The surge: Details

This page provides detail on the surge period, from 1993-94 to 2003-04. It looks at:
  • the numbers on the productivity surge
  • explanations for the surge in terms of changes in output and input growth
  • industry contributions to the surge in productivity growth
  • underlying reasons for the surge
  • what the surge meant for growth in living standards

The numbers

ABS estimates of the three forms of productivity -- labour productivity (LP), capital productivity (KP) and multifactor productivity (MFP) -- are shown in Figure 1
(16-industry market sector) and Figure 2 (12 selected industries). For coverage of the 12- and 16-industry groupings, see Measurement of productivity

Figure 1 is only included for completeness. I will focus on Figure 2, because of its coverage of the baseline period. The average annual rates of productivity growth for the baseline and surge periods are shown on the chart.
Index, 1994-95 to 2014-15, 2013-14=100
Numbers in the chart are annual average rates of growth (per cent per annum). They are calculated from one productivity peak to another productivity peak
Data source: ABS Cat. No.5260.0.55.002

Index, 1973-74 to 2014-15, 2013-14=100
Numbers in the chart are annual average rates of growth (per cent per annum). They are calculated from one productivity peak to another productivity peak
Data source: ABS Cat. No.5260.0.55.002

In comparison to the baseline period, the surge period showed:
  • an increase of over 150% in the rate of MFP growth
  • an increase of around 50% in the rate of labour productivity growth
  • a move away from negative growth in capital productivity to essentially flat capital productivity.
While there was a strong decline in capital productivity in the baseline period, most of the decline occurred from 1973-74 to 1982-83. The annual rate of growth in capital productivity from 1982-83 to 1993-94, however, averaged 0.39%. From this perspective, the improvement in capital productivity in the surge period was not as marked.

Changes in output and input growth

One step in understanding productivity trends is to examine the proximate or immediate contributors in terms of relative movements in output and inputs. For example, the rate of productivity growth could increase because output accelerated more than inputs or because output decelerated less than inputs. There is different information in these possibilities. 

Table 1 shows the essential proximate reasons for the productivity surge. Output accelerated rapidly (an additional 1.8 percentage points of annual growth compared with the baseline period) , but without the same requirement for increased input growth: 
  • the rate of MFP growth accelerated by 1.2 percentage points because the output acceleration was three times the input acceleration (1.8pp output acceleration and 0.6pp combined input acceleration). 
  • labour productivity accelerated 1.1 percentage points (1.8pp output acceleration less 0.7pp labour input acceleration). 
  • capital productivity accelerated 1.6 percentage points (1.8pp output acceleration less 0.1pp capital input acceleration).

Compound average annual rates of growth (per cent per year, columns 1 and 2) and percentage points (column 3)
Source: Calculations from ABS estimates (ABS Cat. No.5260.0.55.002)

That is, the economy -- or at least a major part of the private sector -- was able to expand rapidly, based on improved efficiency of input use. The improvement in productivity of labour and capital were both strong, but capital productivity showed greater acceleration.1 

The industry contributions

A second step in getting to understand the productivity surge is to look at the industry sources of the productivity surge. Data restrictions make this task a little difficult.

The industry MFP growth rates over 1993-94 to 2003-04 are presented in Figure 3.


There are two prominent features.

First, there was strong productivity growth in a number of services industries -- much stronger than in manufacturing, which has traditionally been seen as the principal source of productivity growth.

Second, it was a good period for productivity growth in agriculture.

Compound average annual rates of growth (per cent per year, columns 1 and 2) and percentage points (column 3)
Source: Calculations from ABS estimates (ABS Cat. No.5260.0.55.002)

However, industry productivity accelerations  -- changes in rates of productivity growth from the previous baseline period -- are of greater interest in the quest to explain the productivity growth surge.

The latest available ABS estimates are used in Table 2 to show the acceleration in industry MFP in the surge period (1993-94 to 2003-04). The prior comparison period, 1989-90 to 1993-94, is quite short as a representation of the baseline period, but the latest ABS series of industry productivity estimates only goes back as far as 1989-90. 

There are other sources we can use to check any impressions formed.

Contributions to the acceleration in the productivity of the 12-industries group are a combination of the acceleration in an industry's productivity and the relative size of the industry. (If industries A and B have the same productivity acceleration, but A is twice as big as B, industry A's contribution to the 12-industry group acceleration will be twice that of B's.)

Table 2 suggests the main contributors to the surge in productivity growth were:
  • Wholesale trade, with by far the strongest acceleration (6.3 percentage points) 
  • Construction
  • Accommodation & food services (its contribution weakened by its relatively small size)
  • Manufacturing (small acceleration but the largest industry)
  • Agriculture. 

