How to correctly understand government statistics

(Picture source: Tu Chong Creative)

Zhang Jiankang/text

Government economic statistics have received increasing attention, so that some important data often become the focus of social discussion. Although the government statistics department has repeatedly made clarifications and explanations, whenever important statistical data is released, there are always different voices in the society. Whether various voices are right or wrong requires professionals with highly specialized knowledge to give professional answers.

In June 2023, Xu Xianchun, a special researcher at the National School of Development at Peking University, published the book "Perspective on Chinese Government Statistical Data: Understanding and Application", which strives to provide an overview of the current public, academic and policy levels from the levels of theoretical review, case analysis and practical application. Several government statistics that attract the most attention, including GDP, disposable income of residents, consumption, investment, and import and export, are used to clarify doubts and confusions in order to achieve the purpose of settling disputes and ending disputes. Taking advantage of the publication of the new book, the Economic Observer conducted an exclusive interview with special researcher Xu Xianchun.

"Chinese Government Statistics in Perspective: Understanding and Application"

Xu Xianchun/Author

social science literature press

June 2023

Why there is a contrast between statistical data and intuitive feelings

Economic Observer: Mr. Xu, let’s start with some government statistics just released.

In mid-July, the National Bureau of Statistics released the main data on China’s economic growth in the first half of 2023, and then different voices emerged in society. What do you think about this phenomenon?

Xu Xianchun: It is a relatively common phenomenon that there is a contrast between some people’s intuitive feelings and government statistics. Different groups often look at government statistics from different angles. Different perspectives lead to different feelings about the statistics. For example, per capita disposable income of residents. The national per capita disposable income released by the National Bureau of Statistics is calculated based on data obtained from a survey of 160,000 survey sample households across the country. It represents 160,000 sample households across the country, including high-income households, upper-middle-income households, middle-income households, low-middle-income households, low-income households, etc. Different groups have different feelings about the per capita disposable income of residents across the country. High-income groups may feel that the national per capita disposable income announced by the National Bureau of Statistics is too low, while low-income groups may feel that it is too high. The income of a high-income family may be equivalent to the income of several low- and middle-income families. On average, the income of the 160,000 sample households will most likely not be a median, but more likely to be above the median, so even if Among the 160,000 sample households, the per capita income of a considerable number of sample households will fall below the national per capita disposable income. Therefore, a considerable number of people may feel that the national per capita disposable income announced by the National Bureau of Statistics is on the high side. .

Another example is the CPI increase. The feelings of low- and middle-income groups are more likely to be in contrast with the statistics released by the National Bureau of Statistics. Everyone is very sensitive to the price changes of daily consumer goods such as grains, vegetables, fruits, and pork, but they do not often feel the price changes of durable consumer goods such as color TVs, refrigerators, washing machines, computers, and automobiles. When calculating the CPI increase, daily consumer goods and durable consumer goods will be put into the same basket for calculation. According to the calculated results, if the increase has narrowed, the reason may be that the price of durable consumer goods has dropped, and the price of daily consumer goods has not changed, or even increased. Low- and middle-income families who purchase less durable consumer goods will think that the price announced by the government The statistics do not match the actual price changes you feel.

Economic Observer: Does a similar situation exist in developed countries in Europe and the United States?

Xu Xianchun: Different groups have different feelings about statistical data, and the same problem exists abroad.

In 2003, I accompanied the then Director of the National Bureau of Statistics to visit the Italian Bureau of Statistics. We went to the office of the director of the Italian Statistics Institute, who asked us straight to the point: "Do you know why people are marching in the streets?" We said we didn't know. The director told us that the Italian Statistics Institute had just released Italian economic growth data, which showed that the Italian economy had returned to positive growth. However, the marchers said: "We don't even have jobs yet. How can the economy return to positive growth?" They thought. The Bureau of Statistics figures are fake. But on the other hand, the Italian Ministry of Economy complained: "The Italian economy has already returned to positive growth, and the statistics are too lagging."

Italy is a developed economy, and there is still a big contrast between the feelings of certain groups and the data released by the government.

Why some statistics fight

Economic Observer: Some people question the statistics because they find that data from different channels are very different, which is what we often call the "data fight" problem.

Xu Xianchun: There are inconsistencies between statistical data and there are many situations. Take GDP, for example. Since the establishment of the accounting system, China's regional GDP summary data and the national GDP data calculated by the National Bureau of Statistics have been inconsistent for a long time. Most of the time, the local summary data is significantly higher than the data calculated by the National Bureau of Statistics. So some people question, isn't the national GDP data compiled from regional GDP data? How come the national GDP data is significantly lower than the aggregate regional GDP data?

In fact, national GDP data is not a summary of regional GDP data. From a production perspective, national GDP data is calculated by the National Bureau of Statistics based on industry-by-industry value added and then summarized. Before the fourth national economic census in 2018, China adopted a hierarchical accounting system, that is, the National Bureau of Statistics calculated national GDP data, and regional statistical bureaus calculated regional GDP data in accordance with the accounting methods formulated by the National Bureau of Statistics. Before 2012, there was no direct reporting statistical survey system for enterprises. In some places where the economic development was not very good but they wanted political achievements, it was possible to intervene in the data. Therefore, the summary of regional GDP data was significantly higher than National GDP data calculated by the National Bureau of Statistics.

