Understanding the Complexities of Regional Accounts

Amey Sapre

1. Background

In India, estimates of national income and related aggregates are prepared at the All-India and regional level, i.e. State/UT and District level. The regional accounts comprise the state domestic product (SDP) and district domestic product (DDP) and other selected aggregates such as capital formation, savings, output of the agricultural sector that provide the primary source of information for policy formulation at the sub-national level. Compilation of such aggregates follow an internationally accepted process but differs across countries primarily in terms of data availability and procedures adopted by the statistical agency. Thus, understanding these aggregates requires a round-about understanding of the conceptual framework, processes adopted and state of data availability. National accounts provide ‘estimates’ of macro aggregates and given the strength of the method and quality of input data, all estimates have a margin of error that can be small, large or unknown. In a general discourse, data users are often unaware of such inherent errors and also do not have a sense of their magnitude or direction. While it may not be possible for data users or policy makers to reconstruct such macro aggregates or know the magnitude of errors, it is however useful to know the process of compilation, which, in part, reveals the complexities, quality and limitations of such aggregates.

This small brief describes the relevance of such macro aggregates, particularly for a state-level analysi,s and outlines the complexities in using them for policy purposes.

2. Policy Relevance: Why do we need regional accounts?

In India, individual states are large interlinked economies. For each state government, estimates of state income (the gross domestic product or GSDP), among others, are necessary indicators for preparation of annual budgets, projecting basic public finance aggregates, analysing sources of growth and for local level planning. Specifically, in the case of public finance;

(i) The borrowing space and the limit to which a state can borrow are determined by nominal GSDP

(ii) In the case of resource transfer from the Union to States, the criteria of relative ‘income distance’ is determined by the per capita GSDP of states

(iii) Nominal GSDP is used as a proxy base for determining tax capacity, tax potential and projecting revenues, particularly of indirect taxes

(iv) GSDP is also used as in indicator for determining the share and size of sector-wise economic activities, such as manufacturing sector, services, or specific sectors like tourism, etc.

(v) GSDP estimates are also necessary for preparation of annual state budgets and for presenting the overall state of economy.

(vi) GSDP estimates can provide a reliable summary assessment of the contribution of different institutional sectors in the economy – such as the general government, public sector, private sector and the unincorporated or household sector.

GSDP estimates receive singular attention in policy debates. However, the full set of accounts, i.e. GDP by expenditure, income, capital formation, institutional accounts, the input-output table, expenditure by functions of the government., are in fact more useful in determining the progress and performance of the regional economy. On this front, in India, there has been limited or no progress in compiling the full set of accounts at the state level. Except for GSDP by sectors, no other statistical outputs are available in full at the state level. A brief summary of what is available is as follows.

3. Compilation: What can be and is done

In comparison to national level aggregates, i.e. GDP, regional aggregates have a few qualifications that ought to be noted. Regional estimates (GSDP in this case) can be prepared by two approaches, namely: the income originating within the region and income accruing to the region. In the first approach, the estimates correspond to the income originating to factors of production that are physically located within the geographic boundary of the region. Thus, the GSDP (or final value-added estimates) reflect the net value of goods and services produced within the region. In the second approach, the estimates correspond to the income accruing to the residents of a region, irrespective of whether factors of production are residing or operating within the region. In this case, the GSDP estimates reflect the net value of goods and services available to the residents or attributable to the region.

Conceptually, income accruing is wider in coverage than income originating in the region. Income accruing also provides a true and fair measure of the economic standard of living or per capita aggregates of the region. However, this method is data intensive and requires detailed inter-regional flows of goods & services, factors of production and factor payments. Regional entities such as states and districts have ‘open boundaries’ with free physical and financial flows. Due to non-availability of data that can capture all such flows, it is practically difficult to prepare estimates based on the income accruing concept. Thus, in its present structure, GSDP estimates are prepared by the first approach, i.e. domestic product originating within the geographical boundaries of the region. Both methods, however, can lead to potentially different estimates, primarily because free flow of factors of production, goods and services and factor payments can lead to a significant difference between the income originating and accruing to residents.

Preparation of national account aggregates follow the guidelines of the System of National Accounts (SNA) and a standard template in a country-specific context. Within this scheme GSDP compilation follows the same principles, but differs in terms of the process. There are two accepted practices, i.e. either a top-to-bottom approach in which national totals are estimated first and its parts are allocated to regions based on indicators, or a bottom-up approach, in which estimates are prepared for the region and are aggregated to arrive at national totals. To elaborate, the economy in the national accounts is divided into sectors that represent economic activity (such as agriculture, fishing, forestry, manufacturing, construction, trade, transport, public administration, etc.) and institutional entities that represents who undertakes the activity (such as general government., private vorporates and households). Thus, estimates of value added are prepared for each entity across sectors and aggregated to arrive at national GDP.

In the present system of compilation of GSDP, the approach is primarily top-to-bottom. Except for direct estimation of value addition in the agriculture and allied sectors (crops, livestock, fishing) ,the estimates for all of manufacturing and the services sector are prepared as national totals and allocated by the statistical agency based on state-level representative indicators. Such an allocation process is conventionally followed in case of supra-regional activities such as railways, communication, etc. that flow across regions and cannot be directly computed for a specific region. However, under the present system of compilation, it is followed for nearly all sub-sectors.

4. Practice: What and how we compile
Theoretically, GSDP by production, income or expenditure method is equivalent. However, in practice, it may not be possible to compile the entire GSDP (or GDP for the nation) by all three methods. In India, the choice is dictated by data availability and convenience. A brief outline is as under.

