Millennials as a Social Class?

In my conversation with Bret Weinstein and Michael Lind, I mentioned that I was quite taken with Thibault Muzergues‘s electoral sociology of the West, laid out in his The Great Class Shift. In his schema, Western societies are politically fragmented into four classes. The big winner of neoliberal globalization is Florida’s “Creative” class. This is the professional class that mans all the elite professions, babbles from all the prestige media, and powers all the superstar firms from Google to Goldman. In the US and many other countries, admission into its ranks is governed by prestige schools, as Markovits has argued. The big loser of neoliberal globalization is what Justin Gest has called the New Minority; another name for the white working class which, in America, has been busy killing itself for the past twenty years. Across the West and beyond, the New Minority is behind right-wing populism. The third class is what Muzergues calls the Suburban Middle Class. These are middling small business owners and sales people and suchlike. They go to college, but not selective schools. They live out the American dream in suburbs across the West. This class too is willing to resist the domination of the professional class — it was behind the Tea Party revolt in the United States and the Yellow-Vest revolt in France. Formerly the numerically dominant class, this class plays kingmakers in a world where no single class can dominate on its own. Trump’s coalition, for instance, includes not just the New Minority but a significant chunk of the Suburban Middle Class.

Thibault Muzergues, The Great Class Shift (2020).

What is quirky about Muzergues’s schema is the fourth estate. He calls Millennials a social class, defined by a history of high expectations and bitter disappointment, and the social force behind left-wing populisms across the West. In the United States, it is what powered the Occupy movement, Bernie Sanders’s “working class” revolution, cancel culture, and has been engaged in bitter street warfare with the police. But, as Muzergues documents, it really comes into its own in Southern Europe. One can extend the analysis beyond the West. For instance, the Millennials were the social force behind the early Arab Uprisings, particularly in Egypt, and the Green Movement in Iran.

Muzergues shows how neatly one can make sense of Western politics in the present passage by paying careful attention to the class configurations of political coalitions. I’ll leave you to read The Great Class Shift to see how tidy the schema really is.

In what follows, we restrict attention to US Millennials. We document a number of patterns of interest. First, we document the downward mobility of Millennials relative to earlier generations. We show that, for any given level of educational attainment, Millennials enjoy lower occupational prestige relative to GenXers and especially Boomers. Second, we document that parental socioeconomic status (SES) is more important for Millennials than previous generations. We show that, for any given level of educational attainment, parental SES predicts higher occupational prestige for Millennials relative to earlier generations. In other words, inherited status plays a greater role in the life of Millennials than GenXers and Boomers. Third, we document that Millennials are more likely to lean towards the Democratic Party overall, but this conceals a much stronger class gradient in Millennial political affinities. More educated and higher SES Millennials are much more likely than previous generations to be Democrats; less educated Millennials are more likely to be Republicans. Fourth, Millennials are more likely to be antiracist, as measured by the Symbolic Racism question. Moreover, the class gradient for antiracism is more pronounced than earlier generations.

The upshot of all this is that Millennials are more divided by class than earlier generations. This means that treating the generation as a social class doesn’t actually work, as I argued with Muzergues on Twitter. Bernie’s left-wing populism, BLM and the woke counterrevolution more generally, are projects of professional-class Millennials.

Our econometric strategy is to fit mixed-effects models stratified by generation. I’ll explain as we go along. We begin with occupational prestige. We model occupational prestige as a linear function of educational attainment. We also control for race. In the baseline model, we admit random intercepts by generation. Later, we shall fit proper hierarchical models where we allow the random effects themselves to be linear functions of SES or educational attainment (EA).

Unsurprisingly, we find that the elasticity of occupational prestige against educational attainment is large and robust. More interestingly, the conditional means of the random intercepts by generation reveal the promised pattern. For any given level of educational attainment, Boomers enjoy higher occupational prestige than GenXers, who in turn enjoy higher occupational prestige than Millennials. Note also that Black people punch below their weight — for any given level of educational attainment, their occupational prestige is lower.

Table 1. Baseline Model of Occupational Prestige.
Fixed EffectStd ErrZ
Black-0.1160.010-12.17
Educational Attainment0.4160.004104.89
Random effects:Intercept
Boomers0.095
GenXers0.005
Millennials-0.100
F-stat2873.2
Source: GSS, author’s computations. Response is occupational prestige. Mixed-effects model stratified by generation. All variables except race dummies have been robustly standardized to have mean 0 and variance 1.  
Millennials are downwardly mobile relative to GenXers and especially Boomers.

Having documented the downward mobility of Millennials, we allow the generational intercepts to be themselves functions of SES/EA. This allows us to examine the fortunes of Millennials stratified by class. We find that the gradient of inherited class (operationalized by parental SES/EA) is larger for Millennials than GenXers, and that for GenXers is larger than that for Boomers. What this shows is that parental status is more important for life outcomes for Millennials than it was for GenXers, and especially Boomers.