After the fourth national economic census, the situation has changed a lot: First, unified accounting of regional GDP is adopted. That is, the National Bureau of Statistics, while calculating national GDP data, organizes provincial-level statistical bureaus to conduct unified accounting at the provincial level. Regional GDP data; the provincial statistics bureau organizes the municipal statistics bureau to calculate the city-level regional GDP data; the municipal statistics bureau organizes the county-level statistics bureau to calculate the county-level regional GDP data. The data is determined step by step from the national to the provincial level to the municipal and county levels, and the national GDP data and the regional GDP summary data can be basically connected. Second, a statistical survey system for enterprises to report directly to the Internet has been established at that time. Industrial enterprises above designated size, qualified construction enterprises, wholesale and retail enterprises above designated size, accommodation and catering enterprises and real estate development enterprises above designated size, through The online direct reporting system directly reports enterprise statistical data. Provincial-level, municipal-level, and county-level statistical bureaus are granted the authority to review and summarize enterprise data in the region, but do not have the right to modify it. If problems are found, they need to be traced back to the enterprise. Enterprise You can only modify the data if you think it is wrong, thus avoiding the interference of intermediate links on the statistical data. The third is to conduct sample surveys on some small businesses to make up for some data deficiencies in the past. The fourth is the GDP data between two census years. Generally, the GDP data calculated using regular statistical survey data are revised based on the data of the next census year to ensure the continuity between the regular annual GDP data and the census year GDP data. sex and comparability.

How are national GDP data finally determined? The National Bureau of Statistics will hold a GDP data review and evaluation meeting every quarter. The leaders of the National Bureau of Statistics and the responsible comrades of each business department will review and evaluate GDP data and professional statistical data. If there is any discrepancy between GDP data and professional statistical data, or between professional statistics, If there are contradictions between statistical data, it is necessary to find out the reasons and correct the problematic statistical data.

Various professional departments often send professional statisticians to local areas to check the quality of the data. If they find that the statistical data is inconsistent with the actual situation or even fraudulent, they will resolutely correct it; in 2017, the National Bureau of Statistics established a Law Enforcement Supervision Bureau to enforce the law against statistical fraud. Inspect, investigate and deal with in accordance with the law to ensure the authenticity of statistical data.

Economic Observer: Is there another important factor: the misunderstanding of some important statistical indicators?

Xu Xianchun: Statistics is a highly professional and technical work. For some important statistical indicators, there are often inconsistencies between readers’ understanding and statistical specifications. For example, some people may think that residents’ disposable income is very simple. However, in government statistics, residents' disposable income includes residents' disposable income in fund flow accounting and household disposable income in household surveys. The two statistical indicators are different in terms of basic purposes, caliber range, data sources and calculation methods, so the data performance will naturally be different. For example, from 2018 to 2020, residents’ disposable income in capital flow accounting was about 1.3 times that of residents’ disposable income directly calculated using household survey data, and even reached 1.4 times in some years.

Another example is GDP, which we often use. When it comes to GDP, many people will think that they know it very well, but how many people have a clear understanding of the basic theory and accounting methods of GDP? To give a typical example, in the first quarter of 2020, according to data released by the National Bureau of Statistics, China's economy fell by 6.9% year-on-year, but during the same period, total retail sales of consumer goods fell by 19% year-on-year, fixed asset investment fell by 16.1%, and the goods trade surplus according to customs statistics down 80.6%. Some people questioned how could GDP only drop by 6.9% when total retail sales of consumer goods, fixed asset investment and goods trade surplus all dropped by more than double digits? The GDP decline is certainly underestimated.

In fact, these people are not very clear about the relationship between relevant statistical indicators. They directly use changes in total retail sales of consumer goods to judge changes in consumer demand, directly use changes in fixed asset investment to judge changes in investment demand, and directly use customs statistics on goods. Changes in the trade surplus determine changes in net export demand. Take consumer demand as an example. Consumption demand includes household consumption expenditure and government consumption expenditure. Resident consumption expenditure also includes expenditure on goods and services purchased through currency and expenditure on goods and services not purchased through currency (i.e., virtual consumption expenditure). For example, farmers Self-produced grains, vegetables and fruits belong to household consumption expenditures, but because they do not enter the market and are not traded with currency, they are not included in the total retail sales of consumer goods. Another example is that resident-owned housing services are not commodities and are not included in the total retail sales of consumer goods.

Since most virtual consumption expenditures were not affected by the epidemic, virtual consumption expenditures not only did not decline but increased in the first quarter of 2020. Government consumption expenditure includes expenditure on public services and expenditure on personal consumption goods and services undertaken by government departments. In the first quarter of 2020, public service expenditures fell slightly. The above-mentioned performance of virtual consumption expenditure and public service expenditure in the first quarter of 2020 has to a certain extent slowed down the impact of the epidemic on household consumption expenditure and government consumption expenditure. Therefore, consumer demand did not drop as sharply as the total retail sales of consumer goods, but the decline was relatively small. Small.

