Introduction
The rise of populism and decline of mainstream parties across Western democracies has renewed scholarly interest in the factors driving political dissatisfaction and support for radical parties. A burgeoning literature studies the structural roots of these developments (Anelli, Colantone, and Stanig Reference Anelli, Colantone and Stanig2019; Ballard-Rosa, Jensen, and Scheve Reference Ballard-Rosa, Jensen and Scheve2022; Broz, Frieden, and Weymouth Reference Broz, Frieden and Weymouth2021; Burgoon et al. Reference Burgoon, van Noort, Rooduijn and Underhill2019; Colantone and Stanig Reference Colantone and Stanig2018; Inglehart and Norris Reference Inglehart and Norris2017; Kurer Reference Kurer2020; Margalit Reference Margalit2019; Rodrik Reference Rodrik2018). Recent research in political science emphasizes the role of status loss, particularly so among white men without college degrees, often viewed as “left behind” by an emerging global knowledge economy and increasingly disillusioned by the mainstream parties that promote it. Radical parties have attracted these voters by attributing their status loss to the advancement of other groups such as immigrants, minorities, and women, or to the mismanagement of globalization by an unchecked, political-financial elite (Gest, Reny, and Mayer Reference Gest, Reny and Mayer2017; Gidron and Hall Reference Gidron and Hall2017; Steenvoorden and Harteveld Reference Steenvoorden and Harteveld2018).
We build on these explanations by recasting status loss in terms of occupational change and intergenerational mobility. Postwar economic expansion spawned an era of unprecedented upward mobility. As Lipset (Reference Lipset1959, 11) put it, “in every industrial country, a large proportion of the population have had to find occupations considerably different from those of their parents,” a phenomenon rooted in “a period of prosperity that followed on the heels of great dislocations” (33). As a result, occupational reorientation most often meant occupational upgrading. Leading accounts at the time presupposed that economies would expand indefinitely, sustaining upward mobility for future generations (e.g., Treiman Reference Treiman1970).
However, the rate of upward mobility began to decrease toward the end of the twentieth century, a trend that continues today. Scholars posit that this is largely due to postindustrial occupational change. Deindustrialization, job polarization, and the recent slowdown in the growth of relative demand for high-skilled workers (see, e.g., Autor and Dorn Reference Autor and Dorn2013; Autor, Goldin, and Katz Reference Autor, Goldin and Katz2020; Goos, Manning, and Salomons Reference Goos, Manning and Salomons2009; Oesch Reference Oesch2013) have impaired the social escalator, eliminating decent, middle-skill jobs while at the same time gradually reducing the value of a college degree. Recent empirical studies report a striking contrast between the fortunes of those born in the 50s and 60s and those born in the 80s and early 90s, a pattern that holds both in the United States (e.g., Chetty et al. Reference Chetty, Grusky, Hell, Hendren, Manduca and Narang2017; Hout Reference Hout2018) and across many advanced economies in Western Europe (e.g., Berman Reference BermanForthcoming; Breen and Müller Reference Breen and Müller2020; Bukodi, Paskov, and Nolan Reference Bukodi, Paskov and Nolan2020) including Germany (Hertel Reference Hertel2017).
We contend that a rise in disappointed expectations has followed this large-scale structural shift and that these disappointments shape voter behavior. The upward mobility of postwar cohorts gave way to a societal and political consensus of progress. Most people could reasonably expect to at least meet, if not exceed, status standards set by their parents. We expect that the growing share of citizens who fall short of these expectations attribute some of the blame to mainstream parties.
Relying on German household panel data and machine-learning methods, we propose an innovative empirical approach to overcome a major obstacle to studying status discordance: measuring expectations. We start from the assumption that status standards set by parents have a lasting influence on how individuals understand their own status achievements. We create out-of-sample predictions for respondents’ expected occupational status based on key socioeconomic characteristics of their fathersFootnote 1 (education, job, occupational status, migration background, etc.) and various pretreatment characteristics of the respondents themselves (gender, age, citizenship, region in childhood, school grades, etc.). We argue that such a prediction provides a reasonable proxy for the status position respondents expect to achieve in adulthood. We then calculate the difference between expected status (prediction) and realized outcome (observed) in adulthood. This difference is what we term status discordance. It captures respondents’ status achievements relative to salient socioeconomic reference points in childhood.
Our analysis provides strong and consistent evidence that negative status discordance is systematically related to political disillusionment manifested by higher rates of abstention, lower identification with mainstream parties, and higher support for radical parties. At the same time, we also emphasize variation in reactions to negative status discordance among men and women and among voters of different socioeconomic backgrounds, examining in particular the structural foundations of supporting either radical left or radical right alternatives to mainstream parties.
Our findings shed light on the forces driving recent electoral outcomes in postindustrial societies. We contribute to research examining these outcomes by drawing attention to the role of expectations in shaping generational voting patterns. Due to occupational change and fading prospects for mobility, growing shares of voters are having to adjust expectations that have long been rooted in postwar projections of progress. Our findings suggest that, although for now still outnumbered by the upwardly mobile, these disappointed voters have and will continue to change the electoral landscape of advanced democracies.
The Politics of Expectations
Interest in the factors driving support for radical parties expanded substantially following a string of major radical-right victories in the first decade of the twenty-first century. Building on earlier inquiries into why and when radical parties find electoral success, many recent contributions point forcefully toward the importance of postindustrial occupational change. These explanations were particularly popular in the aftermath of Brexit and the triumph of Donald Trump, events viewed as being propelled by support from voters who have been hit especially hard by deindustrialization, automation, and globalization.
