Firms and Workplaces As a Nexus for Intergenerational Inequality Processes
Firms and their workplaces is the nexus of labor market activities: it is where individuals work to produce services or commodities, it is where they gain experience to enhance their productivity, it is where decisions about their hiring, promotion or lay off are taken, and to large and increasing extent also where wages are set and negotiated, especially for white collar workers. Firms and workplaces are also a foci (Feld 1981) for group dynamics, where actor form alliances to lay claims on resources (Tilly 1999; Tomaskovic-Devey, Hällsten and Avent-Holt 2015). However, most research on inequality and labor market processes has tended to focus on individuals as the unit of analysis, and ignored how they are clustered in these social foci (Lazear and Shaw 2009). The main lessons to be learned is that inequality cannot necessarily be reduced to single parameters of gender wage gaps, immigrant-native wage gaps, class origin advantages or returns human captain and so forth. Swedish matched employer-employee register data offers an exceptional opportunity to further our understandings of these processes. Recent advancements utilizing these data has been made in order to understand wage distributions (Skans Nordström, Edin and Holmlund 2009), gender processes (Hensvik 2014), immigrant-native recruitment (Åslund, Hensvik and Nordström Skans 2014), and immigrant-native wage setting processes (Tomaskovic-Devey, Hällsten and Avent-Holt 2015).
The current project seeks to analyze how firms and workplaces contribute to social mobility processes, i.e., to the advantages or disadvantages an individual’s family background, and the resources associated with it, generates for the individual’s labor market outcomes. Studies of social mobility in both sociology and economics tend to find that individuals labor market outcomes are strongly structured by family background (Björklund and Jäntti 2012): inequality in one generation is to a large extent transferred to the next generation. Individuals of lower ranked class origins tend to earn substantially less than those of more advantaged class origins when one compares individuals with similar or identical education (Hansen 2001; Mastekaasa 2011; Hällsten 2013).
There are many theoretical explanations of these findings, ranging from deliberate actions, for example in social closure and exclusionary practices, to hard-to-observe advantageous characteristics that individuals carry and get rewarded more or less unconsciously (Bowles and Gintis 1976; Groves 2005; Jackson, Goldthorpe and Mills 2005). The focus in this project will be on how social closure practices contribute to social mobility. Relational inequality theory (Tilly 1999; Avent-Holt and Tomaskovic-Devey 2010; Tomaskovic-Devey, Hällsten and Avent-Holt 2015) suggest that social groups interact in workplaces to lay claims on organizations resources such as positions and wages. How successful such strategies can depend on context, for example the extent of formalized personnel policy, merit evaluation processes, how work is organized, wage setting practices, but also the organizations surrounding context. An origin that is shared within a closed social circle may also spur niche competition, and so advantage can become a burden. Some prior research has identified variations in class origin effects across labor market sectors (Hansen 2001), and it seems concentrated to the private competitive parts of the economy, with important variations across occupations and sectors (Hällsten 2013). This project seeks to ratchet these inequalities down to the level of firms and workplaces.
The research questions of the current proposal is to estimate effects of this segregation and social composition, i.e., the effects of social origin composition at different organizational levels for for recruitment, wage setting, and promotion of employees. For example: is it beneficial to have service class origin in an organization with a high representation of this origin? Is it easier to get recruited? Does the same opposite hold for working class origins?
The project will use Swedish population level registers from 1960 until the present (to this date: data exists up until 2012). With the censuses 1960 to 1990, the occupations register from 2001 and onwards, and the multigenerational register which contain parent-child links for large parts of the population born after 1932 it is possible to code occupation or occupation based measures (such as social class or occupational prestige) of parents for 75-85 percent of the working population, depending on birth cohort. The composition of the workforce’s social origin is thus identifiable for a large part of the population.
The firms and workplaces can be identified using their organization or establishment identity number, which is linked to each of its employees. Mergers and splits, which may create errors in these variables can be handled by analyzing labor force streams across firms and workplaces in adjacent year pairs (Statistis Sweden 2015). This information will be used to measure the composition in workplaces. For the individuals, information on class origin is collected from the censuses and the occupation registers as described above. Other important information comes from the education and schooling registers, administrative data on earnings and incomes. The military enlistment register provides important on cognitive ability and some personality characteristics, which are important for social mobility processes (Mood, Jonsson and Bihagen 2013). This data is limited to men in certain cohorts, but using information on uncles or brothers one may also include women the analyses (Grönqvist, Öckert and Vlachos 2010).
The project will use regression-type methods to analyze the composition effects. One important caveat is that composition effects may reflect unobserved characteristics of workers—or firms (Abowd and Kramarz 1999; Abowd, Kramarz and Margolis 1999). It is therefore essential to employ elaborate control strategies. Recent advancement allows the estimation two highly dimensional fixed effects (Abowd, Kramarz and Margolis 1999; Ouazad 2007; Cornelissen 2008); which be either the individual or the family in one dimension, and firm or workplace in the other. To the extent that characteristics of individual and workplace are stable over, or across siblings (for the family fixed effects), this selectivity of unobserved characteristics of workers or firms can be controlled for.