Scott Sumner  

The (white) kids are alright

Great Moments in Economic Hist... The Long and the Short Runs...

This post is about Charles Murray's book entitled Coming Apart, which I have not read. It's a bit presumptuous for me to comment on a book merely based on book reviews. So consider this an initial inquiry, I'm looking to be better educated on these issues.

The reviewers suggest that in Coming Apart, Murray claims that well-educated affluent families are doing better and better, while lower and working class families are struggling along all sorts of socioeconomic dimensions. Furthermore, the success of the upper classes gets passed on to their children though various cultural practices related to child rearing (and perhaps genetics as well.) Reviews suggest that Robert Putnam's new book also points to a decline in social capital, especially among the lower classes.

I presume there must be at least a bit of truth in what these guys are saying, but I'm really perplexed by the graph below, which appeared in The Economist, and is based on a Brookings Institution study:

Screen Shot 2015-06-06 at 2.37.55 PM.png
Let's start with the second and fourth quintiles of income at birth. There appears to be almost no difference in the distribution of future income between white kids born into what some might call the lower and upper middle class. (As an aside, in America the upper middle class actually refers to people in the top quintile, but not the top 1%. I'm following the conventional practice of interpreting the income inequality data literally.)

I find these results to be stunning. The entire middle 60 percent has pretty similar life outcomes. But that's not really what Murray is talking about, at least based on this NYR of Books review:

But Coming Apart is also a book about class. Or more precisely, two contrasting classes of white Americans. One, a "new upper class," includes not just the rich and powerful, since it takes in a generous 20 percent of the population. By my calculations, it starts with families earning $135,724. The other, a "new lower class," is everyone in the bottom 30 percent. Its top income, also by my count, would be $52,057. Nor are these classes wholly economic; Murray adds educational and occupational status to give a more rounded portrayal. Thus everyone in his upper class must have completed college and hold a professional or managerial position. He explains what makes both these classes "new," and why conventional rubrics no longer apply. No discussion is given to the remaining 50 percent, which is odd, since they are literally mid-America and cast most of the votes in presidential elections.
This is an important point, as a casual reader of these book reviews (including me at first) might assume Murray had in mind a "Two Americas" model, whereas he actually seems to have in mind a model of three or more Americas. And the data suggests the middle 60% is pretty egalitarian, at least in an intergenerational sense.

So what about the bottom 30%? The graph above doesn't break the data into deciles, but the bottom 20% still shows a surprising amount of upward mobility, with 16% of white lower class kids rising to the top quintile, instead of 20% in a perfectly equal society. In a sense these findings conform to my casual empiricism; white America has a huge middle class and very small upper and lower classes. So why am I stunned? Because almost everyone I talk to tells me that I'm wrong, that we are becoming a nation of "Two Americas", with a huge gap opening up between lower and upper class, even among white Americans.

Although I always thought the inequality issue was being overstated, I never imagined the data would look this equal for whites, especially among the bottom 4 quintiles. Only in the top quintile does the inequality become somewhat pronounced, but even there 10% of kids of affluent white families end up in the bottom quintile, a not insignificant figure (again 20% would represent complete equality.) How many of those rich kids would end up in the bottom quintile if life outcomes depended entirely on inherited IQ, and not at all on environmental advantages like private schools and tutors? (I suspect that Bryan Caplan is right about education.)

This all made me wonder if there is some quirk in the statistical technique that made this sort of equality almost inevitable. Perhaps, but when you switch to black Americans, you do see the sort of inequality that one might have expected (unfortunately there is too little data to report the top quintile):

Screen Shot 2015-06-06 at 12.07.32 PM.png
I find the difference in the life outcome of bottom quintile black kids and bottom quintile white kids to be stunning. Not the fact of the difference, but the size of it.

Question for those who have read Coming Apart. Does the white data somewhat refute Murray's claims, am I misinterpreting the data, or am I misinterpreting Murray?

One reason I'm still skeptical of the Two America's hypothesis that you see in the popular press is that the income distribution doesn't look like a two-humped camel. Is America really divided between an upper and lower class, or do we have a huge middle class? At a racial level I do think there is some support for the Two Americas view. Blacks and Native Americans and to some extent Hispanics really do earn much less on average than white and Asian Americans. But Murray's book specifically focused on whites---that's the issue I'm examining here.

Robert Putnam focuses on the declining status of America's kids. Again, I don't doubt that his book is full of supporting evidence. But couldn't one find evidence in the other direction? Have test scores been rising? Is teenage motherhood now at the lowest level in all of world history? Isn't the crime rate sharply lower?

This is from a very interesting article on Los Angeles street gangs, entitled "This is How Gangs End":

Los Angeles gave America the modern street gang. Groups like the Crips and MS-13 have spread from coast to coast, and even abroad. But on California's streets they have been vanishing"
Question. What sort of data do Murray and Putnam have for the claim that lower class kids are doing much more poorly?

