1 Introduction and summary *

2 Differences in productivity measures and their use in price control *

3 CEPA’s method of drawing conclusions from the evidence *

4 Methodological issues *

5 The evidence *

6 Partial factor productivity *

7 Conclusions *

1. Introduction and summary

As part of its review of distribution network operators (DNOs) the Office of Gas and Electricity Markets (Ofgem) has published work by Cambridge Economic Policy Associates Ltd (CEPA) entitled Productivity Improvements in Distribution Network Operators. United Utilities Electricity (UUE) has asked Horton 4 Consulting to comment on it.

Work on comparative efficiency has tended to confirm that it is difficult to reach reliable conclusions on the comparative efficiency of DNOs. If, as seems likely, it is not possible to reject the hypothesis that companies are efficient, price controls should be set on the basis that an individual company’s future costs will only differ from present ones in real terms insofar as total industry productivity growth (or factor price change) is likely to differ from that in the economy as a whole. It is therefore to be welcomed that Ofgem has commissioned work on DNO productivity improvements. However, there are several possible definitions of productivity and which one is appropriate depends in part on the use to which it will be put.

CEPA makes a number of estimates of total factor productivity (TFP) growth in sectors it considers to be comparable to DNOs from which it derives a range for future DNO productivity growth that it then compares with a figure for UK general productivity growth. It does so by examining an international data set compiled by the National Institute for Social and Economic Research (NIESR), regulated accounts of companies in the UK and in Norway and the US, and its own surveys of analysts and companies.

It also discusses partial factor productivity growth, sometimes labour productivity and sometimes operating cost efficiency.

While we have some concerns about details of the analysis, the main problem in CEPA’s report concerns the judgmental leap between the evidence presented and the conclusion. CEPA’s central TFP growth estimate is simply the mid-point of an upper and lower bound. The former is derived entirely from performance by companies in an immediately post-privatisation period that cannot be expected to be sustained over the longer term. The latter seems almost arbitrarily selected from a range of figures that would appear to imply a significantly lower number. It follows that CEPA’s central estimate is not reliable.

A number of features of the methods used may affect the results.

• CEPA appears to have used the CCA modern equivalent asset method to measure capital assets employed, which will introduce revaluation effects. We prefer an inventory approach rather than use of figures taken directly from current cost accounts.
• Adjustments are not made for changes in cost allocations that have occurred, although these will have resulted in large overstatements of DNO productivity growth – particularly in 2000/01 when 14% of operating costs were reallocated away from DNOs.
• Outputs are measured in different ways when compiling productivity measures that are then compared with each other.
• When including quality as an output CEPA incorrectly calculates its weight by multiplying the assumed value by the number of lost units. However, the quality service that is delivered is not the number of units that is lost but the number that is not lost.

CEPA’s report concentrates on total factor productivity. It rightly draws few conclusions as regards partial factor productivity or operating cost efficiency. The figures derived from the NIESR data set are all of labour productivity, which is not directly comparable with measured operating cost efficiency but would be expected to be higher. The figures for operating costs are either for the post-privatisation experience of the DNOs themselves, NGC and the water companies or for Norwegian and US electricity distribution companies. Experience of these two last is of modest operating cost efficiency growth of 1.6% and 0.5% respectively. However, there is no similar whole economy measurement with which these can be compared in order to calculate the extent to which the sector outperforms the RPI.

Our assessment of TFP, based on the evidence presented, differs from that reached by CEPA. It is that the forecast of the potential for trend DNO TFP growth is subject to significant error but there are no strong grounds to expect it to be much higher than that for the economy as a whole.

• DNO past performance is subject to substantial measurement error and is influenced by privatisation effects and so is not a good guide to future performance.
• Similar considerations apply to the evidence from other UK utilities.
• Evidence from US and Norwegian distribution companies has unexpected features but is consistent with a growth similar to the UK economy average.
• US, French and German national accounts data suggest TFP growth in the utility sector slightly lower than that for the UK economy and around that for their economies as a whole.
• CEPA’s survey of analysts is small but the results are consistent with TFP growth similar to that for the economy as a whole.
• TFP growth in the UK utility sector was also 0.7% higher than that for GDP as a whole in the 1950-90 period but this is on the basis of a utility output definition that grows faster than that used by Ofgem.
• The composite sectoral TFP estimate (excluding post-privatisation utilities data) gives a TFP growth rate slightly higher than that for the economy as a whole.

