ABS Research Methodology

PRIMARY DATA COLLECTION

ABS collects primary quantitative and qualitative data with a range of interview techniques, whichever is appropriate for the study. For quantitative data we use personal face-to-face, telephone, postal mail, e-mail interviews. For qualitative data we use in-depth interviews and focus groups.

ABS uses a number of carefully selected field agencies. We select the agency most suited to the study in question and can thus offer the best facility available.

Excellence of data collection is a priority for ABS and fieldwork is under the continuous supervision of our senior research team.

TECHNICAL NOTES ON FORECAST METHODOLOGY
GENERATION, TRANSMISSION AND DISTRIBUTION, EQUIPMENT MARKETS

ABS has produced a number of different forecasts of different components of the electricity supply industry and its equipment markets. These were carried out in sequence, starting with generating capacity, transmission and distribution capacity, growth and product demand.

The first forecast was for generating capacity. The model developed for that was reapplied to forecasting transmission and distribution markets, with various refinements.

GENERATING CAPACITY FORECASTS
(ABS Power Predictor 1 and 2)

A number of determinants affect the rate of growth of electrical generating capacity, the most commonly quoted being population growth, urbanisation, economic development, natural resources and environmental factors. One forecasting method is to produce a weight for each factor and to apply the weight to current capacity levels.

Analysis of the data indicates that the most significant single determinant globally is the level of economic development. Some of the other variables can work either way. Increased population and urbanisation in very low income countries can put such a strain on resources that it retards development and the ability to install new capacity, rather than creating more consumers, ergo more electrical capacity. ABS believes that there has been confusion between demand and the ability to supply. Because demand rises it does not necessarily mean that it is met. Almost all undeveloped and developing countries are chronically short of electricity, many to a development-threatening level, until they reach a point at which the economy starts to change into that of a newly industrialised country. Countries like China, India and Pakistan habitually suffer 20-40% power shortages, which are in many cases cited as the biggest single obstacle to economic development. Such countries then usually have problems on the other face of the coin; when installed capacity is raised to meet development needs too much electricity is generated and consumed without controls and the environment is damaged. A different set of variables then starts to operate. Consequently the future growth trend is different for countries at different stages of development. The ABS approach to forecasting electrical generating capacity involves two procedures, approaching the forecasts from two unconnected viewpoints and comparing the results.

1. THEORETICAL STAGE - REGRESSION ANALYSIS

The installed capacity of every country has been tabulated at five year intervals from 1970 to 1995. World totals and capacity for the major countries are available at five year intervals since 1950. Regression analysis was carried out, based on historical capacity levels from 1970 to 1998/99 in four stages; globally and disaggregated at three levels and forecasts were produced to 2020. The following chart shows one of 33 regression analyses carried out, for upper-middle income countries, compared with the final forecasts. The final forecasts, shown in the columns are higher than the statistical historical projection because of the future move of several lower-middle income countries into the upper-middle segment, which the historical data does not account for.

Overall - based on global capacity levels from 1970 to 1995 a global forecast was made to 2020. The effect of evening out is so strong that the world demonstrates almost linear growth. It is at the succeeding stages of this analysis that the regressions provided the most useful trends, when different segments were fitted by different functions, with different gradients.

Disaggregated 1 - countries were classified into four groups within the four World Bank national income categories - high income, upper-middle income, lower-middle income, low income. Regression analysis was carried out on these four groups. This analysis showed clearly that at a certain point of economic development electrical growth decelerates, indicating that the growth of electrical capacity is finite.

Disaggregated 2 - countries were classified by ten geographical regions - Western Europe, Eastern Europe, FSU (Former Soviet Union), Africa, Middle East, Asia, Pacific, North America, Central America, South America. Regression analysis was carried out for 33 sub-sets of groups of countries within economic category.

Disaggregated 4 - in some of the sub-sets of economic group within geographical region, important countries were isolated and separate regression analysis was carried out for these, to explain the effects of broader trends. This was mainly in Asia.

There are inherent variations between countries, every one of which is in a different position on a continuum of electrical development, ranging from the maturity of the industrialised countries when no significant growth will occur or the installed capacity base may even contract, to under-capacity in countries which have just started to develop and which can only go up. Countries and groups of countries therefore have different regression equations; logarithmic, quadratic, power and in one case linear.

2. INDIVIDUAL COUNTRY ASSESSMENT

The limitation of theoretically generated forecasts is that they are limited to specified and quantifiable conditions. They take no account of unforeseen events and lack judgmental sensitivity. An obvious recent example was the Asian financial crisis which erupted in 1998 and which has caused business analysts to downgrade many forecasts. This has been acute because prior to the Asian crisis, forecasts were being based on extremely confident assumptions of growth, disregarding the underlying realities of the Asian countries, many of which have serious structural flaws. Yet for over three years there had been warnings of weaknesses in the Asian economies from a small number of observers.