How to use government statistics correctly

Economic Observer: Can you provide some suggestions on how we should use government statistics?

Xu Xianchun: Government statistical data are valuable economic and social resources. Correct use of government statistical data is essential for objectively and accurately judging the economic and social development situation, formulating scientific and reasonable economic and social development policies, and drawing academic research with important theoretical and applied value. Results are all important. To use government statistics correctly, the following four principles should be grasped:

The first is to correctly select government statistics based on the issues being studied. There are three principles for correctly selecting government statistics, namely representativeness, good quality, and consistency.

The second is to accurately understand the classification standards, survey scope, survey methods and collection methods of government statistical data. The classification standards, survey scope, survey methods and collection methods of government statistical data will be adjusted as the actual situation of economic and social development changes. Only by accurately understanding their changes can they be used correctly.

The third is to accurately understand the scope and calculation methods of government statistical indicators. The caliber range and calculation method of many indicators in Chinese government statistics have changed, and some changes have been very drastic. For example, the caliber range of labor compensation involved in the income-based GDP accounting has undergone two major adjustments. I don’t know. Their changes will directly affect the scientific nature and objectivity of the research results.

Fourth, we must accurately grasp the scope of application of government statistical indicators and the relationship between relevant statistical indicators. Accurately grasping the relationship between relevant statistical indicators is mainly to solve the problem of being unable to obtain the data of the required statistical indicators and being forced to use related statistical indicators that can obtain data to make correct inferences.

Economic Observer: What if historical data is involved?

Xu Xianchun: When using historical data, the most important thing is to pay attention to whether the statistical system of the corresponding statistical indicators has changed. With the rapid development of economy and society and the continuous advancement of science and technology, the statistical system is constantly changing, including the definition of statistical indicators, caliber range, data sources, calculation methods, scope of application, etc., which may all change. If some historical data, especially GDP accounting historical data, undergo major changes in terms of caliber range, data sources, calculation methods, etc., in order to ensure the continuity and comparability of historical data, the National Bureau of Statistics will systematically revise the corresponding historical data. For revised historical data, be careful to use new data and do not use unrevised data. Otherwise, there may be problems of discontinuity and incomparability of historical data, which may affect the scientificity and objectivity of the research results. The National Bureau of Statistics has established a national database in which historical data are revised and comparable. However, the data in the early published statistical yearbooks are likely to have not been revised, and care should be taken when using them.

For historical data that has not been revised, attention must be paid to its continuity and comparability. If you make revisions based on the needs of your research questions, you must understand in detail whether there have been major changes in the caliber range, data sources, calculation methods, etc. of the corresponding statistical indicators, as well as what kind of changes have occurred, and look for corresponding reference materials. For example, the professional statistical yearbook of the corresponding year, because the statistical data released by the professional statistical yearbook is relatively detailed, it is convenient to revise the corresponding data according to the needs of the research problem and in response to changes in the caliber range, data sources, calculation methods, etc. of the corresponding statistical indicators.

Economic Observer: Can Mr. Xu give an example of changes in the content and scope of statistical indicators due to economic and social development?

Xu Xianchun: A typical example is the total formation of fixed assets in GDP accounting using the expenditure method. In the international standards of national economic accounting before 2008, research and development expenditures were treated as intermediate inputs, and only in the 2008 international standards were they treated as fixed capital formation. If it is treated as an intermediate input, it will not be included in the GDP; if it is treated as a fixed asset, it will be included in the GDP.

According to current international standards for national economic accounting, gross fixed asset formation does not yet include data on capital expenditures. However, the importance of data assets has become increasingly apparent. For Internet platform companies like Didi Chuxing, data assets have become their most important assets. Their investment in data collection, storage, development and maintenance has exceeded that in equipment, Investment in office building procurement and maintenance; for some companies that are undergoing digital transformation and upgrading, traditional fixed asset investment is shrinking, but investment in data assets is growing rapidly. I have led my students to investigate more than 80 companies in 11 provinces, most of which are companies in economically developed provinces such as Guangdong, Shanghai, Beijing, Jiangsu, and Zhejiang. The digital transformation trends I have seen are very obvious. Although experts and scholars have different opinions on how to handle data capital expenditures in GDP accounting, it is an indisputable fact that in the era of digital economy, the role of data assets in economic development is rising rapidly, and research on statistical theories and methods must Keep up quickly, otherwise the important role played by data assets in economic development will not be reflected.

The current explosive growth of data and the rapid development of the digital economy have posed a series of challenges to government statistics. For example, how to accurately estimate the value of data assets and their contribution to economic development? For another example, how to comprehensively and objectively measure the added value of the digital economy? Many Internet platforms now provide a large number of free or low-cost services, including search engines, navigation, WeChat communication, online ticket booking, etc. Since they are free or low-cost, the current GDP has not yet fully reflected these services. The value created by the activity. Another example is that Internet platform companies have created a large number of flexible and diverse employment methods, including online ride-hailing, takeaway riders, etc. Most of these gig economy employees have not signed labor contracts with Internet platform companies, and the added value they create is often little or no. fully reflected in GDP.