In an interview in November of 2013, the Vice Director of China’s National Bureau of Statistics, Xu Xianchun, announced that China would soon be converting their method of national accounting to the 2008 SNA standard, from the old SNA 1993 standards. The new standards (which have largely already been implemented in US, Australia, Korea, Nigeria and others) seek to adjust the way GDP is measured in an effort to better reflect the evolving nature of production, work, and capital accumulation. These changes affect everything from how military expenditures are classified, to how water is valued and accounted for on the national balance sheet (if at all). However, for many countries – mostly in the developed world – the largest impact of this change in standards will be on how expenditures on R&D, especially in the information and communication technology (ICT) sector, will be treated.
In the US, the change was made in large part in 2013 when the BEA released the first series of revised NIPAs which included R&D expenditures as an investment, which gets counted in final demand, and not as an intermediate good, which is not counted. Over the entire period of the revision (1929 – 2013+), this change does not materially affect the long-run historical average growth rates. However, the level of GDP has been revised higher going back to 1930. Many other countries are migrating to this standard in 2014 as well, including all of the European Union and Korea.
In the interview, Xu did not mention when exactly China would transition to the new system of accounting. However, reading two most recent GDP reports, the footnotes indicate that they are “currently transitioning” (1H 2014 report), as opposed to “under consideration” in the 2H 2013 report, indicating this process is underway. It is important to note that not all of the specifications under the 2008 SNA guidelines will necessarily be implemented immediately. We would expect the changes to be announced in the beginning of 2015.
Narrowly considering the capitalization of intangible capital from R&D investment, we see three notable impacts: 1) the level and change of GDP, 2) labor share of income, and 3) productivity.
The ultimate impact of this change depends somewhat on which standards of the 2008 SNA China implements and when. However, similar to other countries who have implemented 2008 SNA or similar standards, the largest change will come from the treatment of R&D. The magnitude of the impact on aggregate income and production depends on the economy’s R&D intensity. While China spends a lot on R&D in level terms relative to most other countries, on a per unit of GDP basis, it is still quite low. Other measures of innovation output, such as patents per researcher, patents per capita, publications per capita – all of which lag behind the most developed economies – corroborate China’s developing status. However, the pace of both inputs (investment in R&D) and outputs (patents/publications) in China is increasing faster than the overall economy, meaning on a flow basis the accounting change will represent a growing share of GDP growth.
Using the parameters from several studies by Eurostat and the OECD, our rough estimation is that after the revision, China’s GDP level will increase from a 2011 baseline by about 0.1-0.2%. The impact on GDP growth will be similar for recent years, but should increase as R&D expenditures increase. We extrapolated the trend in R&D expenditures from 1997 and used the resulting coefficient to project forward using the IMF WEO forecasts for China’s GDP. By the end of our forecast period using these growth assumptions, we estimate R&D expenditures as a percent of GDP should reach approximately 3.5% the 1.8% in our 2011 baseline estimate.
Exhibit 1: R&D intensity vs. estimated impact of R&D capitalization on GDP
Source: OECD, Eurostat, China National Bureau of Statistics, author’s calculations
Exhibit 2: China’s R&D expenditures as a % of GDP
Source: China National Bureau of Statistics, author’s calculations
Exhibit 3: Estimated Impact of R&D capitalization on Nominal GDP Growth
Source: China National Bureau of Statistics, author’s calculations
If China’s R&D expenditure intensity increases, so then should GDP growth. Using these simplistic assumptions, we would estimate that China’s GDP growth will be on average 0.15 – 0.20 percentage points higher than the IMF currently projects through 2019 (ceteris paribus), simply due to the capitalization of R&D.
Labor Share of Income
One less-discussed issue is the impact this accounting change (both in China and the US) has on the relative factor shares of income. This is due to the fact that the increases in investment accrue to income to capital inputs, rather than labor inputs. In other words, although the measured income “pie” is getting larger, only the capital “slice” is increasing to make up the difference. In the U.S., the inclusion of intangibles shows an even more pernicious trend in the labor share of income than previously thought.
Using the above framework, we would estimate that the accounting change would subtract an estimated 0.05 percentage points from labor share of income since 1992. While this is not a tremendous change, increasing the labor share of income in the economy is a priority of the government and this change would negatively impact that outcome. It is more likely, however, that other supply-side reforms that the government is enacting will have a much larger, positive impact on labor’s share of income.
Quantifying the impact on productivity is more of a challenge. Growth in productivity, broadly measured in output per hour, can be broadly broken down into three components: labor composition (changing mix of workers and skills), capital deepening (more mechanical or technological inputs), and a residual called multifactor productivity (which historically is assumed to be an interaction of several of these other components). If R&D were counted as an investment, the requisite output increase would contribute to the capital deepening component, largely at the expense of labor and multifactor components. In the US, for example, it is estimated that R&D capitalization caused the labor share of productivity growth from 1995-2003 to drop from 14% to 11%, multifactor productivity share dropping from 51% to 35%, and capital deepening share increasing from 35% to 55%.
In summary, we view this GDP accounting change as a positive step for China. It is not, as has been remarked by some, an attempt by China to bolster its GDP growth data in the face of a measured economic slow-down. Rather, it is part of an international forward-thinking effort to have its national statistics better reflect the nature of economic activity.
 EU countries are classified by R&D expenditures: high, medium-high, medium-low, and low. High: Finland, Sweden. Medium-high: Austria, Netherlands, UK. Medium-low: Belgium, Denmark, France, Germany. Low: Czech Republic, Ireland, Italy, Luxembourg, Portugal, Slovakia, Slovenia, Spain