The consequences of auto industry decline: The Commission’s view

The PC (2015) defines its role as providing ‘independent research and advice to government on economic, social and environment issues facing the welfare of Australians’. The PC (2002) has conducted one previous inquiry into the auto industry as well as an earlier study by its predecessor, the Industry Commission (Australian Government, 1990). Many of the claims in PC (2014a) repeat earlier views that ‘there is nothing special about the passenger vehicle industry’ in terms of its economic contribution relative to other sectors of the economy (Australian Government, 1990: 2). Industry, trade unions, state governments and some critical scholars have often expressed radically different views. For example, some have urged a more interventionist style of government in regional development, including efforts to encourage new investment as well as the rejuvenation of existing industries (Australian Manufacturing Workers Union (AMWU), 2014; Reference SpoehrSpoehr, 2014; Reference TonerToner, 2013).

In its most recent inquiry, the PC (2014a) concludes that the AUD30 billion spent by governments on ‘transitional assistance’ from 1997 to 2012 ‘forestalled’ but did not prevent the announced closure of auto assembly operations in Australia. It frames these closures as part of a necessary ‘structural adjustment’ in the national economy. This adjustment, it claims, will lead to the closure of most of the auto components suppliers in Australia and the gradual loss of up to 40,000 jobs as an ‘upper bound estimate’. This is based on assumptions that 80% of workers in auto assembly and 40% of workers in auto components will be retrenched (PC, 2014a: 198–199). The PC acknowledges that the closures will generate significant social costs, forecasting that many workers will suffer loss of income, lower employment security and, in many cases, unemployment. However, it emphasises its forecast that about two-thirds of retrenched workers would have found a new job within 12 months of redundancy (PC, 2014a: 2).

This forecast is influenced partly by a major study of retrenched workers following the closure of Mitsubishi’s operations in Adelaide in 2004, which led to the loss of around 1100 jobs (Reference Beer, Baum and ThomasBeer et al., 2006). The PC (2014a) focuses on this study’s finding that 74% of workers had found new employment 30 months after retrenchment (PC, 2014a: 310), using this as a rough basis for its forecast that two-thirds of auto workers retrenched during the current transition will find new employment within 12 months (PC, 2014a: 25–26).

The PC’s forecast is also based on its ‘whole-of-economy’ CGE approach. It relies upon a version of CGE called the Monash Multi-Regional Forecasting (MMRF) model. MMRF is ‘multi-regional’ in that it assembles national economic data from eight separate regions (six Australian states and two territories). Data for each region come from Input–Output Transaction Tables (IOTTs) provided by the Australian Bureau of Statistics (ABS), fiscal accounts for each state and territory government, and measures of household income and spending. The IOTT for each region comprises 64 sectors, including 21 manufacturing sectors.

An IOTT is a way of measuring total output in an economy generated by the demand required to produce additional units of output in each of its constituent sectors. In theory, one can use an IOTT to forecast that each additional dollar of output in one sector will generate y dollars of aggregate output. However, the PC (2014a) is highly critical of relying solely on IOTTs to make forecasts about the impact of increasing or reducing sectoral output, citing an earlier report by a PC staffer which emphasises the regular ‘abuse’ of IOTTs by vested interests: ‘Abuse primarily relates to overstating the economic importance of specific sectoral or regional activities … In particular, these applications fail to consider the opportunity cost of both spending measures and alternate uses of resources, and may misinform policy-makers’ (Reference GrettonGretton, 2013: 1). We agree that IOTT results can be presented in a misleading way, especially by those industrialists who have a vested interest in appealing for government subsidies. Closer scrutiny often reveals that output multipliers are less significant than claimed (Reference BarnesBarnes, 2015).

