THE ELECTRICITY SUPPLY INDUSTRY: A STUDY OF AN INDUSTRY IN TRANSITION.

O' Mahony, Mary ; Vecchi, Michela

Michela Vecchi [*]

In 1989 the UK began a process of transferring an almost wholly state-owned electricity supply industry (ESI) into a collection of privately-owned generation, transmission and distribution utilities. Using data from 1960-97, this paper aims to evaluate how the performance of the UK ESI has changed over time and to compare the UK performance with France, Germany and the United States in order to assess the impact of the liberalisation process. The study takes a whole industry approach, combining the four aspects of electricity production - generation, transmission, distribution and supply. The computation of labour and total factor productivity and the impact on consumer prices are used to shed light on how successful the various industry structures have been in raising performance.

Industry structure in electricity supply

Introduction

In recent years many countries have initiated reforms to their Electricity Supply Industries (ESI). This process has been driven both by the technology of the industry and by economic considerations, as well as political motivations. Technological developments have made it feasible for small generators to produce electricity efficiently. Liberalisation of the electricity market is expected to result in important efficiency gains, by achieving a better co-ordination of resources, and reduction in costs and prices. Also, since the Single European Act of 1987, the European Commission has been committed to the implementation of the liberalisation of the network industries. The 1997 European Electricity Directive prescribes common rules for the progressive liberalisation of the national electricity markets within the EU.

The United Kingdom is in the vanguard of this process (Pollit, 1995). The privatisation and restructuring of the ESI, which began in 1989, involved a number of changes to the operation of the industry, the most recent of which is the introduction of competition in the supply of electricity to final consumers in 1999. The primary purpose of this paper is to benchmark the UK ESI against practice abroad and its own previous performance in order to evaluate the impact of the reforms on productivity and prices. Hence, the UK industry is compared to the ESI in Germany, France and the United States, thus evaluating relative efficiency in four countries with very different market structures/regulatory regimes over the period 1960 to 1997. In particular we measure growth in total factor productivity (TFP) over time in the four countries and benchmark relative productivity levels at points in time. This will allow us to shed light on how successful the various industry structures have been in raising performance.

The project takes a whole-industry approach, combining the four aspects of electricity production - generation, transmission, distribution and supply. Previous econometric studies examining the impact of industrial structure and regulation on efficiency tended to compare plants engaged in similar activities, such as generation in plants using similar types of fuel (see for example Pollitt, 1995, 1996; Zeitsch and Lawrence, 1996). But the performance of the ESI depends also on the production strategies pursued in total so that these studies tend to miss out companies engaged simultaneously in generation and transmission/distribution. For example very few studies consider the position in the vertically integrated ESI in France or Germany but, as will be shown below, important lessons can be learned from a consideration of electricity production in these two countries.

The following section presents the main features of the structure of the industry, discussing recent changes that have characterised the operation of the ESI in the four countries. We then describe the methodology used to evaluate labour productivity and TFP and we briefly discuss the data used in the empirical analysis. More detail on methodology and data sources can be found in a larger report underlying this paper (O'Mahony and Vecchi, 2000). The fourth section presents the main results on productivity and the fifth section then considers the additional question of how restructuring of the UK ESI has benefited final consumers by examining changes in the prices of output and fuel used in electricity production. Finally, the conclusion attempts to evaluate performance with reference to different market structures and recent changes.

The structure of the electricity supply industry

The technical structure of the industry

Fuel use and technology in generation

Differences across countries can be observed in the type of fuel chosen for generating electricity. The industry is highly energy intensive so that changes in the relative costs of various types of fuel have impacts on the choice of technology and efficiency with which capital is used. The main types of energy used are coal, oil, gas, hydro and nuclear energy and each type dictates unique technical characteristics of production.

Table 1 shows generation shares by fuel type for the total OECD and the four countries included in this study. In the OECD as a whole coal remains the most important fuel type followed by nuclear. In the 1990s natural gas became an important fossil fuel source, particularly in the UK, whereas the importance of oil was much reduced. Price movements as well as the development of the Combined Cycle Gas Turbines (CCGT) [1] in the late 1980s, have favoured the move towards gas. The UK, the US and Germany remain heavily dependent on coal whereas generation in France is currently almost 80 per cent based on nuclear fuel. The shares of renewable sources of energy other than nuclear and hydro remain small in all countries.

The choices between alternative forms of generation have involved both economic and political considerations. Both the relative prices of different fuels and technical progress have an impact on the type of generation technology used. In recent years relative fuel price movements have favoured a move to gas in the UK, and to a lesser extent in the US, with the rapid development of CCGT technology in the late 1980s. Also in an industry subject to direct or a high degree of government control or regulation, it is hardly surprising that political considerations impact on the choice of generating technology. Thus the impact of the ESI on other industries has been a factor, the most notable example being the dependence of the high cost British coal industry on electricity generators to sell its output. But Britain was not alone in Europe in using one industry to further the cause of another, for example the German electricity industry also 'protects' its coal industry.

Transmission and distribution

The transmission and distribution systems must ensure that generation is equal to demand at every moment in time or risk power failures. With uncertain demand and the probability that sometimes power stations either will break down or be closed for maintenance, it is necessary to devise a transmission system with some in-built reserve capacity, ready to operate at a moment's notice. In a wholly publicly controlled system, retaining the necessary reserve capacity does not present technical problems although there are efficiency considerations. In a privatised environment it is necessary to devise a system which ensures security of supply.

Wholesale markets for electricity vary across countries. Following privatisation the fundamental transmission mechanism in Britain was the spot market, or 'Pool'. In this market prices were set half-hourly so as to equate supply and demand. [2] High-voltage transmission is undertaken by a regulated monopoly, the National Grid Company (NGC). In Germany, until 1998, access to regional grids was either under local monopoly control or determined by contracts among energy producers. The German system has now embarked on a process of liberalisation of access (see Utilities Journal, 2000, for details). In the US the system consists of three major power grids or networks. These consist of high voltage connections between individual utilities designed to permit transfer of electrical energy across the network. The utilities within each power grid co-ordinate their operations and planning and buy and sell power among themselves. The French state owned monopoly Electricite de France (EDF) controls the entire transmissio n and distribution network in that country.

There are also differences in the systems of distribution and supply. In the UK the latter is carried out by the regional electricity companies (RECs) whereas in Germany and the US many generating companies also distribute and supply electricity, although most supply to final domestic consumers is through separate companies, often local monopolies.

Market structure and regulatory regimes

Ownership

The ESI was considered for a long time to have features of a natural monopoly and as such it has been subject to government regulation (in the case of private ownership) or direct public ownership in most countries. In the UK up to the end of the 1980s, and in France up to the current period, public ownership has been predominant. In France, Electricite de France (EDF) still operates under conditions of near monopoly, generating over 90 per cent of the country's electricity and operating the entire French transmission and most of the distribution networks.

