Posted on Friday, September 10, 2010
Before the Current Recession, a Phantom Recovery
Testimony of Pew Research Center Executive Vice President Paul Taylor before the Senate Finance Committee
Chairman Baucus, Ranking Member Grassley and members of this Committee, thank you for the opportunity to testify at this hearing on middle class tax relief, and allow me to make one thing clear at the outset: I am not a tax policy expert. Nor am I an economist, though I have colleagues who are. I am appearing before you today as the principal author of a report the Pew Research Center released last year entitled, "Inside the Middle Class: Bad Times Hit the Good Life." The report combines findings of one of our major national public opinion surveys with the Center's analysis of four decades of demographic and economic trends from the Census Bureau and other sources.
The Pew Research Center does not take positions on policy issues. We are a nonpartisan "fact tank" that generates information we hope will be of value to policymakers. My testimony today summarizes and updates some of the key findings of our 169 page report. Among them:
About half of all Americans think of themselves as middle class. Economically and demographically, they are a surprisingly varied group.
Middle-class Americans believe they've stagnated in the past five years. But they also believe they have a higher standard of living than their parents had.
Income data from the Census say they're right on both fronts. Even before the onset of the current recession, this decade had witnessed the longest stretch in modern U.S. economic history in which median household income failed to surpass an earlier peak (in 1999). However, over a longer haul -- since 1969 -- median household income has risen by 41%, after adjusting both for inflation and changes in household size.
For middle-income Americans, the past several decades have been characterized by rising prosperity and rising inequality. The middle income tier has fallen farther behind the upper income group in both income and wealth.
For the past two decades -- until the current recession -- middle income Americans had been spending more and, relative to their income, borrowing much more. Housing was the key driver of both trends.
Since 1970, the middle income tier in America has shrunk by about 5 percentage points. This small but notable hollowing out has been accompanied by an increase in the share of adults in both the upper and lower income tiers.
Since 1970, the demography of the middle-income tier has changed in much the same way that the composition of the U.S. population as a whole has changed. It is older, better educated, less likely to be white and less likely to be married.
I. How the Middle Class Defines Itself
"Middle class" is a term that is universally familiar but devilishly difficult to pin down. It is both a social and economic construct, and because these domains don't always align, its borders are fuzzy. Is a $30,000-a-year resident in brain surgery lower class? Is a $100,000-a-year plumber upper-middle class? One way to sidestep riddles of this sort is to ask people to label themselves. That's what we did in a telephone survey of a nationally representative sample of 2,413 adults last year.1 It produced a straightforward-seeming result: about half (53%) of all adults in America say they are middle class; and the median annual family income of this group of respondents is $52,285.
But beyond the simplicity of these numbers lies a nest of anomalies. For example, about four-in-ten (41%) adults with $100,000 or more in annual household income say they are middle class. So do nearly half (46%) of those whose household incomes are below $40,000. As for those in between, about a third say they're not in the middle class.
In short, when it comes to self-identifying as middle class, different groups of Americans use different yardsticks. Some cases in point: younger adults and older adults are both more likely than middle-aged adults to describe themselves as middle class, even though, on average, their income levels are lower. Meantime, middle-aged middle-class adults are more likely than those who are younger and older to report financial stresses, even though they have more income. Also, roughly the same percentages of whites (53%), blacks (50%) and Hispanics (54%) self-identify as middle class, despite the fact that the income and wealth of blacks and Hispanics who say they are middle class is much lower than that of whites who say they are middle class. Clearly, declaring oneself to be middle class is more than a statement about income and assets. It's also a state of mind.
Along these same lines, when survey respondents are asked how much money they think it takes for a family of four to lead a "middle-class lifestyle" in their area, their answers array along a sliding scale that correlates with their income. The greater the income, the higher the estimate. Thus, adults in families whose income is between $100,000 and $150,000 a year believe, on average, that it takes $80,000 to live a middle-class life in their area. By contrast, adults in families whose income is less than $30,000 a year believe that a middle-class lifestyle can be had for about $50,000 a year. Analyzing these estimates by the ZIP codes of the respondents yields a similar finding: people who live in communities with a high cost of living think it takes, on average, about $15,000 more to be in the middle class than do people who live in communities with a low cost of living.
