Big Data Age: Is GDP statistics outdated?

Macroeconomic statistics, such as gross domestic product (GDP), unemployment, inflation and trade deficits, are often used as a leading indicator of macro analysis. For some time, GDP has even become a measure of economic success or failure. Nowadays, the internationalization of multinational production and sales, the rapid development of modern service industry, the integration of network economy and real economy, so that traditional macroeconomic data can no longer cover the whole picture of economic development. Recently, international public opinion on how to make up for big data application How to fill this loophole to launch hot discussion. In the big Data age, a series of traditional economic indicators, such as GDP, is really "outdated"?

1. What is the traditional data "missing"?

July 28, 2009, the National Bureau of Statistics issued nationwide urban units working average wage data, average monthly wages in various industries have risen to varying degrees. However, some netizens say their wages are "increased". Since then, the term "increased" has begun to be used in a situation where personal experience is inconsistent with macroeconomic data. Why is there a discrepancy? Experts explained that structural differences, such as regional factors, are likely to lead to some of the things that have not been accounted for. At the same time, average data often ignores individuals at the micro level.

The "unreliable" of traditional macroscopic statistics is not only in one country, but with the further development of globalization, the reference value of some trade data is also greatly reduced. For example, the U.S.-China trade deficit will increase by about 200 dollars every time an iphone is traveling long distances from Foxconn to the United States. In fact, more than 10 companies in at least five countries in the world are supplying iphone accessories, and only about 10 dollars per iphone value is eventually flowing into the Chinese economy.

Zakari Carabel, president of the US Rivertwise Research Institute, said in a recent article on misleading lead indicators that conventional trade statistics do not calculate the value added to the middle stage, and that their portrayal of bilateral trade relations is distorted.

Similarly, the significance of GDP as a statistical indicator is declining. The traditional view is that the growth of GDP is the stability of employment, that is, the increase of residents ' income and so on. However, since 2008, although Britain's GDP growth rate has been roughly zero, but its employment rate has increased. GDP growth in developed countries is not as high as in some developing countries, but we find that the quality gap between these countries is widening.

If a steel mill forms pollution, its cost of cleaning up the pollution will increase GDP, and the health costs of the contaminated workers or residents will increase GDP. Conversely, if a country uses long life LED lights to replace the traditional bulbs, the result is less electricity consumption, and technological progress just reduces the GDP value.

The widespread criticism of GDP is that it does not involve the environmental damage caused by the production of goods and services. Although the growth rate of GDP is seen as a measure of progress, it has never been used to measure happiness or social welfare. Zhaoyan, dean of the Institute of Statistics of Renmin University of China, said that the statistical reflection of centralized system of national accounting statistics is not effective for some such as environmental protection and ecological economy, transportation, tourism, government revenue and expenditure and income distribution regulation, monetary and financial system, science and technology innovation and cultural industry statistical demand.

Diana Coille, a former UK Treasury consultant, said in his February article, "What is missing from the measurement of GDP economic performance", that official economic data, such as national accounts, has a variety of sources, but that surveys of individuals and businesses constitute their backbone. By sending a form to some companies or sending investigators to collect information about prices offered by different stores, these conventional investigative methods are almost impossible to keep up to date as the economic structure itself changes. As an obvious example: the questionnaire does not actually take online shopping into account, and online prices are likely to be lower.

Harvard University scholar Feng Jinming pointed out that the traditional economic statistics have two defects: first, hysteresis, and the second is low frequency. For example: The consumer price index (CPI) in the country generally has a lag period, in China, for example, usually wait until next month around 9th to release the CPI data last month.

Experts say that with the same value demand, people may pay more attention to the sales data of a large electric dealer (such as Taobao) without needing to pay attention to the statistic index of "Social consumer retail sales". In releasing the results, it is not enough to tell others only one result (such as national GDP data), but also need to provide more convenient and efficient use of the user through visualization, interaction and so on, providing more detailed "meaning" information.

2, "garbage" data How to shine again?

Alibaba based on the online retail Price index (ISPI) of Taobao, and based on the real-time data of online transactions, reflects the trend of retail price and trading volume of 10 kinds of goods and service categories, such as food, tobacco and alcohol and supplies, clothing and so on. The index and the official CPI chain Index showed a linkage, and at the key turning point showed a certain leading position, can help to understand inflation, economic growth, residents consumption and other macroeconomic indicators.

"The traditional economic statistics in the future will be large data." "Feng Jinming that in the past production statistics remain at the industry level, or limited to enterprises above the scale, and the future may be for all enterprises; The traditional consumption statistics are mainly based on sample surveys and may be specific to each family or individual in the future; The traditional price statistics contain only thousand commodities, involving tens of thousands of survey sales outlets, And the future may be tens of thousands of kinds of goods, all online vendors and most of the offline sales outlets. With the maturity of large data technology, "sample as a whole" will become a trend, the sampling is more and more unimportant.