Because of this weakness in IOTT analysis, the PC has modified the use of IOTTs within its broader CGE framework (Reference GrettonGretton, 2013: 19). IOTT analysis traditionally assumes fixed prices for factors of production based on constant factor supply. If we assume that factors are scarce, then factor prices must be relative. Taking on board this modified assumption, the PC’s CGE framework assumes that the supply of labour is given by the working age population and labour force participation rate and the supply of capital determined by the average national rate of interest. The CGE framework contains a further four key assumptions:

1. 1. Labour markets have several characteristics of perfect competition;

2. 2. Economic distribution, including regional inequality, is subservient to gross domestic product (GDP);

3. 3. Capital is perfectly mobile within Australia but perfectly immobile internationally;

4. 4. Technology is of secondary importance in influencing capital productivity.

On the first point about labour markets, the MMRF assumes that individual wage-seekers behave in a way predicted by neoclassical economic simulations of perfect competition. In their model, all workers are assumed to know where the best alternative jobs are and can move to wherever the best wages are available. Real wages are determined by the productivity of individual workers, with the most competitive workers seeking out and obtaining the highest paid jobs (Reference GrettonGretton, 2013: 21).

On the second assumption, CGE downplays the importance of geographical economic distribution. While the PC acknowledges the existence of socio-economic disadvantage in some regions, it argues that redistributive public investment generally diverts economic resources from activities where they are more efficiently used and should be considered against ‘community benefit’. Its notion of ‘community’ refers to Australia’s national community, signified by the ‘costs and benefits to the Australian community as a whole’ that result from any change in government policy (PC, 2014a: 29). While the PC (2014a) sometimes refers to local or regional communities by, for example, referring to the places in which disadvantaged workers live (pp. 230–231), its idea of community is primarily national in character and ultimately measured by the level of GDP. Thus, the underlying assumption is that economic distribution is subservient to policy settings that can encourage higher levels of GDP.

Third, the MMRF model assumes that capital is perfectly mobile within Australia but relatively immobile internationally. In this framework, the marginal product of capital is fixed and profits are proportional to the capital-to-labour ratio in each IOTT sector (Reference GrettonGretton, 2013: 21). Thus, capital will flow to the sector or region in which the capital-to-labour ratio is lowest, assuming that capital is fully mobile between all Australian states and territories. To some extent, the CGE model can simulate Australia’s place in the international economy by incorporating the trade balance – that is, by incorporating income flows that are the result of past Foreign Direct Investment (FDI) inflows and outflows – but it does not incorporate the cross-border investment flows themselves, including financial decision-making based upon expected investment flows.

Fourth, the assumption that the marginal product of capital is fixed is underpinned by a further assumption: that technology has a weak influence on capital productivity and growth. In reality, applied scientific knowledge, technology, skill and organisational expertise can strongly influence marginal product. While neoclassical economics has long attempted to incorporate technological change into growth models – most notably through the treatment of total factor productivity in Cobb–Douglas production functions – CGE does not incorporate this type of measure. This issue is partly recognised by the PC. For example, Reference GrettonGretton (2013) makes a clear distinction between ‘inter-industry linkages’, measured via IOTTs, and ‘spillovers’ which are ‘more difficult to identify and quantify’ (p. 2).

There is a vast international literature on spillovers which refers to the transfer of technology, codified knowledge and skills between industries and firms, and less tangible attributes like expertise, management methods and tacit knowledge. Most of the literature attempts to quantify these transfers, with much of it focused on the impact of FDI in developing countries (Reference Blomström and KokkoBlomström and Kokko, 1998; Reference Colen, Maertens and SwinnenColen et al., 2008; Reference Havranek and IrsovaHavranek and Irsova, 2012). But the PC (2014a) seems to be unconvinced by any of this literature:

In the following section, we outline the range of significant problems with the PC’s forecasts. We point to the unrealistic nature of several key assumptions in its CGE framework, its misreading of previous studies of retrenched workers and, in particular, the key omissions from its analysis about the quality of present and future work.

Challenging the Commission’s forecasts

Before critiquing the model’s key assumptions, it is worthwhile noting that the implication that CGE reflects ‘general equilibrium’ analysis is misleading. General equilibrium theory is a wing of neoclassical economics based on highly abstract reasoning about the conditions necessary for Pareto optimality: a purely theoretical situation in which it is not possible to increase the welfare of any individual within a given economy without reducing the welfare of others. Such a situation assumes that all economic resources have been put to use and cannot be allocated any more efficiently (Reference Cross and StrachanCross and Strachan, 2001; Reference KirmanKirman, 1992).

