Firms and the Intergenerational Transmission of Labor Market Advantage

Firms and the Intergenerational Transmission of Labor Market Advantage Per Engzell, Nathan Wilmers Document type Post-print: This manuscript has passed peer review and been accepted for publication by a journal. It may include final edits by the author(s) but no editing or formatting by the publisher. Funding information This research was funded by the European Research Council, Grant Agreement No. 101165962 Markets and Mobility: How Employers Structure Economic Opportunity. Suggested citation Engzell, Per, and Nathan Wilmers. (2025). Firms and the Intergenerational Transmission of Labor Market Advantage. American Journal of Sociology, forthcoming. Date of record September 10, 2025. Terms of use This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International license: https://creativecommons.org/licenses/by-nc-nd/4.0/ Firms and the Intergenerational Transmission of Labor Market Advantage Per Engzell University College London Nathan Wilmers MIT Sloan July 2025 Abstract Pay inequality stems both from differences in workers’ individual characteristics and from firm pay-setting. Some firms capture more value or pass on a larger share of surplus to workers. Yet, intergenerational mobility research focuses on transmission of individual traits and has neglected the question of how firms shape the inheritance of inequality. We use three decades of Swedish population register data to decompose the intergenerational earnings correlation into firm pay premiums and worker effects. One quarter of the intergenerational earnings correlation at midlife is explained by sorting between firms with unequal pay. Employer inheritance accounts for a small share of this firm-based earnings transmission. Instead, high-education and highoccupation workers disproportionately work at high-paying firms. Parental referral networks and the inheritance of industry and labor market context play a supplementary role. As workers with high-education or high-status jobs increasingly benefit from high paying firms, firm premiums constitute a central mechanism through which collective processes drive intergenerational earnings transmission. Acknowledgements: This work received support from the European Research Council grant no. 101165962 (MaMo), Russell Sage Foundation grant no. R-2111-34876, Swedish Research Council for Health, Working Life, and Welfare (Forte) grant no. 2016-07099, Swedish Research Council grants no. 2019-02552 and 2022-02314, and the French National Research Agency (ANR) as part of the “Investissements d’Avenir” program LIEPP (ANR-11-LABX-0091, ANR-11-IDEX-0005-02) and the Université Paris Cité IdEx (ANR-18-IDEX-0001). Richard Breen provided invaluable help at an early stage of the project. Of the many people who provided feedback, Martin Hällsten, Max Thaning, and Wouter Zwysen deserve special mention. For further comments, we thank seven anonymous reviewers and numerous audience members at the RC28 Spring Meeting in Turku, 2021, the European Consortium for Sociological Research Annual Conference in Amsterdam, 2022, the American Sociological Association Annual Meeting in Los Angeles, 2022, and seminars and workshops at Bocconi University, Cornell University, European University Institute, Imperial College London, Institute for Fiscal Studies, Massachusetts Institute of Technology, Oxford Leverhulme Centre for Demographic Science, Rockwool Foundation, Sciences Po, Swedish Institute for Social Research, University of Konstanz, University of Lausanne, University of Wisconsin–Madison, and Queen Mary University of London. Direct correspondence to Per Engzell, UCL Social Research Institute, 55-59 Gordon Square, London WC1H 0NU. Email: [email protected]. 1 1Introduction In capitalist economies, value is created and allocated largely through collective action within firms. Productive collaboration among co-workers can raise revenue and potential pay above what workers’ individual characteristics imply. Monopsonistic employers, in contrast, can suppress pay below workers’ marginal product. Effective collective bargaining by workers at one firm can boost their share of surplus above that received by similar workers in a different firm. Firms are thus a critical vehicle through which collective processes, irreducible to individual traits, translate into unequal outcomes. Yet, a half century of research on intergenerational mobility has largely ignored the role of firms and employers. Instead, researchers argue that social origin influences opportunity via lasting effects on individual worker characteristics: Parents with greater resources are able to invest in their children in numerous ways, and cultivate traits in them such as skill, agreeableness, or work orientation, that are valued in the labor market (Black and Devereux, 2011; Bowles et al., 2009; Farkas, 2003; Jæger and Breen, 2016; Jencks et al., 1979; Morgan et al., 2006; Sewell et al., 1969). This standard approach rests on the assumption that individual characteristics pay offin a competitive labor market. A growing body of research challenges this assumption by studying firms as asiteofemergentinequalityirreducibletomarket-pricedindividualtraits.Working at a highor low-paying firm affects earnings beyond what would be expected from individual worker characteristics (Abowd et al., 1999; Groshen, 1991). Recent changes in the labor market suggest that these firm effects may play a prominent role in intergenerational earnings transmission. Between-firm inequality has grown across many developed countries (Tomaskovic-Devey et al., 2020). Moreover, occupation and worker characteristics are increasingly correlated with working at a high-paying firm (Card et al., 2013; Criscuolo et al., 2020; Wilmers and Aeppli, 2021). This consolidation of potentially independent dimensions of inequality— firm and occupation or education—means that even understanding how individual traits affect intergenerational earnings transmission requires studying firms. We examine the role of firms in stratification in several ways, beginning with a fundamental question: how much do firm differences in pay contribute to intergenerational earnings persistence? We then go on to answer a set of questions on how these firm premiums arise and how they benefit different workers. Do workers from different social backgrounds benefit equally from working at high-paying firms? Is firm-based transmission mostly relevant in the early career or does it carry lasting advantages? Do privileged workers sort into firms with higher profits, firms that pass on a greater share to employees, or both? How does stratification by firms map on to other, more commonly studied dimensions such as education and occupation? And lastly, what helps workers of privileged background sort into high-paying firms? To answer these questions, we bring methods from research on labor market inequality to the study of intergenerational mobility (Abowd et al., 1999). Empirically, we draw on population-wide linked employer-employee data from Sweden, which offer a sufficiently long time span to observe both parents and children at prime working age. By following workers across employers over time, we distinguish 2 the components of earnings attributable to two sources: (a) the firm one works at and (b) durable individual worker traits. The firm component is the premium associated with working at a given firm, conditional on the composition of its workers. The individual earnings component reflects worker traits that are consistently rewarded across firms. We nest this approach in a path model to precisely quantify the contribution of firm pay premiums to the intergenerational earnings correlation. In additional analyses, we test mechanisms by studying how education, occupation, parental co-working networks, and structural labor market positions mediate firm sorting. Our approach uncovers several new facts about the role of employers in intergenerational earnings persistence. Inequality in firm premiums by social background is evident at labor market entry and persists throughout the career. The firm premium among children from privileged backgrounds is relatively constant in absolute terms, but accounts for a shrinking relative part of earnings transmission as individual earnings fan out over the career. Yet, even in mature earnings, firm-based sorting accounts for a quarter of intergenerational persistence. This firm premium advantage enjoyed by high earning-background workers is a function of working both at firms that have a higher value-added per worker and at firms that pass a larger share of value-added into worker earnings. Moreover, it is access to these firms that matters: the earnings boost from working at a high-paying firm is similar for workers from high and low parental earnings backgrounds. What accounts for these patterns? We show that education and occupation account for roughly two-thirds of firm advantages by social background. Most of the remaining third is accounted for in equal parts by referral networks from a parent and inheritance of industry and local labor market. Firms are an especially potent explanation for the part of income inheritance not accounted for by education or occupation: half of the direct effect of parent on child earnings, and most of the class origin gap in earnings, is explained by sorting across highand low-paying firms. We conclude that firm premiums are an important and understudied channel in the intergenerational transmission of advantage. 2Firm premiums and stratification Contemporary employment has bifurcated by firm. Superstars in technology and professional services offer staggering compensation, while outsourced contractors, shrinking manufacturers, and large retailers pay far less. If in an earlier era, Ford Motor Company offered relatively high pay to production workers, now it is largely managers and professionals, at companies like Google and Goldman Sachs, that benefit from employment at high-paying employers (Wilmers and Aeppli, 2021). This bifurcation implies inequality in firm pay premiums: durable, firm-wide differences in pay that affect workers’ earnings. Some firms are high-paying, and consistently pay their workers more across job titles (Groshen, 1991). Firm pay premiums are persistent over time (Engbom et al., 2023) and appear even when comparing firms with similar occupational composition (Groshen, 1991; Wilmers and Aeppli, 2021) and employing workers of similar skill levels (Song et al., 2019). 