Evidence from an influential body of work indeed reveals that certain voters suffered income losses and increased unemployment risk following the shift to a postindustrial economy, which in many cases was overlooked by mainstream political elites (Anelli, Colantone, and Stanig Reference Anelli, Colantone and Stanig2019; Ballard-Rosa, Jensen, and Scheve Reference Ballard-Rosa, Jensen and Scheve2022; Broz, Frieden, and Weymouth Reference Broz, Frieden and Weymouth2021; Burgoon et al. Reference Burgoon, van Noort, Rooduijn and Underhill2019; Colantone and Stanig Reference Colantone and Stanig2018; Inglehart and Norris Reference Inglehart and Norris2017; Kurer Reference Kurer2020; Owen and Johnston Reference Owen and Johnston2017; Rodrik Reference Rodrik2018). Recent contributions to this literature have offered different ways of capturing status loss. Gidron and Hall (Reference Gidron and Hall2017) show how the disappearance of manufacturing jobs and the concentration of knowledge economies in urban centers encourages support for the radical right. The authors also point to the strongly gendered nature of this relationship, which applies in particular to white men without college degrees. Evidence from interviews similarly reveals that white men often view a loss in status relative to gains in status among women or then minorities, a view influenced by the purported replacement of low-skilled native workers with immigrant workers (Gest Reference Gest2016; Hochschild Reference Hochschild2016).
Although the literature on status loss provides an important point of departure, we argue that existing work has neglected long-term changes in economic opportunity in general and intergenerational occupational mobility in particular. Similar to existing research, we seek to capture status loss but propose to do so with a focus on disappointed expectations resulting from declining occupational mobility across generations. We argue that occupational standards set by parents are a particularly salient reference point for understanding status (Cohen Reference Cohen1987; Hill and Duncan Reference Hill and Duncan1987). As the first example of what work is and why people work, parental occupations orient individual understandings of status. In some cases, parental occupations serve as a standard that children can reasonably expect to meet and in other cases as a standard that children should expect to surpass. Parental occupations establish enduring associations between work, income, and standard of living. Children use their parents as a reference point when forming ideas about the financial and social prestige attached to certain occupations, ideas that later play an important role in career choices, earnings, and evaluations about their own achievements. These understandings may change in response to new experiences or expanded awareness about meritocracy or equality, but standards set by parents nonetheless carry lifelong weight, anchoring individual status expectations (Checchi Reference Checchi2006; Dustmann Reference Dustmann2004; Johnson Reference Johnson2002).
Evidence from a longstanding literature on relative deprivation shows that discrepancies between expectations and reality have political consequences (e.g., Burgoon et al. Reference Burgoon, van Noort, Rooduijn and Underhill2019; Geschwender Reference Geschwender1964; Gurr Reference Gurr1970; Kurer et al. Reference Kurer, Häusermann, Wüest and Enggist2019; Mitrea, Mühlböck, and Warmuth Reference Mitrea, Mühlböck and Warmuth2021; Paskov, Präg, and Richards Reference Paskov, Präg and Richards2021). Building on these insights, we propose that the discrepancies resulting from declining occupational mobility affect political behavior because those experiencing negative status discordance feel they have been denied the opportunity to secure the status they had come to expect for themselves based on standards set in childhood—expectations also encouraged by the postwar political consensus. Mainstream political parties have seemed to assume that, in the long run, economic growth would sustain upward mobility to some degree regardless of political interventions.
While such assumptions have suited many well, they have served to alienate the increasing share of downwardly mobile voters. Status discordance captures a process by which individuals become disillusioned with the prospects for social and economic mobility to such an extent that they blame mainstream parties for the lack of opportunities. Thus, it is not only that status-discordant voters are dissatisfied with mainstream incumbents but also that they consider these parties as the source of their current hardships. When disappointed voters lose faith in both standard political responses, whether the government-based policies of the center-left or the market-based solutions of the center-right, it is the political establishment as a whole that has failed (Guiso et al. Reference Guiso, Herrera, Morelli and Sonno2020). The experience of negative status discordance thus heightens the likelihood that these voters envision more radical change than that which is on the policy agendas of mainstream parties.
We expect that status-discordant votersFootnote 2 are more likely to feel politically alienated and dissatisfied with mainstream politics, which may manifest in the form of political abstention and lower levels of identification with political parties more generally. At the same time, the growing literature on status loss suggests that other status-discordant voters will choose to voice their discontent in the political arena (Gidron and Hall Reference Gidron and Hall2017; Kurer Reference Kurer2020), especially when there are compelling antiestablishment party options (Guiso et al. Reference Guiso, Herrera, Morelli and Sonno2020). We thus also expect that negative status discordance increases the likelihood of supporting parties outside the political mainstream.
H1: Status-discordant voters are more likely to feel politically alienated.
H2: Status-discordant voters are less likely to vote for mainstream parties.
H3: Status-discordant voters are more likely to vote for radical parties.
Given that the effects of occupational change vary across different groups in society (Chetty et al. Reference Chetty, Hendren, Jones and Porter2020; Kao and Tienda Reference Kao and Tienda1998; Lareau Reference Lareau2011; Polavieja and Platt Reference Polavieja and Platt2014), we explore the possibility of divergent political reactions to the experience of status discordance.