Despite its name, CGE has little directly to do with this theoretical tradition. Instead, CGE draws links between given levels of national output (GDP) and a country’s national accounts for household spending, private investment, government investment, international trade as well as sectoral data from IOTTs. CGE models what kinds of ‘adjustments’ or ‘shocks’ in these accounts correspond to different GDP levels (Reference Mitra-KahnMitra-Kahn, 2008: 59). In the example of its inquiry into auto industry assistance, the PC uses the MMRF model to contrast the impact of ongoing auto manufacturing output on future GDP in comparison to future GDP in the industry’s absence (PC, 2014b). By linking national output to consumers, firms, governments and trade, this approach is reminiscent of the traditional Keynesian macroeconomic framework and has very little to do with the neoclassical theoretical tradition of general equilibrium.

However, the problems of using a CGE model to analyse industry policy are more substantive than its misplaced name. Most of the key assumptions outlined above are highly unrealistic and some are likely to have led to an underestimation of the auto industry’s impact on the national economy as well as specific regions. First, the assumption that workers are perfectly mobile geographically and solely motivated by higher wages in different locations is contradicted by literature on retrenched workers. Several findings from the Mitsubishi study cited at length by the PC demonstrate that ex-Mitsubishi workers tended to remain in the same areas of Adelaide. For example, Reference BeerBeer (2008) argued that the Mitsubishi study supported international findings that high rates of home ownership contributed to higher rates of unemployment by reducing labour mobility (Reference OswaldOswald, 1996). Of the Mitsubishi study’s sample, 84% were either home owners or paying off a mortgage, 57% had lived in the area for more than a decade and over 80% did not expect to move out of the area (Reference Beer, Baum and ThomasBeer et al., 2006: 62, 65).

Similar findings have come from other major studies of retrenched workers. Reference WellerWeller’s (2009) research on ex-Ansett workers found that most workers avoided ‘speculative migration’ and only 35% had relocated for new employment 5 years after the airline’s collapse. Mobility was linked to age, gender and occupation, with younger, male cabin crew staff forming the largest proportion of relocating workers, while older workers with families or with blue-collar jobs were far less likely to migrate or relocate their homes to secure job opportunities in new locations (Reference WellerWeller, 2009: 248–250). This suggests that employment options are fewer for workers less able to relocate due to their partners’ work commitments, school-age children or their housing situation.

The PC’s second assumption that geographical distribution is subordinate in importance to higher levels of GDP is consistent with its long-time opposition to regional adjustment funds. There have been several such funds established in response to the automotive closures. For example, in Victoria, the Federal and State Government jointly supported the establishment of regional ‘Innovation and Investment Funds’ in Geelong and in Melbourne’s northern suburbs in response to Ford’s closure announcement in 2013. The Government of Victoria has recently extended this to the south-eastern suburbs of Melbourne where many automotive components manufacturers are historically based. The PC’s assumption primarily reflects an underlying belief that economic distribution is less important than growth and that redistributive policies are an unwarranted burden on growth-oriented policies. The Government of SA (2013) put this argument succinctly in its submission to the recent PC inquiry: ‘It is very likely that most regional adjustment programs would not be supported if assessed against a whole-of-economy perspective but equity is also an important policy consideration’ (p. 16).

The PC’s third assumption that capital is perfectly mobile between regions in Australia’s national economy but relatively immobile between Australia and the rest of the world is clearly untenable and reflects the limitations of the CGE approach. For example, it does not take into account the relative certainty that most of the remaining value in the carmaker’s fixed capital investments in Australia will be transferred overseas which all three Australia-based car manufacturers clarified in their submissions to the PC inquiry (Ford Motor Company of Australia Ltd (Ford), 2013; GM Holden Ltd (Holden), 2013; Toyota Australia (Toyota), 2013).

