Over the past three decades, the American economy has grown considerably. Accounting for population growth and price changes, GDP has increased by more than 60 percent. There is a prevailing sentiment, however, that the middle class and the poor have not enjoyed any of the benefits of this sustained growth. A 2007 CBS News poll found that 60 percent of Americans believed that things had gotten worse for the middle class during the past decade. With the financial collapse of 2008 and the recession that followed it, that percentage is surely much higher today. Conventional wisdom holds that things have also gotten worse for those at the bottom, despite efforts to alleviate the living conditions of the poor. As Robert Siegel, of National Public Radio, recently stated, “It is commonplace to say that we have lost the war on poverty.”

Much of this sentiment stems from official measures that paint a bleak picture of the middle class and the poor. Official median household income fell between 1999 and 2004, using the conventional adjustment for inflation. Since then, median income has risen but remains below its 1999 level. Official statistics are even gloomier for the poor: The official poverty rate in 2009 was higher than in 1980.

But this grim portrait of an America in which life has improved for the wealthy and no one else is inaccurate. Over the past three decades, growth in the U.S. economy has produced considerable, and demonstrable, improvement in material well-being for both the middle class and the poor. They, too, have come along for the ride. One just needs to know where—and how—to look.

With analytical measures that are great advancements on the official methods of the past, we have found in our research significant evidence of such improvement among the middle class and poor. Between 1980 and 2009, income and consumption rose by more than 50 percent in real terms for both groups. Living units became markedly larger and much more likely to feature air-conditioning and other amenities. The quality of the cars these families own also improved considerably.

First, we’ve found strong evidence of improvement in the material well-being of poor families. For Americans in the 10th percentile of wealth, income rose by 44 percent between 1980 and 2009. Even this, however, understates improvements at the bottom. Consumption among the poorest (what goods they actually obtained or used) grew even more during this same period. For those in the bottom 20 percent, or the lowest quintile, the size of living units increased by 200 to 250 square feet, and the fraction of these low-income people who enjoy either central or window air-conditioning doubled. The share with other amenities also rose noticeably.

There were similar improvements for the middle class. Their houses have become bigger, rising by more than one-quarter of a room without adjusting for household size, and by seven-tenths of a room when accounting for family size. Since 1989, square footage has risen between 300 and 350 square feet. About half of this increase has taken place since 1999, a period during which official median household income fell.

The share of middle-quintile individuals with air-conditioning rose from 58 percent to 88 percent between 1981 and 2009. Central air-conditioning rose from 27 percent to 67 percent. Since 1989, the share of people in the middle quintile with a dishwasher has risen from 53 percent to 70 percent; and with a clothes dryer from 79 percent to 88 percent. Again, a large share of this increase has occurred since 1999. Since 1989, the incidence of leaks from plumbing and roofs has fallen sharply, as has the likelihood of living amid peeling paint or with a broken toilet.

What accounts for the striking differences between official analyses and our research? First, most analyses of economic well-being rely on narrow income measures that do not reflect the full range of resources, monetary and otherwise, available to the households for consumption of goods and services.

They ignore food stamps and do not take into account resources from important antipoverty programs such as the Earned Income Tax Credit (EITC) and housing or school-lunch subsidies. These excluded benefits are generous and have increased significantly over the past few decades. Official analyses also often rely on underreported measures of income. For these reasons, our analyses include comparisons between official pretax income numbers and figures that incorporate taxes and benefits.

Second, official statistics exaggerate inflation. Obviously, the higher prices seem, the lower the perceived purchasing power of the poor. But in determining the poverty threshold—the statistical point that separates the poor from the rest of the population—the official statistics rely on an index called the Consumer Price Index for All Urban Consumers, also known as CPI-U. There is broad consensus that the CPI-U overstates inflation and therefore significantly understates improvements in economic well-being.¹ When median incomes are calculated, a similarly biased index is used: the Consumer Price Index Research Series, or CPI-U-RS. Over time, bias in both these indices has significantly muted the rise in household income for both the middle class and the poor.

