Investment and economic growth: a dilemma in China

Abstract

The correlation between China’s investment rate and economic growth rate has changed from positive to negative in recent years, and the topic of investment efficiency and overinvestment has regained attention. Based on China’s special institutional arrangements of promotion tournament and considerable discretion in local policy implementation, this article collects national and provincial panel data from 1978 to 2022 and establishes four main findings. First, the increase in investment rate stems in part from the pressure to ‘secure growth’. Second, the negative impact of capital user cost on investment is manifested only before 2007. Third, interregional interaction is very important for provinces’ investment behaviors. Finally, while high investment rate accumulates more capital, it has an adverse effect on the overall productivity. In summary, the Chinese experience of the ‘double-edged sword’ shows that investment in relation to economic growth should be maintained at an appropriate level.

Keywords:

• Investment
• capital user cost
• productivity
• economic growth
• golden rule

JEL CODES:

• O47
• E22

Acknowledgments

We are grateful to the National Natural Science Foundation of China (Grant No. 72003188), the National Social Science Fund of China (Grant No. 18CJL017) for financial support. The authors alone are responsible for all errors.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Investment rate is defined as fixed asset investment as a percentage of GDP. Data on fixed asset investment (in current prices), GDP (in current prices) and GDP Index (previous year = 100) used to compute investment rate and economic growth rate (official) are sourced from the Chinese National Bureau of Statistics website. Data on real GDP at constant national prices (RGDPNA) for calculating economic growth rate (PWT) are obtained from PWT version10.0 (released on June 18, 2021 with data updated to 2019).

2 Years after 2020 are not included here as the latest version of PWT is only updated until 2019.

3 We refer to the method adopted in Bai et al. (2006) and Bai and Zhang (2014a) to update the return on capital data to 2022.

4 China does not appear to be the only exception. Using panel data for 49 economies, including China, that consistently or regularly announce growth targets, Xu and Liu (2017) finds that, on average, a one-percentage-point change in growth targets in these economies during the target management period is followed by a one-percentage-point change in the actual economic growth rate. Further analysis suggests that this "self-fulfilling" growth target stems primarily from a corresponding increase in capital accumulation.

5 After 2018, the Chinese NBS no longer publicly releases the monthly "fixed asset investment" data, but only the monthly "cumulative growth in fixed asset investment" one, and we deduce the corresponding values of the former for each month after 2018 based on the latter.

6 Here the geographical grouping of provinces into central and northeastern regions follow the division of economic regions published by the NBS in 2011. (http://www.stats.gov.cn/ztjc/zthd/sjtjr/dejtjkfr/tjkp/201106/t20110613_71947.htm)

7 There are missing values in the data series of fixed asset investment price index for Tibet, Hainan and Chongqing in some years, which we fill in with national data, and data for Guangdong and Sichuan.

8 Since detailed annual data for total outstanding loans of different maturities are not publicly accessible, we use the arithmetic mean of the interest rate of loans with a maturity of below six months (including six months) and that of loans with a maturity of six months to one year (including one year) as a proxy for the interest rate of short-term loans. Similarly, the interest rate of medium- to long-term loans is proxied with the arithmetic mean of the interest rate of loans with a maturity of one to three years (including three years), that of loans with a maturity of three to five years (including five years), and that of loans with a maturity of above five years. To ensure consistency, we unify the annual data on outstanding loans of various maturities for each province and convert them into the two categories of short-term loans and medium- to long-term loans.

9 http://www.stats.gov.cn/ztjc/zthd/sjtjr/dejtjkfr/tjkp/201106/t20110613_71947.htm.

10 The stimulus package focused efforts on four fronts: expanding government spending, improving urban and rural income, reducing tax, and intensifying financial support. The main body of expanded government spending was a two-year investment plan totaling 4 trillion yuan, which was why the package was often called the “4 trillion-yuan investment plan”. Comprising of 1.18 trillion central government funding, 0.83 trillion local government inputs, 1.41 trillion bank loans and 0.58 trillion nongovernmental contribution from firms, etc., the 4 trillion-yuan investment plan covered the following major spending items: (1) housing for low income groups; (2) rural social safety net and rural infrastructures; (3) railroads, highways, airports, bridges, and other large infrastructure projects; (4) healthcare, culture and education; (5) environmental and ecological projects; (6) innovation and industrial structure upgrading; and (7) Sichuan earthquake reconstruction.