The German ESI has a complex structure involving elements of geography and fragmentation, mixed ownership and partial regulation. It consists of three subsectors, public supply, industrial producers and generation by the federal railway system. Own generation by industry has been traditionally more important in Germany than in other countries, comprising 40 per cent of total output in 1960 falling to about 10 per cent in recent years.

Contrary to most European countries, the ESI in the United States has never gone through a process of nationalisation. Throughout its development, privately-owned utilities have been operating next to publicly-owned ones, the former accounting for a much greater share of the market.

Market structure

In the UK the state-owned company in charge of generation (Central Electricity Generating Board -- CEGB) and the twelve area boards in charge of distribution and supply were privatised in 1990. Following the 1988 White Paper, 'Privatising electricity', the CEGB was divided into four companies on 31 March 1990. These were three generating companies National Power (NP), PowerGen (PG) and Nuclear Electric (NE), and the National Grid Company (NGC). NP and PG were privatised in 1990. The twelve Area Boards were also privatised as the Regional Electricity Companies (RECs). The NGC was transferred to the joint ownership of the RECS and the RECs were sold to the public in December 1990. In 1995 the NGC was floated and the RECs were required to sell most of their shares in NGC.

A market for trading electricity, the Pool, was put in place exposing buying and selling of electricity to market forces. In terms of competition to final consumers it was decided in 1989 to proceed in three stages. The largest consumers were allowed to choose their suppliers from the start. Competition was extended to intermediate consumers in 1994 and to smallest consumers (households) in 1999. Since transmission and distribution were seen as natural monopolies, these were regulated by the Office of Electricity Regulation (OFFER). [3] The Director General of Electricity Supply was given legal duties to protect customers and promote competition. These two areas plus initial supply were regulated by price caps using the RPI-X formula. [4] The generation section of the market was subjected to less direct control with more effort directed towards increasing competition. This has allowed the entrance of several Independent Private Producers (IPPs).

In the US the type and mix of ownership of electricity producers is a complex structure. However, investorowned utilities provide approximately 75 per cent of the electricity sold to final consumers.

Since 1935, legislation in the form of Federal Power and Public Utilities Holding Company Acts has created vertically integrated and regulated monopolies. Utilities have been operating exclusively in areas assigned by the States in return for the universal obligation of providing the service. The regulatory requirements, at the federal and at the state level, have been particularly cumbersome, making the ESI one of the most heavily regulated sectors in the US economy. [5] Rate of return regulation was widely applied across the US. The regulated system worked quite well until the 1970s. Since then the two oil crises and the introduction of environmental regulations imposed greater constraints on firms -- regulation lags placed difficult financial conditions on utilities as the prices they were allowed to charge no longer covered their increased costs.

The first stage towards deregulation of the US ESI was marked by the Public Utilities Regulatory Policy Act (PURPA) in 1978. The act was primarily directed towards encouraging utilities to use resources more efficiently and to direct investments towards renewable resources. However, the act implicitly introduced competition at the generation level, by allowing non-utility companies that were using renewable energy sources to enter the market and by requiring utilities to buy power from them. The move towards a more competitive wholesale market was further encouraged by the Energy Policy Act of 1992. The Act extended the powers of the Federal Energy Regulatory Commission (FERC) in granting authority to private generators.

The introduction of retail competition in the US is facing more obstacles. One of the main issues is that some utilities will suffer the burden of the stranded costs, that is, costs related to pre-deregulation contractual arrangements, and that will not be recovered once competition is fully effective. New generators employing new cost-saving technologies and offering the service at lower prices, will attract new customers, leaving the other generators in major difficulties. The solution that most states are contemplating is to allow a full recovery of these costs, even if this implies a slower process towards competition in the short run. This process has recently run into serious problems, most noticeably those in California, where the price regulations prohibited utilities from passing on fuel price rises to final consumers, eventually leading to power shortages.

The structure of the German ESI up to 1998 was based on the 1935 Energy Law, Energiewirtschaftsgesetz (EnWG) which involved an exclusive territory concept with demarcation contracts between utilities concerning the geographical areas they supply. The Konzessionsvertrag contract gives regional suppliers a monopoly over their local community. In return it pays a large sum to the local communities which are an important source of revenues for a range of social services. Thus local governments have an interest in maintaining this uncompetitive system. A second system of contracts, the Demarkationsvertrage, draws boundary lines between territories, which are the exclusive markets of the electricity companies. Regional electricity companies therefore have a monopoly in their own area but also accept that they cannot compete in any other area, thus making the two types of contracts complementary. In addition to these horizontal contracts there are also vertical ones. These permit bigger industrial companies to buy d irectly from an electricity firm, thus bypassing the regional and local monopolies. The electricity generators also have price contracts with the regional companies. There were also rules of access to the cross-regional grids. In summary, the German system of contracts served to impede competition in the electricity industry.

Regulatory intervention takes a number of forms including controls on investment behaviour, entry and exit and intervention on prices. Investment is controlled by licensing requirements for new plants and capacity changes. Regulation of prices is confined to low-voltage retail sales and the wholesale market is virtually unregulated. The regulation of retail prices normally takes the form of a rate of return regulation with tariffs approved in the form of a price cap.

In Germany the liberalisation process has been slowed down by political considerations and by the complexity of its system. Moreover, the integration and the restructuring of the East German ESI imposed a further burden on the German authorities. In April 1998 the Energy Act came into force, following the EU Electricity Directive. This aims to reform the legal framework of the German ESI with an eye to greater liberalisation and deregulation but, to date, has not led either to any restructuring of the sector or to changed ownership (Bergman et al., 1999). But major changes are foreseen, for example mergers are planned among the larger interconnected utilities (see Utilities Journal, 2000).

In France the liberalisation process has been particularly slow and it has been one of the few countries that did not meet the deadline for translating the EU Electricity directive into national law. Hence, its system is still dominated by the EDF, which produces and transmits 90 per cent of the whole electricity produced in the country. Even the recently proposed law for the liberalisation of the ESI still gives EDF a position of dominance.

Table 2 presents a summary of the main features pertaining to ownership, market structure and regulatory regime in the four countries considered in this study. It illustrates that the four countries span a broad spectrum of structures. Hence a comparison of their relative performance in terms of efficiency and pricing should yield some interesting conclusions on the impact of structure on performance.