3 These premiums are quantitatively important for worker earnings, accounting for 15% to 30% of earnings inequality in developed countries (Criscuolo et al., 2020; Tomaskovic-Devey et al., 2020). Incorporating pay differences between firms requires updating theories of stratification, just as prior mobility research has repeatedly incorporated newly ascendant sources of inequality. Marx argued that ownership of the means of production, rather than inherited title, increasingly determined the distribution of economic resources. Blau and Duncan (1967) countered that the separation of ownership from control required attention to gradational differences between employees of different authority. Goldthorpe (1982) argued that the autonomy of much office work fostered a more durable “service relation” distinct from the labor contract, while Weeden and Grusky (2005) underlined how core class mechanisms such as closure and solidarity can operate at the level of detailed occupations. These shifts in stratification research have thus tracked the rise of capitalists; the managerial revolution; the post-industrial transition; and the subsequent fracturing of class identities. If in contemporary labor markets, defined by superstar firms and skill segregation, economic success requires working at a high-paying firm, how does this affect intergenerational mobility? We theorize how firm premiums affect stratification in a two-stage model. First, firms accumulate resources, either through market power or through the effective combination of factor inputs. Second, workers and other stakeholders bargain over the allocation of these resources (Tomaskovic-Devey and Avent-Holt, 2019). In both of these stages, privileged workers can disproportionately reap the benefits of firm premiums. In the first stage, firms are a distinct basis of inequality—compared to capital ownership, education level, or occupational membership—in that they are direct vehicles of collective action, rather than an individual’s endowment. A group of doctors may coordinate to restrict the supply of medical graduates, as a lobbying effort outside of their usual work, to increase the value of their skills (Weeden, 2002). But a group of co-workers coordinates in virtue of being co-workers. Firms form to facilitate productive activity that imposes excessively high transaction costs when performed as a market-mediated nexus of contracts (Williamson, 1975). As such, in afirststage,firmscanmoreorlesssuccessfullydevelopamarketnicheorcultivate efficient production practices that allow them to accumulate resources. Workers’ socio-economic background can matter in several ways for this resource accumulation stage. Most directly, class solidarity can reduce co-workers’ temptation to shirk or otherwise undermine a firm’s performance. As Burawoy and Wright (1990, p. 265) argue, “firms within which workers can effectively reduce free riding will be the most profitable”. Indeed, this insight motivates extensive research on ways that organizational citizenship behavior boosts firm performance (Podsakoffand MacKenzie, 2014). Class cooperation in monitoring and disciplining co-workers can give owners and managers incentive to gather workers of a common class background. This homogeneity dynamic can benefit high-earnings background workers, as when hedge fund employees from privileged backgrounds rely on “networks of trust and loyalty” to blunt financial risks (Neely, 2022, p. 21). But it 4 can also benefit less privileged workers, as in blue collar union organizing (Wilmers, 2019) or ethnic enclaves (Portes and Sensenbrenner, 1993). Beyond solidarity and monitoring, positive peer effects and human capital complementarities can raise productivity for teams of high-education and highstatus occupation workers, over and above what individual workers on those teams could produce in other contexts (Kremer, 1993; Song et al., 2019; Wilmers and Aeppli, 2021). Insofar as workers of privileged backgrounds obtain better education and higher status occupations, they are most likely to benefit from this additional productivity boost. Finally, even if a firm’s resource gathering advantage is entirely exogenous to the characteristics of its workers—a monopolistic position in a product market for example, or a lucrative patent (Kline et al., 2019)—privileged workers may use networks and strategic knowledge to effectively obtain positions within profitable firms. By strategic knowledge, we mean awareness of these high-resource firms and their selection and hiring processes (Rivera, 2012). All of these first-stage resource accumulation processes increase potential earnings for workers. However, the rewards for workers at a highly productive firm are only potential (Fligstein and Fernandez, 1988). In a second stage, firms then allocate rewards across stakeholders. Here, solidarity again matters. High class workers may bargain more effectively, invoking their educational and occupational credentials to demand a larger share of a firm’s resources. This would mean that firms that employ privileged workers must pay larger firm premiums. In contrast, firms that employ vulnerable workers could pay less, even exercising monopsonistic power to push wages below workers’ marginal productivity. These second stage processes look more like conventional social closure (Murphy, 1988; Parkin, 1979)—a zerosum bargaining setting, in which workers try to capture surplus from a principal, often by restricting labor supply. While in the abstract, these two stages of firm premium development need not deliver advantage disproportionately to privileged workers, recent research provides suggestive evidence that contemporary firm premiums are benefiting privileged workers. In the mid-twentieth century, firm premiums were associated with blue-collar employers, often large and unionized (Cobb and Lin, 2017; Wilmers, 2019). These relatively high-paying jobs for non-college workers provided generational security for low-earning families: union cards and jobs in specific plants were passed from fathers and uncles to sons and nephews (Bryson and Davies, 2019). Similarly, ethnic labor market niches in large cities protected secure public sector or construction jobs for particular immigrant groups (Waldinger, 1999). The monopolization of these firm premiums can provide higher earnings for relatively low-education and -status background workers.1 Since the 1970s however, high-paying blue collar employers have faced increasing cost pressure that has undermined their firm premiums (Sørensen, 2000; Wilmers, 2018). However, firm premiums have not thereby disappeared. On 1In the language of Parkin (1979), this would amount to “usurpationary” social closure, which involves the efforts of subordinate groups seeking to gain access to resources, opportunities, or positions previously monopolized by a dominant group. The prevailing mode of social closure is “exclusionary,” which involves the efforts of a dominant group to restrict access to resources and maintain their privileged status and power. 5 the contrary, in recent decades firm premiums have either grown in importance (Criscuolo et al., 2020) or remained stable (Song et al., 2019) across most developed countries. But, there has been a reallocation of firm premiums from blue-collar workers toward white-collar and professional employees (Criscuolo et al., 2020; Song et al., 2019; Wilmers and Aeppli, 2021). In the place of high-paying blue collar employers, the rise of finance (Krippner, 2011), superstar technology firms (Autor et al., 2020), and high-paying sole-proprietorships (Smith et al., 2019) have meant aquitedifferent industry and occupational composition of high-paying firms. These new sources of firm premiums are often marked by credentialism and social exclusion, as in finance (Neely, 2022), or provide opportunity for direct transmission of high-paying positions, as in family inheritance of sole-proprietorships (Corak and Piraino, 2011). Recent changes in the sources of firm premiums across industries and occupations therefore make them particularly likely to benefit high-income background workers. 3Limits of status attainment and human capital models Theorizing these dynamics linking firm premiums to workers’ class background responds to longstanding calls for incorporating firms and labor markets into models of stratification (Baron and Bielby, 1980; Spilerman, 1977; Stolzenberg, 1978). While early models of status attainment posited the direct involvement of parents in finding a job (Blau and Duncan, 1967), this focus was soon replaced by psychological influences on ability, aspirations and, ultimately, schooling (Sewell et al., 1969; Sewell and Hauser, 1975). The resulting model faced repeated criticism for its onesided focus on the individual (Baron, 1984; Featherman, 1981). Yet, a compelling alternative failed to emerge and the status attainment model remains, implicitly or explicitly, the primary backdrop for stratification research. The field today therefore remains preoccupied with education and other forms of human capital (Bloome et al., 2018; Breen and Müller, 2020). This is not without reason, as education is both the main conduit of social reproduction and a vehicle for intergenerational mobility (Hout, 2012; Torche, 2015). However, recent research emphasizes that human capital is not the only source of advantage for privileged children (Bernardi and Ballarino, 2016; Witteveen and Attewell, 2017, 2020). Among UK managers and professionals with similar educational background, high-class children end up higher-paid, in part by sorting to larger and more centrally located firms (Laurison and Friedman, 2016). In the US, while education accounts for the bulk of occupational status transmission, it only accounts for about half of the intergenerational persistence in family income (Torche, 2016). Even the classical finding that high-skilled labor markets operate more meritocratically is weaker when income is studied, and when accounting for selection processes into education (Fiel, 2020; Zhou, 2019). These findings highlight the importance of understanding processes that take place over and above education in studying intergenerational earnings transmission. Labor economics provides a tool to do so through the use of linked employer6 employee data (Abowd et al., 1999; Song et al., 2019). Observing the same individuals working across multiple firms, and the same firms employing successive individuals, reveals that some firms consistently provide higher pay for the same kind of work. Conversely, some workers consistently earn a high salary no matter where they work. These two aspects allow us to identify a part of earnings attributable to firm pay-setting, and one attributable to durable individual characteristics. Applications of this model in economics have tended to offer an interpretation of these two components in terms of firm productivity, on the one hand, and worker human capital or “ability” on the other (Card et al., 2018). We are careful to avoid this interpretation. As we have argued, the firm-specific component of pay reflects a complex combination of market power, complementarities among factor inputs, and contested profit allocation decisions. Similarly, the individual component reflects anything that helps the worker gain a persistent advantage on her co-workers, including human capital but also personal connections, cultural capital, entitlement, or status characteristics (Correll et al., 2017; Khan, 2010).2 For these reasons, pay differences within and between firms do not map on to asimpledistinctionbetweenmeritocraticandnon-meritocraticpay.Nevertheless, we show that both the role of education, and the effects of social background over and above education, can be clarified by attention to pay differences between firms. Doing so highlights that education, far from simply conveying human capital, also serves as a powerful sorting device in matching to employers—in line with longstanding sociological accounts (Bills, 2003; Collins, 1979; Domina et al., 2017). Before turning to our quantification of the role of firm premiums in intergenerational earnings transmission, we distinguish specific mechanisms through which firm premiums may be distributed unequally among workers of different backgrounds. 