Carpe Diem

Explaining income inequality by household demographics


This week, President Obama called America’s “dangerous and growing” income inequality the “defining challenge of our time,” and he plans to put the issue of income inequality at the center of his agenda during the remainder of his second term. The president is not alone in his concern about income inequality – there’s been a lot of discussion in recent years on the issue, especially concerns about “increasing income inequality.” That concern is demonstrated by more than 1 million search results for the term “increasing income inequality” on Google. However, there’s apparently not as much attention or concern about “explaining income inequality” (there are only 178,000 Google search results for that term), and that’s the topic that this post will attempt to address.

Most of the discussion on income inequality focuses on the relative differences over time between low-income and high-income American households, but it’s also instructive to analyze the demographic differences among income groups at a given point in time to answer the question: How are high-income households different from low-income households?

The chart above (click to enlarge) shows some key demographic characteristics of U.S. households by income quintiles for 2012, using updated data from the Census Bureau (here, here and here, and see my previous versions of this analysis for years 2009 and 2010 and 2011).

Below is a summary of some of the key demographic differences between American households in different income quintiles in 2012:

1. Mean number of earners per household. On average, there are significantly more income earners per household in the top income quintile households (2.04) than earners per household in the lowest-income households (0.45). It can also be seen that the average number of earners increases for each higher income quintile, demonstrating that one of the main factors in explaining differences in income among U.S. households is the number of earners per household. Also, the unadjusted ratio of average income for the highest to lowest quintile of 15.8 times ($181,905 to $11.490), falls to a ratio of only 3.5 times when comparing “income per earner” of the two quintiles: $89,169 for the top fifth to $25,533 for the bottom fifth.

2. Share of households with no earners. Sixty-one percent of U.S. households in the bottom fifth of Americans by income had no earners for the entire year in 2012. In contrast, only 3% of the households in the top fifth had no earners in 2012, providing more evidence of the strong relationship between household income and income earners per household.

3. Marital status of householders. Married-couple households represent a much greater share of the top income quintile (77.5%) than for the bottom income quintile (17%), and single-parent or single households represented a much greater share of the bottom 20% of households (83.0%) than for the top 20% (22.5%). Like for the average number of earners per household, the share of married-couple households also increases for each higher income quintile.

4. Age of householders. Almost 8 out of every 10 households (79.5%) in the top income quintile included individuals in their prime earning years between the ages of 35-64, compared to fewer than half (47.3%) of household members in the bottom fifth who were in that prime earning age group last year. The share of householders in the prime earning age group of 35-64 year olds increases with each higher income quintile.

Compared to members of the top income quintile of households by income, household members in the bottom income quintile were 1.6 times more likely (23.5% vs. 14.8%) to be in the youngest age group (under 35 years), and more than 5 times more likely (29.2% vs. 5.7%) to be in the oldest age group (65 years and over).

6. Work status of householders. More than four times as many top quintile households included at least one adult who was working full-time in 2012 (78.2%) compared to the bottom income quintile (only 18.2%), and more than five times as many households in the bottom quintile included adults who did not work at all (67.3%) compared to top quintile households whose family members did not work (12.5%). The share of householders working full-time increases at each higher income quintile.

7. Education of householders. Family members of households in the top fifth by income were six times more likely to have a college degree (77.2%) than members of households in the bottom income quintile (only 12.9%). In contrast, householders in the lowest income quintile were 26 times more likely than those in the top income quintile to have less than a high school degree in 2012 (26.7 % vs. 1.0%). As expected, the Census data show that there is a significantly positive relationship between education and income.

Bottom Line: Household demographics, including the average number of earners per household and the marital status, age, and education of householders are all very highly correlated with household income. Specifically, high-income households have a greater average number of income-earners than households in lower-income quintiles, and individuals in high income households are far more likely than individuals in low-income households to be well-educated, married, working full-time, and in their prime earning years. In contrast, individuals in lower-income households are far more likely than their counterparts in higher-income households to be less-educated, working part-time, either very young (under 35 years) or very old (over 65 years), and living in single-parent households.

The good news is that the key demographic factors that explain differences in household income are not fixed over our lifetimes and are largely under our control (e.g. staying in school, getting and staying married,etc.), which means that individuals and households are not destined to remain in a single income quintile forever. Fortunately, studies that track people over time indicate that individuals and households move up and down the income quintiles over their lifetimes, as the key demographic variables highlighted above change, see CD posts here, here and here. And Thomas Sowell pointed out earlier this year in his column “Income Mobility” that:

Most working Americans who were initially in the bottom 20% of income-earners, rise out of that bottom 20%. More of them end up in the top 20% than remain in the bottom 20%. People who were initially in the bottom 20% in income have had the highest rate of increase in their incomes, while those who were initially in the top 20% have had the lowest. This is the direct opposite of the pattern found when following income brackets over time, rather than following individual people.