Hypothetically, an annual misstating of inflation by just 1 percent would lead to a 33 percent difference in determining median incomes between 1980 and 2009. Our best evidence indicates that the annual bias over much of this period has been even greater. In 1996, the Boskin Commission, a group of eminent economists appointed by the Senate Finance Committee, concluded that the annual bias in the CPI-U was 1.1 percentage points that year, but 1.3 percentage points in preceding years. In our results, we make changes to correct for such bias by using an adjustment that subtracts 0.8 percentage points from the growth as figured in the CPI-U-RS index each year.

Finally, we believe that consumption provides a more appropriate measure of well-being than does income. It reflects resources that may have been accumulated over time, whereas income may vary in the short term. Consider, for example, a retired couple who own their home outright and who live off of savings. Clearly, their income will not reflect the totality of their actual material well-being.

Income measures also fail to capture disparities in consumption that result from differences in the amount of credit available and credit used. Even if income remains the same, government safety-net programs affect the ability of households to consume because they diminish the need for households to save for a rainy day. (If you know you’re going to receive free health care from the time you’re 65, you do not have to restrict your spending in the prior two decades to save for health-care expenses, for example.) Consumption is more likely to capture income from self-employment and to reflect private and government transfers. Much of the effort to improve income measures involves making them more like consumption measures, but it is much easier to begin with consumption data. That’s what we did.

Our preferred measures of income and consumption as well as data on housing and cars all tell the same story:²The material circumstances of the middle class have improved dramatically since 1980. It is crucial to note the pattern of differences between the results from official reports and those from our improved measures. We’ve therefore provided three different assessments for comparison, which you can see in Figure 1.

First, official pretax median household income grew by 20 percent in real terms between 1980 and 2000, but then fell by 5 percent during the 2000s. The second measure of pretax money income differs from the first in several ways: Resources are measured at the family, rather than household, level and adjusted for differences in family size and composition following National Academy of Sciences recommendations. The median for this second measure follows a pattern very similar to that of the official measure, although the level of the median is about 15 percent higher.

The third measure, unlike the first two, accounts for inflation using our adjusted inflation index rather than the biased CPI-U-RS. When the adjusted index is used, median income is seen to have risen by 46 percent between 1980 and 2009, compared with 17 percent when the biased measure is used.

It’s not the rich 1 percent who have been working to stall the well-being of the middle class; it’s bad statistics.

In terms of car ownership—the largest and most expensive consumer product—those in the middle-income quintile, or the middle 20 percent, have fared well. Car ownership rates for this quintile are high—close to 95 percent—and have not changed noticeably during the past three decades. But the fraction of families with more than one car rose by more than 4.4 percentage points between 1999 and 2009. Furthermore, these cars were much more likely to have features such as power brakes, power steering, and a sunroof. The number of cars with air-conditioning rose from 47 percent to more than 83 percent between 1981 and 2004. These changes in vehicle characteristics understate the full pattern of quality improvement. There have been widespread improvements in braking and acceleration, as well as in the adoption of air bags, antilock brakes, sophisticated sound systems, and other features. Any discussion of material well-being must take into account these advances in amenities. If the material comforts of high earners can be cited as the excesses of a carefree elite, then honesty demands consideration of the dramatically improved products that are increasingly available to Americans across the economic spectrum.

All our results indicate a notable rise in the material well-being of the poor during the past three decades. Using our preferred measures of income and consumption (and addressing bias in previous methods of calculating inflation), we’ve shown a sharp decline in poverty. We’ve also shown that there have been noticeable improvements in the quality of living units and cars for poor families.

The official measure of pretax money income for America’s poorest 10 percent indicates only modest gains over the past three decades. We see virtually no change between 1980 and 1993. The 10th percentile for the official measure then rose by 19 percent between 1993 and 1999, but it fell in real terms in the most recent decade. But, after making the same adjustments made to the middle-class results, we can see that the income of the 10th percentile has actually risen by 40 percent.

The consumption numbers indicate significantly improved well-being for those near the bottom over most of our time period, as we can see in Figure 2. Between 1980 and 2009, consumption among the 10th percentile rose by 54 percent in real terms, while after-tax income plus noncash benefits grew by 44 percent. The changes in consumption have often differed from the changes in income, particularly in recent years. During the 2000s, consumption at the 10th percentile grew by 18 percent, while after-tax income plus noncash benefits grew by less than 4 percent.

