Dual Embeddedness: Informal Job Matching and Labor Market ...

4 downloads 7565 Views 288KB Size Report
Aug 7, 2013 - nonsearchers receive their information through informal social ... pling of education and job training systems in CMEs (particularly in Germany) ..... training is defined as having received a degree through the vocational educa-.
Dual Embeddedness: Informal Job Matching and Labor Market Institutions in the United States and Germany Steve McDonald, Richard A. Benton, David F. Warner

Social Forces, Volume 91, Number 1, September 2012, pp. 75-97 (Article)

Published by Oxford University Press

For additional information about this article http://muse.jhu.edu/journals/sof/summary/v091/91.1.mcdonald.html

Access provided by North Carolina State University (7 Aug 2013 12:34 GMT)

Dual Embeddedness

Dual Embeddedness: Informal Job Matching and Labor Market Institutions in the United States and Germany Steve McDonald, North Carolina State University Richard A. Benton, North Carolina State University David F. Warner, University of Nebraska – Lincoln

D

rawing on the embeddedness, varieties of capitalism and macrosociological life course perspectives, we examine how institutional arrangements affect network-based job finding behaviors in the United States and Germany. Analysis of cross-national survey data reveals that informal job matching is highly clustered among specific types of individuals and firms in the United States, whereas it is more ubiquitous in Germany. These differences are linked to (1. loosely regulated and hierarchical employment relations in the United States that facilitate network dominance in specific economic sectors and (2. coordinated market relations, tight employment regulations and extensive social insurance system in Germany that generate opportunities for informal matching but limit the influence of network behavior on employment characteristics. These findings illustrate how social institutions shape access to economic resources through network relations. Sociologists have long recognized the importance of personal contacts in job allocation processes, in terms of both workers gaining access to information about openings (Granovetter [1974]1995; Royster 2003) and employers seeking to fill vacancies through network-based ties (Fernandez et al. 2000). Informal job matches are indeed quite distinct from formal matches in terms of the personal characteristics of the applicants as well as the types of jobs into which they are hired (for a review, see Marsden and Gorman 2001). At the same time, these actors (both individuals and firms) are embedded in a broader set of economic institutions. The frequency, character and ­consequences of informal job matching are likely to be highly contingent on the national institutional context.

Dale Dannefer, Mark Nance, and Anna Manzoni provided detailed comments on earlier versions of the paper. We also thank Arne Kalleberg and the anonymous Social Forces reviewers for their careful consideration and feedback, which no doubt helped to improve the paper. This research was supported by funds from North Carolina State University’s Faculty Research and Professional Development Grant. An earlier version of this paper was presented at the 2010 meeting of the Southern Sociological Society. © The Author 2012. Published by Oxford University Press on behalf of the University of North Carolina at Chapel Hill. All rights reserved. For permissions, please email: [email protected].

Social Forces 91(1) 75–97, September 2012 doi: 10.1093/sf/sos069 Advance Access publication on 19 July 2012

75

76 Social Forces 91(1)

Few studies, however, have analyzed the interplay between economic institutions and informal job matching in a comparative framework and thus have missed the opportunity to demonstrate how institutional arrangements influence the extent and character of network-based job allocation processes. Accordingly, the present study compares job matching in the United States and Germany, using survey data from the National Longitudinal Survey of Youth (NLSY) and the German Socio-Economic Panel (GSOEP). These two countries are ideal candidates for comparison because each maintains a distinct set of economic institutions, with more flexible and loosely regulated employment relations in the United States compared with the tighter institutional control of the German labor market. We draw from three distinct, but compatible, theoretical approaches to understand how both formal and informal economic actions are conditioned by institutional contexts. First, the new economic sociological literature on embeddedness describes how economic behavior is embedded within networks of social relations (Granovetter 1985) and how economic processes are influenced by societal institutions (Polanyi [1957]2001). However, most of the empirical research that has emerged from this literature has focused either on the role of social networks in microlevel economic phenomena or on the interplay between social institutions and economic processes (Krippner and Alvarez 2007). Recently, researchers have begun to examine the ways that political and cultural institutions shape economic actions – a concept that Baker and Faulkner (2009) have referred to as double embeddedness. Here, we expand on this approach by focusing on how economic institutions structure the ways that individuals and firms engage in network-based market behavior – in particular, informal job finding. We illustrate how economic action is dually embedded, unfolding in networks of social relations that are themselves set in specific national, social and economic institutional arrangements. Second, we draw on the “varieties of capitalism” literature (Hall and Soskice 2001). This perspective recognizes that economic behavior is influenced by the presence of distinct configurations of capitalist institutions, with actions guided primarily by competition in some countries and by strategic collaboration in others. Institutions not only constrain action but also provide resources for particular types of action. Finally, we utilize a macrosociological life course perspective (Mayer 2009; Settersten 2003), which brings together the themes highlighted in the literatures on embeddedness and varieties of capitalism. Institutional arrangements structure networks of relations – which supply specific and often limited resources to individuals (Dannefer 1992) – both creating and constraining opportunities for individual action. Accordingly, life course researchers emphasize the need for cross-national comparisons to examine how specific institutions and policies organize life course processes (Mayer 2009). For example, research on the long-term consequences of unemployment across societies reveals that differences in institutional environments can have an enduring impact on careers (Gangl 2004, 2006). Along these same lines, we view job matching as a life course process that involves agency (on the part of ­workers and firms alike) within the context of specific institutional arrangements and

