Reliability of GDP and per capita income numbers in Sri Lanka
Thursday, 20 March 2014 00:00
-
- {{hitsCtrl.values.hits}}
By Raja B.M. Korale
There has been an interesting exchange between the columnist at Daily FT, W.A. Wijewardena (WAW), and the Director of Statistics of the Central Bank (DSCBSL) on the differences between the Per Capita Income (PCI) estimated from national accounts, and PCI calculated from household income and expenditure surveys. Such interactions are useful to facilitate the public to understand the crucial issues involved in the estimation of the size of the economy through national accounts and welfare levels of people ascertained from household surveys.
This note was prepared not in any way to dilute the issue that is raised in the paper prepared by WAW as to whether the current GDP aggregates and GDP growth rates could have been overestimated, a central issue evident from his presentation. This brief response was written to add to the material on this issue in the belief that they will be useful for their resolution.
As regards the specific issue of the reported difference between the Per Capita Income (PCI) estimate based on GDP/GNP and the Per Capita Income (PCI) estimate based on the Household Income and Expenditure Surveys (HIES), the difference or the deviation that is reported can be explained.
These are two different indicators based on different concepts and definitions as also explained by both WAW and DSCBSL. Hence, their values will also be different and it is not necessary that the difference or deviation between the two numbers should be zero, even if the estimates prepared are measures that are complete, comprehensive and reliable and accurate. The PCI of Households can be treated as only a part of the PCI based on National Income. Thus, that the values of these two indicators should be numerically close is not a requirement.
System of National Accounts (SNA)
For the purposes of the System of National Accounts (SNA), institutional units that are resident in the economy are grouped together into five sectors, viz., non-financial corporations, financial corporations, government units including social security funds, Non Profit Institutions (NPIs) Serving Households (NPISHs), and Households. The five sectors together make up the total economy. (See SNA 2008 Edition, paras 1.9 and 1.10). The UN guidelines issued on the collection of data on household income and expenditure broadly follow the SNA framework but there are variations.
Thus, according to the SNA framework, the household sector is only one of the five sectors and therefore the contribution of the household sector to the economy will always be smaller than the aggregate amount in respect of all five sectors. The relative size of the household sector depends on the size and relative shares of the other sectors. The private enterprise, banking and government sectors make significant contributions to the GDP. Thus, the value of the household survey based PCI cannot approach the PCI computed from GDP estimates.
According to the GDP estimates prepared by DCS, PCI based on GDP estimates, which is computed by dividing the GDP (or GNP as the case may be), by the estimated population, the estimate for the year 2012 is reported as $ 2,923 per person.
The per capita household income is derived by dividing the estimated household income by the estimated population and the HIES 2012-2013 had estimated the PCI as Rs. 143,184 ($ 1122/approximately) per year.
Thus, the two estimates differ and the household survey based estimate is only 38.4% of the GDP based PCI estimate. The overestimation of the GDP which is discussed below, had contributed to lower the reported value to what is recorded here, but in fact its value could have been higher.
It should be evident from the above comments that it would be more appropriate to compare the HIES household based PCI with the corresponding indicator that is prepared in respect of the GDP based per capita value of the GDP component “personal consumption expenditure”, and its trends, it should bear a better relationship with HIES based PCI estimates.
The relative shares of the five sector estimates and their trends should also provide indications on the completeness and reliability of the estimates. A review of the overall GDP estimates and its five components and trends in the movements of the estimates should indicate the stability of the estimates, the departures that should be examined and explained and most importantly whether they in fact reflect the current status of the economy.
Overestimation of growth rates
The issue raised by WAW on the possible overestimation of GDP aggregates and GDP growth rates should receive attention to correct the issue. A brief examination of the methods in use indicates that they are likely to overestimate the economic growth rates. The present methodologies in use which were driven by some government policy compulsions have raised these issues.
An examination of the movement of the CPI and the GDP deflator might be useful to provide indications. The GDP deflator is a measure of the prices of all goods and services, while CPI covers only goods and services bought by consumers. In the past couple of decades, the rate of increase in the value of the GDP deflator had been generally lower than that of the CPI based inflation rate until the last few years. However, during the past few years the rate of increase of the CPI has been lower than that of the GDP deflator reflecting that the inflation rate is underestimated.
As the CPI is used as a deflator in the compilation of constant price estimates of GDP, the use of a reduced CPI that is underestimated would lead to raise both the GDP aggregates as well as the GDP growth rates. The exclusion of certain important consumer items such as alcohol and tobacco on policy considerations and the use of a basket of goods and services that will lower the proportion of food expenditure in the basket all lead to the underestimation of the CPI. (See Sri Lanka Economy Transition: A Tribute to Jayantha Kelegama, 2009: Inflation: Measurement and Trends, p.265-287). According to Engel’s Law, named after the 19th century German Statistician and Economist Ernst Engel, the proportion of income spent on food tends to decline as a person’s income increases; in other words, lower income households spend a greater proportion of their available income on household food expenditure. The exclusion of liquor and tobacco etc and the use of the Colombo District based basket of goods and services with a lower food ratio would lower the CPI and its growth trends. Especially in the context of a paucity of current databases, the CPI has been and will be used in the preparation of GDP estimates which would result in the overestimation of reported GDP aggregates and GDP growth rates.
The data and estimation methods in use to compile the current GDP aggregates and GDP growth rates should be examined and studied to ascertain the inconsistencies and the extent of overestimation of these macro-economic aggregates and indicators and assist the concerned authorities to review and resolve them.
Insufficient data sources and undeveloped statistical techniques: IMF
The International Monetary Fund (IMF) had commented in their Article IV Report of April 2013 that the statistical data base that is currently available and the methodology in use do not support the compilation of adequately complete and reliable national accounts and other macro-economic aggregates and indicators.
“The national accounts suffer from insufficient data sources and undeveloped statistical techniques. The country does not have periodic comprehensive benchmarks or a system of regular annual surveys of establishments. A statistical business register, which would serve as the main basis for conducting sample surveys, is not available. As a result, the few surveys that are conducted do not have good sample frames.”
“However, detailed data needed to measure both output and intermediate consumption are mostly unavailable or not collected. As a result, the estimates of gross value added are prepared directly relying on outdated fixed ratios established from the base year 1996, often with outdated studies or ad hoc assumptions. The methodology for deriving GDP at constant prices is not satisfactory.”
The reported position by IMF as above is not favourable especially in the context where the concerned institutions had produced professional staff who had earlier been engaged as international advisers on national accounts statistics by these international organisations to upgrade the statistical development programs in foreign countries.
Thus, it is necessary to accept that the data sources that are currently available are not adequate to produce the type of reliable and robust national accounts estimates that are needed for policy planning and implementation. That the current methodologies and techniques in use will have to be reviewed, and that new methods and estimation procedures should be adopted to produce the needed macroeconomic aggregates and indicators have been highlighted.
The fact that the professional staff needed to be trained and upgraded to adopt the recommended SNA framework, concepts, definitions, compilation methodology and estimation procedures is concomitant for their adoption. As is evident, the lack of resources for generating the data bases and the staff capacity issues has no doubt resulted in the creation of situations where the recommended methodologies could not be adopted.
The Government should examine these issues in depth without delay and institute requisite remedial measures including the provision of resources for collecting and compiling statistical data bases and for staff training and development; statistical program management and supervision, to restore the credibility and reliability of the statistical estimates and indicators produced and released by the Government.
(The writer was the former Director of the Department of Census and Statistics; ADB Lead Statistics Consultant; ILO Senior Employment and Sampling Consultant; and World Bank Senior Statistics Adviser in Technical Assistance Projects in several countries)