4Firm premium mechanisms of intergenerational transmission There are several links that tie high-earning family background workers to highpaying firms. First, privileged workers can obtain credentials and occupational standing that either support firm premiums or gain workers access to high-paying firms. Second, high-premium firms and industries can be obtained through parental referral and co-worker network ties, as a type of social capital. Lastly, children may work for similar employers to their parents, even in the absence of activated network ties, simply due to a common opportunity structure. In the following, we elaborate these mechanisms, before testing them empirically. 4.1 Educationand occupation-driven firm advantages Prior stratification research has demonstrated that children of high-earning parents have more education and work in higher-status occupations than children of 2Swift (2005) argues that traits that are “constitutive of the person” can serve as a legitimate basis for differential rewards in the labor market. While our decomposition is well-suited to identify such traits, we remain less optimistic about the ethical significance of this distinction. 7 lower-earning parents. What does this mean for firm premiums? Concentrated employment of high-education and high-status occupations can itself anchor firm premiums. In O-Ring models of production, complementarities among high-performing workers can drive higher productivity and pay than any individual worker would receive in other contexts (Cornelissen et al., 2017; Kremer, 1993). Beyond pure productivity effects, concentrated employment of positionally powerful workers can increase collective bargaining power (Tomaskovic-Devey and Avent-Holt, 2019). Note that these effects, while a function of education, are distinct from an individual worker’s pay-offto human capital. Emergent productivity and bargaining power here benefits groups of highly educated and high-status workers over and above their outside options as individuals competing on the labor market. Occupational and educational advantages can also sort workers to better firms (Hirsch, 1977; Thurow, 1975). One US analysis finds that a quarter of the higher education pay premium is attributable to sorting across high-paying firms (Engbom and Moser, 2017). Indeed, at the higher end of the institutional hierarchy, a small number of universities serve as pipelines for elite employers (Binder et al., 2016; Bühlmann et al., 2022; Rivera, 2016; Weinstein, 2018). In the Scandinavian context we study, field of study can be a more powerful source of stratification than institution (Borgen and Mastekaasa, 2018; Hällsten, 2010; Hällsten and Thaning, 2018). This pathway can route high-earning background workers to high premium firms. Beyond levels or even fields of education, children of high-earning parents may cultivate cultural dispositions that gain the confidence of gatekeepers at highpremium firms (Armstrong and Hamilton, 2013). As long as class-based dispositions are domain general, like ease and entitlement, they can both benefit children as individual traits and in the firm sorting process (Bourdieu, 1984; Khan, 2010; Lareau, 2011). Indeed, recent qualitative studies of elite employers all emphasize the importance of tacit cultural performance to getting in and staying employed. A study of accounting, broadcasting and architecture firms found that behavioral codes, often learned in elite educational environments, gave high-class background workers an advantage (Friedman and Laurison, 2020). In hedge funds, upper-class men succeed by embodying stereotypes about high risk-tolerance and trustworthiness (Neely, 2022). Consistent with this mechanism, a contemporaneous study of Israel finds that high-earning background children disproportionately sort to high-paying firms and that this sorting is largely due to those children’s individual traits (Dobbin and Zohar, 2023). Through both value creation and capture, privileged workers’ educational and occupational advantages can deliver them disproportionately to high premium firms. 4.2 Job referral networks and biased sorting Social networks are a ubiquitous source of inequality (Bourdieu, 2000; DiMaggio and Garip, 2012; Lin, 2002; Small, 2009) and the job search process is no exception (Castilla et al., 2013; Granovetter, 1974; Marsden and Gorman, 2001; Trimble and Kmec, 2011). While much research has focused on “weak ties,” acquaintances with 8 6.2 Path model Having estimated the separate firm and individual worker contributions to earnings, we explore transmission in a path model. Does intergenerational earnings transmission work through the inheritance of individual traits as assumed in traditional intergenerational mobility research? Or is a significant part of intergenerational advantage instead due to employment at a high-paying firm? Figure 1 displays this model, where the superscript ⇤denotes the parental generation. We model earnings as a function of firm and worker components in each generation: Y⇤ i=a1F⇤ i+b1W⇤ i+✏1i,(3) Yit =a2tFit +b2tWit +✏2it,(4) where Y⇤ iis the father’s average earnings from 1990–1999, Yit is the child’s earnings at age t,andtheright-handsidevariablesarethecorrespondingfirm (F⇤ i,F it)andworker(W⇤ i,W it) components. We are interested in the following pathways: Firm-to-firm: Y⇤ i F⇤ i!Fit !Yit Worker-to-firm: Y⇤ i W⇤ i!Fit !Yit Firm-to-worker: Y⇤ i F⇤ i!Wit !Yit Worker-to-worker: Y⇤ i W⇤ i!Wit !Yit We estimate the importance of each of these paths of transmission by taking the product of each intergenerational correlation of earnings components (denoted by ⇢with ffirm and wworker subscripts) and the importance of each component for overall earnings within each generation (aand bfrom Equations 3 and 4) (Duncan, 1966). Akeyfeatureofourmodelisthatfirmandworkercharacteristicsmaybe correlated both across and within generations, as represented by the bidirectional arrows in Figure 1. This correlation is implicit in our use of bivariate correlations as opposed to partial regression coefficients for the intergenerational parameters ⇢ff, ⇢fw,⇢wf,and⇢ww. We elaborate on this point in Appendix A8, where we display an alternative decomposition that separates the partial regression coefficients and the covariance between father firm and worker components. This gives identical results for the total magnitude of each pathway, but divides these effects into direct and indirect paths. We also test for interactions between family background and firm premiums in determining children’s earnings, to validate the assumptions of the path model. While our model allows for within-generation correlation between firm and worker components, we take no strong stand on the nature of this correlation. It may stem from workers influencing firms, firms influencing workers, or a combination of both, which is not well-captured in our path diagram. Explicitly incorporating these as causal paths would make for a richer model, but at the expense of analytical tractability. To keep the graph acyclical, imposing a path in one direction (e.g., from worker to firm within generation) rules out the possibility of a path in the opposite direction. Therefore, while our model is couched as a causal diagram, 15 we interpret it as providing a descriptive decomposition of the intergenerational earnings correlation. Given a set of assumptions that we describe further in Appendix A9, the intergenerational earnings correlation at child’s age tcan then be decomposed as: ⇢t=corr(Y⇤ i,Y it)(5) =⇢ff,t ·a1·a2t | {z } firm-to-firm +⇢wf,t ·b1·a2t | {z } worker-to-firm +⇢fw,t ·a1·b2t | {z } firm-to-worker +⇢ww,t ·b1·b2t | {z } worker-to-worker . Equation 5 shows that the relative contribution of each mechanism to the overall intergenerational correlation depends not only on the correlation between father’s component and child’s component (the ⇢terms), but also on the importance of those components for each generation’s earnings (the aand bterms). For example, a high correlation between father’s firm premium and son’s firm premium may account for asmallshareoftheoverallintergenerationalcorrelationifthefirmpremiumisa small component of overall earnings. 6.3 Firm contributions to mobility by career stage So far, we have abstracted from the dimension of children’s age. The 30-year span of our data lets us observe fathers and children over several years near mid-career earnings. Here we describe how we implement this focus on mid-career earnings, and how we extend this analysis to study earnings mobility throughout the early career. In all our analyses, we treat parental social background as time-invariant. We do so by collapsing father’s earnings Y⇤ iacross the first 10 years in our data, 1990–1999. Fathers are in the age range 30–59, which is a time when male earnings are representative of lifetime income and the children would have been in their formative years. Taking multi-year averages centered at prime working ages for earnings is common practice in the literature on intergenerational mobility and key to avoid common biases of attenuation and life cycle bias (Fox et al., 2016). Because the average of a uniform variable is generally not uniform, we reimpose the rank transform within (child) gender–age cells to retain the interpretation of coefficients as rank-order correlations. We create measures of fathers’ firm-level component F⇤ i and individual worker-level component W⇤ iin the same way, by collapsing predicted values over the years 1990–1999 and repeating the rank transform. For children, we perform both a time-invariant and a time-variant analysis of earnings. First, in our main analyses, we treat earnings as time-invariant by averaging earnings ranks across ages 38–42, which provides a good proxy for lifetime earnings (Nybom and Stuhler, 2017). Here we exclude the 1978 cohort who is only followed until age 41. We average the firm and individual worker fixed effects (Fit,Wit)overthesameages,38–42. Likeforthefathers,wereimposetherank transform after taking multi-year averages. Second, in a time-variant analysis, we study annual earnings for each year between the ages 25 and 40. This means following children from 2000–2015 for our oldest cohort (born 1975) and 2003–2018 for our youngest (born 1978). These age-specific estimates allow us to track the importance of different earnings components to mobility across career stages. 