Gender
As noted above, rates of downward mobility are most pronounced among men without college degrees. While prospects for upward mobility gradually began fading for men, women instead made unprecedented socioeconomic gains. The decades since World War II have seen a rise in women’s college attendance at similar if not higher rates than men, which has helped to boost their prospects in white-collar and professional occupations (Breen and Müller Reference Breen and Müller2020; DiPrete and Buchmann Reference DiPrete and Buchmann2013). Although women continue to face numerous barriers in the labor market (Goldin Reference Goldin1990; Reference Goldin2021; Sauer et al. Reference Sauer, Valet, Shams and Tomaskovic-Devey2021), women’s occupational experiences in recent decades relative to their historical exclusion from the labor market, we expect, makes them less vulnerable to the political effects of status loss (Gingrich and Kuo Reference Gingrich, Kuo, Kemmerling, van Kersbergen and Marx2021).
There are a number of factors that motivate this reasoning. To begin with, women are less prone to the effects of negative status discordance because, compared with men, they are less likely to experience it in the first place (Hertel Reference Hertel2017). We empirically confirm this below (see Figure 2 and Figure A8 in the Supplementary Materials). Furthermore, even among women who do experience status discordance, such an experience may feel less like status loss. There are two reasons for this. First, because of the considerable socioeconomic gains that women have made and broader movements in favor of gender equality, status discordance may not register as such because of the positive societal developments occurring simultaneously. The empowering and emancipatory experience of increasingly equal access to paid employment (see, e.g., Orloff Reference Orloff1993) may moderate political reactions among the individual women who do experience status loss as conceptualized in this article. Second, our measure imperfectly captures women’s status expectations, which, in comparison with men’s, may be influenced to a lesser degree by the occupation of their father and to a greater degree by that of their mother. Thus, we expect that women who experience status discordance as we have measured it will not undergo the same intense experience of status loss because their father’s occupation might not represent their primary or their most salient point of reference.
H4: The relationship between status discordance and political alienation is less pronounced among women than among men.
Socioeconomic Background
Second, the effects of occupational change depend heavily on educational background, which may result in distinct voting behavior across different socioeconomic groups. For the most part, occupational change has favored individuals with professional, college-educated backgrounds, leaving few prospects for those from blue-collar families and those without college degrees. At the same time, even those with college degrees face fierce competition and rising income inequality in the knowledge economy.
We hypothesize that status-discordant voters with weaker socioeconomic backgrounds are more likely to vote for radical right parties. These parties tend to hearken back to earlier times marked by economies dependent on routine work, clearly demarcated social and cultural identities, and traditional workplace hierarchies—times when manual workers earned a respectable standard of living and even had some prospects for career mobility (Gidron and Hall Reference Gidron and Hall2017; Steenvoorden and Harteveld Reference Steenvoorden and Harteveld2018). For the parents of status-discordant voters without college degrees, postwar labor markets provided ample economic opportunity and security. This is no longer the case, and radical-right parties’ promises to undo recent societal change is thus an effective way of attracting these voters. Of course, the anti-immigrant and EU-skeptic views of radical right parties are also important, but we consider such positions as part of their broader platform calling for a return to earlier—preimmigration, pre-EU—times.
In contrast, status-discordant voters with more privileged socioeconomic backgrounds are more likely to vote for radical left parties. Although these voters are overall less likely to experience downward mobility, they may well experience status loss relative to their parents. Consider, for example, an individual who grew up in a financially stable home headed by a lawyer or doctor and who seamlessly attended and graduated from college. Even for them, occupational trajectories after college have grown more challenging due to rising competition following the “massification” of higher education (Ansell and Gingrich Reference Ansell, Gingrich, Manow, Palier and Schwander2018), the declining college wage premium (Autor, Goldin, and Katz Reference Autor, Goldin and Katz2020), and the fact that some work in public or arts sectors may be intellectually or socially rewarding but poorly remunerated and with limited prospects for career mobility (Ansell and Gingrich Reference Ansell, Gingrich, Rosenbluth and Weir2021). Radical-left parties appeal to these voters not by calling for a return to earlier times but instead by calling for a radically different economic system to combat growing inequality and to economically reward those with goals other than profit growth, be they environmental, sociocultural, or humanitarian (Rooduijn et al. Reference Rooduijn, Burgoon, van Elsas and van de Werfhorst2017).
In other words, socioeconomic background distinguishes status-discordant voters by their reference points—by the contexts in which their status expectations were formed. Whether the economic and political context that supported the occupational trajectory of their parents would present status-discordant voters with the kinds of opportunities that they feel are lacking today is a primary factor influencing whether these voters look to the left or right for answers.
H5: The relationship between status discordance and support for radical-right parties is more pronounced among respondents from weaker socioeconomic backgrounds.
Declining Occupational Mobility and Status Discordance
Why do disappointed status expectations result in such remarkable political disruption now but not so much before? While such disappointments have always been a feature of any society, they were less frequent and thus less politically relevant. In the prosperous and highly mobile postwar period, most people experienced status upgrades relative to their parents. In contrast, the trend of declining upward mobility in postindustrial democracies suggests that a growing number of people are falling short of expectations based on parental standards. Declining mobility, then, matters for politics not only because it is becoming more pervasive but also because expectations are still undergoing adjustment to a postindustrial context in the shadow of postwar prosperity.
