Poverty in the United States-1

As I attempt to resume posting on a regular basis, I naturally paid a visit to the Census Economic Briefing Room and skimmed down the list of indicators. Poverty rates for all families and individuals are up for the fifth consecutive year, with an additional 5.5 million Americans now in poverty.¹ This survey was released on Tuesday, but, curiously, does not include recent data on racial distribution of poverty, except for Hispanics.

Naturally, I collected the data I could on poverty and made a chart (below).

First, let's discuss the latest news: not only has poverty risen substantially (the thick red line) three full years into a recovery, median wages for male workers declined 2.3%, while median wages for women declined 1%. Women are still catching up to men in wages, but only because their wages are not declining as fast as those of men (CBPP). The low unemployment rate reflects a declining rate of workforce participation (BLS), with the private sector employment growing at 0.8%, and hours per worker still significantly lower than summer of '01 (all data SA, private sector).

Poverty typically is measured using the poverty rate (“headcount”). This type of measure is incomplete: it ignores the depth of poverty... A useful measure of the depth of poverty is the poverty gap, which can be calculated by subtracting the average income among the poor from the poverty line and then dividing this difference by the poverty line. The poverty measure we use, which we refer to as the “poverty level,” is calculated as the poverty rate multiplied by the poverty gap. If 15 percent of the population lives in households with incomes below the poverty line, the poverty rate is 15.0. If the average income among the poor is two-thirds of the poverty line, the poverty gap is .333. The poverty level is then 15.0 x .333 = 5.0.
[p.12] The study is preoccupied also with what I regard as the most meaningful measure of income for such a study, the pre-tax, pre-transfer income of households.

The IKS study includes regression of poverty levels on GSP (gross state product) per capita, employment, and unemployment (p.19, 20). The effect of GSP and employment are rather large and highly significant (employment more so than GSP per capita); unemployment, oddly, is less so—probably because official unemployment rates do not reflect "discouraged workers," who are more likely to belong to impoverished households. Table 3 (p.21) illustrates that employment (expressed as hours worked) and 10th-percentile wages were highly significant in predicting poverty, but less so in predicting changes in poverty levels. The impact of the minimum wage level is ambiguous, both tiny and statistically insignificant, but in states where the statutory minimum wage is the federal minimum, i.e., where it is lowest, and where a signifincant proportion of the population is working at the minimum wage, increases in the minimum wage have a highly significant impact in reducing poverty (p.16).

Higher employment rates and higher levels of per capita gross state product are strongly associated with higher levels of hours worked in bottomincome- quartile households and with higher tenth-percentile wage levels, which in turn are strongly associated with lower pretax-pretransfer absolute poverty. The states with the lowest levels of market absolute poverty as of 2000–2002 were New Hampshire, Minnesota, Maryland, Connecticut, Colorado, Iowa, Nevada, New Jersey, Virginia, and Wisconsin... All of these states had both high levels of hours worked among low-income households and high tenth-percentile wage levels, with the possible exception of Iowa on the latter. States that did well on one of these dimensions but not on the other tended to have somewhat higher levels of poverty. Massachusetts, for example, had the second highest level of tenth-percentile wages but ranked much lower on hours worked. Nebraska had the highest level of hours worked in low-income households but one of the lowest tenth-percentile wage levels. These two states were only slightly better than average in their levels of pretax-pretransfer absolute poverty.
[p.25]

(PART 2)

NOTE: 1 The definition of poverty is based on a formula assigning different nutritional requirements to households with members of the specified age and sex. The Orshansky Method in use begins with minimal nutritional requirements, calculates the cost (average for all urban areas), and multiplies that by three to establish the threshhold. The official poverty definition counts money income before taxes, including wages, salaries, interest, dividends, self-employment income, welfare payments (TANF), unemployment insurance, and social security payments.

There has been discussion (see linked essay) of adopting a "standard budget" approach; the advantage of a standard budget is that, rather than use simplifying assumptions about the share of food expenditures in the family budget, it involves a comprehensive basket of items that can be adjusted to reflect regional conditions. Discussions of poverty measures: Economic Policy Institute; different estimates of poverty based on different methodologies, US Census Bureau; poverty and income page, Center for Budget & Policy Priorities.

2 During the period 1984-1993, the Hispanic population of the USA grew at an average annual rate of 4.6%; that of the rest of the USA, at 0.7% (aggregate US growth for the period was 1.0%). This growth is especially surprising when one realizes it was mainly concentrated in a few states and followed hard on the heels of another major expansionary wave.

The 1984-1993 wave was stimulated by the Simpson-Mazzoli Immigration Act, which gave an amnesty for undocumented residents of the USA. In '93, so many Hispanics evidently took advantage of the amnesty that the community grew 13% in a single year. The prior wave was probably stimulated by falling oil prices.

Posted by James R MacLean at September 2, 2005 09:34 AM