The bottom line is that while about half the country considers itself middle class, their judgments are influenced by the laws of relativity. Bearing this in mind, our report undertook a parallel analysis of what it means to be middle class -- this one driven by economic and demographic data rather than by self-definition. Using Census figures, we divided Americans into three income tiers -- low, middle and high. We defined the middle income tier as consisting of adults who live in a household where the annual income falls within 75% to 150% of the national median, a common set of boundaries for analyses of income dispersion.2
This analytical frame enabled us to examine how the middle income tier has changed over time in its size, demographic traits and standard of living. Has it gotten smaller or larger since 1970? How has it fared economically -- both in absolute terms, and relative to the income tiers below and above? Also, by lining up these Census-based trends alongside our survey results, we were able to make some judgments about how well public perceptions square with the underlying economic realities.
II. The Middle Class Sees Short-Term Stagnation, Long Term Progress
Even before the current recession settled in, the American middle class felt stuck in its tracks. The Pew Research survey was conducted in January and February of 2008, during the early days of what was then a still-not-yet officially declared recession. Even so, a majority of self-identified middle class survey respondents said then that in the past five years, they either hadn't moved forward in life or had actually fallen backward. Though there are some differences by income class, the total U.S. population, in the aggregate, felt essentially the same way: some 25% said they hadn't moved forward in life, and 31% said they had fallen backward. This was the most downbeat short-term assessment of personal progress in nearly half a century of polling by the Pew Research Center and the Gallup organization.
However, the public's judgments about its economic mobility aren't all negative. When we asked respondents to measure their progress over a longer time frame, their glasses became more than half full. Two-thirds of all respondents said they had a higher standard of living than their parents had when their parents were their age. Just 14% said their standard of living was lower.
In both of these sets of judgments -- short term stagnation, long term progress -- those who self-identify as middle class fall out where one would expect on the socio-economic ladder. They are more upbeat about their progress than are those who self-identify with the lower class, but less upbeat than those who self-identify with the upper class.3 This pattern repeats itself on a wide range of questions about personal finances that we posed in this survey. For example, the percentage of people who said they "live comfortably" ranges from 66% of the self-defined upper class to 39% of the self-defined middle class to just 9% of the self-defined lower class. And when we asked people if they had cut back on household spending in the past year because money was tight, 75% of the lower class said they had, compared with 53% of the middle class and just 36% of the upper class.
III. Trends in Household Income, 1970-2007
One of the most robust findings of our report is how closely people's perceptions of their economic progress over time square with economic trends from the Census and other key government data sources. When people in the middle class say they haven't moved forward in life in recent years, the economic data say they're right. And when they say that they're doing better than their parents did when they were the same age, the economic data once again say they're right.
Let's first look at the long term trends. From 1970 through 2007, median annual household income increased by a total of 41%, after adjusting both for inflation and for the decline in household size.4 Most economists agree that trends in inflation-adjusted median household income are the best single indicator of a population's changing standard of living. But they are not the only measure. Our report also examined changes over time in household wealth and household expenditures. All three indicators tell essentially the same story: over the past several decades, there has been measurable progress in America's standard of living.
However, trends for the population as a whole do not always explain what has happened to different sub-groups within the population. Have some done better than others? And, in particular, how has the middle-income tier fared since 1970 compared with groups above and below?
When it comes to household income, the middle income group has not fared quite as well as the lower income group (those in households with annual incomes below 75% of the median) or nearly as well as the upper income group (those in households with annual incomes above 150% of the median). Since 1970, the median income of the middle group has risen by 40%, compared with 44% for the lower group and 52% for the upper group (all rates adjusted both for inflation and changes in household size).
A similar analysis of the change over time in the median net worth of these three income tiers produces a more pronounced contrast in fortunes. The median net worth of the upper income tier rose by 115% from 1983 to 2007, compared with a rise of just 68% for the middle income tier and 44% for the lower tier. Thus, over several decades, the middle and lower income groups have been making absolute economic progress while experiencing relative decline.