Figure 1 provides a graphical contextualization of our argument. The x-axis shows an individual’s intergenerational reference point—that is, an individual’s expected occupational status based on parental background. The y-axis shows the actual occupational outcome realized by that same individual. If the realized status perfectly coincides with the expected status, this individual observation would lie on the 45-degree line. If the realized occupational status exceeds (falls below) the intergenerational reference point, we would instead observe positive (negative) status discordance.
Figure 1. Status Discordance When Absolute Mobility Declines (Simulated Data)
Now, any given society consists of a large number of such comparisons of individual expectations and outcomes, resulting in a distribution of observations as the one simulated in Figure 1. If all observations aligned on the 45-degree line, this would be a society with zero mobility; parental background would be a perfect predictor of individual status. In reality, there is a fair amount of mobility and the pattern in most postindustrial societies more closely resembles the depicted distribution following the dashed line.Footnote 3
The indicated downward shift of the black dashed line illustrates the background condition for our argument—namely, the decline in occupational mobility. Almost all those born in the postwar era managed to exceed expectations based on the economic circumstances of their parents, which in this figure would result in a population consisting almost exclusively of individuals experiencing positive status discordance, thus lying above the 45-degree line. However, due to declining occupational mobility, the respective shares of positive and negative status discordance are becoming increasingly similar. That is, the downward shift of the dashed line directly increases the number of people facing negative status discordance. Therefore, the pool of disappointed citizens is steadily growing, which we argue helps explain why we see political disruption at this point in time but not before.
Finally, we want to again emphasize that, as the figure illustrates, status discordance can occur across the occupational hierarchy, at all status positions. Our focus is strictly on any deviation from an expectation based on parental background.Footnote 4 Appreciating the reach of occupational change is important to understanding its effects. How disappointed expectations shape voting behavior—that is, how disappointed voters ultimately express their discontent with mainstream parties—depends heavily on the socioeconomic context in which status expectations were formed (see H5 above).
Status Discordance: Concept and Measure
For the purpose of our analysis, we need an empirical measure that captures deviations in realized socioeconomic status from expected socioeconomic status based on parental background. As we have argued, this intergenerational reference point is critical because it shapes individual status expectations. Two individuals with the same absolute socioeconomic status are likely to think very differently about their achievements depending on the socioeconomic context in which they were raised and the status expectations this generated.
The empirical challenge thus lies in acquiring an estimate of expected status outcomes as proxied by parental background. We propose an innovative approach to arrive at an empirical estimate of such status expectations. We draw on rich household panel data from the German Socio-Economic Panel (SOEP 2019) and recent advances in machine learning to create out-of-sample predictions of occupational status given parental background. The SOEP consists of a large number of sub-datasets, which can be combined by a unique personal identifier. Of particular relevance for our endeavour is the detailed biography and life history data on respondents themselves and a questionnaire dealing specifically with intergenerational aspects and thus covering a large range of variables on parents’ sociodemographics and life circumstances.Footnote 5
More specifically, we use earlier waves of SOEP to train a random forest algorithm (Breiman Reference Breiman2001) to predict an individual’s occupational status on the basis of an array of variables capturing their father’s socioeconomic status as well as various pretreatment childhood characteristics of the respondent themselves. Tree-based models like random forests are a valuable tool for prediction that make few assumptions on the data-generating process and very flexibly handle nonlinearities, interactions between covariates, or the inclusion of irrelevant covariates (Montgomery and Olivella Reference Montgomery and Olivella2018). We have evaluated the performance of different learning methods (regression trees, bagging, random forest) and different model parameters on our dataset and decided based on the mean squared error (MSE)—the most commonly used measure for assessing model accuracy (Gareth et al. Reference Gareth, Witten, Hastie and Tibshirani2013)—to proceed with a random forest algorithm. We detail the model selection and optimization process in Appendix A1 of the Supplementary Information.Footnote 6
After training the algorithm, we create individual predictions of occupational status for all respondents in the latest available SOEP wave from 2018. This latest release contains information on respondents’ voting behavior in the 2017 German federal election, which allows us to study the relationship between mobility perceptions and political behavior.Footnote 7 The difference between the realized occupational status and status expectations proxied by predictions based on parental background is our measure of intergenerational status discordance (ISD):
$$ {\mathrm{ISD}}_i\hskip1.5pt =\hskip1.5pt {\hat{\mathrm{y}}}_i\hbox{-} {\mathrm{y}}_i, $$
where y i is the realized individual socioeconomic status, captured by the International Socio-Economic Index (ISEI), a widely used measure of occupational status. In contrast to categorical indices of occupations, such as social class schemes, the ISEI is a continuous, unidimensional measure, created by optimally scaling occupation as an intervening variable between education and income (Ganzeboom, De Graaf, and Treiman Reference Ganzeboom, De Graaf and Treiman1992). More intuitively, the ISEI can be understood as a weighted sum of mean education and mean income for each occupational group, adjusted for age to account for life-cycle effects. A detailed comparison of the available indices of class, occupational prestige, and occupational status shows that the ISEI accounts well for what drives the process of intergenerational occupational mobility (Ganzeboom, De Graaf, and Treiman Reference Ganzeboom, De Graaf and Treiman1992, 7).
ISEI is also well-suited to capturing both the social and economic dimensions of status. In the context of occupational change, status positions encompass more than just material factors because they reflect the social boundaries made salient by political processes that protect some workers but leave others vulnerable to economic precarity and social isolation. Given the growing socioeconomic disparities between individuals with a college education and those without, ISEI provides a nuanced indicator of how education, occupation, and income have become tightly bundled and together greatly influence social positions, life chances, and access to scarce and valuable socioeconomic resources.
