LBO: 01/05/2006"
The claims of two different growth rates for Sri Lanka by two agencies have highlighted inherent shortcomings in the preparation of national income accounting and also added fuel to sceptics who do not believe economic growth numbers.
Last week, the Central Bank of Sri Lanka said the country's economy grew by 6.0 percent just a day after the ministry of finance said it grew by 6.2 percent.
The ministry of finance used the national accounts prepared by the Department of Census and Statistics, while the Central Bank generated its own numbers.
But in practice, both agencies largely use the same sources and share information to a great degree.
So what is the correct growth figure?
"There is no correct growth figure," says H N Thenuwara, Director of Economic Research at the Central Bank.
"A difference of 0.2 percent is not statistically significant."
Conspiracy theory
The Central Bank cancelled the press conference, citing the attempted assassination of the country's army commander.
But the Finance Ministry which released its own Annual Report with the 6.2 percent number, went ahead with the Press conference, chaired by Treasury Secretary P B Jayasundera.
This led to a theory among some media persons, that the Central Bank cancelled the traditional press conference that accompanies the release of the Annual Report, because it did not want to be drawn into a controversy over the two numbers.
Sri Lanka's Central Bank which suffers from fiscal dominance problems and has difficulty controlling inflation as a result, has become increasingly silent during the last two years.
Earlier in the year it cancelled the press conference that accompanies the 3rd quarter GDP numbers.
The press conference that accompanies the monthly monetary policy statements were cancelled much earlier.
But official insist that there is nothing sinister or secret about the two GDP growth numbers, and that it can all be explained.
Estimate
GDP numbers are essentially estimates, says Thenuwara.
Agencies that compute national account numbers always call them 'GDP estimates', though laymen and the media refer to them as 'GDP' in absolute terms.
A large part of the growth difference for 2005 is related to the base year, used by the two government agencies.
The size of the economic output generated within the country which is measures by the Gross Domestic Product (GDP) is first measured by calculated products and services of at current prices (Gross National Product or GNP includes income earned from abroad and sent back by Sri Lankan citizens and companies).
Then the effect of inflation is taken off by adjusting with a 'GDP deflator' to come up with 'constant' prices.
Coming with the GDP deflator is a tedious task, with each sector being measured separately.
"In a sector like tea, where lots of data is available, we can find the volumes and prices and find the effect of prices rises easily," explains Central Bank's Director of Statistics, Anila Dias Bandaranaike.
"For other sectors like construction we have to use indices."
The Central Bank came up with a GDP at current prices of 2,366 billion rupees, and a GDP deflator of 9.9 percent, while the Census Department came up with current GDP of 2407 billion and a GDP deflator of 10 percent.
Different Base
The Central Bank uses a 1996 base and the Census Department a 1998 base, using respective consumer economic surveys of those years.
A statistical model of the economy (essentially a snapshot of the economy) is also constructed at the time, using actual information and surveys.
In the model used by the Census Department based on the 1998 survey, agriculture has a weight of 12.2 percent, while the Central Bank's data structure has a weight of 17.2 percent.
A tale of two economies
.................Census Department........Central Bank
Agriculture............12.2..................17.2
Industry...............30.5..................27.0
Services...............57.3..................55.8
(Sources : Central Bank, Department of Census and Statistics)
The data structure used by the Census department has 15 sub-sectors, while the Central Bank has only 9.
For example, the Central Bank sub-divides the services sector into four sub-sectors; wholesale and retail, transport storage and communication, financial services, real estate and business services, public administration defence and other social and personal services.
The Census Department has additional sub-sectors such as, ownership of dwellings, private services, hotels and restaurants in its structure of the services sector.
Agriculture, a stagnant sector, contributes little to economic growth, while services and industry brings in most of the growth.
Therefore, a data structure which has a higher weight for services and industry, like that of the Census Department model, may generate a higher GDP number.
GDP numbers are estimates to a large degree, except for areas like tea, where meticulous record-keeping is the norm, as well as the financial sector, telecommunications and large factory industry.
In theory, the growth rates between two consumer economic surveys should lead to the new structure of the economy in later years.
But in practice, since there a lot of estimates, and assumptions and their effects can get compounded over several years.
Statisticians use surveys to take a 'new picture' of the economy from time to time and use it as an error-correcting mechanism.
Statisticians however, dislike words like 'errors', for them such techniques are a standard part of their work.
What is important is for users of information to use the same data series for comparison purposes across time.
But for ordinary laymen, the two GDP growth numbers provide ideal fodder to re-inforce the adage of 'lies, damn lies and statistics'.