Production Approach: In this approach, first the value of total production is estimated, which constitutes the Gross Value of Output (GVO) and the value of Intermediate Consumption (IC) is deducted to arrive at Gross Value Added (GVA). This approach is typically followed for commodity producing sectors like, Agriculture (including Livestock), Forestry, Fishery, Mining & Quarrying and Manufacturing.

Income Approach: In this method, the income accrued to the factors of production like, land, labour and capital in the form of factor payments, i.e. rent, wages and salary (including any social security benefits), interest and profit (or loss, as a residual) of the entrepreneur are used for arriving at total value added. This method is typically followed for sectors such as; Electricity, Gas and Water Supply, parts of Trade, Hotels & Restaurants, Transport, Storage & Communications, Financing and Insurance, Real Estate, Ownership of Dwellings, Professional Services, Public Administration and other personal services.

Expenditure Approach: This method measures income at the stage of disposal, i.e. similar to accounting for final expenditures. The method based on the commodity flow approach and is used in the Construction Sector.

Although the methods followed are consistent across states, the quality of data, particularly for the agricultural sector and indicators used for allocation may differ. The broad structure is summarized as follows;

I. Organized Sector which includes
(a) General Government (GG) – Estimates for GG are prepared from the State Budgets or are distributed in the proportion of the Central Government Employees.

(b) Public Financial/ Non-Financial Corporations
1. Departmental Enterprises (DE) or Departmental Commercial Undertakings (DCU) – Estimates are prepared from State Budget or distributed in the proportion of the Central Government Employees.
2. Non-Departmental Enterprises (NDE) or Non-Departmental Commercial Undertakings (NDCUs) – Analysis of Profit and Loss statement of NDCU for States or All-India estimates are allocated on the basis of State wise number of employees and value of assets.

(c) Private Financial/ Non-Financial Corporations
1. Private Incorporated Enterprises and 2. Quasi-corporations: All India estimates are allocated using indicators (based on ASI and NSSO’s unincorporated enterprise survey)

II. Households or Unorganized sector: Estimates of services (including quasi corporate sector) and manufacturing sector are prepared for the base year by multiplying the value added per worker by labour input and extrapolating these benchmark estimates with suitable indicators for other years. Information on Taxes and subsidies on products are provided by the NSO to respective states.

A detailed version of the process of compilation is available in compilation manuals of Directorate of Economics and Statistics (DES) of respective states, CSO (2015), NSO (2019), NSO (2020) and Sapre & Bhardwaj (2023).

5. Complexities: What we have

Although GSDP estimates provide a summary assessment of the state economy, the estimates hide several complexities that data users may not conceive. In a top-to-bottom approach, sector-wise allocations to states depend on the strength of the indicator chosen. For instance, in the organized part of the economy, i.e. General Government and the Private Corporate sector, data availability is annual, but a top-to-bottom allocation process does not accurately capture the changes in the regional economy. In the case of the organized manufacturing sector, the national totals are compiled from the MCA21 database that collates financial statements of available firms. These totals are then distributed across states for each manufacturing activity as per their shares in the value added compiled from the Annual Survey of Industries. Undoubtedly, when MCA21 and ASI based estimates present a contrasting picture, the sector-wise estimates of GSDP start to show a disconnect from trends of the states.
Similarly, non-availability of source data on a regular basis across entities introduces significant measurement error. This problem is serious in case of the unincorporated or the household sector. These estimates are prepared only in the base year or the year when the unincorporated enterprise survey is available. For intertwining years, estimates are extrapolated, largely using indicators of the organized sector and then distributed across states, again using indicators. For example, in the services sector, national total of GVA from hotels, restaurants for private corporate and household sector are distributed based on tourist arrivals (domestic + foreign). In case of ownership of dwellings, estimates of houses are based on 2011 census and gross rentals obtained in the base year (2011) are moved using housing index (CPI) and number of housing by the inter-census decadal growth rates. While these may plausible or practical, estimates only have a large margin of error as we move further from the base year. As the process continues, the degree of error moves from large to unknown.
GSDP estimates are also confronted by large revisions when the national series undergoes a base year revision. New methods, databases etc. are introduced to improve the depth and coverage of aggregates. However, they also bring complexities especially in doing a bottom-up approach. NSO (2020) highlighted the problems with an increasing top-to-bottom approach and the need for a State’s actual data to be incorporated to the extent possible so that the true picture of the State economy is appropriately portrayed.

6. Challenges: What we face

Changes in the economy are a continuum and capturing such qualitative and quantitative changes is a challenge for the statistical system. Compiling macro aggregates is a combination of methods, data and timely improvements in the processes. Changes in the past decade, particularly in areas of telecommunication, digital services, e-commerce, electronic payment systems and e-commerce based service delivery have significantly altered the services sector and composition of employment. Thus, in effect, it must have also changed sources of value addition and growth. Similarly, if use of digital technology has changed business of transport, retail trade, hoteling, delivery of personal services, education, coaching, etc., it ought to have impacted the services sector across states. Our compilation processes have a limited ability in capturing such deep and wide changes at the national level, especially when a large part of the economy (45%), i.e. the household sector’s estimated output, are moved forward from the base year.
The use of macro aggregates may be necessary for policy purposes, but they have a limited ability to explain the state of the economy. As of now, the challenge of building the full set of accounts for a fair assessment of the regional economies is far more serious than putting the estimates to policy use.

Amey Sapre is Associate Professor at NIPFP. The views are of the author and not of the organization. This article is an abridged and revised version of Sapre & Bhardwaj (2023) Status and Compilation issues in national accounts statistics, NIPFP Working Paper, No. 379 Email: amey.sapre@nipfp.org.in