TECHNICAL NOTES

 The tenth national accounts release and covers the years 1993 to 2005.  It is an update of national accounts estimates prepared by the National Institute of Statistics (NIS), Ministry of Planning (MOP), with the assistance of the International Monetary Fund (IMF) Multisector Statistics Advisor (MSA). It includes revised estimates for 2001 to 2004, and the preliminary estimate for 2005.

The updating of the national accounts series is done side by side with improvement of the coverage, compilation techniques and source data, while trying to maintain the consistency of the whole series. Updates of the data series used in previous estimation rounds, as well as, new data series were used in this estimation round.

Data Sources

The development of the national accounts is a continuous process.  The national accounts database is a compilation of information utilized in the estimation of national accounts.  NIS is improving and maintaining the national accounts database with assistance from the IMF MSA. The main sources of data for the current series can be classified under three broad categories:

            Administrative data produced as by-product of government agencies’ administrative function. The main administrative data utilized are:

• Agriculture production volumes and wholesale commodity prices data from various departments within the Ministry of Agriculture, Forestry and Fisheries (MAFF);

• GSP exports data from the Ministry of Commerce (MOC) until 2003 ;

• Electricity supply data from Electricite’ de Cambodge (EDC);

• Construction approval permits data compiled by the Phnom Penh Municipality, provincial offices and the Ministry of Construction, Land Management and Urban Planning (MOCLMUP);

• Ministry of Public Works and Transport (MOPW&T) transport and port authorities data;

• Ministry of Post and Telecommunications (MOPTC) post, telecommunications, mobile phone subscribers and Internet services subscribers data;

• Visitor arrivals data compiled by the Ministry of Tourism (MOT) from Ministry of Interior (MOI) immigration data;

• Government finance statistics (TOFE) and other regulatory data on casinos, gambling, and lotteries from Ministry of Economic and Finance;

• Exchange and interest rates, monetary and financial statistics and balance of payments (BOP) statistics compiled by the National Bank of Cambodia (NBC);

• Non-government organizations actual expenditure and foreign investment approvals data from the Council for the Development of Cambodia (CDC); and

• International trade data from the Customs and Excise Department (CED) of the Ministry of Economy and Finance (MEF).

            Census, survey and other data collected and compiled by NIS, with financial and technical assistance from various international agencies:

• Socio-Economic Surveys of Cambodia for 1993-1994, 1996, 1997, 1999 and 2004;

• Survey of Industrial Establishments (SIE) for 1993, 1995 and 2000;

• Demographic Survey of 1996;

• Consumer Price Indices (CPI) for Phnom Penh from 1994 onwards and Urban Cambodia CPI from 2000 onwards;

• Producer Price Indices (PPI) from 2000 to 2003;

• Quarterly Survey of Establishments (QSE) for Manufacturing Establishments and Hotels 1996-1999;

• Monthly Construction Permit Survey (MCPS) until 2003;

• Survey of Retail Trading Units (SRTU) 1996-1999;

• Labor Force Surveys in Phnom Penh, 1997-1998 and Cambodia in 2000 and 2001;

• General Population Census 1998 and revised Population Projections, Cambodia 2004 and 2005;

• Agriculture Finance Survey 2002, and Aquaculture and Abattoirs Survey 2005 and 2006;

• Informal sector surveys in 2002, 2003, 2005 and 2006 covering textile manufacturers, smuggled fuel retailers and other retailers, agriculture commodity wholesalers and other wholesalers, guesthouses, water and land transport operators, and Internet cafes; and

• Surveys of casinos in 2003 and 2005 and 2006, and studies of domestic and international tourist expenditure in 2005.

        Special studies, reports and publications, including one-shot surveys and studies undertaken jointly by other government agencies and international agencies such as:

• Food and Agriculture Organization (FAO) on agriculture margins and farm-gate prices;

• Mekong River Commission (MRC) fishery studies and surveys;

• MAFF studies and surveys on agriculture margins and farm-gate prices; and

• MOT tourist expenditure surveys conducted in 1995, 1999, 2001 and 2002; and

• IMF World Economic Outlook (WEO) commodity prices.

Data series used in this estimation round were mainly based on preliminary or final 2005 administrative data from all source areas and revised data for earlier years for some source data.

It is extremely important to maintain surveys/data collection activities, which are particularly critical for national account estimation. While there have been significant improvements in accessing better source data in the last two years, resulting in a marked improvement in the quality of the estimates, this estimation round was still adversely affected by the lack of source data for critical areas. The lack of annual establishment or enterprise survey data continues to prevent the independent estimation of intermediate costs estimates, with compilers having to continue relying on benchmark value added to output ratios. The lack of annual household survey data adversely impacts on household consumption expenditure and output estimates (especially for services industries), as well as price deflators for rural areas. The lack of annual Labor Force Survey data continues to affect GDP by factor income estimates, with the need to rely on broad industry CSES 2004 labor force and income estimates for 2004 onwards. Finally, the continuing lack of producer, trade, export and import price indices continues to adversely affect the quality of constant price estimates for all years. A regular survey collection program is essential for training respondents to provide reliable data and for developing skills in collection and critical analysis, as well as, publishing survey results as stand alone statistics.