2. Differences in productivity measures and their use in price control

Previous price controls have assessed the scope for cost reduction during the coming price control period primarily by attempting to assess the extent to which companies’ productivity is below an efficient level. The extent to which the efficient level might itself move in the future has been a secondary consideration to which comparatively little attention has been paid.

If, as seems likely, it is not possible to reject the hypothesis that companies are efficient and there is no scope for catch-up cost saving, the potential for efficiency growth will be that of the industry as a whole. Cost savings would be possible, relative to an economy-wide price index like the RPI, if industry productivity growth exceeded that for the economy or if the relative price of the industry’s inputs falls. Price controls should therefore be set on the basis that an individual company’s future costs will only differ from present ones in real terms insofar as total industry productivity growth (or factor price change) is likely to differ from that in the economy as a whole.

Further work on comparative efficiency has tended to confirm that it is difficult to reach reliable conclusions on the comparative efficiency of DNOs. It is therefore to be welcomed that Ofgem has commissioned work from CEPA on DNO productivity improvements. However, there are several possible definitions of productivity and which one is appropriate depends in part on the use to which it will be put.

Previous DNO price control reviews have used a "building block" approach that has necessitated a disaggregated view of productivity.

• A judgement has been taken of the scope for improvement in an operating cost productivity measure that is the ratio of controllable operating costs deflated by the retail price index (RPI) to a weighted sum of three outputs – customer numbers, units distributed and line length. The question has been whether the particular circumstances of the electricity industry are likely to result in greater gains in efficiency than those in the economy as a whole that are embodied in the RPI. There has been debate as to whether this partial measure will be affected by changes in the capital output ratio or by unusual movements in relative factor prices but no specific adjustments have been made for those reasons.
• Capital cost productivity has also been assumed to rise relative to that in the economy as a whole in that the unit price of installed pieces of capital equipment has sometimes been assumed to fall in real terms.
• In the last (1999) review financing costs were assumed to fall and the cost of capital was reduced.
• These efficiency judgements have been used merely to forecast costs and an X factor has been calculated to equate net present values of expected costs and revenues. There has been no systematic attempt to derive X from productivity trends.

The productivity measures assessed in CEPA’s report include more general measures whose application would be different. It discusses "partial factor productivity", relating to operating costs, but its major conclusions concern growth in a measure of total factor productivity and the difference between it and that for the economy as a whole, which "over the longer term …. it can be appropriate for the X factor to approach".

Most productivity analysis attempts to relate physical outputs to a weighted sum of physical inputs such as labour or materials or the services of items of capital equipment. Given a view of likely productivity growth of this sort a forecast of cost movements relative to the RPI can be obtained, but only with additional assumptions about productivity growth in the economy as a whole and likely movements in the prices of the inputs relative to the RPI. Given an appropriate starting price level, X could be set equal to forecast industry productivity growth less that for the economy as a whole less relative input price increases.

CEPA’s work on DNOs relates output in physical terms to the money value of costs deflated by the RPI and therefore combines physical productivity changes with the impact of relative price movements. The resulting productivity estimates are compared with similarly derived measures for other utilities in the UK and overseas and with national accounts data for the US, France and Germany. The method implicit in the national accounts data differs in that expenditures on inputs are deflated separately, not by a general measure such as the RPI, and the productivity estimate attempts to derive only the physical measure.

Bearing these conceptual differences in mind we now proceed to discuss:

• CEPA’s method of arriving at its conclusions (section 3);
• The theoretical problems involved in making the estimates (section 4);
• The estimates themselves (section 5);
• Partial factor productivity (section 6); and
• Our conclusions (section 7).