Dual Embeddedness

structural conditions. These institutions and structures shape the extent to which agency is possible (Settersten 2003). These three theoretical orientations offer complementary insights into our understanding of the interplay between economic institutions and market behaviors across social contexts. By synthesizing these approaches, we predict that variations in labor market regulation, links between training and employment and welfare state policies influence the ways in which workers seek out and are allocated to labor market positions. Few researchers have attempted to make direct comparisons of informal job matching across countries. The crossnational evidence identifies differences in the frequency of informal matches across countries, but finds few differences in the types of individuals that engage in informal searches or the kinds of jobs that they enter informally (De Graaf and Flap 1988). These null findings reflect the failure to examine both the supply side and the demand side of informal job matches. Recent research in the United States has explored the important distinction between the two key types of informal job matching: (1. informal job seeking (supply side) and (2. informal recruitment (demand side). By virtue of being informally recruited into new jobs, nonsearchers – people who change jobs without engaging in a job search – are fundamentally distinct from active informal job seekers, in terms of both personal and career characteristics and in the kinds of labor market positions into which they are hired (Kmec et al. 2010; McDonald and Elder 2006). Cross-national differences in regulatory, training and economic entitlement environments should therefore alter the prevalence, processes and employment characteristics associated with informal job searching and nonsearching.

Prior Research on Informal Job Matching Informal job matching has received considerable scholarly attention (Marsden and Gorman 2001). Informal hiring is highly clustered in the United States, as the search for jobs through personal connections has been linked to a variety of characteristics, including gender, race, education, work experience, occupation, and wages (e.g., Mouw 2003; Smith 2000). Most research on informality in the labor market has focused on supply-side job searching behaviors of individuals (cf., Granovetter [1974]1995). Recent research on nonsearching has generated renewed interest in informal recruitment processes. Nonsearching occurs when people change jobs without engaging in a job search.1 While some nonsearches result from formal recruitment practices such as the use of executive search consultants (Finlay and Coverdill 2002), in most instances (about 83 percent) nonsearchers receive their information through informal social interaction (Granovetter [1974]1995; Hanson and Pratt 1991; McDonald 2010). Therefore, nonsearching is an informal recruitment process whereby individuals are not actively looking for work but receive unsolicited information about job opportunities from their personal connections (Granovetter [1974]1995). Research on nonsearching emphasizes demand-side informal recruitment ­practices (both in its conceptualization and operationalization)2 and therefore balances

