Job Growth and Replacement Needs in Nursing Occupations
Appendix 2: Methodology for estimating replacement needs
Methodology for estimating gross replacement
Gross replacement needs are estimated using data mainly from the Labour Mobility, survey for 2000 (ABS 2000b). This is supplementary to the monthly Labour Force survey, and is conducted every second year. Apart from exclusion of persons in some minor categories, for example, those aged 70 and over, the scope of the survey is the same as that for the Labour Force. The survey is retrospective in nature in that respondents are asked details about their labour experience in the previous twelve months. Among other information, respondents in the February 2000 survey were asked about their:
- Labour force status in February 1999;
- Labour force status in February 2000;
- Occupation in February 1999 if employed at that time; and
- Occupation in February 2000 if employed at that time;
Such information can help estimate the flows shown in Figure 43 and hence the gross replacement needs in each occupation.
If there is expansion in employment in an occupation then replacement needs in it can be determined as the sum of the outflows B, C and D, and if there is a contraction then it can be determined as the sum of the inflows E, F, G, H, I and J. A method for determining these two sums is presented below.
A method to estimate the sum of the outflows is considered first. Outflows B and C can be estimated directly from the data but D cannot be. There are two possible methods of estimating the sum of the outflows. The first method involves decomposing D into three sub-outflows:
- Out of the labour force but remaining in the population;
- Deaths; and
- Emigration.
Each outflow is then estimated separately. Estimates of the first flow can be obtained from the Labour Mobility survey data but the other two cannot be and alternative data sources would be necessary in order to estimate them. Estimates of outflows due to deaths could be made using the age-sex specific death rates published in ABS (1999a). Since the mortality rates for employed persons are generally lower than for others in the population, this method is likely to overestimate the outflows due to deaths. An additional problem is that the mortality rates even among the employed vary significantly by occupation. Estimates of outflows due to emigration could be made using data on permanent and long-term departures of Australian residents published in ABC (2001c). Although very reliable data are collected on the total number of emigres, in terms of their occupation the data are less reliable because the collection is based on self-reporting. A problem of more concern in the context of estimating emigration outflows is the lack of precise information on the emigre’s labour force status and occupation as at February 1999. Therefore, although some sort of estimates of outflows due to deaths and emigration is possible, there are doubts regarding the accuracy of these estimates.
The second method involves estimating the sum of the outflows B, C and D indirectly. If the total employment in the occupation at February 1999 is known, then the sum of these outflows can be estimated as the difference between total employment and outflows A, which are the number of stayers in the occupation. Employment by occupation at February 2000 can be directly obtained from the survey data but that at February 1999 it cannot be. This is because persons who in February 1999 were in a given occupation but are no longer in the population at February 2000, due to death or emigration, are outside the scope of the survey. One way to estimate employment by occupation at February 1999 is to proportionately scale up the observed 1999 employment by occupation from the Labour Mobility survey to reflect the true total for 1999. Estimates of the true total can be obtained from the Labour Force survey for February 1999, with the scope restricted to those persons aged less than 70. It is reasonable to ask why are the estimates of employment by occupation from the Labour Force survey not used directly. The reason for this is that, in spite of the Labour Mobility being a supplementary survey to the Labour Force, estimates of employment from the two surveys can vary substantially at the four-digit occupation level, and especially for small occupations. This is because different systems of weights are applied to the sample values from the two surveys to obtain population estimates.
The second method of estimating the sum of the outflows is more straightforward and uses data that are more consistent with each other and from similar sources. Hence it has been the method of choice to produce estimates for this report.
The method for estimating the sum of the inflows is more straightforward. Even though it is not possible to observe the individual inflows shown in Figure 43 from the Labour Mobility data, E and F can be observed jointly and so too can G, H, I and J.
The above allows estimates for total outflows from, and total inflows into, each occupation over the period for which the data are collected. To make projections outflow and inflow rates are calculated and applied to the current period data. For example, if projections are required for 2002 then, first, employment for 2000 and 2001 are compared to determine if there has been an expansion or contraction. If there has been an expansion then the outflow rate is applied to 2001 base data to obtain total replacement needs for 2002; otherwise the inflow rate is applied. The projections for a five-year period are simply five times this.
Methodology for estimating net replacement
The necessary flows of individuals to and from an occupation needed to estimate net replacement can be derived from data on stock of employed persons by age at two points in time using the cohort-component method (Eck 1991; Williams and de Grip 1993). The method allows estimation of flow rates in or out of an occupation by considering the net changes in cohort sizes over a given period. It has been widely used for calculating survival ratios in demography (ABS 1992; Davenport and O'Leary 1992; Kippen and McDonald 2000; Pollard, Yusef and Pollard 1974; Shryock and Siegel 1980). It has also been used to calculate student progression rates through courses in higher education (Shan and Burke 1999) and wastage rates in health manpower studies (Office of International Health 1979).
