The Status of GDP Statistics
In early 2019 problems with Japan's Monthly Labor Survey were revealed, drawing attention to the trustworthiness and role of statistical data. Serving on the Statistics Commission, the author of this article not only finds these events impossible to forget, but also personally regrets what happened, as someone who shares the responsibility for providing better governmental statistics.
However, this article will not explain in detail the uncovered problems with the labor statistics. Instead, its purpose is to use these problems as an opportunity to rethink the importance and role of statistics, as well as the issues currently impacting Japanese governmental statistics.
The volume of Japanese governmental statistics is immense, with 53 different fundamental statistics alone. These statistics are managed by different ministries and their characteristics cannot be easily summarized. This article will therefore focus on GDP statistics, which receive the highest attention among government data and are used to formulate various policies.
GDP, a nation's gross domestic product, is one index within an accounting system used for measuring the entirety of a country's economic activity, called the System of National Accounts, and this System of National Accounts itself consists of processed data compiled using primary statistics and the like created from administrative record information and questionnaires submitted directly to statistical agencies (Note 1). This means that the accuracy of the individual primary statistics used for estimates of national accounts also impacts the accuracy of the national accounts. Considering the accuracy, role, and issues of GDP values in this light leads to a broader discussion of issues with many governmental statistics.
The History and Changing Role of GDP Statistics
The System of National Accounts, i.e. the system of accounts that includes GDP statistics, is actually not that old. The lesson that it was necessary to comprehend trends in the entirety of the economy derived from the experience of the Great Depression of 1929. In other words, policymakers during the Great Depression of almost a century ago did not have data showing how much the whole economy had contracted. Starting in the 1930s, economists like Simon Kuznets, Richard Stone, and James Meade who would later receive the Nobel Prize in Economics contributed to the establishment of national accounts, and the estimation methods of the system of national accounts were established by the United Nations after World War II. The history of GDP statistics hence does not yet exceed the length of a human life (Note 2).
Just as a human life is full of changes, however, the role of GDP data, too, has changed greatly over a period of approximately 80 years. GDP statistics after World War II were often used as an index for measuring the state of recovery of countries involved in the war. GDP data were thus expected to capture how much the scale of an economy had expanded via the economic activity of the entire country.
However, once the economies of developed nations had recovered and reached a certain economic scale, economists started looking to GDP as an index of short-term economic changes. Since GDP changes started to also impact stock markets, economists sought to understand the state of the underlying economy as quickly as possible and began to demand earlier delivery of GDP data. This resulted in an emphasis on rates of change, such as seasonally-adjusted quarter-to-quarter changes, rather than on the scale of the economy.
For compilers of statistics, GDP is merely one part of the System of National Accounts and was originally estimated on the basis of first-order data and the like covering as much economic activity of a given country as possible in order to construct this system. These GDP data are called yearly estimates and the GDP series for the previous fiscal year is published each year in December. What we often see in newspapers and the like on the other hand, are called preliminary quarterly estimates and estimated with yearly estimates as a starting point and focusing on rates of change in monthly and quarterly primary statistics, etc. Private sector users want these quarterly preliminary estimates published as quickly as possible. In the case of Japan, there are primary and secondary preliminaryestimates, with the primary estimates published approximately one and a half months after the end of the quarter on which the estimates are based. These primary preliminary estimates only came into being in the twenty-first century. The breadth of revisions for primary preliminary estimates, secondary preliminary estimates, and yearly estimates is frequently debated and the trade-off between accuracy and speed has yet to be resolved. (Note 3).
Imminent Drastic Changes to GDP Statistics
Since the first version of the international standards for GDP statistics was established in 1953, the standards have been revised three times: in 1968, 1993, and 2008. In 2015 International Association for Research in Income and Wealth and the OECD jointly held a conference regarding the future of national accounts, and a look at the speech transcripts shows that demands for national accounts have increased each year. Van de Ven (2017), who presented at the conference, summarizes the long-term issues for GDP statistics as follows:
- How should value added for the entirety of a single country be measured amid increasing globalization and international fragmentation of businesses?
- How should knowledge-intense economies stemming from the IT revolution be captured?
- How should the distribution of income and wealth be integrated within the system?
- Will GDP become able to reflect improvements in living standards and quality of life?
With regard to item 1, there exists a large volume of research discussing the relationship with productivity improvements of individual businesses in conjunction with the globalization of business activities. Furthermore, there is the question of how to use an international input-output table to connect the results of this research to estimate the GDP of an entire individual country, and the solution will not be easy.
As far as item 2 is concerned, the integration of such factors into national accounts through estimates of intangible assets such as software and R&D expenses is already underway. In the future, approaches such as appraising the value of creations in the arts sector as intangible assets will likely advance as well, including in Japan.
Item 3 is an extremely important question, but the distribution within institutional sectors of national accounts, such as businesses or households, can be measured to a certain extent. The main problem is the distribution of income brackets, which may be addressed by including other statistics such as the Comprehensive Survey of Living Conditions.
Item 4 can be considered an issue that goes beyond conventional GDP. The development of a new index of national wealth as introduced by Managi and Tanaka (2018) is a representative example, but in order to decide whether such an index can become an index that goes beyond GDP, which is limited to economic activity, it will likely be necessary to compare and evaluate the range of used it will have on the part of the private sector and the government.
Expanding Statistics as Infrastructure
Though it has significant problems as outlined above, the GDP index is unlikely to disappear. Without GDP statistics, we would largely lose macroeconomic empirical analyses of GDP fluctuations and growth. In this sense, promoting an expansion of national accounts through, for example, the development of satellite accounts, seems to be a more appropriate direction for the future.
However, the lack of personnel in the statistics sector, as revealed by the recent problems with statistics, is grave. There is an argument that the aging of Japan's physical infrastructure has been a factor slowing down the recovery after the recent natural disaster, and governmental statistics, as intangible social infrastructure, can be seen as causing structural fatigue as well. As is evident from recent debates, it is difficult for intangible assets to respond to new technologies when they are not complemented by human assets and the like. In this sense, cultivating and maintaining qualified personnel in the statistics sector is essential for preserving and expanding this intangible infrastructure. In recent years the government has touted evidence-based policymaking, and the maintenance and expansion of underlying governmental statistics is urgently needed in order to maintain such policies as well.
October 21, 2019