RDP 9505: Labour-Productivity Growth and Relative Wages: 1978-1994 1. Introduction

During much of the 1980s, labour-productivity growth in Australia was unusually slow (see Figure 1). Between March 1983 and June 1991 (which are equivalent points in the business cycle), output per hour worked in the non-farm economy increased at an annual rate of just 0.68 per cent per year. This compares with average annual labour-productivity growth of 1.34 per cent over the previous five years, and 2.51 per cent growth over the subsequent three years. This period of slow labour-productivity growth occurred at the same time that employment was growing rapidly. The most frequently offered explanation for this combination of favourable labour-market outcomes but low productivity growth is that the Prices and Incomes Accord held down real wages and led to a substitution of labour for capital.[1] The growth of the service sector and measurement problems are also sometimes cited as explanations for the 1980s productivity slowdown.

Figure 1: Labour Productivity in the Non-Farm Sector
(March quarter 1978 = 100)
Figure 1: Labour Productivity in the Non-Farm Sector

This paper uses industry-level productivity and wages data in an attempt to improve our understanding of labour-productivity outcomes over the period since 1978. It does not contradict the explanation based on changes in the relative price of labour, but argues that the story is richer than that suggested by the simple factor-substitution explanation. The paper pays particular attention to the slowdown in labour-productivity growth over the second half of the 1980s. Over that period, four industries – construction, wholesale and retail trade, finance, property and business services and recreation, personal and other services – experienced declines in the level of labour productivity. The paper examines possible reasons for these declines and examines the contribution of these industries to the aggregate productivity slowdown.

The industry data raise a number of interesting issues. Foremost amongst these is the issue of how to measure output in the service industries. This problem is graphically illustrated by the wholesale and retail trade industry. Despite the adoption of new technologies and rationalisation within the industry, the measured level of labour productivity fell over the second half of the 1980s. To a significant extent this fall was the result of the deregulation of shopping hours, which led to an increase in opening hours and employment. While deregulation is unlikely to have led to more goods being processed through the checkout (the standard measure of output), it certainly made shopping more convenient. While statisticians attempt to make adjustments for improvements in the quality of goods, no adjustments are made for improvements in the quality of services. The result of this is that deregulation of shopping hours led to a reduction in measured output per hour worked, but to an increase in many people's living standards. As the service sector continues to expand, contradictions of this sort will become more frequent.

While measurement problems in other service industries adversely affect the measured level of real output, they can only explain the productivity slowdown if the service sector's share of total employment increased substantially or, somehow, the measurement problems became worse in the 1980s. There is some evidence that measurement problems did in fact become more severe in the finance, property and business services sector and, in particular, in the wholesale and retail trade industry. Measurement problems appear to have played a much smaller role in explaining the slowdown in productivity growth in the recreation, personal and other services sector. In this sector, compositional shifts appear to have been important. In addition, declines in real product wages allowed rapid employment growth, despite the fact that the average level of labour productivity of the new workers was less than the average level of the existing workers.

Outside the service sector, the construction industry, and to a lesser extent the manufacturing industry, also made significant contributions to the 1980s slowdown. Working in the other direction, faster rates of productivity growth in electricity, gas and water, communications and transport and storage acted to push up productivity growth. These favourable effects were, however, more than offset by developments in other sectors.

The industry data also provide some insight into the relationships between relative productivity performance and wage outcomes. Under enterprise bargaining arrangements, wage increases for individual enterprises and industries are increasingly justified in terms of the individual firm's or industry's productivity performance. Such a relationship between productivity growth and wages has an obvious appeal. Ultimately, however, differential rates of productivity growth between industries do not appear to lead to substantially different rates of increases in wages across industries. Instead, differences in productivity growth affect relative prices; slow productivity growth in hairdressing does not lead to stagnant real wages for hairdressers, but instead to an increase in the relative price of a haircut. The examination of the wage data also suggests that real output growth in the finance, property and business services sector has been significantly underestimated.

The remainder of the paper is structured as follows. Section 2 examines trends in labour-productivity growth over the period from March 1978 to June 1994 and the contribution of various industries to changes in labour-productivity growth.[2] Section 3 then analyses productivity trends in the wholesale and retail trade, recreation, personal and other services and construction sectors. Section 4 follows with an examination of the interactions among wages, productivity and prices using industry-level data. Finally, Section 5 provides a summary and concluding remarks.


See EPAC (1989), Dowrick (1990), Dixon and McDonald (1992) and Phipps, Sheen and Wilkins (1992). [1]

The choice of time period is governed by the availability of data on hours worked by industry. [2]