2. CEPA’s method of drawing conclusions from the evidence

CEPA attempts to forecast future DNO productivity growth by considering:

• Past DNO productivity growth (calculated from regulated accounts);
• Past productivity growth in other UK utilities (calculated from regulated accounts);
• Past productivity growth in electricity distribution in other countries (calculated from regulated accounts);
• Past productivity growth in comparable sectors of the economy and in gas, electricity and water in the United States, France and Germany (calculated from national accounts);
• Forecasts by analysts and comparable companies.

This produces a number of estimates from which CEPA derives a range for future DNO total factor productivity growth that it then compares with a figure for UK general productivity growth derived from the national accounts.

While we have some concerns about details of the analysis (see sections 4 and 5), the main problem in CEPA’s report concerns the judgmental leap between the evidence presented and the conclusion.

Having obtained a set of TFP estimates CEPA then says:

In other words, an upper rate of 3.4% is set by UK utility performance, a lower bound of 1.4% by performance in Germany and splitting the difference gives the central estimate.

However, this reasoning is seriously flawed:

• The upper bound consists almost entirely of the post-privatisation performance of the DNOs themselves and of other privatised industries, which of course involves much higher productivity growth than any likely sustained trend rate. The two figures given are the DNOs’ own 1991-2001 performance and the national accounts figure for "UK utilities" 1990-99. The former consists of the post-privatisation period, is likely also to be affected by data errors and includes the probably anomalous year 2000-01 (see section 4.2). The latter is actually the gas, electricity and water sectors, which include not only the DNOs themselves, but also the recently privatised generators, NGC, Scottish electricity companies, English and Welsh water companies and British Gas.
• The lower bound of 1.4% is said to be the figure for gas, electricity and water in Germany for the last ten years. Figure 33 on page 47 of CEPA’s report actually gives 1.2% as the volume-adjusted rate. The similar figures for France and the United States are 1.5% and 0.2%. There is a separately calculated figure for US electricity distribution of 2.2% and one for Norway of 0.2%. The seven analysts surveyed produced a range of estimates from minus 0.2% to 2% with a median of 1.5%. Undescribed forward-looking analysis of twelve companies was said to produce a figure of 2.3%. These figures represent practically all the quoted recent results other than those for companies immediately after privatisation. It is, to put it mildly, unclear why CEPA adopted a lower bound of 1.4% from studying the numbers. The simple average of the seven estimates is 1.3%, and so another commentator might pick CEPA’s lower band as a central estimate.

It follows that CEPA’s central estimate, which is the mid-point of an upper bound set by post-privatisation performance and a seemingly arbitrarily selected lower bound, is not reliable. The figures cited would seem to imply a significantly lower number.

2. Methodological issues

The estimation of total factor productivity growth is not a simple matter. Both outputs and inputs must be measured and weighted together. The movement in the ratio can then be calculated but some of that may be due to other factors, notably economies of scale resulting from increases in output.

In this section we discuss problems relating to:

• The measurement of capital factor inputs;
• Operating cost uncertainty;
• Weighting;
• Unmarketed outputs such as quality; and
• The effect of output growth.

1. Capital inputs

As CEPA explains, neither historic cost capital asset values nor the regulatory asset base is likely to provide a suitable estimate of the asset base, whose return and depreciation should be included as a capital factor input. CEPA says it uses current cost values but the method used is not described in detail.

We prefer an inventory approach rather than use of figures taken directly from current cost accounts. Current cost values taken from regulatory accounts will include the effect of revaluations (other than those for RPI changes) that may suggest changes in capital input that did not in fact take place.

Even using an inventory method, there are several possible approaches. For example, in other work, we have constructed an estimate of DNO network assets by calculating asset value and depreciation estimates from the reported CCA network values at vesting and network investment since then. Constant price values are derived by indexing using the RPI. Post-vesting assets are assumed to depreciate 2.5% a year for forty years. Pre-vesting assets are assumed to have an age profile ranging from 0 to 40 years that can be represented by a straight line that is tilted so as to produce an average age equal to that reported at vesting. Depreciation in each subsequent year can be calculated from that assumed age profile.