Acceleration in rate of MFP growth in the surge period, 1993-94 to 2003-04, compared 
with the period 1989-90 to 1993-94
Industry value added in 1993-94 is used to calculate relative size
Source: My calculations from ABS estimates (ABS Cat. No.5260.0.55.002 and Cat. No. 5204.0)

The Productivity Commission (PC 2009) compared industry MFP growth rates over 1993-94 to 1998-99 with rates over 1988-89 to 1993-94, using an earlier series of ABS estimates (with slightly different industry definitions) that started earlier, but was subsequently superseded. The order of industry accelerations was similar to the one shown here, although manufacturing and retail trade were more prominent. 

Using that previous ABS series, Parham (2004) also showed that productivity accelerated in the finance sector,  but from 1989-90 rather than from 1993-94. 

In summary, while a definitive picture cannot be formed, the main productivity accelerations and contributions to the overall surge in productivity growth appear to have come from:
  • wholesale trade
and otherwise not necessarily in order
  • construction
  • manufacturing
  • agriculture.
The underlying reasons                                 

The truth is there is no way to explain the productivity surge definitively. For a start, there are too many difficulties in defining and measuring variables (for example, how do you measure the intensity of microeconomic reforms in different years?). 

It is a matter of establishing the most plausible explanation(s) from a careful assessment of available evidence. To me, the key plausibility test, given the kind of evidence that can be marshalled, is whether an explanation fits the industry pattern of productivity acceleration.

Three principal explanations have been put forward:
  • microeconomic reforms
  • technological advances, in the area of information and communications technologies (ICTs)
  • increased work intensity.
Commentators have tended to focus on one or the other. But, in fact, there are connections between these explanations.

I will outline the nature and plausibility of each of them.

Microeconomic reforms

'Microeconomic reform' is another one of those terms that can mean different things to different people.

My take is that micro reforms:
  • take effect in specific product and factor (labour and capital) markets 
  • bring greater emphasis to the role of markets and less to the role of government interventions in guiding what is produced in the economy and how it is produced
    • or, at least, introduce new (often 'market friendly') ways to pursue policy objectives that reduce the unintended adverse effects of existing policy approaches
  • serve an underlying objective to improve the living standards of Australians as a whole and over the long term through additional growth in real income 
  • serve this objective by generating production gains (generating more income) and consumption gains (increasing the consumption opportunities from given income)
  • generate production gains by fostering the use of labour, capital and other resources in activities and in ways that result in more total output and income.2
Some examples of areas of microeconomic reforms are provided in Box 1.

Box 1: Areas of microeconomic reforms

Microeconomic reforms generally give more emphasis to markets in determining what is produced in the economy and how it is produced. 

Some key areas of reform, introduced in progressive phases since the 1980s, have been:
  • less capital market deregulation
  • phased reductions in trade restrictions
  • commercialisation (and some privatisation) of government business enterprises
  • less unnecessary business regulation
  • less labour market regulation
  • pro-competitive regulation of natural monopolies

From the outset, the production gains associated with microeconomic reforms were about improving productivity. The intention was to improve productivity by encouraging 'better' use of resources in given production activities (technical efficiency) and by facilitating the allocation of resources to activities in which they are 'better' utilised -- where 'better' means generating more total output and income.

Microeconomic reforms were introduced progressively to meet various structural problems in the Australian economy -- prevalence of small-scale production facilities, outdated technologies, poor management and work practices, misallocation of investment and employment and a production culture that was inward looking (in an international sense) and resistant to change. These structural problems left the Australian economy weakly placed to prosper in a more competitive and globalised world.

Microeconomic reforms were to operate in various ways -- especially open businesses to more competitive pressures, reduce unnecessary business regulation and reduce inflexibilities in the operation of factor (capital and labour) markets. The intention was to sharpen market incentives to be more productive, improve the allocation of resources and encourage a more dynamic and outward-oriented business culture.

There have been challenges in marshalling evidence on the effect of microeconomic reforms on Australia’s productivity performance. It is difficult to characterise the strength of a phased and multifaceted program of reforms over time, let alone assign it a quantitative measure. Isolating the effects of reforms from other influences (such as macroeconomic reforms) is also difficult.

Some of the evidence in favour of a link between reforms and productivity came as part of the assessment of the case for proceeding with a program of reforms. General equilibrium modelling was the main vehicle. These modelling exercises showed substantial gains in economy-wide and long-term effects in the form of additional GDP and consumption. However, they were prior assessments of assumed initial impacts, rather than post-implementation evaluations of the effects of reforms on actual productivity trends.

