Productivity and efficiency measures

A range of indicators can be used to determine the performance of a firm or industry. The conventional indicators of the performance of private firms include the level of rate-of-return and prices. In Figure 8.1, the rate-of-return for the Australian gas transmission/distribution/retail industry is shown for the period 1990 to 1999 (earnings before interest and tax/total assets). In this figure, it can be seen that the rate-of-return for the industry was reasonable in the 1990s. Another indicator of the industry’s performance is the level of demand growth. Figure 8.2 next shows the growth rate of demand for natural gas and real GDP.

Figure 8.1 Gas industry earnings (before interest and tax/total assets), percentage Source: Australian Bureau of Statistics (1981-1999)

Figure 8.2 Growth of real GDP and demand for natural gas, percentage

Source: Australian Bureau of Statistics (1981-1999); Australian Gas Association (1982-1995, 1996-1999)

Over the period 1985 to 1999, growth in gas demand averaged slightly higher than that of real GDP (annual average of 4.2 per cent compared with 3.9; Figure 8.2). In Figure 8.2, it can be seen that even though overall demand for gas was greater than that of real GDP, gas demand fluctuated more.

Another indicator of firm or industry performance relates to the level of prices. Traditionally, gas prices in Australia have been divided into separate rates for residential, commercial and industrial classes. These figures in constant dollar terms are presented in Figure 8.3 for the period 1980 to 1999. Comparative prices are also given in Tables 8.2 and 8.3. Final real gas prices remained fairly steady throughout the 1990s (Figure 8.3).

The performance indicators discussed so far do not provide definitive indication of the change in performance of the gas industry. Utilities such as gas operate in markets that lack prices and costs determined under competitive conditions. This was certainly the case for the gas industry in Australia before the reforms of the 1990s, and still exists for specific components of the industry such as distribution and transmission. In such instances, the usual market indicators of performance such as profitability cannot be used to accurately gauge an industry’s economic performance. It is possible that these financial indicators

Reform and productivity change 169

Figure 8.3 Real prices of gas, $A1990, $/GJs Source: Australian Gas Association (1982-1995, 1996-1999)

Table 8.2 Capital city natural gas prices (1994 and 1999)

might be more a consequence of the distortions themselves rather than the performance of the industry in question. In these circumstances, the level and change of productivity and efficiency is a more appropriate indicator of an industry’s performance. In the case study of Australia’s gas industry discussed in this chapter, a similar process of analysis has been undertaken to what was discussed in Chapter 4 of this book.

Table 8.3 Descriptive statistics for Australian natural gas sectors in each state (1985/6 to 1998/9)

Efficiency can be defined as the degree to which resources are being used in an optimal fashion to produce outputs of a given quantity. In general, there are three main aspects of economic efficiency: technical; allocative; and scale (Farrell 1957). Technical efficiency occurs when the maximum output possible is delivered using given inputs. Allocative efficiency measures factor proportions given relative factor prices. Scale efficiency refers to optimal size. Economic efficiency is the product of technical and allocative efficiency, and helps to identify whether cost reductions are possible. In contrast, productivity is a measure of the physical output produced from the use of a given quantity of inputs. Productivity often varies as a result of improvements in production technology, differences in the technical efficiency of an industry and the external operating environment in which production occurs. Productivity change is a dynamic indicator of the change in outputs relative to inputs; while productivity growth will reflect changes in economic efficiency as well as technological change.

The most common method of determining levels of productivity and efficiency is to create index numbers. These can generally be constructed without the need for statistical estimation of a production or cost function. The most commonly used index numbers are those that indicate the partial factor productivity of a firm or industry. Partial factor productivity measures generally relate to an industry’s output to a single input factor. For example, in the natural gas industry, gas throughput per employee is an example of a labor-based partial productivity measure (see Figure 8.4 below). Another commonly used labor

Figure 8.4 Labor productivity, PJs per employee Source: Australian Gas Association (1982-1995, 1996-1999)

Figure 8.5 Capital productivity, PJs per km of pipeline Source: Australian Gas Association (1982-1995, 1996-1999)

productivity index in the gas industry is an index of customers per employee. Capital productivity measures, however, are often difficult to calculate in the gas industry due to the problems in measuring capital inputs. Two commonly used indicators of capital productivity are the number of customers and the gas throughput per kilometer of mains.

In relation to the Australian gas industry, there have been several studies that have calculated partial productivity measures. Examples of labor productivity and capital productivity measures can also be found in the Bureau of Industry Economics’ international performance indicators - gas supply, and the Australian Gas Association’s gas distribution industry performance indicators, and Gas Statistics Australia.¹

Although relatively easy to calculate, the partial factor approach has a disadvantage in that it can be misleading when reviewing the change in productivity of an industry. For instance, it might be possible for an industry to raise productivity in terms of one input, at the expense of reducing the productivity of other inputs. Over the years, most industries have substituted capital for labor in the production process. Therefore, indexes of output to labor would tend to overstate the growth of overall productivity (i.e. the combined productivity of labor, capital and other factors). Where this applies, a partial measure, such as labor productivity, may inaccurately portray the total change in productivity. If the process has simply involved a substitution of capital for labor, then a TFP indicator that estimates a more modest increase in overall productivity would be a more appropriate measurement of productivity.