The claims of two different growth rates for Sri Lanka by two agencies have highlighted inherent shortcomings in the preparation of national income accounting and also added fuel to sceptics who do not believe economic growth numbers.
Last week, the Central Bank of Sri Lanka said the country's economy grew by 6.0 percent just a day after the ministry of finance said it grew by 6.2 percent.
The ministry of finance used the national accounts prepared by the Department of Census and Statistics, while the Central Bank generated its own numbers.
But in practice, both agencies largely use the same sources and share information to a great degree.
So what is the correct growth figure?
"There is no correct growth figure," says H N Thenuwara, Director of Economic Research at the Central Bank.
"A difference of 0.2 percent is not statistically significant."
Conspiracy theory
The Central Bank cancelled the press conference, citing the attempted assassination of the country's army commander.
But the Finance Ministry which released its own Annual Report with the 6.2 percent number, went ahead with the Press conference, chaired by Treasury Secretary P B Jayasundera.
This led to a theory among some media persons, that the Central Bank cancelled the traditional press conference that accompanies the release of the Annual Report, because it did not want to be drawn into a controversy over the two numbers.
Sri Lanka's Central Bank which suffers from fiscal dominance problems and has difficulty controlling inflation as a result, has become increasingly silent during the last two years.
Earlier in the year it cancelled the press conference that accompanies the 3rd quarter GDP numbers.
The press conference that accompanies the monthly monetary policy statements were cancelled much earlier.
But official insist that there is nothing sinister or secret about the two GDP growth numbers, and that it can all be explained.
Estimate
GDP numbers are essentially estimates, says Thenuwara.
Agencies that compute national account numbers always call them 'GDP estimates', though laymen and the media refer to them as 'GDP' in absolute terms.
A large part of the growth difference for 2005 is related to the base year, used by the two government agencies.
The size of the economic output generated within the country which is measures by the Gross Domestic Product (GDP) is first measured by calculated products and services of at current prices (Gross National Product or GNP includes income earned from abroad and sent back by Sri Lankan citizens and companies).
Then the effect of inflation is taken off by adjusting with a 'GDP deflator' to come up with 'constant' prices.
Coming with the GDP deflator is a tedious task, with each sector being measured separately.
"In a sector like tea, where lots of data is available, we can find the volumes and prices and find the effect of prices rises easily," explains Central Bank's Director of Statistics, Anila Dias Bandaranaike.
"For other sectors like construction we have to use indices."
The Central Bank came up with a GDP at current prices of 2,366 billion rupees, and a GDP deflator of 9.9 percent, while the Census Department came up with current GDP of 2407 billion and a GDP deflator of 10 percent.
Different Base
The Central Bank uses a 1996 base and the Census Department a 1998 base, using respective consumer economic surveys of those years.
A statistical model of the economy (essentially a snapshot of the economy) is also constructed at the time, using actual information and surveys.
In the model used by the Census Department based on the 1998 survey, agriculture has a weight of 12.2 percent, while the Central Bank's data structure has a weight of 17.2 percent.
A tale of two economies
.................Census Department........Central Bank
Agriculture............12.2..................17.2
Industry...............30.5..................27.0
Services...............57.3..................55.8
(Sources : Central Bank, Department of Census and Statistics)
The data structure used by the Census department has 15 sub-sectors, while the Central Bank has only 9.
For example, the Central Bank sub-divides the services sector into four sub-sectors; wholesale and retail, transport storage and communication, financial services, real estate and business services, public administration defence and other social and personal services.
The Census Department has additional sub-sectors such as, ownership of dwellings, private services, hotels and restaurants in its structure of the services sector.
Agriculture, a stagnant sector, contributes little to economic growth, while services and industry brings in most of the growth.
Therefore, a data structure which has a higher weight for services and industry, like that of the Census Department model, may generate a higher GDP number.
GDP numbers are estimates to a large degree, except for areas like tea, where meticulous record-keeping is the norm, as well as the financial sector, telecommunications and large factory industry.
In theory, the growth rates between two consumer economic surveys should lead to the new structure of the economy in later years.
But in practice, since there a lot of estimates, and assumptions and their effects can get compounded over several years.
Statisticians use surveys to take a 'new picture' of the economy from time to time and use it as an error-correcting mechanism.
Statisticians however, dislike words like 'errors', for them such techniques are a standard part of their work.
What is important is for users of information to use the same data series for comparison purposes across time.
But for ordinary laymen, the two GDP growth numbers provide ideal fodder to re-inforce the adage of 'lies, damn lies and statistics'.