77

78 Social Forces 91(1)

the individualist perspective on job allocation apparent in job search theory in economics (Devine and Kiefer 1991) and status attainment research (Kerckhoff 1976), offering a more complete picture of informality in the labor market. Prior research indicates that nonsearching is common, particularly among individuals who enter high-wage, managerial jobs (Elliott 2000, Granovetter [1974]1995; McDonald and Elder 2006). The extent to which this feature of the U.S. labor market extends to other societies remains uncertain. International evidence on job matching comes primarily from studies of individual countries without making cross-national comparisons (e.g., Bian 1997; Chua 2011; Gerber and Mayorova 2010). Only three studies have compared informal job matching across countries with distinct institutional environments. De Graaf and Flap (1988) found that personal contacts were used most frequently in the United States, followed by Germany, and then the Netherlands. Franzen and Hangartner (2006) analyzed data from an international survey of individuals in 28 different countries. Consistent with the results from the earlier study, American workers were more likely than German workers to have found jobs through their personal connections. Finally, in his study of job searching across 14 European countries and the United States, Pellizzari (2010) found informal searching to be more prevalent in Germany than in the United States – the opposite of what was found in the prior studies. Altogether these studies found few cross-national differences in the employment characteristics associated with informal job matching. Overall, these studies do not resolve the question of which contexts tend to maintain more informal activity. Also, there is little evidence of divergent outcomes associated with informality in the United States and German contexts. We see two main problems with the international evidence. First, cross-national research has not yet distinguished between informal search and informal recruitment. The surveys used in De Graaf and Flap (1988) as well as Franzen and Hangartner (2006) did not ask respondents whether they had engaged in an active search. It remains likely that many of the informally recruited respondents were included in the informal search category for the analyses. However, when asked about how they found their jobs, nonsearchers may have selected response categories that were treated as formal searches – such as applying directly to firms, having been contacted by an employer or providing a response of “other search.” Alternatively, nonsearchers appear to have been included as the reference category to informal job seekers in Pellizzari (2010). Failing to distinguish between formal search, informal search and nonsearch may obscure the role of informality in the labor markets and has perhaps contributed to the inconsistent research findings. Second, these prior studies lack the requisite theoretical perspective to make a more comprehensive assessment of why one would expect to see differences in informal labor market activity in distinct institutional environments.

Varieties of Capitalism and Institutional Arrangements The varieties of capitalism approach distinguishes between coordinated market economies (CMEs) such as Germany and liberal market economies (LMEs)

Dual Embeddedness

such as the United States. First, CMEs manage economic action through strategic collaboration and nonmarket arrangements such as extensive networks of firms and actors. By contrast, LMEs organize their economic behavior primarily through competitive markets driven by supply and demand equilibria (Hall and Soskice 2001). Second, stronger labor market regulations and union presence in CMEs restricts employers’ options for dealing with exogenous shifts in supply and demand for goods and services. Firms in LMEs have a more flexible set of options for dealing with labor. Third, there is tendency toward tighter coupling of education and job training systems in CMEs (particularly in Germany) than in LMEs. Fourth, CMEs offer greater security for employed workers (through extended employment contracts) and unemployed workers (through generous unemployment insurance), whereas worker protections in LMEs are more circumscribed. Finally, conservative economic policies in some CMEs (like Germany) create incentives for preserving the single-earner family model, whereas LMEs tend to have greater heterogeneity in work-family relations. These institutional differences provide a distinct set of resources and constraints for economic actors, influencing the ways that life events (such as job changes) are experienced (Mayer 2009). They also structure the opportunities for and consequences of informal hiring. Below we offer a series of expectations regarding how these two institutional contexts might affect informal job matching (see Table 1 for a summary).

Across Country Differences in Informal Job Matching A careful read of the evidence might lead one to expect greater informality in the United States,3 but the arguments developed in the context of the varieties of capitalism literature lead us to consider the alternative possibility: that informal job matching is likely to be more common in Germany than in the United States. The coordination of economic activities in Germany institutionalizes the creation and maintenance of network relationships (Hall and Soskice 2001). For example, the works councils – required in German firms with more than five permanent employees – promote internal efficiency through information exchange and consultation (Freeman and Lazear 1995). These institutional features facilitate “niche production” among firms rather than direct competition, which further promotes inter-firm collaboration by encouraging the sharing of information (Hall and Soskice 2001). These market-based relationships regulate the employee/employer affiliation as well. The United States has lower labor union density than Germany (12.5% compared with 27%, respectively; Visser 2006), which leads to greater unilateral control by top management. The coordination of economic activity should facilitate the informal flow of information between workers and employers. These institutional arrangements set the stage for informal job finding by fostering social relationships and creating a context conducive to interpersonal interaction (for example, see Small 2009). Alternatively, LMEs foster more competitive relations among actors. Firm interrelationships in the United States tend to be based on market relationships and enforceable contracts, which may lead to greater distrust and social closure in

