Who Gets and Who Gives Employer-Provided Benefits? Evidence from Matched Employer-Employee Data

Friday, 3 July 2015: 8:30 AM-10:00 AM
TW2.3.03 (Tower Two)
Tali Kristal, University of Haifa, Haifa, Israel
Despite their importance for workers’ wellbeing as an essential component of workers’ total compensation, and although they augment income inequality among workers, our understanding of the distribution and determinants of employer-provided benefits, such as pension plans, is still limited.  This paper aims to fill these lacunas in stratification research by developing a structural model of benefit determination and by analyzing inequality in benefits-coverage and benefits-level across and within organizations, based on a unique Israeli register data of workers matched to their workplaces.

My objectives are threefold.  First, I develop an explicit conceptualization of the structure which underlies benefits determination based on the concept of economic rent.  I put forward rent-extraction by powerful workers as a causal mechanism that produces within-workplaces variation; and workplaces’ rent-sharing, attributable to profitable firms and diffusion of employment practices, as a mechanism that generates between-workplaces variation.  Second, I focus on the workplace in the analysis of benefits, emphasizing that workplaces are the primary site for the generation of stratification (Baron and Bielby, 1980), in particular benefits inequality.  My third objective is to provide a first rigorous analysis of benefits-level determination based on the unique Israeli dataset that includes scanty data on employer cost benefits at the individual level in Israel.  Much as in other liberal and coordinated welfare states, in the hybrid Israeli liberal-coordinated state (Kristal, 2013) employer-provided benefits is a major component of workers’ total compensation and a fundamental source of inequality.     

To empirically test the study’s arguments, I make use of matched employer-employee data that have enabled studies to simultaneously analyze the effect of individual and workplaces’ characteristics on earnings and employment outcomes (Avent-Holt and Tomaskovic-Devey, 2012; Sørensen, 2007).  We know little, however, about these effects on benefits.  The Israeli matched employer-employee dataset is constructed from governmental registers and maintained by Israel’s Central Bureau of Statistics (CBS) for research purposes.  It is arguably the most reliable information on employer-provided benefits, as it is based on administrative records of the income tax collection system – reports of employers regarding their employees.  Moreover, unlike other studies on benefits inequality, it includes information at the individual-level on employer cost benefits that makes it possible to capture the level of benefits and not only their existence.  Another major advantage in the employer-employee dataset is the link between employer and employee that permits a more detailed delineation of the process by which workplaces’ characteristics affects individual labor-market outcomes. 

The Israeli employer-employee data contains a wealth of demographic and employment information characterizing workers, as well as rich information on their workplaces.  Since the matched employer-employee dataset does not include information on workers’ occupation and education level, both of them significant for any explanation of labor market inequalities, I asked the CBS to combine the employer-employee data for the year 2008 with half of the sample from the 2008 census of population, which yields a sample size of about 150,000 employed wage and salary workers aged 25-64 with positive wages who worked at least one month.  Apart from details about demographic factors, the Census also includes information on the level of education and 3-digit occupational codes, which is determined by the type of work performed by the interviewed person at his place of work and classified according to the ILO classification of occupations. 

In the analyses I employ a multilevel modeling strategy.  I estimate hierarchical logistic models to predict the odds of obtaining benefits and hierarchical linear models to predict the level of benefits among those workers who obtain them.  To empirically test the effect of workers’ characteristics and workplaces features on benefits, I use a random-intercept hierarchical model (also referred to in multilevel modeling as an intercept-only model).  To test for an interaction between individual and workplace effects, I analyzed the effects of workplaces’ features on the educational gap in benefits.    

In support of the theoretical model, I find that workers with greater bargaining power for rent-extraction (i.e., FTFY, service-class occupations, unionized, and educated workers) are more likely to obtain valued benefits within workplaces, independently of wages.  I also find that profitable (as indicated by firms’ size and economic scale, and financial firms) and formalized workplaces (i.e., state-owned, public organizations and older organizations) are more likely to share their rent by providing valued benefits for all their workers, educated workers in particular.  I conclude that benefits exacerbate disparities arising from wages, except for women who have higher odds of obtaining valued benefits than comparable men. 


Avent-Holt, Dustin and Donald Tomaskovic-Devey. 2012. “Relational Inequality: Gender Earnings Inequality in U.S. and Japanese Manufacturing Plants.” Social Forces 91:157-180.

Baron, N. James and William T. Bielby. 1980. “Bringing the Firms Back in: Stratification, Segmentation, and the Organization of Work.” American Sociological Review 45:737-765. 

Kristal, Tali. 2013. "Slicing the Pie: State Policy, Class Organization, Class Integration, and Labor’s Share of Israeli National Income." Social Problems 60:100-127.

Sørensen, Jesper B. 2007. “Bureaucracy and Entrepreneurship: Workplace Effects on Entrepreneurial Entry.” Administrative Science Quarterly 52:387-412.