In ten years of producing reports about the energy sector in every country, ABS has amassed a considerable amount of information about most countries and basic information for the remaining few. We have accumulated knowledge and personal experience of many of them. It was decided to use this knowledge to approach the forecasting from two, opposite directions. The first was generated theoretically, from a global base downwards, as shown above. The second was a laborious process, independently to scrutinise every country in the world and to assess judgementally its future potential and what it can achieve. The individual forecasts were accumulated into the same groups and sub-sets as those used in the regression analysis and the results compared. There was close agreement in most groups, excepting the low income countries. This enabled us to examine areas of difference critically and to review individual forecasts. The particular circumstances affecting forecast results are discussed in the text of the Power Predictor report. The final outcome are the individual country forecasts.

As well as producing forecasts of generating capacity, ABS has compiled a database of generating projects now actively under construction. 1,320 projects are listed and there are further projects of which we have notice but not definite confirmation. A final check on the forecasts was available from this project database, although this was not used to generate the forecasts. This was particularly useful for small countries, where one large project has distorted the general rends. For example, what is the explanation for a country with installed capacity of 100 MW and little apparent need for more, suddenly building a 500 MW power station? In such cases we have examined the situation to determine the cause. The project record may be wrong, the forecast may be too low, the electricity may be for export, or major industrial expansion may be in the pipeline. Whatever the reason, we have attempted to establish the best explanation which reconciles conflicting evidence. Purely theoretical forecasting would fail to do this.

CORRELATION BETWEEN FORECASTS AND THE PROJECT DATABASE

The project database has been compiled empirically from as many sources of information as available. They are independent sets of information, based on separate sources and constructed differently from the forecasts. It is therefore an excellent cross-check to compare them.

The total project capacity for completion in the five-year period 1999 to 2003 is 493 GW. It averages 109 GW a year for the first four years but tails off to 56 GW in 2003. The reason for this is that reports on project completion's five years ahead are bound to be lower than reports for the more immediate future. The net effect of this is that the figure of 493 GW from 1999 to 2003 is about 10% low for the total period and an average of 109 GW represents a truer figure.

The independently estimated forecasts are 486 GW for the five years for 1996-2000 and 578 GW between 2001 and 2005, averaging 106 GW a year from 1996 to 2005.

The comparison of 109 GW from the project database and 106 GW from the theoretical estimates are within 3.3% of each other, representing a correlation of 96.7% from the two independent sources. This is a substantial improvement on the already excellent correlation of 91.7% obtained in Power Predictor I.

CALCULATIONS FOR REPLACEMENT MARKETS

Market estimates have been made for replacement of obsolete equipment, which is not contained within the forecasts of installed capacity growth. These are particularly important in the mature markets, where installed capacity may even decline as more efficient plant is built and changes in capacity will seriously undervalue the size of the market growth. The estimate for replacement is based on a plant life of 30 years. It is difficult to arrive at a global average for the lifespan of plant. In well maintained utilities plant is monitored and written-off according to an assigned life, which in many cases is 30 years. In developing countries maintenance may be so poor that plant deteriorates immediately after installation. There have been reports of cases where plant is unusable after five years. Paradoxically, in many of these countries money is not available to replace such equipment and it has to be repaired repeatedly, despite the high cost of doing this. Lifetimes can then become extended way beyond practical limits.

The replacement market in 2000 is calculated as 0.0333 6 installed capacity in 1970. The lifespan has a huge effect on the replacement market size when the installed base is rising. The period 1970-1975 saw an increase of 453 GW in world capacity. By decreasing life from 30 to 25 years, the proportion to be renewed in 2000 would be increased from 3.333% to 4.0% and would be based on an world installed base of 1,624 GW in 1975 instead of 3.33% of 1,171 GW in 1970, giving an estimate of 64,960 MW compared with 39,000 MW. This disparity evens out after systems reach a mature plateau.

METHODOLOGY OF THE TRANSMISSION & DISTRIBUTION FORECASTS
(The T&D Report 2000)

The methodology of the market size estimates and forecasts of development of the transmission and distribution systems consisted of four stages, combining a top-down and a bottom-up approach. This methodology was found to be highly effective in creating the forecasts of electrical generating capacity in ABS Power Predictor.

Stage 1

All available data about electrical installed capacity, transmission and distribution that we could locate was entered into a master database. Basic information about electricity generation and consumption is included in the database for every country. Original information about transmission and distribution is included in the database for 116 countries, varying in extent. In some cases it is extensive and over a period of time, in other cases it consists of circuit lengths of transmission lines at one point in time. The database contains information on the following areas; installed generating capacity (MW), HV transmission line lengths by voltage (kV), MV and LV distribution line lengths by voltage (kV and V), transmission substation capacity (MVA), where possible for transmission and distribution separately, capital expenditure on transmission and distribution. The ideal is to have all of this data for each country but this does not exist. The database contains comprehensive entries containing information in all these sub-sectors for 32 countries.

