Using official data from the Australian Bureau of Economic Statistics and a formal growth accounting framework, this paper shows that the rapid accumulation of information processing and communication technology (ICT) capital over the last two decades in Australia has played a significant role in explaining the impressive, structural acceleration of labor productivity. The following statistical data are also included: household income, expenditure and savings, labor market, fiscal indicators, credit aggregates, capital and financial account, external assets and liabilities, export by commodity group, and so on.


Using official data from the Australian Bureau of Economic Statistics and a formal growth accounting framework, this paper shows that the rapid accumulation of information processing and communication technology (ICT) capital over the last two decades in Australia has played a significant role in explaining the impressive, structural acceleration of labor productivity. The following statistical data are also included: household income, expenditure and savings, labor market, fiscal indicators, credit aggregates, capital and financial account, external assets and liabilities, export by commodity group, and so on.

I. Is Australia a “New Economy”?1

A. Introduction

1. The question of whether Australia is a “new economy”2 has been drawing increasing attention over the recent past. First, Australia has recorded an impressive growth performance over the 1990s, outstripping even that of the US (Table I.1). The post—1995 acceleration of labor productivity growth in the US was the trigger for the burgeoning of the “new economy” literature. In Australia, most studies to date have focused on the role of the wide ranging structural reforms since 1985 in bringing about the acceleration in productivity3. More recently, however, the quest for the fundamental factors underlying Australian productivity growth has focused attention on the growth—enhancing effects of the adoption of information processing and communication technologies.4

Table I.1.

Output, Labor Productivity and Total Factor Productivity Growth, Selected Countries

(Average annual percent change)

article image
Source: OECD (Scarpetta, Bassamini, Pilat and Schreyer. Working Paper n.248, 2000).

Trend GDP per capita, total economy.

Trend GDP per employee, total economy.

Tread TFP, business sector, estimated as Solow residual with time varying factor shares.

2. Another impetus for the interest in determining Australia’s new economy credentials is the weakness of the Australian dollar over most of 2000, which led many commentators to wonder whether the international investment community was penalizing Australia for being an “old economy,” in particular for the lack of a large ICT—producing sector.

3. The objective of this paper is to analyze the contribution of ICT capital to the Australian productivity performance within a formal, growth accounting framework. This is done by using official data released by the Australian Bureau of Statistics, which adds rigor to the analysis, but requires caution when comparing the results with those of other countries due to measurement differences, and with cross—country studies that are based on consistent but somewhat arbitrary definitions of ICT outputs, inputs and prices.5

4. This paper concludes that the rapid process of ICT capital accumulation in Australia has started to pay off in terms of output and labor productivity growth. The ICT capital contribution to growth has increased dramatically in the last decade, and ICT capital now accounts for about two—thirds of the total growth contribution from capital deepening.

5. While most of the labor productivity acceleration of the late 1990s has been due to an increase in the rate of growth of total factor productivity (TFP),6 an important result from the perspective of establishing Australia’s “new economy” credentials is the existence of a positive relationship between ICT capital accumulation and TFP growth across Australian industries. While this is not definitive proof that Australia is benefiting from the positive network externalities associated with the usage of new technologies, the existence of such a relationship, at a minimum, leaves open the possibility that the recent TFP upsurge may mark the start of a new phase of growth for Australia.

6. The rest of the paper is organized as follows. The next section presents the formal growth accounting framework that has been generally used to identify the main channels through which ICT affect aggregate productivity. Within this section, special emphasis is given to the measurement of capital services and the contribution to growth of ICT capital. Section C presents the results of some studies that have analyzed the impact of ICT on productivity and output growth in the US. Section D replicates the growth accounting exercise for a group of Australian industries, since determining the impact of ICT on productivity requires examining what is happening at an industry level. Section F concludes.

B. A Growth Accounting Framework For the New Economy

7. In order to evaluate the contribution of new technologies to output and labor productivity growth, the following, well established, growth accounting equation can be used:


Based on a constant returns to scale production function (α+Ø=l) and perfect competition in the goods and labor markets, output (yt) growth can be accounted for by increasing use of capital, (Kt) and labor (Lt) inputs, each weighted by their share of total income (α and Ø), and by a residual (At), commonly named total factor productivity (TFP), which captures any growth in output that is not associated with input usage, that is, any disembodied technical change.

