WikiLeaks Document Release http://wikileaks.org/wiki/CRS-RS22627 February 2, 2009 Congressional Research Service Report RS22627 Who Are the "Middle Class"? Brian W. Cashell, Government and Finance Division October 22, 2008 Abstract. There is no consensus definition of "middle class," neither is there an official government definition. What constitutes the middle class is relative, subjective, and not easily defined. The mid-point in the distribution is the median, and in 2007 the median household income was $50,233. How far above and below that amount the middle stretches remains an open question. The U.S. Census Bureau has published figures for 2007 breaking the income distribution into quintiles, or fifths. The narrowest view of who might be considered middle class based on that presentation would include those in the middle quintile, which includes households with income between $39,100 and $62,000. A more generous definition might be based on the three middle quintiles, those households with income between $20,291 and $100,000. Surveys suggest that from 1% to just over 3% of the population consider themselves to be upper class. Comparing those figures with the income distribution would put the dividing line between middle and upper class close to, if not above, $250,000. Similarly, survey responses suggest that the lower end of the middle class might be close to $40,000. ¢ http://wikileaks.org/wiki/CRS-RS22627 Prepared for Members and Committees of Congress ¢ There is no consensus definition of "middle class," neither is there an official government definition. What constitutes the middle class is relative, subjective, and not easily defined. The mid-point in the distribution is the median, and in 2007 the median household income was $50,233. How far above and below that amount the middle stretches remains an open question. The U.S. Census Bureau has published figures for 2007 breaking the income distribution into quintiles, or fifths. The narrowest view of who might be considered middle class based on that presentation would include those in the middle quintile, which includes households with income between $39,100 and $62,000. A more generous definition might be based on the three middle quintiles, those households with income between $20,291 and $100,000. Surveys suggest that from 1% to just over 3% of the population consider themselves to be upper class. Comparing those figures with the income distribution would put the dividing line between middle and upper class close to, if not above, $250,000. Similarly, survey responses suggest that the lower end of the middle class might be close to $40,000. http://wikileaks.org/wiki/CRS-RS22627 M uch of the legislation considered by Congress is in the name of the so-called "middle class." But there is no consensus definition of middle class. Neither is there an official government definition, and it is not the aim of this report to establish one. What constitutes the middle class is relative, subjective, and not easily defined. Most people likely have decided views as to whether they are middle class. At the same time, those who refer to the middle class have a rough idea whom they have in mind. How closely these two definitions correspond is another matter. In some contexts, the term middle class may refer to a group with shared values or views, but much of the time it is intended to refer to those who fall within a particular range of incomes. For example, a tax cut proposal may be promoted on the grounds that it would benefit the middle class. Where the distinction is based solely on income, the term "middle income" is sometimes used. This report will use the terms middle class and middle income interchangeably. What defines the middle class is discussed here solely in terms of income. This paper attempts to put the term middle class, or middle income, into some perspective. It begins by presenting available income distribution data as a way of identifying what constitutes middle income. Then, it presents the results of some subjective surveys in an attempt to identify http://wikileaks.org/wiki/CRS-RS22627 what income levels are considered to fall into the middle class. Finally it discusses some economic findings that may help explain how those in the middle class feel about their place in the distribution. For the purposes of this report, the measure of income used is household money income.1 The Census Bureau publishes income distributions for both households and families, with households including slightly more of the population. Households are relevant to the discussion here because members pool whatever income they have and can be presumed to share a common standard of living.2 Any discussion of the middle income necessarily starts with the very middle. The mid-point in the distribution is the median household income, and in 2007 it was $50,233. How far the middle stretches above and below that amount is the issue. While distribution figures are available for fairly small income classes, what many analysts find most useful are data showing the distribution by "quintile." The population of households is divided into fifths, and these quintiles are arranged from lowest to highest income. Table 1 shows what income levels separate the five quintiles from one another in 2007, and also show what share of total household income is accounted for by each quintile. In addition, the Census Bureau provides separate data for the top 5% of the households in the income distribution. 1 Money income accounts for a wide range of income sources, but it is usually incomplete. Money income includes income from earnings, interest and dividends, Social Security, and other forms of social insurance. It does not include the value of non-money benefits such as food stamps or housing subsidies. Neither does it include capital gains. 2 In 2007, there were an estimated 116,783,000 households in the United States. 