Similarly, the PC’s fourth assumption – that technology is not an important consideration in forecasting the impact of industry policy – reflects the weakness of the CGE approach. There is evidence from international studies that technology transfers through vertical spillovers – in other words, transfers between different industries – play a significant role in economic growth and development (Reference Blomström and KokkoBlomström and Kokko, 1998). While foreign corporations are likely to prevent competitors in the same industry from adopting their technology or production methods (horizontal spillovers), they are also likely to pass on vertical spillovers through forward linkages (firm to consumer) and backward linkages (firm to supplier), whether by design or as an unintended consequence of their regional operations (Reference Colen, Maertens and SwinnenColen et al., 2008).

Vertical spillovers via backward linkages often depend upon the physical presence of a lead firm’s manufacturing operations. Reference Havranek and IrsovaHavranek and Irsova (2012) collated 3626 estimates from 57 separate studies of vertical spillovers, finding that a 10% increase in ‘foreign presence’ correlated with a 1.2% increase in domestic suppliers’ productivity, although they caution that the relationship is highly dependent upon the characteristics of the host country, the type of industry and the type of FDI (Reference Havranek and IrsovaHavranek and Irsova, 2012: 1389). Auto manufacturing is one such industry in which, given the right institutional arrangements, foreign firms can generate significant spillover benefits for domestic producers. Global carmakers usually dominate ‘relational’ value chains characterised by close collaboration with hundreds of supply firms (Reference Sturgeon, Van Biesebroeck and GereffiSturgeon et al., 2008). They pass on quality standards through design specifications and frequent inspection of product quality. Spillovers occur through product imitation, training workers in supply firms or via Research and Development (R&D) facilities established by lead firms and competition between suppliers who compete for contracts with lead firms (Reference Colen, Maertens and SwinnenColen et al., 2008: 18).

Because these spillovers are generally difficult to quantify in price terms, most cannot be incorporated into a CGE method. This problem seems to reflect the PC’s (2014a) concern when it claims that there are no ‘sufficiently robust’ methods available for measuring spillover effects (p. 87). However, the PC contradicts the logic of this position by claiming that the value of spillovers from Australian auto manufacturing is lower than the opportunity cost of government subsidies to auto firms (PC, 2014a: 13). Clearly, one cannot make such relative ‘economic value’-dependent statements while rejecting the idea that such ‘value’ can be calculated in the first place.

In contrast, many of the submissions to the PC inquiry also acknowledge that spillovers are difficult to quantify but draw opposing policy conclusions about the need for government subsidies (AMWU, 2014: 19; Australian Industry Group (Ai Group), 2013: 8). The former Federation of Automotive Products Manufacturers (FAPM, 2013)^{Footnote 1} used a case study approach to emphasise the importance of auto manufacturing spillovers. It points to examples of technology like Australian-designed engines, brakes and chassis, innovations transferred to other industries like carbon-fibre products and three-dimensional (3D) printing and ‘lean process’ management techniques which are used in transport and banking (also see Allen Consulting Group, 2013). Similarly, the Government of SA (2013) argues that auto manufacturing generates spillovers for the food processing and defence sectors.

The contestable assumptions at the heart of the PC’s modelling suggest that it has underestimated the impact of the industry’s demise. For instance, if workers are less likely to be able to fill appropriate alternative employment opportunities than its CGE model assumes, then forecasts for ongoing unemployment effects are likely to be worse than the PC has predicted. Furthermore, if income flows from future investment foregone due to the assembly plant closures do not eventuate, then there is likely to be less demand for future jobs than the PC predicts, also contributing to higher-than-expected unemployment.

A further problem with the PC’s forecasts relates to its treatment of literature on retrenched workers. As outlined in the previous section, the PC (2014a) cites the Mitsubishi study as part of the rationale to justify its forecast that two-thirds of retrenched auto workers will find new employment within 12 months (pp. 25–26). This reflects a particularly narrow interpretation of this wide-ranging study; it also reflects the lack of attention the PC has paid to other important studies of retrenched workers. Previous studies have found, for example, that older workers, women and workers from some Non-English Speaking Backgrounds (NESB) suffer greater disadvantage from retrenchment (Reference Webber and CampbellWebber and Campbell, 1997; Reference Webber and WellerWebber and Weller, 2001).