There is one more trend analysis that fills out the story line. When we look at changes in median household income by individual decade rather than over the 40-year span, we can see why so many Americans feel they are stuck in their tracks -- and why they felt this way even before the onset of the current recession.
As the chart to the right illustrates, the middle income tier has made virtually no gains at all in median household income during this decade (at least not through the end of 2007, the last year for which data for this analysis are available), and neither have the groups above or below them. For the typical American household, the Great Recession that began in December 2007 was preceded by a less dramatic but equally unusual economic phenomenon: a Phantom Recovery. Inflation-adjusted median household income has remained below its 1999 peak in every year since then, first during the shallow recession at the start of this decade and later during the economic "recovery" that ran from the end of 2001 through the end of 2007. In fact, the eight-year period from 1999 to 2007 is the longest in modern U.S. economic history in which median household income failed to surpass an earlier peak. The Census Bureau has not yet released household income data for 2008, but the recession has likely kept this indicator below its 1999 peak for another year, and may keep it there on into the next decade.
The bottom line is that the middle class has it right. Over the long haul, its standard of living has risen. But over the course of the past decade, it hasn't.
IV. The Middle Class Borrows More and Consumes More
One of the important changes in the economic life of the middle class over the past quarter century has been the extent to which it has engaged in higher levels of consumption and taken on a higher burden of debt. At least until the current recession, housing was the key driver of both trends.
From 1980 through 2006, middle income families increased their inflation-adjusted expenditures by 15%. In part, that's because the cost of many of the anchors of a middle class lifestyle -- housing, health care, education -- rose more sharply than inflation. It's also because many new goods and services that didn't exist a few decades ago -- high speed internet, cable and satellite subscriptions, high definition television, among others -- have become commonplace consumer items. (As the chart to the right illustrates, many in the middle class tend to overestimate just how commonplace these items are -- creating a "possessions perception gap" that may have fueled some of this consumption).
The 15% growth in real consumption by the middle income over this time period is about the same at the change experienced by lower income families (16%) but only about half the rate of growth experienced by upper income families (32%).
Somewhat more so than for the income groups above and below them, the middle tier paid for its increased consumption by taking on more debt, most typically in the form of bigger mortgages and second mortgages. Thus, the median debt-to-income ratio of middle income households more than doubled from 1983 to 2004; these rates also rose for the other lower and upper income groups, but not quite at the accelerated pace of the middle group.
Obviously, housing values, expenditure patterns and debt habits are much different today than they had been just a few years ago. It's beyond the scope of this report to analyze these more recent developments, in part because of the lag time in the availability of some of the key data. The story told by the numbers presented here essentially ends before the start of the current recession. But it helps to provide a context for -- and perhaps to foreshadow -- current developments.
V. The Middle Income Tier: Then and Now
Since 1970, the middle income tier in America has shrunk by about 5 percentage points.
In 1970, 40% of all adults in this country lived in a middle income household, with "middle" defined as one where the income falls within 75% to 150% of the median. By 2006, just 35% of adults were in the middle income tier.
This small but notable hollowing out of the middle has been accompanied by an increase in the share of adults in both the lower income category and the upper income category. The rise in share has been greater over this time period for the upper group (to 32% in 2006 from 28% in 1970) than for the lower income tier (to 33% in 2006 from 31% in 1970). Looking at these changes by age group shows that the trends have been very different for the youngest and oldest adults. The 65 and older group has moved ahead during the past 36 years; the 18-to-29 year old group has fallen behind. Among the older group, just 45% were in the lower income tier in 2006, down from 58% in 1970. Among the younger group, 39% were lower income in 2006, up from 30% in 1970.
Some demographic groups have improved their income status since 1970; others have fallen behind.
The period since 1970 has seen a distinct sorting of many different demographic groups into different income tiers. In addition to the elderly, the groups that have gained the most include blacks and native-born Hispanics. Married adults have also done well, while the never-married have fallen behind.