Information has been obtained from utility company reports, national industry association reports, national government statistics, international agencies. As a guide to the extensive coverage of the database, the information derived from this exercise is summarised in the review of all countries contained in Section 2 of this report, although this section does not contain all of the information in the database. With the exception of low voltage line lengths in the United States, every figure in this section has been extracted from one of the above sources. The National Electrical Reliability Council of the United States (NERC) publishes the most comprehensive information and analysis of the country's electricity industry provided by any country, with one exception; data about distribution lines below 4 kV is not collected and cannot be obtained from the regional offices of NERC. ABS conducted a survey of 191 out of the 3,199 electricity utilities in the US and projected an estimate of LV line lengths from the sample data. The sample was selected with a stratified probability design. Generally there is less information available about low voltage distribution than about transmission.

The data available to the analyst is strangely variable. Some very small national utilities have published exemplary reports for many years, with all the information required. There are several in Africa, the Middle East and Asia. These countries are extremely valuable in providing benchmarks for similar economies where only the most elementary electrical statistics exist. Most of the industrialised countries publish excellent and comprehensive information but by no means all. Extremely detailed information is now becoming available about the larger former Comecon countries but is limited for some of the smaller ones. Many countries, especially those with a large variety of utilities in both public and private ownership, provide little more than a national overview of the most basic nature.

The data came in so many forms and from so many sources that many conversions were required, the most common being from route lengths to circuit lengths and from miles to kilometres. The sources were not all for the same year, so care was taken to ensure that data was recorded in the right year, where necessary being projected forwards or backwards to a base year for forecasting. The base year used was 1995. In many cases we have information for 1998 and even in a few cases for 1999. This is a valuable check on the forecasts based on 1995.

Stage 2

A set of models was created;

Installed generating capacity, MW

Transmission line lengths, circuit km

Distribution line lengths, circuit km

Transformer capacity, MVA

Capital expenditure on transmission and distribution, US$

These models take 1995 as the base line for every country and forecasts were made to 2010. Where past data is available it has been entered and where not available reverse regressions have been made. Actual capital expenditure is entered for a total of 67.7% of the final projected world estimate in 1995. The global market was modelled on different criteria; installed generating capacity, transmission line lengths and transformer capacity and a series of projections made. The variation between projections of total market size arising from the different criteria was between 95% and 106%. Time series projections have been made on the basis of a number of variables.

The market can be viewed in two basic elements; new installations and replacement of existing installations. New installations have been forecast on the basis of capacity growth and economic development, taking into account such factors as rural electrification in developing countries. There are different reasons for replacement.

Stage 3

The final stage of analysis was a bottom-up examination of every estimate in the spreadsheets. Each figure was scrutinised in the light of market experience and carefully evaluated. Particular attention has been paid to evaluation of the actual achievements of governments against published development plans and targets. This is an area requiring considerable experience of the markets because the outcomes are not always obvious. An example is a comparison between India and Taiwan. Some knowledge of the Indian power market will reveal very quickly that India has not achieved a single power target in the Development Plans throughout the past fifty years and it is wise to be conservative about the targets in the current Ninth Plan, which runs from 1998-2003. In comparison, Taiwan has an extraordinary record of economic development and with a very efficient power utility, it is easy to accept that targets will be met. However, Taiwan is also experiencing a growing tide towards grass roots democracy and rapidly escalating public awareness about environmental issues. This is largely because of the pace of economic development which has lead to unchecked exploitation of resources and lack of controls. As a result the development plans of Taipower have frequently been interrupted by protesters objecting to the use of land for power schemes and appealing decisions.

EQUIPMENT DEMAND FORECASTS

Interviews with executives in the power and engineering industries have been conducted to construct a set of ratios of the composition of total capital cost for five components; land purchase or way rights, design and engineering, construction, equipment purchase and finance cost. There are wide variations in the composition of these costs and comments in the relevant section of the report should be noted. Average figures should not be used without careful evaluation of each situation but as an indicator. For example, there is a land cost in constructing distribution networks for greenfield housing developments but there is no additional land cost in up-grading either underground cables or overhead lines. Important issues in the transmission and distribution sector were explored in these interviews. A set of models was constructed for product composition within the market when producing the first edition of this report in 1996. These models have been reviewed and adjustments made in 1999.

These product demand forecasts were then entered into a table with the matrix - production - exports + imports = market. Several problems exist with constructing such a matrix. The import and export figures contain some contradictions. Some countries do not supply the data. The figure that Country A may record as its imports from Country B may not be the same as the figure that Country B shows as its exports to Country A. The imports and exports of capital goods and high value equipment are seldom regular but vary according to projects in hand, especially in the smaller countries. We therefore recorded and averaged the imports and exports for the most recent three years from the export figures of the largest 51 exporters of HV electrical equipment.

Another problem is that production data is not available for all countries. Over the relevant product groups, production data is available for about one third and in one case covers nearly a half of global production. Thus for these countries and products we were able to compare two production figures, the value reported from national statistics and the estimate derived from the market forecast, using the above equation. This provided a validation of the forecasting technique and accuracy and where necessary adjustments were made.