Productivity measurement

A first evaluation of the relative performance of the ESI compares productivity growth across the four countries. Labour productivity is one of the most straightforward ways to evaluate industry performance. This measure is simply expressed as output divided by the total number of workers. However, labour shares in value added are between 0.25 and 0.35 in the ESI, compared to about 0.7 for the aggregate economy (O'Mahony, 1999) so that labour productivity can only give a partial picture of the performance of the electricity industry. A more complete analysis of the industry performance can be obtained by computing Total Factor Productivity (TFP).

In this paper we adopt the approach to productivity measurement pioneered by Jorgenson and others (1987). As applied to the ESI, the general framework is a production function for gross output:

Y = Y(K, L, M, t) (1)

where Y is gross output, t is time, and K, L, M stand for capital, labour, and intermediate inputs, respectively. The growth of TFP over time, TFPG, can then be measured by a Tornqvist index:

[TFPG.sub.t] = [delta]ln[Y.sub.t] - [w.sub.Lt][delta]ln[L.sub.t]

-[w.sub.Kt][delta]ln[K.sub.t] - [w.sub.Mt][delta]ln[M.sub.t] (2)

where

[w.sub.X] = 1/2([w.sub.Xt] + [w.sub.[X.sub.t+1]]), X = K, L, M (3)

and the [w.sub.X] are the elasticities of output with respect to the inputs. Under the assumptions of perfectly competitive product and factor markets and constant returns to scale these elasticities can be measured by the share of each input in total revenue. With constant returns to scale, payments to each input exhaust the value of output so that the input shares in any one year sum to one.

It is important to note that output Y, and the inputs K, L, F and M, are themselves aggregates and can be further broken down if desired and possible. In the present study both [Y.sub.t] and [M.sub.t] are indices of, respectively, aggregate output and intermediate materials.

The ESI is a multi-output industry where electricity is supplied to different types of consumers at different prices. The Tornqvist index of aggregate output is given by:

[delta]ln[Y.sub.t] = [s.sub.it] [[[sigma].sup.3].sub.i=1] [delta]ln[y.sub.it] (14)

i = residential, industrial, other

where, as in equation (3), [s.sub.it] is the average revenue share, across two periods, for output i.

Similarly to equation (4), we construct a Torqvist index for fuel input:

[delta]ln[F.sub.t] = [[micro].sub.jt] [[[sigma].sup.4].sub.i=1] [delta]ln[f.sub.jt] (5)

j = coal, oil, hydro, nuclear

where the [[micro].sub.j]s are the average share, across two periods, of each fuel input in total fuel costs. In this study, labour input (hours worked), capital and materials other than fuel are all treated as single aggregate inputs.

We also present relative levels of productivity between two countries in a particular benchmark year. Relative output levels are estimated as follows. Letting j denote country and u the numeraire country (in this paper taken to be the US) then relative output in benchmark year T is given by:

[[Y.sub.j,T]/[Y.sub.u,T]] = {[P.sub.u,T]/[P.sub.j,T]} [[R.sub.j,T]/[R.sub.u,T]] (6)

where Y is aggregate output, P denotes output prices and R is nominal total revenues. The first term on the right-hand side, the average price ratios, is derived using geometric means of quantity-weighted price ratios for the three types of consumer, using both domestic country and US weights. Similarly cross-country TFP comparisons can be readily calculated, using equation (6) and a formula analogous to equation (2) with ratios across countries rather than time (see the discussion in O'Mahony, 1999).

The assumptions underlying the above calculation of TFP, and in particular those that output price equals marginal cost or that the industry operates with constant returns to scale, may not hold in practice. Output price in most countries is set by the regulator so the former certainly does not hold. In the case of the latter what are called differences in TFP might also be due to economies of scale. Also it is widely believed that generation and transmission are subject to increasing returns to scale (see Christensen and Greene, 1976; Gollop and Roberts, 1981 for US evidence; and Burns and Weyman-Jones, 1994, for evidence of economies of scale in UK distribution).

To calculate the sensitivity of the estimates to some of these problems we use a method employed in Hall (1990) where inputs are weighted according to their cost rather than revenue shares. If there are supernormal profits or if profits are affected by the actions of regulators, then the input shares should be calculated as payments to factors divided by marginal cost times quantity, rather than prices times quantity (revenue). The advantage of the revenue share approach is that capital's contribution can be measured as a residual having subtracted wage payments and materials costs from the value of output. In using cost shares we need to derive an appropriate user cost of capital, which involves additional problems.

Data sources and methods

The empirical analysis is based on annual data (1960-1997) for the whole industry. We distinguish three types of output, electricity supplied to domestic residents, industry and other sectors (largely commercial and government).6 Four types of fuel are also distinguished, coal, oil, gas and nuclear energy. In addition other intermediate inputs (largely water supply and business services) are also included. Labour input is measured as annual total hours worked.

Capital input is measured using the perpetual inventory method based on the real value of investment flows rather than the more commonly used generating capacity measures. The latter do not allow for capital accumulation in the transmission, distribution and supply sectors and do not take account of the varying cost of capital equipment according to type of fuel used in generation. Depreciation is assumed to be exponential using rates employed in the US national accounts for all four countries. Capital expenditures generally do not allow for future disposal costs of generating capacity. This is a particular problem for nuclear generation where future disposal costs are uncertain and may turn out to be much larger than anticipated. Given the relative 'newness' of nuclear technology, long asset lives for nuclear generators and the fact that many of the costs of using this technology have yet to be realised, it is difficult to give a precise estimate of the impact of disposal costs. Therefore we employed an extr eme assumption that the 'true' cost of nuclear generators was in fact twice the amount spent, assuming this additional expenditure should have been set aside to cover future disposal costs.

The implementation of equation (2) also requires data on payments to factor inputs. Total payments to labour are readily available for the four countries as are prices for coal, oil and gas -- the latter can be combined with quantities used to derive a total cost for each of these fuel inputs. A problem arises in the case of nuclear fuel since the quantity measure, uranium, represents only a fraction of the actual cost of nuclear fuels. Adjustments were made to include both front end costs, conversion of raw uranium into uranium hexafluoride ([UF.sub.]6), enrichment and fuel fabrication, and back end costs, the process of storage and disposal of the spent fuel. An adjustment factor to allow for costs of materials other than fuel was included, based on data from production censuses.

The calculation of relative levels of fuel input in 1993 follows the approach used to calculate relative levels of output given in equation (6) above. Hence the total costs of fuel in each country in domestic currency is converted to US$ using the relative prices of total fuel input. These in turn are calculated using quantity weighted price relatives for each of the four types.