Various approaches can be used to measure TFP, which can lead to different empirical results and interpretations. The most common TFP approach is to estimate the ratio of the index of the weighted sum of outputs (weighted usually by revenue shares) by the index of weighted sum of inputs (weighted usually by cost shares). The main difficulty in using this approach, however, is that it is data-intensive. There has been no attempt to use this method to determine the TFP of the Australian natural gas industry, although it has been used in other countries (e.g. Bishop and Thompson 1992). It has also been applied on a number of occasions to the Australian electricity industry (e.g. Australia, Industries Assistance Commission 1989; Australia, Industry Commission 1994; Lawrence, Swan and Zeitsch 1991; Swan Consultants 1991). During the 1990s, the Steering Committee on National Performance Monitoring of Government Trading Enterprises used this method to calculate the TFP of the various government-owned electricity companies operating in Australia (Australia, Steering Committee on National Performance Monitoring of Government Trading Enterprises 1992, 1998), but did not apply this to the government-owned gas transmission and distribution pipelines.

Another method of determining TFP is to use DEA to apply the Malmquist procedure. This approach can be applied to panel data to calculate indices of TFP change, technological change, technical efficiency change and scale efficiency change, as discussed in Fare, Grosskopf, Norris and Zhang (1994), as well as in Chapter 4 of this book. The idea behind this method of efficiency analysis is to use data collected for the industry sectors or firms, and to derive what is known as the ‘best-practice frontier’. What constitutes a best-practice frontier can change, therefore, it is important to incorporate it into process of estimating the production process. The Malmquist TFP index is one method of doing so.

In effect, the Malmquist TFP index derives an efficiency measure for one year relative to the prior year, while allowing the technical progress frontier to shift.² This approach allows the decomposition of productivity change into technological change and technical efficiency change. Technical efficiency change under constant returns-to-scale can be further decomposed into scale efficiency and pure technical efficiency under variable returns-to-scale. Scale efficiency can be defined as the extent to which an organization can take advantage of returns- to-scale by altering its size toward the optimal scale (which is defined as the region in which there are constant returns-to-scale in the relationship between outputs and inputs). In contrast, pure technical efficiency is determined by the difference between the observed ratio of combined quantities of output to input, and the ratio achieved by best-practice institutions that can be attributed to managerial practices and not scale efficiency.

DEA is a non-parametric linear programming technique that estimates organizational efficiency by measuring the ratio of total inputs employed against total output produced for each organization. This ratio is then compared with others in the sample group to derive an estimate of relative efficiency. DEA identifies the most efficient providers of a good or service based on their ability to produce a given level of output using the least number of inputs. Other organizations in the sample group receive an efficiency score determined by the variance in their ratio of inputs employed to outputs produced relative to the most efficient producer in the sample group. DEA is therefore a measure of relative productivity among the sample group.

In contrast with other measures of productivity, DEA only requires data on the physical quantities of inputs employed and output produced, if technical and scale efficiency indicators are to be estimated. To estimate allocative efficiency, factor prices are also needed. Hence, the information requirements for DEA are fewer and less cumbersome than conventional TFP analysis, and the issue of allowing for different accounting treatments across organizations does not arise. Thus, in this study the DEA Malmquist approach was used to determine the TFP measurements of the Australian states’ natural gas distribution/ retail sectors over the period 1985 to 1999. Distribution/retail is the focus of this study for two reasons. First, data for the extraction and treatment component of the industry is not readily available. Transmission is not included because the inputs in this segment of the industry are heavily dependent upon the distance from the source of natural gas and city markets. For instance, in the Victorian market there is a relatively short distance between the Gippsland Basin and the city of Melbourne compared with the distance between the Cooper Basin in Central Australia and the markets of Adelaide, Sydney and Brisbane. A great deal more is therefore required in terms of both capital and labor inputs to transport the same amount of gas to the latter markets than the former. This ‘spatial’ problem is one not often encountered in gas distribution. In Australia, gas reticulation is confined to the state capitals and larger regional centers; gas distribution networks, therefore the gas industry, supply populations with similar densities in each state.

In the past, this approach of calculating TFP has not been used in the context of the Australian gas industry. For example, the Bureau of Industry Economics’ report on the gas industry (Australia, Bureau of Industry Economics 1994) used DEA to benchmark the performance of the Australian gas industry against overseas firms, but only for a single year rather than evaluating changes in performance over time. This approach has also often been used in relation to studying gas transmission/distribution in other countries such as the United Kingdom (e.g. Price and Weyman-Jones 1996).

The primary source of data for this study is the statistical publication of the Australian Gas Association. In terms of determining the level of output for the Australian gas industry, the amount of gas throughput in peta-joule in each state has been used. Inputs used include the capital stock and labor employed. However instead of using some sort of accounting method to estimate the stock of capital, physical kilometers of gas mains have been used (as per the approach used by the Bureau of Industry Economics 1994; and Price and Weyman-Jones 1996). In addition, as factor prices were not available, the estimates are of TFP decomposed into pure technical efficiency change, scale efficiency change and technological change, but not allocative efficiency. The final dataset used is composed of information from natural gas sectors in five states (New South Wales, Victoria, Queensland, South Australia and Western Australia) for the period 1985 to 1999. This is a balanced panel dataset with 75 observations.³ Table 8.3 presents descriptive statistics for each output and input variable in the sample data; all of the Malmquist TFP indices were derived using the DEA program (see Coelli 1996).