Do social networks increase labour supply elasticity?

Abstract

Using an original data set including measures of social networks of migrants in China, we find that social networks increase the elasticities of labour supplies for migrant workers. The effects differ for men and women and can help explain part of the importance of developing social networks to improve labour market outcomes.

Notes

¹ When computationally feasible, maximum likelihood with the Heckman correction term is preferred to the two-step procedure because it produces efficient coefficients (Deaton, 1997).

² The coefficients would be unbiased but inefficient because the disturbance terms V_i and have heteroskedastic variances over a cross-section of individuals, since ln w_i appears in both Equations 1 and 2. As a result, interequation covariance is downwardly biased, so we compute heteroskedastic-consistent robust standard errors and also correct for clustering, as we are using a household survey with more than one observation per household.

³ Social capital was rescaled by a factor of 800 to lie between 0–100.

⁴ In brief, we first estimate a sample selection corrected MLE for hourly wages, including social networks and social capital in the specification (see Mroz, 1987 for a detailed discussion of the methodology). There are different determinants for the structural labour supply equation. The exclusion restriction used in the wage equation was whether an individual was content in the present job. The presence of children in the household was the exclusion restriction used in the participation equation. Health status is included in the participation equation as a significant determinant of participation. The second stage involves predicting wages from the earnings function that standardizes for observable characteristics. We then use predicted wages in the participation equation to find the responsiveness of hours to a change in the wage. The coefficient on predicted wage allows us to calculate the elasticity of the labour supply function, evaluated at the mean of hours and wages.

⁵ We do not report reservation wages for migrants, as they are atypical in that those who come to urban areas tend to have low or zero reservation wages. There were nine respondents in the sample who supplied positive hours for zero wages, for instance. This could have been due to measurement error, but also underscores the unique nature of the sample.