It’s highly likely that most of today’s high-income, college-educated, married individuals who are now in their peak earning years were in a lower-income quintile in their prior, single younger years, before they acquired education and job experience. It’s also likely that individuals in today’s top income quintiles will move back down to a lower income quintile in the future during their retirement years, which is just part of the natural lifetime cycle of moving up and down the income quintiles for most Americans. So when we hear the President and the media talk about an “income inequality crisis” in America, we should keep in mind that basic household demographics go a long way towards explaining the differences in household income in the United States. And because the key income-determining demographic variables change over a person’s lifetime, income mobility and the American dream are still “alive and well” in the US.

22 thoughts on “Explaining income inequality by household demographics

  1. There were also a lot of very logical reasons why Germans were in trouble during the 1920-30s. Blaming non-Aryans had more popular appeal. Emotions sell. Thinking requires effort.

  2. Dr. Perry:

    From the blurb on Amazon for Tyler Cowen’s book “Average is Over”: “The widening gap between rich and poor means dealing with one big, uncomfortable truth: If you’re not at the top, you’re at the bottom.”

    Could you put that comment in perspective in light of your post?

  3. would be interesting to add a row that says – “has been arrested” and another that says “arrested for drugs”

    and perhaps a 3rd – dollars of entitlements received.

    oh and a fourth: gets earned income credit
    and itemizes deductions

  4. I think what is more telling is the Second & Fourth Quintile. The top & bottom always includes the outliers that will never fit in.
    The average pay per earner in the Second was about 75% of the Fourth but the second only had half the earners & were twice as likely to be single. They both had a part-time worker about the same.
    Since these groups both had about 22-23 under 35 biggest difference is the 18% jump in over 65. The main reason why some people are in the second vs. fourth quintile is that they are retired singles living modest but probably not bad lives.
    While that is not everyone’s dream retirement, it is not a bad way to go & probably worth more “inequity”.

  5. more than five times as many households in the bottom quintile included adults who did not work at all (67.3%)“….

    So jobs in the prison laundry and kitchen don’t count, eh?

    • Nice chart from the EPI…which was founded by Robert Reich and some Dukakis flunkies. Not exactly truth seeking. It’s meaningless, unless of course you are a class warrior and need some ammo. Its not an apples to apples comparison. If you follow the people earning $47K in 1979, and find out what those same people are earning now…people move in and out of those quintiles. There was also a huge increase in the divorce rate and a gajillion other variables which also changes the story.

      • Let’s ignore any elephants in a room, like “a rapid concentration of wealth at the extreme high end,” stagnant or falling real wages with rising productivity, or a decline in the real minimum wage, while teen unemployment increased (represented by the steep decline in the teen labor force participation rate and unemployment rate).

    • Higher U.S. wages may attract top computer engineers from India.

      However, prices don’t have to fall, profits can rise instead.

      More productive low-skiled immigrants, with a stronger work ethic, can benefit the U.S. economy.

      And, a higher minimum wage can attract them.

  6. MP –
    I agree with most of your post. But I like Charles Murray’s analysis (of this issue with just White America) in Coming Apart, especially when he compares 1960 with today, and Fishtown (Phila — which I grew up next neighborhood over) and Belmont (MA).
    The point that he makes, and you miss, is that over time the nation has become less socially mobile — we are more likely to live in ‘bubbles’ of our own ‘class’ today, and will likely stay there. He thinks the American sense of community is coming apart. You make the dynamics apear as a natural phenomenon.

    Values like education, marriage, civics, etc. are not temporary in consequences. As Murray points out, and I agree (even growing up in a place like Fishtown), in 1960 we were a little more of a ‘classless’ society or at least mixed more in comparison to today where we are more likely to live in a ‘bubble’. So if the trend continues, the outlook is not good.

  7. It would be interesting to see the effect of the baby boomer aging. Like us or not the baby boomers were the last generation that grew up with a more unified culture. AND before the entitlement class.

    It would also be interesting to see the employer, i.e. government, self employed, small business, large corporation.

  8. Thanks for cogent analysis of attributes associated with income inequality by quintile. Marital status, age of householders, work status, and education of householders all have meaning as we think through proposed policies that could be undertaken.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>