16 6.4 Dissecting the firm premium Why exactly does parental background link up to firm premiums? Our theoretical framework highlights two ways that firms can boost individual earnings: firstly, profitable firms generate a greater resource pool overall, and secondly, firms make different decisions about what share of that pool gets passed on to employees. Workers from high social origins could benefit from both: they may have better access to high-profit firms and also be better positioned to bargain for compensation. The Swedish company register contains value added per worker, which offers a measure of potential firm premiums before profit allocation decisions are made. We use this to create two versions of our firm ranking: one for the part of the premium predicted by value added, and another for the residual representing redistribution conditional on firm profit. We correlate both with father’s earnings rank to test whether privileged workers benefit from matching to firms with higher profit, higher pass-through, or both. Next, we ask two questions about how firm-based advantages overlap with education and occupation. First, how much of the economic returns to education and occupation are mediated by firm sorting? We use “returns” loosely to refer to observational earnings differences across degrees and occupations, rather than the causal coefficients typical of returns-to-schooling estimates. Second, how much of residual influence of social background on earnings—that is, not mediated by education or occupation—is explained by firm sorting? We look both at the direct association between parent and child earnings, and differences in pay by social class background. In all these analyses, for children we focus on mature earnings around age 40, and therefore make use of the latest available information on education and occupation in 2019. Our data on education consist of 44 discrete levels and 358 detailed fields, which together make up 1,993 unique categories.3Occupational data consist of Swedish census codes at the 3 digit (150 categories) and 4 digit (431 categories) level. Education data are for the most part reported directly by the degree-granting institution. Occupational data come from various sources, including payroll data from large or public-sector employers, as well as surveys of small enterprises. Complete data on occupation is available in 1990, the year Sweden performed its last Census. We use this information to construct fathers’ social class as described below. Post-1990, the occupation register draws on a range of sources available at different intervals. Most important is the wage structure statistics, which covers public employers and large private employers (more than 500 workers). For smaller companies, Statistics Sweden surveys a sample of 47,000 annually, with complete coverage achieved over a period of 5 years (SCB, 2019). To capture observational returns to education and occupation in a single coefficient, we first regress the individual’s earnings rank around age 40 on a set of indicator variables reflecting the detailed categories, and store predicted values from this regression. We then calculate the correlation between predicted and ac3In our earlier analysis of field-dependent earnings trajectories, we use a coarser categorization of 118 fields necessary to estimate these trajectories reliably. When using field of study as a control variable, we face no such constraint, and thus use the most detailed categories available. 17 tual earnings rank. The resulting correlation coefficient can be interpreted as a measure of the proportion of earnings attributable to education and occupation. Thereafter we calculate partial correlations that account for the child’s firm component of earnings, to see how much of the returns to education and occupation are mediated by firm sorting. To study direct effects of social origin, we first focus on the residual association between father and child earnings that remains when the influence of the child’s education or occupation is partitioned out. Moreover, we look at the class gap in earnings within occupations (Laurison and Friedman, 2016). For father’s class, we rely on 1990 Census data and the Swedish socioeconomic classification (SEI), similar to the Erikson–Goldthorpe class schema (Carlsson et al., 1974). We group the middle and higher salariat (46, 56) together with professionals (60), and contrast them against remaining categories: workers (11–22), lower salariat (36), farmers (79), and entrepreneurs (89). The class pay gap is defined as the difference in earnings percentiles of children from high and low class, once detailed education or occupation are accounted for. 6.5 Explaining sorting into high-paying firms In further analyses, we go on to test a set of theoretically relevant explanations for how workers sort to firms by social origin: sorting by human capital, job referral networks, and structural locations in the labor market. First, we ask how much of the firm sorting advantage is explained by sorting on educational and occupational attainment. We regress the child’s firm rank Fit on father’s earnings rank Y⇤ i, and see how their correlation changes with the same detailed educational and occupational controls as above. We also test whether firm sorting is due to unobserved individual worker characteristics. High-paying firms may be particularly good at identifying and attracting high productivity workers. If this is the case, then firm sorting advantages for privileged-background workers may be due to their individual characteristics. To test this, we control for the child’s individual component of earnings Wit in regressing the child’s firm component on father’s earnings. We operationalize job referral in two ways. First, we look at direct referral as the incidence of working at the same workplace as the father. Second, we identify indirect referral as finding work with a senior colleague who at some point earlier has worked with the father. To construct these networks, we make use of workplace identifiers rather than firm identifiers as in our main analysis, and restrict links to workplaces of 1000 people and fewer. This is to make it more plausible that a social tie will actually exist, which is doubtful at the level of firms or for very large workplaces. To ensure that these overlaps do not simply capture other mechanisms such as geographical propinquity we also identify a group of “placebo” hires. In these, there is a worker connecting the father’s and child’s workplace but no shared overlap in time, the overlap with the father occurred after the overlap with the child, or the worker entered the child’s workplace after the child was hired. Again, we regress the child’s firm component of earnings on father’s earnings and inspect 18 $JULFXOWXUHRWKHU )LQDQFHDQGEXVLQHVV (GXFDWLRQDQGUHVHDUFK +HDOWKSXEOLFDGPLQVHUYLFHV 6XSHUVWDUILUPV /DUJHILUPV +LJKYDOXHDGGHG +LJKRSHUDWLQJSURILW 0DMRULW\LPPLJUDQW 0DQXIDFWXULQJ &RQVWUXFWLRQ (QHUJ\ 7UDGHDQGFRPPXQLFDWLRQ    0HDQILUPSUHPLXPUDQN    0HDQIDWKHUHDUQLQJVUDQN Figure 2: Types of firms by firm premium and social selectivity. Note: Mean firm premium rank shown on the vertical axis, mean father earnings rank shown on the horizontal axis. Circles are shown with size proportional to their employment share, and the diagonal line shows the weighted regression line of best fit. how the coefficient differs for groups who experience job referral versus those who do not. To address the transmission of structural location in the labor market, we study geographic immobility, industry overlap, and firm inheritance. In these analyses, we focus on the father’s and child’s main employer, defined as as where the father held the longest tenure throughout the period 1990–1999, or the son throughout 2000–2019. Hence, we do not directly observe whether a father referred a child to their firm. Especially with large employers, overlap could occur by chance, so this is a broader mechanism than the referral networks identified above. Geographic immobility occurs when the father’s and child’s main employer are both located in the same municipality, and sector inheritance when they are located in the same 1-digit industry.4Here, too, we regress the child’s firm component of earnings on father’s earnings and inspect how the coefficient differs for groups who share municipality, industry, or employment with their father with those who do not. 4Sweden consists of 290 municipalities, the population of which averaged about 30,000 in 2000. The 1-digit level of the Swedish industry classification (SNI) encompasses 10 categories: agriculture, construction, culture and services, education and research, energy, finance and business, health care, manufacturing, public administration, trade and communication, as well as an “unspecified” category. 