PS. If our kids are doing better than we think, might it reflect the age-old bias that grown-ups have about the young? When I hear older people talk about the young they often seem to me to sound slightly hysterical. I think to myself (without saying this out loud) "Why are you so horrified by the sort of behavior that you and your friends used to engage in as teens and young adults?"

Comments and Sharing

COMMENTS (23 to date)
E. Harding writes:

There's one issue I'm seeing with the graphs: the median income for Blacks is much lower than for Whites. So if we're to measure income mobility for U.S. Blacks, it may be best not to use income quintiles for all Americans, but for Blacks only. Same for Whites. This is to see how strong within-race Black income mobility is, as compared with the (as you can see, low) Black income mobility within the entire U.S.
"Have test scores been rising?"
-For Asians (on ACT and SAT). Are you referring to Flynn effect?
Also, I'd really be interested in seeing data about Chinese, Sub-Saharan African immigrant, and Vietnamese income mobility in the U.S..

ThomasH writes:

I suspect the reason for the perception of the "two classes" perception is that those who write and read about this are mainly ~ 9th decile people lookinig at the gap between them and the top mil-ile.

Scott Sumner writes:

E. Harding, All good points. I wasn't just thinking of the Flynn effect---I also seem to recall reading about modest increases on some of the other standardized tests.

So am I correct in assuming that the fact that 16% of poor white kids ends up in the top overall quintile is misleading, because a lower percentage would end up in the top white quintile?

Thomas, That could be.

Airman Spry Shark writes:

Do you have a link to the Economist article? I'd like to see what is said around the graphics as they're not particularly intuitive (i.e., I find it quite confusing that the 1st quintile, displayed to the top & left of the bubble plots, is the poorest).

E. Harding writes:

I don't know of any big rise in U.S. standardized test scores other than for Asians.
Scott, I think you'd be correct on that.

Jim Dow writes:

I think ThomasH has it. There's a top group with the advantages we think of with the "upper class", there's a middle group that is functional but without those advantages, and then there's a bottom group that often is very dysfunctional. Those at the top have problems differentiating between the two groups farther down.

Lorenzo from Oz writes:

A better URL for the gang piece is:

Richard writes:

You're looking at the income they're born into, and then outcomes at 40. But most parents are under 40. So you're comparing 20-30 year olds to 40 year olds in the chart, which seems wrong.

Also, poor immigrants and higher birthrates at the bottom keep inflating the lower class, which probably matters too.

blink writes:

Seconding Airman Spry Shark, do you have links to the Economist article or the original data?

In particular, I would like to know the percent of whites/blacks in each quintile (for reasons noted by E. Harding). Also, I would like to compare these to overall mobility for the entire population. This should provide an easy-to-interpret Markovian matrix which can serve as a benchmark.

Josiah writes:

I don't know about Putnam, but Murray doesn't talk much about income in Coming Apart. His focus is more on things like marriage and single parenthood, community involvement, criminality, and self-reported happiness. He does talk about employment and disability rates, which are indirectly about income, but he doesn't speak on income directly.

Gaspard writes:

Brookings article

Sadly the journalists and the thinktanks they rely on will keep churning stuff out on mobility using very noisy and misleading income data because that's all they have.

As I understand it there is more churn among whites between quintiles because their income position is more loosely correlated to their overall class position taking into account assets, education, family networks. The 10% who went from top to bottom quintile can be people in transitions of some kind rather than real drop outs.

Gregory Clark in his book The Son Also Rises, found that reversion to the mean does not support Murray's idea of a permanent cleavage but that true mobility is way slower than these single generation income snapshots indicate, which fits with what would seem intuitive

Curt Doolittle writes:


Murray, like most conservatives, is studying, and conveying observations about our change in NORMATIVE capital, not income or consumption.

Deviation from northern european traditional norms is a luxury good ( the absolute nuclear family, delayed marriage, delayed reproduction, high investment parenting, the manorial/protestant work ethic, hight trust from homogeneity, truth-telling/testimony ).


(a)While you haven't read the book, the fact that you, who are one of our very best (IMHO), immediately assume the mainstream bias that income (an easily visible measure) is somehow meaningful rather than justification of priors provides a more valuable insight into the 'mathiness' of mainstream economics, than murray's book does about the destruction of the family as the central unit of inter-temporal reproduction and temporal production that was in no small part, caused by that mainstream bias and 'mathiness'.

(b) No economic hypothesis can be 'true' in the sense that it is descriptively complete, and therefore free of error, bias, and deception, if we fail to account for the full spectrum of costs in the full spectrum of time frames. That is after all, the only measure of costs: opportunity costs. So solving for income or consumption demonstrates a selection bias, under the assumption that all negative externalities are less 'bad' than the 'good' produced by observable increases in income and consumption.