We do not know precisely what method CEPA has used but it appears to have been the CCA modern equivalent asset method, which will introduce revaluation effects.

2. Operating cost uncertainty

Operating costs are normally much more easily measurable but, in the case of the DNOs, there are still problems. Throughout the 1990s there have been changes in the methods of transfer pricing and cost allocation between businesses. The trend has been to move the allocation of costs away from the distribution business, which will tend to overstate productivity growth. Indeed the accounting changes following the 1999 distribution price control review are likely to have been a significant cause of the large apparent cost reduction between 1999-00 and 2000-01 noted by CEPA.

The transfer of costs to supply in the 1999 DNO review (net of the increased capitalisation also assumed) represented a reduction in standard controllable costs of more than 14%. This followed earlier transfers of costs such as meter reading, advertising, corporate and IT.

The situation is exacerbated because there is reason to believe that DNO costs in the early 1990s were overstated. Domah and Pollitt "note that there was a rise in real unit distribution and supply controllable costs by about 15 per cent immediately after privatisation in 1990. The cost remained at a high level until 1994–95, after which there was a dramatic fall. The rise in controllable cost of £358 million (in nominal terms) between 1989–90 and 1990–91 represents a 21 per cent increase in nominal terms or a 7 per cent rise in real controllable costs per unit distributed." In other words, given an expectation of continuing productivity growth, total operating costs in the regional electricity companies immediately after privatisation were about 10% higher than might have been expected. The reason for this has not been established clearly but it is likely that there was an element of provisioning and other means of bringing costs forward to the 1990-94 period when the price control regime was relatively generous and reported profits would otherwise have been higher.

Relative to the present level of operating costs, DNO costs at the start of the period contained substantial elements that are now allocated elsewhere and were probably also temporarily increased in the immediate post-privatisation period. Therefore the fall in costs and the estimate of productivity growth is overstated.

3. Weighting and money values

The contrast between CEPA’s approach of deflating money values by the RPI and that in the national accounts data, where inputs and outputs are separately deflated, was discussed above in section 2.

However, CEPA does not treat all costs as a single monetary unit but deflates capital and operating costs separately and weights them together. It does so using weights from the average over the period rather than using the weights for each individual year. The difference this makes has not been reported.

4. Quality and other outputs

The DNO output term is weighted together using the weights Ofgem used in its operating cost equations, but omitting line length because of data problems. Whether the same weighting is appropriate when considering total costs is debatable. The same or similar methods appear also to have been used by CEPA for US and Norwegian electricity distribution and for BT but this differs from what has been used in the national accounts utilities comparisons and for the water companies and NGC, all of which use units delivered. We have not checked the figures but units delivered tend to rise more rapidly than customer numbers and so the difference in method would tend to increase the figures derived for those comparators.

The addition of quality as an output, which has a significant impact on some of the results, raises considerable problems. Some of these relate to its measurement but the greatest is the decision of what weight to assign to it. In water a cost based approach is used (and is discussed in 5.3 below) but in other industries, including electricity distribution, the weight is derived from customer valuation – a price times a volume. While difficult to do, the value of a marginal increase in quality can be estimated and a price derived. The real problem lies in assessing the volume of the service that is delivered. CEPA incorrectly multiplies the assumed value of a lost kWh (£2.80) by the number of lost kWh (14 million) to derive a weight of about £40 million or 2%. However, this is the wrong weight as can seen from the fact that, if more effort is put into quality and fewer units are lost, the weight would be reduced. The quality service that is delivered is not the number of units that is lost but the number that is not lost. Use of that figure (i.e. all units successfully delivered) would produce an absurdly high weight for quality but the method that produces the 2% weight is flawed.

An alternative method would be to exclude quality-related costs, thereby removing the need to adjust output. However, this would be difficult to do, particularly as regards capital assets.

5. Scale

As CEPA explains, if there are economies of scale, some productivity improvement can occur purely as a result of output increase. Depending on the means of conducting the price control calculations and, in particular, on whether the price control formula assumes scale effects, the measure of productivity growth that should be used may be that after removing that part which is due to scale.

