Good news about redistribution?

Economists, investors and business leaders devote considerable effort to peeling back the layers of causality that impede economic growth. Among the causes are high taxes that divert money from productive investment. High taxes, in turn, are caused by large government redistribution, which is justified as a countermeasure for income inequality and poverty.

This paper summarizes new research that peels back one more layer of the causal chain and shows that in the United Sates, official portrayals of income inequality and poverty are overstated by factors of five or more. One cannot generalize these results directly to other advanced nations, but some of the underlying mechanisms that created the overstatements likely exist elsewhere. Until further research shows otherwise, official claims with respect to the extent of poverty and income inequality in advanced nations should be treated with some skepticism.

Counting all income and taxes

Calls for more redistribution cite differences in money income as measured by the Census Bureau Current Population Survey (CPS). Money income includes earning from work or business, Social Security, some retirement plans, cash public assistance, dividends, and interest. The first data column in Table 1 shows the distribution of household money income by income fifths or quintiles. See table 1

Table showing distribution of household income by source, United States, 2015
Click to enlarge.

Differences in the average money income are offered as proof of significant inequality. For example, the ratios section of the table shows that average money income for the top quintile is 16.2 times larger than for the bottom quintile. But these comparisons exclude $1 trillion in government transfer payments to lower-income households and do not account for taxes that reduce the spendable income for higher-income households by as much as 50 percent.

These missing elements are not mistakes; they are documented exclusions with known but largely ignored features. If the government were to raise taxes on the wealthy and transfer all the additional money to the lowest income group through more food stamps, the official metrics of inequality would not change because neither the new taxes taken from the top nor the additional money given to the bottom would be used in the calculations.

The “More complete estimates” columns in Table 1 are the result of research that enhances the Census data to fill many, but not all, of the gaps in the money income measure.

The improved estimates begin with the distribution of market income across its quintiles, followed by the distribution of Social Security, other transfers, and taxes across those same households and then the resulting distribution of final net income. For example, the lowest income fifth makes only 2.2 percent of the total market income. Their income is augmented by a 34.3 percent share of Social Security and Medicare and 41.5 percent of other transfers.

These households contribute only 1.1 percent of all taxes. The result is that the lowest quintile receives 12.9 percent of spendable income available for consumption and savings.
The highest income quintile earns 57.7 percent of all market income, receives only 10.4 percent of Social Security and Medicare payments, pays 65.3 percent of all taxes, and is left with 39.3 percent of the final net income.

The ratios in the last three lines of Table 1 show the degree of inequality in the components. Average market earnings in the highest income group are 26.6 times those in the lowest. But the lowest income group gets 3.3 times more in Social Security and Medicare. It also receives 41.5 percent of other transfers while the highest group gets essentially none. The highest income group pays 61.2 times more taxes. The net result is that the highest group averages only 3.0 times more spendable income than the lowest. As in the next-to-last column shows, government redistribution has cut the ratio gap between the top and bottom quintiles by 88.6 percent. The last column shows that the gaps in more complete statistics are 81.3 percent smaller than in the Census money income. See figure 1

Average income, transfers, and taxes by income groups, 2013
Figure 1

Figure 1 illustrates the dynamics of redistribution. On average, households with $63,136 in earned market income get to keep it all. They pay taxes averaging approximately $17,000 per year but on average also get an equal amount of government transfers. Households with earned income less than $63,136 constituted 52.5 percent of all households and on average received net benefit payments from government. Some of the 47.5 percent of households above the break-even point also got transfer payments, but they paid more taxes than the transfers they received.

Net taxes from the top 47.5 percent of households raised average spendable incomes for the lower groups to near the median earned-income level and also to pay for all government services.

The Gini coefficient measurement of inequality

Advocates to reduce income inequality use a computation called the Gini coefficient to claim that income inequality in the United States is greater than in other developed democracies. The Gini coefficient has a theoretical minimum of 0.000 that represents no inequality, with everyone having exactly the same income. Its theoretical maximum is 1.000, representing total inequality with one person having all the income and everyone else having nothing.
Figure 2 displays Gini coefficients from the Organization for Economic Cooperation and Development (OECD) for seven large economies plus Denmark and Sweden, the two OECD nations with the smallest (least unequal) Gini coefficients. (For now, ignore the USA Complete bar.)