In other words, if we stack all possible forms of capital by the length of the production cycle and it's corresponding consumption or decay, then what is the net change?

The conservative mind is biased to the long term, to saving, to risk, and to disgust. It is a reproductive strategy - a very masculine one perhaps - and the absolute nuclear family is central to it. And it was a very expensive reproductive strategy to develop - which is why was unique.

He does not make the leap (not being an economist) to the extremely damaging suggestion that we move people to capital (a heavy industrial era bias) and it's destruction of the family and its impact upon norms, instead of moving capital to people (a post-heaving-industrial economy) in order to preserve and expand normative capital.

America's dirty secret is that pervasive consumption is an insufficient reward for loneliness and isolation. Americans are heavily drug dependent for the sole reason that they are the most lonely and isolated peoples on earth, for whom the media is a poor substitute for friends and family. The absolute nuclear family is necessary, perhaps, but it can only persist within a civic society. The civic society is a product of the absolute nuclear family. It cannot exist otherwise.

So what is the cost of the destruction of the family in pursuit of income and consumption?

What will be the cost of 40% of american women on anti-depressants?

Mathiness is most visible in the selection bias demonstrated by measuring temporally differential income rather than inter-temporarily differential consumption. But that is not the most important effect of quantitative pseudoscience: it is the destruction of long term capital in favor of short term consumption and the placement of faith in technology to rescue us from the consequences of it.

So, it is not so trivial a question as you suppose.

It's an illustration of everything that is wrong with modern macro's mathniess.

It's not the use of math. It's measuring in favor of bias.

Curt Doolittle
The Propertarian Institute
Kiev, Ukraine

BC writes:

Scott is right about the striking upward mobility for Whites. Whites in the bottom 3 quintiles are all more likely to move up than down, and those in the 4th are about equally likely to stay the same or move up than to move down. The fact that those born in the poorest quintile are almost uniformly likely to end up in any quintile is amazing(ly good).

Equally striking is the extreme downward mobility for Blacks. At all quintiles except the lowest, Blacks are far more likely to stay the same or move down than to move up. One positive might be that half of the poorest (Quintile 1) Blacks move up rather than being "trapped" in poverty. The downward mobility of 3rd and 4th quintile Blacks is very surprising to me.

ThomasH is partially right about the Two Americas perception. The disconnect arises because, if there are two Americas, it's probably divided along racial lines as Scott describes. However, Big Government programs to redistribute from Whites to Blacks are not politically saleable, hence the rhetorical focus on inequality between the top 1% and "everyone else" (which really means the next highest 1%).

Scott Sumner writes:

Airman, Sorry, I added the link.

E. Harding, Thanks, I've probably relied too much on news accounts. I suppose there is also the composition bias to consider, as more students now take the test.

Lorenzo, Thanks for the link.

Richard, Maybe, but it's hard to believe the researchers would not have held age constant. After all, they have the projected quintile at age 40, not the actual quintile. Presumably they look at parents income at 40 as well. Can someone check the original study?

Josiah, Good point, but surely his hypothesis has powerful implications for income inequality.

Gaspard, Good point, In other posts I've argued that income inequality data is virtually worthless, because of all sorts of problems, including the ones you mention. In addition, it combines wage and capital income, which makes zero sense.

Curt, I think you read lots of things into the post that were not there. In other posts I've indicated that income is an almost useless indicator. I only mentioned it here because other people have become interested in income inequality. It's not my interest. But if others are interested, what do they say about this data?

I've travelled around the world and lived in a few other countries. I haven't noticed that Americans are particularly lonely, but I suppose that's possible. There are 315 million Americans, of many different cultures, and I obviously have not met them all.

BC, Interesting points.

JK Brown writes:

I just read Theodore Dalrymple's 'Life at the Bottom'. It is a series of essays on his observations of his patients at his English hospital and also his prison work. Very illuminating, if only because he writes of the British poor whites but describes the same behaviors that, in America, are attributed to poor minorities, especially Blacks. He, also, makes the distinction between the economically poor and what he terms the "underclass".

The underclass are economically poor but also of habit. They most often speak and act as if they have no agency in their lives, disdain education to the point of abusing fellow students/citizens who exhibit intellectual habits, act on impulse with no thought to past or future, etc.

Dalrymple's underclass and the habits picked up by Murry's lower and working classes are very similar. I believe traditional values and the underclass values are the two Americas. In fact, from Dalrymple's book, I see it as two Anglospheres. if not Western civilizations. What's worse is the habit of the underclass are not limited to the economically unfortunate. They are progressing into the upper classes, and when the family money runs out or is withheld, with similar economic consequences. This spread is caused by the presentation/standardization of underclass habits by the media, pop stars and celebrities.