Table 2. Mixed-Effects Models of Occupational Prestige.
Fixed EffectStd ErrZ
Black-0.0970.011-9.11
Educational Attainment0.4110.00590.88
Random effects:InterceptParental SES
Boomers-0.2330.029
GenXers-0.3340.042
Millennials-0.4370.054
F-stat1728.0
Fixed EffectStd ErrZ
Black-0.1010.011-9.48
Educational Attainment0.4130.00492.88
Random effects:InterceptParental OP
Boomers-0.1000.026
GenXers-0.1990.031
Millennials-0.3010.053
F-stat1561.5
Source: GSS, author’s computations. Response is occupational prestige. Mixed-effects models stratified by generation. All variables except race dummies have been robustly standardized to have mean 0 and variance 1.  

Okay, so we have documented the downward mobility of Millennials and the increasing importance of inherited class for life outcomes over three generations. Now I want to document the partisan affinities of the generations.

The baseline model for partisan affinity (higher means Democrat) is a washout. Educational attainment is a stronger predictor of affinity for the Democrats than SES. But the generational random intercepts all vanish.

Table 3. Baseline Models of Partisan Affinity.
Fixed EffectStd ErrZ
Black0.3970.01525.95
Educational Attainment0.0290.0064.52
Random effects:Intercept
Boomers0.000
GenXers0.000
Millennials0.000
F-stat213.5
Fixed EffectStd ErrZ
Black0.4020.01625.42
SES0.0050.0100.52
Random effects:Intercept
Boomers0.001
GenXers0.000
Millennials0.000
F-stat204.3
Source: GSS, author’s computations. Response is partisan affinity (higher is pro-Democrat). Mixed-effects models stratified by generation. All variables except race dummies have been robustly standardized to have mean 0 and variance 1.  

In contrast, once we allow the random intercepts to vary by SES/EA, we pick a clear generational pattern. Note first that higher SES predicts affinity for the GOP, while higher EA predicts affinity for the Democrats. Black always predicts higher affinity for Dems. What is interesting is the pattern of the slopes of the random effects. For Boomers, higher SES predicts affinity for the GOP and less so for GenXers. But higher SES (marginally) predicts affinity for the Dems among Millennials! The pattern is much sharper when we admit fixed effects for SES and race, and model the random effects as a function of EA. Educational attainment is a much stronger predictor of affinity for Dems among Millennials relative to GenXers, and of GenXers relative to Boomers. Educational attainment has a stronger class-partisan signal than SES.

Table 4. Mixed-Effects Models of Partisan Affinity.
Fixed EffectStd ErrZ
Black0.4030.01625.55
Educational Attainment0.0410.0085.07
Random effects:InterceptSES
Boomers0.011-0.046
GenXers-0.005-0.035
Millennials-0.0060.003
F-stat125.6
Fixed EffectStd ErrZ
Black0.4050.01625.64
SES-0.0280.012-2.37
Random effects:InterceptEducational Attainment
Boomers0.0210.027
GenXers0.0000.031
Millennials-0.0220.077
F-stat117.1
Source: GSS, author’s computations. Response is partisan affinity (higher is pro-Democrat). Mixed-effects models stratified by generation. All variables except race dummies have been robustly standardized to have mean 0 and variance 1.  

The upshot is that the political affinities of Millennials are more strongly conditioned by class than older generations. Professional class Millennials are strong Democrats; working class Millennials are strong Republicans. This is the fundamental issue with positing this generation as a social class.

The class stratification of Millennials goes deep. Above all, professional class Millennials are much more likely to be woke antiracists than working class Millennials, as we shall now document.

Political scientists measure “Symbolic Racism” as degree of agreement with the following statement.

Irish, Italians, Jewish and many other minorities overcame prejudice and worked their way up. Blacks should do the same without special favors.

GSS “Symbolic Racism” question.

We have argued that this is essentially a gotcha question that mines working class ideology—Michele Lamont’s script of the ‘disciplined self’. And we have documented before than Millennials score much lower on this scale. In that exercise, we did not control for race or status. Here we replicate the result with both controls. In our baseline models with random intercepts, we find that Millennials are dramatically more antiracist than GenXers and Boomers.

Table 5. Baseline Models of Symbolic Racism.
Fixed EffectStd ErrZ
Black-0.3470.026-13.34
Educational Attainment-0.1710.011-15.45
Random effects:Intercept
Boomers0.046
GenXers0.046
Millennials-0.092
F-stat100.5
Fixed EffectStd ErrZ
Black-0.3570.027-13.24
SES-0.2010.017-11.83
Random effects:Intercept
Boomers0.061
GenXers0.044
Millennials-0.105
F-stat72.1
Source: GSS, author’s computations. Response is degree of agreement with Symbolic Racism statement. Mixed-effects models stratified by generation. All variables except race dummies have been robustly standardized to have mean 0 and variance 1.  

Whether we control for SES or EA, Millennials stand out. This gives us a robust clue as to the social basis of the woke counterrevolution.

