«City Economy 14 Examining the relationship between commuting patterns, employment growth and long term unemployment in the Sydney Major Statistical ...»
migrated into the same area. The goodness of fit measures (adjusted R2) indicate that the relationships are strong (0.95 and 0.85, respectively for in-commuting and in-migration). So both out-migration and out-commuting occur in areas where employment losses arise.
The employment growth had only a small impact on the change in unemployment however (1000 extra jobs reducing unemployment by 5 via reductions in the unemployment rate but increasing it by 20 as a result of demographic processes (among them the hidden unemployed). Notably, 81 workers dropped the out of the labour force via participation rate changes for every 1000 jobs created. Given the surprising nature of this result, we are exploring the sensitivity of different age cohorts. It should be emphasised that the adjustments are over a 5 year period and they indicate that at least for some males (particularly those unemployed in 1996) the response to employment growth has been extremely muted.
The constant terms in each equation, inasmuch as we can interpret them as indicating the extent of adjustment that is not explained by the percentage change in employment, tell us that the labour force grew by 4.0 (on average) as a result of natural increase; by 1.3 per cent (on average) as a result of net in-migration; shrunk on average by 1.4 per cent due to declining labour force participation rates.
Bailey and Turok (2000: 642) suggest that “part of the explanation for these changes must lie in the changes for different occupational groups. With a rising proportion of jobs in the professional and other skilled occupations, it would be expected that a larger proportion of the employment opportunities would be taken by in-commuters. This reflects the fact that the more advantaged population cohorts have greater choice of housing and transport and as a result tend to commute longer distances than the more disadvantaged segments of the population. Equally, the declining participation rate for the resident populations is consistent with a smaller proportion of less skilled job opportunities.
The regression models can be extended by adding more control variables to the right hand side. As a first step, we included a metropolitan dummy which took the value of 1 for a metro region and 0 otherwise (based on the Sydney MSR geography). Following Bailey and Turok (2000: 642), we also controlled for occupational structure as a “means for assessing the extent to which different occupational groups were able to adjust to employment change.” In this regard, the percentage of manual male workers in total male employment for each area was also included. The relationship between occupational structure (as measured here) and employment change is predicted to be strongly positive in a growth phase.
The inclusion of the occupational structure does not significantly alter the estimated labour market responses to employment change which are shown in Table 3, although the goodness of fit of the regressions improves in some cases. The statistically significant negative coefficients (on the labour force participation and change in the unemployment rate components) suggest that areas which had higher percentage of manual employment have lower labour force participation rates and less reduction in unemployment as a result of employment change. The natural increase component is marginally significant (t-stat = 1.85) and positive. Bailey and Turok (2000: 642) also note a similar result for the UK and drawing on the work of Armitage (1997) suggest that “this is likely to reflect the higher fertility rates which occur in areas with higher concentrations of manufacturing industry”.
Female results The female results for the labour market adjustment responses to employment change are shown in Table 5. The results are in contrast to those for males (Table 3). Overall, the labour force responses due to demographic processes are smaller for women than they are for men. Further, the prior expectation was that women would be less likely to respond to employment change through migration or commuting by comparison to their male counterparts. However, while the net inmigration response is lower for females (219 jobs per 1,000 extra jobs compared to 364 for males), the in-commuting coefficient (highly statistically significant) tells a different story and indicates that for every 1,000 jobs generated there are (on average) 832 extra female in-commuters to the area (compared to 744 for males). Indeed, in-commuting is the main female response to employment change. Bailey and Turok (2000) found that the main response for women was in changing labour force participation rates. Given that our data is for a period of consolidated employment growth (in contrast to Bailey and Turok, who studied a period of employment loss) we would expect the cyclical labour force responses to be muted. The results confirm this expectation. For every 1000 jobs created 35 women leave the labour force.
The main picture to emerge from the results is that women rely more heavily on commuting across regions relative to men to gain income-earning opportunities in response to employment growth.
Similar to the male results, employment growth had only a small impact on the change in female unemployment however (1000 extra jobs reducing unemployment by 2 via reductions in the unemployment rate but increasing it by 14 as a result of demographic processes. Given the adjustments are over a 5 year period the response of unemployed females (in 1996) to employment growth has been extremely muted.
The notable difference in the constant terms for women is in the participation rate response. The constant terms indicate the extent of adjustment that is not explained by the percentage change in employment. For females, this component suggests a 2.3 per cent (on average) labour force increase as a result of labour force participation responses compared to the shrinking male response.
Table 6 reports the results of the extended regressions along the same lines as those reported in Table 4 for males. The metropolitan dummy impacts significantly only on the change in unemployment due to demographic processes response for females. The results suggest that females in metropolitan regions enjoy greater reductions in unemployment relative to their non-metropolitan counterparts when employment is growing. There are no other statistically significant differences in the labour market responses for females between the metropolitan and non-metropolitan areas.
As in the male case, the inclusion of the occupational structure variable in the female regression does not significantly alter the estimated labour market responses to employment change which are shown in Table 5, although the goodness of fit of the regressions improves in some cases. The natural increase component is strongly significant and positive and dominates the employment change response. It is also consistent with the male result discussed earlier. The other responses are not sensitive to the inclusion of this measure.
