«City Economy 14 Examining the relationship between commuting patterns, employment growth and long term unemployment in the Sydney Major Statistical ...»
Also cities with high shares of manual workers experienced less out-migration and greater increases in inactivity, in response to a given level of employment change. The authors attribute these results to a number of factors. First more qualified individuals have higher incomes and are able to commute greater distances. In addition, women tend to be more constrained than men due to their higher level of domestic responsibilities, and greater incidence of part-time work. Second, less qualified workers are alleged to experience greater barriers to migration than professional and managerial employees, which can be attributed to income levels, moving costs and barriers to migration arising from the social housing system. Bailey and Turok (2000: 648) suggest that there are likely to be few direct benefits for residents from creating professional and managerial employment, unless a greater percentage can be obtained for this group. First, there are few unemployed residents in these occupations. Second, the potential applicants for these jobs have wide commuting fields and hence significant choice about housing location. On the other hand, job creation for less qualified workers brings direct benefits. Over half of the jobs are obtained by residents previously unemployed or inactive; while more than a quarter go to in-migrants or those who would have migrated out. Few jobs are lost to commuters.
US researcher, Renkow (2003) also employs the labour market accounts framework to explore the determinants of the components of the labour market adjustment process across both urban and rural counties in the USA over a period 1980-90. His study of county-level data covers North Carolina and adjacent counties in Virginia, South Carolina, Georgia and Tennessee. The motivation for his study is both the question of who secures new jobs created in a particular county, but also the public finance considerations, since 44 per cent of local public expenditures in rural North Carolina are funded by residential property taxes, but equally the demand for public services might change too. If new jobs are secured by new residents (in-migrants) rather than residents, the increase in property taxes is accompanied by an increase in demand for public services. However, Renkow does not identify net in-migration by decomposing the change in the labour force.
Commuting patterns and employment growth City Economy 14 - 4 City Economy 14 Renkow (2003: 506) regresses the changes in in- and out-commuting, the change in the labour force and unemployment on a metro dummy, the local changes in employment and the labour force, and the changes in the labour force, the relative wage and relative housing costs within the each county’s commuting zone. Renkow shows that changes in commuting patterns and the size of the labour force take up most of the labour market adjustment associated with employment change, as opposed to the unemployment rate, which is consistent with the work of Owen et al. (1984).
Significant differences in the pattern of labour market adjustment are found between rural and metropolitan counties. The significant take up of new jobs via in-commuting suggests that the leakages associated with employment shocks may be substantial (Renkow, 2003: 510). The author concedes that the geographical unit chosen, namely counties, may influence the results with a larger unit leading to a smaller leakage.
THE LABOUR MARKET ACCOUNTS MODEL
The labour market accounts framework decompose the movements in working age population (WAP) and labour force (LF) for a particular area to determine who fills the jobs arising from changing employment levels. The approach is useful for analysing the amount that any particular community enjoys higher incomes as a result of employment growth (Barkley et al., 2002) as well as providing the basis for measuring the shortfall of jobs in a local area (Bailey and Turok, 2000).
Figure 1 presents a stylised version of the LMA framework to show the seven labour force sources (components) for workers in local employment. Following Barkley et al. (2002), local residents who are currently not in the labour force may choose to become economically active (A) by increasing their labour force participation. Local unemployment residents may gain local employment (B). Local residents who are in employment (locally or not) may take additional jobs (C), or they may quit and take new local jobs (D, E). Residents from outside the local area may also in-commute (F) or ‘in-migrate’ (move into) the local area (G) and take employment there.
Figure 1 Allocation of new jobs among components of the labour force Source: Barkley et al, 2002.
The system of labour market accounts used in this paper draws on the contemporary approach of Bailey and Turok (2000), which is compatible with the stylisation presented in Figure 1. Bailey and Turok (2000, p.637) note that employment change over time in an area gives rise to three interrelated changes, namely changes in the number of economically active residents, which Commuting patterns and employment growth City Economy 14 - 5 City Economy 14 incorporates the level of net in-migration, changes in the number of these residents who are unemployed and changes in net commuting flows.
Then ΔE ≡ ΔLF − ΔU − ΔC (1) where E denotes employment in the local area, LF is the local resident labour force (or the number of economically active local residents), C is the level of net out-commuting, U denotes the level of unemployment of local residents and the symbol Δ denotes the change in levels.
Similarly Bailey and Turok (2000: 638) note that changes in unemployment can also be broken down into the component associated with the change in the demographic process and that arising from the change in the unemployment rate, so that ΔU ≡ ΔU d + ΔU r (4) The final component of the accounts arises from the change in the net in-commuting associated with the local area (ΔC), which, subject to data availability, can be broken in into gross changes in
in-commuting minus gross changes in out-commuting. More simply it can be written as:
ΔC ≡ ΔEr − ΔEl where ΔEl denotes the change in local (SLA) employment and ΔE r denotes the change in the level of employment of residents, some of which is local. Then successive substitution of (2), (3) and (4)
into (1) yields the following identity:
ΔE ≡ ΔNI − ΔNM + ΔLFr − ΔC − ΔU d − ΔUR (5)
DATA SOURCES AND DESCRIPTIONSStatistical Local Area (SLA) data for this paper are drawn from a number of sources. Census data for a range of demographic characteristics (age, sex and occupation, as well as the SLA based economic and socio-demographic characteristics are taken from the Basic Community Profile (BCP), the Time Series Profile (TSP) and the Working Population Profile (WPP). These data were collected on the 7 August 2001 and persons are counted on an enumeration, rather than usual residence basis. While the Greater Metropolitan Sydney study area, officially comprises 70 SLAs, only 55 SLAs were recorded in the customized data table supplied by the ABS (involving the aggregation of Sydney, Newcastle, Blacktown, Sutherland Shire and and the removal of SLAs in the upper and northern Hunter). These 55 SLAs became the new basis for our study using Census data for 1996 and 2001. The SLA based Basic Community Profiles (BCPs) for the two years yield the resident populations of the SLAs by age (and sex) and hence the WAP, that is workers 15 and over, the labour force participation rate and the levels of employment by industry and occupation and unemployment.