8. A first set of refinements to this equation can be made by distinguishing between ICT and non—ICT capital stocks, replacing labor input with a quality—adjusted index (q), and expressing all variables in per capita terms (lower—case variables indicate rates of growth of each variable less the rate of growth of unadjusted labor), resulting in the following equation:


9. A further refinement of the basic equation can be made by splitting the TFP growth into three components. First, a spillover effect related to the usage of ICT capital (θ) is introduced in order to single out “super—formal” returns, that is, the returns associated with this type of capital which are not paid to anyone and thus are part of TIT. Second, following Domar (1961), TFP growth can be disaggregated between growth in the ICT—producing sector and in the rest of the economy, using the two sectors’ shares of total gross output as weights:


10. This equation allows us to distinguish three channels through which ICT affects output and labor productivity growth: 1) via its role as a capital input; 2) via the TFP increase in the ICT—producing sector, and 3) via the TFP increase associated with the spillover effects related to the usage of the new technologies.7

11. Much of the discussion in the new economy literature has focused on the distinction between the usage of ICT (affecting labor productivity though the first and third channel) and its production (working via the second channel). In some sense, it can be argued that the more important measure of the new economy is the third channel, which is associated with externalities that cause an economy—wide increase in TFP. In this case, the adoption of ICT would be equivalent to an upward shift of the production function and lead to higher, long—term output and productivity growth rates. On the other hand, the ICT contribution to growth through the other two channels could be interpreted as a one—time transition to higher levels of productivity that occurs when new types of capital goods become available and start replacing old vintage of capital, and labor. In this case, a high contribution of new technologies to growth could indicate that such capital has reached a sufficiently large proportion of the total capital stock to be quantitatively significant.

C. The U.S. Benchmark

12. Before turning to the evidence for Australia, productivity estimates for the US are presented as they serve as an important benchmark for other countries (Table I.2).8 The recent, sustained acceleration of labor and total factor productivity in the US has been interpreted quite differently by different researchers, largely because of the different methodologies adopted in estimating equation [3]. Oliner and Sichel (2000), and to a lesser extent Jorgenson and Stiroh (2000), lean toward the view that ICT has played a significant role in generating a fundamental change in U.S. economy’s growth. On the other hand, Gordon (2000) and Bosworth and Triplett (2000) represent the more agnostic view that the ICT “revolution” has not had the same Impact of the general—purpose technologies introduced in the past century, such as railways and electricity.

Table I.2.

Contribution to Growth, United States and Selected Countries

(In percent)

article image

Selected country estimates are based on Goldman-Sacts, October 2000.

Aced. is the difference in labor productivity growth between 1996-1999 and 1991-1995 for other and Schell; between 1995-1998 and 1990-96 for Iorgenaon and Strict; and between 1995-1999 and 1972-1993 for Gordon.

13. More specifically, despite some difference in methodology,9 Oliner and Sichel and Jorgenson and Stiroh obtain similar estimates, and attribute around a quarter percentage point of the acceleration in labor productivity since 1995 to the ICT production process (TFP growth in the ICT sector) and a half percentage point to capital deepening, all of which is due to the accumulation of ICT capital. Together, ICT contributes for around three quarters of the recent labor productivity acceleration.10

14. In contrast to the other two studies, Gordon focuses on identifying the cyclical component of the U.S. labor productivity surge and finds that about half of the U.S. labor productivity acceleration after 1995 has been a cyclical phenomenon. Disaggregating the trend labor productivity acceleration, he finds that half is accounted for by the TFP increase in computer production and the other half by ICT capital deepening. Hence, in sharp contrast with the first two studies, Gordon sees no TFP growth outside the ICT sector, and takes this as proof that ICT cannot be considered a significant, pervasive technological breakthrough.

D. The New Economy Contribution to Growth in Australia

The First Channel: ICT Capital Deepening

15. This section estimates the growth accounting equation [3] for Australia using official Australian Bureau of Statistics (ABS) data on output, inputs and prices, and analyzes the role played by ICT in the Australian labor productivity growth, starting with the contribution of ICT capital deepening.

16. A key step in assessing the new economy credentials of Australia is to obtain a measure of disembodied TFP, corresponding to an upward shift of the technological frontier. To do this, it is crucial to express both capital and labor in quality adjusted terms, to take into account the fact that technology and other factors tend to improve the quality of the factors of production over time. In particular, this means that the ICT capital stock in [3] must capture changes in the quality of capital associated with investments in ICT capital goods.

17. A first step to capturing capital quality changes associated with investments in ICT capital is to recognize that calculating price changes by comparing identical products over time (as normally done) would completely miss the quality (that is, output) improvement associated with the rapid succession of progressively more powerful speed, memory, disk capacity and many other features of computers and ICT hardware. Thus, price indexes of such products should be calculated by taking into account the change in the products’ characteristics, through a so called “hedonic” function. Doing this amounts to redistributing the growth of nominal investments in ICT from prices to volumes and, therefore, to embody technological changes in capital stock.

18. The U.S. has used hedonic price indexes for computer equipment since 1985 and only a few countries have followed this methodology to deflate their investments data. In Australia, the ABS has adopted hedonic prices compiled by the U.S. Bureau of Economic Analysis (BEA) to deflate investments in Computers and Peripherals, while nominal investments in software are deflated by an index that is assumed to decline by 6 percent per year (an estimate also used by Statistics Canada and constructed by observing the trend of software prices for popular PC software).