We estimate status expectations,
$ {\hat{\mathrm{y}}}_i=f\left({\boldsymbol{X}}_j,{\boldsymbol{X}}_i\right) $
, based on a set of variables of parental background,
X
j
, and various potentially relevant characteristics of the respondent,
X
i. In the full model,
X
j
includes father’s occupational status, father’s major occupational group, father’s education level, and father’s year of birth, and
X
i
is a vector of variables including basic demographic and childhood circumstances of the respondent themselves, all of which can be considered pretreatment covariates (age, gender, migration background, citizenship, region of school degree, urban/rural childhood location, school grades at age 15). We present variable importance plots of the underlying random forest model in the Supplementary Materials to this article.
In addition, we calculate a reduced version of the algorithm that drops variables with a relatively high prevalence of missingness (region of school degree, school grades). In the Supplementary Materials, we show the respective variance importance plot. Although these variables improve the predictive analysis by increasing the explained variance in occupational status, their inclusion results in a lower number of observations in the following models of the main analysis. We provide a full set of results based on this reduced predictive model in the Supplementary Materials to demonstrate that our results do not hinge on specific choices related to this trade-off.
Validation and Descriptives
Operationally speaking, our proposed measure is distinct from subjective social status, which is the guiding concept for some of the relevant literature discussed above. Common attitudinal measures for subjective status suggest that it captures where individuals locate themselves in the social hierarchy based upon the way others in society treat and value them. In contrast, our concept of status discordance captures how individuals view their achievements relative to expectations based on parental background. This performance is not relative to parental performance per se but rather how individuals have fared relative to others with similar starting points in terms of parental background. So, rather than capturing downward mobility in absolute terms, negative status discordance reflects a sense among certain individuals that they were denied the opportunity to make the status achievements that they observe among socioeconomic peers from childhood. In this way, subjective status and status discordance serve as indirect, but distinct, indicators of downward mobility.
Despite these operational differences, however, it is likely that many of those who report a low subjective social status because they feel less valued in society are likely to have also experienced some degree of status discordance. While we are not able to directly evaluate this correlation due to the general scarcity of attitudinal items in the SOEP and the absence of a question on subjective status, we still attempt to confirm some of the intuitive properties of our proposed measure before proceeding with the main analysis. First, we show in the Supplementary Materials that negative status discordance results in systematically lower levels of various measures of life satisfaction. The relationship is fairly strong, as the coefficient’s magnitude is comparable to that for the influence of income on satisfaction. While life satisfaction is an admittedly rough proxy for our core concept, it is reassuring that status discordance to some extent shifts subjective assessments in the expected direction. Second, we leverage a special subset of the SOEP to validate our measure against a more direct proxy. Starting in 2000, the SOEP began asking first-time respondents who entered the panel at the age of 17 about their career aspirations. The SOEP enumerators then coded their open-ended responses into ISEI. This provides an opportunity for us to calculate a similar measure of status discordance based on a comparison of realized occupational status in 2018 against verbalized status expectations at the age of 17 rather than predicted occupational status. The two measures correlate strongly and robustly (see details in the Supplementary Materials). Of course, this small and disproportionately young subsample does not lend itself for populationwide inference but still serves as a valuable opportunity to validate our measure.
Next, we present some descriptive information on the distribution of status discordance and discuss variation across various dimensions. Panel (a) in Figure 2 displays the overall distribution of the measure resulting from the procedure described in the previous section. As expected, most respondents achieve an occupational status that corresponds closely to predictions based on parental background, resulting in most ISD values clustering around zero. Deviations from the predictions occur to a similar extent in both directions, meaning that we have a comparable number of negative status discordance (downward mobility) and positive status discordance (upward mobility) in the total sample.Footnote 8 Panel (b) confirms our expectation of a gendered pattern: women in recent decades have caught up with men when it comes to occupational trajectories, resulting in more frequent positive status discordance. However, the tails of the distribution add interesting nuance in that exceptionally successful careers relative to socioeconomic origin are still more common among men.
Figure 2. Distribution of Intergenerational Status Discordance
Panel (c) shows the expected pattern with respect to education: negative status discordance is much more common among those without a college degree. As mentioned previously, although highly educated voters do experience downward mobility, completing a college degree strongly increases the chances of a higher socioeconomic status and thus the probability of achieving a position in society that matches or exceeds expectations based on parental background. Panel (d) shows variation among respondents based on migrant background and shows that, on average, respondents with German backgrounds enjoy advantages that translate into more successful occupational careers relative to socioeconomic peers with migrant backgrounds.Footnote 9
The two lower panels reveal important variation with respect to place of residence and moving patterns since childhood. Panel (e) shows that, even today, respondents who grew up in East Germany, on average, face less favorable mobility prospects compared with those who lived in the West in 1989. Both the roots and persistence of this pattern have been described impressively in a recent sociological account on German reunification from an East German perspective (Mau Reference Mau2019). Finally, panel (f) demonstrates that spatial mobility serves as a source of occupational mobility. Negative status discordance is more common among individuals who never left the place where they grew up.
In the Supplementary Materials to this article, we attempt to capture the influence of social networks and collective organization on occupational mobility by looking at distributions of status discordance by union membership and church attendance. These additional figures indeed provide some tentative evidence for the presence of network effects in that individuals active in either organization are characterized, on average, by slightly more positive mobility trajectories. All in all, these distributions align well with our intuition and demonstrate the face validity of our original measure.
