What Has Happened to the Incomes of the Middle Class and Poor? Part Three: The Bronze Age

This essay is the third in a series examining income trends for middle-class and poor households. In the first, I showed that once federal safety net programs are fully considered, the incomes of the middle class and poor are no lower today than they were in 2007, prior to the Great Recession. The second essay argued that while income surveys show little growth from 2000 to 2007, approaches that may correct for underreporting of income show faster income growth during this “lost decade” than during the 1980s and 1990s. In this essay, I will begin to explore longer-term trends in household income. 

The story over the long-run is one of strong improvement in living standards, driven in part by greater work and higher wages among women, but in part by continued gains in male earnings, declining family size, lower taxes, and more federal and employer benefits. While the range of estimates is wide, a reasonable conclusion from the totality of the evidence is that poor and middle class households are at least 30 percent richer today than their counterparts from 35 years ago.

As discussed in the previous essays, when economists analyze income trends, they compare years that represent similar points in the business cycle. Typically, they compare peaks, when income reaches a short-term high on the eve of a recession. As we will see in the final essay of the series, income growth was especially high during the “Golden Age” of the 1950s and 1960s. Because income growth in recent decades has been so much slower –call it the “Bronze Age”—many analysts, such as those at the Economic Policy Institute, focus on income growth since the 1973 or 1979 peaks. Using the earlier year is problematic, however, because it came after President Richard Nixon had implemented price controls in advance of the 1972 election. The 1973 income peak is artificial—the consequence of holding inflation down by fiat before letting it accelerate again.

Using 1979 as a starting point also makes sense because the Current Population Survey (CPS) includes information on noncash benefits and taxes starting with that year. As we have seen, these are an important part of the story in interpreting trends.

It is useful to begin once again with the official Census Bureau figures. The “money income” definition includes all private sources of income, the earnings and pensions of government employees and retirees, and cash benefits from the federal government. The latter includes income from Social Security, unemployment, and welfare checks, to name three of the most important cash transfer programs.  As in the earlier essays, I use a better inflation adjustment than that incorporated into the official figures, relying on the Bureau of Economic Analysis’s Personal Consumption Expenditures deflator.  Median money income rose by 25 percent from 1979 to 2007. That is to say, the income in the middle of the household income distribution was 25 percent higher in 2007 than in 1979 after taking the rising cost of living into account.

If we take a step back, we can consider pre-tax and -transfer income so that the federal cash benefits are excluded. My estimates from the CPS and CPS-based estimates from CBO concur. If no adjustment is made for declining household size, median pre-tax and -transfer income rose by 20 to 23 percent, depending on whether or not the value of employer-provided health insurance is included in “income” and whether the median household is considered or the median person’s household income. 

Household size, however, fell over this period. Declining and delayed fertility and marriage led to fewer children per family. When households have fewer mouths to feed, a given amount of income goes a longer way, so even if incomes had not grown, Americans would have been better off. To compare the living standards of two households of different sizes, economists typically assume that if one household is twice as big as another, it needs only 1.4 times the income of the other to be as well off (1.4 is the square root of two).  This adjustment reflects the reality that additional members of a household do not require an equivalent increase in expenditures.  The same mortgage or rent needs to be paid whether a household loses or gains a member. Grocery expenses increase with more people, but separate gallons of milk need not be purchased for each person. Heating bills will not be twice as high if the size of the household doubles. 

For these reasons, economists often divide household incomes by the square root of household size to put them all on the same “needs” scale before computing medians and looking at trends. After adjusting for household size in this way, both my estimates and those of the CBO indicate that median market income rose 26 to 30 percent between 1979 and 2007.

The Census Bureau money income figure excludes noncash transfer payments, such as those from food stamps (the Supplemental Nutrition Assistance Program), public housing benefits, Medicaid, and Medicare. Including these noncash payments, median pre-tax post-transfer income rose 30 to 33 percent without adjusting income for household size and 37 to 42 percent after taking household size into account. Once again, my estimates and those of CBO are in agreement.

Finally, taxes fell over this 28-year period, leaving middle-class households with more disposable income for a given amount of pre-tax income. I estimate that median post-tax and -transfer income rose by 31 to 35 percent before adjusting for household size and by 40 to 43 percent after the adjustment. CBO’s estimates, unlike those in the CPS data, account for corporate and excise taxes but they do not account for state taxes. The CPS data account for state income taxes, but they do not include tax credits in determining liability for either federal or state income tax. Both estimates include payroll taxes. The CBO figures suggest that median post-tax and -transfer income rose by 42 to 49 percent, depending on whether or not household size is adjusted.

