nonfamily household have individual poverty rates based on individual income; finally, poverty status is not measured for children under 15 who are not related to the householder. In this method, the census is assuming that children under 15 not related to the householder may not have access to the householder’s income so their poverty status should not be measured. For purposes here, the desire was to exclude the least number of children from poverty status measurement. Return to text.Calculations
from data contained in Table No. 678: Families With Earners—Number and Earnings: 1980 to 1993. (1996). Statistical abstract of the United States, 1995. Washington, DC: U.S. Government Printing Office, 434. Return to text.For
purposes of this analysis, children are defined as persons 18 years or younger and are not necessarily related to the head of the household, as they were in the Bureau of the Census category definition used in the previous sections. Return to text.