Compilation Methods

The estimates in this release have been expressed in current and constant prices for all years. Estimates at constant prices refer to the values for current year valued at base year 2000 prices. The general methodology adopted for the estimation of GDP at constant 2000 prices is the deflation technique. Current price estimates are deflated using various price indices.  

GDP is estimated using two approaches, which are presented in two main tables in Part III. One is GDP by economic activity (production measure) (Tables 2a-e) and the other is GDP by expenditure share (Tables 3a-e). GDP levels estimated using the expenditure approach are used to countercheck the GDP levels estimated using the production approach (GDP by economic activity). A statistical discrepancy is included in the computation of GDP by expenditure share. In the course of the estimation, one of the main concerns of a national account compiler is to reduce the statistical discrepancy to a minimum.

A number of revisions have been made to the series for 2001 to 2004 using new and revised data from NIS, ministries and other organisations. With the accumulation of 13 years worth of data from various sources the opportunity has also been taken to look at trends in the time series and to make some adjustments in recent years. These revisions have improved consistency between the production and expenditure series and improved the year on year movements for some items in the series. They have also helped to reduce the statistical discrepancy to around one percent of GDP in both current and constant prices.

As in previous estimation rounds, each important sector of the economy was estimated using a worksheet specifically designed to allow for the utilization of whatever information was available on the sector. The worksheets consist of three parts, the estimation of gross value added (GVA) using the production approach, the use and supply analysis, and the estimation of GVA at constant prices.

To capture the changes in the economy through the years, various indicators have been developed. Employment, sales and average compensation indices were compiled from the results of short surveys conducted by NIS. These were used to interpolate and extrapolate results of one-shot surveys and studies. These indicators were used together with the various administrative data that were provided by the ministries and government agencies. In this estimation round an effort has been made to improve the consistency of the estimation techniques over time for each series.

Agricultural sector current price estimates are based on gross value added to output ratios, derived from irregular MAFF and NIS surveys and adjusted for significant changes in input costs (e.g. fuel prices), applied to production data sourced from various departments within MAFF and balance of payments (BOP) data for exports of agricultural products, with adjustments for under-coverage (e.g. commercial agribusinesses). For crops and livestock production, MAFF production data and NIS data on household consumption of fruit and vegetables, and commercial agribusinesses production are used in estimating output. A supply/use estimation model is used for fisheries production, based on MAFF data, population projections and water level data, adjusted for under-coverage and cross-checked with fish consumption expenditure plus exports estimates. For forestry, hunting and gathering, a combination of MAFF forestry, NBC BOP exports and MOCLMUP construction data are used for logging, household consumption expenditure on charcoal and firewood data; and NIS data on hunting and gathering are used.

In estimating current and constant price value added, adjusted farm gate prices and price indices are calculated from retail and wholesale prices collected by MAFF, the NIS Phnom Penh (1994 to 2000), Urban CPI (2000 onwards), and household final consumption expenditure implicit price deflators are used. It is assumed that farm gate prices move along parallel lines with MAFF wholesale and NIS retail prices data. For export-oriented production such as rubber and logging, export prices indices are used.

 Industry and services sectors’ current price estimates are also largely based on gross value added to output ratios derived from the 1993, 1995 and 2000 SIE results or other sources and adjusted for significant changes in input costs, are applied to output estimates for each industry. In a couple of instances value added per worker is used. 1993 benchmark ratios, adjusted over time, have continued to be used for rubber manufacturing; electricity gas and water; and services industries; the 1995 benchmark ratio, adjusted over time, has been used for all years for construction; and 2000 benchmark ratios, adjusted over time, have been used for all other manufacturing industries from 2000 onwards. Smoothed ratios based on the 1993 and 2000 benchmark ratios have been used for 1994 to 1999. However, adjustments have been made to the ratios for most industries from 2001 onwards to reflect the impact of increased fuel costs.

Output for most production industries are benchmarked to the 1993, 1995 and 2000 SIE results and extrapolated using a variety of movement indicators. Mining output is based on construction output movements and NIS data for other non-construction related mining (e.g. gems, gold, salt). Food, beverages and tobacco outputs are largely based on household consumption expenditure estimates; textiles, wearing apparel and footwear outputs are based on adjusted GSP exports data and informal sector production estimates; wood, paper and publishing output is based mainly on domestic use, and exports of sawn timber; and rubber manufacturing output is based mainly on NBC BOP rubber exports and MAFF production data. Metallic and non-metallic products manufacturing outputs are largely based on construction output movements. Other manufacturing GVA is estimated based on GVA movements for manufacturing industries, excluding textile, wearing apparel and footwear manufacturing.