79

80 Social Forces 91(1)

Table 1: ​Summary of Study Expectations and Rationale Across-Country Differences In Germany, market coordination facilitates network connections in job matching processes. In the U.S., market competition increases reliance on formal job search processes. •  ​  ​Informal job matching will be more common among German workers than among American workers. Extensive social safety net in Germany reduces the economic urgency to find new jobs, leading to more nonsearching. Lack of social safety net in the United States generates economic urgency to find new jobs, leading to more active job searching. •  ​  Nonsearching will be greater among German workers than among American workers. Within-Country Differences Family Conservative German policies and institutions decommodify family labor and promote single-earner family arrangements. Due to a lack of family-based economic policies and incentives, U.S. employment decision making is sensitive to shifts in family arrangements. •  ​  ​In the United States, nonsearching will be more likely among married individuals and parents. •  ​  ​In Germany, family characteristics will not be associated with informal job matching. Education Tight coupling between education and job training in Germany effectively signals job qualifications. General skills imparted in the United States hinder the signaling of job qualifications (particularly among poorly educated workers) leading to a greater reliance on informal hiring. •  ​  ​In the United States, informal job matching will be more likely among poorly educated workers. •  ​  ​In Germany, informal job matching will not be associated with education. Employment Tight regulatory control and employment protection keeps informality from dominating segments of the German labor market. Lack of regulatory control in the United States and uncertainty about qualifications leads to social closure and clustering of informal job matching activity in specific sectors of the labor market. •  ​  ​In the United States, informal job searching will be most common among workers in blue collar occupations, while nonsearching will be most common among workers in managerial occupations and will be associated with higher wages. •  ​  ​In Germany, informal job matching will not be associated with type of occupation or with hourly wages.

Dual Embeddedness

network relationships in the United States (cf. Kanter 1977). This is not to suggest that informal searching and matching activity is impossible or even unlikely in competitive environments (for example, see Fernandez et al. 2000), only that these activities are perhaps more common in coordinated markets. The prevalence of different types of informal matches also likely varies across the two countries. Compared with United States workers, German workers have substantially greater security, with long-term employment agreements (“standard indefinite contracts”) and legislation restricting contracting and setting explicit guidelines for termination that protect job security (Gangl 2002, 2006; Kurz et al. 2006). Also, the United States offers a less extensive set of unemployment transfers than does Germany (Gangl 2004; McManus and DiPrete 2000), which may further contribute to insecurity among American workers. As such, both the likelihood of (DiPrete 2002) and the long-term negative consequences associated with job dislocation (Gangl 2004) are greater in the United States. The elevated job insecurity among U.S. workers is likely to lead to more active job searches and fewer nonsearches. For example, recent evidence suggests that perceptions of job insecurity reduce organizational commitment and increase turnover intentions (Sora et al. 2010). Overall, German workers should be more likely than American workers to change jobs via the nonsearch process, due to the strength of the worker protections (which reduce the economic urgency to search for jobs) and the cooperative labor market institutions (which assist in fostering informal connections among and between workers and employers).

Within Country Differences in Informal Job Matching In general, given their different institutional arrangements, we expect informal job matching to be less commonly associated with personal and career factors in Germany than in the United States. The greater flexibility and weaker regulation in the United States creates an environment with larger disparities in job rewards and job security (Kalleberg 2009). This situation generates opportunities for informal market activities to dominate among certain kinds of workers and among certain sectors of the labor market. We therefore anticipate greater clustering of informal job matching behavior in the United States. By contrast, the tighter institutional control over the German labor market likely leads informal matching to be distributed more randomly across individuals and firms. Below we offer several explicit expectations concerning the patterning of informal job matching in these two institutional contexts.