The adjustment is to subtract (1-η)/η times the change in output, where η is the elasticity of costs with respect to scale. CEPA reports a regression estimate of 0.7 for the scale elasticity in distribution in a log equation and Ofgem’s assumptions of £25 million fixed operating cost and linear form imply DNO controllable operating cost scale elasticities ranging from about 0.35 to 0.6 depending on the size of the company. The figure actually used in the adjustment is 0.85. This is justified mainly on the basis of findings in other countries. We do not wish to contest the figure but merely point out the degree of uncertainty and the possible inconsistency with what Ofgem assumes elsewhere.

4. The evidence

Given the methodological problems it is not surprising that results are uncertain and varied. CEPA presents and discusses them intelligently but then proceeds to draw unwarranted conclusions.

1. The UK economy

TFP growth in the UK economy is estimated at around 1.4% pa, 1.3% after a small adjustment for economies of scale. Although there is bound to be some uncertainty over the figure, it appears to be soundly based and stable. It lies between the (unadjusted for scale) 1974-99 TFP trend growth figures for the United States and France of 1% and 1.5% respectively.

However, all these TFP growth estimates are calculated on a different basis from the accounting analysis in sections 5.2, 5.3, and 5.4.1 below.

2. DNOs

CEPA derives a TFP growth of 4.2% but this is of little help in making forward projections. Firstly, performance in the period has been affected by post-privatisation improvements for which there will be much less scope in the future. Secondly the estimate is uncertain since it depends on assumptions on the method of compiling the capital stock, the size of the economies of scale effect, the treatment of quality and the consistency of the regulatory accounting data over time. Significantly different views can be taken on all these factors. CEPA reports the impact of removing the single year 2000-01 from the calculation as reducing the average annual growth figure from 4.2% to 3.2%.

3. Other UK utilities (accounting data)

Three other groups of UK utilities are investigated – NGC, the water companies, BT and Railtrack. These too are of little help in making forward projections for the same reasons.

Firstly, although the immediate post-privatisation periods are excluded for the water companies and BT, the results will still be affected by privatisation effects, strongly so in the case of NGC and, probably, Railtrack.

Secondly, the estimates are uncertain.

• The output of the first two is assessed on the basis of units delivered rather than the composite (⅔ customers, ⅓ units) used for DNOs.
• The results for water companies depend entirely on the treatment of quality. When this is given a 42% weight (based on costs) TFP growth from 1994-95 to 20001-02 is 7.7% pa, with a 2% weight (based on customer valuation) the answer is 0.1% and, if quality is excluded, TFP growth is negative. Both the cost and customer value weights appear to have been calculated incorrectly. The cost weight is a share of capital expenditure, not total costs or even total assets. The customer value weight appears to result from a similarly flawed calculation as that for DNOs discussed in 4.4 above.

In any case, the results are so diverse as to be of little use in making projections for DNOs even if they were good indicators of future productivity growth in those companies. The water company estimate can be minus 0.3% or 7.7%, depending on the treatment of quality, NGC is said to be 2.4%, Railtrack, based only on four years (1997-98 to 2001-02), 2.9% and BT, based on the same four years, 13.2%.

1. International comparators

International comparisons are made for electricity distribution from regulated accounts for the United States and Norway and for gas/electricity/water ("utilities") from national accounts data for the United States, France and Germany.

0. Electricity (accounting data)

While much accounting data for Norway and the US was available from the regulatory authorities CEPA has had to construct estimates of CCA assets from HCA data. We have not examined the data but wonder if either the method or special features of the time period analysed may in some way have affected the results. While opex productivity is calculated at 0.5% in the US and 1.6% in Norway, the results for capital assets are very different (4% and minus 1.1% respectively) with the result that the US TFP estimate is 2.2% and that for Norway 0.2%

1. Gas, electricity and water (national accounts data)

The National Institute of Economic and Social Research (NIESR) has compiled measures of productivity growth by sector and for the economy as a whole for the UK and for a number of other countries. These NIESR estimates have been carefully prepared but require the use of assumptions that may not be correct. The results are therefore uncertain. For example, the 1974 oil price rise may have made a part of the capital stock redundant in some countries so the increase in inputs (from a lower base than assumed) may be understated and TFP growth exaggerated.