19. Table I.3 shows that ICT investments in Australia have grown at double digit rates since the middle of 1980s. In particular, growth in investment in hardware (computers and peripherals) has been very high, reflecting the marked reduction in hardware prices. The much higher growth of investments in ICT capital compared to other types of capital has caused a significant increase of the of ICT share of non—building investments. In 2000, investment in hardware and software accounted for over 20 percent of total non building investment (Chart I.1), compared to about 6 percent in 1986. However, a note of caution is warranted about the sustainability of this process, particularly since, in the second half of the 1990s, investment in hardware has not accelerated significantly, despite a sharp increase in the rate of decline of prices.

Table I.3.

Investment in ICT, Australia

(In percent)

article image
Source: Staff estimates based on ABS data.
Chart I.1.
Investment in Computer Hardware and Software as a Share of Total Investment, Australia.

20. It is interesting to examine how Australia stacks up against other advanced economies in ICT capital accumulation. To overcome the problems associated with cross—country comparability of data, some authors (Schreyer, 2000, and Daveri, 2000) have used U.S. hedonic prices to deflate other countries IT nominal investments.11 The result of this exercise is reported in Table I.4. The figures for Australia are broadly similar to the ones reported in Table I.1 as far as investment in hardware is concerned (this should not be surprising since, as noted above, ABS has adopted U.S. hedonic prices). On the other hand, the growth of investment in software is considerably lower, reflecting the much smaller rate of decline of software prices in the U.S. compared to the 6 percent assumed by the ABS.

Table I.4.

Investment in ICT. Other Countries

(In percent)

article image
Source: Daveri (2000) Nominal investments are from the WITSA/IDC database, with the exception of the US (data from BEA). WITSA (World Information Technology and Service Alliance) is a communication of 32 ICT industries association around the world, IDC (International Data Corporation) is a research and consulting company on ICT industries.

Invesment in hardware, software and communication equipment.

21. Keeping these differences in mind, the evidence that emerges from the two tables is that the pace of investment in ICT during the 1990s has been higher in Australia than in any other industrialized country. Having started from a lower share of total non—residential investment, the higher pace of ICT accumulation has allowed Australia to (at least partly) catch up with most of these countries.

22. The next step in constructing a quality—adjusted capital stock is to express the capital stock in “efficiency” units.12 This is done by recognizing that older capital goods provide fewer productive services than new ones, as their efficiency declines with age following the release of new, more efficient, models or simply because of “wear and tear.” For each capital asset a measure of its productive capacity can be obtained through the perpetual inventory equation:


where It, j is the (real) investment in capital asset j at time t and Õ i, j is the parameter that captures the rate at which efficiency of the investment in asset j made in t-i is written down over time. Kt, j is defined as the “productive” capital stock since it measures the income—generating capacity of the assets. As such it differs from the net capital stock, which is more a “wealth” indicator as it reflects the current market value of the asset (Oliner and Sichel, 2000). The difference between the two measures of capital is especially relevant for ICT capital assets, as their productive efficiency decline much less with age13 but their economic depreciation (value losses) is high.

23. The concept of capital input that is appropriate for equation [3] is the flow of productive services that each asset provides during a period of time. The basic assumption here is that the flow of capital services is proportional to the asset’s productive capital value, and thus declines with age as well.14 Given the heterogeneity of capital goods, each with a different age—efficiency profile, total capital services must be obtained as the weighted average of individual flows. As in equilibrium “rental prices” (the prices that would be charged to rent a unit of capital) are equal to marginal productivities of the capital goods, the contribution of every assets to total capital services is weighted by an estimate of its rental price.

24. Weighting capital services by rental prices has an immediate consequence for the ICT contribution to output and labor productivity growth.15 In equilibrium, rental prices for ICT capital tend to be much higher than for other assets, as computers depreciate rapidly and have large negative capital losses (thus, ICT capital must have a greater marginal productivity if it is to be held at all). This implies that investments in ICT goods receive a larger weight in the estimate of capital services than in the estimate of the net capital stock. The related, positive quality change in the stock of capital implies a stronger contribution of capital deepening to growth, at the expense of TFP.

25. At the same time, the combination of higher marginal productivity and falling prices implies that ICT capital has a very high q—ratio, inducing firms to substitute towards this type of capital. Using rental prices allows this substitution process to be captured in the estimate of capital services. A summary indicator of the compositional change of capital stock towards the more productive ICT capital is obtained for Australia by looking at the difference between growth of capital services and growth of net capital stock. Chart I.2 confirms that the rapid accumulation of ICT assets has brought about a positive compositional effect in Australia’s capital stock, especially in the second half of the 1990s.

Chart I.2.
Australia - Changes in the Composition of Capital

(Average Annual Percent Change)


26. The relative contribution of ICT capital on growth within equation [3] depends on its share of total income rr). This share is more likely to be large if ICT is a great share of total capital and/or if the relative rental price between ICT and other capital assets is high. While the rapid decline of ICT prices has reduced the relative rental price for ICT goods (Chart I.3 shows the steady fall in the relative rental prices for selected industries in Australia), the massive investment in ICT capital goods during the 1990s has raised the share of ICT in the total productive capital stock. The net result of these opposing forces has been a sharp increase of the ICT capital income share, especially for software (Table I.5).