Main Findings
Status Discordance and Political Alienation
As a first step in our analysis of the political effects of status discordance, we show simple bivariate correlations between a ranked percentile version of ISD (see Footnote footnote 10 below) and different indicators of political alienation from mainstream parties. Figure 3 shows the relationship between status discordance and (a) abstention and (b) voting for a radical party in the 2017 German federal election. Here, radical parties include those from both the left (“Die Linke”) and the right (“Alternative für Deutschland”).
Figure 3. Intergenerational Status Discordance (Percentile) and Voting Behavior
The figure reveals strong positive correlations for both indicators of political alienation. Those in the lowest percentiles of the ranked ISD variable—that is, individuals whose socioeconomic status is higher than expected based on parental background, have a close-to-zero probability of abstention and show low levels of support for radical nonmainstream parties. In contrast, the level of support strongly and steadily increases with percentile ranks including higher numbers of citizens who have experienced negative status discordance.
We consider this clear and consistent pattern, in line with our expectations, as the first piece of evidence in support of our core hypothesis that the disappointment of status expectations increases the likelihood of dissatisfaction with mainstream politics. As a next step, we examine the relationship between status discordance and political alienation in a more robust multivariate setting running linear probability models. The main explanatory variable is the raw ISD value.Footnote 10 We include sociodemographic (age, gender, migration background) and socioeconomic (education, income, employment status) characteristics of individuals in the models. In addition, following the insights from Figure 2, we include variables capturing the East/West location of a respondent in 1989 and the geographic mobility patterns since childhood as relevant predictors of the explanatory variable and likely determinant of the dependent variable. Finally, Bundesland-fixed effects take care of regional variation in the contemporary electoral landscape of Germany.
Following our theoretical expectations, the first models presented in Table 1 focus on political alienation measured by abstention (column 1), general political disenchantment, proxied by whether respondents feel close to any party in the party system (column 2), and support for radical parties in the federal election 2017 (column 3). We supplement the first set of models with an analysis of support for mainstream parties, which should yield the empirical mirror image of the results on support for radical parties. We show results from a broad definition that includes the by now fairly established Green party (column 4). The coefficients for ISD thus report the change in the probability to support a given party family (or the probability to identify with a party) for a one-unit increase in negative status discordance net of current socioeconomic status. That is, here we are not interested in whether more educated or higher-income respondents are more likely to support a given party but rather in the role of status discordance compared with otherwise similar respondents.
Table 1. Intergenerational Status Discordance and Political Alienation
Note: *** p < 0.001; ** p < 0.01; *p < 0.05.
The results in Table 1 demonstrate that the strong correlations presented before hold when controlling for sociodemographic and socioeconomic characteristics of respondents. A one-standard-deviation increase in ISD (SD = 15.5) is associated with a 1–2 percentage point increase in abstention and/or support for radical left or right parties. The magnitude of the effect is notable given low baseline probabilities, particularly when it comes to voting for radical parties (about 14%). The association with (lacking) identification with political parties goes in the same direction and is even slightly more pronounced. The last column of Table 1 shows the mirror image of this relationship—that is, the pronounced negative correlation between negative status discordance and support for mainstream parties. Here, a standard-deviation increase in status discordance reduces support for mainstream parties by about 3 percentage points. The inclusion or exclusion of the Green party in this definition hardly changes the result, reflecting the relatively established nature of the German Greens with their considerable experience in government.
This first set of evidence thus confirms our main expectation that the disappointment following declining mobility prospects has political implications. Net of other variables that capture absolute socioeconomic conditions, downward mobility—at least when conceptualized as intergenerational status discordance—serves as a significant driver of political alienation and dissatisfaction with mainstream parties.
Status Discordance and Party Choice
As a next step, we extend our focus beyond radical vs. mainstream voting and examine support for particular parties. In further disaggregating voting behavior, our goal is to begin exploring the possibility of variation among status-discordant voters. Figure 4 summarizes results from various models assessing the relationship between status discordance and having voted for a specific party in the 2017 general election. We include the same set of individual-level controls as in the previous analysis to keep current socioeconomic conditions of respondents constant (full regression tables are provided in the Supplementary Materials).
Figure 4. Intergenerational Status Discordance and Party Support (Bundestagswahl 2017)
Intergenerational status discordance indeed carries considerable explanatory power when it comes to party choice. In line with our expectations, negative status discordance is consistently related to higher values of support for radical parties. Radical-right parties in particular attract the downwardly mobile, evidenced by robust positive associations between status discordance and support for those parties. In terms of magnitude, the coefficient of status discordance indicates a larger effect than that of log income and age. This favorable comparison to important and well-known predictors of radical-right voting underscores the importance of expectations in shaping voter behavior. At the same time, level of education, perhaps the most consistent factor associated with voting for radical-right parties, is clearly a more powerful driver of support for the Alternative für Deutschland (we display standardized coefficients in the Supplementary Materials). The coefficient for radical-left parties is weakly positive but estimated imprecisely.Footnote 11 In additional analyses below, we expand on this finding by exploring variation in reactions to status discordance by educational background.