But the OECD explicitly excluded Medicaid transfers and state and local taxes from the United States calculation. These exclusions constitute a significant upward bias in the United States coefficient. The “USA Complete” Gini coefficient in Figure 2 corrects for these omissions, thereby reducing the coefficient to 0.23, lower than any of the comparison countries. See figure 2

Graph showing Gini coefficients of spendable income in advanced nations, including more complete USA coefficient
Click to enlarge.

The incidence of poverty has fallen significantly

Measuring the incidence of poverty starts with a poverty threshold, below which people are designated “poor.” The U.S. government stipulates poverty thresholds in terms of money income. They were first calculated for 1963 at three times the cost of an adequate economical diet. Since then, each threshold has been escalated by the rate of change in the Consumer Price Index for All Urban Consumers (CPI-U).

The Census Bureau uses the CPS family money incomes to identify families with incomes below their relevant thresholds. For 2015, it reported that 13.5 percent of the population lived below their poverty thresholds, the same as the average for the preceding 48 years. See figure 3

Graph showing percent of population below poverty threshold, 1947-2015

During the 15 years before President Lyndon B. Johnson called for the War on Poverty, the measured poverty rate had fallen from 34.8 percent to 19.0 percent. Two years after his speech and before most of the new programs were fully implemented, the rate had dropped to 14.7 percent, well within the range that would prevail for the next 48 years.

The apparent failure to reduce the incidence of poverty and the end to the systematic improvement trend of the mid-20th century is at least in part the result of flaws in the way poverty is measured.

In 1963, almost all welfare benefits were monetary payments which were counted in the CPS money income. But the War on Poverty and subsequent programs were excluded from the CPS measure of money income because they were considered “in-kind” aid. The $1 trillion-plus in annual transfer payments that are not counted in the official inequality measurements are also omitted from the income used to measure poverty. In effect, the new programs, by definition, could not improve the official measure of poverty, even as they were raising the spendable incomes of low-income people.

Escalating the poverty thresholds by the CPI-U overstates the money needed to maintain the standard of living established by the 1963 poverty benchmark. This excess is not the result of some mistake. It arises from well-known, but widely ignored, technical methods used in price-index construction. The Bureau of Labor Statistics publishes an alternative research estimate of consumer price change, the CPI-U-RS, that avoids some of these limitations. Bruce Meyer and James Sullivan have integrated the CPI-U-RS with other research to estimate poverty rates using a reduced-bias calculation of price change.

The official poverty measure stopped declining in the 1960s. Reduced-bias measures continued to decline for another 30 years until the turn of the millennium. The bias-adjusted poverty measure stabilized just below 5 percent, less than half the minimum 11.1 percent achieved by the official number in 1973.

Figure 4 shows the average annual income for the lowest 20 percentiles for 2015. The lower curve is the CPS money income used to calculate poverty. The official threshold line intersects the CPS money income curve at point A, the official poverty rate of 13.5 percent of the population.

The “CPS + Excluded Transfers” curve reflects the higher actual incomes resulting from a fuller accounting of transfer payments. It intersects the official poverty threshold just below 3 percent of the population, point B. Omitting transfer payments has caused poverty to be overstated more than four-fold. See figure 4

Graph showing incidence of poverty adjusted for excluded transfers and biased CPI

Point C identifies the 4.5 percent result from applying the reduced-bias poverty threshold to the official money income definitions. Finally, point D shows that using both the improved threshold and the more complete income estimates yields a 2 percent incidence rate, almost seven times smaller than the official measure.


More than 50 years after the United States declared the War on Poverty, poverty is almost entirely gone. Government spends more than $1 trillion annually in the name of reducing poverty. Yet the measurement system continues to report no reduction in poverty, and substantial taxpayer dollars go to people who are not poor.

Public understanding has also been misinformed by similar false signals about income inequality. The official estimates are deficient on several dimensions and exaggerate the degree of income inequality.

The misleading official measures continue be cited in efforts to increase transfer payments and raise taxes. A new, informed debate is needed on the size and structure of redistribution that comport with the facts. The good news is that the challenge is much smaller than we had been led to believe.