Family Characteristics Conservative family institutions in Germany create incentives for preserving gendered work patterns and the single-earner family model. Germany’s welfare state reinforces women’s primary responsibility for childrearing and housework through the gendered German labor market (Buchholz and Grunow 2006; Esping-Andersen 1999), delegating unpaid family labor more often to German women (Esping-Andersen 1999). For example, childcare availability is limited for children younger than 3 years old, yet Germans have relatively generous

81

82 Social Forces 91(1)

state-sponsored parental leave (Buchholz and Grunow 2006; Kurz et al. 2006), which helps to decommodify family labor and reduce female labor force participation (Drobnicˇ et al. 1999; Esping-Andersen 1990). Alternatively, the lack of institutional supports in the United States affects how people are connected to jobs, providing few direct supports to working-age individuals based on marriage, parental status or employment (health insurance and Social Security are notable exceptions). The economic decision making of American workers is therefore likely more sensitive to shifts in family relations and obligations. Unmarried U.S. workers may also be more likely to engage in an active job search as they have a greater economic urgency to enter the labor market and to change jobs in an effort to secure greater economic rewards. Parents should be less likely than childless individuals to engage in an active search. Prior research finds higher rates of nonsearching among individuals who have exited the labor market for child care reasons in the United States (McDonald 2005). By contrast, marital and parental statuses should play a minimal role in influencing the informal ways that Germans find their jobs, as generous family supports offer few economic incentives for reentry into the labor market.

Education Germany has extensive vocational training that combines work with education (Blossfeld and Stockmann 1998/1999; Kurz et al. 2006) and promotes the development of firm-specific skill sets. By contrast, education and training systems in the United States impart more general skills (Kerckhoff 2003). Employers in the United States, therefore, tend to be skeptical of educational credentials as signals of job relevant skills (Miller and Rosenbaum 1997). Because the German system ensures that even poorly credentialed workers have received vocational training, German workers are more quickly integrated into the labor force than American workers (Blossfeld and Stockmann 1998/1999; DiPrete et al. 1997). To compensate for a lack of information about training, informal hiring should be more common among lesser educated workers in the United States, whereas education should be unrelated to informal hiring in Germany.

Employment In the United States, blue-collar occupations such as machine operators and laborers require lower levels of training. Also, as mentioned above, hiring authorities in the United States rely heavily on noneducational criteria (such as network connections) for evaluating job candidates in low-wage and bluecollar occupations (Holzer 1996; Moss and Tilly 1996). At the same time, this informality in hiring extends only to informal job searching, not to informal recruitment through the nonsearch process (Elliott 2000; McDonald and Elder 2006). Employers in blue-collar occupations often hire candidates through job referrals, but rarely expend the effort needed to informally recruit candidates. Consequently, informal job searching in the United States should be more common among workers in blue-collar occupations.

Dual Embeddedness

Employers in the United States labor market have a history of engaging in social closure practices to limit access to high-wage positions (Stainback et al. 2010). Informal recruitment is an effective way in which economic actors can hoard opportunities for in-group members (cf. Elliott 2000; Kanter 1977). Thus, gaining access to high-wage jobs often involves being recruited through informal connections. While most people can ask around among their friends and colleagues for assistance in finding a job, only a select few can receive job-finding help without seeking it out (Granovetter’s (1974[1995]). People who receive unsolicited job leads through routine social interactions have access to more high-status contacts in their occupational networks than individuals who do not receive unsolicited job leads (Lin and Ao 2008). Consequently, nonsearching is positively associated with wages, while informal job searching is not (McDonald and Elder 2006; Mouw 2003). Yet not all high-wage occupations in the United States exhibit high levels of informality. Informal hiring is relatively uncommon among professional and technical occupations because of certification requirements that offer appropriate skill signals (Holzer 1996). Executive and managerial skills, however, are more difficult to assess, since there are fewer explicit certification requirements (Kanter 1977). Employers often attempt to bridge this uncertainty by emphasizing candidates’ reputation and network connections/ affiliations. Thus, nonsearching should be more common among workers in managerial positions and associated with higher wages in the United States. We expect less clustering of informal hiring by occupation and wages in Germany. First, extensive training and apprenticeship systems in Germany reduce uncertainty involved in assessing the skills of job applicants and the need to rely on informal signals of merit. Therefore, there is little reason to suspect that informality would vary significantly across occupational categories. Second, German employment protection legislation leads to a more risk-averse hiring system. The costs of hiring someone who does not fit or perform well are greater in Germany due to the long-term employment contracts. Consequently, German firms should be less willing than U.S. firms to commit high salaries to informally matched workers. The greater flexibility enjoyed by U.S. firms reduces the costs of a poor match, making them more likely to offer lucrative wages to well-connected workers. Moreover, given collective bargaining arrangements, German salaries tend to be tied more to differences in job tenure than to other employee characteristics.