In a post-implementation assessment, the Productivity Commission (PC 1999) noted the productivity surge was largely confined to Australia, suggesting the uplift in performance had domestic origins. It related the industry locus of productivity growth to the implementation of reforms and also provided industry and firm-based case study evidence. It found that some of the structural problems in the Australian economy had eased – resources were being more productivity allocated, production was becoming more specialised, management practices and workplace arrangements were improving, use of up-to-date technologies and innovation had increased, and the use of workforce skills had lifted.

Parham (2004) presented updated evidence that highlighted new service industry contributors to the productivity surge. He also reviewed evidence from cross-country studies (by the OECD and IMF) and several academic studies of the Australian economy. With one important exception (see below), these provided support for policy reforms having a strong positive effect on Australia’s productivity growth.

Information and communications technologies (ICTs)

The role of ICTs in productivity growth came to attention in explaining the elevated productivity growth of the US in the 1990s. Initially, the distinction between ICT equipment and internet services and related start-up companies was lost. But the focus did turn to equipment (essentially connected computers and software), especially as the ‘tech boom’ turned to ‘bust’.

ICT equipment could lift productivity in three ways:

  • technological advances in the production of ICT equipment enables multifactor productivity gains in equipment manufacturing industries
  • investment in use of ICTs can raise the capital stock, providing a capital deepening (increased ratio of capital to labour) source of labour productivity growth in a diverse range of using industries
  • the use of ICTs can provide a platform for ICT users to develop their own innovations in products and processes, as part of 'businesses transformation', leading to multifactor productivity gains in using industries.
These mechanisms are interrelated. Production gains in ICTs led to lower prices, which raised investment. Investment was also related to the extent of transformation gains that could be realised.

Parham, Roberts and Sun (2001) took a standard growth accounting approach to assess the latter two effects in Australia. (There is little ICT production in Australia.) This approach was updated in PC (2004) and then Parham and Segarajasekaran (2016). The latter found an IT capital deepening contribution of 0.8 of a percentage point to annual labour productivity growth of 2.9 per cent between 1993-94 and 2003-04. That's big.

Parham and Segarajasekaran found the strongest IT and productivity links in Finance and insurance services, Information media and telecommunications, Wholesale trade and Retail trade. They also found some casual evidence that use of IT had improved the productivity of non-IT capital. In particular,  the ICT platform could enable firms to reduce their requirements for land and buildings. Think of ATMs, hone and internet banking and the closure of bank branches.

Other analytical approaches have to be used to measure MFP gains associated with business transformation. The growth accounting technique cannot isolate any ICT-related gains from other sources of MFP gains.

Based on its own econometric analysis and a review of other studies, the Productivity Commission (PC 2004) settled on an ICT contribution of around 0.2 of a percentage of annual MFP growth during the productivity surge.

ICTs appear to have a strong link to the industry footprint of the productivity surge. In particular, Johnston et al (2000), identified the transformational role that ICTs enabled in the wholesale and retail industries. ICTs have also been important in promoting stronger productivity growth in the finance industry, as well as media and telecommunications.

The OECD cross-country study, mentioned above in relation to policy reforms, provided a further key insight. The OECD’s Growth Project was set up essentially to understand why the US’s rate of productivity growth had accelerated in the 1990s, while that of Europe had not. This outcome went counter to the favoured model of 'catch-up', whereby European countries could count on faster productivity growth as they caught up to the productivity leader, the US.

The study found ICTs played a central role. However, production and use of ICTs alone were not sufficient to explain the performance differences. After all, some European countries also produced ICTs and the technologies were available for use in all European countries.

The OECD (2003) found there were particular policy and institutional features that separated the good performers from the poor performers. (Incidentally, Australia was identified as one of a very few countries to have benefited from the transformational use of ICTs.) The factors the OECD identified as driving and enabling ICT-related productivity gains were competition, trade exposure and flexibility in product and factor markets.

This conclusion reinforces the view that microeconomic reforms in Australia played a joint role, along with the ICT ‘technology shock’, in fostering much stronger productivity growth.  

Work intensity

John Quiggin has consistently put forward the view that the role of microeconomic reforms in Australia’s productivity surge has been misplaced. In his view, microeconomic reforms weakened workers’ bargaining power, leading to increased work intensity—longer hours and increased pace of work. He has contended increased work intensity has been mismeasured as productivity growth because of unrecorded increases in hours worked and because work had reached an unsustainable pace (Quiggin 2000, 2001).

I examined Quiggin’s case in Parham (2004).