In the basic TFP equation capital's share is estimated as a residual. But in order to assess the sensitivity of the estimates to the underlying assumptions, in particular constant returns to scale, it is necessary to have independent estimates of capital's share. This is achieved by multiplying the value of the capital stock by its user cost, using the well-known formula in Hall and Jorgenson (1967). [7]

Annual data series for the required quantities and prices of outputs and inputs were generally taken from national sources including each country's national accounts, statistical yearbooks and publications of energy statistics. The main data sources are listed in Appendix A. Further details on methods, assumptions employed, and exact data sources are given in O'Mahony and Vecchi (2000).

Results

Labour productivity in the ESI

We begin our assessment of relative performance of the ESI in the four countries by examining changes in labour productivity, defined as output per hour worked, shown in table 3. The average performance of the UK over the whole period has been very high, with rates of growth over 5 per cent. A similar performance can be observed in France. Germany and the US show lower growth overall. In all four countries labour productivity growth is lower in the period post-1979 than the average achieved in the 1960s and 1970s. Only France shows better performance in the 1980s than in the 1990s. In the remaining three countries labour productivity growth is higher in the latter decade but the acceleration is much greater in the UK, coinciding with the restructuring and privatisation of the industry. Labour productivity growth in the post-privatisation period in the UK is above the average for the whole sample and above the other country's performance. It is this surge in labour productivity growth that is often argued to b e evidence supporting the benefits of the changing structure of the UK ESI. Growth in labour productivity in the ESI in the 1990s has been particularly high relative to the about 2 per cent norm achieved for aggregate GDP or manufacturing. In fact similarly high growth rates are generally observed only in sectors where considerable restructuring and change of ownership have occurred, most notably in gas supply, communications and individual manufacturing sectors such as iron and steel (O'Mahony, 1999).

Total factor productivity estimates

We now come to a consideration of the main results of this paper, relative total factor productivity (TFP) performance, which takes account of differences in capital intensity and variations in the fuel mix in electricity generation. Table 4 presents a cross-country comparison of rates of growth of TFP, which can be interpreted as the extent to which output has grown having taken account of the growth in all factor inputs. In all four countries TFP growth is on average positive, indicating that improvements in the methods of organising production are important as a source of growth in the ESI. Over the entire period TFP growth in Britain is better than that achieved by Germany and only slightly below rates in the US. France is characterised by the best performance with an average rate of growth about 60 per cent higher than that achieved in the UK or the US. Hence experience of using the new nuclear technology and economics of replication are likely to have been important in France.

In terms of TFP growth, the UK performance in the post-privatisation period looks unimpressive, in contrast to the results for labour productivity growth. In fact, the rate of growth of TFP in the last ten years is only 1.2 per cent, a figure below the UK period average of just over 2 per cent. It is also considerably lower than that achieved by France or the US, and marginally lower than rates in Germany. Aggregate input growth was almost static in Britain in the period post-1989 with the very large increases in capital investment being counter-balanced by reductions in labour input. Hence productivity improvements were important in contributing to output growth in that period. However, the bottom line is that privatisation and restructuring of the UK ESI have not, to date, delivered the improvements in productivity which were expected at the outset.

We next consider relative levels of total factor productivity, shown in chart 1, which benchmarks to US levels in 1993. In the most recent year for which reliable data are available, the ESI in France produces the greatest amount of output for its level of factor inputs, followed closely by the US. Britain produces about 15-20 per cent less output for its inputs than either of these countries but achieves more than the German ESI which only manages to produce half the output per unit of input of the French. Over time the UK ESI achieved almost complete convergence to US levels by the end of the 1970s but fell behind again in the subsequent two decades. The French ESI, on the other hand, underwent a process of almost continuous convergence on US productivity levels and took over as the productivity leader in the early 1990s. The US has, however, gained some ground on the French in the past few years. The productivity record of the German ESI can only be described as dismal with that country's position relative to the other three much worse now than in the 1960s.

In summary, over the entire period since 1960, TFP growth rates in Britain were on a par with those achieved in the US and better than in Germany. Growth rates in Britain were however lower in the post-privatisation period so that restructuring did not yield the expected productivity gains. Using preliminary data available up to the third quarter of 2000 suggests no trend improvement in UK productivity in more recent years. The results show the French state-owned monopoly outperforming the other three countries, in particular following the widespread use of nuclear technology in the late 1970s. However relative performance depends on the validity of the underlying assumptions. Hence we now look at the sensitivity of the results to such assumptions.

Sensitivity analysis

We now consider the extent to which the results above are sensitive to the assumptions underlying the basic TFP calculations. A number of sensitivity tests turned out to have only minor impacts. [8] In terms of fuel input there may be problems in the method used to construct the nuclear component in going from raw uranium to total nuclear input. We experimented with a number of assumptions and found that the TFP results were not very sensitive to this. For example, increasing the conversion costs by a third reduced the French TFP growth in the period 1979 to 1989 by about 0.4 percentage points per annum and in the period since 1989 by about 0.2 percentage points. This, however, by no means diminishes the French advantage. Changes in this assumption have very little impact on other countries, given their low share of nuclear fuel, or in France in other periods. Variations in the shares of other materials inputs or the depreciation rate on capital similarly had little impact.

We next consider the implications of using cost rather than revenue shares to weight input contributions in calculating TFP. [9] To do so we estimated total costs as the sum of payments to labour, intermediate inputs and capital, the latter derived by multiplying capital stocks by the user cost of capital. The results are shown in Appendix table A1. Use of the alternative cost shares makes almost no difference to the results for the UK and the US. The French results do change for the period 1960 to 1979; the cost shares imply higher TFP growth in this period but this may be a function of assumptions needed to calculate capital for France prior to 1981 when investment flows were not readily available from the national accounts.

The two sets of TFP estimates show somewhat greater differences in the case of Germany. The use of cost shares reduces TFP growth in all periods, with the greatest difference in the period 1979-89, rendering the comparative performance of the ESI in that country even worse than the situation represented in table 4 above. This suggests that monopoly profits are probably more prevalent in that country, consistent with our observations on the industry structure.

Output and fuel prices in the ESI

In the evaluation of performance in the ESI we also consider changes and cross-country differences in output and fuel prices, as presented in table 5. One of the theoretical claims in favour of the privatisation process and the move towards a more competitive market structure is a decrease in the price level and hence increased welfare for final consumers. Therefore, we expect prices to be lower in the US, where the ESI is mainly under private ownership, and in the post-privatisation period in the UK. Some commentators, e.g. OXERA (1999) have pointed to lower output prices in real terms in the UK as a measure of the success of the restructuring program. But others, such as Newbery and Pollitt (1997), have suggested that these lower prices are likely to be a consequence of lower real fuel prices and that Britain has not performed better than other countries in this respect. This section considers these claims.