Table I.5.

Income Shares of ICT Capital, Australia (In percent)

article image
Source: Staff estimates based on ABS data.
Chart I.3.
Australia - Relative Rental Prices for Computers and Peripherals Selected Sectors.

27. Once capital services and asset income shares are estimated for ICT and non—ICT assets, their contribution to output and labor productivity growth can be quantified within equation [3].16 Table I.6 shows that the ICT contribution to growth has been increasing steadily in Australia over the last 30 years, both in absolute and relative terms. While ICT capital (the sum of hardware and software) accounted for about 2 percent of capital deepening in the first half of 1970s, it explains about two thirds of capital deepening in the 1990s. Further, while the acceleration of labor productivity in the second half of the 1990s has coincided with the growth of TFP, the acceleration of ICT capital deepening has accounted for almost 0.2 percentage points of labor productivity growth and has offset the negative contribution coming from the slower rate of accumulation of other types of capital. Another piece of evidence that the incessant rate of ICT accumulation in Australia over the last 20 years has finally paid off is mat, on average in the last 5 years, ICT capital has contributed to output growth for around 0.9 percentage points per year.

Table I.6.

Contribution to Growth, Australia.

(In percent)

article image
Source: Staff estimates based on ABS data.

28. Comparing the figures in Table I.6 with those in Table I.2 shows that, in the period 1995-2000, ICT capital deepening contribution to labor productivity growth in Australia was broadly the same as in the U.S., considerably larger than in the UK and the Euro area, and smaller only compared to Japan.

29. A rough estimate of the ICT capital contribution to labor productivity growth through the ICT—related improvement in the quality of labor (see footnote 7) shows that this is not likely to be a very large factor for Australia. The wage differential between high—skilled and other workers in Australia has increased by 4 percentage points from 1986 to 1998 (De Laine, Laplagne and Stone, 2000). Over the same period, the share of high—skilled workers over total employment has increased by 12 percentage points. Even identifying all high—skilled workers with those using ICT at work, these numbers suggest that ICT has contributed about 0.5 percentage points to the increase of the average wage (and, thus, of labor productivity) over the period 1986-1998, only 0.04 percentage points per year. This result also suggests that the overall contribution of labor quality to the labor productivity growth in Australia has been probably of a second order magnitude.17

The Second Channel: The Production of ICT

30. Table I.6 shows a relatively large contribution to output growth from TFP, reflecting disembodied technical changes, organizational and managerial changes, and other aspects that are not captured by factor accumulation. An assessment of the second and third channels through which ICT or the new economy might affect growth requires a more detailed analysis of this TFP dynamic.

31. For the purpose of this discussion, a broadly defined ICT—producing sector is made up of those industries involved in the production of computers, peripherals, software and any other means of information processing and communication. This sector in Australia is not separately identified in official statistics. However, an indication of the relative importance of the second channel in equation [3] may be obtained by looking at others statistics about the ICT sector in Australia relative to other countries.

32. An OECD study18 ranks industrialized countries according to 4 key indicators, the ratio of employment, value added, R&D and trade in their ICT sector with the totals for the business enterprise sector. According to this study, Australia belongs to the low ICT intensity group, with two medium (R&D and trade) and two low (value added and employment) ratings (Table I.7). The relatively low weight of the ICT sector in the Australian economy suggests that this is unlikely to play a significant role in the growth accounting equation [3].

Table I.7.

Indicators of ICT Sector, 1997, Selected Countries

(In percent)

article image
Source: Measuring the ICT sector, OECD. 2000.

The Third Channel: Spillovers Effects from ICT Capital Accumulation

33. As pointed out in Section B, one test for the existence of a “new economy” is whether, in addition to its contribution to capital deepening, the larger use of ICT capital has caused an increase in TFP growth and whether the TFP increase is sufficiently broad—based and extended across the economy in order to support the possibility of spillovers and externalities associated with the usage of ICT capital. An analysis of the dynamics of TFP in different industries of the economy is thus needed to assess the importance of the third channel in equation [3]. The next section extends the growth accounting exercise for Australia to a sectoral level.

34. Before presenting the results of the analysis of sectoral TFP growth, however, it should be stressed that while the main focus of this paper is on ICT investment (spending on ICT by the business sector) much of the recent debate on the “new economy” has focused on total spending on ICT (by households and the public sector as well as businesses) taken as a proxy for the diffusion of new technologies across the population. A ranking of OECD countries based on their total spending on ICT as a share of GDP (using data on ICT nominal investments from the WITSA/IDC database) indicates that in 1997 Australia was second only to New Zealand and well above the U.S. and all the other OECD countries. Other indicators for the degree of penetration of ICT across the economy are reported in Table I.8, and all show the high degree of usage of new technologies by Australian households relative to other industrialized countries.