As indicators of trends in disposable income, these post-tax and -transfer estimates accord with the median household consumption estimates of Bruce Meyer and James Sullivan. Meyer and Sullivan adjust for household size in a different way than CBO and I do, and their estimates only go back to 1980. They also use a different inflation adjustment than I do. However, once I apply my inflation adjustment, the increase from 1980 to 2007 is 44 percent. Using their preferred inflation adjustment, the increase is over 60 percent.

The first two essays in this series found that the incomes of the poor tracked the incomes of the middle class fairly closely. Is that also true going back to 1979? As the table below shows, it appears that poor households saw somewhat slower growth than did the middle class, though the conclusion is sensitive to how Medicaid and Medicare benefits should be valued. 

I looked at the 20th percentile of household income—the income for the household that is richer than 20 percent of households but poorer than 80 percent of them. I find that pre-tax and -transfer income growth at the 20th percentile lagged that of the middle class—rising just 12 to 16 percent between 1979 and 2007. In contrast, the CBO estimates suggest that the pre-tax and -transfer income of the bottom 20 percent of Americans rose 51 percent. The difference is due to the inclusion in the CBO figures of people in households with no such income. Since more people in the bottom 20 percent today have a worker in the household than in 1979, there are fewer zero-dollar incomes pulling the average down. My estimates do not include people in households with no pre-tax and -transfer income. My figures and those of the CBO are more congruent after transfers and taxes are taken into account. The pre-tax, post-transfer incomes of the poor rose by 22 to 30 percent, depending on whether household size is controlled or not and whether the focus is on households or persons. These figures only value federal health benefits as income if a household otherwise meets a minimum income threshold. The embedded assumption is that health benefits have no value as income until a household can meet its basic needs. If, on the other hand, we assume that all households with federal health benefits fully value them as income, the increase in the 20th percentile from 1979 to 2007 was 33 to 46 percent.

Finally, taking taxes into account as well, the post-tax and -transfer income of the poor rose by either 22 to 28 percent (using the more conservative valuation of Medicaid and Medicare) or 32 to 44 percent (using the full valuation). Meyer and Sullivan’s estimates—using my preferred inflation adjustment—indicate a roughly 34 percent increase in consumption at the tenth percentile. Using their preferred adjustment for inflation, the increase is 50 percent.

It is often claimed that median incomes would have fallen if not for the increase in work among wives. Such accounts point to official Census Bureau estimates showing stagnation in male earnings and contrast them to the official household income estimates. Observers increasingly point to figures from the Brookings Hamilton Project showing steep declines in male earnings after accounting for the increase in non-working men over time. Some observers also object to household-size adjustments on the grounds that smaller families may reflect a growing share of Americans being unable to afford the number of children they prefer.

It is certainly the case that men’s earnings have grown less than household incomes have, and women’s earnings by more than household incomes. However, claims that median male earnings have declined since 1979 must rely on methodological decisions I have shown elsewhere to be inappropriate.  Employment rates among women have increased across the industrialized world, coinciding with rising educational attainment, delayed marriage, and declining fertility. All indications are that these trends reflect the preferences of many women for the satisfaction, independence, and additional income provided by a job over the under-valued and, for many women, unrewarding life of fulltime homemaking and childrearing. 

More to the point, the CPS trends for households broken down by the marital status of the head all show sizable income gains without any household-size adjustments. For all households, the increase in this measure of pre-tax, post-transfer income was 30 percent from 1979 to 2007. For households in which the head was married, the increase was 44 percent. Among households where the spouses were separated, median income grew by 34 percent; for those with a widowed head, 50 percent; among those with a divorced head, 33 percent. Finally, among households with a never-married head, the increase was 34 percent from 1979 to 2007, and the increase for households with a never-married male head was 29 percent.

The fact that incomes have risen by at least 30 percent in the middle and at the bottom feels disappointing in large measure because of the remarkably strong income growth of the Golden Age of the 1950s and 1960s. The next installment of this series will extend our view of income growth back to that earlier era and contrast it with the Bronze Age.