His contention is partly true. Microeconomic reforms would have increased the intensity of work, perhaps mainly in reducing overmanning and avoidable on-the-job ‘leisure’. To this extent, the increase in work intensity would be a genuine source of productivity improvement.

The pertinent issue is whether the increased intensity had become unsustainable, as Quiggin claims. While Quiggin cited evidence that workers’ stress levels had increased, this did not establish that the pace of work had become unsustainable. Moreover, there is other evidence that the trend toward longer hours of work had petered out by the beginning of the productivity surge period.

Furthermore, additional hours worked would not be mismeasured -- at least in principle. The ABS hours worked series is sourced from a household survey in which respondents -- the workers -- are asked to nominate their hours of work. This would include overtime — paid and unpaid.

On top of these points, the work intensity argument does not rest easily with the industry locus of the productivity surge. It is by no means clear that increased work intensity was responsible for the productivity acceleration in the wholesaling industry, as the prime example. Hours worked in wholesale trade declined during the productivity surge period4 as operations were streamlined and became more technology enabled.

Quiggin (2004) predicted the appearance of increased productivity in the 1990s would pass. He claimed (Quiggin 2011) that the productivity slump in the 2000s vindicated this position. However, again, the industry locus is troublesome. The slump was particularly centred on the mining industry (see The slump). Did Australia’s productivity growth disappear because mining workers found a way to reduce the pace of work during the mining boom? I don’t think so. 

The consequences for living standards

Productivity is an intermediate objective, which can be pursued in the quest to increase living standards. Productivity growth is the key source of improvements in living standards over the long term.

The importance of productivity growth in this context is aptly captured by the often-cited words of Paul Krugman:

    'Productivity isn't everything, but in the long run, it is almost everything. A country’s ability to improve its standard of living over time
    depends almost 
entirely on its ability to raise its output per worker’ 
    P. Krugman (1992), The Age of Diminished Expectations: U.S. Economic Policy in the 1980s, MIT Press, Cambridge.

The importance of productivity in Australia's growth in living standards can be illustrated by using a simple relationship between productivity and average income. It takes advantage of the fact that GDP represents both output and income generated in the economy (Box 2).

In words, the growth in average income is equal to the sum of growth in GDP per hour and growth in hours per capita.

GDP per hour is a broad measure of labour productivity.

Hours per capita (across the whole population) is referred to as 'labour utilisation'. It captures demographics (the proportion of the population of working age), labour force participation, the employment/unemployment rate and average hours worked.3

Box 2 Decomposition of growth in average income

The growth in average income can be decomposed into contributions from productivity growth and growth in labour utilisation.

This decomposition of growth in GDP per capita illustrates one aspect of the Krugman adage -- productivity isn't everything. Increases in labour utilisation are an alternative source of improvement in living standards. Consequently, a measure that increased the employment rate, but had a neutral (or negative) effect on productivity, would (or could) improve living standards.

Using this decomposition formula, we see in Table 3 that GDP per capita accelerated by 1.1 percentage points from the baseline to the surge period.5 That's a 70% increase in the rate of growth in average income. Labour productivity growth accounted for two-thirds of the acceleration.

Growth rates (per cent per annum) and percentage point contributions
Calculated as differences in natural logs
Source: My calculations based on data from ABS Cat. No. 5204.0

In short, additional productivity growth in the surge period brought a substantial increase in the rate of growth in Australian living standards.

To finish off, the decomposition also illustrates the other aspect of Krugman's adage -- productivity is almost everything in the long run. When the decomposition is applied to the entire period of the last four decades, GDP per hour worked accounts for very close to all of the growth in GDP per person. [Link]

  1. From Table 1, the acceleration in labour productivity, 1.11 pp, is the acceleration in output (1.80 pp) less the acceleration in labour (0.69 pp). The acceleration in capital productivity is 1.79 pp (1.80 pp less 0.01 pp). 
  2. See PC (1999), Chapter 2 for the context, objectives, nature and implementation of microeconomic reforms. Banks (2012) discusses a more general notion of 'reform' and what makes for successful reform.
  3. H/P = W/P*R/W*E/R*H/E where the first term on the right hand side is the proportion of the population of working age, the second is the participation rate of those of working age, the third is employment rate of those participating and the final term is the average hours worked by those in employment. 
  4. The ABS productivity data records a 16% decline in hours worked in wholesale trade between 1993-94 and 2003-04. 
  5. The 1970s brought some major changes in labour utilisation, when unemployment grew. I consider the period from 1981-82 to be more representative of the baseline period. Taking the growth rate in average income from 1974-75, instead of 1981-82, makes little difference. However, the contribution of labour productivity increases and the contribution of labour utilisation turns negative.

Key references