Data and Methods To test these propositions, we use survey data from two sources. The U.S. data are from the NLSY, a nationally representative panel survey of 12,686 individuals aged 14-22 years in 1979. Follow-up interviews were conducted annually until 1994 and biennially thereafter. The German data come from the GSOEP – a nationally representative survey that began in 1984 and has followed 5,921 households annually, with replacement samples added at various years since the survey’s inception. Both datasets contain detailed information on job matching. However, several distinct design features of the two datasets required ­several

83

84 Social Forces 91(1)

restrictions to ensure comparability. First, the NLSY only collected detailed information on job matching between 1994 and 2000. Therefore, we restricted the analyses to this time period. Second, the GSOEP contains a wider age-range of respondents than the NLSY, so we restricted our analyses to the ages of respondents in the NLSY (29-37 in 1994 and 35-43 in 2000). Third, the GSOEP is a household survey and therefore contains information from both the household head and from his/her spouse/partner (when applicable). By contrast, the NLSY is a survey of individuals; household members were randomly selected for inclusion into the sample. To account for this difference, we randomly selected a single member from multiperson GSOEP households for inclusion in the analysis. Fourth, NLSY respondents report on up to five jobs each survey wave, whereas GSOEP respondents report only on their primary jobs, so we focused only on respondents’ primary jobs. Fifth, while the GSOEP ascertained both internal and external job changes (intra-firm and inter-firm mobility), the NLSY asked only about transitions to different employers. The analysis was therefore limited to external job changes. We arranged the data in a person-period format where individual respondents contribute an observation for each unique primary job from 1994-2000. We began with a sample of 8,492 individuals who contributed 15,066 jobs in the NLSY and 1,323 individuals who contributed 1,983 jobs in the GSEOP. To be included in the analysis, cases had to meet several additional criteria. First, since the focus of this analysis was on the different matching methods, we excluded individuals self-employed or employed in family businesses because such job matches are distinctive in that they were created by respondents or by their family members. This excluded 1,275 jobs in the United States and 203 jobs in Germany. Second, we used listwise deletion to eliminate cases that lacked information on any of the variables used in the analysis (excluding 1,397 U.S. and 145 German jobs).4 These refinements resulted in an analytic sample of 7,392 individuals and 12,394 jobs in the NLSY data and 1,126 individuals and 1,635 jobs in the GSOEP data. As such, U.S. respondents have an average of 1.68 jobs, whereas the average is 1.45 for German respondents – indicative of the lower rates of inter-firm mobility in the German labor market. Table 2 shows the descriptive statistics for the analytic samples.

Measures The dependent variable is a categorical indicator of three ways that respondents were matched to their current jobs. In the United States, respondents were asked, “Were you looking for work when you were offered this job?” We classified respondents who said no as having secured their jobs through the nonsearch process. Respondents who said yes were then asked, “Which of the [following] methods. . .led to your being offered your job?” We considered respondents who stated that they had “contacted friends or relatives” as having secured their job through an informal search. The remaining respondents used a formal search, which includes the use of employment agencies, searching the want ads and applying directly to employers. The German survey asked similar questions that we used to provide a comparable set of categories. Respondents were asked,

Dual Embeddedness

Table 2: ​Study Variables and Descriptive Statistics From U.S. and German Samples United States Variable

Mean

Germany

SD

Mean

SD

Dependent variables  ​ ​Formal search (reference category)

0.52



0.43

—*

 ​ ​Informal search

0.21



0.17

—*

 ​ ​Nonsearch

0.27



0.40

—*

0.49



0.50



Personal characteristics  ​ ​Female

0.39



 ​ ​Blacka

0.31



 ​ ​Hispanica

0.19



 ​ ​Other racea

0.11

— 0.11



0.54



0.67

—*

0.59



0.70

—*

 ​ ​White (reference

 ​ ​Not German

category)a

nationalityb

 ​ ​Married  ​ ​Any children under age 16  ​ ​Education

13.07

2.38

12.18

2.65*

 ​ ​Vocational training

0.63



0.67

—*

 ​ ​Years of experience

12.46

4.38

10.32

5.58*

 ​ ​Managerial occupation (reference category)