7002 ,selitniuQ emocnI dlohesuoH .1 elbaT elitniuQ fo egnaR emocnI erahS emocnI elitniuQ fo elitniuq mottob 192,02$ naht ssel %4.3 elitniuq dnoces 001,93$ ot 192,02$ %7.8 elitniuq elddim 000,26$ ot 001,93$ %8.41 elitniuq htruof 000,001$ ot 000,26$ %4.32 elitniuq pot 000,001$ naht erom %7.94 %5 pot 000,771$ naht erom %2.12 .uaeruB susneC .S.U ,ecremmoC fo tnemtrapeD .S.U :ecruoS These figures make it possible to consider a more inclusive definition of the middle than median income. A narrow view of who might be considered middle class would include only those in the middle quintile, those households with income between $39,100 and $62,000. But it seems unlikely that so small an income range would correspond with many impressions of who is http://wikileaks.org/wiki/CRS-RS22627 middle class. A more generous definition might be based on the three middle quintiles, those households with income between $20,291 and $100,000. That group accounts for 60% of all households and 46.9% of all household income. Perhaps the broadest definition of middle class to be had from these numbers would be to add the part of the top quintile just up to the point where the top 5% begins. That would put those households with income between $20,291 and $177,000 in the middle class. That group accounted for 75.4% of all household income in 2007. One other consideration as to where to draw the lower income limit for the middle class is the official poverty threshold. There are multiple poverty thresholds depending on size of household and ages of children. Given that the average household size in 2007 was 2.6 people, the poverty threshold for a family of three might be appropriate. That threshold was $16,530 in 2007. Whether a household just above the poverty threshold might be considered part of the middle class seems subject to debate. Clearly the further a household is above or below the median income, the more subjective its inclusion in the middle class becomes. ¢ Since who is middle class is a subjective question, the way the term is used might help define it. Thus, opinion surveys might provide some basis for identifying the relevant income bounds. A number of surveys in recent years have asked people to indicate to what social class they consider themselves to belong. The classes they are usually asked to choose from are lower, working, middle (sometimes divided into lower and upper middle), and upper. For example, a New York Times survey, in May 2005, asked people to identify their social class.3 The results showed only 1% of those surveyed considered themselves to be upper class, 67% considered themselves to be middle or upper-middle class, 35% were working class and 7% were lower class. That the official poverty rate in 2007 was 12.5% suggests that either some who are in poverty consider themselves to be working class (which may be because they believe their poverty status to be temporary), or there were problems surveying those at that end of the distribution. In the Census Bureau's detailed household income distribution data, the highest income class is $250,000. The share of households with income over $250,000 accounts for 1.9% of all households. If the survey answers correspond to the income data, then the self-reported middle class, broadly defined, includes households with income over $250,000. Similarly, comparing the survey responses with the income data puts the lower end of the middle class just over $40,000 in 2007.4 The National Opinion Research Center (NORC) at the University of Chicago has asked people to identify their social class in a number of surveys beginning in 1972. The cumulative results of those surveys can also be used to compare the income distribution with self-reported class divisions. The average of the NORC survey data between 1972 and 2006 suggests that 3.3% of the population consider themselves to be upper class. That would put the dividing line between http://wikileaks.org/wiki/CRS-RS22627 middle and upper class at just over $200,000 in 2007. The NORC surveys indicated that 4.8% of the population considered themselves to be lower class, and 47.8% classified themselves as working class. That would put the dividing line between working class and middle class at about $52,500, again assuming a correspondence between the survey data and the income distribution. The Pew Research Center sampled opinions regarding middle class attitudes.5 They found that 53% of Americans considered themselves to be middle class. Comparing the proportion of those who identified themselves as middle class could include incomes from about $25,000 up to $95,000. When those who considered themselves to be either lower or upper middle class the income range expanded to include those with income near $10,000 up to just under $250,000. Given the income quintile data presented in the previous section, the survey responses seem to suggest that while some households in the second quintile ($20,291 to $39,100) might be considered middle class, the term middle class might be more likely to refer to those in the middle and fourth quintile ($39,100 to $100,000), and to many of those households between the 80th and 95th percentile ($100,000 to $177,000) in the distribution as well. Among the many assumptions economists make to facilitate analysis is that of diminishing marginal utility of income. This concept refers to the assumption that as income increases, each additional dollar yields less satisfaction than the one that came before. With respect to middle 3 The New York Times published a series of articles in May 2005 under the title "Class Matters," which are available at http://www.nytimes.com/indexes/2005/05/15/national/class/index.html. 4 Because the upper end of the income distribution is a small proportion of the population, it is possible this group is under-represented in surveys like this. That would mean relatively greater measurement error in their response. 