In the 1990-99 period calculated TFP growth in the gas/electricity/water sector was reported by CEPA as 0.2% in the US, 1.5% in France and 1.2% in Germany. The electricity sector in the US recorded higher TFP growth (1.7%) than gas and water but the US utility sector had low growth in the longer 1974-99 period when even electricity had only 0.5% pa TFP growth. France’s utility TFP growth in 1974-99 is estimated at 2.1%.

Over an even longer 1950-99 period US utility TFP growth is calculated at 1.4% (electricity 1.7%). TFP growth in France was very rapid in the 1950s and 1960s, when utility output was growing at almost 10% per annum, producing a reported 4.4% for the period as a whole.

The simple average of utility TFP growth estimates for the US, Germany and France 1990-99 was 1% and for US and France in the longer period 1974-99 0.9%. If the US electricity sector is used, rather than the utility sector, the figures are raised to 1.5% and 1.3% respectively.

2. UK sectoral estimates

Similar figures for the UK gas/electricity/water sector show 3.4% for 1990-99, 2.8% for 1974-99 and 2.5% for 1950-99. This implies growth of 3.4% in the post-privatisation 1990-99 period, 2.5% in 1974-90 and 2.2% 1950-74. CEPA also quote their own rather different trend estimates (3.2%, 2.0%, 1.2%) which imply 3.2% in the recent period, 1.3% 1974-90 and 0.4% 1950-74. However, we cannot reproduce these results.

CEPA also reports a TFP growth estimate for DNOs by disaggregating their functions, comparing them with other sectors in the economy (18% construction, 36% engineering, 28% utilities, 9% business services and 9% communications) and calculating the weighted sum of 1990-99 TFP growth in those sectors.

Like CEPA, we do not find this approach particularly convincing.

• The method is suspect because it implicitly assumes distribution is more intensive in engineering, communications, business services and construction than are utilities in general but the method used to derive the weights did not attempt to test whether that was the case but merely assigned activities to those categories.
• The weighted sum calculated is 2.0% but this depends in part on the (circular) use of the high post-privatisation utility TFP growth. If the 2.5% figure is substituted the estimate falls to 1.8% and use of the 1.3% CEPA 1974-90 trend produces an estimate of 1.5%.

1. Further analysis of the NIESR database

Since we did not fully understand the derivation of the figures CEPA calculated from the NIESR database, in particular the "trend" figures quoted in figure 35, we calculated TFP growth estimates for gas/electricity/water and the whole economy for the UK, US, France and Germany for 1950-99 and a number of sub-periods using the NIESR data set.

We estimated trend growth rates for the periods 1950-74, 1974-90, and 1990-99 (and for the longer periods 1974-99 and 1950-99) by regressing the log of the NIESR estimate of TFP against a constant and a time trend. This produced the trends given in table 5.6.1.



  

We then adjusted them for economies of scale using elasticities of 0.85 for electricity, gas and water and 0.95 for GDP, which resulted in the figures shown in table 5.6.2.

 



 

 Table 5.6.3 shows the differences between TFP growth in the utilities sectors and that for the economy as a whole.



 

Thus it appears that gas/electricity/water TFP growth in the UK was about 0.7% higher than that of GDP in 1974-90 and around 2% higher in the post-privatisation period. However, this does not seem to have been the case in the United States, France and Germany where, apart from in France in the 1950s and 1960s when output increased eight-fold, experience has varied somewhat but TFP growth seems to have been around that for the economy as a whole.

It should be remembered that all these measures use a definition of utility output that is likely to have grown significantly faster than that used for distribution by Ofgem and so a likely downward adjustment to productivity growth is required.

2. Surveys

CEPA reports the results of their surveys of analysts and companies tersely.