Modeling the random effects as functions of SES and EA reveals the familiar pattern. The intercept for Millennials is always lower. However, SES gives us no handle on the question; educational attainment does. The slope for SES is virtually the same for the three generations, around -0.08. The slope for EA is around -0.12 for Boomers and GenXers, while it is -0.18 for Millennials.

Table 6. Mixed-Effects Models of Symbolic Racism.
Fixed EffectStd ErrZ
Black-0.3610.027-13.54
Educational Attainment-0.1440.013-10.67
Random effects:InterceptSES
Boomers-0.071-0.078
GenXers-0.073-0.080
Millennials-0.216-0.078
F-stat49.8
Fixed EffectStd ErrZ
Black-0.3640.027-13.67
SES-0.0890.020-4.40
Random effects:InterceptEducational Attainment
Boomers-0.204-0.123
GenXers-0.203-0.121
Millennials-0.309-0.184
F-stat39.7
Source: GSS, author’s computations. Response is degree of agreement with Symbolic Racism statement. Mixed-effects models stratified by generation. All variables except race dummies have been robustly standardized to have mean 0 and variance 1.  

The interpretation of these patterns is that Millennials are more antiracist than older generations but they are also more divided by class — as captured by educational attainment, not SES. In other words, the key divide among Millennials is educational, not money. Those with more education under their belt are more read into woke ideology than those with less. That being said, the dominant pattern is that the whole generation is more antiracist than GenXers and Boomers.

We have shown that Millennials are indeed more downwardly mobile than older generations. We have also documented that inherited class, in the sense of parental socioeconomic status, is of greater importance to the life outcomes of Millennials than it was for GenXers and especially Boomers. Ie, class matters more for Millennials than older generations. Moreover, we have shown that class partisan polarization is greater among Millennials than older generations; especially along the educational attainment ladder. Finally, we have shown that Millennials are dramatically more likely to be antiracist than GenXers and Boomers. And on this question too we found that the class gradient is steeper for Millennials than older generations.

These patterns help make sense of the present conjuncture. Yes, along with the New Minority, Millennials are losers of late neoliberalism. But the generation is itself more divided than previous generations. It is professional-class Millennials who were behind Occupy and Bernie, going some way to explain why the revolution was aborted. They are also behind BLM and the woke wing of the Democratic Party. But working-class Millennials, who sit at the unhappy intersection of the generation forsaken by fortune and the New Minority, are opposed to the woke counterrevolution — talk about intersectionality. Both sides in the street warfare we have seen this summer belong to this generation.

In future work, I’ll try to identify the diagnostic characters of the Suburban Middle Class, Muzergues’s kingmakers.


Postscript. I am satisfied with the models for occupational prestige. But I am unsatisfied with the models for partisan affinity and symbolic racism. Specifically, I don’t think we should admit fixed-effects for SES/EA in Tables 3-6. Instead of reporting the full results, I am going to share the main results in the form of four graphs.

Controlling for race and sex, we stratify by generation, and allow our random effects to vary with SES. This yields a much stronger pattern. Whereas Boomers with higher socioeconomic status lean a bit Republican and those with lower status lean Democrat, the gradient vanishes for GenXers. For Millennials on the other hand, the class gradient is very, very large. Millennials lower down the socioeconomic ladder are much more likely to be Republican; those above, to be Democrat.

The pattern with educational attainment is similar to what we had before. This is because we had fixed effects for SES, which, as we have seen above, is only important for Millennials. The big update is that the class-partisan polarization among Millennials relative to Boomers and GenXers is much more pronounced that previously reported.

Controlling again for race and sex, the gradient of Symbolic Racism against SES is less pronounced for Millennials; while the intercept is still large.

Again, for educational attainment, we recover our earlier result because SES contains a weaker class signal than educational attainment.

Again, the big update is that the class-partisan polarization among Millennials relative to Boomers and GenXers is much more pronounced that previously reported.

2 thoughts on “Millennials as a Social Class?

  1. I wouldn’t say Millenials have an affinity for democrats so much as we absolutely hate both worthless parties (the democrats marginally less so), and actually the entire garbage country too, and we think democracy is a lie, which it undeniably is.

  2. Thanks. Some of the charts above answer questions I had about how the political-vs-education/$$ relationships are (or are not) modulated by {generation or age}.

    A not-entirely-serious aside: Many analyses of ‘generation’ in the US highlight the divide between boomers and millennials. I suppose because of the parent-child dynamic makes it a natural contrast. But what about Gen X? Came of age when the US economic-political system was at its modern high point in the 80’s-90’s, and due to become the center of mass of likely voters in the next decade. There’s got to be something there besides just the bland middle. As a member of this age group (tho not originally from the US) I find it Just a little disappointing. Is it more a function of age — too old to get excited by radical reform, too young to be scared of it… perhaps the trends less a matter of historical movement, more just each generation going through the same shifts over time. For example those with $$ in the bank shifting more conservative as the years go by, a constant pattern? Or are Gen-X’ers just boring?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s