Table I.8.

Indicators of ICT Diffusion, Selected Countries

article image

OECD, Information Technology Outlook (quanting National Statistical Offices).

Internet Software Continuum (ISC) (

OECD. The figures cater to the cost of success to Internet for 20 hours including VAT.

Computer In dusty Alma mach Inc. (

OECD, Information Technology Outlook.

E. A Cross-Industry Analysis

35. While the figures in Table I.8 are an important signal of the degree of diffusion of ICT in everyday life, it would be simplistic to base an assessment of a country’s ability to benefit from a significant ICT contribution to output growth solely on such indicators. In order to make such a judgment a further step is required, namely to establish a link between a country’s capacity to adopt new technologies and its ability to “incorporate” them in the production process.19

36. This can be assessed by applying the growth accounting equation [3] to the eight market sector industries: Agriculture, Forestry & Fishing; Mining; Manufacturing; Electricity, Gas & Water, Construction; Retail Trade; Wholesale Trade; Transport & Storage; Communications; Accommodation, Cafes & Restaurants; Finance & Insurance; and Cultural & Recreation Services.20

37. Two caveats are necessary before proceeding. The first one is to keep in mind the well—known problems in measuring the output of service industries. A couple of examples illustrate the issue at stake.21 One of the most relevant innovations related to the investment in ICT capital within the banking sector is the diffusion of ATMs, which, among other benefits, has allowed great saving of time, permitted transactions 24 hours a day and greatly reduced the need to carry cash balances. Nonetheless, such quality (and, thus, output) improvements are generally not captured in the national account statistics (with the only exception of the U.S.). Similarly, in the retail sector, the shift from department stores to lower—price outlets is typically treated as a reduction in quality, not prices. Such a shift (and the related value of increased product variety) is explicitly ruled out in the construction of price indexes which follow a specific product in a specific store (the same way in which price indexes of computers follow the “box,” rather than what is inside it, thereby missing the increase in power and speed).

38. The second caveat is that a fully satisfactory estimate of sectoral TFP would require using industry gross output, and treating material inputs as a separate factor of production in the growth accounting approach. However, in the absence of this series for Australian industries, the estimates below are based on industry value added.22 As a large part of the output of the service sector is sold as an intermediate input to other industries, errors in measuring their output will likely result in an over—estimation of the labor and total factor productivity gains accruing to the receiving industry. While in the aggregate this is irrelevant, as under—estimation of productivity gains for one industry is offset by over—estimation for another one, a growth accounting exercise conducted at the sectoral level using a value added concept of output may end up erroneously allocating productivity gains across different industries23.

39. For the purpose of this paper, these problems are especially relevant because the largest purchases of ICT capital in Australia are by the service sector. Table I.9 shows that, in 1987-2000, communications, finance and insurance, accommodation cafes and restaurants and cultural and recreational services have invested in computers and peripherals at a much stronger pace than other industries. The same picture is obtained by looking at the dynamics of the ICT capital stocks. Table I.10 shows that the three most “ICT—capital intensive” sectors are finance and insurance, construction and wholesale trade for software, and finance and insurance, retail trade and cultural and recreational services for hardware.

Table I.9.

Investment in Hardware (Computers and Peripherals), Volumes

(In percent)

article image
Source: Staff estimates based on ABS data.
Table I.10.

ICT Capital Services. Share of Total Capital and Rates Of Growth (Annul Averages), 1995-2000

(In percent)

article image
Sources: Staff estimates based ABS data.

40. The main objective of this section is to examine the cross—industry relationship between ICT capital and TFP growth. The data on investments and capital stocks would suggest a larger relative ICT capital contribution to labor productivity growth in the service sector industries. Are these also the industries reporting the highest increases in TFP?

41. Table I.11 shows that the industry with the highest TFP growth in the most recent period 1995-2000 was wholesale trade, followed by agriculture, communication services, and finance and insurance.24 Labor productivity growth in these industries has also benefited from a relatively high contribution from ICT assets, but with two exceptions. The first one is agriculture, with a relatively low contribution from both hardware and software. The second one is wholesale trade, which has been taking relatively less advantage from hardware25.

Table I.11.

TFP and ICT Contribution to Labor Productivity Growth, 1995-2000

(Average annual rates of change)

article image
Source: Staff estimates based on ABS data.

42. Overall, the data from Table I.11 provide mixed signals on the existence of a positive relationship between TFP and ICT capital. On the one hand, labor productivity and TFP growth have been very large in some service industries which have been intensely investing in ICT technologies. This is a relevant result per se, first because it seems to contradict the so—called “ICT paradox” (largely identifiable with the slow labor productivity growth in the ICT—intensive service sector), and second because it occurs despite the problems in output measurement that were stressed above, and that are probably causing an underestimation of the productivity gains in the service sector.26

43. While the large ICT capital accumulation and the very strong increase in TFP in some of the industries considered (such as finance and insurance) support the notion of a new economy, Table I.11 also shows that there have been some industries which have experienced a modest TEP growth even with a relatively substantial contribution to growth from ICT assets, particularly cultural and recreational services. However, abstracting from this sector and plotting together TFP growth rates and the ICT share of total capital services in each industry suggests that there is indeed a positive relationship between TFP growth and ICT capital intensity across Australian industries27 (Chart I.4).