Only seven analysts replied to the survey and their answers are widely dispersed. The highest TFP estimate was 2%, the lowest minus 0.3% and the median 1.5%. Given the apparent skewedness of the distribution it may be a reasonable guess that the mean was lower than the median. The analysts do not appear to have been asked for TFP projections for the UK economy as a whole.

Twelve company projections were considered. CEPA do not describe the method used to convert the data but they calculate that three chemicals companies expected 3.1% TFP growth, three oil companies 1.1%, two metals companies 2.8% and four engineering companies 2.1%. It is not clear why these companies where chosen or why they should be expected to perform differently from the economy as a whole. It would appear that the resemblance to DNOs is one of capital intensity but the figures cited throughout the report do not suggest that capital efficiency growth is normally high.

4. Partial factor productivity

CEPA presents some figures for partial factor productivity. These sometimes consider operating costs deflated by a general price index as an input and sometimes give figures for labour productivity. The two are not directly comparable. The latter is normally higher than the former since real wages tend to increase and so labour input deflated by wages will rise less rapidly than the same money expenditure deflated by a general price index.

The figures derived from the NIESR data set are all of labour productivity. The figures for operating costs are either for the post-privatisation experience of the DNOs themselves, NGC and the water companies or for Norwegian and US electricity distribution companies. Experience of these two last is of modest efficiency growth of 1.6% and 0.5% respectively. However, there is no similar whole economy measurement with which these can be compared.

Operating cost efficiency growth will differ from that for total factor productivity because of the impact of changes in the proportion of capital relative to other inputs (capital substitution). The rate of growth of operating cost productivity will be equal to that for TFP plus the difference between the rate of growth of capital and other inputs times the elasticity of output with respect to capital. Thus the likely rate of growth of operating cost efficiency can in principle be derived from likely TFP growth given a forecast of the increase in capital assets (other than that for other purposes such as quality improvement) and an estimate of the elasticity. However, this is not simple and, unsurprisingly, is not addressed in CEPA’s paper.

CEPA report DNO capital stock as increasing around 6.5% per annum relative to opex in 1991/92-2001/02. This sort of change in relative factor inputs is not expected to continue and so the difference between DNO TFP and opex efficiency growth is likely to be less marked.

5. Conclusions

CEPA has produced an interesting report but it appears to be one of work in progress. There appear to be inconsistencies in some of the numbers, the methods used are not fully described and the conclusions drawn are not justified.

Our assessment, based on the evidence presented, is that the forecast of the potential for trend DNO productivity growth is subject to significant error but there are no strong grounds to expect it to be much higher than that for the economy as a whole, which has been around 1.3%.

• DNO past performance is subject to substantial measurement error and, since it is influenced by privatisation effects, would not (even if correctly measured) be a good guide to future performance.
• The evidence from other UK utilities is also not a good guide to future DNO performance. It is strongly affected by the assumptions used and/or derived from limited data periods. The results are diverse and affected by privatisation.
• Evidence from US and Norwegian distribution companies has unexpected features but is consistent with a growth similar to the UK average. The simple average is 1.2%.
• NIESR estimates derived from US, French and German national accounts data suggest TFP growth in the utility sector slightly lower than that for the UK economy. The simple average growth 1990-99 was 1%. Apart from in the period of rapid growth in France 1950-70, utility TFP growth in those countries has been around that for their economies as a whole.
• TFP growth in the UK utility sector, apart from having been stronger in the 1990-99 post privatisation period, was also 0.7% higher than that for GDP as a whole in the earlier 1950-90 period but this is on the basis of a utility output definition that grows faster than that used by Ofgem.
• CEPA’s composite sectoral TFP estimate excluding post-privatisation utilities data gives a TFP growth rate (1.5-1.8%) slightly higher than that for the economy as a whole (1.4%).
• CEPA’s survey of analysts is small but the median TFP figure (1.5%) and the downward skewed distribution are consistent with TFP growth similar to that for the economy as a whole.
• The evidence from company accounts produces a higher figure (2.3%) but - given the lack of definition of the concepts used, explanation of method applied, justification for the selection of the twelve particular companies or explanation why these might be expected to differ from the economy as a whole – we do not consider that much weight can be placed on it.