Modelling physicians' labour supply: comparing a discrete choice

IZA Logo

IZA Seminar

Place: Schaumburg-Lippe-Str. 9, 53113 Bonn

Date: 12.04.2011, 12:15 - 13:30


Presentation by 

Guyonne Kalb (Melbourne Institute of Applied Economic and Social Research)


There are only a few studies of physicians' labour supply that use a structural labour supply model, and no studies of Australian physicians' labour supply exist. The aim of this paper is to examine the non-pecuniary and pecuniary determinants of physicians' labour supply using a sample of medical doctors from the first wave of the new Medicine in Australia: Balancing Employment and Life (MABEL) longitudinal survey of doctors. Our basic sample of analysis includes 1,359 male and 1,254 female GPs and 2,089 male and 845 female Specialists from the first wave conducted in 2008. We take a discrete choice approach in which the labour supply decision is viewed as a choice from a number of discrete alternative combinations of income and leisure time (defined as total hours available minus working hours). We allow a choice from ten labour supply levels, each of which is associated with a specific level of net household income. Net income is computed at each labour supply point by taking account of the Australian tax and transfer system. Preference parameters are estimated which indicate the value of net income versus the value of leisure time. These preferences are allowed to vary by age, State of residence, presence of dependent children, age of youngest child, health status, presence and working status of spouse, and we have included unobserved heterogeneity in the preference for income and leisure. A conditional logit is used to obtain estimates for the parameters of a quadratic utility specification for male and female GPs separately and for male and female Specialists separately. Despite some loss of information, the discrete choices can be an accurate representation of reality, especially when individuals have limited choices regarding working hours. A major advantage of the discrete choice approach compared to the continuous labour supply model is that it simplifies estimation when one wants to include the details of complex (non-linear and/or non-convex) budget constraints arising from the tax and transfer system. We calculate the wage elasticity and the marginal effects on labour supply for each variable together with the corresponding confidence intervals for each individual. The preliminary findings show that, on average, the wage elasticities are significantly negative, but small, for all four groups, ranging in size between -0.07 for male and female GPs, -0.09 for male Specialists and -0.10 for female pecialists. Within each doctor group there are substantial differences in wage elasticities, with values ranging from positive to negative: roughly in between -0.5 and +0.5. The results show significant differences in the marginal effects of non-pecuniary factors between GPs and Specialists, as well as between men and women. For instance, female doctors prefer to work fewer hours when an additional child is present, with a larger effect for GPs than for Specialists. Conversely, male GPs and Specialists prefer to work more hours when an additional child is present. We also estimate a reduced form specification by regressing log wage, log income, and a number of individual and household characteristics on log hours. We compare the results of the two different specifications. The effects of individual and household characteristics are in the same direction and of similar relative magnitude as in the structural model. Wage elasticities are quite comparable, on average, although in the reduced form specification, by assumption, all individuals are estimated to have the same wage elasticity.

For more information, please contact