Chart I.4.

Australia - TEP Growth and ICT Share of Capital Services, 1995-2000


44. Clearly, the correlation showed in Chart I.4 does not suggest a causal link between investments in ICT and TFP growth. There could be other, common factors that are driving both variables in the period of time considered, such as the structural reforms being implemented in Australia since the mid 1980s. The counter argument to this observation could be that the essence of the new economy story for Australia is the fact that these reforms have led to organizational changes, improved management techniques, and a more business—friendly institutional environment, Although these factors would be reflected in a long—term, sustainable increase of TFP growth, at the same time, the reforms have encouraged the trend towards an upgrading of new technologies in production processes and more active interest in product and process innovation, with resultant strong incentives for ICT capital accumulation.

45. In other words, the key argument of this story is that, especially in the context of the more favorable environment produced by structural reforms, the sustained accumulation of ICT capital was likely to be associated with a profound reorganization of economic activities, aimed at taking advantage of the network externalities allowed by the new technologies. These externalities are in theory capable of changing the nature and the boundaries of firms, allowing activities that are currently carried out within hierarchical structures to take place by market transactions, reducing intermediation, squeezing out monopoly rents, reducing inventory costs, moving activities from the market to household and from households to the market. In particular, the commercial use of Internet, both at a business—to—business and business—to—consumer level, would be mostly responsible for this shift in the technological paradigm.

46. While it is safe to conclude that it is still too early for the statistics to capture the full potential of the ICT network externalities, the more limited objective pursued in this paper is to use a traditional growth accounting framework to examine whether there at least signs of the positive impact of ICT capital on TFP growth, following the proponents of the new economy story. At an aggregate level, the analysis suggest that ICT capital accumulation has certainly played a great role in the productivity acceleration recently experienced by Australia. Taking the analysis to the sectoral level shows that, despite the measurement problems, one cannot rule out the possibility that ICT capital had a part in the TFP acceleration across industries. On the whole, these results suggest that the market concerns that Australia is still an “old economy” are misplaced.

F. Conclusions

47. Using official data from the Australian Bureau of Economic Statistics and a formal growth accounting framework, this paper shows that the rapid accumulation of ICT capital over the last two decades in Australia has played a significant role in explaining the impressive, structural acceleration of labor productivity after 1995. About half of the labor productivity growth in the last 5 years can be explained by capital deepening, and around two—thirds of total capital deepening can be attributed to ICT capital (hardware and software), A comparison with other countries shows that the ICT capital contribution to productivity growth in Australia has been as large as in the U.S. and well above the average for the countries in the Euro area.

48. A key test for the new economy story is whether ICT capital accumulation has caused an economy wide acceleration in TFP growth. Extending the growth accounting framework to the sectoral level, this paper shows that, despite measurement problems, there is a correlation between TFP growth rates and ICT share of capital services across industries. This leaves the door open to the possibility that Australia will be in a position to benefit from the network externalities associated with the diffusion of ICT in economic activity.


  • Australian Bureau of Statistics, 1999-2000, Business Use of Information Technology.

  • Bassanini, Andrea, Stefano Scarpetta and Ignazio Visco, 2000, “Knowledge, Technology and Economic Growth: Recent Evidence from OECD Countries,OECD Working Paper, No.259.

    • Search Google Scholar
    • Export Citation
  • Bean, Charles, 2000, “The Australian Economic Miracle: a View From the North,” in The Australian Economy in the 1990s, Reserve Bank of Australia.

    • Search Google Scholar
    • Export Citation
  • Bosworth, Barry and Jack Triplett, 2000, “What’s New About the New Economy? IT, Economic growth and Productivity,Brooking Institution. (

    • Search Google Scholar
    • Export Citation
  • Bosworth, Barry and Jack Triplets, 2000, “Productivity in the Services Sector,Brooking Institutions, forthcoming in Services in the International Economy ed. by Robert M. Stern, (

    • Search Google Scholar
    • Export Citation
  • Council of Economic Advisors, 2001, Economic Report of the President, Washington, US.

  • Daveri, Francesco, 2000, “Is Growth an Information Technology Story in Europe Too?,IGIER Working Paper, n.168.

  • De Laine, Craig, Patrick Laplagne and Susan Stone, 2000, “The Increasing Demand for Skilled Workers in Australia: The Role of Technical Change,Staff Research Paper, (Canberra: Productivity Commission).

    • Search Google Scholar
    • Export Citation
  • DeLong, Bradford and Lawrence Summers, 1990, “Equipment Investment and Economic Growth”, National Bureau of Economic Research, Working Paper Series, NO. 3515

    • Search Google Scholar
    • Export Citation
  • Domar, Evsey, 1961, “On the Measurement of Technological Change”, Economic Journal, (December).

