US Middle Class Has Disappeared into Higher-income Groups; Recent Stagnation Explained by Changing Household Demographics?

Over the last few years, I’ve featured a series of posts analyzing the distribution of US household incomes by the shares of various income categories, see the most recent post here. In those posts I alternatively calculated the category “middle-income” as: a) $25,000 to $75,000, b) $35,000 to $75,000 and c) $50,000 to $100,000 based on Census Bureau data on household income available here. Regardless of the exact measure of “middle-income,” I was able to show that the share of “middle-income” American households has been declining since the mid-1960s — not because more households were falling into a “lower-income” category, but rather because more and more US households were moving into an “upper-income” category.

A recent New York Times article has re-visited this topic, and the article’s title reveals their rather pessimistic conclusion that the “Middle Class Shrinks Further as More Fall Out Instead of Climbing Up.” Here’s how the article starts:

The middle class that President Obama identified in his State of the Union speech last week as the foundation of the American economy has been shrinking for almost half a century. In the late 1960s, more than half of the households in the United States were squarely in the middle, earning, in today’s dollars, $35,000 to $100,000 a year. Few people noticed or cared as the size of that group began to fall, because the shift was primarily caused by more Americans climbing the economic ladder into upper-income brackets.

But since 2000, the middle-class share of households has continued to narrow, the main reason being that more people have fallen to the bottom. At the same time, fewer of those in this group fit the traditional image of a married couple with children at home, a gap increasingly filled by the elderly.

This social upheaval helps explain why the president focused on reviving the middle class, offering a raft of proposals squarely aimed at concerns like paying for a college education, taking parental leave, affording child care and buying a home.

The New York Times uses the income category $35,000 to $100,000 (in constant 2013 dollars) as its definition of “middle class” household income, and the chart above shows that income group’s share of all US households in every year from 1967 to 2013, along with the annual shares of “lower-income” households (below $35,000) and “upper-income” households (above $100,000). Here’s what the data show:

1. From 53% in 1967 (and a slightly higher 54.7% in 1968), the share of “middle-income” US households earning between $35,000 and $100,000 (in 2013 dollars) has been gradually decreasing over the last half-century as the NY Times points out. Over the 47-year period between 1967 and 2013, the share of “middle-income” American households has fallen by ten percentage points, from 53% in 1967 to about 43% in 2013. So the Census Bureau data on household income reveal a definite decline in the share of “middle class” households in America, confirming the reports we hear all the time about the “disappearing middle class” (47,000 results from a Google search).

2. But where did the disappearing middle class go? Did they fall into the lower-income category? Not at all, because the share of “lower-income” US households earning less than $35,000 also fell over the last 47 years by about five percentage points, from 39.3% of all US households in 1967 to 34.4% of households in 2013.

3. The biggest shift in household income between 1967 and 2013 took place for the share of American households earning $100,000 or more, which almost tripled from only a 7.7% share of US households in 1967 (1 in 13 households) to a 22.5% share of US households in 2013 (almost one in four). That nearly 15 percentage point increase in the share of US households earning $100,000 or more between 1967 and 2013 (from 7.7% to 22.5%) came about from a 10 percentage point decline in middle-income households (from 53% to 43%) and a 5 percentage point decline in lower-income households (from 39.3% to 34.4%). So as I concluded before, a large part of the middle class did disappear, but they didn’t fall down to a lower-income class, they rose into the upper-income class!

4. Here’s another way to understand the dynamic income shift over the last half century that elevated millions of American households into a higher income category. Whereas “middle class” US households were so numerous in 1967 that they outnumbered “upper class” US households by a ratio of almost 7-to-1, so many American “middle class” households had moved to the “upper class” by 2013 that the ratio of middle-income to upper-income households had fallen to less than 2-to-1. Stated differently, in 1969 there were almost 700 “middle class” US households to every 100 “upper-income” households; but by 2013 there were fewer than 200 “middle class” households per 100 “upper-income” households reflecting the movement of millions of US households who had climbed up the economic ladder to a higher income group. It was an amazing period of increased prosperity, upward mobility and an unprecedented increase in the number and share of American households going from “middle class” to “upper class.” The NY Times does acknowledge that many Americans were rising into higher income brackets through the last part of the last century, but focuses mostly on the recent stall in that rise since the turn of the century.

It’s evident in the graph above that the income shares of US households have stabilized in recent years and that the upward trend in the share of upper-income US households flattened out starting around the turn of the century. It’s also the case that the share of US households in the lower-income category stopped declining around 2000 and has increased from 30.7% in 2000 to 35% in 2011 before declining to 34.4% in 2013. In other words, over the last decade it does appear that the previous positive trends that were characterized by a shift of US households from lower-income and middle-income categories into the upper-income category have stalled. But that recent stagnation shouldn’t stop us from appreciating the fact that over a longer period of time, Census data on income distribution reveal evidence of rising income levels for a rising share of American households as the chart above illustrates.

Question: What accounts for the recent struggles of the middle class and stagnating median household income (see chart above) that the New York Times describes as a “social upheaval” from more households “falling to the bottom”? Here are five possible demographic reasons that could explain why median household income has stagnated over the last decade, and which therefore might help explain the claims about a “shrinking and struggling middle class.”

1. The Census Bureau measures “household income” and not “household compensation.” As fringe benefits, especially medical care coverage and health insurance, make up an increasing share of employee compensation, it’s likely that employee compensation, and therefore “household compensation,” has been increasing over time for most US households even though money household income has been stagnant.

The size and composition of households have changed over time in fundamental ways that distort a comparison of “median household income” over time. For example:

2. The average household size has decreased from 3.28 members in 1967 to 2.62 members in 2000 to a record low 2.54 members in 2013.

3. The share of households with two earners has decreased from 45% in 2000 to only about 39% in 2013, while the share of households with one earners increased from 35% to 37%, and the share of households with no earners increased 20% to 23.7% over that same period (see chart above). Those changes have taken place over the last decade during the exact same period that median household income has stagnated or fallen.

4. Retirees as a share of the US population have increased from 14.75% in 2003 to a record high 16.6% in 2013. That demographic shift has also taken place at the same time that real US median household income has declined (see chart above).

5. The number of average weekly work hours per household declined by more than 10% between 2000 and 2013, from 48.3 hours in 2000 to 43.3 hours in 2013, which would potentially help explain a large part of the 8.6% decline in real median household income over that period (see chart above).

Bottom Line: In other words, important demographic changes that have taken place over the last decade or longer might account for stagnating household income because the average US household today, compared to a household even a decade ago, is: a) smaller, b) has fewer earners on average and is more likely to have no earners, c) is more likely to be a retiree household on a fixed income, d) contributes fewer average weekly work hours, and e) receives a greater share of their compensation in the form of non-taxable fringe benefits.

Therefore, isn’t it possible that the “social upheaval” that has taken place over the last decade is really an “upheaval” in the size, composition and characteristics of a typical US “household” and not necessarily an era of reduced economic opportunities and less upward mobility for the middle class? At the very least, in any discussion about the “middle class” and “household income” we have to recognize that an “American household” is a dynamic concept that is constantly evolving and changing over time, and therefore distorts any comparisons between household incomes today to those of a decade or a generation ago.

Note: The New York Times did some ten-year demographic income trend comparisons based on household differences in age, education, race and family status, but didn’t really address the issue that the composition and characteristics of US households have changed quite dramatically over time, especially in the last decade.