Skip to main content

Table 3 Transfer function models from the first period 1961 to 1982

From: Physician and nurse supply in Serbia using time-series data

Dependent variable (labels and name)

Potential predictors in start model (name only)

Significant predictors in final model (name only)

Model type

Stationary R2

Number of outliers

Q-stat (P value)

Z-stat (P value)

Physicians (y1)

x1, x2, x3, x4,x5, x6

x2

TF (0,1,0)

0.63

0

1.52 (0.96)

0.81 (0.52)

Nurses (y2)

y1, x1,x2, x3,x4, x5, x6

none

ARIMA (0,2,0)

0.76

1

4.67 (0.59)

0.93 (0.35)

Inpatient care discharges (x3)

y1, y2,x1, x2,x4, x5, x6

none

ARIMA (0,1,0)

0.96

3

8.30 (0.22)

0.89 (0.40)

Outpatient care visits (x4)

y1, y2,x1, x2,x3, x5, x6

y1

TF (0,1,0)

0.61

0

6.00 (0.42)

0.70 (0.72)

Students enrolled in the first year of studies (x5)

y1, y2,x1, x2,x3, x4, x5, x6

x6

TF (0,1,0)

0.98

4

0.60 (0.99)

0.55 (0.92)

Graduated medical doctors (x6)

y1, y2, x1, x2, x3, x4, x5, x6

none

ARIMA (0,1,0)

0.85

3

3.85 (0.70)

0.84 (0.48)

  1. Legend: stationary R2 - measure of goodness of fit of model. Range is from negative infinity to 1; Q-stat - is Ljung-Box Q statistics that test the null hypotheses of no autocorrelation in residual series; Z-stat - is Kolmogorov-Smirnov statistics that test the null hypotheses of normal distribution of residual series.