  • Goldman—Sachs, 2000, “The ICT Revolution—New Data on the Global Impact,Global Economics Weekly (October).

  • Gordon, Robert, 2000, “Does the “New Economy” Measure up to the Great Inventions in the Past?,Journal of Economic Perspectives, Vol. 14 (Fall), No. 4

    • Search Google Scholar
    • Export Citation
  • Gullickson, William and Michael J. Harper, 1999, “Possible Measurement Bias in Aggregate Productivity Growth,Monthly Labor Review (February) pp. 4767

    • Search Google Scholar
    • Export Citation
  • Jorgenson, Dale and Zvi Griliches, 1967, “The Explanation of Productivity Change,Review of Economic Studies, Vol. 34, No. 3.

  • Jorgenson, Dale and Kevin Stiroh, 2000, “Raising the Speed Limit U.S. Economic Growth in the Information Age,Brooking Papers on Economic Activity, 31:1.

    • Search Google Scholar
    • Export Citation
  • OECD, 2000, Science and Technology Outlook.

  • OECD, 2000, Measuring the ICT Sector.

  • Oliner, Stephen and Daniel Sichel, 2000, The Resurgence of Growth in the late 1990s: is Information Technology the Story, Federal Reserve Board Working Paper (February).

    • Search Google Scholar
    • Export Citation
  • Parham, Dean, 1999, “The New Economy? A New Look at Australia Productivity Performance,Staff Research Paper, (Canberra: Productivity Commission).

    • Search Google Scholar
    • Export Citation
  • Productivity Commission, 1999, “Microeconomic Reforms and Australian Productivity: Exploring the links,Staff Research Paper, (Canberra: Productivity Commission).

    • Search Google Scholar
    • Export Citation
  • Salgado, Ranil, 2000Australia: Productivity Growth and Structural Reforms,IMF Country Report 00/24 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Scarpetta Stefano, Andrea Bassanini, Dirk Pilat and Paul Schreyer, 2000, “Economic Growth in the OECD Area: Recent Trends at the Aggregate and Sectoral Level”, OECD/ECO Working Paper, n.21.

    • Search Google Scholar
    • Export Citation
  • Schreyer, Paul, 2000, “The Contribution of Information and Communication Technology to Output Growth: a Study of the G7 Countries,OECD/DSTI Working Paper, 2000/2.

    • Search Google Scholar
    • Export Citation

Annex I.1 Weighting Capital Services by Rental Prices

Ignoring taxes and fiscal incentives, the rental price for the asset j at time t can be estimated as:


In equilibrium, the gross return on investing one dollar in asset j (the term in brackets on the right hand side of this expression) must be sufficient to cover opportunity costs (r, the net nominal rate of return on capital, assumed equal for all assets), the loss in market value with ageing (dj, the depreciation rate for asset j) and the capital gains or losses associated with a change in price of the asset (π j, the rate of change of the asset j price, Pt, j). The rate of growth of capital services, the appropriate measure of capital input to be used in equation [3], is thus obtained as:




The official ABS data used in this paper distinguish between 15 types of assets. These are: Computer software, Computers and peripherals, Road vehicles, Other transport equipment, Industrial machinery and equipment, Electrical and electronic equipment, Other plant and equipment, Non—dwelling construction, Ownership transfer costs. Inventories, Land, Livestock, Mineral exploration, and Artistic originals.

For each of them the ABS provides different estimates of productive stocks and rental prices depending on the industry where they are used and, for each industry, on whether they belong to the corporate or unincorporated sector. Following equation [7], rental prices are used to weight the rates of change of each asset’s productive capital stock, both within and across industries. The sum of these weighted rates of change gives the aggregate capital services. For the industry analysis the same methodology is followed, this time weighting only corporate and unincorporated productive capital stocks within the same industry.

If the growth contribution of asset j (i.e., ICT capital) is to be seen separately in equation [3], it would be the product of this asset’s capital services and its share of total income. In turn, this share can be expressed as the product of the share of total capital income accruing to asset j and the share of capital over total income:



Table I.2.1.

Source of Labor Productivity Growth, by Industry

article image
article image

Prepared by Roberto Cardarelli (x38059), who is available to answer questions.


This term is used here in its narrow sense, that is, to indicate the fundamental changes in economic activities which have been associated with the diffusion of information processing and communication technologies (ICT).


All studies reported in footnote 3 are unanimous in their interpretation of the acceleration in TFP growth as the product of the structural reforms implemented since the mid—1980s. Parham (1999) shows that this acceleration, far from being cyclical, corresponds to a structural break in the Australian growth path; as a result, in 1998 labor productivity was 15 percent higher than it would have been had the economy remained on its historical growth trajectory.