11 We refer to Zhang et al. (2005) and Liu (1998) to obtain the Mincer coefficients for primary school, junior high school, senior high school, secondary vocational school, and vocational college and above (all in relation to being illiterate or semi-illiterate), which are 7.5%, 16.1%, 24.8%, 31.7%, and 38.8%, respectively. Next, we estimate age-specific composite Mincer coefficients (${\mathrm{Mincer}\left(\mathrm{a}\right)}_{\mathrm{it}},$ defined as: ${\mathrm{Mincer}\left(\mathrm{a}\right)}_{\mathrm{it}}={\mathrm{primr}\left(\mathrm{a}\right)}_{\mathrm{it}}\mathrm{*}0.075+{\mathrm{secr}\left(\mathrm{a}\right)}_{\mathrm{it}}\mathrm{*}0.161+{\mathrm{highr}\left(\mathrm{a}\right)}_{\mathrm{it}}\mathrm{*}0.248{+\mathrm{techr}\left(\mathrm{a}\right)}_{\mathrm{it}}\mathrm{*}0.317+\left({\mathrm{collr}\left(\mathrm{a}\right)}_{\mathrm{it}}+{\mathrm{univr}\left(\mathrm{a}\right)}_{\mathrm{it}}\right)\mathrm{*}0.388,$ where $\mathrm{primr}\left(\mathrm{a}\right),$ $\mathrm{secr}\left(\mathrm{a}\right),$ $\mathrm{highr}\left(\mathrm{a}\right),$ $\mathrm{techr}\left(\mathrm{a}\right),$ $\mathrm{collr}\left(\mathrm{a}\right)$ and $\mathrm{ }\mathrm{univr}\left(\mathrm{a}\right)$ respectively represent the share of population aged $\mathrm{a}$ with an educational level of primary school, junior high school, senior high school, secondary vocational school, vocational college, and undergraduate and above.

12 First, we convert GDP into constant US dollars based on the exchange rate calculated with GDP in constant local currency of China and US, which are sourced from the WDI database (updated in December, 2019). Next, we obtain relative income for each province across the sample period based on the population size data for the corresponding province and the US (gathered from the NBS website and the WDI database).

13 Data for total imports and total exports (both in current US dollars) over 1978-1993 are collected from the China Compendium of Statistics 1949-1999, those over 1993-2022 are retrieved from the NBS website. We convert the two series of data into total imports and exports in current yuan based on the US dollar to yuan buying and selling rates, which are calculated based on data at national level (i.e., total imports and total exports given both in US dollars and yuan, collected from the NBS website).

14 Both GDP and inventory are measured in 1978 constant prices. A description of how the former is calculated can be found in previous text. The latter is computed as follows: following Bai et al. (2006), we use the perpetual inventory method to obtain total capital stock and physical capita stock (both in 1978 constant prices) based on total capital formation and total fixed capital formation as proxies for new investment. The data needed for the calculation are from the NBS website and the China Compendium of Statistics 1949-2008.

15 Employed population data are retrieved mainly from the China Compendium of Statistics 1949-2008, supplemented with data from provincial statistical year books, Macrochina, and Statistical Communiqué on the National Economic and Social Development; working-age population aged 15-64 are estimates (see previous description).

16 Local fiscal revenue (in current prices) are sourced from data on the general public budget revenue of local governments published on the NBS website; GDP (in current prices) data for 1993-2022 are from the NBS website and those for 1978-1992 are from the China Compendium of Statistics 1949-2008.

17 Data for 1978-2004 are collected from the China Compendium of Statistics 1949-1999, those for 2005 from Macrochina, those for 2006-2022 from the NBS website, and manual request for public disclosure from provincial branches of the NBS.

18 Data for 1993-2022 come from the NBS website and those for 1978-1992 from the China Compendium of Statistics 1949-2008.

19 Urban population and total population data for 2005-2022 are sourced from the NBS website, those for 1978-1994 are from the China Compendium of Statistics 1949-2008.

20 Here, slowdown is defined as follows: for any given province i in year t, if the average TFP growth rate of the year t + 1 and t + 2 is below 95% of the average TFP growth rate of the year t-1 and t-2, year t is considered to have recorded a slowdown in TFP growth rate. We have also considered the case of 99% and the conclusions are similar. We choose to use values lagged two periods for the following reason: the annual rate of fixed assets put into use (i.e., the ratio of newly added fixed assets to fixed asset investment) is generally reported to be above 50% (collected from the NBS website) at national, regional, or sectoral level, meaning that in most cases it takes two years for investment in fixed assets to be transformed into newly added fixed assets. We have also used values lagged three periods, and the conclusions are more or less consistent.

21 It should be noted that, in the context of the golden rule, a higher investment rate translates into a higher level of income, but not necessarily a high level of consumption, and the investment rate that maximizes the level of consumption should be at a moderate level. It is not entirely comparable with what we discuss here.

22 Our estimates are much smaller than those given in Diao et al. (2012) and Ouyang and Peng (2015). Through computable general equilibrium modelling, Diao et al. (2012) reported a substantially reduced GDP growth rate of 2.9% in 2009 in the absence of the stimulus package. Ouyang and Peng (2015) found in their linear and semiparametric estimation of treatment effect that China’s economic growth rate between 2008 and 2010 would have been 3.2% lower on average.

Additional information

Notes on contributors

Qiong Zhang

Qiong Zhang is an associate professor from the School of Public Administration and Policy, Renmin University of China, Beijing, China. Her research areas include labor, population, and economic growth. She has recently published papers in journals such as American Economic Review, Journal of the Asia Pacific Economy, Journal of Development Studies and Population Studies.

Zhongwen Zhang

Zhongwen Zhang is an assistant professor from the School of Applied Economics, Renmin University of China, Beijing, China. His areas include economic growth and development, and the Chinese economy. He has recently published papers in journals such as Journal of Management World, the Journal of World Economy, Journal of Financial Research and Statistical Research.