As for why the habits prevail in minority communities in the US and envelop fewer Whites, at least in media reports, well, the habits derived from Liberal intellectuals' ideas and pontifications and since the 1960s, the African-American community and to an extent other minorities have had a well organized and professional apparatus to promote and encourage the habits which sell well in poorer communities. The targeting of minorities may have acted as a buffer for poor Whites, who may have naturally resisted going ghetto. This latter is becoming less of a barrier as time goes on.

Miguel Madeira writes:

"I believe traditional values and the underclass values are the two Americas"

"Traditional" is really the better word? After all, in many issues (age of marriage, number of kids, nuclear vs. extended family, probably more...), it is the underclass values that are the "traditional" values, and the "elite" values are the "modern/liberal" values.

Scott Sumner writes:

JK, Yes, that's a popular theory, and I certainly agree that cultural habits matter. But what about the data in the post---doesn't it undercut that theory?

Miguel, Good point, and add in gay marriage.

Ali Bertarian writes:
Scott Sumner writes:

JK, Yes, that's a popular theory, and I certainly agree that cultural habits matter. But what about the data in the post---doesn't it undercut that theory?

I guess it depends upon what you mean by "undercut." If a large enough percentage of the 23% of the 1st quintile who remain in the 1st quintile have the characteristics to which Dalrymple refers, (90%?) then I would not call that an undercut.

Mercer writes:

I read Coming Apart over a year ago. He wrote little about income in the book. He was focused on illegitimate children and male employment. I recall he said near the end of the book the middle was becoming like the bottom.

His analysis was the trends he decried were from male laziness and the sixties culture. I think the trends are the result of stagnant male wages alongside soaring female wages for low skilled workers.

JK Brown writes:

I don't see the data undercutting the theory. The adoption of the "underclass values" (lack of agency, disdain for education, illegitimacy) is far more prevalent in the Black community from top to bottom via rappers and other "role models. Purely, anecdotal but it appears Black parents in even the upper middle class have to fight to keep their kids from adopting the habits of the underclass.

This culture does impact Whites but mostly at the lower quintiles and they are also likely to seek to differentiate by resisting the values.

Even if the underclass habits are thrown off as the individual ages, many of the decisions such as leaving school, serial illegitimacy limits the ability of the late "bloomer" to undo the damage to their economic prospects.

In today's economy, imagine the economic prospects of individuals steeped in an attitude toward education such as Dalrymple relates here;

In the past their ignorance was purely passive: the mere absence of knowledge. Of late, however, it has taken on a more positive and malign quality: a profound aversion to anything that smacks of intelligence, education, or culture.
J.V. Dubois writes:

This is a good article, but I am afraid that you may have been caught in a trap that you yourself detected in the past but forgot about it. Some of the stuff was covered by other people, like difference betwen black/white and average income quintiles. But some stuff was covered by you in the past.

Just one example: child of two students of medicine vs child of single teenage mother. Both are kids of lowest quintile from income standpoint, but no sane person would say they have equal chances for wealth mobility.

So if you have a population where people study longer (and thus have no official income) then you will expect larger upward mobility since these kids (and their parents) had more years to have children.

There may be other vast differences. If income is counted as household income, then again we can expect differences based on size of the household. Single child will have it easier than middle child with 8 siblings.

In the past you convinced me that inequality data are so mired with errors and conceptual problems so they are borderline useless. So when working with these data one should expect result to be in line with the famous rule: garbage in - garbage out.

Brian writes:


The mobility data you show is extremely misleading and not a valid basis for speculating about relative mobility between blacks and whites. As others have noted, the data separate out distinct populations with very different income quintiles, but group them all by the same quintiles.

To show how misleading this can be, imagine doing the same by combining U.S. data with data from Denmark. Denmark is one of the highest mobility nations in the world. But if grouped in U.S. quintiles, virtually no one would make it to the 5th quintile because the Danish 5th quintile is at a much lower level, in the lower range of the U.S. 4th quintile. This data would make Denmark appear to have little to no top-bottom or bottom-top mobility.

Something similar is happening in this data for blacks versus whites. If blacks were rated against their own quintiles, their mobility would appear to be much higher, perhaps even higher than whites. I'm not saying that's correct either, just that the data as presented is certainly misleading.

Finally, I would note that U.S. mobility data is strongly consistent with the expectations from a meritocracy. That fact is hard to reconcile with any "two Americas" model of inequality.

Granite26 writes:

Sorry I'm a little late to the party.

My take on the book was that upper class individuals had structural, habitual, behavioral benefits that are not being mixed back into the populace at large. Conversely, modern society is making bad, destructive behaviors more acceptable and prevalent in lower society (from a conservative bent, this is making it easier to have welfare babies that survive but don't thrive).

Because modern society is becoming more stratified, especially with men and women from similar school background marrying, the upper class, the leadership class, is isolated from what's actually going on.


Comments for this entry have been closed
Return to top