The Mitsubishi study found that most workers were older married men with limited formal education. Most were long-term ‘process workers’ on middle incomes, although there was also a significant minority of higher-skilled engineers (Reference Beer, Baum and ThomasBeer et al., 2006: 11). The initial survey of 373 workers found an unemployment rate up to 25%. Of those workers with new jobs, only 4% had secured permanent positions. In all, 35% were employed casually and 15% were on sub-annual, short-term contracts (Reference Beer, Baum and ThomasBeer et al., 2006: 16–17). The report’s findings emphasised ‘social capital’ – defined in terms of the loss of workplace-related social interactions as well as lower-than-average trust towards government and business – as well as mental health and stress indicators (also see Reference Verity and JolleyVerity and Jolley, 2008). While many workers reported that their new jobs were better than before (44% of the sample), Reference BeerBeer (2008) showed most of these workers were moving into more insecure forms of employment. After 18 months of redundancy, 38% had worked at least two jobs in the preceding year, with some working in up to six different positions (Reference BeerBeer, 2008: 325).

A further consideration in examining the PC’s narrow focus on the Mitsubishi study’s unemployment findings is that when Mitsubishi closed local operations, many retrenched workers found new jobs with the same skill or occupational profiles, partly because competitor firms were expanding to absorb market share (Reference Beer, Baum and ThomasBeer et al., 2006). This is a far less likely option for contemporary auto workers, given that all remaining motor vehicle assembly operations will close. There is also likely to be reduced demand for their skills in other manufacturing companies given the decline in total manufacturing employment in recent years (ABS, 2015).

These findings also suggest that a major omission from the PC’s entire framework concerns the quality of work. For instance, its theoretical assumption that technology is not of primary importance in the economic development process is significant as technology spillovers impact upon the occupational and skill profile of labour markets, which impact upon the quality of work generated through manufacturing demand. While the PC (2014a) suggests there is a case for providing targeted government assistance for the most disadvantaged workers, it does not explain what kind of assistance this should be or how much funding the federal government should allocate, beyond a suggestion than these matters ‘are likely to be more appropriately addressed through broader economic and social policies’ (p. 231). Ultimately, this interlude is overwhelmed by the PC’s (2014a) forecast that most ex-auto workers will find new jobs without government assistance: ‘The labour market in Australia is dynamic – many employees lose their jobs in any one year and many people who are jobless are hired’ (p. 26). In the following section, we posit an alternative research agenda for exploring the impact of the auto industry’s demise by focusing on work quality especially in closure-affected cities and regions. We introduce the concept of ‘social spillovers’ – quality of life indicators influenced by the presence or absence of particular industries – in order to frame this agenda.

Social spillovers and decent work: The case for a new research agenda

In their research on retrenched Mitsubishi workers, Reference Beer, Baum and ThomasBeer et al. (2006) include analysis on the nature of new employment contracts, including whether or not new jobs were permanent, temporary, full-time, part-time or casual. This is indeed a crucial issue in measuring the quality of new employment. However, observing and measuring this impact require the inclusion of research methodologies ignored by the PC and largely underplayed even in the sociological and geographical literatures on retrenched workers. The social and economic impacts of retrenchment on individuals and households unfold as a process over many years, calling for longitudinal research methodologies to reveal workers’ trajectories (Reference Weller and WebberWeller and Webber, 1999). For example, the study of ex-Ansett workers is one of the few studies of retrenchment to use a longitudinal approach, enabling the researchers to demonstrate quality-of-work impacts. One of the key findings from this research was that participants generally rated their new job to be worse in terms of employment and working conditions, including pay, working hours, job satisfaction and social interaction (Reference Weller and WebberWeller and Webber, 2004: 321).^{Footnote 2}

A vital research and policy agenda is to build upon and extend this approach by framing the quality of work as a subset of, and intrinsically linked to, the quality of life (Reference WatsonWatson, 2012). Such an approach uses the ILO’s DWA as a framework for the development of indicators to measure the social and economic well-being of workers and communities experiencing the impacts of industrial decline. These indicators, which also take account of development theories and principles, can be seen as a way to measure the social spillovers that may be lost through manufacturing divestment or alternatively that might be gained through alternative investment and employment opportunities.