The methodology used to estimate net replacement in nursing occupations basically follows that developed in Shah and Burke (2001). The method is only briefly reviewed here as details can be found in the aforementioned paper.
To explain how the cohort-component method can be used to approximate
net replacement, first net outflows are defined. Suppose for simplicity
that the annual employment data in an occupation are available by five-year
age groups, 
represents
the size of cohort of age a at time t and 
represents the size of the same cohort five years later. Then the net
five-year flow from this cohort is represented by the change in the size
of the cohort over this period. In other words, it is given by:
.
If the size of the cohort has decreased (
)
then we say there has been a net outflow of (
),
otherwise the net outflow is zero.
In most occupations there are likely to be more leavers from older aged cohorts than there are entrants to them. Furthermore, new entrants to an occupation are likely to be found mainly in the younger cohorts and re-entrants mainly in the older cohorts. This means that, in general, net outflows from the older aged cohorts are likely to be positive, and will be made up of leavers less (mostly) re-entrants. The sum of the net outflows over all cohort components approximates the number of leavers from an occupation less (mostly) re-entrants, and thus provides an estimate for net turnover or replacement needs in the occupation.
The method outlined is applied if employment in the occupation has been expanding. If employment has been declining, then the sum of the net outflows is reduced by the contraction in employment because in this case not all those who leave are replaced.
The calculation of net replacement in an occupation in which there has been employment growth is illustrated in Table 9. The stylised example includes only five age cohorts. It is assumed that a person retires from the workforce when he/she reaches 40 years of age. In the example, employment increases by 20 over the 1990-1995 period, and hence no adjustments are made to net outflows. Net outflows only occur from the oldest three cohorts. For example, the net outflow from the cohort aged 25-29 is 40. The sum of net outflows, which equals 200, approximates net replacement in this occupation over the five-year period. Overall there are 220 new jobs and 90 are estimated to be taken by persons aged 15-19, 80 by persons aged 20-24 and 50 by persons aged 25-29.
Table A5. Hypothetical example of calculations of net replacement
|
Age cohort 1990 |
Employment 1990 |
Age cohort 1995 |
Employment 1995 |
Employment Change 1990-95 |
Net outflow 1990-95 |
|
15-19 |
90 |
90 |
|||
|
15-19 |
100 |
20-24 |
180 |
80 |
0 |
|
20-24 |
200 |
25-29 |
250 |
50 |
0 |
|
25-29 |
240 |
30-34 |
200 |
-40 |
40 |
|
30-34 |
180 |
35-39 |
120 |
-60 |
60 |
|
35-39 |
100 |
40 and over |
0 |
-100 |
100 |
|
Total |
820 |
840 |
20 |
200 |
Definition of cohorts
Net outflows from an occupation are likely to vary by background characteristics of individuals. There is ample evidence of the variation in the flows of individuals into and out of an occupation by sex. Women are more likely to temporarily withdraw, often for a number of years, from the workforce to have and raise families. A number of these women re-enter the workforce at a later time. The age at which women retire is also generally lower than that for men. Stromback (1984) provides some Australian evidence that shows patterns of leaving a job vary widely with the sex of a worker. Here age and sex define cohorts as do Willems and de Grip (1993), but Eck (1991) does not distinguish between male and female cohorts.
Retirement age
The model requires an assumption to be made regarding when an individual is likely to retire from the labour force. Since the oldest age group is 70 years and over in the publicly accessible Labour Force survey data, the model assumes a retirement age of 70. This means the net outflow rate from the 70 years and over age group will be 100 per cent. In Australia even though compulsory retirement age of workers has been relaxed in many sectors of employment in recent times, most people do tend to retire by the age of 70 years and therefore the assumption is unlikely to have much effect on the estimates of net replacement demand.
Adjustment for shrinking occupations
As explained earlier, if employment in an occupation is declining then not all those who leave are replaced. In such cases to compensate for the decline, the net outflows are reduced uniformly across all cohorts.
Business cycle correction
The flow of workers from an occupation into unemployment or to outside the labour force is affected by the business cycle. These flows are likely to be large during recessions. In Shah and Burke (1991) the net outflow rates are averaged over a number of periods to smooth out the business cycle, but in the latest refinement of the model a simple exponential smoothing algorithm is used to smooth the flow rates and to make projections.
Labour force structural change correction
Future changes in the labour force structure are likely to affect flow rates, too. Adjustments for it are made by adding to the forecasted flow rate the difference in the labour force growth rates in the current and the next period. The projections of the labour force for future periods are obtained from ABS (1999b). The corrections applied are sex-age specific only as the projected size of the labour force by occupations is unavailable.
Contents | Previous [highered/nursing/templates/includes/report_footer.htm]