19 7Results To motivate our analysis, Figure 2 plots various types of firms by their pay premium and the degree of social selection into the firm. The vertical axis shows the average firm pay premium, and the horizontal axis shows the average father earnings rank among employees. Circles are displayed with size proportional to their employment share, and the diagonal shows a weighted regression line of best fit. An important source of differences in pay is industry. Agriculture and, to alesserextent,construction,recruitpredominantlyfromthebottomhalfofthe father earnings distribution. But while agriculture offers a lower firm premium than other industries, construction is close to the average firm premium rank. As such, the construction industry provides relatively high-paying jobs for children from the bottom half of the father earnings distribution. At the other extreme, finance and business recruit from the top half of the father earnings distribution and offer pay above the mean firm premium rank. Most other industries draw from father’s earnings ranks that average out to the middle of the distribution, but differ in the pay that they offer workers.5For example, workers in education and research as well as health, public administration, and service—in large part public-sector jobs—get paid less. Conversely, firms in energy and manufacturing pay a higher premium. These firms, then, are likely to offer upward mobility for children from the broad middle classes. This is notable given that manufacturing employment has declined in Sweden like elsewhere (see Appendix A10). We also distinguish firms in the top decile of firm size, value added, operating profit, or firm premiums (“superstar firms”). With the exception of large firms, these categories predominantly hire privileged workers and offer significantly higher pay than other firms. Another finding emerges for immigrant-dominated firms, defined as those where the majority of workers have an immigrant background (first or second generation). These firms overwhelmingly recruit from the lower half of the father’s earnings distribution, yet they offer higher premiums than expected given their workforce composition—possibly due to ethnic solidarity facilitating profitsharing within the firm (cf. Åslund et al., 2021). In Appendix A11, we provide a more detailed breakdown of how firm premiums depend on sector, industry, and occupation. Work in the public or private sector alone explains 9% of the variation in firm premiums and 1-digit industries explain 20%. Sector and industry together explain 22% of the variation, or 23% with an interaction term between the two. Occupation explains 32–35% depending on whether a 3or 4-digit classification is used. The combination of occupation, sector, and industry together explain 40% of the variation in firm premiums. 5That the typical worker is recruited from the middle does not imply that recruitment is uniform across the whole range of father earnings. In Appendix Figure A21, we inspect recruitment from both tails of the distribution, revealing further nuances. For example, manufacturing, as well as education and research, underrepresent children from both poor and rich households. 20     &KLOGUDQN       )DWKHUHDUQLQJVUDQN )LUP :RUNHU (a) Sons     &KLOGUDQN       )DWKHUHDUQLQJVUDQN )LUP :RUNHU (b) Daughters Figure 3: Mean child firm and worker rank by father earnings rank. Note: The figure shows the child’s expected firm and worker rank by father earnings rank. Father earnings rank measured 1990–1999, child firm and worker rank measured at age 38–42. Father’s earnings are associated both with children’s individual earnings and the relative pay of the employer they sort to. 7.1 The contribution of firms to overall intergenerational earnings persistence We have established that sorting into different types of firms varies by social background, but substantial sorting likely occurs within these broad firm categories as well. To provide a more detailed view, Figure 3 presents the expected rank of children’s firm and individual worker components of earnings across percentiles of father’s earnings. Here we also separate results by child gender. Figure 3 shows that fathers’ earnings have a substantial association with both the individual worker and firm components of child earnings. For sons, the individual worker rank association is 0.24 with fathers’ earnings, comparable to the earnings correlation itself, at ⇢=0.24 (see Appendix Table A13). yyy The firmlevel association is smaller but nevertheless substantial, at 0.15.Fordaughters,the worker and firm associations are more similar: the individual worker rank association is 0.20 and the firm-level association 0.16,comparedtoanoverallearnings correlation of ⇢=0.20 (see Appendix Table A14). So, father’s earnings predict not only the individual worker component of children’s earnings, emphasized in prior intergenerational mobility research, but also the firm premium component. Given these strong correlations, we next ask how much firm pay differences contribute to the overall persistence of earnings across generations. To do this, we estimate all of the separate terms in Equation 5 of our decomposition framework above. We bring these components together in Table 1, which details the proportional contribution to the earnings correlation of each of the four paths: firm-to-firm, worker-to-firm, firm-to-worker, worker-to-worker. Filling in the path model is critical, as, consistent with earlier work on the two-way worker-firm fixed effect model, the worker component is substantially more important to explaining overall earnings variation (as captured for sons in the firm coefficient a2=0.36 compared to 21 Table 1: Decomposition parameters. Estimate (percent) Parameter Sons Daughters firm-to-firm = ⇢ff·a1·a20.016 (6.80%) 0.007 (3.91%) worker-to-firm = ⇢wf ·b1·a20.037 (16.4%) 0.036 (19.2%) firm-to-worker = ⇢fw ·a1·b20.013 (5.66%) 0.008 (4.45%) worker-to-worker = ⇢ww ·b1·b20.162 (71.1%) 0.137 (72.4%) ff+ wf + fw + ww = corr(Y⇤ i,Y i)0.228 (100%) 0.189 (100%) ⇢ff=corr(F⇤ i,F i)0.120 0.070 ⇢wf =corr(W⇤ i,F i)0.126 0.150 ⇢fw =corr(F⇤ i,W i)0.048 0.031 ⇢ww =corr(W⇤ i,W i)0.263 0.218 a1=corr(F⇤ i,Y⇤ i|W⇤ i)0.363 0.366 b1=corr(W⇤ i,Y⇤ i|F⇤ i)0.837 0.835 a2=corr(Fi,Y i|Wi)0.355 0.289 b2=corr(Wi,Y i|Fi)0.738 0.751 Note: Path components of the intergenerational earnings correlation as described in Figure 1. Father earnings rank measured 1990–1999, child earnings rank measured at age 38–42. The most important path is worker-to-worker, followed by worker-to-firm, firm-to-firm, and firm-toworker. Even in mature earnings, firm-based mechanisms account for more than a quarter of intergenerational earnings persistence. the worker coefficient b2=0.74). Although father’s earnings strongly predict son’s firm premium, the firm premium, for both generations, accounts for less variation in earnings than does the worker component. Taking these different correlations with overall earnings into account, we find that the worker-to-worker path accounts for between 71% (sons) and 72% (daughters) of the total intergenerational earnings correlation. Firm-related pathways account for the remaining quarter of intergenerational mobility. This decomposition shows that while worker characteristics are the dominant source of intergenerational earnings transmission, firm premiums play an important role in stratification. Extending this analysis to workplaces reveals that as much as two thirds of intergenerational earnings resemblance is driven by workplace segregation, reducing the direct worker-to-worker link to 61% for sons and 59% for daughters (Appendix A3). Table 1 also provides our first evidence into the mechanisms through which firm premiums affect intergenerational mobility. Although the underlying firm-tofirm (⇢ff)andworker-to-firm(⇢wf)correlationsaresimilar,theworker-to-firmpath is more important for the overall intergenerational correlation. This is because a father’s individual worker component is a more influential predictor of earnings than is his firm premium rank. Specifically, the firm-to-firm and firm-to-worker paths each account for about 6–7% of the intergenerational earnings correlation for sons, and about 4% for daughters. Markedly more important is the worker-to-firm path which alone accounts for more than half of the share explained by firms: 16% of the overall correlation for sons and 19% for daughters. 22     &KLOGHDUQLQJVUDQN       )DWKHUHDUQLQJVUDQN &KLOGILUPTXDUWLOH 4 4 4 4 (a) Equal returns       'HQVLW\       )DWKHUHDUQLQJVUDQN &KLOGILUPTXDUWLOH 4 4 4 4 (b) Unequal sorting Figure 4: Father and son earnings by quartiles of the firm premium. Note: The left panel shows expected son earnings rank by father earnings rank at different quartiles of the son’s firm premium. The right panel shows the density of father earnings rank at different quartiles of the son’s firm premium. Father earnings rank measured 1990–1999, son firm and earnings rank measured at age 38–42. Differential sorting, not differential rewards, drives inequality. Daughters shown in Appendix Figure A23. This first set of results shows that children of high-earning parents disproportionately work at high premium firms. While this process is not the main channel of intergenerational earnings persistence, it accounts for around a quarter—a quantitatively meaningful share. We further show that the main source of this firm-related intergenerational transmission is the worker-to-firm path: parents with valuable worker characteristics have children who work at high-paying firms. 7.2 Who benefits from working at a high-paying firm? Do workers from highand low-earnings backgrounds benefit equally from working at high-paying firms? Testing this matters for two reasons. First, any policy attempt to increase mobility by equalizing access to employers would have to proceed on this assumption. Second, our path model assumes linear, additive effects, so violations thereof would challenge its interpretation. Theoretically, this assumption could break down in different ways. Working at a high-paying firm could diminish parental influence on a child’s earnings, akin to how obtaining a college degree is believed to act as a leveller (Karlson, 2019; Torche, 2011). Alternatively, the influence of parental background could loom especially large in these settings, as suggested by literature on the “class ceiling” in elite managerial positions and professional services (Friedman and Laurison, 2020). We test this in Figure 4, left, which shows a separate father–son earnings association at each quartile of the firm premium. This figure reinforces the importance of firms for overall earnings: sons from the same father earnings percentile earn 20 percentile points more when working for a top-quartile firm compared to 23 abottom-quartilefirm. 6Moreover, it appears that the linear, additive model is aremarkablyaccuratefit,asthereislittledetectableheterogeneityoftheslope across different levels of firm premiums. Corresponding results for daughters are shown in Appendix A12. They reveal that firm premiums in the top quartile appear especially important to boost daughters’ earnings but the finding of largely parallel father–child associations of earnings at different firm premiums holds for daughters as well. In sum, firm premiums seem to benefit privileged children mainly in that they are more likely to find work with a high-paying firm, but once inside the firm, highand low-earnings background workers derive a similar boost (or penalty) to their earnings. Inequalities in access, not in outcomes, drive the firm-based advantage of workers from privileged backgrounds. Figure 4, right, visualizes the very unequal probabilities with which workers of different parental earnings background enter firms with high or low pay premiums. The figure shows that this sorting is not limited to a specific part of the firm premiums distribution: across the distribution of firm, unequal sorting by father’s earnings emerges. 