The authors are employees of the Commonwealth Treasury. We would like to thank Andrew Johnson, Paddy Jilek and Laurie Antioch for earlier work on a similar topic. Of course any errors and omissions are the responsibility of the authors. The views in this paper are those of the authors and are not necessarily those of the Government or the Commonwealth Treasury.
The labour market is a key area which determines how the macroeconomy responds to various shocks and policy changes. Yet it is also an area of great uncertainty and opinions differ on how to treat particular equations. It is quite clear from the work on segmented labour markets that behaviour at the aggregate level can sometimes have very complex roots. TRYM works at a high level of aggregation and is designed to explore the links between the labour market and other macro markets in a concise and coherent way. To do so requires some simplifying assumptions. (Hence TRYM cannot be used to explore questions of detail in the labour market area. For example, there are no relative wage effects in TRYM and no distinction between demographic or occupational groups.) However, TRYM does have significant advantages over simple partial analyses of the labour market to answer broad macro questions. For example, a simple partial analysis of the labour market might conclude that wage changes will only have a relatively minor effect on employment in the short to medium term, and that very large changes in the real wage would be required to eliminate unemployment at any particular time. Full model analysis on the other hand indicates that the employment response is relatively large and that only relatively small changes in real wages are required to reduce unemployment. This follows from the financial market reaction to reduced inflationary pressures coming from the labour market. As the economy has become more open over time, the interest rate and exchange rate effects on activity have become more important in transmitting any given wage shock to unemployment. The model also has the advantage of being able to explore the link between labour market imbalances and imbalances in other areas of the economy.
This paper re-examines the specification of the behavioural equations in TRYM particularly the wage equation, and takes the opportunity to represent some TRYM results on the effects of a change in the NAIRU. This paper is based on work in progress and is presented to obtain feedback on the work done before final specifications are incorporated into the model.
Figure 1: Stylised Representation of the Labour Market in TRYM 10
2.5 The Unemployment/Vacancy (U/V) Relationship or Beveridge Curve 11
* indicates the test has failed at the 5% confidence level.
** inspection of the equation residuals revealed greater volatility in the period prior to 1980. The failure of the test at the 5% confidence level may therefore be data related. The vacancy data used was the ABS spliced vacancy series. The structural break occurs in the early 1980s. 13
Labour Demand 19
3.2 The Private Business Sector 19
3.3 Labour Demand Equation 20
Labour Supply 24
4.3 Labour Supply Equation 25
A TRYM Model Simulation of The Effects of a Lower NAIRU 30
5.1 Long Run Effects 31
5.2 Short To Medium Run Impact 32
5.21 Wages, Employment and Prices 32
5.22 Government Policy Responses 34
5.23 Financial Markets 34
5.24 Demand and Output 35
5.25 External Sector 37
5.26 Net Lending and the CAD 38
5.3 Medium Run Adjustment 39
The deterioration in the performance of many OECD economies in 1970s and 1980s created new challenges for macroeconometric modelling. Macroeconomic models of this time proved unable to deal with supply shocks prevalent in the 1970s (eg) the two oil shocks. As a consequence, there was widespread recognition of the need to focus on the supply side of models. The test for model builders was to develop a framework that could explain short-run fluctuations, whether driven by demand or supply shocks, while ensuring that the model provided a long run solution that was internally consistent, both empirically and theoretically1.
One of the most important aspects of the renewed focus on the supply side has been the re examination of the labour market behaviour spurred on by the rise in and persistence of high levels of unemployment in most OECD economies. During the last ten to fifteen years a number of new theories have been developed to provide insights into how the labour market behaves and why persistence in unemployment might develop. Despite this work, labour market behaviour, particularly wage behaviour, remains a major source of uncertainty on the supply side of macro models. It is also a key area in determining short term responses, for example, of inflation to changes in demand and hence to policy trade-offs and results.2
The TRYM model was designed as a simulation model, based on aggregate data that takes account of the major interactions and linkages in the macroeconomy. TRYM therefore has an aggregated, simplified interpretation of the Australian labour market. This consists of three behavioural equations. These are: a wage setting equation (an expectations augmented Phillips curve), a labour demand equation, and a labour force participation equation (labour supply). In the TRYM model, households or workers make decisions on labour force participation and wage demands (independent from their consumption/savings choice), while firms make decisions about labour demand (employment levels), prices and investment. Investment, prices and employment are estimated jointly to ensure consistency with profit maximisation in the long run given estimated production technology.
The re-examination of the labour market specification presented at the June 1993 TRYM Conference was motivated by the desire to more fully integrate the specification of the various labour market equations and to improve their explanatory power. In particular it was to (by equation):
· Wages: examine the possibility of simultaneity in the wage and price relationship incorporated in the model3; and to explore the possibility that changes in the search effectiveness of the unemployed as evidenced by the changes in the unemployment/vacancy (UV) relationship had led to changes in the link between unemployment and wages. (A number of recent studies both in Australia and overseas have excluded long term unemployment from the unemployment measure used in modelling wages on the basis that the long term unemployed exert little leverage on the wage bargaining process - ie are not search effective.)
· Labour Demand: incorporate information about the demand for labour contained in unfilled vacancies data. Unfilled job vacancies represent the difference between the demand for labour and actual employment. Fluctuations in vacancies may mean changes in the demand for labour do not translate directly into changes in employment. Incorporation of the vacancies data should therefore help to estimate the true impact of factors which affect the demand for labour. In particular, it should help tie down the effect of wages.
· Labour Supply: re-examine the specification of the equation, in particular to explore the effect of demographics on the participation rate and to re-examine the role of after-tax real wages on labour supply decisions.
Given the source of potential changes to the wage and labour demand equations, the specification of the labour market has been extended to include an equation to endogenise unfilled vacancies. This equation captures the outward movement in the unemployment/vacancy relationship or Beveridge curve which occurred in the 1970s.
The paper is arranged in the following manner. Sections 2, 3 and 4 examine the wage, labour demand and labour supply equations respectively. Each section has a short introduction which briefly reviews the theoretical and empirical literature. Section 5 presents a full TRYM model simulation of the impact of a lower NAIRU, and how wage restraint flows through the economy. Section 6 draws some conclusions from this analysis.