Labour market models and their use in projecting vocational education and training requirements

Because much of the policy relating to vocational education and training (VET) has continued to evolve since the radical changes experienced in this sector as a result of industry restructuring of the 1980s, many of the objectives associated with policy development within this area have not been as clearly defined or articulated as perhaps would be desirable. For example, it is difficult to locate clear objectives attached to VET policy in either the relevant legislation or the annual reports of the Australian National Training Authority (ANTA). In part this reflects the rapidly changing industry (and therefore, training) environment of the 1980s whereby insufficient time was available for consideration of the appropriate roles of public and private sectors in VET provision.

Similarly, while ANTA has indicated that model-derived employment forecasts are an important component of planning processes in the VET sector, it would appear that nowhere is the role of labour market models and occupational forecasting clearly defined. Historically there has been use worldwide of large-scale economic models for manpower planning and VET purposes for over fifty years.

In part response to policy directives, the energies of the VET sector appears to have focussed energies more on 'organisational efficiency', mobility and equity matters than on identifying which training services offered by the sector are most valuable and relevant to the requirements of industry.

The substantial economic literature on labour market behaviour and policy identifies several reasons for government intervention in VET areas: labour market information has 'public good' characteristics and would therefore be underprovided if left to the private sector sources; capital market imperfections may prevent young and unskilled workers from investing appropriate amounts in VET.

The prevailing view of VET research worldwide is that current and timely information on changing labour market conditions improves the decision-making of all current and potential labour market participants, especially where long lead time in skill acquisition is required.

This information must be available at a highly disaggregated level and should identify likely imbalances between demand and supply in occupational labour markets rather than simply attempt to make employment forecasts.

Such labour market information must build upon both occupational demand and supply data, taking into account workforce outflows, population growth by age and sex, net migration, changes in participation rates, projections of skill categories within new labour supply, and effects of labour supply substitutions between occupations.

This information should also take account of the likely changes in wage relativities and their impacts, as well as the related issue of the time lags required to overcome market imbalances.

A key issue, which appears not to be not addressed in VET-related economic literature or policy statements, is that the value (to the economy as a whole) of removing imbalances in one occupation rather than another, depends upon factors such as the relationship between price and marginal cost for the product concerned.

Although most State training authorities purchase and/or develop their own model-based occupational profiles, the use of labour market models appears to be at best ad hoc and at worst ignored.

Reasons reported for the use of occupational profiles related less to the need for forecast accuracy, but rather took account of factors such as heavy weightings given to expenditure patterns in previous years, political pressures to 'consult' widely and pressure from particular groups (for example, industry training advisory bodies [ITABS]) to be consulted.

A further constraint restricting the use of labour market models is reported to be a lack of staff with the economic and econometric training necessary to apply modelling techniques to VET planning.

It was reported that significant staff cuts have taken place in the VET sector in all States, and that these cuts have had significant impact upon staff with modelling and related quantitative skills.

Economic models used for occupational forecasting should include the following characteristics:

• The separate parts of the model should have overall consistency.
• The model should make use of the informational content of economic theory.
• Parameters describing behaviour within the model should be estimated appropriately from market data.
• It should be possible for independent researchers to replicate results claimed to have been derived from a particular model.

Forecasts obtained using economic models depend upon both how well the model used approximates behaviour in the economy, and upon how well-forecasted are the trends in the outside (exogenous) variables that 'drive' outcomes from the model.

Two types of forecasting exercise may be distinguished: those that build upon the 'most likely' values and trends of outside (exogenous) variables; and scenario analysis, where the consequences of a particular, but not necessarily likely, economic shock (such as a hydrocarbon tax) are investigated.

Whichever types of forecast are derived, it is extremely difficult to provide information on the probability of any particular variable taking a value in a given range. At best, sensitivity analysis may be undertaken to suggest the extent to which outcomes may vary in different situations. Alternatively, historical performance may be taken into account.

For some time Australia has been, and remains, at the forefront of economic modelling developments, both in relation to large-scale disaggregated economy-wide models and to smaller dynamic macroeconometric models.

In particular, the Monash and Murphy models used widely in the VET sector are large-scale disaggregated models with dynamic macroeconomic sectors that substantially meet the model criteria detailed above. The slightly smaller National Institute of Economic and Industry Research (NIEIR) model only partially meets these criteria.

The smaller-scale multi-country McKibbin G-Cubed models are also leading-edge examples and, as with the macro-focussed Treasury TRYM model, also substantially meet the model criteria above, but lack the disaggregated detail so necessary to VET requirements.

Existing Australian modelling methodology does not model labour supply adequately for VET purposes and, at best, gives minimal direct information about likely occupational labour market imbalances.

Both the Monash and Murphy models contain some dynamic aspects, and the Monash model has signalled development of supply-side considerations. Nevertheless, at this stage only the Murphy model offers some insights relating to labour market adjustment behaviour, insights which cannot however, be deduced by the subscribers themselves from the forecast information currently provided, even though this could easily be achieved.

Overseas occupational forecasting methodology, such as that undertaken in Canada, has a far more detailed framework for dealing with adjustment in labour supply and may provide pointers for further developments in Australian labour market models.

The Canadian methodology, however, has significant limitations in relation to demand-side modelling, the level of disaggregation and also to the use of flow projections. Specifically, while labour market flows are predicted, no information is offered about the skills environment in which these flows take.

Notwithstanding the limitations applying to all current occupational forecasting methodologies, it must be understood that a coherent organisation of theory, understanding and data within a modelling framework is generally a superior method of forecasting to one based upon unstructured use of 'expert' opinion on particular markets or sectors.

By worldwide standards, Monash University offers an outstanding resource base for methodological developments of the kind signalled in this report. In addition, there is every indication that these resources are being used in a proactive manner to enhance Monash methodology based upon the identification and understanding of VET sector needs.

1. Drawing on the insights available from models of labour market behaviour and economic policy analysis, the role of the public sector in some VET sector activities, specifically those relating to labour market supply and demand, should be reconsidered, with a view to clarifying:
   • whether the focus of the public sector should be to provide information on projected labour market imbalances in order to improve industry decision-making in the various occupational labour markets
   • under what circumstances projected imbalances can be addressed efficiently through private sector demand-side and supply-side decisions, without public sector interventions in these markets
   • for those occupational labour markets where public sector interventions (other than information provision) can be justified, the basis upon which the relative importance of imbalances in different markets can be established
2. On the basis of the first point above, ANTA should review its definitions of the objectives of VET policy and of the key performance measures that derive from policy objectives.
3. It is desirable that the national and State agencies concerned with the design and application of VET policy maintain within their organisations, some expertise in the application of economic analysis, modelling and forecasting to VET issues.
4. A key research priority is to improve the labour market dynamics of models, especially in regard to the treatment of occupational labour supply and replacement demand, as well as to likely occupational wage adjustments. In this way, the capacity to forecast occupational labour market imbalances will be increased.
5. A further research priority is to develop the model's capacity to indicate the relative value of reducing labour market imbalances for one occupation as opposed to another.
6. It is desirable that the research priorities identified in 4 and 5 above be undertaken in a way that utilises the leading-edge skills of the Australian economic modelling community, especially those exemplified by the Monash and Murphy methodologies.
7. Incentives should be provided for a focussed program of research which, while encouraging further developments on the Monash and Murphy models, should avoid encouraging duplication of research activity between different modelling agencies.
8. The possibilities for collaborative model development should be explored but not to the extent of precluding an ongoing process of model comparison and evaluation.