Inter-relationship Between China’s Input-Ouput Estimation, production-based GDP and Expenditure-based GDP
Department of National Accounts, NBS
PENG Zhilong
China started it’s GDP estimation from production side in 1985, GDP estimation from expenditure side in 1992, and input-output estimation in 1987 (which had started been before, but it was in 1987 that NBS began to compile I-O table formally). The three accounts have been greatly improved through years of efforts, data source is getting more richful and accounting methods is becoming increasingly normative, which could provide more and more information. However, an inevitable outstanding question is that the data source and measurement of these accounts are lack of close inter-relationship, and the accounting contents appear to be independent with each other in practice, therefore it is difficult to guarantee the consistency of the results.
I. The relationship between input-output estimation and GDP at production and expenditure approach
1. What are the contents of input-output survey?
I-O survey targets at major products of enterprises and at enterprises or corporations for service industry, which belong to a kind of major survey on the whole. There are several structures contained in the survey, including：the structure of intermediate input of products (enterprises or corporations)；the structure of value-added (the proportion of depreciation, compensation of employees, net production tax and operating surplus)；the structure of household consumption；the structure of fixed capital formation; the structure of inventory; and the structure of imports and exports.
2、The characteristics of input-output table compiling
China’s I-O table uses gross output value (output) and value-added by industries (94 industries) in GDP as the level in the first and third quadrant of I-O table respectively, then calculate each sub-item of intermediate input and value-added with the structure data obtained from I-O survey. NBS uses household consumption, government consumption, gross fixed capital formation, changes of inventory, imports and exports in expenditure GDP as the level in the second quadrant of I-O table, then combining with the structure data obtained from I-O survey to determine each column vector of final use.
For example：The intermediate input of industry and corresponding gross output value are obtained from I-O survey, in which , thus we could derive the intermediate input structure by , and the value-added structure could be calculated directly using survey information, that is . We could find gross output value and value-added in GDP, therefore the intermediate input of industry could be derived by, similarly, the value-added vector of industry could be calculated by. The calculation of data in the second quadrant operates the same.
Obviously, at present, China’s I-O survey and I-O table measurement are conducted after GDP estimation at production and expenditure approach, then use gross output value (output) and value-added in production-based GDP, combining with household consumption, government consumption, gross fixed capital formation, changes of inventory and imports and exports in expenditure-based GDP to determine the main aggregates in I-O table, while I-O survey just provides some structure ratios which are used as estimating base but not the base for GDP accounting. Further more, I-O survey just captures some small structure ratios, such as the internal-structure of intermediate consumption and value-added. According to the current relationship between GDP and I-O survey or I-O table, it is apparently that GDP serves I-O table compiling and related data from GDP turn into the control data of corresponding part in I-O table, rather than I-O survey or I-O table serves GDP estimation and creates certain control data for GDP estimation.
II. The relationship between GDP at production approach and GDP at expenditure approach
China’s production-based GDP is not the sense of GDP at production approach in deed. Actually, except agriculture and industry, other industries’ value-added are estimated using income approach (distribution approach).
1. Annual Estimation
There are three aspects where GDP at production and expenditure approach have correspondence: First, government (administrative and institutional units) output and government expenditure, which use the same data source in their estimation, that is ‘Final Annual Accounts of administrative and institutional units’ compiled by Ministry of Finance；Second, owner occupied dwellings, which is not only treated as output of corresponding sub-industry in production-based GDP, but also as part of household consumption in expenditure-based GDP. In practice, the part of owner occupied dwellings is added to the urban and rural household consumption; Third, FISIM, certain method is used to identify the part of FISIM serving for household, and plus it to household consumption.
2. Quarterly Estimation
In quarterly GDP estimation, nearly no relationship is indicated between the data from production and expenditure approach. In quarterly GDP estimation from production side, the following formula is used massively to calculate the value-added by industries indirectly: value-added by industries = base number of the same period last year × growth rate of the representative operational indicator of the industry × elasticity ratio. In which, elasticity ratio = growth rate of value-added of the industry last year / growth rate of the representative operational indicator of the industry.
III. Difficulties and doubts
1. The relationship between specific statistics and GDP Estimates
In GDP estimation, we usually need to use a lot of specific statistics, which are compiled and published by specific statistical departments，for example, statistics of gross output value and value-added in agriculture, statistics of industrial value-added, statistics of gross output value in construction, statistics of sales output value in wholesale and retail trades, statistics of operating income in hotel and catering services, statistics of investment in fixed assets, statistics of total retail sales of commodities, statistics of imports and exports. If use these data directly in GDP estimation (which were required to be adjusted with GDP coverage before), both gross data and growth rate would appear great discrepancy between production-based GDP and expenditure-based GDP. Under such circumstance, how to treat the relationship between GDP accounting and specific statistics? Whether we could do man-made adjustments with specific statistics? For example, multiplying specific statistics by a ratio unequal to 1？
2.How to apply the information of I-O table and I-O survey to annual and quarterly GDP calculations?
3. How to use the information of I-O table and I-O survey to control the discrepancy appears between the production-based GDP and expenditure-based GDP?
4. How to keep the consistency of the deflators between GDP production and expenditure approach?
5. Is there any perfect method to evaluate the reliability of GDP growth rate, such as single-indicator evaluation and multi-indicators evaluation?
4