Nurses and physicians: a longitudinal analysis of mobility between jobs and labor supply

Abstract

We estimate a dynamic discrete choice model of registered nurses’ labor supply. A distinguished feature of our model is that the random terms in the utility functions are correlated over time and jobs (habit or job persistence). Past options and not only the past optimal choices matter for the current choices. Given observed incentives and institutional constraints on offered hours, we find that nurses are mobile when they are young (less mobility than among physicians), but there is also a weak tendency of higher mobility again when they are approaching retirement age. Wage increases have a modest impact on labor supply. The overall elasticity for nurses is close to zero. These low elasticities shadow for stronger responses, shifting labor away from part-time jobs in the public and private sector toward full-time jobs in the private sector. A change in taxation away from the progressive tax system toward a flat tax of 28 % gives registered nurses a very modest incentive to shift their job to private hospitals. For physicians, the impact is stronger.

Appendices

Appendix 1: Tax functions

See Tables 6, 7 and 8.

Table 6 Tax function, 1997

Table 7 Tax function, 1998

Table 8 Tax function, 1999

Appendix 2: Data

In this appendix, the data and estimates for physicians are copied from Andreassen et al. (2013) (Tables 9, 10, 11, 12, 13, 14, 15).

Table 9 Sample selection

Table 10 Number of married physicians and married nurses in the different sectors

Table 11 Average weekly hours across sectors

Table 12 Distribution between short part-time and long part-time among married nurses

Table 13 Mean of the explanatory variables for married physicians and married registered nurses

Table 14 Mean consumption for married physicians and married nurses by sector

Table 15 Mean leisure for married physicians and married nurses by sector

Appendix 3: Observed transition rates for nurses, transition rates for physicians are given in Andreassen et al. (2013)

See Tables 16, 17, 18 and 19.

Table 16 Transitions of married nurses from 1997 to 1998

Table 17 Transitions of married nurses from 1998 to 1999

Table 18 Transitions of married nurses from 1997 to 1998

Table 19 Transitions of married nurses from 1998 to 1999

Appendix 4: Wage equations and selection effects for nurses, for physicians see Andreassen et al. (2013)

We estimate wage equations for all individuals for the 3 years 1997, 1998 and 1999. The wage equations for physicians are documented in Andreassen et al. (2013). The wage equations for nurses are documented below. We take sample selection into account by including the predicted choice probabilities as explanatory variables in the wage equations. These probabilities were the predictions resulting from a simple multinomial logit estimation of sector choice. We show the mean of the explanatory variables used in the estimation of the choice probabilities in Table 20. The estimates of the choice probabilities are given in Table 21, and the resulting average predicted probabilities are given in Tables 22, along with the means of the other variables used in the wage equations. The logit estimations were done on all nurses in a given year, while the wage equations were estimated on all working nurses with observations of wage income (Table 23). The estimates of the wage equations are given in Table 24. The wage equations for all nine work sectors have been estimated simultaneously using maximum likelihood, allowing for correlation between the different wages. The parameters ${\sigma }_1$ to ${\sigma }_9$ are the variance parameters mentioned in the main paper, and the parameters ${\kappa }_1$ to ${\kappa }_9$ are the parameters allowing for correlation between sectors. As shown in tables, these correlation factors are not found to be significant, indicating that there is not much residual correlation between the different wages after correcting for the other explanatory variables. In general, being a woman reduces wages, while wages increase with age. Table 25 shows the mean and predicted hourly wages for nurses and derived from predictions using the estimated wage equations reported above. Table 26 shows the predicted wages for physicians based on the wage equations documented in Andreassen et al. (2013). The wages for doctors are higher and vary more than the wages of nurses.

Table 20 Mean of the explanatory variables for the logit estimation

Table 21 Logit estimates of choice of sector and hours (job type)

Table 22 Sample selection for logit estimation and estimation of wage equations

Table 23 Mean of the explanatory variables for the wage equations

Table 24 Estimated coefficients of the wage equations for nurses 1997–1999

Table 25 Mean and median predicted hourly wages for nurses

Table 26 Mean and median predicted hourly wages for physicians