The accumulation of ICT capital may be affecting growth also by inducing an improvement in labor quality, a channel that is not explicitly identified in [3]. De Long and Summers (1991) stress that employees learn new skills and more efficient production methods after the installation of new equipment. As at least part of the quality change in labor is caused by the ICT capital accumulation, the framework described by equation [3] tends to underestimate the real contribution of ICT to growth.


An important caveat is that methodological differences in the level of statistical sophistication limit the scope for reliable international comparison of productivity data. However, this is a less serious concern for Australia, because the ABS has closely followed methodologies and conventions similar to those in the US for measuring ICT capital stocks and prices.


Jorgenson and Stiroh use a broader concept of output, including consumer durables and imputed service flows from owner—occupied houses.


Both studies identify the key source of ICT contribution to growth with the high efficiency gains experienced in the ICT sector (especially semiconductors), which has been reflected in a sharp decline in computer prices. This in turn has led to massive computer investments, as both firms and household sought to cut costs by substituting toward relatively cheaper inputs.


In more detail, the authors construct “hedonic” price indexes for ICT investments in several countries under the assumption that the rate of change of ICT prices with respect to other capital goods is the same in each country as in the US. The same methodology is followed by the Goldman Sachs study whose estimates are presented in Table I.2.


The methodology described here was first proposed by Jorgenson and Griliches (1967) and is now adopted by several national statistical agencies, such as the US BLS and the ABS. In particular, the ABS has started releasing data on productive capital stocks in 1998 and revised these estimates in 1999. The data on capital stocks used in this paper are still defined as experimental.


ABS follows the BLS in adopting an hyperbolic age—efficiency function, according to which the efficiency of ICT assets declines by small amounts at first and rapidly at the end.


This assumption amounts to ignoring the fact that utilization of capital services may vary depending on the stage of the business cycle. In this paper, the business cycle’s contribution to the total factor productivity estimates is neutralized by applying a Hodrick-Prescott filter to both output and inputs in equation [3].


Annex I.1 describes more in detail the aggregation process.


As for the other variables in [3], output is measured in terms of valued added, labor input is the number of hours worked, the labor income share is the ratio of compensation of employees plus net taxes on labor to total income, and the capital income share is 1 minus the labor income share. All data are taken from the Australian System of National Account 1999-2000 and refer to the market sector, a special industry grouping comprising only those industries (listed in section E) for which output can be satisfactorily estimated.


In the same way as for capital services, an index for “labor services” can also be estimated as the weighted average of hours worked by different age—gender—education groups, with the weights equal to the different groups share of total labor compensation. Estimating such index for Australia, Bassanini, Scarpetta and Visco (2000) also conclude that its contribution to output growth has been relatively modest.


An example of the need to take these figures with a pinch of salt comes from the 1999-2000 “Business Use of Information Technology” survey of business use of computers and the Internet, published by the ABS. This survey shows that, as of June 2000, about 60 percent of Australian firms had Internet access, but only 23 percent of them used the Internet for activities associated with buying goods and 28 percent in activities associated with selling goods.


For each of these industries capital input is obtained as described by equation [7] in Annex I.1. As the hours worked at a sectoral level are available only from 1986, the results are obtained only from this year onwards. The labor income shares for each industry are obtained as the ratio of the compensation of employees to the industry value added.


Bosworth and Triplett (2000) provide a more detailed discussion of this issue.


In contrast to the “sectoral” approach to industry productivity proposed by Domar (1961), and based on the “gross output” concept, using a net output concept (such as value added) implies that the intermediate inputs purchased from other sectors are subtracted from the sector’s gross output rather than recognized as a separate factor of production (in addition to the primary factors, labor and capital). By providing an explicit role for intermediate goods and services as a source of industry growth, the “gross output” concept allows the aggregate TPF gains to be correctly allocated among industries (this is also the method adopted by the BLS in estimating productivity for US industries). For a detailed treatment of these issues, see Gullickson and Harper (1999).


In analyzing the link between ICT growth and TFP, this paper thus restricts its focus to within—industry effects and ignores possible spillovers effects across industries. For example, network gains may occur between the manufacturing and wholesale sectors, implying that growth in ICT usage in the wholesale sector could produce TFP gains in the manufacturing sector.


Estimates of the sources of labor productivity growth for the market sector industries in different periods of time is reported in Annex I.2.


This can be explained by the low ICT capital intensity of agriculture (the lowest for software and the second lowest for hardware) and the relatively low hardware capital intensity of the wholesale trade sector (Table I.10).


The acceleration in labor and total factor productivity in the service sector is not solely an Australian phenomenon, as recent estimates for the US also show a pick—up in some of these industries, in particular in wholesale trade, retail trade and finance (see Council of Economic Advisors, 2001).


The coefficient of correlation, measuring the degree of linear correlation between TFP growth rates and ICT shares of capital, is positive and equal to 0.3. Chart I.4 plots TFP growth with ICT capital intensity, and not the growth of ICT capital, as spillovers effects are likely to become relevant only when the stock of ICT capital becomes sufficiently large.