Much in contrast, mainstream parties are not appealing to voters who experience negative status discordance. While there is no clear relationship with mainstream left parties, the association with mainstream-right parties is clearly negative. This means that support for CDU, CSU, and FDP is significantly higher among voters who experience positive status discordance. Support for the mainstream right among those on the winning side of mobility trends accords with longstanding theoretical expectations, particularly those elaborated upon by Bendix and Lipset (Reference Bendix and Lipset1959) and Piketty (Reference Piketty1995), and with more recent empirical studies (Alesina, Stantcheva, and Teso Reference Alesina, Stantcheva and Teso2018; Benabou and Ok Reference Benabou and Ok2001). It is worth noting that this pattern is robust when looking at former West and Eastern German regions separately.
Variation in Status Discordance
The results thus far confirm the thrust of the first set of hypotheses in that the experience of intergenerational status discordance goes hand in hand with a pronounced sense of political alienation. People who are not able to meet status expectations formed in childhood are characterized by higher rates of abstention, lower political identification with political actors, lower support for mainstream parties, and a much higher propensity to vote for radical antiestablishment parties, especially radical-right parties.
However, as we previously proposed, given that the effects of occupational change vary widely across groups in society, reactions to status discordance may take on divergent patterns. In the following sections, we examine our two remaining hypotheses by exploring the extent to which the effects of status discordance differ by gender and socioeconomic background.
Gender
As demonstrated earlier, occupational change has carried different consequences for the career trajectories of men and women. Over time, women have experienced occupational gains in absolute terms and also in relative terms compared with men (see additional analyses in the Supplementary Materials), resulting in an overall less frequent incidence of negative status discordance among women (see Figure 2). In light of the strong relationship between political alienation and status discordance demonstrated above, this gendered pattern of intergenerational mobility is the first factor driving overall lower levels of political alienation among women, and vice versa among men.
In addition to different baseline rates of downward mobility, for those women who do experience status discordance, such experiences may be moderated by broader developments toward gender equality that have served to boost women’s socioeconomic prospects across advanced democracies. That is, women’s individual experiences of status discordance may feel less like status loss because their points of reference differ from those of men. Beyond this substantive difference in reference points, our ISD measure relies on fathers’ occupational status and thus may not fully capture the reference points most salient for women.
To test our hypothesis that the relationship between status discordance and political alienation is less pronounced among women, we estimate the interaction effect of gender and ISD on support for radical parties. As Figure 5 shows, the marginal effect of status discordance is indeed much weaker among women. It is no longer statistically different from zero with respect to the radical right and is virtually zero with respect to the radical left. The strong relationship between status discordance and radical voting is thus first and foremost a story about downwardly mobile men, which accords with recent findings about the particularly pronounced salience of occupationally based status threats among men (Gingrich and Kuo Reference Gingrich, Kuo, Kemmerling, van Kersbergen and Marx2021).
Figure 5. Status Discordance and Radical Party Support, by Gender
Socioeconomic Background
In addition to differences among women and men, the effects of occupational change turn heavily on socioeconomic background. We thus expect that reactions to status discordance and subsequent party preferences will vary along this dimension. That is, although a voter with a blue-collar background who now faces poverty may have the same ISD rank as a college-educated voter who ends up in the service class, although they share an increased likelihood to support parties outside the mainstream, we expect that their socioeconomic background influences whether they look to the left or right for answers.
To test our final hypothesis, we use two proxies that capture distinct aspects of voters’ socioeconomic backgrounds. We distinguish respondents by their own educational background and by the occupational status of their fathers (full regression tables in the Supplementary Materials). Figure 6 shows marginal effects of status discordance conditional on whether an individual has a college degree or not. The results reveal that a college degree strongly decreases the likelihood of supporting a radical-right party among respondents experiencing negative status discordance. Although its positive effect on the likelihood to vote for a radical-left party is weaker and imprecisely estimated, this analysis nonetheless supports our hypothesis that socioeconomic background influences party choice.
Figure 6. Status Discordance and Radical Party Support, by Education Level
Figure 7 provides a different proxy for socioeconomic background, showing the marginal effects of status discordance conditional on father occupational status—that is, an interaction of our explanatory variable with a pretreatment variable. Again, the analysis confirms that divergent reactions to status discordance are rooted in socioeconomic background. Respondents who experience downward mobility relative to fathers who worked middle- and, especially, low-skill jobs are significantly more likely to support radical-right parties, whereas we do not find robust evidence of radical-right support among those from high-status origins. In contrast, the probability of supporting a radical-left party increases with the occupational status of respondents’ fathers. Support for the radical left is in fact confined to voters who experience downward mobility relative to fathers at higher positions in the occupational hierarchy. As with the preceding analysis, these coefficients are estimated with a considerable amount of uncertainty and we do not want to overstate their robustness. Despite these limitations, we think that the broader pattern revealed by this analysis—how parental occupational status conditions the relationship between status discordance and political behavior—is consistent and informative.
Figure 7. Status Discordance and Radical Party Support, by Father Status
This pattern carries numerous implications about why the radical left appeals to those from privileged backgrounds and the radical right to those from weaker socioeconomic backgrounds. The former group includes voters who cannot replicate the economic prosperity of their parents despite having a university degree. In contrast to their parents’ generation, they face steep competition from a growing supply of college graduates and increasingly globalized markets, which results in mounting wage pressure even in highly profitable sectors of the economy. This group also includes individuals who seek satisfaction in areas that transcend traditional ideas of career and material success—for example, the nonprofit or arts sectors—and who give up economic ambitions for such work. Radical-left parties appeal to such voters by criticizing elite financial institutions and calling for a sweeping reform of capitalist economies.