Output prices

We look first at the electricity price level in each country, taking into account the different final users of electricity. In all countries residential users are charged the highest price, with the only exception being the UK in the 1960s and mid-1970s when prices for the residual sector were marginally higher. Electricity is a very important input into industrial production, hence all countries recognise the necessity of keeping prices low for this substantial group of users. Over time we can observe similar trends across the four countries. Prices decreased throughout the 1960s, increased after the two oil shocks and decreased again in the mid-1980s and in the 1990s. In the UK prices decreased at a faster rate in the past ten years than in other countries. In terms of price levels, US prices are among the lowest in the sample, being above French prices only during short intervals. The residential sector is in this case the one where prices are comparatively lower. This seems to support the claim that a sys tem based on private companies can result in lower prices. However, in France, the monopolistic structure of the industry and its public ownership has also resulted in low prices since the early 1980s. In the UK the slowdown in electricity prices started in the mid-1980s and prices are still on a downward trend with the best performance in the residential sector.

Fuel prices

The debate about post-privatisation prices in the UK is still very active. Despite the decline in prices shown by the data, authors argue that consumers would have been better off had privatisarion not taken place because the price reduction has been the result of falling fuel prices and technology advances, in particular the shift to gas cycle combined heat and power generation. Therefore, the fall in prices would have taken place regardless of the change in the ownership structure of the industry (Newberry and Pollit, 1997; Yarrow, 1992). Some argue that the fall in prices would have been even stronger had privatisation not taken place and would have affected residential consumers more positively (Branston, 1999). We therefore turn our attention to fuel prices, to see how these have varied across the four countries.

In the 1970s there was a noticeable rise in oil and gas prices with a less pronounced increase in nuclear and coal prices. Coal prices followed a decreasing trend in the United States from the late 1970s onwards, while they increased in the European countries, in particular in the United Kingdom and in Germany. In France and in the US coal is the cheapest fuel input after nuclear, while in the UK and Germany the price of coal, oil and gas fluctuates more and the relative cost of using one input or the other varies according to the reference period. In the late-1990s the price of coal drops in both countries. Across all countries we can also observe a decrease in gas prices, particularly in the European countries Nuclear fuel prices declined in the mid-1980s and his trend continued throughout the 1990s.

The United States is characterised by the lowest price of gas and coal throughout the whole sample. The price of gas in the UK has experienced a strong downward pattern compared to the other countries and in recent years has been slightly above the US level. The widest cross-country variations can be observed in the price of coal. In Germany the price of coal was only slightly above other countries at the beginning of the sample period but has risen considerably from the early 1980s, while a decrease can be observed in the remaining countries.

To what extent are changes in fuel prices responsible for the observed changes in output prices? To answer this we calculate the growth in average output prices minus average fuel prices as follows:

[delta] ln [P.sub.t] = [[sigma].sub.i][s.sub.i]ln([p.sub.it] / [p.sub.i,t-1])-[s.sub.f] [[sigma].sub.j][[micro].sub.j] ln([[p.sup.f].sub.jt] / [[p.sup.f].sub.j,t-1]) (7)

where p and pf are output and fuel prices, respectively, [s.sub.i], is the share of output i in total revenue, [[micro].sub.j] is the share of fuel type j in total fuel costs and [s.sub.f] is the share of fuel in total revenue, averaged over periods t and t-1. [10] Deflating (7) by the consumer price index gives a measure of real, fuel-adjusted output prices. The results are shown in table 6.

These calculations show that fuel-adjusted real prices declined on average over the entire period in all four countries with the greatest decline in the US. Since 1979 France has outperformed the UK and the US, with the rate of decrease highest in Germany post-1989. In the entire period since privatisation, the rates of decrease achieved in the UK have not been similar to that achieved by the other three countries, justifying the criticisms of other authors that the changes did not yield the expected benefits to consumers immediately.

However there are differences within the privatisation period. Up to 1993 fuel-adjusted real output prices in fact rose in Britain and the UK was alone in the four countries in showing this increasing trend. But after 1993 these prices began to decline rapidly in Britain. Hence there may be reasons to believe that the benefits to consumers are now being felt, although with lag. It is possible to update these price changes for the UK to the end of the third quarter 2000. In the period from 1997 to this recent date real fuel-adjusted prices declined at an average annual rate of about 2.6 per cent which is lower than that achieved in the previous four-year period. In 1993 the regulator insisted that the large power producers divested some of their capacity and there was considerable entry by independent power producers. Thus the move to more competition in the generation sector probably accounted for much of the subsequent reduction in price. It is probably too early yet to say if the changes in retail competiti on have had a significant impact but the evidence to date suggests that this may be less important than competition in generation in yielding welfare gains to consumers.

Conclusions

The results of this paper suggest that the productivity record of the UK ESI after privatisation can be summarised as being unremarkable. Although labour productivity growth rates were very high in the period since privatisation, this was more than counterbalanced by increases in other inputs. Relative to its own past experience, or that in other countries, productivity in the UK privatised industry does not appear to have shown any pronounced improvement.

If there has been a benefit to consumers from privatisation, it appears to have occurred only in the final years of the study, when real electricity prices seem to have fallen by more than elsewhere and have been more than accounted for by decreases in fuel costs. Since there was no corresponding upsurge in productivity in this period, it is likely that the reduction in prices occurred primarily from increased competition in the generating part of the industry.

The results presented on total factor productivity suggest that in terms of country rankings, France outperforms all four countries, with little difference between the UK and the US and Germany, who performed particularly badly. These results appear to be robust to changes in the assumptions underlying the calculations. It remains to ask what general conclusions emerge from this on industry structure and regulation.

Firstly on ownership, the results do not support the argument that private ownership per se leads to better productivity performance. The best performer is the publicly-owned French industry and the achievements of the formerly publicly-controlled CEGB in the UK were not particularly bad by international standards. The German ESI, which operates under a mixture of private and public ownership, shows the worst performance.

An argument against public ownership is that stateowned firms can suffer from excessive political intervention. There is ample evidence of state interference in the UK CEGB but this does not appear to have been particularly detrimental to productivity performance. But the degree of political intervention would appear to have been greater in Germany, where much of the industry is privately owned. In that country the industry is very fragmented and much of the intervention is at a localised level. This suggests that it is the relative bargaining power between politicians and the firms in the industry that is important. Managers of a large publicly owned monopoly may be in a better position to withstand pressures by politicians to interfere in pursuit of broader political aims.

The second issue relates to the impact of market structure on incentives towards efficient production. Again the French results put a question mark over the notion that a competitive environment is preferable to a monopoly situation. The French success was, however, due to a government directed move to producing with the technology which at that time generated large productivity gains. On the other hand, the poor relative performance of Germany, with its widespread local monopolies and state enshrined demarcation contracts, provides evidence against anti-competitive structures.