7.3 When do firm premiums matter? How does the importance of these firm and worker transmission components change over a career period? Some sources of persistence—like firm inheritance and direct parental aid in job search—should be strongest at the early career stage, before tapering out by mid-career. Figure 5 tests this by decomposing the intergenerational earnings correlation into its path components at each year of a child’s career. Firm sorting is indeed most important early in the career and mediates the bulk of earnings transmission in the first few years of labor market entry. This fades gradually to about a quarter of the total correlation by age 40 (for relative numbers, see Appendix Figure A9). This decline across the career in the firm-related share of earnings transmission is due to the increase in importance of the worker-to-worker path, as the child’s worker component becomes a more important portion of their earnings. At the same time, firm-to-firm transmission faces a small decline in absolute terms and alargerelativedecline. Havingaparentemployedatahigh-earningfirmmainly benefits workers when they are young. In contrast, the worker-to-firm transmission path grows in absolute terms and retains an important share of intergenerational earnings persistence up through age 40. Note that this pathway, in which children of high-earning parents end up working at high-premium firms, could be important both at early and later career stages. Direct parental aid should matter mainly early on, when parents are still active and engaged with their children’s labor market attainment. Indirect human and social capital sorting advantages, in contrast, could pay offthroughout a child’s career. The continued importance of the worker-to-firm pathway up through 6This gap in earnings results from differences in firm pay, but also sorting of workers into firms, which we address in a later section. Studying heterogenous returns while accounting for selection is a complex task, as highlighted by studies on higher education and mobility (Fiel, 2020; Karlson, 2019; Zhou, 2019). 24 Table 2: Association between father earnings and son firm sorting, with controls. b/se b/se b/se b/se Coefficient 0.149 0.046 0.030 0.014 (0.003) (0.003) (0.004) (0.004) Education – XXX Occupation – XXX Worker FE – XXX Direct referral – – XX Indirect referral – – XX Geographic immobility – – – X Industry inheritance – – – X Firm inheritance – – – X N 112931 112931 112931 112931 Note: Rank correlation between father earnings and son firm premium, controlling for mechanisms. Father earnings rank measured 1990–1999, son firm rank measured at age 38–42. Codes for education and occupation comprise 1,993 and 431 categories, respectively. Terms for referral, immobility, and inheritance include an interaction with father earnings rank. Human capital advantages account for roughly two-thirds of firm advantages by earnings background, with most of the remaining third accounted for by labor market referral and inheritance of opportunity structure. Daughters shown in Appendix Table A10. with the exception that industry inheritance is less common and concentrated to the top of the father’s earnings distribution (Appendix Figure A28). To test how these forms of structural overlap shape firm sorting, we regress the son’s firm component of earnings around age 40 on the father’s earnings rank, and partition the sample by geographic immobility, industry inheritance, and employer inheritance. For sons who find their main job in the same municipality, industry or firm as their father, the rank correlation between father earnings and the son’s firm component of earnings is 27%, 61%, and 115% stronger, respectively. When these groups are excluded, the coefficient is reduced by 12%, 17%, and 6% respectively, from the baseline of 0.15. Thus, geographic immobility and industry inheritance explain a substantial share of the sorting advantage. By contrast, firm inheritance is relatively inconsequential as this group is small enough that removing it from estimation does not substantially alter the association in the remaining population. 7.8 Multivariate results In Table 2, we bring the above explanations of sorting together in one model. Corresponding results for daughters appear in Appendix Table A10. As before, we regress the sons’s firm component of earnings around age 40 on the father’s earnings rank. We then add stepwise controls starting with education, occupation, and the worker component of earnings. We use the most detailed variables: 1,993 education codes and 431 occupation codes. For the subsequent variables reflecting referral networks, structural location, and firm inheritance we include both the main term and an interaction with father earnings rank. In this way, the remaining coefficient 31 captures persistence in the comparison group who do not experience job referral or structural persistence. Controlling for detailed education, occupation, and the worker component of earnings, together explains two thirds of the firm sorting advantage by social origin (69%). Adding controls for direct and indirect referral explains another tenth of the remaining association (11%), as does adjusting for structural location and firm inheritance (11%). Only a small proportion of the firm sorting advantage remains unexplained (9%). The proportions explained for daughters are similar, with 63%, 9%, and 12% in each step, leaving a somewhat larger proportion of the firm sorting advantage unexplained (17%) (Appendix Table A10).8These results reinforce those of our previous analysis: Privileged workers primarily secure positions at high-paying employers by means of their education and occupation, with referral networks and the inheritance of industry and labor market context playing a supplementary role. 7.9 Additional analyses and robustness Workplace premiums. Our data contain two types of identifiers related to the individual’s employment: one for firm/employer and one for workplaces. We use both for distinct purposes. In our earnings decomposition, we focus on employers for consistency with previous literature and as most theoretical mechanisms operate at this level. Conversely, we use workplace identifiers to define parental social networks, which are more plausible at this level. For completeness, we reassess our main results using workplace identifiers for the earnings decomposition in Appendix A3. The results assign a larger importance to workplace stratification than firm stratification, as would be expected given the finer granularity of this measure. Around 40% of the mature intergenerational correlation in earnings is explained by sorting between workplaces that offer unequal pay, compared to 29% for firms in our main analysis. Mothers’ earnings. Throughout our main analysis, we focus on fathers, typically the main breadwinners in the parent generation. Using family income would complicate the link with employers, as it involves multiple employers as well as sources beyond employment. Instead, we examine the role of mothers’ earnings separately in Appendix A4. This leads to lower intergenerational correlations for sons but not daughters, while the relative breakdown of earnings components remains similar. Essentially, both fathers and mothers contribute to earnings persistence overall and specifically to firm-based transmission. Appendix Table A5 shows moderate spousal correlations between fathers’ and mothers’ earnings, comparable in size to intergenerational correlations. This implies that mothers’ earnings contain important additional information not present in fathers’ earnings, highlighting the importance of gendered paths of transmission as a topic for future research. 8Partly, this might be due to us having operationalized several key mechanisms with reference to the father, whereas mothers may be more influential for daughters. 32 Worker-varying firm premiums. Our main analysis assumes firm premiums are uniform across workers, unaffected by individual characteristics. We also find no evidence that payoffs to firm premiums differ by social origin. However, firm premiums may vary along other dimensions, such as education and sex. In Appendix Figure A13, we address this by splitting each firm identifier into two separate categories: one for workers with tertiary education and another for those without. This approach, which treats highand low-educated workers within the same firm as belonging to different firms, more than doubles the firm component of earnings transmission. This suggests that firms do differentiate between worker types in the premiums they offer. However, interpreting these components as purely firm-level effects is difficult, as broader labor market pay differences by education become embedded in the firm component. By contrast, when firms are split by male and female workers, the decomposition results remain largely unchanged (see Appendix Figure A14).9 Time-varying firm premiums. Our main analysis assumes that firm premiums remain constant throughout the observation period, and are the same for parents and children. This is a strong assumption given evidence of shifting firm premiums over time (Criscuolo et al., 2020; Song et al., 2019; Wilmers and Aeppli, 2021). In Appendix Figure A15, we allow firm premiums to vary by reassigning firm identifiers every 5 years, treating each firm as a different unit in 1990–1994, 1995–1999, and so forth. The findings closely resemble those in our main results section. One explanation might be the high rate of firm turnover in our data, as evidenced in Appendix Figure A19. Turnover occurs not only due to businesses going bankrupt or closing down but also because of mergers, acquisitions, changes in ownership, or other types of restructuring. For these reasons, violations to the assumption of constant firm premiums appear mitigated by the limited life-span of firm identifiers. Network referral and worker rank. We have focused on mechanisms that can explain the matching of high parental earnings workers to high-paying firms. However, access to valuable social capital may influence both the firm premium and an individual’s rank within the firm. In Appendix A15, we therefore test the effect of job referral networks on individual worker rank. Doing so reveals that referral improves worker prospects through the rank achieved within a firm, and not just the average pay of the firm—albeit with somewhat less force, especially for weak ties. For sons, direct referral explains 8% of individual earnings transmission, as opposed to 10% of firm-based transmission. Indirect referral explains 4% of individual earnings transmission, as opposed to 14% of firm-based transmission. By contrast, geographic immobility, industry inheritance, and employer inheritance do little to explain the individual worker rank. These results suggest that gaining 9Since earnings rank is constructed within sex, age, and earnings year, labor market-wide pay differences by sex are excluded by design, unlike the education pay premium. 