Construction estimates are based on output for the private sector for 1993 benchmark and 2002 expanded data, with a composite movement indicator of imports of construction materials and Phnom Penh building approvals being used to estimate output for intervening years, plus government expenditure on construction. Compilation of private sector construction value added estimates from 2001 onwards are now based on modelled actual expenditure on construction compared to building approvals data used previously. Output for electricity, gas and water is based on household consumption estimates and administrative data from EDC, supplemented by NIS investigations during provincial field visits.

Output for most services industries are largely based on household final consumption expenditure (HFCE) and international tourist expenditure (ITE) estimates, and to a lesser extent on administrative data and informal sector estimates. For Trade output, HFCE and ITE data are used for retail trade, and crops, livestock and fish production and BOP imports and exports data are used for wholesale trade estimates, with SIE, informal sector survey and NIS fuel price data used for GVA ratios. Finance and public administration output estimates are based on administrative data (provided to the NIS by MEF and NBC in late April 2006). For real estate and business services output estimates are based on HFCE, BOP and coverage adjustment data. For other services outputs, CDC NGO expenditure, HFCE and TOFE data are used.   

In most cases, HFCE implicit price deflators, and Phnom Penh or Urban CPI indices are used to derive constant price estimates. For export-oriented industries such as textile, clothing and footwear (GSP exports component), rubber and wood manufacturing, HFCE deflators are crosschecked with WEO and trading partner import price indices or export prices indices. For construction a composite price index of imported materials and wages is used. For finance, interest rate movements crosschecked with Urban CPI movements, are used. For public administration a composite of HFCE implicit price deflators and a wages index is used.

Household final consumption expenditure is the main component of GDP by expenditure share. The HFCE series are compiled separately for Phnom Penh, other urban and rural areas, and are benchmarked to the 1993-94, 1999 and 2004 CSES. The expenditure weights for 1994 to 1999 are smoothed weights between the 1993 CSES and 2000 (price adjusted from the 1999 CSES) weights. Results of intervening household expenditure surveys such as the 1996 and 1997 CSES have been used to validate the smoothed weights. The expenditure weights for 2001 to 2003 are smoothed weights between the 2000 and 2004 benchmark weights. So that 1993, 1994, 1999, 2000 and 2004 HFCE estimates are based on actual survey data, while HFCE for 1995 to 1998, 2001 to 2003, and 2005 are modelled estimates. From 2000 onwards the HCE price deflators are based on the Urban CPI. For 2005 onwards, population growth rates, expenditure trends between 2000 and 2004, and increases in Phnom Penh and Urban CPI prices are used to estimate HFCE. Adjustments are made for under-reporting of expenditure by households (e.g. meals away from home and take away food, alcohol and tobacco, gambling, and adult entertainment and personal services). Data collected by CDC on actual NGO expenditure are used for private non-profit organizations serving households on the expenditure side of the accounts. For government final consumption expenditure, the government TOFE accounts from MEF are utilized.

The commodity flow approach is used in the estimation of expenditures on durable equipment. This is done by monitoring the supply and use of capital goods crosschecked with capital imports data. For construction, the expenditure approach is used based on government actual expenditure and adjusted private sector construction permits data. Changes in inventories of livestock and poultry are estimated using the data from MAFF, and the parameters calculated from the results of the Livestock Survey conducted by MAFF/FAO. 

The national accounts international trade in goods and services estimates are fully consistent with the BOP estimates (provided to the NIS in late April 2006) from 1998 to 2005. Exports and imports of merchandise trade compiled by CED are also utilised in verifying BOP data. Adjustments made by NBC for unrecorded exports and imports of merchandise are used directly. The NIS made some additional adjustments to unrecorded trade and re-exports for 1993 to 1997 in order to ensure consistency with revised coverage of BOP data for later years.

The NIS is very grateful for the continued support and cooperation of the line ministries and other agencies in providing data for use in the national accounts. We acknowledge the difficulties surrounding the collection of such data, as respondents are often reluctant to provide a true picture of production. However, in the absence of any other data sources the NIS has no choice but to use this information in the GDP estimates. In order to improve the quality and credibility of the National Accounts, it is vital that financing and training are provided to enable the collection of independently sourced data. The new Statistics Law gives the NIS the authority to collect the data and to guarantee that an individual’s or business’s data will remain confidential and not able to be identified in the published results. However, the lack of sufficient funding for statistical data collection, processing and dissemination within government continues to be the main cause for the remaining shortcomings in the coverage and quality of Cambodia’s national accounts estimates.

Proposed Improvements for the Next Estimation Round

        A number of improvement are proposed for the national accounts estimates over the coming year. These improvements include:

• Incorporating the results of the CSES 2004 in relation to updating informal sector production outputs, intermediate consumption, value added and other aggregates.
• Implementing a regular manufacturing establishments survey.
• Incorporating the results of the new annual household survey that is expected to commence in October 2006.
• Continuing to improve access to and quality of other source data, especially for prices and services sector.
• Continuing to improve compilation methods, especially for services sector estimates.
• Continuing to expand dissemination of national accounts data.
• Continuing to improve consistency with SNA 1993 standards