Quality of life relates to a broad range of measures that go beyond the scope of this article. Development theory such as the capabilities approach originally developed by Reference SenSen (1985, Reference Sen2005) and further elaborated by Reference NussbaumNussbaum (2000, Reference Nussbaum2003) offers useful frameworks for analysing complex economic, social, political and environmental dimensions that enhance individuals’ capabilities and their capacity to enjoy a good life. This takes into account income, distribution, gender equality, education, housing, physical and mental health, the quality of the physical environment as well as more subjective or less-easily quantifiable indicators like political freedom, freedom from discrimination and violence, and the realisation of human rights.

The quality of work represents a subset of this much broader social agenda. In response to increased globalisation and the new challenges facing labour movements internationally, the ILO developed the DWA in 1998. The DWA framework involves ‘the promotion of opportunities for women and men to obtain decent and productive work, in conditions of freedom, equity, security and human dignity’ (ILO, 1999). The DWA provides a coherent set of policies for employment promotion and protection, security and income support, the promotion of equality of opportunity and access to rights at work (Reference RogersRogers, 2007). It brings together economic and social policy frameworks and provides an opportunity to be specific about policy goals in national or regional contexts (Reference FieldsFields, 2003). Importantly, the DWA moves beyond employment to capture the experience of all of those who work, including the unemployed, self-employed and household-based and other forms of unpaid labour. It is also a global agenda that moves beyond the concept of legal rights to incorporate basic human rights and principles that are beyond national laws (Reference SenSen, 2000). One of the underlying concerns is that work on a global scale continues to become less secure, with marked increases in short-term contract work, irregular hours, temporary and part-time work and a commensurate decline in stable, full-time jobs (ILO, 2015).

Indicators have been developed to assist in the measurement of achievement towards a DWA for each ILO member country. Some of these DWIs have been developed by ILO statisticians while others have been developed by non-governmental organisations (NGOs) with tripartite support from governments and academic institutions.^{Footnote 3} The main DWIs include familiar measures like working age population, unemployment and youth unemployment as well as a broad range of updated measures, including informal and precarious employment, low pay and working poverty, excessive working hours, occupational injury, social security spending and pension rights, union density, collective bargaining coverage and days lost to industrial conflict. There are also a range of additional DWIs which link to these main measures, including average real wages, minimum wages, under-employment, the gender wage gap, discrimination and public health care, as well as new measures under consideration, including paid annual leave, maternity protection, employment of people with disabilities, sick leave and freedom of association. Furthermore, a range of ‘legal framework indicators’ has been developed to accompany these statistical measures (ILO, 2008). Such an approach goes well beyond the framework adopted by the PC, which treats labour as an exchangeable factor of production in the neoclassical sense.

The DWA is broadly consistent with the ILO’s social democratic tradition and its Polanyian approach, which treats labour as embedded in social relations and social institutions rather than a commodity which can be bought and sold like any ‘regular’ good or service (Reference LercheLerche, 2012). The DWA is also a response to the onset of neoliberal economic policy outcomes since the 1980s, including decentralised labour market policies, less generous welfare assistance for the poor and the unemployed, privatisation of public assets and falling union density.

In the context of the quality of work in Australian, particularly in regions affected by the looming automotive manufacturing closures, studies would need to generate a set of DWIs to compare and contrast work quality in auto assembly and components manufacturing firms today with the quality of alternative work in the future for those workers retrenched during the current transition. They would also need to map these changes over several years in order to construct a longitudinal database, as well as to develop non-workplace-based indicators to incorporate the well-being of retired or unemployed workers and family members. Such indicators can be framed as social spillovers which measure the impact of investment and employment creation on social and economic well-being, in contrast to more conventional spillover measures which focus on the transfer of technology, skills and knowledge. The social spillovers of auto manufacturing, measured by a set of DWIs, can be compared over time with the social spillovers from future work or employment in affected cities and regions.