33 access to a firm through personal contacts, especially one’s parent, allows workers to fast-track their careers at the expense of co-workers.10 8Conclusion Much research in social stratification has focused on education, occupation, and upward mobility inside firms. But rising between-firm inequality has divided the economy between high-paying superstar firms and low-wage laggards. These labor market changes have been accompanied by empirical methods that identify unequal firm premiums by controlling for firms’ worker composition. Moreover, recent qualitative research emphasizes that firm sorting can be influenced by social background, whereby advantaged job applicants obtain entry to elite firms. In this paper, we clarify how inequality in firm pay premiums affects intergenerational mobility. Empirically, we find that the endowment of individual traits accounts for the majority of earnings transmission, consistent with the main thrust of stratification research. Yet, more than a quarter of earnings transmission occurs via inequality in firm premiums. Even by age 40, children from privileged origins continue to work in higher-paying firms than their less fortunate peers. This firm advantage reflects high-earning background workers enjoying employment at firms that both have higher value-added and also that share a larger portion of this value-added with workers. The main way that privileged workers sort to high-paying employers is through education and occupation. Network referral and inheritance of structural position in a local labor market or industry play supplementary, but smaller roles. The relative importance of these mechanisms may be specific to the Swedish context. Active labor market policies and robust labor unions contribute to a system that promotes formal job search methods over informal connections, distinguishing it from many other developed nations (Bachmann and Baumgarten, 2013). This makes us think that, if anything, Sweden is a “least-likely” case that provides a lower bound on the importance of contacts. While Sweden has an egalitarian system of education funding, careers are heavily stratified by field of study, something that begins in secondary school when family influence still looms large (Dahl et al., 2023). This early emphasis on specialization may influence the significance of educational field of study in employer sorting processes. In countries more heavily stratified by institution, school prestige may gain in importance and future work should study this issue with comparative data. It is important to clarify that the large earnings transmission component attributable to time-invariant individual traits does not necessarily reflect meritocracy or human capital. High earnings pay-offtraits could be individual skills, intelligence, personality or ability. But they also include consistent discrimination by gender or ethnicity and cultural capital that garners deference across multiple domains. Likewise, firm premiums can reflect marginal productivity, insofar as productivity is a 10Alternatively, they may indicate that employers use networks to identify high-quality candidates who would have earned similarly high wages in the absence of a referral. Further longitudinal research should distinguish these explanations. 34 function of peer effects or collaboration. Our focus is on distinguishing the role of firm premiums from the broad set of individual endowments and attributes, rather than distinguishing merit per se from unearned privilege. Notably, children from privileged origins enjoy a firm premium advantage regardless of the sources of parents’ earnings advantage. Parents with valuable individual characteristics support and cultivate their children to obtain jobs at high-paying firms, even when the parent did not work at such a firm. These results show how fragile “achieved” rewards in one generation can harden into “ascribed” or class-based advantages in the next. Parents who achieve a high individual standing by dint of luck or skill face a problem in how to pass on their advantage. For example, the skills that gained the parent their position may be costly to transfer, or turn out to be obsolete. Strategic knowledge about the education system or labor market, along with the mobilization of network ties, may be less costly ways to pass on this status. Educational attainment explains an important part, up to half, of firm sorting by social background. Meanwhile, firm premiums explain only about a quarter of the returns to education. However, firms explain a much larger portion of the intergenerational earnings advantage not mediated by education or occupation. This residual is of growing concern: as access to education and white-collar jobs reaches an increasing share of the population, privileged parents will be pressed for new means to reproduce their success. We show that firm sorting explains half or more of the direct effect of parent on child earnings after education and occupation is partitioned out. Firm premiums also explain most of the class origin gap in earnings that much recent scholarship has focused on (Laurison and Friedman, 2016). The firm-based approach that we pursue holds promise for understanding the role of networks in the reproduction of economic success. About a third of sons from homes above median income have shared a workplace with their father, and a similar proportion with a former colleague of the father. These referrals yield a payoff,notthroughpermanentemploymentbutratherasstepping-stonejobsonthe path to a lucrative career. Inheritance of a main employer is rare and surprisingly inconsequential. While parental referral is transient, children’s benefits in terms of firm sorting are lasting. These results are consistent with existing work on the returns to social capital over the life course (McDonald and Elder Jr, 2006). The role of firm premiums as a critical component of the apparent education premium should change how we think about education in the labor market. Educational advantage may be attributable in part to skills, but also to an advantage in firm sorting via networks, productivity complementarities and strategic knowledge. Awealthofqualitativeworkalreadytestifiestotheimportanceofthesemechanisms (Ashley, 2022; Ho, 2009; Rivera, 2016), yet quantitative stratification research lags behind with its focus on individuals over matching processes. A firm-based approach is a promising way forward. The use of new data sources and tools could identify pipelines between elite educational institutions and elite employers. Another implication of our work concerns how the distribution of firm premia shape intergenerational advantage across different segments of the economy. We have provided initial analysis down these lines, by showing that most industries recruit fairly evenly across social origins and inherited advantage appears consistent 35 across the firm premium distribution. Nevertheless, future research may identify labor markets in which firm-mediated advantage is weaker or interrupted altogether. Such research could also help distinguish the kinds of firms and the sources of firm pay premiums that provide mobility opportunities from the kinds that lock in advantages for already privileged workers. While trends such as skill segregation and superstar firms will amplify the social background returns to this process, policies of centralized bargaining and minimum wages could dampen them. Future research should also assess how the influence of firms on intergenerational mobility varies with the overall importance of firms in determining workers’ pay. Sweden is a conservative case here, as firm premiums account for a relatively small share of earnings inequality compared to other developed countries (Criscuolo et al., 2020). But, more broadly, across most rich countries, earnings inequality is increasingly between-firms (Tomaskovic-Devey et al., 2020), due to a combination of increased positive worker sorting across firms and persistently important firm pay premiums (Card et al., 2013; Song et al., 2019). In economies in which firms substantially determine worker earnings, such as the US, we expect firms’ role in intergenerational mobility to be amplified. This points to a research agenda on how comparative work and labor organization might shape the inheritance of advantage. Here our analysis connects to Relational Inequality Theory (Rosenfeld, 2021; Tomaskovic-Devey and Avent-Holt, 2019), which emphasizes the organizational and institutional determination of earnings inequality. While that work has uncovered determinants of pay-setting and of firm premiums that go beyond competitive market price determination, it has focused on explaining inequality associated with ascriptive characteristics like race, sex, and immigration status (Tomaskovic-Devey et al., 2015). It suggests that social closure and resource pooling across firms can be an important source of wage inequality. We add that firm-specific organizational and institutional deviations from competitive market wages can contribute to intergenerational earnings transmission. Beyond contributing to inequality, firm premiums affect intergenerational stratification. Since its origins in the 1960s, status attainment research has consistently focused on ways that high-status parents endow children with education and other individual resources valued in the labor market. Although this research program has proven extraordinarily productive, it has also obscured ways that social reproduction is influenced by structural features of the labor market. A strong undercurrent of stratification research has questioned this focus on the individual, hypothesizing labor market segmentation as a driving force behind intergenerational stratification (Baron, 1984; Baron and Bielby, 1980; Featherman, 1981; Spilerman, 1977; Stolzenberg, 1978). Recent availability of large-scale administrative microdata and new empirical methods make it possible for the first time to test this hypothesis. Our findings demonstrate the promise of incorporating firms and workplaces into an analysis of intergenerational mobility and should inspire further work toward this aim. 36 References Abowd, John M., Francis Kramarz, and David N. Margolis. 