India’s National Accounts
It is emphasised that in view of the essentially de-centralized character of the Indian Statistical System, the continental size of the country with large diversities and federal character of polity, the Indian System of National Accounts to include regional accounts at the State level and below. The term National Accounts Statistics (NAS) in this overview should be interpreted in the inclusive sense.
The NAS is a framework that provides an internally consistent description of National macro economy based on the processing of data generated by the entire National statistical system. The estimates of National income and related aggregates and accounts are derived statistics that draw on basic data available from different primary sources. The primary sources consist of data generated as a by-product of public administration system (such as land records, collection of direct and indirect taxes, civil registration of births and deaths, etc.) as well as data collected directly through censuses and sample surveys conducted by official agencies of the Central and State Governments. For certain newly emerging activities such as software, where official statistical system is not currently in place, NAS also draws on selective non-official sources. The accuracy and quality of the National account estimates depend on (a) geographical coverage and quality of primary data; and (b) the methods, procedures and approximations used in translating the primary data into NAS framework. While the underlying concepts and methodology of compilation has been mostly standardised under the United Nations’ System of National Accounts (UN-SNA), procedures and approximations are shaped by the country-specific data collection system. For making the estimates comparable over time and internationally, the National Accounts Division (NAD) of the Central Statistical Organisation (CSO) maintains detailed, well documented methods and procedures unchanged till the revision of the base year when efforts are made to bring about improvements in these respects while bridging data gaps and introducing newly available better quality data.
Given the use of wide-ranging data sources with varying quality in different spheres in the compilation of National Accounts in all the countries of the world, weaknesses in the National statistical system get reflected in the NAS. In the federal political framework, the Indian Statistical System is decentralised in character so that NAS necessarily have to depend on a large number of autonomous source agencies. NAD often finds itself unable to make source agencies appreciate the requirements of National Accounts for timely reporting as well as for additional data or wider coverage. In a continental country like India, regional diversities in public administration and the importance given to collection, maintenance and dissemination of statistics also influence their quality. Data used for Central fund allocation to States create their own problems of reliability when they are generated by the concerned administrative departments. National Sample Survey Organisation (NSSO) has been carrying out periodic surveys that provide input into the National income estimation. While the surveys give reasonably good estimates at the all-India level, they have to grapple with the basic character of the Indian economy while collecting information about employment, incomes, trade and profit margins. This relates to predominance of self-employment in agriculture as well as non-agricultural rural and urban areas where connection of earning members to the labour market is loose and amorphous, workers often engage in multiple activities during the year, income streams are irregular, enterprises do not keep accounts and entries into and exits out of enterprises are common. Over the years, the involvement of State statistical bureaus in conducting surveys and type studies for evolving norms for National income estimation has also been on the decline. It is also important to point out that survey-based estimates have their inherent limitation in terms of known sampling errors and unknown non-sampling errors.
Standard texts on National Accounts do mention independent checks on National account aggregates by estimating them through three alternative methods: income, expenditure and commodity flow. Data limitations however, do not permit these independent consistency checks. Often, certain National account identities are used to derive certain components as residuals. For example, aggregate PFCE in the Indian system of National Accounts.
In India, the basic (gross or net) domestic product estimates at factor cost by industry or sector of origin can be broadly classified into two broad categories from the viewpoint of differences in database. Direct estimates are based on annually available statistics on a regular basis so that they reflect year-to-ear variations in the concerned economic activities. The second broad category of indirect estimation has to be resorted to when regular annual statistics are not available. In such cases, periodic benchmark survey based estimates are derived for the survey year and are extrapolated backward or forward on the basis of (often) -physical indicator of activity in the sector. The degree of approximation in this context critically depends on the sensitivity of the indicator in reflecting year-to-year variations in the concerned economic activity. By type of institutions, direct estimates mostly relate to public (of which Government proper is a component) and private corporate sector so that the estimates relating to them usually constitute what is usually referred to as “organized” sector or segment of the economy. Indirect estimates mostly relate to households (including non-profit institutions serving households) and constitute the residual ‘unorganized’ sector or segment of the economy.
While direct estimates are based on annually available statistics, their translation into National account aggregates often requires the use of certain norms, rates and ratios or other assumptions. In the absences of timely availability of annual estimates, advance, quick or provisional estimates often resort to readily available indicators of activity in the sector. Their revisions after the use of regular annual statistics sometimes bring about major changes in the provisional estimates. In such cases, the fault lies with the quality of data provided by the source agency supplying provisional indicators as well as delays by the source agency generating annual regular statistics. NAD of CSO however unfairly finds itself at the receiving end of the criticism.
The CSO has been publishing the basic documentation of methods of National account compilation in the periodical publication, National Accounts Statistics – Sources and Methods. It is released after each major revision of the estimates at updated price-base. Two publications so far available relate to 1970-71 and 1980-81 price series. The latest revision has been undertaken with 1993-94 price base. The share of “direct estimates” in aggregate GDP rose from 57.6 per cent in the 1970-71 base series to 63.7 per cent in the 1980-81 base series and further to 89.6 per cent in 1993-94. In the latest series, the sectoral share of direct estimates varies between 100 per cent (mining, registered manufacturing, electricity, gas and water supply, railways and public administration and defence sectors) to 26.6 per cent (forestry sector).
The detailed documentation in the foregoing sections at the sectoral, sub-sectoral and regional level of methods and data sources for compilation of NAS was meant to bring out the underlying problems and difficulties most of which are shared by many other countries in the world. Steps for improvements should obviously be directed toward improving the quality, coverage and timeliness of “direct” estimates while raising their share in a given aggregate or sub-aggregate at the National and regional level. “Indirect” method of estimation has to be resorted to in sectors and activities mostly marked by self-employment in small Own-account or household enterprises (requiring periodical survey-based benchmark estimates) or where large regional diversities in economic practices exist (requiring type-studies). In these cases, often times, the physical indicators used for extrapolating the benchmark year estimate backward and forward are themselves interpolated estimates thereby further increasing the degree of in-directness. While type studies necessarily have to be regionally dispersed and decentralized, the possibility of carrying out annual surveys obviously suggests itself to reduce the degree of in-directness. However, the costs and benefits of doing so Centrally need to be weighed carefully. Uneven development of survey capabilities across States would favour Centralized arrangement. While this argument is indeed persuasive, two considerations go against it. One, the decentralized character of the Indian Statistical System that has also been stressed in terms of reference of the Commission. Two, Centrally carried out sample surveys with uniform survey methods have possibly come in the way of innovative experimentation in survey methods to capture the regionally unique features. Eliciting cooperation of the States in carrying out requisite State level sample surveys and type studies would not only improve the quality of regional accounts but would also improve the reliability of National estimates.