On the gender front, men and women have moved in different directions, depending on marital and work status. Working husbands and working wives both have seen their income positions improve since 1970, but the gains have been greater for working husbands. Among those who are not married, the gender pattern is reversed: single working women's income position has improved since 1970, while single working men's income position has declined. Other groups that have not fared well are young adults, people in lower-skilled jobs, people with less educational attainment, and immigrant Hispanics. The decline for this last group is mainly the result of a heavy influx of low-skilled immigrants, rather than downward mobility among immigrants already in the U.S.
Since 1970, the middle income tier has gotten older, better educated, less likely to be white and less likely to be married.
Demographic changes in the middle income tier since 1970 are very similar to the changes in the U.S. adult population as a whole. The average age for middle income adults was 45 in 2006, up from 41 in 1970 (comparable figures for the full adult population are 46 in 2006 and 44 in 1970). In 1970, 88% of the middle income group was white; by 2006, just 71% was white (comparable figures for the full adult population are 86% in 1970 and 70% in 2006). The ethnic group that moved heavily into the middle income tier during this period was Hispanics: in 1970, they made up just 3% of the middle tier; by 2006, they were 13%.
In 1970, more than three-quarters (76%) of the middle income group were married; by 2006, just 57% were married. But the biggest demographic change has come in levels of educational attainment. In 1970, just one-in-five middle income adults had at least some college education; by 2006, more than half did. As noted on the previous page, never married adults and those with less educational attainment have been among the groups suffering the biggest losses in income status over this period.
Appendix: Adjusting for Household Size
Household income data reported in this study are adjusted for the number of persons in a household. That is done in recognition of the reality that a four-person household with an income of, say, $50,000 faces a tighter budget constraint than a two-person household with the same income. In addition to comparisons across households at a given point in time, this adjustment is useful for measuring changes in the income of households over time. That is because average household size in the United States has decreased from 3.1 persons in 1970 to 2.5 persons in 2007, a drop of 19%. Ignoring this demographic change would mean ignoring a commensurate loosening of the household budget constraint.
Following the practice of other researchers, this study does not make this adjustment on a simple per capita basis -- on the theory that it does not cost twice as much to run a four-person household as it does to run a two-person household. Instead, it adjusts by using a ratio based on the square root of household size. In practical terms, this means that the cost of a one-person house is set at 100; the cost of a two person household is set at 141; the cost of three-person household is set at 173, the cost of a four-person household is set at 200, etc.
As discussed in the main body of the report, adjusting for household size has an effect on trends over time in income since 1970. However, once the adjustment has been made, it is immaterial whether one scales incomes to one-, two-, three- or four-person households. Regardless of the choice of household size, exactly the same results would emerge with respect to the trends in the well-being of lower, middle and upper income groups.
For a more detailed discussion of this and other methodological issues that arose in our research, go to the Appendix on Page 163 of our 2008 Middle Class report.
1. The survey was conducted from Jan. 24 through Feb. 19, 2008 among a nationally representative sample of 2,413 adults. The survey design included an over-sample of blacks and Hispanics, as well as a dual sample frame of respondents reached via landline (1,659) or cell (754) phone. All data are weighted to produce results from a representative sample of the full adult population. Margin of sampling error is plus or minus 2.5 percentage points for results based on the total sample at the 95% confidence level.
2. The boundaries of this middle tier vary by household size; in 2007, they were $45,676 to $91,351 for a family of three, which is close to the typical household size in the U.S. All dollar figures inflation-adjusted to January 2009.
3. Throughout this report, we combine respondents who say they are "upper" and "upper middle" into a single "upper class" category and we combine respondents who say they are "lower middle" and "lower" into a single "lower class" category.
4. In 1970, the typical household had 3.1 people. In 2007, the typical household had just 2.5 people. This trend is the result of a decline in fertility (leading to smaller families) and a decline in the years that adults spend being married (leading to more single person households). Without making an adjustment for this change, the increase in median household income from 1970-2007 would be just 21%. For more on how we made this adjustment, see appendix.