The very large increase in labour productivity in the UK post-restructuring suggests that the changes did affect incentives to efficient use of this input. But the increased investment during that period, when there was evidence of overcapacity in the industry, runs counter to this argument. The latter may have resulted from the historical accident of the development of a cheaper technology, combined cycle gas turbines, coincidental with the restructuring of the industry. It remains to be seen if overall productivity increases in the future, when the older coal-fired plants are retired. Finally TFP growth rates in the US, although not spectacular in any period, probably present the best evidence in favour of the competitive thesis. Without any major restructuring, as in the UK, or redirection of the industry to a dependence on a particular type of technology, as in France, the US largely maintained a reasonable rate of growth in output for given inputs in the period under consideration in this paper.

The third issue is the extent to which different forms of regulation have impacted on the industry. Both the US and Germany have used rate of return regulation so that the very different outcomes in these two countries imply no clear judgements on its benefits. On the other hand the RPI-X form of regulation in Britain also does not seem to have delivered its expected productivity improvements. But the impact of regulation depends not only on the price control mechanism employed but also on the general approach taken by the regulators. In the UK, regulation of the transmission and distribution components have been different from that applied to generation. The results in this paper indicate that the total environment in which the industry operates, its industrial structure, the technology used, and the general political climate are likely to have a greater impact on productivity than the particular type of regulatory framework employed.

In more general terms the reduction in final consumer prices in the UK in recent years suggests that more competition may ultimately have beneficial effects on consumer welfare. It is too early to ascertain the extent to which the recent drive towards competition in supply will reinforce this trend but this is likely to lead to further reductions in price. Ultimately however, increases in productivity will be required to deliver long-term benefits and these have yet to surface.

Finally, this paper has concentrated on a narrow measure of productivity within the ESI. aWe believe this is important in assessing how effective the industry is in producing an output which is of fundamental importance in modern industrial society. However it neglects other considerations of a broader nature, the two most important of which are its impact on other industries and on environmental pollution. There is no doubt that privatisation of the ESI in the UK has had a devastating effect on the coal industry. The German government has made a decision not to go down this route. Calculating the costs on the physical environment of the different strategies adopted by the countries is an extremely difficult exercise and beyond the scope of this paper. But we do not wish to suggest that these are insignificant. In this respect the German industrial structure has led to more expenditure on pollution-abating technologies than in the other countries. It may be that the typical German consumer believes that the b enefits from this outweigh the loss in productivity.

(*.) National Institute of Economic and Social Research. This study was supported by funding from the Leverhulme Trust. We would like to thank Nicholas Oulton. Alan Horncastle, Catherine Waddhams and colleagues at NIESR for comments.

NOTES

(1.) This technology involves operating at a lower scale than either coal-fired or nuclear stations, has short construction times and low capital operating costs.

(2.) The pool system has been substituted by a market-base system for wholesale power called the New Electricity Trading Arrangements (NETA).

(3.) Recently amalgamated with the gas regulator to form a single body, OFGEM.

(4.) This form of regulation is also called price regulation. The prices are fixed by a regulator and they are not allowed to rise above a certain level. The price cap is fixed for a certain period of time, usually 3-4 years. During this period the firm is not allowed to increase its price by more than a pre-determinate percentage per year. This percentage is given by the formula RPI-X, that is the rate of growth of prices must not exceed the growth rate of the retail price index minus some anticipated rate of technological change. Compared to the rate of return regulation employed in many countries, firms have downward price flexibility in a price-cap regime and, at least in theory, firms have an incentive to keep their costs low and increase efficiency, so they can earn large profits.

(5.) A large number of contributions to the literature, often critical towards regulation, has been published in the United States, see, for example, Stigler and Freeman, 1962; Stigler, 1971, 1973; Baumol, 1982. One of the main issues highlighted was the finding that regulation did not result in lower prices and concerns about the possibility that the regulator is subject to strong political pressures. Moreover, the rate of return regulation, prevalent in the US, gave firms little incentive to increase efficiency, since it virtually guaranteed the reimbursement of all their costs (Arocena and Waddams Price, 1999).

(6.) Exports are treated as a separate output for France, the only country where they are significant as a share of total output.

(7.) In this paper we ignore any influence from differences in investment tax rates and credits.

(8.) Details of the various sensitivity analyses undertaken are available in O'Mahony and Vecchi (2000).

(9.) The estimation of payments to capital are problematic, in particular they are dependent on the interest rate used. Also, the capital gains term in the user cost formula makes the year-on-year variation greater when cost shares are used, particularly in the I 970s. For these reasons we prefer to present estimates based on weighting inputs by revenue shares as our main results.

(10.) This is the fuel input component in the 'dual' approach to TFP measurement which begins with a translog cost function rather than a production function (Chambers, 1988).

REFERENCES

Arocena, P. and Waddams Price, C. (1999), 'Generating efficiency: economic and environmental regulation of public and private electricity generators in Spain', Applied Economics, 28, pp. 29-39.

Baumol, W.J. (1982), 'Contestable markets: an uprising in the theory of industry structure', American Economic Review, 72(1), pp.1-15.

Bergman, L., Brunekreeft, G., Doyle, C., von der Fehr, M.N., Newbery, D.M., Pollit, M. and Regibeau, P. (1999), A European Market for Electricity? Monitoring European Deregulation 2, London, Centre for Economic Policy Research.

Branston, J.R. (1999), 'A counterfactual price analysis of electricity privatisation in England and Wales', Industrial Development Policy Discussion Paper, no. 7, Institute for Industrial Development Policy.

Burns, P. and Weyman-Jones, T.G. (1994), 'Regulatory incentives, privatisation, and productivity growth in UK electricity distribution', Centre for the Study of Regulated Industries, Technical Paper No. 1.

Chambers, R.G. (1988), Applied Production Analysis: A Dual Approach, Cambridge, Cambridge University Press.

Christensen, L. and Greene, W.H. (1976), 'Economies of scale in U.S. electric power generation', Journal of Political Economy, 84 (4), pp. 655-76.

Gollop, F.M. and Roberts, M.J. (1981), 'The sources of economic growth in the U.S. electricity power industry', in Cowing, T.G. and Stevenson, R.E. (eds), Productivity Measurement in Regulated Industries, New York, Academic Press.

Green, R. and Newbery, D.M. (1997), 'Competition in the English electricity industry', Oxford Review of Economic Policy, 13(1), pp. 27-46.

Hall, R.E. (1990), 'Invariance properties of Solow's productivity residual, in Diamond, P. (ed.), Growth/Productivity/Unemployment: Essays to Celebrate Bob Solow's Birthday, Cambridge, MA, The MIT Press.

