Purpose
Over the last forty years New Zealanders, along with people in most other developed countries, have experienced big changes in the jobs they do and the places they live and work. These changes include:
- a decline in the number of manufacturing jobs;
- an increase in the quantity and renumeration of information-intensive work, often requiring tertiary education training;
- increased participation of females in the paid workforce;
- the rapid growth of large “super cities”; and
- internal migration to locations with desirable amenities, especially a favourable climate.
These trends have simultaneously changed the nature of work and the location of workers. Some urban areas have transformed themselves and thrived in the new environment, while others have found the transition difficult and have stagnated. Moreover, the transition has been accompanied by changes in income inequality as the relative importance of different skills and talents has changed and as the returns to those skills have varied across regions.
This paper uses census data to document and analyse the changing nature of jobs in regional New Zealand between 1976 and 2013. While the material is largely descriptive, its aim is to unravel the effects of several different forces on the evolution of jobs, towns and cities. This paper is not designed to make predictions about either the future of work or the future of regions. Rather, by documenting the evolution of regional employment patterns in New Zealand over the last forty years it aims to help understand how New Zealand has got to its current situation.
Methodology
To explore these issues, the paper uses census data covering employment in 30 urban areas and 65 separate industries between 1976 and 2013. Using the Stats NZ classification (based on UA13), we identified three large cities, 14 medium sized urban area ranging in size from Blenheim (2013 population of 29 298) to Hamilton (203 448), and 13 small urban areas ranging from Greymouth (9660) to Timaru (27 051). The total population of the medium sized areas in 2013 was 1 039 000, while the population of the small areas was 213 000. The population of Auckland, Wellington, and Christchurch were 1 335 000, 377 000 and 353 000 respectively.
City-industry employment data are available at five-year intervals from census returns for the years 1976-2013.5 We aggregate these data into 65 industries representing two- or three-digit industry codes (based on ANZSIC06), for example dairy product manufacturing (5 061 employees nationwide in 2013); supermarket and grocery stores (37 704 employees); banking and finance (35 793 employees); or education (129 324 employees). The employment numbers include working proprietors. One of our main results traces the effect of employment shocks to two-digit industries over five-year periods from 1976 to 2013. The most interesting findings concern the effect of employment changes in the primary and manufacturing sectors, which employed 25% of workers nationally in 1976.
A key part of our methodology is to identify the industries that are ubiquitous to all urban areas and the industries that are concentrated in certain locations and can be considered city-specific. To do this, we use employment census data from 1976 to 2013 to calculate location quotients for each industry in each urban area. A location quotient is the fraction of a location’s workers employed in a certain industry relative to fraction of national employment in that industry. The location quotients for the “ubiquitous” industries that are found in all urban areas are near one, and consequently the cross-city variance of the location quotients of these ubiquitous industries is low. In contrast, the cross-city variance of the location quotients of tradeable industries that are concentrated in certain urban areas is large. By ordering industries according to their cross-city variances, industries that are concentrated in a few areas but absent from others can be systematically identified.
We used a methodology based on Bartik (1991) to trace the effect of employment shocks on city-level employment. The Bartik methodology defines a city-industry employment shock as the amount of employment a city would lose or gain in a particular industry if that industry contracted or expanded at the national average rate. This approach automatically adjusts a city’s employment change for the over-or under-representation of an industry in that city. We focus on the effects of employment shocks to tradeable industries (identified as industries with high location quotient variances) on total employment and on employment in other industries. Much of the focus concerns shocks to manufacturing industries as these comprise a large fraction of the industries in the tradeable sectors.
The detailed location and industry data are used in several other ways. In section 2.3, they are used to trace the location of jobs in expanding sectors. In section 2.4, they are used to estimate employment churn, the extent that people switch jobs from one industry to another, in different sized cities. In section 3.1 the data are used to see how city-level employment in different industries is affected by local demand, to examine the extent non-tradeable goods are in fact more affected by local demand conditions than tradeable goods. In section 3.2 the employment data are used to analyse how nationwide shocks to manufacturing and primary industries affected employment in other sectors, to ascertain if the way cities responded to the largely negative manufacturing employment shocks depended on the characteristics of the industries. In section 3.3 the data are used to see whether employment growth in different cities was significantly affected by their initial industrial specialisations, or whether other factors were more important. Lastly, in section 3.4 we use the data to examine the extent that urban areas are becoming more or less industrially diversified through time.
Key Results
The results in this paper suggest that the employment dynamics of New Zealand’s smaller urban areas are quite different to those of the larger ones. At the start of the period large and small urban areas had quite different types of manufacturing industries, with smaller areas disproportionately focussed on industries that processed rural products.
Almost all urban areas suffered from the long manufacturing downturn that started in the late 1970s, but when small areas lost jobs in their rural-processing manufacturing industries they found it much harder to create new jobs in different industries than when larger ones lost jobs in their manufacturing industries. Manufacturing job losses in most large and medium sized urban areas did not reduce employment overall because employment in other industries expanded, particularly in the personal and professional service sectors.
The different response to the loss of manufacturing jobs accentuated the differences in the employment patterns of different sized urban areas. Small and medium-size areas are now, relative to large areas, much more specialised in manufacturing than they used to be and undertake very different types of manufacturing than large areas. They have also suffered because the sectors that have expanded nationally, such as the finance or the professional services sectors, have disproportionately expanded in Auckland. In theory, accountants could have displaced manufacturing workers in Wanganui, Invercargill, or Napier. In practice, they have not, because accounting firms prefer to be located in Auckland. Since the work of many of the new expanding industries is best done in big cities, the sectoral shift of the economy has made it difficult for many small and medium sized urban areas to expand.
The economies of most cities and towns became more diversified as manufacturing declined and service sectors expanded. In this paper we created a formal measure of the extent that cities and towns have diversified rather than specialised and show there are only a few examples of urban areas that have become more reliant on specialist industries since 1976. Rather, most areas became more diversified, and more like each other. Small and medium sized urban areas with distinctive employment patterns are less common than they were. As migration between areas is easier when all areas have similar jobs, the reducing importance of city-specific industries may have catalysed the shift of jobs from slow-growing areas to climate-favoured fast-growing areas.
The analysis in this paper leads to a number of policy-relevant conclusions. Three key ones stand out. First, the decline of manufacturing and the increasing importance of several new service industries has tended to favour large cities, as there has been a rapid expansion in sectors such as the professional services sector that prefer to be located in big cities. New Zealand’s experience is consistent with overseas trends. It is probable the disproportionate growth of large cities has occurred because these industries derive greater agglomeration benefits from locating in big centres than previously important industries such as manufacturing. Global trends like these are unlikely to be overcome by regional interventions aimed at encouraging the development of industries in locations where agglomeration benefits do not exist.
Second, the speed at which urban areas recover from negative employment shocks to their specialist industries depends on the type of industry receiving the shock. For instance, it seems to be much more difficult to recover from adverse shocks that hit rural processing industries than shocks that hit other types of manufacturing industries. If the government wishes to help regional economies recovering from employment downturns, it should recognize that the transition path out of some industries is harder than others.
Lastly, the paper has focussed attention on the similarities and differences of small and large urban economies. The most obvious similarity is that all towns produce similar non-tradeable goods – goods that have a low cross-city location quotient variance – whereas they produce different specialities. For this reason, government programmes aimed at enhancing the way non-tradeable businesses improve their productivity are likely to produce the widest regional benefits, as they have the potential to improve productivity in many sectors everywhere. There is growing recognition of this principle in regional development strategies around the globe