"Compared to traditional economic statistics, the major data-induced changes are mainly in four aspects: faster, more accurate, broader and finer." "These characteristics benefit future industry policies and macroeconomic decisions," Feng Jinming said.

With the popularization of computer and Internet and the development of electronic commerce, more and more economic behavior is recorded. As large data-related technologies mature, the amount of "junk" data that the public and private companies have accumulated in the past is likely to glow again. For example, using traffic accidents and crime data to guide the distribution of police forces, using consumption and tax data to guide the distribution of income, using passenger data to guide railway and civil aviation deployment, using Internet keywords to disseminate data for epidemiological prevention and so on.

Zhaoyan that in the large data age, the government can set up a data centralized platform at the national level to co-ordinate the management of economic and social data, including the space of economic and social statistics, covering all aspects of social life, can be based on the management of various sectors of the service industry administrative records, business statistics, regulatory information, The scientific statistic method System of statistical accounting in the construction industry.

"Statistical analysis can not be on data, but also to Suegen, in-depth analysis of the reasons behind the changes in data." To achieve this, we need to pay attention to macro-analysis, but also to focus on micro-analysis. "said Jiantang, director of the National Bureau of Statistics.

The traditional average index conceals the specific development trend of region and individual. For example, it would be wrong to treat unemployment as a national problem, because there would be a big difference in employment trends as geographical, gender and educational levels changed. But none of these problems is reflected in the unemployment rate, and the policies adopted through the unemployment rate started off in the wrong direction.

Experts say these indicators have little effect on small businesses or individuals. Individuals decide whether to start a business or buy a house now should not refer to the unemployment rate or the number of houses in the country. For people who want to open a clothing store or restaurant, CPI often has no reference value. On the contrary, entrepreneurs should pay attention to the local market dynamics and the trend of the industry. 30 years ago, such statistics could be quite difficult, and today it is only a few hours on a computer. What we need in the big Data age is to tailor metrics specifically for the specific needs of governments, businesses, communities and individuals, which is now possible.

3. How does GDP "hug" big data?

November 19, 2013, the National Bureau of Statistics and Baidu, Alibaba and other 11 enterprises signed a large data strategic cooperation framework Agreement. The aim is to jointly promote the application of large data in government statistics, and continuously enhance the scientific and timely nature of government statistics. Jiantang pointed out that in the past, traditional statistical methods were designed by statistical professionals to design statistical forms, and companies were found to investigate production data from the directory library. However, in the large data age, there are many business entities in the directory database can not find, but there are trading activities and transaction data. These data are readily available, and are massive, unstructured, and non-standard. Statistical departments to use existing data, such data can be obtained daily, is a statistical treasure.

"At present, an era of mass production, sharing and use of large data is coming. This is mighty, unstoppable historical trend, who has a large data, who occupies the commanding heights, has achieved the initiative. Jiantang said that for the government, large data will become the information base of macro-control, national governance and social management.

However, the statistics of large data are not so simple to achieve. Zhaoyan points out that in the current information data sharing platform of various departments, administrative records, business statistics, activity statistics, financial information, etc., are not complete can adapt to the transformation of large data analysis.

"Inconsistent indicators, inconsistent indicators, inconsistent time, spatial inconsistency, inconsistent indicators, inconsistent classification, coding inconsistent, such a messy database, basically not even regular statistical collation, statistical description and analysis can not do." Zhaoyan said that in the large data age, China needs to promote the concept of sharing, cooperation and synergy, so that the government functional departments break the traditional constraints of their own camps, and truly open the sharing of departmental data to achieve the maximization of the overall interests. For a long time, various departments of our country pay much attention to their own interests, the so-called sharing work is only to deal with. But in the era of large data, this will directly restrict the development of China's economic and social and industrial upgrading.

"It is necessary to point out that the large data in traditional economic statistics is complementary, not a substitute." Feng Jinming said that the traditional statistical methods in economic growth, taxation, trade, income distribution and other areas of the leading advantage of statistics, and large data in terms of prices, inflation, unemployment, consumption and other aspects of the statistical advantages.

So how should GDP statistics be perfected in the big data age? In this regard, experts believe that the GDP accounting data should be optimized to ensure the accuracy and authenticity of GDP accounting data. The results of GDP accounting in the future can not only be total and speed, but also must have various industries, industries, types and different regions, different regions of the fine items, the data of the sub-item, in order to meet the different objects, different industries personalized needs. The government statistics will change from the grasp of "macroscopic" to the application of "microcosmic", and the GDP will gradually become the popular index that both macro and micro are applicable. At the same time, it should be more vivid and understandable by graph, image, map and animation to show the data size, explain the relationship between data and development trend, and provide people with easy to understand and easy to use. (Guangming journalist Chen Heng)