The effect of hours of work on social interaction

Henry Saffer, Karine Lamiraud

20 February 2008



Do you know who your friends are? Have you seen them lately? Data from both the United States and France show that some important forms of social interaction are on the decline (Putnam 1996; Blanpain and Pan Ké Shon 1998). While membership in social groups has remained relatively stable over time, there has been a decline in visiting friends, neighbours, and relatives. This decline in visiting is not simply due to friends switching to email communication and socializing at work. Evidence of a true decline in friendship is provided by McPherson, Smith-Lovin and Brashears (2006), who document a decline in the reported number of close friends over the past 20 years. Understanding the determinants of the decline in visiting has attracted interest in both the academic literature and in the popular press. It raises concerns on both sides of the Atlantic because social interaction is thought to have positive effects on the mental and physical health of individuals and the efficiency of economic institutions.

Are work and friends complements or substitutes?

An intuitively plausible reason offered for the recent decline in social interaction is growth in hours of work per capita. In particular, the increase in female labour force participation has increased hours of work per capita, which could result in less social interaction. However, it has also been argued that individuals who work longer hours are more inclined to both civic engagement and visiting with friends and neighbours. This could occur if there were an important unobserved third factor such as ambition that affects both working hours and social contacts. For example, an individual who is ambitious may choose to work long hours and to participate in civic organisations and meet with friends and neighbours more than a less ambitious individual. In this case, hours of work and social interaction would be positively related.

The theory of household production, developed by Gary Becker (1965), provides the basis for an empirical model of social interaction. Becker’s theory emphasises the role of time in consumption and that time is a limited resource. We (Saffer and Lamiraud, 2008) employ Becker’s theory to derive a demand for social interaction. This demand function, like any other demand function, shows that the quantity of social interaction demanded depends on its own price, the price of other goods, income and taste. The price of social interaction is positively related to the individual’s valuation of their non-working time. This price is usually approximated by the individual’s wage. However, in our study, we assume that the price of non-working time is a function of the supply and demand for this type of time. As hours of work increase, the supply of non-working time decreases. This raises the price of non-working time. Education is also an empirical proxy for the price of time. Education is assumed, to varying degrees, to increase productivity. An increase in the productivity of time reduces the time cost of social interaction.

An empirical examination

Empirically isolating the effect of hours of work on social interaction requires an exogenous change in hours of work. Our research focuses on France’s enactment of a new employment law that mandated an exogenous decline in hours of work. The 1998 legislation reduced the legal number of hours worked per week from 39 to 35. The employment law consisted of three parts: the first part covered firms with more than 20 employees, the second part covered firms with 20 or fewer employees, and the third part covered civil servants. Firms of more than 20 employees were required to conform to the law by January 2000, while small firms and civil servants were covered by January 2002. The changes in hours of work resulting from this law are exogenous to individual characteristics.

The empirical results clearly show that the employment law reduced average hours of work by 1.5 to two hours per week. These results are consistent with findings of Estevao and Sa (2006). However, the results show no evidence that these extra hours went to increased social interaction. That is, hours of work are not found to be an important determinant of social interaction. This remains true for sub-samples defined by gender, marital status, and children.

What shapes social interaction?

Human capital, however, is found to be an important determinant of social interaction. The effect of human capital, as measured by education and age, is positive for membership activities but negative for visiting relatives and friends. This is not an intuitive result and requires some explanation. One possibility is that this effect results from the productivity-enhancing aspect of education. Membership activities, like employment, are goal-oriented. Education increases productivity both at work and in membership activities. However, education has little effect on the productivity of time spent visiting. Thus, an increase in education results in greater productivity in membership activities and greater utility for the individual. To put this more intuitively, education makes membership activities more interesting and visiting less interesting. This shifts social interaction to membership activities and away from visiting.

Other factors were also found to be important determinants of social interaction. Higher income increases memberships and decreases visiting, which seems consistent with the education effect. Marriage tends to reduce all social interactions, which suggests that a spouse is a substitute for other social interactions. Children have a positive effect of membership in school and church groups, which is probably the result of complementarity between these activities and child care. Males tend to have less of all social interactions, which is a familiar result.

Finally, a comparison between France and the United States shows that the response to human capital and other variables are much the same in both nations. Since the time data show that visiting has declined while education has increased, it is possible that the true cause of the decline in visiting is rising education. Trends in social interactions, it seems, are not driven by a simple trade-off between work and play but by education and choices in consuming different types of socializing.


Becker, G. “A Theory of the Allocation of Time”, Economic Journal, 75(299): 493-517, 1965.
Blanpain, N. and Pan Ké Shon, JL. “1983–1997 Les Français Parlent de Moins en Moins,”, INSEE Premiere, 1998.
Estevao, M., and Sa, F., "Are the French Happy with the 35-Hour Workweek?," IMF Working Papers, 06/251, International Monetary Fund, 2006.
McPherson, J, Smith-Lovin L. and Brashears M. “Social Isolation in America: Changes in Core Discussion Networks over Two Decades”, American Sociological Review, vol. 71, June 2006.
Putnam, R., “The Strange Disappearance of Civic America”, The American Prospect, Winter, 1996.
Saffer, H. and K. Lamiraud, “The Effect of Hours of Work on Social Interaction”, NBER Working Paper 13743, January 2008.



Topics:  Labour markets

Tags:  working hours, social interaction

Professor of Economics at Kean University in New Jersey

Assistant Professor of Health Economics at the University of Lausanne