5 Pew Research Center, Inside the Middle Class: Bad Times Hit the Good Life, April 2008, available at http://pewsocialtrends.org/pubs/706/middle-class-poll. income it is meaningful because, if true, it means that there are greater gains in satisfaction to be had moving up into the middle class than there are to be had moving up from the middle to the upper end of the income distribution. It might not be unreasonable to say that those who consider themselves middle class are relatively content, at least with their economic situation. But while a middle-income household may be well above a subsistence level of income, the satisfaction or happiness afforded by that income may also depend on where that income level fits into the overall income distribution. The idea that happiness depends on both the absolute and the relative level of income is known as the relative income hypothesis. If individuals care about where they are situated in the overall distribution, that would seem evidence of the economic importance of a middle-income group with a shared stake in the health of the economy. Interest in the economics of happiness has led to a number of studies that have found evidence of the importance of one's place in the overall distribution of income. While ultimately happiness is subjective, one study analyzed what would seem to be an objective measure of happiness (or, more accurately, unhappiness). Daly and Wilson looked at how changes in income may have influenced suicide rates.6 They found evidence that changes in relative income had a significant http://wikileaks.org/wiki/CRS-RS22627 influence on suicide rates. The authors used three measures of relative income: the ratio of income at the 90th percentile to median income, the ratio of median income to income at the 10th percentile, and the ratio of income at the 90th percentile to income at the 10th percentile in the distribution. Among their findings was that an increase in the 90/10 ratio, interpreted as an increase in overall income inequality, was not associated with an increase in suicide rates. However the study also suggested that an increase in the 50/10 ratio was associated with a decline in the suicide rate, and that an increase in the 90/50 ratio was associated with an increase in the suicide rate. This was considered evidence that relative income had an important influence on happiness. In particular, those at the middle (50th percentile) were happier the larger the gap was between them and those at the lower end (10th percentile) of the distribution. At the same time those at the middle were less happy the larger the gap was between them and those at the upper end (90th percentile) of the distribution. Luttmer also presents evidence that relative income is an important determinant of happiness.7 He used survey data that matched individuals' self-reported happiness with income data. He found that an increase in one's own income raised reported happiness, but that an increase in one's neighbors' income had a negative effect. Further he found that a decline in one's own income resulted in about the same decrease in happiness as occurred when one's neighbor experienced an increase in income. 6 Mary Daly and Dan Wilson, "Keeping Up with the Joneses and Staying Ahead of the Smiths: Evidence from Suicide Data," Federal Reserve Bank of San Francisco Working Paper 2006-12, April 2006, 41 pp. 7 Erzo F.P. Luttmer, "Neighbors as Negatives: Relative Earnings and Well-Being," Quarterly Journal of Economics, August 2005, pp. 963-1002. Easterlin surveyed the literature addressing the connection between happiness and income and developed a model to explain it.8 He found a number of studies showing that at any given time cross section comparisons were likely to show that higher income is correlated with greater happiness. At the same time, it was also typical that as income rose over the course of one's lifetime there was no corresponding increase in happiness. Easterlin suggested that self assessments of happiness at a particular income level are dependent on aspirations. As incomes rise, so do aspirations. The rise in aspirations has an effect on happiness that tends to offset the effect of rising income. This hypothesis would explain the apparently contradictory evidence that in cross section analysis, happiness rises with income, but that over time as income rises happiness is relatively stable. The conclusions of these studies might not seem robust, since they are based on crude self assessments of happiness or well being. But if they are correct, they may be relevant to the notion of a middle class. Together they suggest that happiness, insofar as it is determined by income, depends on one's status in the overall distribution. The Daly and Wilson study looked at specific points in the distribution to reach their findings and related them to median income (50th percentile). But it might not be unreasonable to apply their conclusions to a larger group at the middle of the distribution. In other words, the happiness of the middle class, however that might http://wikileaks.org/wiki/CRS-RS22627 be defined, could be argued to depend on both keeping up with the Joneses and staying ahead of the Smiths as they put it in their article. The happiness of those who identify themselves as middle class would seem to depend on what happens to those with less as well as those with more income. No attempt to identify the middle class in the income distribution can be expected to yield a precise answer. But the term is used so often that it is worth the effort to attach some numbers to it. If the middle class is taken to be those who have more than enough to afford basic necessities, it can be presumed to exclude those at or near the poverty thresholds. Surveys indicate many people felt an income near $40,000 was the minimum to be considered middle class. On the other end, surveys suggested that those with income approaching $200,000 might still be considered middle class.9 Whatever else they may have in common, those who constitute the middle class may have, more or less, similar sentiments regarding their position in the income distribution. Being well above the bottom is a source of satisfaction. But, when those at the upper end of the distribution fare better than they do, it is a source of consternation. 8 Richard A. Easterlin, "Income and Happiness: Towards a Unified Theory," The Economic Journal, July 2001, pp. 465-484. 9 If those who consider themselves to be upper class were under-represented in the survey, then the actual number of those who consider themselves to be upper class would be larger. If that is the case, the actual level of income that divides the middle from the upper class would be lower than is suggested by the survey. Brian W. Cashell Specialist in Macroeconomic Policy bcashell@crs.loc.gov, 7-7816 http://wikileaks.org/wiki/CRS-RS22627