This paper is the first to apply the Labour Market Accounts framework to Australian data. At this stage the results must be considered to be tentative. The heavy reliance of both women and men on changes in commuting patterns is in part a consequence of the choice of SLAs as the basic spatial units of analysis. By contrast, Bailey and Turok (2000) examined 28 urban areas, so that intra-urban changes in the spatial pattern of commuting would be suppressed in their work. The net incommuting variable is constructed via employment figures from the WPP and BCP. These data warrant further exploration.
The main picture to emerge from the results is that women rely more heavily on commuting across local areas relative to men to gain income-earning opportunities in response to employment growth.
The employment growth had only a small impact on the change in unemployment for both males and females. The metropolitan dummy only impacts significantly on labour force participation with the metropolitan regions having higher levels of labour force activity than the non-metropolitan regions. Within this single equation framework the commuting results are separate from those relating to participation.
The inclusion of the occupational structure does not significantly alter the estimated labour market responses to employment change which are shown in Table 3, although the goodness of fit of the regressions improves in some cases. The statistically significant negative coefficients (on the labour force participation and change in the unemployment rate components) suggest that areas which had higher percentage of manual employment have lower labour force participation rates and less reduction in unemployment as a result of employment change. The natural increase component is marginally significant and positive.
The next phase of this research is to focus more closely on the dominance of net in-commuting in the male and female regression results. Increasing net in-commuting can arise if in-commuting increases and/or out-commuting declines. Disentangling these sub-components is important because the two flows have significantly different implications for the local area labour forces.
REFERENCESArmitage, B. (1997) ‘Variations in fertility between different types of local area’, Population Trends, 87, Spring, pp. 20-28.
Australian Bureau of Statistics (ABS) (2001), Demography NSW, Cat no. 3311.1, Government Publishing Service Canberra.
Australian Bureau of Statistics (ABS) (2001), Deaths, Australia, Cat no. 3302.0, Government Publishing Service Canberra.
Australian Bureau of Statistics, (ABS) (2002), Census of Population and Housing Basic Community Profile 2001, Catalogue No. 2001.0, Government Publishing Service, Canberra.
Australian Bulletin of Statistics (various issues) Death, Cat.3302.0.
Bailey, N. and Turok, I. (2000) ‘Adjustment to Job Loss in Britain’s Major Cities’, Regional Studies, 34(7), pp. 631-653.
Barkley, D.L., Henry, M.S. and Warner, M. (2002) Economic Development: The Role of Local Labor Market Adjustments, Southern Rural Development Center, Mississippi.
Commuting patterns and employment growth City Economy 14 - 16 City Economy 14 Denniss, R. and Watts, M.J. (2001) ‘Regional Labour Markets: Naturally Less Efficient?’, in E.
Carlson and W.F. Mitchell (eds.) Achieving Full Employment, supplement to Economic and Labour Relations Review, 12, pp.166-182.
Green, A. E. and Owen, D. (1991) ‘Local labour supply and demand interactions in Britain during the 1980s’, Regional Studies, 25, pp. 295-314.
Gordon, I. (2003), ‘Unemployment and Spatial Labour Markets: strong adjustment and persistent concentration’, in Geographies of Labour Market Inequality, Martin, R and Morrison, P.S. (Eds), Routledge, London.
Mitchell, W.F. and Bill, A., (2004) ‘Spatial dependence in regional unemployment in Australia’, in Carlson E. (ed) A Future that works: economics, employment and the environment, Proceedings of the 6th Path to Full Employment Conference and the11th National Conference on Unemployment, pp. 312-326.
Mitchell, W.F. and Bill, A., (2005a), ‘A Spatial Econometric Analysis of the Irreversibility of Long Term Unemployment in Australia’, Working Paper No. 05-05, Centre of Full Employment and Equity, Newcastle.
Mitchell, W.F. and Bill, A. (2005b) ‘Who Benefits from Growth? Disadvantaged Workers in Growing Regions?’, paper presented to 2005 Australia New Zealand Regional Science Association International annual meeting, Manukau, New Zealand, September.
Mitchell, W.F. and Carlson, E.A. (2005) ‘Exploring employment growth disparities across metropolitan and regional Australia’, Australasian Journal of Regional Studies, forthcoming November.
Owen, D.W., Gillespie, A.E. and Coombes, M.G. (1984) ‘Job Shortfalls in British Local Labour market Areas: a classification of labour supply and demand trends, 1971-81’, Regional Studies, 18, pp. 469-488.
Renkow, M. (2003) ‘Employment Growth, Worker Mobility and Rural Economic Development’, American Journal of Agricultural Economics, 85(2), pp. 503-13.
Scott, J. and Johnson, T.G. (2000) ‘The Community Policy Analysis Network: A national infrastructure for community policy decision support’, Journal of Regional Analysis and Policy, 28, pp. 49-63.
Tobler, W. (1970) ‘A Computer Movie Simulating urban Growth in the Detroit Region’, Economic Geography, 46, pp. 234-40.
Turok, I. and Edge, N. (1999) The Jobs gap in Britain’s Cities: Employment Loss and Labour Market Consquences, Joseph Rowntree Foundation, York.