Thursday Dec 12, 2024
Thursday Dec 12, 2024
Friday, 7 March 2014 00:55 - - {{hitsCtrl.values.hits}}
GDP compilation
The need for National Accounts Statistics (NAS) comes with the desire of having a numerical value for the size of an economy together with the structural breakdown. Such an estimate forms the basis for identifying developments in the economy; such as whether the economy is expanding or is it not growing at previous levels, what are the sectors that support growth and what areas are moderating? This information is a crucial component for establishing future policy decisions within a country for both public sector and private sector.
Further, the NAS provides a way for comparing the level of economic development within countries and regions. Are people of a particular country better off in monetary terms than those of another; is a question that could also be answered through these estimates. For this purpose there should be a consistent and globally acceptable methodology in compiling NAS.
The United Nations’ System of National Accounts (SNA) serves this purpose in compiling the National Accounts for a given economy by spelling out the framework and guidelines on producing NAS. In consideration of the global developments UN has been publishing a methodical document from 1947, with the latest release being published in 2008.
Currently, the Department of Census and Statistics of Sri Lanka (DCS) principally follows the 1993 SNA version and is in the process to elevate to the 2008 SNA, which is an updated version of the 1993 SNA. SNA recommends a holistic approach in compilation of national accounts that is it does not consider each individual transaction independently but rather considers the total economic output of a country.
The per capita GDP
Consequently, the per capita GDP (GDPPCI) estimate is derived by dividing the total output, obtained from the system of national accounts, by the mid year population estimate of the country. This provides statistics that are used internationally to assess the stages of development of individual countries.
Per capita household income based on HIES
Meanwhile, household income and expenditure surveys are carried out to understand the movement in more micro level indicators such as income distribution. In the case of Sri Lanka the Department of Census and Statistics carries out the Household Income and Expenditure Survey (HIES) with three year intervals to capture the developments of the household sector.
The latest survey was carried out in 2012/13 reveals information on demographic characteristics together with household expenditure and income. Accordingly, HIES also provides an estimates on per capita household income through the survey results.
So these two indicators on per capita income (PCI) seem to be estimating the same thing for the general public. Since the HSPCI includes any source of income such as transfer incomes, which are not included in the GDPPCI, then in fact the HSPCI could be larger than the GDPPCI.
Though, a layman could feel that these two estimates are the same, but someone with a reasonable knowledge in the subject matter can understand the difference. Hence, we will look in to the sources of discrepancies between the PCI estimates from the GDP and Household Surveys (HS).
Why do the two PCI estimates differ?
Retained/undisbursed profits of the corporate sector
A source of deviation is the retained/undisbursed profits of the corporate sector. Institutions that make a profit through the creation of value added output would retain some of it for future investments with the corporations. Hence, these profits by the corporate sector would not completely flow into the household sector and these would not be reflected in household income estimated through surveys but would be included in the GDP estimate of the country. The Non-Profit Institutions Serving Households (NPISH) are also a component of the economy and their contribution is included in the compilation of national accounts estimates; however this sector is not reflected in a household survey.
Role of Government
Another consideration is the role of the Government within the economy, which provides welfare benefits. The cash transfers by the Government towards the households would be reflected in a household survey. However, collective consumption such as health, education and defence provided by the Government, would not be included in household data while they are included in national income data. Capturing of personal tax information would again be disputed between the two income estimates as taxes deducted at the source of income received by an individual might not be included in a household survey.
HIES is a sample survey
Survey data would carry with it a sampling bias with different cultural and social attitudes, which could be manifested in the reluctance in providing information on income.
Contd. on page 17
It is a well-established fact that respondents are less willing to divulge income information, particularly in countries where the tax collection framework is not comprehensive. Hence, this would lead to biasness towards under-stating income earnings. This issue is addressed in household surveys by validating income estimates through expenditure information provided by the households; as the expenditure of a household should be financed through the sources of income.
However, surveys carry with them other inherent characteristics as well, which could impact the findings. The fatigue in answering a long set of questions in describing a detailed list of household characteristics could lead to complacency in answering questions at the later stages of the questionnaire. Further, the memory recall issue is also another contributing factor towards survey results; for instance questions on spending across a span of one year would be more susceptible to this issue again leading to the accuracy of survey results.
Time has become a more valuable asset as countries develop, leading to reluctance to spend time on a survey, which would not yield any direct benefit to the respondent. Although this could be observed at all income levels; people who earn a higher income would be more unwilling to respond to such surveys as this could be further augmented with the reluctance to reveal actual income details. Hence, survey data could include a bias towards low-income households. All these facts, on reluctance to divulge income, memory recall issues and limited time, would contribute towards bringing down the estimates on income derived through household surveys.
Country experiences
Table 1 shows data on such estimates from a number of countries around the globe covering OECD, East Asian, Southeast Asian and South Asian countries.
However, even in the set of countries considered there are significant differences between the two PCI estimates, with GDPPCI being considerably higher than HSPCI. It should be noted that most of these countries estimate GDP by the income approach as well; hence income data of these countries show that the household surveys only reflect a share of the total income that is generated within the economy. Hence, Sri Lanka is not an exception in terms of having a higher GDPPCI, in fact developed as well as developing countries share this similar characteristic in their respective statistics as well.
Therefore, there are a number of factors that can contribute to the estimate of the HSPCI compared to the GDPPCI. Further, this has been the similar experience around the globe as well, highlighting the fact that Sri Lanka is not an exception but rather the norm in this aspect.
This fact is further reiterated in Table 2 through the historical progress of the HSPCI in comparison to the GDPPCI, showing that a similar trend has been observed even prior to 2005 when the writer of the article published in the Opinion column of the Daily FT was employed at the Central Bank of Sri Lanka (CBSL) and then the CBSL was responsible in compiling both GDPPCI and HSPCI.
Director
Statistics Department
Central Bank of Sri Lanka