In contrast, status-discordant voters from weaker socioeconomic backgrounds prefer radical-right parties’ promise to bring back the conditions that prompted postwar growth and created economic opportunities for workers without college degrees. In the decades since, job prospects for the children of such workers have undoubtedly worsened. Common examples are semiskilled routine work in factories and back offices vulnerable to displacement due to automation and technological change. The once-promising occupational trajectory of relatively well-paid and secure routine employment has all but disappeared for current labor market entrants without a college degree, forcing them instead into dead-end and often poorly remunerated jobs in the service sector. Radical right parties are attractive for these voters because they advocate for a return to the times in which jobs not requiring a college education still had integrity and carried great prospects for upward mobility.Footnote 12
Conclusion
We have argued that the disappointed expectations following fading prospects for mobility have political consequences and that these consequences shed light on the political disruptions unfolding across Western democracies today. To support our argument, we introduced the innovative concept of status discordance, which we capture empirically as the difference between status expectations and status outcomes. Using parental socioeconomic background as the reference point for status expectations, we demonstrate that the experience of disappointing these expectations—an experience becoming increasingly frequent—is an important factor underlying political alienation and dissatisfaction with mainstream parties. Our evidence, based on German household panel data, reveals that the downwardly mobile are more likely to abstain from voting and to vote for radical parties.
We also demonstrated that the effects of status discordance vary by gender and socioeconomic background. First, we showed not only that women experience status discordance less frequently but also that, even when they do, they are not likely to vote for radical parties. We then distinguished between status-discordant voters based on socioeconomic background: downwardly mobile voters with weaker socioeconomic backgrounds tend to support radical-right parties, whereas those from more privileged backgrounds gravitate toward the radical left. Although we do provide some tentative exploration of the attitudinal mechanisms linking negative status discordance to a specific party choice, this remains a fruitful pursuit for future research. More generally, increasing disparities across groups in access to opportunities for mobility likely have profound political consequences that warrant further exploration.
We believe our results carry implications beyond the German context. Downward mobility is increasing across postindustrial democracies, as are electoral victories among radical parties—trends that our results indicate may be linked. Although trends in downward mobility as well as party systems vary across Western democracies, our findings suggest that even in two-party systems such as the US, the growing share of downwardly mobile voters across all socioeconomic backgrounds may help to explain the recent rise of candidates such as Donald Trump and Bernie Sanders, whose political platforms diverged markedly from the usual centrist tendencies of American politics.
More generally, we hope our analysis draws attention to the importance of expectations when it comes to political behavior and to the link between social mobility and political (in)stability. Upward mobility was an undervalued cornerstone of postwar politics. As increasing shares of citizens face declining opportunities for mobility, disappointed expectations are on the rise, affecting voter behavior and electoral outcomes. Importantly, the disappointment we capture is distinct from the usual democratic dissatisfaction in that it cannot be addressed through short-term measures or replacement of party leaders. It instead reflects a sense among a growing group of voters that the political solutions prevailing in recent decades are out of touch with the encroaching reality of downward mobility across postindustrial societies. Although change is inherent to democratic politics, the degree of political disruption taking place in advanced democracies is remarkable. At the same time, looking ahead, the long-run evolution of absolute mobility might also imply that this current phase of disruption is exceptional and transient. More recent birth cohorts who were already born into societies with less optimistic or stagnant mobility prospects likely adjust their socioeconomic expectations to this new reality. As a result, we might see fewer instances where expectation and realized outcome deviate to an extent that creates politically relevant grievances. In any case, given that ongoing occupational and educational trends seem likely to continue, as the task of social and economic inclusion looms, party politics in advanced democracies may never be the same.
Supplementary Materials
To view supplementary material for this article, please visit http://doi.org/10.1017/S0003055422000077.
Data Availability Statement
Research documentation and code that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/GHZFJM. Limitations on data availability are discussed in the text.
Acknowledgments
Previous versions of this paper have been presented at ECPR 2019, SVPW 2020, Social Status Workshop Zurich/Milano (virtual), University of Minnesota’s Comparative Politics Colloquium (virtual), ECPR Joint Sessions 2020 (virtual), and King’s College QPE Online Spring Seminar Series. We thank the participants as well as Tarik Abou-Chadi, Per Engzell, Geoff Evans, Lukas Haffert, Sophie Hill, Arno Van Hootegem, Peter Hall, Minyoung Kim, Giorgio Malet, Alex Mierke-Zatwarnicki, Brian Nolan, David Samuels, Denise Traber, David Weisstanner, and Delia Zollinger for their very helpful comments.
Funding Statement
Thomas thankfully acknowledges funding from a Berggruen-Weatherhead Postdoctoral Fellowship at Harvard University, from the Swiss National Science Foundation (Grant, No. 185204), and the Deutsche Forschungsgemeinschaft (DFG – German Research Foundation) under Germany’s Excellence Strategy – EXC-2035/1 – 390681379. Briitta thankfully acknowledges funding from a Visiting Research Scholarship at WZB Berlin Social Science Center, a Center for European Studies Dissertation Fellowship at Harvard University, and a Weatherhead Center for International Affairs Research Grant.
Conflict of Interest
The authors declare no ethical issues or conflicts of interest in this research.
Ethical Standards
The authors affirm this research did not involve human subjects. The research is based only on secondary analysis of publicly available data that have been stripped of all identifiable information.
Open access