Hall, R.E. and Jorgenson, D.W. (1967), 'Tax policy and investment behaviour', American Economic Review, 57, pp. 391-414.

Jorgenson, D.W., Gollop, F.M. and Fraumeni, B. (1987), Productivity and U.S. Economic Growth, Amsterdam, North-Holland.

Muller, J. and Stahl, K. (1996), 'regulation of the market for electricity in the Federal Republic of Germany', in Gilbert, R.J. and Kahn, E.P. (eds), International Comparison of Electricity Regulation, Cambridge, Cambridge University Press.

Newbery, D.M. and Pollitt, M.G. (1997), 'The restructuring and privatisation of Britain's CEGB - was it worth it?', The Journal of Industrial Economics, XLV(3), pp. 269-303.

OECD (1998), Electricity Information, Paris, OECD.

O'Mahony, M. (1999), Britain's Relative Productivity Performance, 1950-1996: An International Perspective, London, National Institute of Economic and Social Research.

O'Mahony, M. and Vecchi, M. (2000), 'The electricity supply industry: a study of an industry in transition', Final report to the Leverhume Trust.

OXERA (1999), Guide to the Economic Regulation of the Electricity Industry, Oxford, The OXERA Press,.

Pollitt, M. G. (1995), Ownership and Performance in Electric Utilities: International Evidence on Privatisation and Efficiency, Oxford University Press, Oxford.

-(1996), 'Ownership and efficiency in nuclear power production', Oxford Economic Papers, 48, pp. 342-60.

Stigler, G. (1971), 'The theory of economic regulation', Bell Journal of Economics and Management Science, 2, pp. 3-21.

-(1973), 'Economic competition and political competition', Public Choice (Fall), pp. 9 91-106.

Stigler, G. and Freeman, C. (1962), 'What can regulators regulate? The case of electricity', Journal of Law and Economics, 5 (October), pp. 1-16.

Utilities Journal, January 2000, OXERA.

Yarrow, G.K. (1992), British Electricity Prices since Privatisation, Oxford, Regulatory Policy Institute.

Zeitsch, J. and Lawrence, D. (1996), 'Decomposing economic inefficiency in base-load power plants', Journal of Productivity Analysis, 7, pp. 359-78.

Table 1 Electricity production, 1996: fuel type as a per cent of
total generation
 Thermal Hydro Nuclear Other a Total
 Coal Gas Oil
OECD 39.2 12.4 7.6 16.2 24.2 0.5 100
UK 42.8 23.9 4.0 1.4 27.7 0.1 100
US 53.1 13.7 2.6 10.4 19.7 0.6 100
Germany 55.3 8.8 1.5 4.9 29.2 0.4 100
France 6.1 0.8 1.5 13.7 77.8 0.1 100
Source: Electricity Information, OECD, Paris, 1998 edition.
(a)Other includes: geothermal, solar, tide, wave, ocean, and
wind energy.
Table 2 The structure of the ESI in the four countries
 UK
Market structure To 1990: monopoly;
(horizontal) post-1990: oligopoly
 (generation); monopoly
 (transmission); local
 monopoly (supply 1998);
 competition (supply
 from 1999)
Market structure Separate components
(vertical)
Ownership To 1990: public (CEGB);
 post-1990 private
Regulation RPI-X
 US France
Market structure Oligopoly (generation Monopoly (EdF)
(horizontal) and transmission);
 local monopoly (supply);
 competition in some
 areas of generation and
 supply
Market structure Partly vertically integrated Vertically integrated
(vertical)
Ownership Private, public, mixed Public
Regulation Rate of return Public control
 Germany
Market structure Oligopoly (generation and
(horizontal) transmission); local
 monopoly (supply)
Market structure Partly vertically integrated
(vertical)
Ownership Private, public, mixed
Regulation Rate of return, legal
Table 3 Labour productivity growth in the ESI (% per annum)
 UK US France Germany (a)
60-97 5.37 3.97 5.21 3.88
60-79 6.07 4.40 6.25 5.94
79-97 4.63 3.53 4.12 1.70
79-89 2.45 2.24 4.70 0.98
89-97 7.36 5.13 3.39 2.60
(a)The results in this and subsequent tables refer to the
former West Germany.
Table 4 TFP growth rates, 1960-97 (% per annum)
 United Kingdom United States France Germany
1960-97 2.05 2.20 3.38 1.84
1960-79 2.54 1.85 3.08 1.52
1979-97 1.53 2.57 3.69 2.17
1979-89 1.78 2.60 4.17 2.84
1989-97 1.22 2.55 3.09 1.32
Table 5 Real output and fuel prices, (1993 US $)
 United Kingdom United States
 Resid. Ind. Other Resid. Ind.
Output prices
60-97 0.110 0.077 0.109 0.095 0.052
60-79 0.105 0.082 0.118 0.100 0.048
79-97 0.116 0.073 0.101 0.090 0.058
79-89 0.121 0.081 0.109 0.097 0.065
89-97 0.110 0.063 0.090 0.082 0.049
 France Germany
 Other Resid. Ind. Other Resid. Ind. Other
Output prices
60-97 0.091 0.119 0.062 0.086 0.185 0.108 0.143
60-79 0.096 0.121 0.063 0.088 0.182 0.112 0.146
79-97 0.087 0.116 0.060 0.084 0.187 0.103 0.140
79-89 0.095 0.122 0.066 0.089 0.194 0.111 0.148
89-97 0.077 0.110 0.053 0.077 0.182 0.094 0.131
 Coal Oil Gas Coal Oil Gas Coal Oil Gas
Fuel prices
60-97 1.32 1.72 1.50 0.78 1.66 1.08 1.09 1.51 2.23
60-79 1.21 1.54 1.13 0.81 1.52 0.77 1.16 1.54 1.94
79-97 1.45 1.94 1.90 0.77 1.84 1.43 1.03 1.49 2.54
79-89 1.74 2.60 2.35 0.93 2.36 1.73 1.27 1.90 3.06
89-97 1.10 1.06 1.32 0.58 1.15 1.04 0.74 0.93 1.82
 Coal Oil Gas
Fuel prices
60-97 1.96 1.89 2.44
60-79 1.66 1.75 2.21
79-97 2.29 2.08 2.68
79-89 2.45 2.46 3.27
89-97 2.12 1.56 1.86
Notes: Prices are average levels over
time periods, converted using the 1993
dollar exchange rate and to real levels
using each country's consumer price
index as a deflator. Output prices,
$ per GwH, Fuel prices, $ per mtoe.
Table 6 Real price growth after adjusting for fuel price growth
 UK US France Germany
1960-97 -1.49 -1.69 -1.36 -1.45
1960-79 -2.11 -2.32 -1.60 -2.47
1979-97 -0.82 -1.03 -1.12 -0.27
1979-89 -0.41 -0.75 -1.05 1.17
1989-97 -1.34 -1.38 -1.20 -2.07
1989-93 1.76 -1.05 -0.45 -1.49
1993-97 -4.44 -1.72 -1.96 -2.64

Appendix: Data Sources

United Kingdom

Gross output, own purchases and net output in TWh: source: Energy Trends (ET), Dept. of Trade and Industry and Digest of Energy Statistics (DES), Dept. of Energy, various issues.