1999. “High Wage Workers and High Wage Firms.” Econometrica 67:251–333. Andrews, Martyn J, Len Gill, Thorsten Schank, and Richard Upward. 2008. “High wage workers and low wage firms: negative assortative matching or limited mobility bias?” Journal of the Royal Statistical Society: Series A (Statistics in Society) 171:673–697. Armstrong, Elizabeth A and Laura T Hamilton. 2013. Paying for the Party.Harvard University Press. Ashley, Louise. 2022. Highly Discriminating: Why the City Isn’t Fair and Diversity Doesn’t Work. Policy Press. Åslund, Olof, Cristina Bratu, Stefano Lombardi, and Anna Thoresson. 2021. “Firm productivity and immigrant-native earnings disparities.” IZA Discussion Paper No. 14960. Autor, David, David Dorn, Lawrence F Katz, Christina Patterson, and John Van Reenen. 2020. “The fall of the labor share and the rise of superstar firms.” Quarterly Journal of Economics 135:645–709. Bachmann, Ronald and Daniel Baumgarten. 2013. “How do the unemployed search for a job? – Evidence from the EU Labour Force Survey.” IZA Journal of European Labor Studies 2:22. Baron, James N. 1984. “Organizational perspectives on stratification.” Annual Review of Sociology 10:37–69. Baron, James N. and William T. Bielby. 1980. “Bringing the Firms Back in: Stratification, Segmentation, and the Organization of Work.” American Sociological Review 45:737–765. Beggs, John J and Wayne J Villemez. 2001. “Regional labor markets.” In Sourcebook of Labor Markets, edited by Ivar Berg and Arne L Kalleberg, pp. 503–529. Springer. Bernardi, Fabrizio and Gabriele Ballarino. 2016. Education, Occupation and Social Origin: A comparative analysis of the transmission of socio-economic inequalities. Edward Elgar Publishing. Bibb, Robert and William H Form. 1977. “The effects of industrial, occupational, and sex stratification on wages in blue-collar markets.” Social Forces 55:974–996. Bihagen, Erik. 2008. “Does Class Matter Equally for Men and Women? A Study of the Impact of Class on Wage Growth in Sweden 1999–2003.” Sociology 42:522– 540. 37 Bills, David B. 2003. “Credentials, signals, and screens: Explaining the relationship between schooling and job assignment.” Review of educational research 73:441– 469. Binder, Amy J, Daniel B Davis, and Nick Bloom. 2016. “Career funneling: How elite students learn to define and desire ‘prestigious’ jobs.” Sociology of Education 89:20–39. Björklund, Anders. 2000. “Going Different Ways: Labour Market Policy in Denmark and Sweden.” In Why Deregulate Labour Markets?, edited by Gøsta EspingAndersen and Marino Regini, pp. 148–180. New York: Oxford University Press. Black, Sandra and Paul Devereux. 2011. “Recent Developments in Intergenerational Mobility.” In Handbook of Labor Economics,editedbyOrleyAshenfelterand David Card, volume 4, pp. 1487–1541. North Holland Press, Elsevier. Blau, Peter M and Otis Dudley Duncan. 1967. The American occupational structure. New York: John Wiley & Sons. Bloome, Deirdre, Shauna Dyer, and Xiang Zhou. 2018. “Educational inequality, educational expansion, and intergenerational income persistence in the United States.” American Sociological Review 83:1215–1253. Borgen, Nicolai T and Arne Mastekaasa. 2018. “Horizontal stratification of higher education: the relative importance of field of study, institution, and department for candidates’ wages.” Social Forces 97:531–558. Bourdieu, Pierre. 1984. Distinction: A social critique of the judgement of taste. Harvard University Press. Bourdieu, Pierre. 2000. The Social Structures of the Economy.London:Polity. Bowles, Samuel, Herbert Gintis, and Melissa Osborne Groves. 2009. Unequal chances: Family background and economic success. Princeton University Press. Bratberg, Espen, Jonathan Davis, Bhashkar Mazumder, Martin Nybom, Daniel D Schnitzlein, and Kjell Vaage. 2017. “A comparison of intergenerational mobility curves in Germany, Norway, Sweden, and the US.” The Scandinavian Journal of Economics 119:72–101. Breen, Richard, Carina Mood, and Jan O Jonsson. 2016. “How much scope for a mobility paradox? The relationship between social and income mobility in Sweden.” Sociological Science 3:39. Breen, Richard and Walter Müller. 2020. Education and Intergenerational Social Mobility in Europe and the United States. Stanford University Press. Bryson, Alex and Rhys Davies. 2019. “Family, place and the intergenerational transmission of union membership.” British Journal of Industrial Relations 57:624– 650. 38 Bühlmann, Felix, François Schoenberger, Lena Ajdacic, and Fabien Foureault. 2022. “Elite recruitment in US finance: How university prestige is used to secure top executive positions.” British Journal of Sociology 73:667–684. Burawoy, Michael and Erik Olin Wright. 1990. “Coercion and consent in contested exchange.” Politics & Society 18:251–266. Card, David, Ana Rute Cardoso, Joerg Heining, and Patrick Kline. 2018. “Firms and labor market inequality: Evidence and some theory.” Journal of Labor Economics 36:S13–S70. Card, David, Jörg Heining, and Patrick Kline. 2013. “Workplace Heterogeneity and the Rise of West German Wage Inequality.” Quarterly Journal of Economics 128:967–1015. Carlsson, Gosta, Robert Erikson, Carl Löfwall, and Bo Wärneryd. 1974. “Socioekonomiska grupperingar.” Statistisk tidskrift 5:381. Carniero, P, Sarah Cattan, Lorraine Dearden, Laura van der Erve, Sonya Krutikova, and Lindsey Macmillan. 2020. “The long shadow of deprivation: Differences in opportunities across England.” Social Mobility Commission Research Report. Castilla, Emilio J, George J Lan, and Ben A Rissing. 2013. “Social networks and employment.” Sociology Compass 7:999–1026. Cheng, Siwei and Xi Song. 2019. “Linked Lives, Linked Trajectories: Intergenerational Association of Intragenerational Income Mobility.” American Sociological Review 84:1037–1068. Cobb, J Adam and Ken-Hou Lin. 2017. “Growing apart: The changing firm-size wage premium and its inequality consequences.” Organization Science 28:429– 446. Collins, Randall. 1979. The credential society: An historical sociology of education and stratification. Columbia University Press. Corak, Miles. 2013. “Income inequality, equality of opportunity, and intergenerational mobility.” Journal of Economic Perspectives 27:79–102. Corak, Miles and Patrizio Piraino. 2011. “The Intergenerational Transmission of Employers.” Journal of Labor Economics 29:37–68. Cornelissen, Thomas, Christian Dustmann, and Uta Schönberg. 2017. “Peer effects in the workplace.” American Economic Review 107:425–56. Correia, Sergio. 2019. “REGHDFE: Stata module to perform linear or instrumentalvariable regression absorbing any number of high-dimensional fixed effects.” Correll, Shelley J, Cecilia L Ridgeway, Ezra W Zuckerman, Sharon Jank, Sara Jordan-Bloch, and Sandra Nakagawa. 2017. “It’s the conventional thought that counts: How third-order inference produces status advantage.” American Sociological Review 82:297–327. 39 Criscuolo, Chiara, Alexander Hijzen, Cyrille Schwellnus, Erling Barth, Wen-Hao Chen, Richard Fabling, Priscilla Fialho, Balazs Stadler, Richard Upward, Wouter Zwysen, Katarzyna Grabska-Romagosa, Ryo Kambayashi, Timo Leidecker, Oskar Nordström Skans, Capucine Riom, and Duncan Roth. 2020. “Workforce composition, productivity and pay: the role of firms in wage inequality.” OECD Economics Department Working Papers. Dahl, Gordon B, Dan-Olof Rooth, and Anders Stenberg. 2023. “High school majors and future earnings.” American Economic Journal: Applied Economics 15:351– 382. Dahl, Molly W and Thomas DeLeire. 2008. “The association between children’s earnings and fathers’ lifetime earnings: estimates using administrative data.” University of Wisconsin-Madison, Institute for Research on Poverty. DiMaggio, Paul and Filiz Garip. 2012. “Network effects and social inequality.” Annual review of sociology 38:93–118. Dobbin, Caue and Tom Zohar. 2023. “Quantifying the Role of Firms in Intergenerational Mobility.” CESifo Working Paper No. 10758. Domina, Thurston, Andrew Penner, and Emily Penner. 2017. “Categorical inequality: Schools as sorting machines.” Annual Review of Sociology 43:311–330. Duncan, Otis Dudley. 1966. “Path analysis: Sociological examples.” American Journal of Sociology 72:1–16. Engbom, Niklas and Christian Moser. 2017. “Returns to education through access to higher-paying firms: Evidence from US matched employer-employee data.” American Economic Review 107:374–78. Engbom, Niklas, Christian Moser, and Jan Sauermann. 2023. “Firm pay dynamics.” Journal of Econometrics 233:396–423. Engzell, Per and Carina Mood. 2023. “Understanding Patterns and Trends in Income Mobility through Multiverse Analysis.” American Sociological Review 88:600–626. Erikson, Robert and John H Goldthorpe. 1992. The constant flux: A study of class mobility in industrial societies. Oxford University Press. Erikson, Robert and Jan O. Jonsson. 1996. Can Education Be Equalized? The Swedish Case in Comparative Perspective. Boulder, CO.: Westview Press. Farkas, George. 2003. “Cognitive skills and noncognitive traits and behaviors in stratification processes.” Annual Review of Sociology 29:541–562. Featherman, David L. 1981. “Social stratification and mobility: two decades of cumulative social science.” American Behavioral Scientist 24:364–385. 40 Williamson, Oliver E. 1975. Markets and Hierarchies: analysis and antitrust implications: a study in the economics of internal organization. The Free Press. Wilmers, Nathan. 2018. “Wage Stagnation and Buyer Power: How Buyer-Supplier Relations Affect U.S. Workers’ Wages, 1978 to 2014.” American Sociological Review 83:213–242. Wilmers, Nathan. 2019. “Solidarity within and across workplaces: How crossworkplace coordination affects earnings inequality.” RSF: The Russell Sage Foundation Journal of the Social Sciences 5:190–215. Wilmers, Nathan and Clem Aeppli. 2021. “Consolidated advantage: new organizational dynamics of wage inequality.” American Sociological Review 86:1100–1130. Witteveen, Dirk and Paul Attewell. 2017. “Family background and earnings inequality among college graduates.” Social Forces 95:1539–1576. Witteveen, Dirk and Paul Attewell. 2020. “Reconsidering the ‘meritocratic power of a college degree’.” Research in Social Stratification and Mobility 66:100479. Zhou, Xiang. 2019. “Equalization or selection? Reassessing the “meritocratic power” of a college degree in intergenerational income mobility.” American Sociological Review 84:459–485. Zimmerman, Seth D. 2019. “Elite colleges and upward mobility to top jobs and top incomes.” American Economic Review 109:1–47. Zwysen, Wouter. 2016. “Crowding out of disadvantaged young adults in Germany: background matters depending on local labour market.” European Sociological Review 32:662–674. 47