A simple comparison of the WAPs over the five years yields the natural increase in the WAP from individuals getting older minus any deaths in that age group plus the level of net in-migration. The natural change in the WAP can be obtaining from age adjusting the 1996 WAP. In order to calculate Commuting patterns and employment growth City Economy 14 - 6 City Economy 14 the rate of natural labour force change SLA level death rates were devised using SLA Demography NSW, 3311.1, 2001 and Deaths, Australia 3302.0, 2001. An estimate of deaths across the age distribution for men and women in each SLA over the 5 year period is obtained by calculating the implied deaths across the age distribution from the age and sex specific NSW death rates and reconciling through pro-rata adjustment the implied total number of deaths in each SLA with the total official recorded annual deaths in each SLA. The estimate of total deaths across the age distribution for both males and females enables the computation of net in-migration by sex.
The Working Population Profiles (WPP) for each census year yield the local (SLA) levels of employment by sex, which contribute to the computation of the change in net in-commuting over the 5 year period. Data does not permit complete disaggregation of labour market accounts by occupation. Complete analysis would require unemployment by occupation and gender for each spatial area.
For SLAs within the NSW Greater Metropolitan Region, a customised table of employees broken down by occupation minor group (3 digit) by sex was also obtained from the ABS for the purposes of calculating changes in commuting patterns. A further table of Journey to Work (JTW) data was supplied by the NSW Department of Planning, providing a matrix of commutes on the night of the 1996 and 2001 Census. It counts employees travelling from their home SLA to their work SLA, broken down by occupation and sex within the Greater Metropolitan Sydney Area.
MODELLING LABOUR MARKET RESPONSES TO EMPLOYMENT GROWTH
Overview of labour market responses Table 1 presents the summary statistics of the labour market responses to employment change between 1996 and 2001 for males and females. The areas gained on average 8.6 per cent of their male labour force and 8.8 of their female labour forces over this period due to demographic changes with Net in-migration dominating (4.7 per cent for males and 5.3 per cent for females). In this Table 1 Summary statistics of labour market responses to employment change, 1996-2001 Mean Std. Dev. Maximum Minimum
Commuting patterns and employment growth City Economy 14 - 7 City Economy 14 growth period, male labour force participation changes reduced the available labour force on average across the areas whereas female labour force participation increased. It is clear that on average, the employment growth has only had a muted impact on the unemployment of residents.
While changes in net in-commuting were a positive addition to the labour forces for the study areas for males (1.7 per cent on average), they represent a dominant labour market response for females (6.8 per cent on average). This is the notable result from our study and bears further analysis.
To shed more light on these results, we examine the LMA decompositions for variation among the areas within the study region outlined in Section 4. Figure 2 shows the individual LMA components of the change in employment between 1996 and 2001 for males while Figure 3 shows the LMA components for females. It is clear that the Sydney SLA dominates the other SLAs in both cases.
The variation of values across areas is greater for males. There is also clearly more variation in labour market adjustment displayed for both males and females in the components that involve people moving either their residence or travelling to work via commuting. The muted response of unemployment revealed in Table 1 also translates in a lack of variation in the unemployment responses across the SLAs, the Sydney SLA included. It is clear that the employment growth experienced over this period (1996-2001) was not particularly beneficial in terms of reducing unemployment.
In terms of the net in-commuting, there is much greater variation for males between the areas that experienced net out-commuting and those that gained from this behaviour in the form of net incommuting. Table 2 ranks the SLAs in descending order for the net in-commuting component and provides some indication of which SLAs gained workers from this source and which lost workers.
There is very little correlation across males and females of the sorted lists of SLAs.
Change (as % of LF in 1996) Figure 3 Change in employment by SLA, 1996-2001, Female residents Source: Authors’ own calculations from Equation (5).
Regression analysis of labour market responses Bailey and Turok (2000: 639) use regression models “to examine the relative strength of the relationships between employment change and each of the labour market adjustment variables.” Several models were explored. The first equations estimated separately for males and females involve regressing each of the labour market adjustment components outlined in Section 3 expressed as a percentage of the 1996 labour force on the employment change between 1996 and 2001 (for males and females, respectively) expressed as a percentage of the respective 1996 labour forces.
The slope coefficient measures the response of that particular labour market adjustment mechanism to employment change. The constant term may be interpreted as measuring the labour market adjustments that are not attributable to employment change although their robustness in this context is debatable. We also seek to determine whether the initial occupational structure of an area impacts on adjustment? It is expected that areas with higher proportions of manual workers would experience lower levels of adjustment. This leads to the further expectation that the adjustment processes of men and women are different. While this arises partly as a result of occupational differences, women are more likely instrumentally attached to the labour force.
Male results The male results for the labour market adjustment responses to employment change are shown in Table 3. As outlined in the introduction, the sample period (1996 to 2001) was a period of consolidated growth (following the 1991 recession).
The results show that there is considerable adjustment to employment change in the form of net inmigration and net in-commuting with the latter dominating. The results confirm those of Bailey and Turok (2000) who found similarly for the UK. By way of interpretation, for every 1000 male jobs created in an area, net in-commuting by men rose by 744 and 364 economically-active men