Output prices. These were taken from EI, ET and DES.

Labour input and compensation. Employment series were taken from data in Labour Market Trends, previously the Department of Employment Gazette and Labour Statistics Yearbooks. Average annual hours are for total utilities (electricity, gas and water) from O'Mahony (1999). The Annual Censuses of Production, ONS and CSO provided data on labour compensation per head, which was used instead of total compensation as there was a major break in the latter after 1989. This was then multiplied by total employment to derive an estimate of the wage bill and hence labour's share of revenue.

Fuel input, quantities, million tonnes of oil equivalent. This is available for four inputs: coal, oil and gas and nuclear. From 1986 the series were taken from ET, including own generators for coal and gas, pre-1986 trends are for public supply. Nuclear fuel was converted to lbs of uranium using a conversion factor of 1.97 millions of lbs per mtoe - this factor was based on the average for the US.

Fuel prices. The primary sources were ET, DES and the Annual Abstract of Statistics (AA), ONS and CSO. From 1960 to 1962, price indices for British coal were used from AA. The UK data do not present a series for the prices of uranium so that the US price series was used, converted to [pound] using the market exchange rate.

But the 1997 price per lb of uranium was taken as the unit value of imports.

Capital stocks. These were based on real investment series from the national accounts, see O'Mahony (1999) for details, and were updated using investment data from the 1999 edition of The Blue Book , ONS.

United States

Output, quantities and prices. The sole data source is historical series given in the Annual Energy Review (AER) 1998, Energy Information Administration.

Labour input. Employment and hours from O'Mahony (1999) updated to 1997 using data from Statistical Abstract of the United States (SA), US Department of Commerce. Labour compensation per head was derived from hourly wage data for manual workers in Electric Services from Employment and Earnings, US Bureau of Labor Statistics. These were supplemented for non-manual workers using the ratio of wages of non-manual to manuals for total utilities (electricity, gas and water) from the same source.

Fuel input, quantities, AER. Uranium purchases in million lbs are available in SA.

Fuel prices, coal, oil and gas, in $ per toe were derived from EI annually from 1984 to 1997. Before 1984 we used price indices for coal, crude petroleum and industrial gas from SA. The price per lb of uranium was taken from SA.

Capital stocks. Construction of the PIM was based on investment data from the US, Bureau of Economic Analysis, purchased directly or more recently from their web-site, see O'Mahony (1999) for details.

Germany

This refers to the former West Germany throughout but in the final few years (1995 to 1997) the data sources generally include the former East Germany so that growth in the former is assumed to follow that in the total unified Germany.

Output, quantities in TWh. From Statistisches Jahrbuch (SJ), Statistisches Bundesamt, various issues.

Output prices. Producer price indices exist for five categories, households, industry, agriculture, low voltage other and high voltage other. In fact only two tariffs exist, one for households and small businesses and one for industry (Muller and Stahl, 1996) although these tend to vary across regions. The price index for 'other' was calculated as a weighted average of households and industry with weights for the former being consumption of agriculture and 'other purchasers' and weights for the latter comprising the remainder of consumption (transport and distribution). Since these are price indexes it was necessary to splice the prices from Electricity In formation for 1990 converted from $ to DM to obtain a price in DM per kwh.

Labour input. Employment and hours, from O'Mahony (1999), originally from the national accounts (Volkswirtscaftliche Gesamtrechnungen (vG)) and from hours series developed in research institutes in Nurnberg (IAB) and Berlin (DIW). Total labour compensation is taken from VG.

Fuel quantities. This is available in SJ, divided into coal (hard and soft (lignite)), natural gas, oil and nuclear.

Fuel prices. Price indices from 1969 to 1997 for coal and fuel oil, and from 1975 for natural gas, were derived from the series materials prices for use in industry, from SJ. SJ presents a series for price of imported nuclear fuel from 1976.

Capital input. Capital stocks were estimated using real investment data from VG combined with historical series, see O'Mahony (1999) for details.

France

Output, quantities in TWh. From Annuaire Statistique de la France AS, INSEE.

Output prices. These were based on a mixture of indices and prices in francs per kWh from AS and EI.

Labour input. Employment: Series 'Personnel des exploitations de production et de distribution d'electricite', from AS. Hours: Average annual hours worked are taken from O'Mahony (1999) and refer to the total utilities sector (electricity, gas and water). Labour compensation: for 1989 onwards compensation per head was taken from the French production censuses, La situation de l'industrie, INSEE. Before this the only information available on average employee remuneration was for total utilities in AS.

Fuel input. Series in mtoe for three types, coal, oil, natural gas and for tonnage of uranium are available in AS.

Fuel prices. Price for coal in $ per toe from El, before that we used the gross price index for coal from AS. The prices for oil and gas were only available in index form from AS and these were gross industrial use rather than specific to the ESI. The price per lb of imported uranium was based on trade unit values, available through Eurostat and in earlier use followed those in Germany, converted using the exchange rate.

Capital input. National accounts real investment data were only available for total utilities (electricity, gas and water). Nominal investment data for the ESI were taken from the French production census from 1981 and deflated using price indices for investment goods for total utilities based on data obtained directly from INSEE (see O'Mahony, 1999, for details). Investment in electricity represents over 80 per cent of total utility investment. Before 1981 the data were constructed from the change in MW of capacity, weighting capacity for each type of fuel use by its relative construction costs from the OECD NEA studies (OECD 1983, 1986, 1989, 1998). A starting stock in 1960 was derived as a proportion of the 1960 stock for total utilities calculated in O'Mahony (1999) using relative employment as the factor of proportionality. Capacity in MW was taken from AS.

Appendix Table AI.
A comparison of TFP growth rates under alternative factor share
assumptions. (Percentage point difference between growth
rates using revenue shares and growth rates using
cost shares.)
 UK US France Germany
1960-97 -0.08 0.03 -0.19 0.23
1960-79 -0.05 0.04 -0.42 0.11
1979-97 -0.11 0.02 0.04 0.37
1979-89 -0.16 0.13 0.02 0.61
1989-97 -0.05 -0.12 0.08 0.07