Taxes and transfer payments (such as the Earned Income Tax Credit) have had an important impact on the resources of those near the bottom. The 10th percentile of after-tax income plus noncash benefits rose noticeably more than did the 10th percentile of pretax income, and the results showing after-tax income and noncash benefits show a consistently higher level of material well-being for those at the bottom. Elsewhere, we show that this difference is accounted for by taxes rather than noncash benefits (for more details, visit nd.edu/~jsulliv4/FiveDecades4.01.pdf).

When we looked at poverty rates, we found striking results. According to the official measures, the poverty rate in the United States between 1980 and 2009 increased by more than 1 percentage point. By our calculations using our adjusted inflation measure, poverty levels fell by more than 3 percentage points. And, even more striking, over the past three decades, a measure of what we might call “consumption poverty”—or the fraction of individuals whose consumption falls below an inflation-adjusted poverty threshold—fell by nearly 10 percentage points, a decline of 76 percent.

Like their middle-class counterparts, those in the bottom income quintile are living—and driving—much better than they were 30 years ago. Their houses have become bigger, rising by 0.16 rooms on average without adjusting for household size, and by half a room when accounting for family size. Housing-unit size rose after 1999, but by a smaller amount than over the previous decade.

The share of bottom-quintile individuals with any air-conditioning rose from 41 percent to 83 percent between 1981 and 2009. Central air-conditioning rose by 39 percentage points from 15 percent to 54 percent. Since 1989, the share of people in the bottom quintile with a dishwasher rose from 22 percent to 42 percent, and the prevalence of clothes dryers for the bottom quintile rose from 48 percent to 68 percent. Most of this increase occurred after 1999, a period during which official poverty rose. Over the period since 1989, the incidence of plumbing and roofing leaks fell sharply, as did the likelihood of living amid peeling paint or with a broken-down toilet.

Over the past three decades, car-ownership rates grew more noticeably among the poor than among the middle class. In 1981, 69 percent of all households in the bottom quintile owned at least one car. By 2009, 76 percent did. There is also evidence that the quality of cars owned has improved. Among the poorest households, the fraction of cars with power brakes, power steering, or other features rose sharply between 1981 and 2004. The fraction of cars with air-conditioning rose from 47 percent to 77 percent between 1980 and 2004. The share of cars with power brakes, power steering, and air-conditioning was about the same for the bottom and middle quintiles by 2004.

These patterns stand in sharp contrast to the noticeable decline in living standards over the past few years, which have hurt both the poor and middle class. Since 2007, median consumption has fallen by 5 percent, and consumption poverty has increased by 21 percent. Many of the calls for radical policy reform have been based on these short-term patterns. But the national crisis as it is currently framed—as one of declining well-being for the middle class and the poor—is not supported by rigorous analysis.

If discussion of wealth inequality is to bear fruit and lead to effective policy, it must begin with this honest assessment and not the politicized impressions that have made genuine debate impossible. The recent declines in the material circumstances of the middle class and the poor are due to a severe recession and a sluggish recovery. These short-term setbacks, while very real, do not offset the long-run picture of American progress that is broad and dramatic.

1 Four types of biases have been identified. Outlet bias refers to the inadequate accounting of the movement of purchases toward low-price discount or big-box stores like Walmart. Substitution bias refers to bias in the use of a fixed list of purchasable items when people substitute away from high-relative-price items so the old market basket becomes less relevant. Quality bias refers to inadequate adjustments for the quality improvements in products over time, while new-product bias refers to the omission of or long delay in the incorporation of new products into the CPI.

2 Our analyses of material well-being draw upon data from several large, nationally representative surveys covering the period from 1980 through 2009. Our consumption measures and information on automobiles were derived from the Consumer Expenditure survey (CE), which provides quarterly information on family spending. We used the surveys from the first quarter of 1980 through the first quarter of 2010. Measures of income came from the Annual Social and Economic Supplement to the Current Population Survey (CPS), which provides data for approximately 100,000 households annually in recent years. We used data from the 1981 through 2010 surveys, which provide family income information for the previous calendar year. The results that follow were drawn from known and respected data sets.