0.07



0.04

—*

 ​ ​Professional/technical occupation

0.18



0.27

—*

 ​ ​Administrative support occupation

0.17



0.12

—*

 ​ ​Sales occupation

0.09



0.10



 ​ ​Service occupation

0.16



0.07

—*

 ​ ​Construction occupation

0.11



0.19

—*

 ​ ​Operator/laborer occupation

0.23



0.22



12.21

8.93

Job characteristics  ​ ​Occupation

 ​ ​Hourly wages (in

dollars)a

 ​ ​Hourly wages (in Deutche Marks)b

30.44

18.89

 ​ ​Firm size

1.97

0.88

1.98

0.99

 ​ ​Part-time

0.16



0.30

—*

 ​ ​Union

0.09



0.16

—*

0.74



0.18

—*

 ​ ​Permanent contract  ​ ​Public sector

0.13



0.23



0.14



Sampling characteristics  ​ ​Black supplementa  ​ ​Hispanic

supplementa

85

86 Social Forces 91(1)

Table 2 ​continued United States Mean  ​ ​Military supplementa

0.02

Germany

SD

 ​ ​East German supplementb  ​ ​Foreign/immigrant

supplementb

 ​ ​1994



0.18



0.13

—*

0.13



0.18



0.13

—*

0.11



0.18



0.13

—*

0.13



0.16



0.23

—*

 ​ ​1999  ​ ​2000 (reference category)

0.28 —

 ​ ​1997  ​ ​1998

SD

0.48

 ​ ​1995  ​ ​1996

Mean



Observations (jobs)

12394

1635

Individuals

7392

1126

Note: SD = standard deviation. SDs not included for binary variables. *: Significant difference (p  .6). We also examine the extent to which workers have permanent employment contracts, a feature unique to the German labor market. A single dummy variable for public sector measured job sector location. Finally, we also controlled for survey design features. We included dummy variables for the various oversamples in both datasets. We controlled for the survey year with a series of dummy indicators, with 2000 serving as the reference category.6

Analytic Strategy Given the nominal nature of our dependent variable and the fact that respondents could contribute multiple observations, as described above, we estimated multinomial logit random effects models to assess differences in the types of individuals and jobs that are associated with informal job matching between the United States and Germany. This approach permits us to estimate an error term to account for random individual effects across multiple observations, while also allowing single observation individuals to be retained as part of the analyses (Petersen 2004). We used the GLLAMM procedure in Stata11® to calculate the random effects regression estimates (Skrondal and Rabe-Hesketh 2003).

87

88 Social Forces 91(1)

Results To what extent is informal job matching more or less prevalent in the United States and Germany? The descriptive evidence presented in Table 2 suggests that, on average, informal searching is more common in the United States, while nonsearching is more prevalent in Germany. These mean differences are statistically significant according to an independent samples t test. However, the observed differences in informal job matching prevalence could be due to compositional differences between the two countries. Therefore, we pooled the data from the two samples into a single file and regressed the type of job match on a dummy variable for country (Germany = 1; U.S. = 0). All of the variables that appeared in both the NLSY and GSOEP data were also included as covariates in the random effects multinomial logistic regression model, plus the hourly wages and work experience variables were combined across the two samples. These models provided estimates of the differences in the prevalence of informal job matching across the two countries (results not shown). Figure 1 presents the predicted probability of informal searching and nonsearching (vs. formal searching) for the two countries. The prevalence of informal job searching is not significantly different between the two countries, despite being slightly higher in the United States. However, 40 percent of Germans received their jobs through nonsearching, compared with 27 percent of Americans. U.S workers were significantly more likely to have found a job though formal searching than German workers (52% vs. 43%, respectively). While informal recruitment is more common in Germany than in the United States, the character of informal recruitment might be distinctive in these two contexts because of the different institutional environments. For example, informal recruitment in the United States often involves employee referral networks (Fernandez et al. 2000), whereas in Germany informal recruitment may involve Figure 1. ​Predicted Probabilities of Job Matching in the United States and Germany

60% 50% 40% 30% 20% 10% 0%

Formal search*

Informal Search United States

Non-search*

Germany

Note: *Statistically significant difference between countries (p